Knowledge Graph of Low-Carbon Technologies in the Energy Sector and Cost Evolution Based on LDA2Vec: A Case Study in China

Abstract

Climate change has attracted global attention, highlighting the critical role of low-carbon technologies in addressing environmental challenges. Due to the multidisciplinary nature, complexity, and diversity of research content on low-carbon technologies, a comprehensive overview is still limited. This paper uses bibliometrics analysis to discuss the research status and hotspots of low-carbon technology from a macro-perspective. The LDA2Vec topic recognition model is adopted to identify key technical terms, and CiteSpace software 6.3.1 Advanced Edition is used to conduct in-depth analysis of the development trajectory of low-carbon technology. After checking the frequency of the relevant keywords, four key techniques were identified. In order to further analyze the research results, the learning curve theory is used to predict the cost development trend of key low-carbon technologies. The results show that: (i) low-carbon technologies play a key role in the energy sector and have a potential impact on policy making, and the cost of related technologies will be significantly reduced in the next few years. (ii) Global low-carbon technologies have entered an important period of development, but remaining challenges need to be addressed by optimizing technological performance. (iii) It is very important to strengthen the research on hydrogen production technology and photovoltaic power generation technology; the cost reduction in hydrogen production technology is still significant and there is room for further optimization. (iv) To effectively address the high costs and technical barriers associated with emerging low-carbon technologies, increased funding for research and development is critical.

1. Introduction

The rise in extreme weather events globally underscores the urgency for immediate climate action. The announcement in September 2020 outlining China’s goals of reaching a “carbon peak” by 2030 and “carbon neutrality” by 2060 reflects China’s commitment to responsible leadership and high-quality development. On 22 September 2020, at the 75th session of the United Nations General Assembly, Xi Jinping put forward the vision of China striving to peak carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060, and stressed the importance of the implementation of this goal many times at important conferences at home and abroad [1]. Achieving the “double carbon” target (i.e., carbon peak and carbon neutrality) hinges on the adoption of low-carbon technologies, which serve as the cornerstone for feasible carbon emission reduction. In order to give full play to the key supporting role of low-carbon technologies in achieving the goal of “dual carbon”, China has formulated a series of policies and guidelines such as the Implementation Plan for Peaking Carbon through Science and Technology (2022–2030), which points out the direction for the development and application of low-carbon technologies [2].

Escalating global pressure to combat climate change has propelled the momentum toward green and low-carbon transformations, bolstered by strengthened policy initiatives. This paradigm shift is epitomized by advancements in wind power, photovoltaic energy, electric vehicles, efficient energy storage, and permanent magnet motors. The economic benefits of emission reductions from green and low-carbon technologies replacing conventional technologies can be used as reference indicators for technology selection, promotion investment, and the formulation of market incentive policies, some of which have already enacted relevant policies for the substitution of fossil energy fuels by low-carbon fuels [3]. Meanwhile, the advancement of clean energy and low-carbon technologies constitutes a pivotal foundation for China to realize its “double carbon” objectives, serving as an indispensable prerequisite for the comprehensive construction of a socialist modern nation [4].

With the increasingly serious problem of global climate change, low-carbon technology has become the key to promote sustainable development. Low-carbon technology is a kind of carbon intangible asset that has a low-carbon contribution value and can continuously bring economic benefits to enterprises, which mainly include carbon sources and carbon sinks. In terms of carbon sources, it includes improving energy efficiency and reducing the proportion of fossil energy utilization. Carbon reduction and carbon-free technologies mainly include IGCC (coal gasification combined cycle) poly generation, ultra-supercritical processes, and photovoltaic technology. A carbon sink mainly refers to the separation of carbon dioxide in the atmosphere through carbon capture, biological absorption, and other technologies, and then removal through carbon storage, pressure, storage in places isolated from the atmosphere, or direct absorption by organisms [5]. In the field of energy, the research on the identification, monitoring and cost development law of low-carbon technology has important theoretical support and practical significance for improving the development of energy technology. In the development process of a new energy technology, although the promotion mode is not the same due to the inherent differences in its technology, it has a similar development path, and the cost is gradually reduced under the effect of the learning efficiency generated in the process of research and development and production, and it goes through the process of basic research—R&D production—large-scale mass production [6]. The study model based on the learning curve is helpful to the formulation of relevant industrial policies by analyzing historical data and studying the change in learning rate of various influencing factors.

However, a common challenge encountered by most low-carbon technologies in China is the absence of core technologies and inadequate research and development capabilities, leading to bottlenecks in their advancement. The development of low-carbon technologies in China started relatively late but has progressed rapidly and covers a wide range of industries. Nevertheless, China still lacks distinctive core technologies of its own, with its research and development capabilities in low-carbon technologies being comparatively modest [7]. China has a significant gap compared with other countries, and attaining the “double carbon” targets continues to pose numerous hurdles [8]. In addition, the cost of low-carbon technologies is an important consideration for their development and a crucial factor in the future expansion of the low-carbon technology industry. Consequently, a thorough examination of the current status of low-carbon technology development in China, alongside an analysis of the cost trends of key low-carbon technologies, holds significant implications for both the technological progress and the strategic transition towards a low-carbon economy in the country. At present, most research studies on hotspot identification in the field are still based on citation analysis and vocabulary analysis. Due to the limitations of the methods themselves, the results of hotspot identification in such research studies are limited in terms of accuracy and objectivity. Based on the LDA2vec topic model, this paper constructs a hotspot recognition index system of influence feature dimension and attention feature dimension from the level of semantic content and semantic relationship, which improves the accuracy of recognition results and is innovative.

Energy types that are affordable, clean, stable, and sustainable are the types of energy that capture global interest for the transition to renewable energy sources and are the means to combat greenhouse gas (GHG) emissions and climate change [9]. To examine how the cost curve of low-carbon technologies in China will evolve in the future, this paper will concentrate on the development trends within the realm of low-carbon technologies in China. Commencing from an exploration of the evolution of keywords pertinent to low-carbon technologies, we will leverage the Web of Science literature database to scrutinize the developmental trajectory, prevailing focal points, and trends of low-carbon technologies in China. Furthermore, we will delineate the cost curves associated with pivotal low-carbon technologies. Against the backdrop of the “double carbon” targets, the analysis of the knowledge graph and cost fluctuations of low-carbon technologies seeks to elucidate their significance in carbon governance, offering valuable insights for the advancement of low-carbon technologies and the reduction in costs associated with key low-carbon technologies in China. In this paper, the LDA2Vec model is adopted, combining the advantages of LDA, to extract text topics effectively, and of Word2Vec, which can fully reflect the characteristics of semantic relations between words. This model can map the semantic information of documents or articles into a continuous space, thus providing a more comprehensive method for topic modeling and semantic representation. Consequently, studying the knowledge graph of low-carbon technologies based on intelligent algorithms and analyzing their cost changes carry significant implications for carbon governance.

2. Overview of Low-Carbon Technology Research in the Energy Sector under the Background of the “Double Carbon” Targets

2.1. Current Development Status of Low-Carbon Technologies in the Energy Sector

In the context of the global low-carbon era, low-carbon development has become the unanimous choice of all countries. China vigorously developing low-carbon technology, has important practical significance, not only to achieve energy saving, emission reduction goals, and industrial transformation needs, but also to actively respond to climate change and change the economic development mode needs. The research and development, application, and diffusion of low-carbon technologies can bring about major changes to the current economic and social system based on high carbon technologies, which is the key to solving the problem of greenhouse gas emissions and global climate change, and is the engine of developing a low-carbon economy [10].

Although China’s low-carbon technology has made great progress on the whole, developed countries are still in a leading position in most areas. Jiang Yuguo and others (2014) believe that there are still the following problems in China’s low-carbon technology: for example, there is not enough awareness of the importance of low-carbon development, the ability to innovate and promote low-carbon technology is seriously insufficient, and there is still a big gap between the accumulation of low-carbon technology in China and Western developed countries, which makes the risk of technology research and development in China much higher than in other countries [5]. Liu et al. (2024) posit that low-carbon technology encompasses the research and development of energy-saving and emission-reducing techniques, which are capable of improving energy utilization efficiency and reducing carbon emissions [11]. Specifically, those who lead in technology will gain advantages in future international competition [12]. Zhao Xiangqin and colleagues (2024) have pointed out that promoting green and low-carbon technological innovation and transformation is a crucial measure that takes into account both high-quality economic development and high-level ecological and environmental protection [13]. Cai Jun (2024) believes that it is necessary to guide, encourage, and facilitate enterprises to increase their investment and application in environmental protection technology research and development, fostering a concept of low-carbon and green development [14]. Zhang Xian and Guo Siyao (2021) suggest that China’s achievement of carbon neutrality goals primarily relies on technological progress [15]. But compared with countries and regions such as the European Union, the United States, and Japan, which started early in low-carbon energy technology, China still lags behind in the application and promotion of low-carbon technology [16]. Specifically, the development of photovoltaic power generation technology has the greatest demand for capital. The electrical machinery and equipment manufacturing industry upstream of the key low-carbon technology industry chain will face severe risks of insufficient capital and labor supply [17].

From a scholarly standpoint, Liu Ping, Yang Weihua, and colleagues have conducted a comprehensive assessment of the advancements in carbon neutrality and emission reduction technologies, encompassing three distinct dimensions: high-efficiency recycling, zero-carbon energy, and negative emission technologies. They have thoroughly analyzed the merits, demerits, and viability of diverse low-carbon technologies, providing valuable insights into their potential trajectories and emphasizing their pivotal significance in attaining carbon neutrality objectives [18]. Subsequently, numerous domestic and international researchers have embarked on a series of investigations into the evolution of these low-carbon technologies [19].

Regarding low-carbon technological innovation, Wang Mingyue and others (2019) have argued that it is influenced by both internal and external factors. They suggest that the construction of various cooperation networks can dissipate uncertainties in the technological innovation process and reduce innovation inputs [20]. Within the technological landscape, low-carbon technologies have pervasively penetrated diverse domains encompassing energy transition, industry, construction, and transportation.

2.2. Methods for Identifying Low-Carbon Technology Themes

The transition to sustainable energy has become an urgent global issue, and there is increasing attention on the development and popularization of low-carbon technologies. However, due to the broad scope and complexity of low-carbon technologies, identifying emerging and promising ones has always been a challenge. In recent years, literature mining methods have been widely used to analyze and mine the scientific literature, aiming to identify key information and development trends from a large amount of the literature [21]. Among the various topic modeling methods, Latent Dirichlet Allocation (LDA) stands as a preeminent technique. However, a crucial prerequisite for its effective application in topic recognition modeling is the specification of the desired number of topics. Despite advancements in perplexity-based and non-parametric iteration methods, the current research landscape lacks a straightforward approach for determining the optimal number of topics within the LDA model [22]. To address this limitation, Wang et al. (2019) advocated for a novel approach that transforms the conventional challenge of selecting topic cluster counts into a clustering problem [23]. This strategy aims to enhance the optimization of news text-oriented topic recognition models. This method uses Word2Vec to embed words in corpus texts, exploring the superior performance of word-related relationships and expressing the implied semantic relationships between topic corpora. Du et al. (2022) proposed a novel text representation method based on word embedding enhancement, which further formed a comprehensive process framework for topic recognition in news texts [24]. Departing from conventional approaches to topic recognition, the proposed framework harnesses probabilistic topic models, such as Latent Dirichlet Allocation (LDA), in tandem with word embedding models like Word2Vec and GloVec. This integrated approach enables a comprehensive extraction and fusion of the text’s topic distribution, semantic nuances, and syntactic intricacies. In summary, advanced literature mining techniques, including Latent Dirichlet Allocation (LDA) and Word2Vec, are pivotal in identifying key trends and promising technologies in low-carbon research, thereby facilitating the economic evaluation of technologies like carbon capture and storage (CCS) and low-carbon power generation. Jung and Lee (2022) emphasize that text mining methods, such as semantic network and main path analysis, are instrumental in uncovering significant research trajectories and core technologies within the vast corpus of the scientific literature [25]. These methodologies not only enhance our understanding of technological advancements but also guide strategic investments in technologies that offer the greatest potential for carbon mitigation and economic viability.

2.3. Research on the Cost of Low-Carbon Technologies in the Energy Sector

Against the backdrop of the “dual carbon” goals, low-cost, low-carbon technologies will become the mainstream of future energy technology development. Therefore, research on the cost of low-carbon technologies is also a significant area that cannot be ignored. Sandra Dermühl and her team (2023) conducted a comparison of the most promising low-carbon hydrogen production technologies from an economic perspective. The analysis revealed that carbon capture and storage (CCS) technology has the lowest unit cost for hydrogen production among them [26]. Li, Y et al. (2014) examined the cost-effective pathways for achieving power decarbonization in China by utilizing various low-carbon electricity technologies and identified the lowest-cost decision through the Generalized Cost-effective Low-carbon Electricity Generation Expansion Planning (GEP) model [27]. The global energy system needs to undergo a revolutionary transformation from today’s fossil fuel-based system to a low-carbon energy system by deeply decarbonizing all energy-demanding sectors. Mohideen et al. (2021) provided a technical overview of low-carbon energy systems, production, and end-use services from the perspective of the hydrogen economy, aiming to develop a sustainable energy future. Although these methods are at different stages of development, it is necessary to address the technical barriers associated with each core conversion technology to accelerate commercialization and facilitate the transition to a circular carbon economy [28]. Napp et al. (2019) proposed a linear cost optimization model for the global energy system to explore the role of advanced technologies in achieving deep decarbonization of the energy system, as well as to provide technical details on how to achieve rapid and deep carbon intensity reduction across energy-demanding sectors [29]. Drawing on their expertise, Morris and colleagues (2019) refined the Economic Prediction and Policy Analysis (EPPA) model at the Massachusetts Institute of Technology. To assess the implications of various potential Emissions Trading System (ETS) strategies on the deployment of carbon capture and storage (CCS) technologies, they integrated the most recent evaluations of China’s power generation technology costs. This enhancement offers a nuanced understanding of the economic dynamics and environmental impacts associated with diverse ETS approaches and their role in advancing CCS technologies [30].

Based on the above analysis, it can be seen that current research mostly focuses on a single field or the development trend analysis of a particular low-carbon technology. There is a lack of comprehensive analysis on the macro-development trajectory of low-carbon technologies in China and the prediction of future hotspots in various fields of low-carbon technologies. Therefore, this study aims to identify representative low-carbon technologies in various fields through literature mining and analysis, and utilize the co-occurrence network and learning curve theory to determine the costs of low-carbon technologies in different fields at various stages. This will provide important support for the low-carbon and green development of various fields in China and the implementation of the carbon peak strategy.

2.4. Development Status of Carbon Emission Trading

Carbon emission trading (CET) creates a system for pricing carbon by setting a cap on total emissions and distributing or auctioning emission allowances. These allowances can be traded, forming a market where their price is determined by supply and demand. He and Song (2022) highlight that this cap-and-trade system effectively internalizes the cost of carbon emissions, prompting firms to factor in these costs in their operational and investment decisions [31].

CET facilitates carbon emission pricing and catalyzes broader economic and environmental transformations through market creation. Lv et al. (2023) demonstrate that the introduction of carbon trading schemes leads to increased investment in environmental protection measures and low-carbon technologies, as firms seek to minimize their carbon liabilities and capitalize on cost savings from reduced emissions [32]. Zhang et al. (2020) provide evidence that carbon trading policies in China have significantly improved market efficiency and led to substantial emission reductions by creating economic incentives for firms to exceed regulatory compliance and invest in cleaner alternatives [33]. Moreover, Xuan et al. (2020) argue that the cascading effects of CET extend beyond individual firms to the broader economy, promoting a shift towards a low-carbon development pathway and contributing to long-term environmental and economic sustainability [34].

3. Theoretical Basis and Data Sources

3.1. Learning Curve Method

In this paper, the learning curve method is selected as the research approach to analyze the cost change trends of various low-carbon technologies. Through statistical analysis of keywords in the field of low-carbon technologies, technologies with a higher frequency of keywords (frequency > 100) are selected as key technologies. The learning curve is a graphical representation with the horizontal axis representing the number of repetitions and the vertical axis indicating the progress of various measurements during the learning process. The theory behind the learning curve suggests that, as tasks are repeatedly performed, they become more efficient and faster, leading to a gradual reduction in the time required to produce a unit of a product. In the context of this study, the new energy industry usually faces high technology costs in its initial development stage, and this trend of cost decline can be described by the learning curve [6]. There are many kinds of new energy industries in China, including solar energy, wind energy, hydro energy, geothermal energy, biomass energy, nuclear energy, and other forms of energy. Among them, some industries are still in the initial stage, and are gradually transforming from small-scale and intermittent production to integrated and continuous large-scale production [35]. With the continuous maturity and improvement of production technology and the accumulation of production management experience, the cost of these technologies still has a lot of room to fall. Reducing the cost to a level acceptable to the market is one of the important prerequisites for commercialization.

3.2. LDA2vec

The integration of LDA2vec into the learning curve method (LCM) significantly enhances the analytical capabilities for understanding and predicting the evolution of technologies and processes. The learning curve method, traditionally used to model the improvement in performance or reduction in cost of a technology over time, benefits from the nuanced topic modeling and semantic analysis provided by LDA2vec [36].

The LDA2vec model is an advanced machine learning framework that synergistically integrates the principles of Latent Dirichlet Allocation (LDA) and Word2Vec to enhance topic modeling and semantic analysis. LDA is a generative statistical model that classifies words into topics based on their co-occurrence patterns within a corpus, while Word2Vec generates dense vector representations for words, capturing their semantic relationships through context-based embeddings [37]. LDA2vec merges these methodologies by aligning topic distributions with word2Vec, thereby enabling the extraction of coherent and interpretable topics that reflect both probabilistic topic structure and semantic similarity [38]. This model offers improved scalability and accuracy in identifying nuanced themes within large textual datasets, making it a valuable tool for natural language processing and information retrieval in complex domains [36].

Specifically, consider a corpus of journal articles designated as $\mathrm{D}=\left\{{\mathrm{d}}_1,{\mathrm{d}}_2,\dots \dots ,{\mathrm{d}}_{\mathrm{n}}\right\}$, whose vocabulary $\left\{{\mathrm{w}}_1,{\mathrm{w}}_2,\dots \dots ,{\mathrm{w}}_{\mathrm{N}}\right\}$ comprises a total of N words. Through the application of Latent Dirichlet Allocation (LDA) on D, we derive the latent topics $\left\{{\mathrm{t}}_1,{\mathrm{t}}_2,\dots \dots ,{\mathrm{t}}_{\mathrm{T}}\right\}$ within the journal articles and the corresponding word probabilities for each topic ${\mathrm{t}}_{\mathrm{i}}$. Notably, the probability of the ${\mathrm{t}}_{\mathrm{i}}$th word within topic $\mathrm{t}$ is denoted as ${\mathsf{\theta }}_{{\mathrm{i}}_{\mathrm{j}}}$. Furthermore, Word2Vec is employed to train D, transforming each word in the vocabulary into a vectorized representation $\left\{\mathrm{v}\left({\mathrm{w}}_1\right),\mathrm{v}\left({\mathrm{w}}_2\right),\dots \dots ,\mathrm{v}\left({\mathrm{w}}_{\mathrm{N}}\right)\right\}$ of fixed length. To generate topic vectors $\mathrm{v}\left({\mathrm{t}}_{\mathrm{i}}\right)$, we sort the words in topic ${\mathrm{t}}_{\mathrm{i}}$ according to their probability and select the top h words with the highest probability. Subsequently, we normalize these topic words, readjusting their probabilities as weights, as demonstrated in Formula (1). The topic vector is then obtained by summing the products of the h word vectors and their respective weights, as outlined in Formula (2).

$${\mathrm{w}}_{{\mathrm{i}}_{\mathrm{j}}}={\displaystyle \frac{{\mathsf{\theta }}_{{\mathrm{i}}_{\mathrm{j}}}}{{\sum }_{\mathrm{n}=1}^{\mathrm{h}}{\mathsf{\theta }}_{{\mathrm{i}}_{\mathrm{n}}}}}$$

(1)

$$\mathrm{v}\left({\mathrm{t}}_{\mathrm{i}}\right)={\sum }_{\mathrm{n}=1}^{\mathrm{h}}{\mathrm{w}}_{{\mathrm{i}}_{\mathrm{n}}}\times \mathrm{v}\left({\mathrm{w}}_{{\mathrm{i}}_{\mathrm{n}}}\right)$$

(2)

Furthermore, the document vector is calculated according to Formula (3), where $\mathrm{c}$ represents the number of words in the document.

$$\mathrm{v}\left({\mathrm{d}}_{\mathrm{i}}\right)={\displaystyle \frac{{\sum }_{\mathrm{n}=1}^{\mathrm{c}}\mathrm{v}\left({\mathrm{w}}_{{\mathrm{i}}_{\mathrm{n}}}\right)}{\mathrm{c}}}$$

(3)

Consequently, each document can be represented by its distance distribution from all topics in the semantic space. We measure the similarity between a journal article and a topic using cosine similarity in the semantic vector space, selecting the topic with the smallest cosine distance as the primary topic for that article. The distance calculation is detailed in Formula (4).

$$\mathrm{distance}\left(\mathrm{v}\left({\mathrm{d}}_{\mathrm{i}}\right),\mathrm{v}\left({\mathrm{t}}_{\mathrm{i}}\right)\right)=\left|\mathrm{v}\left({\mathrm{d}}_{\mathrm{i}}\right)-\left.\mathrm{v}\left({\mathrm{t}}_{\mathrm{i}}\right)\right|\right.={\displaystyle \frac{\mathrm{v}\left({\mathrm{d}}_{\mathrm{i}}\right)-\mathrm{v}\left({\mathrm{t}}_{\mathrm{i}}\right)}{‖\mathrm{v}\left({\mathrm{d}}_{\mathrm{i}}\right)‖‖\mathrm{v}\left({\mathrm{t}}_{\mathrm{i}}\right)‖}}$$

(4)

In conclusion, this article demonstrates that through the LDA2Vec model, we can not only capture the relationship between documents and topics but also preserve contextual connections. This approach enables us to determine the themes of journal articles not solely based on high-frequency topics but instead on themes that better align with the textual content while maximizing the preservation of the document’s semantic nuances.

3.3. Data Sources

This study primarily employs bibliometric analysis techniques to conduct its research. Bibliometrics takes the characteristics of bibliometrics as the research object, and uses the measurement methods of mathematics and statistics to describe and evaluate the distribution of the literature, statistical rules, and predict the direction of the literature. It is a cross-science of set mathematics, statistics, and philology [39]. Bibliometrics, leveraging network theory and graphical tools, transforms bibliographic information into a visual and organized data format, demonstrating its efficiency in conducting a comprehensive review of specific disciplines. It utilizes intuitive network graphs to exhibit the knowledge connections among elements such as “countries, research institutions, and keywords”, and quantitatively reveals the core characteristics of these connections [40]. This paper uses the bibliometric analysis method of CiteSpace to explore the knowledge graph of low-carbon technologies in the field of energy. In this paper, literature data related to low-carbon technologies in the field of energy are retrieved from the Web of Science database, the data are imported into CiteSpace software, keyword emergence analysis and breakthrough point analysis are carried out on the literature [41], and, finally, the knowledge map related to low-carbon technologies in the field of energy is obtained.

This section will adopt bibliometric methods and primarily select the Web of Science database as the source for data retrieval, with the retrieval date set as 1 June 2023. The primary reason for choosing the Web of Science database is its status as a globally recognized platform reflecting scientific research standards. It comprises databases such as SCIE, SSCI, and citation indexes, as well as the JCR Journal Citation Reports and ESI Essential Science Indicators, all of which enjoy a strong reputation in the global scientific and educational fields.

Firstly, within the search box of the Web of Science database, keywords were entered based on the categories of low-carbon technologies outlined in the “National Key Promotion Low-Carbon Technology Catalog” issued by the Ministry of Ecology and Environment of China. Combined with the exploration of the definition of low-carbon technologies and relevant theoretical research, this article has compiled a list of keywords related to low-carbon technologies, including biomass firing power technologies, gas firing power technologies, nuclear power generation technologies, oil firing power technologies, renewable energy power generation technology, solid fuel power generation technology, energy storage and power generation technology, carbon sink technology, and carbon capture and storage power technology.

The search categories were limited to articles and reviews, with the publication date of the papers restricted from 1 January 2003 to 1 January 2023. Under these search conditions, a total of 166,440 journal articles were retrieved.

By systematically reviewing and conducting a comparative analysis of the domestic and international literature on low-carbon technologies, this article aims to clarify the current research status and explore future research trends. Data results were obtained from the Web of Science search interface, focusing on refined subject headings such as “category, publication year, and highly cited papers in the field”. Statistical analysis was conducted using Origin 8.1 and Microsoft Excel 2016 to analyze factors such as literature type, research hotspots, and contributions from different countries.

4. Analysis of the Knowledge Mapping of Low-Carbon Technology Research in the Energy Sector

4.1. Temporal and Spatial Distribution of Research Outcomes

The co-occurrence analysis of publishing countries using CiteSpace is presented in Figure 1. The co-occurrence network of country associations reveals a certain pattern of international collaboration, with distinct connections among nations. In the presented figure, each node serves as a representative of a country, where the node’s size corresponds to the magnitude of publications originating from that particular country. The lines interconnecting these nodes depict the collaborative ties between different nations, while the thickness of these lines serves as an indicator of the number of collaborative publications between them.

The figure depicts 339 connections among 138 countries, with a network density of 0.0359, indicating a relatively concentrated distribution of research in the field of low-carbon technology across various countries and regions worldwide. The direct connections between nodes reveal numerous collaborations among nations and institutions. In terms of publication frequency, the United States (USA) has the highest contribution, with 2128 articles, signifying its active engagement in low-carbon technology research in the energy sector. Other prominent countries include China, the United Kingdom, Germany, Italy, Spain, Australia, Canada, Japan, and India.

From the perspective of the centrality strength of each node (see

Table 1. Ranking of countries (regions) by centrality intensity.

Time	Countries (Regions)	Centrality	Frequency
2003	FRANCE	0.43	336
2003	AUSTRALIA	0.33	501
2004	CZECH REPUBLIC	0.31	50
2003	POLAND	0.29	223
2015	TANZANIA	0.27	5
2003	SOUTH AFRICA	0.25	101
2003	USA	0.24	2128
2003	PEOPLES R CHINA	0.24	1925
2003	PORTUGAL	0.21	128
2003	NEW ZEALAND	0.21	36
2004	GREECE	0.19	136
2005	CROATIA	0.19	46
2010	HUNGARY	0.19	35
2007	MONACO	0.18	2
2010	TUNISIA	0.18	12
2012	ALGERIA	0.18	19
2003	UK	0.17	1302
2020	SIERRA LEONE	0.17	2
2013	NIGERIA	0.16	35
2003	AUSTRIA	0.15	159
2006	IRAN	0.15	185
2006	JORDAN	0.15	14
2016	BRUNEI	0.15	13
2003	SWEDEN	0.14	271
2014	QATAR	0.14	39
2004	ESTONIA	0.1	48
2012	COLOMBIA	0.1	34
2003	RUSSIA	0.09	140

), France has a centrality strength of 0.43, which is the country with the highest centrality strength, indicating that France has a relatively obvious intermediary role in the field of low-carbon technology research, with many studies conducted through this node, and its influence on the research network structure is strong. France plays an important role in promoting the development of global low-carbon technology and promoting international cooperation. China’s centrality strength is 0.23, which shows it also has a certain influence and control in this field, but ranks behind countries such as Australia, the Czech Republic, Poland, Tanzania, and South Africa, with the same centrality strength as the United States, indicating that China does not yet occupy a core position in the research of low-carbon technology. These major countries and regions with high centrality strength have made early strategic layout and long-term accumulation in various fields of low-carbon technology, while China has started relatively late in the research and application of low-carbon technology, which makes China lack advantages in technology research and development, application, and international cooperation. These countries also have rich international cooperation experience in the field of low-carbon technology, and are able to establish close cooperative relationships with other countries to jointly promote the development and application of technology. Although China and the United States have different research frequencies, their centrality strengths are the same, indicating that both countries recognize the importance of low-carbon technology research. In specific low-carbon technology fields, such as clean energy, photovoltaics, and green transportation, both countries have conducted in-depth research and cooperation. However, the low-carbon policy of the United States is relatively flexible and focuses more on the role of market mechanisms; while China’s low-carbon policy focuses more on government guidance and support.

This enlightenment tells us that to achieve the strategic goal of “dual carbon”, we need to vigorously develop green and low-carbon technologies. The research interaction between Chinese research institutions in the field of low-carbon technology should be further strengthened and improved, especially in strengthening research cooperations with developed countries in the field of low-carbon technology. At the same time, domestic research institutions in China should also further enhance their academic influence and gradually establish their authority, in order to strive for more valuable research results in the field of low-carbon technology research [42].

In recent years, China has witnessed remarkable advancements in green and low-carbon technologies, where pivotal clean energy technologies have exhibited profound superiority. Nevertheless, a substantial disparity persists between China and developed economies, such as those in Europe and the United States. According to Bloomberg NEF’s data, China’s investment in low-carbon energy transition accounted for a significant 35.2% of the global total in 2021, thereby occupying a leading position worldwide. Furthermore, the National Energy Administration’s statistics reveal that China’s newly installed capacity of wind power and photovoltaics reached 101 million kilowatts in 2021, maintaining its top rank globally for consecutive years.

China’s strengths are primarily concentrated at the industrial level, yet there is still a noticeable gap in the overall level of green and low-carbon technologies compared to advanced countries. According to the latest technology forecast report by the China Science and Technology Development Strategy Research Institute, an assessment of the development status of green and low-carbon technologies across multiple key areas such as energy storage, hydrogen energy, renewable energy, nuclear energy, and the energy internet revealed that 19.7% of China’s green and low-carbon technologies currently reach international leading levels, while 54.4% are on par with the international average. However, 25.9% of the technologies still lag behind the international average, indicating a significant gap of approximately 7.3 years between China’s average level of green and low-carbon technologies and that of internationally leading countries [43].

4.2. Identification of Hotspots and Changes in Trends

Conducting statistical analysis and co-occurrence analysis on the frequency of keywords in papers is an effective strategy to identify the current research hotspots in the field [44]. Currently, low-carbon technologies play a crucial role in promoting the global energy transition. Understanding the research hotspots and development frontiers of low-carbon technologies is of great guiding significance for carrying out related research [45].

Figure 2 shows the 15 global bursts of low-carbon technology research in the energy sector between 2003 and 2022. Through burst detection analysis of keywords, one can identify the research foci and hotspots within specific time periods, as well as the distribution and evolution of research themes. The red lines represent the specific duration during which these keywords became academic research hotspots, light blue indicates periods before the emergence of these nodes, and dark blue indicates the periods when the nodes began to appear. The initial research focused on climate change, particularly emissions of greenhouse gases such as carbon dioxide. With the advancement of technological innovation, from 2006 to 2014, basic research has become increasingly mature, and relevant topics such as carbon dioxide combustion, capture and storage, and nuclear power have been deeply explored. By exploring how CO₂ combustion can reduce carbon emissions by improving the combustion process, carbon capture and storage technology shows the potential of this technology to reduce the concentration of carbon dioxide in the atmosphere. It is noteworthy that the research intensity of carbon sequestration and carbon capture and storage reached 15.98 and 51.31, respectively. An analysis of the literature related to these 15 outbreaks found that foreign research papers continue to occupy a prominent position, and, in recent years, more and more researchers have been exploring related issues, focusing on flue gas, CO₂ capture, and transport.

In Figure 3, the keywords extracted from the Web of Science search results are clustered into nine categories. Cluster #0 pertains to energy storage research, Cluster #1 to life cycle assessment, Cluster #2 to CO₂ emissions, Cluster #3 to emissions, Cluster #4 to performance, Cluster #5 to carbon capture and storage, Cluster #6 to techno-economic analysis, Cluster #7 to carbon dioxide, Cluster #8 to CO₂ capture, and Cluster #9 to smart grid. Among them, energy storage research focuses on how to improve energy storage efficiency and cost-effectiveness, life cycle assessment explores the environmental impact of fuel and raw material alternative technologies, and CO₂ emissions research focuses on reducing CO₂ emissions from carbon capture, use and storage technologies. In addition, performance focuses on improving the efficiency and effectiveness of various low-carbon technologies, while carbon capture and storage focuses on developing and optimizing carbon capture technologies and their secure storage. Finally, smart grid explores how advanced grid management techniques can be used to optimize energy distribution and improve the flexibility and reliability of the power system. The comprehensive analysis of these research clusters shows the multi-dimensional structure of low-carbon technology research and builds a comprehensive scientific research framework. These nine clusters constitute the research structure of the Web of Science database on low-carbon technologies in the energy sector. When compared with the “National Key Low-Carbon Technology Promotion Catalog” published by the National Development and Reform Commission, it covers five categories: non-fossil energy technologies (12 items), fuel and raw material substitution technologies (11 items), non-CO₂ emission reduction technologies such as process engineering (5 items), carbon capture, utilization, and storage technologies (2 items), and carbon sequestration technologies (3 items).

Through a comparative analysis of research hotspots on low-carbon technologies in the energy sector from the Web of Science database and the “National Catalog of Low-Carbon Technologies for Key Promotion”, it is found that the research focus on low-carbon technologies in the energy sector at home and abroad is basically consistent. Future development will mainly focus on the research and development of more efficient, economical, and reliable renewable energy technologies, including hydropower, wind energy, solar energy, biomass energy, geothermal energy, and ocean energy, as well as advanced power generation and integrated application technologies for these energy sources.

4.3. Analysis of Keyword Frequency

Through further statistical analysis of the number of published articles, we have identified nine technology-related keywords, as shown in Figure 4. This section will delve deeper into the analysis of these key technologies.

Figure 4 shows that “renewable energy” and “nuclear energy” had significant growth before 2022. By analyzing the keywords in the literature within the research field, we track and explore the hotspots and key technologies in low-carbon technology research.

We have constructed a keyword cloud for the relevant literature in the field of low-carbon technology. In this cloud, the larger the keyword appears, the higher its frequency of occurrence, as shown in Figure 5. Through the analysis of the top 30 keywords, we can derive the main research hotspots and directions of low-carbon technology. Regarding the research foci, “energy”, “power”, “CO₂”, and “renewable” are the overall cores of low-carbon technology research. In the energy sector, promoting the transformation from fossil fuels to new energy sources, including hydrogen, storage, biomass, solar, water, wind, nuclear, and other key technologies, are currently the main research hotspots in the field of low-carbon technology. Among them, the frequencies of storage and hydrogen energy appeared 15,460 times and 11,852 times respectively, making them core keywords in the field and also important components of China’s low-carbon technology system.

In the industrial sector, key technologies such as energy cycling (cycle), electrification (electricity), carbon dioxide capture, and carbon capture and storage (CCS) are the research hotspots. In the mid-to-late stages of the 13th Five-Year Plan, China’s industrial energy use and carbon emissions entered a “new growth cycle”, with the energy consumption and carbon emissions of traditional high-energy and high-emission industries rising instead of falling. Promoting the reinvention of low-carbon processes and negative emission retrofitting has become the main direction of China’s industrial technological transformation. To promote the early achievement of carbon peak in the industrial sector, relevant departments have issued the “Implementation Plan for Carbon Peak in the Industrial Sector”, which emphasizes strengthening technologies related to deep decarbonization and enhancing the carbon neutrality capabilities in the industrial sector.

4.4. LDA2vec Keyword Cluster Analysis

LDA2vec is a method that combines Latent Dirichlet Allocation (LDA) and word2vec to perform topic modeling and keyword clustering. By applying LDA2vec to the keywords in the low-carbon technology field, we can identify clusters of related keywords and further understand the structure and relationships within the field. This analysis can help researchers gain a deeper understanding of the research topics and subtopics within low-carbon technology, as well as identify potential new research directions.

By building an LDA2Vec topic model, this study conducted a thematic analysis of the acquired keywords related to low-carbon technology, as shown in

Table 2. Keyword frequency.

Sort	Frequency	Keyword	Sort	Frequency	Keyword
1	38,906	energy	16	5691	fired
2	37,412	power	17	5616	CCS
3	17,404	CO2	18	5375	water
4	15,460	storage	19	5191	wind
5	14,641	gas	20	5125	nuclear
6	13,732	carbon	21	4819	cycle
7	11,927	renewable	22	3823	fossil
8	11,852	hydrogen	23	3791	grid
9	11,165	fuel	24	3548	natural
10	10,072	capture	25	3424	fuels
11	10,061	electricity	26	3416	operation
12	9065	coal	27	3302	capacity
13	6223	combustion	28	2924	air
14	6193	biomass	29	2815	solid
15	5905	solar	30	2754	combined

. We selected the optimal number of topics as three, Keyword clustering reflects the relationship between different topics. Through systematic analysis, the mutual relationship and internal logic between different topics can be deeply revealed, according to which Figure 6 is drawn. As shown in Figure 6, topic 0 includes terms such as “gas”, “heating”, and “hydrogen”, indicating that topic 0 is related to hydrogen production technology. Similarly, we can infer that topic 1 is related to solar power generation technology, while topic 2 is associated with energy storage technology. Currently, countries worldwide are gradually increasing their research and development investments in the field of clean energy, focusing on the development of key technologies such as energy storage, hydrogen energy, and solar energy.

Hydrogen energy, as a clean energy source that is abundant, environmentally friendly, and efficient, is a crucial tool for helping China achieve its carbon peak and carbon neutrality goals and promoting energy structure adjustment and industrial transformation and upgrading. Therefore, key technologies for green hydrogen production have become a new research hotspot. Domestic scholars have conducted extensive research on hydrogen production processes such as electrolytic water, biomass, and nuclear energy, and green hydrogen production technology is developing towards diversification. According to data from the China Hydrogen Energy Alliance, China’s hydrogen production exceeded 33 million tons in 2019, an increase of 60% compared to 2018. In 2021, China’s hydrogen production exceeded 34 million tons, ranking first globally. In regards to the prospective advancements in hydrogen energy, China has promulgated the “Medium- and Long-Term Plan for the Advancement of the Hydrogen Energy Industry (2021–2035)”, emphasizing that hydrogen energy is a pivotal component of the forthcoming national energy landscape.

Solar photovoltaic power generation is a clean energy technology that converts solar energy directly into electrical energy. Yao Yubi and others summarized and analyzed the spatial and temporal distribution characteristics of solar energy resources in China and the potential assessment of solar energy resources in various regions from the perspectives of richness, timeliness, and stability [17]. The conclusion proved that China has abundant solar energy resources, with vast reserves, wide coverage, and tremendous development prospects, making it one of the most competitive new energy sources. In recent years, the global focus on environmental protection and sustainable development has been increasing, which has prompted the more widespread application and promotion of solar photovoltaic power generation technology. China has also proposed, in the “14th Five-Year Plan for Modern Energy System”, to promote the application of renewable energy represented by solar energy.

Energy storage is a pivotal technology in the energy revolution, serving as a crucial support for achieving carbon peak and carbon neutrality goals. Therefore, energy storage technology has received significant attention from the academic community [46]. Presently, China’s energy storage industry is transitioning from the early stages of commercialization to scaled development. To support the development of energy storage, the Chinese government has issued a series of policies covering electricity prices, planning, industry, fiscal and taxation, market rules, and the allocation of new energy storage [47]. By the end of 2018, the cumulative operational capacity of global energy storage projects had reached 180.9 gigawatts (GW), with pumped hydro energy storage accounting for a significant majority of 94%. Domestically, the cumulative operational capacity of energy storage projects stood at 31.3 GW, comprising 17.3% of the global market share [48].

5. Evolution of Low-Carbon Technology Costs Based on the Learning Curve

The learning curve can be used as a tool to measure the relationship between low-carbon technology progress and cost reduction, its research can reveal the cost reduction patterns of specific low-carbon technologies over time and experience, and this analysis can help identify which technologies have the potential to become future research hotspots. Amidst the escalating severity of environmental challenges, the vigorous pursuit of renewable energy sources, notably wind power, has gained worldwide recognition. China, particularly, has emerged as a frontrunner, establishing itself as the world’s largest and fastest-expanding market for wind power generation.

Through statistical analysis of key terms in the field of low-carbon technology, this chapter selects hydrogen production technology, solar power generation technology, and energy storage technology as crucial technologies based on such analysis.

China’s low-carbon technology is still in its infancy, transitioning from “small-scale, intermittent, and cottage-style production” to “integrated, continuous, and large-scale production lines”. In the early stages of development, the cost of low-carbon technology is often high. To promote technological advancement, reduce unit costs, and accelerate the maturation of the research, production, and consumption markets for low-carbon technologies, the learning curve theory is commonly used internationally to estimate the technological learning rate and predict the evolution of technology costs accordingly. Accurately calculating the learning rate of low-carbon technology is crucial for understanding the patterns of cost reduction in technology investment, making research and development investment decisions for the low-carbon technology manufacturing industry, planning the market launch strategies for new products, and formulating government subsidy policies [35].

With the gradual maturity and improvement of production processes, as well as the accumulation of production management experience, there is still significant room for reducing the technical costs. The downward trend of the cost of low-carbon technologies can be described by a learning curve. The process of cost reduction for technologies follows this learning curve. Assuming that the cost of continuous and large-scale, low-carbon technologies will decrease along the learning curve as the cumulative market capacity (annual output, installed capacity) increases, its evolution trend can be described by a limiting efficiency model, as shown in the following formula [49]:

$$\mathrm{c}=\mathrm{c}\prime +∆\mathrm{c}\times {\left({\displaystyle \raisebox{1ex}{$\mathrm{N}$}\!\left/ \!\raisebox{-1ex}{${\mathrm{N}}_0$}\right.}\right)}^{-\mathsf{\beta }}$$

(5)

$\mathrm{c}$: The long-term marginal cost of low-carbon technologies; ${\mathrm{c}}^{\prime }$: the lowest achievable limiting cost of low-carbon technologies, $\mathsf{\Delta }\mathrm{c}$: the difference between the initial cost and the limiting cost; $\mathrm{N}$: the cumulative scale of low-carbon technology usage; ${\mathrm{N}}_0$: the initial scale of low-carbon technologies, $\mathsf{\beta }$: the learning coefficient of costs.

Therefore, based on Formula (5), this chapter will estimate the costs of China’s hydrogen production technology, energy storage technology, and photovoltaic power generation technology. Using data such as their initial and ideal scales, costs, cumulative annual installed capacity, and cumulative annual production, simulations will be conducted on the costs and scales of important low-carbon technologies, aiming to predict the future development of low-carbon technologies.

5.1. Low-Carbon Technical Parameter Setting

According to Formula (5), only by determining the initial and ultimate costs of low-carbon technology can we obtain $\mathsf{\Delta }\mathrm{c}$.

Initial cost: The baseline cost for each technology is set as the unit cost in the reference year. Specifically, the initial costs are determined by the average price of hydrogen production, photovoltaic power generation, and electrochemical energy storage in 2010, respectively: 16 CNY/ton, 1 CNY/KWH, and 3.7 CNY/KWH.

Limit cost: According to the learning curve theory, the economic cost of various low-carbon technologies will decline slowly after it is reduced to a certain extent, and will not fall indefinitely. In essence, the cost function represented in Formula (5) possesses a distinct lower bound, indicating that as the scale of operation increases, the cost of low-carbon technology tends to move towards this minimal threshold value.

From the perspective of technical cost optimization selection, for hydrogen production technology, according to relevant studies, the cost of hydrogen technology in 2030 can be as low as 11 CNY/ton. Furthermore, the collaborative report titled “China 2050 Photovoltaic Development Outlook (2019)” by the Energy Institute of the National Development and Reform Commission, alongside Longi Shares and Shaanxi Coal Group, underscores that the cost of photovoltaic power generation in China currently stands at CNY 0.13 per kilowatt-hour (KWH), marking a significant milestone in renewable energy technology. Consequently, in this paper, we shall adopt 0.13 CNY/KWH as the baseline cost for photovoltaic power generation technology, representing its lowest economic threshold. In addition, according to Formula (5), it is also necessary to determine the initial scale ${\mathrm{N}}_0$ of low-carbon technology and the cumulative scale $\mathrm{N}$ of low-carbon technology use.

Initial Scale ${\mathrm{N}}_0$: This article will use 2010 as the base year to define the initial scale of each technology. The 2010 hydrogen volume is set at its initial scale, which is 12.54 million tons. The amount of carbon dioxide captured in 2010 is taken as the initial scale of carbon capture technology, that is, 20 million tons. In addition, in the context of this analysis, the initial scale is defined as the installed capacity of wind power and photovoltaic power generation in the year 2010. This baseline figure serves as the starting point for examining subsequent trends and developments: 31.07 million kilowatts and 860 million kilowatts, respectively.

Cumulative scale $\mathrm{N}$: This paper will take each low-carbon technology development plan as the cumulative scale to determine the cost development trend of each low-carbon technology. Amidst China’s aspirational goal of achieving carbon neutrality by 2060, the esteemed China Hydrogen Energy Alliance forecasts a substantial increase in the nation’s annual hydrogen demand. Specifically, by 2030, this demand is predicted to soar to 37.15 million tons, accounting for roughly 5% of the overall final energy consumption. Additionally, projections indicate that the envisioned carbon capture capacity will reach 60 million tons in 2025, while the cumulative installed capacity of photovoltaic power generation is anticipated to attain 73 million kilowatts during the same timeframe. Consequently, we adopt 37.15 million tons, 73 million kilowatts, and the respective cumulative scale data as benchmarks for our analysis.

5.2. Learning Curve Analysis

Based on the above data, the cost of hydrogen technology in 2030 can be as low as 11 CNY/ton, the corresponding data of hydrogen production technology can be brought into Equation (5) to obtain $11=9.9+(16-11)\times {(3715/1254)}^{-\mathsf{\beta }}$, and the learning coefficient of hydrogen production technology can be obtained as 1.58.

Similarly, in order to achieve the “China 2050 Photovoltaic Development Outlook (2019)” plan, forecasting the photovoltaic power generation costs for 2030, we estimate a cost of CNY 0.13 per kilowatt-hour (KWH) [50], reflecting significant progress in technological advancements and economies of scale. When the corresponding data of photovoltaic technology are brought into Equation (5), which can obtain $0.13=1+(1-0.13)\times ({73000/86)}^{-\mathsf{\beta }}$, the learning coefficient of solar power technology is 0.63. According to relevant reports, the national energy storage cost will be reduced to 9 million CNY/GW in 2030. The corresponding data of energy storage technology will be brought into Equation (5), $900=3700+(3700-900)\times ({120/18.41)}^{-\mathsf{\beta }}$ can be obtained, and the learning coefficient of energy storage technology is 1.85.

Based on the above minimum cost and learning rate, the annual cost prediction curve of hydrogen production in China can be expressed as Equation (6):

$${\mathrm{C}}_1=9.9+5\times {(\mathrm{N}/1254)}^{-1.58}$$

(6)

The anticipated cost trajectory for photovoltaic (PV) power generation in China is succinctly encapsulated in Equation (7), providing a quantitative framework for assessing future cost developments:

$${\mathrm{C}}_4=0.13+0.87\times {(\mathrm{N}/86)}^{-0.63}$$

(7)

The cost prediction curve of China’s energy storage technology can be expressed as Equation (8):

$${\mathrm{C}}_4=900+2800\times {(\mathrm{N}/18.41)}^{-1.85}$$

(8)

5.3. Production Forecast

China’s annual hydrogen production has experienced significant growth since 2000, starting at 7 million tons and nearly doubling by 2010 with an average annual growth rate exceeding 6%, reaching approximately 12.54 million tons in 2010. According to data from the China National Coal Association, China’s hydrogen production in 2012 was 16 million tons, indicating a steady growth trend. Since 2018, China’s annual hydrogen production has exceeded 20 million tons, reaching 25 million tons by 2020 and surpassing 33 million tons in 2021. Additionally, based on relevant data, China’s hydrogen production in 2017 was approximately 19.15 million tons and exceeded 20 million tons by 2019. According to forecasts by the authoritative China Hydrogen Alliance, under the carbon neutrality target, China’s hydrogen demand is expected to reach 37.15 million tons by 2030, accounting for about 5% of the total end-use energy consumption.

Furthermore, based on previous research data [50,51,52], China’s cumulative installed capacity of photovoltaic (PV) power generation was 860,000 kW in 2010, 6.5 million kW in 2013, 28.05 million kW in 2015, 77.42 million kW in 2017, and 306.56 million kW in 2022. Similarly, China’s cumulative energy storage capacity was 18.41 GW in 2010, 28.9 GW in 2017, 31.3 GW in 2018, and 34.6 GW in 2019.

Based on the aforementioned data, the production forecast curves for China’s annual hydrogen production, cumulative PV power generation capacity, and cumulative energy storage capacity from 2010 to 2030 can be derived. These forecasts utilize exponential curves to fit the data and future plans. As illustrated in Figure 7, the forecast curves indicate that all three low-carbon technologies show a rapid upward trend. First, this trend aligns with the national low-carbon development goals and supports the significant development of green and low-carbon technologies to achieve carbon targets. Second, these technologies receive substantial national support and represent sustainable green and low-carbon advancements. The results in Figure 7 will also serve as an important data source for forecasting cost curves.

5.4. Cost Prediction of Low-Carbon Technology

By substituting the projected annual hydrogen production, cumulative photovoltaic (PV) installed capacity, and cumulative energy storage installed capacity obtained in Section 5.3 into Equations (2), (3) and (4), respectively, we can derive the future cost estimates for hydrogen production technology, photovoltaic technology, and energy storage technology. The results are shown in

Table 3. Cost prediction results of various low-carbon technologies.

Given Year	Hydrogen Production Cost (CNY/ton)	Photovoltaic Power Generation Cost (CNY/KWH)	Energy Storage Technology Cost (10,000 CNY/GW)
2010	14.16	0.52	3918.18
2011	13.80	0.45	3468.531
2012	13.48	0.40	3085.87
2013	13.18	0.36	2760.219
2014	12.9	0.33	2483.083
2015	12.65	0.30	2247.235
2016	12.42	0.28	2046.523
2017	12.21	0.27	1875.714
2018	12.02	0.25	1730.352
2019	11.84	0.24	1606.646
2020	11.68	0.23	1501.369
2021	11.53	0.23	1411.777
2022	11.39	0.22	1335.532
2023	11.27	0.22	1270.647
2024	11.15	0.21	1215.428
2025	11.05	0.21	1168.435
2026	10.95	0.21	1128.444
2027	10.86	0.21	1094.41
2028	10.78	0.21	1065.447
2029	10.71	0.20	1040.798
2030	10.64	0.20	1019.82

, and the cost development trend is shown in Figure 8. Upon a thorough analysis of Table 3 and Figure 8, it is evident that the anticipated expenditure for hydrogen production technology, photovoltaic power generation technology, and energy storage technology in the year 2030 stands at CNY 10.64 per ton, CNY 0.20 per kilowatt-hour (KWH), and CNY 10.1982 million per gigawatt (GW), respectively. These data provide a comprehensive overview of the estimated costs associated with these key renewable energy technologies in the near future.

Using the installed capacity and historical installed cost data, the learning rate of hydrogen production technology, solar power generation technology and energy storage technology is calculated. Drawing upon the data from the National Energy Administration’s annual power bulletin and relevant online resources, we have conducted an in-depth analysis utilizing the learning curve methodology. Our calculations reveal that the learning rate for hydrogen production technology stands at 33.5%, while solar power generation technology exhibits a learning rate of 64.8%, and energy storage technology registers a learning rate of 27.7%. It is noteworthy that the learning rate for solar power generation technology lags behind that of hydrogen production and energy storage technologies, primarily due to the elevated costs associated with photovoltaic raw materials and the relatively low energy conversion efficiency, as well as the significant investment and maintenance expenses.

From the cost development trend. The cost of hydrogen production technology still has some room for decline, and the decline rate has not slowed down significantly. For carbon capture technologies, the decline is predicted to level off around 2026. Upon a thorough analysis of the cost progression of energy storage technology and photovoltaic power generation technology, it becomes apparent that their cost evolution closely resembles the trajectory of carbon capture technology, displaying a distinctive pattern of an “initial decline, culminating in stabilization”. Notably, despite this stabilization, a persistent albeit gradual downward tendency remains evident, implying significant potential for further technological advancements in both energy storage and photovoltaic power generation domains.

Furthermore, given the disparity in the measurement units of various technology costs, we have conducted a comparative analysis of their cost reduction rates, as depicted in Figure 9. In terms of the overall trend, the combined reduction rate of these three technologies exhibits a declining pattern [53]. Nevertheless, the cost reduction rate of photovoltaic power generation technology exhibits minor fluctuations, potentially influenced by market dynamics and policy interventions. Focusing specifically on cost reduction, the rate pertaining to hydrogen production technology remains relatively stable, hovering around 1% to 3%. This is because hydrogen production technology started early and is relatively mature. However, to achieve large-scale, efficient, and sustainable hydrogen production, there are still some challenges [54], such as energy consumption, cost, storage and transportation, and further cost reduction requires more technological breakthroughs and innovations. Conversely, the costs associated with photovoltaic power generation and energy storage technologies are experiencing a more rapid decline. Thanks to advancements in solar cell efficiency and a reduction in manufacturing expenditures, photovoltaic power generation technology has progressively gained competitiveness. Regarding energy storage technology, breakthroughs in battery innovation have bolstered the efficiency and reliability of energy storage systems, while the expansion of production scales and process optimizations have further contributed to cost reductions.

In general, compared with traditional fossil energy power generation technologies, new energy technologies have obvious advantages in future development prospects. With technological advancements and the achievement of scale economies, the costs associated with new energy power generation technologies are progressively diminishing, thus enhancing their market competitiveness. It is anticipated that, in the foreseeable future, new energy technologies will gradually supersede traditional fossil energy technologies globally, emerging as the primary mode of energy supply. However, in the interim, these technologies confront several challenges, including the refinement of energy storage solutions and the establishment of robust transmission networks. To guarantee the unwavering security and stability of energy supply, it is imperative to address these challenges during the advancement of new energy generation technologies.

6. Conclusions and Policy Suggestions

Amidst the steadfast progression of global aspirations towards low-carbon development, the proliferation of innovative low-carbon technologies holds paramount significance in attaining the “dual carbon” objectives. This underscores the imperative need for technological advancements that align with the ever-evolving demands of environmental sustainability. This paper searched for nine keywords related to mainstream technologies and used 166,440 journal articles retrieved from the Web of Science as research data. In the initial phase of our research, we embarked on a macroscopic examination of the evolving trajectories of low-carbon technologies, leveraging the CiteSpace software for in-depth analysis. To conclude, upon examining the frequency of pertinent keywords, four pivotal technologies were identified. Utilizing the learning theory, a technology cost learning curve was constructed to comprehensively assess the evolving cost trajectories of these critical low-carbon technologies. We found the following:

(1)

Based on a thorough analysis of literature sources from the Web of Science database, this study emphasizes the critical role of low-carbon technologies in the energy sector and their potential influence on policy-making. The prominence of research papers in this area highlights the academic community’s active involvement in exploring low-carbon technologies. China’s centrality score of 0.23 indicates significant influence and control in this field, yet it lags behind countries like Australia, the Czech Republic, Poland, Tanzania, and South Africa, and has the same centrality strength as the United States. This suggests that China has not yet assumed a central position in low-carbon technology research, presenting an opportunity to enhance international collaboration, thereby fostering knowledge exchange and innovation globally. This research also provides valuable insights into the maturity and cost evolution of key low-carbon technologies, which are crucial for policymakers. The learning curve analysis predicts a significant decrease in the costs of these technologies in the coming years, although the rate of reduction varies. For instance, hydrogen production technology, despite its current high cost, shows considerable potential for further reduction, while photovoltaic power generation and energy storage technologies are on a promising trajectory to become cost-competitive alternatives to traditional fossil fuels in the near future.

(2)

Through a comprehensive statistical analysis of published articles, we have studied multiple technical keywords, primarily related to cutting-edge fields such as carbon dioxide recovery, energy storage, and renewable energy technology, including but not limited to carbon sequestration technology and energy storage technology.

To further understand the development trajectory of these low-carbon technologies, we adopted the learning curve method. Our analysis indicates that global low-carbon technology has entered an important period of development. Therefore, further research is necessary to overcome the remaining challenges, optimize technical performance, and ultimately accelerate the transition to a low-carbon economy.

(3)

In our comprehensive analysis of the cost evolution of key low-carbon technologies, we carefully drew a learning curve to predict costs until 2030. Our research findings indicate that, although the cost of carbon capture technology is decreasing at the fastest rate, the cost reduction in hydrogen production technology is still significant and has not slowed down significantly, indicating sufficient space for further optimization.

Our research findings emphasize the importance of strengthening research and application of hydrogen production technology and photovoltaic power generation technology. By reducing technology costs, and by improving its accessibility and applicability, we can accelerate the transition to a low-carbon energy model, which is in line with the desire for global sustainable development. This strategic emphasis on key low-carbon technologies is not only consistent with our theme modeling based on LDA2Vec, but also emphasizes the crucial role of these technologies in shaping the future of the energy industry.

(4)

To effectively address the high costs and technical barriers associated with emerging low-carbon technologies, such as hydrogen production and energy storage, it is crucial to increase funding for research and development. Such investment will accelerate the advancement of these technologies, reduce costs, and facilitate their widespread adoption. Enhancing international collaboration is equally vital, given China’s current relatively modest intermediary centrality within the global low-carbon technology network. By engaging in joint research projects and fostering knowledge sharing, China can expedite the development of core technologies and strengthen its global competitiveness in the low-carbon sector. Additionally, the implementation of financial incentives, including tax breaks, subsidies, and grants, is essential to encourage the adoption and further development of low-carbon technologies. Promoting the integration of low-carbon technologies with renewable energy sources, such as solar and wind power, should be a priority to reduce reliance on fossil fuels, decrease carbon emissions, and support the transition to a sustainable energy system. Furthermore, fostering market-based mechanisms like carbon trading is critical for creating a favorable environment for low-carbon technology development. Establishing a robust carbon trading system will provide economic incentives for emission reductions and encourage the development and adoption of low-carbon technologies. Policymakers should adopt a long-term perspective by setting clear and ambitious carbon reduction targets and ensuring that policy frameworks are in place to support the sustained development of these technologies. By implementing these policy suggestions, China can enhance its leadership in the development of low-carbon technologies, make significant progress toward achieving its carbon reduction goals, and contribute to global efforts in combating climate change.

Although this paper comprehensively assesses the cost trajectories of key low-carbon technologies, some limitations remain.

(1)

The hybrid model chosen in this paper takes advantage of the advantages of LDA and Word2Vec, but inevitably integrates their complexity. When determining the optimal number of topics and interpreting the relationship, the model parameters (such as the number of topics and vector dimensions) may be subjective and affect the results. At the same time, this method ignores the influence of market, regulation, and innovative technologies in predicting technology cost according to the assumptions of this paper.

(2)

Low-carbon technology is multidisciplinary in nature, involving engineering, economics, policy research, etc. However, the bibliometrics and LDA2Vec methods selected in this paper cannot achieve multidisciplinary integration in the analysis of different data types, and the analysis problems are limited, which may lead to the omission of key interdisciplinary perspectives.

(3)

This paper focuses on China’s progress in low-carbon technologies and the insights provided for the Chinese market may not be applicable to other regions with different economic, environmental, and regulatory contexts.

According to the above limitations, future research can be carried out on the following aspects.

(1)

Careful consideration of the complexity of LDA and Word2Vec methods, adjusting the model parameters to achieve more objective results, and other optimization techniques such as cross-validation can be used to select the best model parameters (such as the number of topics and vector dimensions). Through these methods, the performance of the model under different parameter settings is evaluated to find the optimal combination of parameters. At the same time, in order to consider the influence of factors such as market, regulation, and technological innovation, external data sources and characteristics can be introduced, combined with the knowledge of domain experts to expand the model, so that it more fully reflects the actual situation.

(2)

In order to overcome the limitations of a single approach in multidisciplinary integration, analytical tools and techniques from other disciplines can be combined. For example, expert systems, case study analyses, or qualitative research methods may be introduced to supplement the results of quantitative analyses. In addition, interdisciplinary teams can be organized to conduct research to ensure that perspectives from different disciplinary fields are taken into account. In this way, the multi-dimensional nature of low-carbon technologies can be more fully captured and key interdisciplinary perspectives are not overlooked.

(3)

In order to improve the universality of the research results, multi-regional or international comparative studies can be conducted. By collecting data on low-carbon technologies from different countries and regions and applying the same analytical methods, technological progress in different economic, environmental, and regulatory contexts can be explored.