A Disruptive Technology Identification Method for New Power Systems Based on Patent Evolution Analysis

Abstract

Disruptive technologies have been employed in various fields with a strategic priority in several countries as the core driving force of the fourth industrial revolution, significantly impacting the development of new power systems. It is a kernel to effectively identify the future potential of disruptive technologies. To overcome the subjectivity and limitations of existing disruptive technology identification methods, we propose a disruptive technology identification method based on patent evolution analysis. Firstly, the evolution matrix of the patent data is calculated. Afterward, we dig into the characteristics of disruptive technologies to build a more targeted identification index system. Finally, the fields of electric power communication and energy generation are selected as typical cases to build the patent data sets. The future development of the identified technologies, including the identified quantum technologies and controlled fusion, is analyzed. The results demonstrate that the proposed model can identify the key technologies of new power systems accurately and contribute to completing the industrial upgrading and transformation more rapidly.

1. Introduction

Disruptive technologies are innovations from existing technologies. At the national level, disruptive technologies have become a powerful driving force for economic and social development and a core element of scientific and technological innovation and industrial upgrading [1]. At the industry level, disruptive technologies have distinctive industry characteristics, which can be determined by analyzing patent data in related fields. As an emerging technology [2], disruptive technologies follow a different technological trajectory compared to existing technologies and have high auxiliary performance, which can create niches and marginal markets and eventually dominate unexpected new application areas, often bringing great challenges to existing enterprises. In today’s technology development environment and situation, how to establish a reasonable and efficient method to effectively identify disruptive technologies in the industrial field based on key factors, such as the trend, direction, and impact of technology development [3], is an important issue.

There are many methods used for disruptive technology identification; for example, Vojak proposed a method for identifying potential disruptive technologies using technology roadmaps based on historical research cases [4]. Delmer Nagy et al. first identified the innovation points and characteristics of the target technology by organizing the opinions of technology experts and market managers. They then assessed the market value and compared it with current market technology [5]. J Sun et al. used the theory of the solution of inventive problems (TSIP) to construct a model to predict technology prospects in terms of technological maturity and systematic concerns [6]. H Carlsen et al. learned from an interdisciplinary group of experts on the organized workshops to evaluate disruptive technologies through scenario analysis [7]. Guangzu Bai et al. used research papers cited by core patents, as well as papers and their citations, to analyze the impact of disruptive technology [8]. However, there still exists subjectivity and limitations in specialized fields according to the above approaches. To alleviate such issues, this paper addresses the problem that the existing research does not pay enough attention to the identified disruptive technologies in terms of future research development and the trend of change. We propose an identification method based on patent evolution analysis, which not only considers the comparative assessment of the patented technology and the currently existing technical content but also considers the future research development of the patented technology and the trend of change that receives attention. This presented method can better analyze the patent development metrically.

With the goal of “carbon peaking and carbon neutral”, energy and power systems need to undergo a low-carbon transformation. Advanced cyber systems should continuously enhance the reliability of the equipment to ensure high-quality access to power system communication [9], which can improve stability and promote the sustainable development of power systems [10]. At present, social energy resources are experiencing a long-term shortage, which is coupled with environmental pollution problems; thus, the efficient utilization of renewable energy is urgently required to actively build a renewable generation power system [11,12]. Researching and deploying disruptive energy generation technology [13] is essential to achieve clean energy production [14,15].

Therefore, in view of the importance of the above-mentioned fields for new-generation electric power systems, this paper selects the patents in the field of electric power communication and energy generation technology as cases to identify disruptive technologies in the fields and analyze the future potential of the related technologies to better grasp the key technologies and efficiently complete industrial upgrades and transform the new power systems.

2. Establishment of Disruptive Technology Identification Models

The process of disruptive technology identification based on patent evolution analysis in this paper mainly includes four parts: (1) the acquisition and pre-processing of patent data, (2) the establishment of an evolution matrix of patent data and the calculation of related index values, (3) disruptive technology determination, and (4) future development analysis. The specific identification process is shown in Figure 1. First, the patent text dataset in the relevant technical fields is obtained from the patent database, which is pre-processed to obtain the original patent text dataset based on time-series characteristics and cited frequency. Then, we calculate the similarity of text abstracts between two-by-two patents using cosine similarity [16] to construct the patent similarity matrix. Subsequently, the patent novelty matrix is obtained based on the relationship between the novelty and similarity of the patent database. Additionally, the patent evolution matrix is then constructed by calculating the novelty in each time slot one by one with the patent disclosure time as the breakpoint. Finally, we determine the index value of each patent based on the established disruptive technology identification index system and perform disruptive discrimination and future development analysis on the identified technologies.

2.1. Data Acquisition and Preprocessing

In order to obtain a scientific and effective target domain for the patent data set, this paper selects patents related to power systems from the patent database of China’s National Knowledge Infrastructure (CNKI). This patent database includes both Chinese and overseas patents, with a total of more than 140 million patents since 1970. Standardized searching information, including the patent name, patent abstract, patent number, patent claims, patent specification, inventor, patentee, patent application date, and patent publication date, was retrieved during the search until 1 November 2022. The obtained patents were sorted according to the year of publication, and the annual distribution diagram of authorized invention patents in the field of power systems can be obtained, as shown in Figure 2. The result contains 55,241 patent records in the field of power systems.

From the annual distribution chart, it can be seen that the first patent in the field of electric power systems was granted in 1985, and then it was in the latent period of development until 2001. From 2001 onwards, electric power systems entered a building period, and the technology in this field received attention from researchers, with the annual number of granted patents rising steadily. In recent years, the number of patents granted annually related to power systems has been in a state of rapid growth and shows a rising development trend, which indicates that the research and application in the field of power systems are in a very active state and have good development potential.

Through the analysis of keyword co-occurrence, we can obtain the research hotspots in the field according to the high-frequency keywords through the Vosviewer, which can realize the mapping of scientific knowledge and show the relationship between the structure, evolution, and cooperation of the knowledge domain. The keyword co-occurrence graph in the power system is shown in Figure 3. The larger the icon, the higher the frequency of occurrence. The line of adjacent nodes indicates the co-occurrence relationship between keywords. Different colors indicate different clusters. From the graph, high-frequency keywords that can be extracted include electric power system, generator, battery, information, etc. We can conclude that the hotspot directions of power systems mainly include energy generation, switching, energy storage, power communication, etc.

2.2. Disruptive Technology Recognition Algorithm

The disruptive technology recognition algorithm first calculates the similarity between patents, constructs the similarity matrix, and transforms the similarity matrix into the novelty matrix. Then, patent issue time is taken as the breakpoint to build the patent evolution matrix.

2.2.1. Patent Similarity Matrix

Before the computation of the novelty matrix, the similarity should be calculated. Cosine similarity has good performance in different corpora, so this paper uses cosine similarity to calculate patent similarity. First of all, the Python tool is used to preprocess the patent data, and the patent publication time is sorted to obtain the patent data set based on time series. Then, the Term Frequency–inverse Document Frequency (TF-IDF) [17] algorithm is used to find out the keywords of each patent, which is used to construct the keyword matrix set of the patent; and the word frequency vector of each patent is formed, respectively, to calculate the cosine similarity between the text vectors. A symmetric matrix of patent similarity corresponding to two pairs of patent documents is obtained, and the value of patent similarity is between 0 and 1. The larger the value, the greater the similarity.

The process of constructing the patent similarity matrix in this paper is as follows: first, assume that two data points in a dimensional space are $A=(a_1,a_2\dots ,a_i)$ and $B=(b_1,b_2\dots ,b_i)$ and then calculate the TF-IDF value of each word as the weight with the formula shown in (1).

$$TF-IDF=t{f}_{ij}\times id{f}_i=\frac{t{f}_{ik}\times \log \left(\frac{N}{n_k}+0.01\right)}{\sqrt{{\displaystyle \sum _{k=1}^m}\left(f_{ik}{}^2\times {\left[\log \left(\frac{N}{n_k}+0.01\right)\right]}^2\right)}}$$

(1)

where $t{f}_{ij}$ is the number of times that the featured item $t_i$ appears in the patent $d_j$, $id{f}_i$ is the reciprocal of the occurrence of the featured item $t_i$, $N$ is the total number of patent sets, $n_i$ is the number of patents in which the featured item $t_i$ appears, and the number of feature items, patents, and patent sets, respectively, and 0.01 is the empirical constant.

Then, cosine similarity is used to calculate the similarity of $A$ and $B$, as shown in Equation (2).

$$S(A,B)=\cos (\theta )=\frac{{\displaystyle \sum _{i=1}^n(a_i\times b_i)}}{\sqrt{{\displaystyle \sum _{i=1}^n(a_i}{)}^2}\times \sqrt{{\displaystyle \sum _{i=1}^n(b_i}{)}^2}}$$

(2)

where $S(A,B)$ is the similarity between the two data points $A$ and $B$, and $a_i$ and $b_j$ denote the TF-IDF values of the keywords in the two data points.

Finally, the similarity matrix of the patent texts can be established with the TF-IDF model and cosine similarity. The main process of patent similarity matrix construction is as follows:

(1)

The TF-IDF algorithm is used to perform the keyword calculation of the pre-processed patent text data set, extract the keyword list of each patent text data $P=\left\{w_1,w_2\dots ,w_n\right\}$, and construct a keyword co-occurrence matrix $D_p=\left\{P_1,P_2,\dots P_n\right\}$ with the number of patent documents and keywords according to the TF-IDF value of each keyword.

(2)

The word vector model is used to process each patent text $P=\left\{w_1,w_2\dots ,w_n\right\}$, and the word vector $V_p=\left\{v_1,v_2,\dots v_n\right\}$ of each patent text is constructed according to the processed word sequence. All of the patent text vectors $D_v=\left\{v_1,v_2,\dots v_n\right\}$ constitute the patent text vector set.

(3)

According to the cosine similarity formula, the similarity between two vectors in the patent text vector set is calculated as $S(V_i,V_j)$.

(4)

The $n\times n$ dimensional patent text similarity matrix $D_s={(d_{ij})}_{i,j=1,\dots ,n}$ is established according to the calculation result of the cosine similarity. The form of $n\times n$ dimensional patent text similarity matrix is constructed as shown in Formula (3):

$$D_S=\left(\begin{array}{ccccc}1& S(V_1,V_2)& \cdots & \cdots & S(V_1,V_n)\\ S(V_2,V_1)& 1& \cdots & & ⋮\\ ⋮& ⋮& \ddots & ⋮& ⋮\\ ⋮& & \cdots & 1& S(V_{n-1},V_n)\\ S(V_n,V_1)& \cdots & \cdots & S(V_n,V_{n-1})& 1\end{array}\right)$$

(3)

2.2.2. Patent Novelty Matrix

The determination of novelty can be established through the existing technology and patent similarity analysis. The calculation of the novelty of the technology described in the patents in two patent sets is based on the similarity between one patent and all previous patents. The greatest similarity of a patent to a previous patent may be considered “old”, whereas novelty refers to the particular part of a patent that is not similar to the previous patent in terms of its greatest similarity.

The unique parts are taken as the novelty of the patent text at the time of publication. We calculate the proportion of the unique parts of the patent that is not similar to one of the previously published patents by subtracting the maximum similarity between the one and other published patents. The formula used to calculate patent novelty when disclosed is shown in Formula (4):

$$NO{V}_{pub}=1-Max(S_{i(n)}),n<i$$

(4)

where $NO{V}_{pub}$ is the novelty of the patent when it is disclosed, $S_{i(n)}$ is the similarity between the patent $P_i$ and other patents disclosed before the publication of $P_n$, and $Max(S_{i(n)})$ represents the maximum similarity of each patent $P_i$ disclosed before the patent $P_n$.

The $n\times n$ dimensional patent text novelty matrix is constructed with the novelty coefficients between the patent texts arranged in the time series, $D_{NOV}$, as shown in Formula (5):

$$D_{NOV}=\left(\begin{array}{ccccccc}NO{V}_{pub1}& & & & & & \\ NOV(p_2,p_1)& NO{V}_{pub2}& & & & & \\ ⋮& ⋮& \ddots & & & & \\ NOV(p_i,p_j)& ⋮& ⋮& NO{V}_{pubi}& & & \\ ⋮& ⋮& ⋮& ⋮& \ddots & & \\ ⋮& ⋮& \cdots & \cdots & \cdots & NO{V}_{pubn-1}& \\ NOV(p_n,p_1)& \cdots & \cdots & \cdots & \cdots & NOV(p_n,p_{n-1})& NO{V}_{pubn}\end{array}\right)$$

(5)

2.2.3. Patent Evolution Matrix

After constructing the patent novelty matrix, the evolution algorithm of patent novelty is used to calculate the change process of patent novelty over time. The standard for judging patent novelty is the publication date. When constructing the patent evolution matrix, the publication year will be used as the criterion to make time slices of the patent data set. This paper analyzes the novelty evolution of patented technology from the publication year to the present, and we can obtain the evolution of patent technology development and provide a new perspective for identifying disruptive technologies.

In this paper, the year of patent novelty is expressed by Formula (6), which evolves over time.

$$NO{V}_{(i,y)}=\{\begin{array}{c}1-Max(S_{i(n)}),n<y+1,(y_i<y)\\ 0,(y_i<y)\end{array}$$

(6)

where $NO{V}_{(i,y)}$ is the novelty of the patent in terms of the year, $S_{i(n)}$ represents the similarity between patent $P_i$ and each patent $P_n$ disclosed before the year $y+1$, and $Max(S_{i(n)})$ represents the maximum similarity between the patent and each patent disclosed before the year. If the patent is not publicly published in the year, the calculation result is null.

We construct the patent evolution matrix by sorting and integrating the publication time series of each patent, as shown in Formula (7):

$$D_{NOV}{}_{(p_i,y_{{}_n})}=\left(\begin{array}{ccccc}NOV(p_1,y_1)& NOV(p_1,y_2)& \cdots & \cdots & NOV(p_1,y_n)\\ NOV(p_2,y_1)& NOV(p_2,y_2)& \cdots & & ⋮\\ ⋮& ⋮& \ddots & ⋮& ⋮\\ ⋮& & \cdots & NOV(p_{i-1},y_{n-1})& NOV(p_{i-1},y_n)\\ NOV(p_i,y_1)& \cdots & \cdots & NOV(p_i,y_{n-1})& NOV(p_i,y_n)\end{array}\right)$$

(7)

2.3. An Indicator of Disruptive Technology

In this paper, establishing the evaluation index system is an important task for the study of disruptive technology identification. A reasonable and effective index system is not only conducive to obtaining highly effective evaluation results but also has an important impact on exploring the nature of the development of disruptive technology. In this paper, disruptive technologies are identified from the perspective of patent evolution and the characteristics of disruptive technologies. Compared with other technologies, disruptive technologies have high novelty, strong development momentum, and great influence [18]. Therefore, the effective recognition of related technologies is reflected through their radical novelty, relatively rapid growth, and prominent impact. The established system of disruptive technology discriminators is as follows:

(1)

In the process of disruption, novelty and growth are prominent in the early stage and weaken over time. We characterize the radical novelty of disruptive technologies with the evolution of patent novelty. Therefore, the novelty of the described patents is defined as follows: if $NO{V}_{pub}$ is less than 0.5, the novelty is weak; if $NO{V}_{pub}$ is between 0.5 and 0.8, the novelty is medium; and if $NO{V}_{pub}$ is greater than 0.8, the novelty is strong.

(2)

The relatively rapid growth is represented by the average annual growth of similar technologies. The relatively rapid growth is due to the fact that disruptive technologies tend to grow faster compared to other technologies in the same field. The growth of disruptive technologies is represented by the average annual growth of similar technologies through the constructed patent similarity matrix. The greater the average annual growth, the faster the development of disruptive technologies.

(3)

A prominent impact is represented by the number of patent citations [19]. The higher the citation frequency of a patent, the more subsequent technologies will be affected by it and the more important the technology covered by the patent [20]. In this paper, the patents that accounted for the top 10% of all patents cited in the year of publication were used to characterize the prominent impact of disruptive technologies.

3. Case Studies

3.1. Theoretical Verification

In this section, disruptive technology is identified in the field of power generation before 2015 to test the validity of the method. From

Table 1. The patent-related index values in the power communication field before 2012.

Patent No.	Patent Code	Publish Time	Publication Novelty	Testing Novelty	Evolutionary Year	Annual Growth	Cited Frequency Sort
1	CN1223355	1999	1	0.6	16	12.7	Top 30%
…	…	…	…	…	…	…	…
128	CN102834605A	2012	1	0.823	3	16.27	Top 10%
…	…	…	…	…	…	…	…
200	CN204918497U	2015	1	0.713	0	13.27	Top 20%

, the trend of the decreasing novelty of patent CN102834605A indicates that the patent received widespread attention after publication. The annual growth of this patent is 16.27, which indicates that the technology present in this patent has undergone great development. The cited frequency accounts for the top 10%. The wind power technology covered in this patent is identified as a disruptive technology. As shown in Figure 4, wind power technology shows a tendency to grow explosively in the electricity generation sector. The results verify the correctness of the proposed method.

3.2. Power Communication Technology as an Example

In this section, we first obtain the patents in the field of power communication technology and then use the algorithm to obtain the similarity matrix, the novelty matrix, and the patent evolution matrix for the field.

3.2.1. Data Acquisition

The patents related to electric power systems in the field of electric power communication are extracted from the patent database of CNKI. The annual distribution of authorized invention patents in the field of electric power communication can be obtained after statistical analysis and collation, as shown in Figure 5.

3.2.2. Patent Similarity Matrix

The patent text dataset in the field of power communication was obtained after pre-processing. The TF-IDF algorithm is used to extract the keywords present in the patent documents. We construct a keyword co-occurrence matrix of patents with the number of behavioral documents and the number of keywords and generate the word frequency vector of the patent text. The similarity between the patents in pairs is calculated with the cosine similarity to obtain the patent similarity matrix (381 × 381), as shown in

Table 2. The patent similarity matrix.

Patent No.	1	2	4	4	5	6	…	381
1	1	0.004	0.005	0.005	0.010	0.011	…	0.006
2	0.004	1	0.140	0.146	0.137	0.153	…	0.003
3	0.005	0.140	1	0.194	0.185	0.215	…	0.016
4	0.005	0.146	0.194	1	0.151	0.182	…	0.010
5	0.010	0.137	0.185	0.151	1	0.229	…	0.007
6	0.011	0.153	0.215	0.182	0.229	1	…	0.012
7	0.015	0.002	0.019	0.008	0.051	0.836	…	0.012
…	…	…	…	…	…	…	…	…
381	0.006	0.003	0.016	0.010	0.007	0.012	…	1

. The first row and the first column indicate the number of each patent. The calculated result indicates the similarity value between each patent. The patent similarity is between 0 and 1, and the larger value indicates a higher similarity between the two patents.

3.2.3. Patent Novelty Matrix

Patent novelty is obtained by subtracting the maximum similarity novelty measure of a patent from its related patents. After calculating the patent text similarity, the novelty at the time of disclosure is calculated as the proportion of the distinctive part of a patent. In order to study the process of the novelty change of a patent after public publication, the date of publication can be regarded as a point in time. The novelty matrix in the field of power communication can be obtained by the novelty calculation formula, as shown in

Table 3. The patent novelty matrix.

Patent No.	1	2	3	4	5	6	…	381
1	1	0.996	0.995	0.995	0.990	0.989	…	0.897
2	0.996	1	0.860	0.854	0.854	0.847	…	0.842
3	0.995	0.860	1	0.806	0.806	0.785	…	0.781
4	0.995	0.854	0.806	1	0.849	0.818	…	0.818
5	0.990	0.854	0.806	0.849	1	0.771	…	0.249
6	0.989	0.847	0.785	0.818	0.771	1	…	0.164
7	0.985	0.847	0.785	0.818	0.771	0.164	…	0.823
…	…	…	…	…	…	…	…	…
381	0.897	0.842	0.781	0.818	0.249	0.164	…	1

, where the first row and the first column indicate the number of each patent, and the calculated result indicates the novelty value among each patent.

3.2.4. Patent Evolution Matrix

We construct a patent evolution matrix by calculating the change in the novelty of patents from year to year. The patent evolution matrix is constructed as shown in

Table 4. The patent evolution matrix.

	Patent No.	1	2	3	4	5	6	…	381
Year
2002		1
2008		0.996	1
2009		0.990	0.854	0.806	0.849	1
2010		0.985	0.847	0.785	0.818	0.249	0.164	…
2011		0.985	0.847	0.785	0.818	0.249	0.164	…
2012		0.969	0.847	0.785	0.818	0.249	0.164	…
2013		0.947	0.847	0.785	0.818	0.249	0.164	…
…		…	…	…	…	…	…	…
2022		0.897	0.842	0.781	0.818	0.249	0.164	…	1
$NO{V}_{pub}$		1	1	0.806	0.849	1	0.164		1

. Most of the patents have high novelty at the time of disclosure. As time evolves, some hot research areas have less novelty due to the appearance of identical patented inventions.

3.3. Energy Generation Technology as an Example

In this section, we first obtain the patents in the field of energy generation technology and then use the algorithm to obtain the similarity matrix, the novelty matrix, and the patent evolution matrix for the field.

3.3.1. Data Acquisition

We extract the patents in the field of energy generation from the patents related to electric energy systems in the CNKI, including the relevant invention patents and utility model patents. We obtain the annual distribution of authorized invention patents in the field of energy generation after statistical analysis and collation, as shown in Figure 6.

3.3.2. Patent Similarity Matrix

We can obtain the patent text dataset after processing by combining the TF-IDF model and cosine similarity. The similarity matrix of a patent text can be accessed as the symmetric moments of patent text similarity (426 × 426), as shown in

Table 5. The patent similarity matrix.

Patent No.	1	2	4	4	5	6	…	426
1	1	0.054	0.029	0.012	0.078	0.003	…	0.009
2	0.054	1	0.038	0.004	0.034	0.010	…	0.010
3	0.029	0.038	1	0.016	0.029	0.004	…	0.031
4	0.012	0.004	0.016	1	0.017	0.023	…	0.004
5	0.078	0.034	0.029	0.017	1	0.019	…	0.015
6	0.003	0.010	0.004	0.023	0.019	1	…	0.021
7	0.050	0.062	0.008	0.008	0.082	0.015	…	0.020
…	…	…	…	…	…	…	…	…
426	0.009	0.010	0.031	0.004	0.015	0.021	…	1

, where the first row and the first column indicate the number of each patent, and the calculated result indicates the similarity value between each patent.

3.3.3. Patent Novelty Matrix and Patent Evolution Matrix

After calculating the similarity of patent texts, we can obtain the novelty matrix in the field of energy generation by calculating the proportion of unique parts that are not similar to previously published patents, as shown in

Table 6. The patent novelty matrix.

Patent No.	1	2	3	4	5	6	…	426
1	1	0.946	0.946	0.946	0.922	0.922	…	0.871
2	0.946	1	0.962	0.962	0.962	0.962	…	0.857
3	0.946	0.962	1	0.984	0.971	0.971	…	0.784
4	0.946	0.962	0.984	1	0.983	0.977	…	0.877
5	0.922	0.962	0.971	0.983	1	0.981	…	0.695
6	0.922	0.962	0.971	0.977	0.981	1	…	0.663
7	0.922	0.938	0.971	0.977	0.918	0.985	…	0.793
…	…	…	…	…	…	…	…	…
426	0.871	0.857	0.784	0.877	0.695	0.663	…	1

, where the first row and the first column indicate the number of each patent, and the result of the calculation indicates the result after the novelty of each patent has changed over time.

The patent evolution matrix can be constructed by calculating the change in the novelty of patents from year to year, as shown in

Table 7. The patent evolution matrix.

	Patent No.	1	2	3	4	5	6	…	426
Year
2008		1						…
2011		0.946	0.962	0.984	1			…
2012		0.922	0.938	0.840	0.977	0.918	0.985	…
2013		0.883	0.875	0.840	0.977	0.850	0.961	…
2014		0.883	0.875	0.840	0.947	0.850	0.958	…
2015		0.883	0.857	0.840	0.943	0.850	0.850	…
2016		0.883	0.857	0.840	0.943	0.850	0.850	…
…		…	…	…	…	…	…	…
2022		0.871	0.857	0.784	0.877	0.695	0.663	…	1
$NO{V}_{pub}$		1	0.962	0.984	1	0.918	0.985	…	1

.

4. Discrimination and Analysis

In this section, we first process the relevant data obtained in the previous section to obtain the index values of the patent data, then we discriminate the disruptive technologies through the discriminatory indexes proposed in this paper, and finally, we analyze the future development of the identified disruptive technologies.

4.1. Identification of Technical Results

According to the disruptive technology identification index system built based on its characteristics, attributes, and the citation frequency ranking of related patents in the previous section, the related patent indicators in the field of power communication and energy generation can be obtained, as shown in

Table 8. The patent-related index values in the power communication field.

Patent No.	Patent Code	Publish Time	Publication Novelty	Testing Novelty	Evolutionary Year	Annual Growth	Cited Frequency Sort
1	CN2503163	2002	1	0.897	20	18	Top 30%
2	CN201134298	2008	1	0.842	14	25	Top 20%
…	…	…	…	…	…	…	…
367	CN115314270A	2022	0.973	0.973	0	14	Top 10%
…	…	…	…	…	…	…	…
381	CN218469730U	2022	1	1	0	0	Top 20%

and

Table 9. The patent-related index values in the energy generation field.

Patent No.	Patent Code	Publish Time	Publication Novelty	Testing Novelty	Evolutionary Year	Annual Growth	Cited Frequency Sort
1	CN201084000	2008	1	0.871	14	26.79	Top 10%
2	CN201947494U	2011	0.962	0.857	11	33.72	Top 30%
…	…	…	…	…	…	…	…
355	CN111075529B	2022	0.973	0.973	0	21	Top 10%
…	…	…	…	…	…	…	…
426	CN115313528A	2022	1	1	0	0	Top 20%

, respectively, where the annual growth is the growth of similar patents. The following conclusions are shown below:

First, from the perspective of radical novelty, two patent technologies, No. 367, in the field of electric power communication (CN115314270A), and No. 355, in the field of energy generation (CN111075529B), were both published in 2022. They had high novelty at the time of public publication, with novelty coefficients of 0.973 and 0.973, respectively, and thus belong to high-novelty patented technologies at the time of public publication. The technical novelty of these two technologies tends to decrease in the process of development. It has been illustrated that these two high novelty patents received wide attention from technical researchers after publication. They are referred to by technicians in the field and provide a theoretical reference for the subsequently patented technologies. Such patented technologies have good development potential and radical novelty compared with other technologies in the field.

Secondly, from the perspective of relatively rapid growth, the number of the average annual growth of similar technologies, granted No.367 (CN115314270A) and No. 355 (CN111075529B), are 21 and 14, respectively, in the fields of electric power communication and energy generation. They both have similar patents within a relatively short period, which is just after their publication. It has been illustrated that both are in a state of rapid growth. These two patent technologies have faster growth rates and stronger development momentum. Such patented technologies have good development potential and relatively rapid growth compared to other technology fields.

Thirdly, from the viewpoint of outstanding influence, the citation frequency of the two patents, No. 367 (CN115314270A), in the field of electric power communication, and No. 355 (CN111075529B), in the field of energy generation, are in the top 10% of the citation frequency ranking table in the relevant fields. They have been highly cited in a short period of publication time, and it can be seen that both technologies have an outstanding impact.

According to the identification index system of disruptive technologies constructed in this paper, two patent technologies, No. 367 (CN115314270A), in the field of electric power communication, and No. 355 (CN111075529B), in the field of energy generation, perform very well compared to other technologies in the three discriminative indexes, which shows the obvious characteristics. Therefore, these two patented technologies are judged to be disruptive technologies in the fields of electric power communication and energy generation, respectively, and these two patented technologies are retrieved through their patent number: (1) Quantum key-based hierarchical encryption method and communication method for electric power business and a Brayden cycle energy generation system for pulsed fusion reactors: this patent mainly applies quantum technology to the field of electric power communication, and improves the encryption efficiency and the use efficiency of quantum keys under the premise of guaranteeing data security. (2) A Brayden cycle energy generation system for pulsed fusion reactors: this patent belongs to the field of nuclear fusion reactor technology, which can provide durable, efficient, and clean energy, and thus completely solve the energy problem.

4.2. Analysis of Technical Development

Through the presentation of patent evolutionary analysis methods, disruptive technologies in the fields of electricity, communications, and power generation are identified as quantum technology and controlled nuclear fusion, respectively.

(1) The quantum technology

With the development and the deepening applications of quantum technology, the future of quantum communication, quantum computing, quantum measurement, quantum materials, etc., in electric power systems will have a wide range of application prospects and bring about profound changes to power systems. Especially in the field of electric power communication, quantum communication technology can be used to carry out the application and build a network architecture. It is essential to ensure the security of the electric power communication network and guarantee the safe and stable operation of the power grid. With the deepening of digitalization and the continuous improvement of system coordination requirements, power systems in all aspects of computing capacity requirements will undergo exponential growth. Traditional computing methods will not be able to handle the existing amount of computing. However, quantum computing technology can overcome this problem as it can be used to handle large-scale computations. The latest research results are constantly applied in the field of power quantum measurement, but the current academic research is blossoming. The key technology research has not yet reached a consensus. Most of the technologies are still in the laboratory research stage.

The quantum technology revolution is receiving more and more attention as it continually achieves new breakthroughs. The advent of quantum technology has spawned technological change, and device developments continue to change the face of the world. It is gradually becoming the cornerstone and driving force of social leapfrog development, providing fundamental security services for the development of the information society and completely solving the problem of information security in terms of infrastructure.

(2) The controlled nuclear fusion

Reviewing the history of nuclear fusion research, it can be seen that the research on magnetic confinement fusion has developed rapidly and made great achievements in the past 20–30 years. The scientific feasibility of developing fusion energy has been confirmed in tokamak-type magnetic confinement fusion devices with breakthroughs in tokamak-type magnetic confinement fusion devices. The design of the international thermonuclear experimental reactor (ITER) has been able to give mankind a bright future, but achieving the practical application of fusion energy is still a long and uneven journey.

Controlled nuclear fusion has some incomparable advantages over other energy sources: abundant sources of raw materials, a non-polluting process of nuclear fusion, and safety and reliability. As fossil fuels such as coal, oil, and natural gas will eventually be depleted, people are looking forward to new energy sources. Controlled fusion reactions can produce vast amounts of energy. Since this energy source is clean, safe, and uses inexhaustible seawater as raw material, controlled fusion is the main hope for human energy in the next century.

5. Conclusions

We take the fields of electric power communication and power generation as examples to identify disruptive technology according to the proposed patent evolution analysis and the identification index system. This paper accomplishes and achieves the following: (1) the effective identification of disruptive technologies in the fields of electric power communication and energy generation and (2) completes the analysis of the future potential of the above disruptive technologies so as to grasp the key technologies and facilitate industrial transformation and upgrades more quickly.

However, there are two shortcomings in this paper. Firstly, the similarity matrix is calculated on the basis of the similarity between the keywords, which are directly derived from the patents. It does not take into account the semantics of the keywords. The similarity matrix will be inaccurate if the extracted keyword clusters do not represent the characteristics of the whole document well. Additionally, the proposed index system only contains the novelty, growth, and impact of the patents, which is not comprehensive enough from the perspective of discriminating. Therefore, how to conduct patent intelligence analysis at both macro and micro levels and how to construct multi-dimensional identification indicators of the patented technologies are the focus of the next research.