Industrial Sustainable Development: The Development Trend of Programmable Logic Controller Technology

Abstract

Programmable Logic Controllers (PLCs) are indispensable for current and future industrial development, especially in smart factories, smart home technology, automated production lines, and machinery manufacturing. This study presents the trends in PLC software and hardware development through a technology roadmap and offers relevant suggestions to help industries achieve sustainable development, enhance market competitiveness, and provide references for research. Through expert interviews and fuzzy Delphi analysis, this study points out that future PLC development needs to focus on editing interfaces, syntax, Central Processing Units, Memory Units, and Communication Modules. Specific recommendations include visualizing regional/global label settings and connection settings, adding Python, JAVA, LabVIEW, and Scratch syntax, improving instruction execution speed, expanding program and expansion capacities, and adopting dual-channel Ethernet and connections to external networks and wireless networks. Fuzzy hierarchical analysis shows that Communication Modules are the most important component, followed by Central Processing Units and syntax expansion, and, finally, program and expansion capacity enhancements. These suggestions aim to promote product innovation and social environment demand evaluation, enhance product competitiveness, and achieve sustainable development goals.

1. Introduction

In 2011, in Germany, Hannover Messe introduced the concept of a new industrial era—the Internet of Things (IoT). This vision of “digitalized and intelligent” industrial development began to take root in various countries, leading them to research and formulate national industrial development directions and indicators. This era is regarded as the fourth industrial revolution, known as Industry 4.0 [1,2]. Its key technologies include Cyber-Physical Systems (CPSs), which integrate digital information technology with physical devices; the Internet of Things (IoT), which connects various physical devices into a complete network system; and the Internet of Services (IoS), which forms an interconnected network system of various social services. Smart Factories combine these three key technologies to further enhance factory productivity and overall efficiency [3].

In 2015, the United Nations announced the “2030 Sustainable Development Goals” (SDGs), comprising 17 goals. Among them, Goal 8 (“Decent Work and Economic Growth”) and Goal 9 (“Industry, Innovation, and Infrastructure”) align with the development process of Industry 4.0. In addition to enhancing industrial competitiveness, flexibility, efficiency, and quality through technological advancements, the development process also emphasizes sustainability, aligning with the 2030 SDGs. Sustainability means meeting the needs of current and future generations [4,5,6].

In 2016, Denmark’s Universal Robots noted that societal demand had shifted from quantity to personalization. In the same year, Japan proposed “Society 5.0” for 2030, a human-centered approach linking Artificial Intelligence (AI), robotics, IoT, and other technologies to create an intelligent society. This forecasted the era of big data utilization. Consequently, in 2019, the Japan Business Federation revised the Corporate Action Charter as the action guideline for Japanese companies. Three years later, in 2021, Rockwell Automation at the 30th Automation Fair proposed that individual companies should go beyond Industry 4.0 to formulate a blueprint for Industry 5.0. They identified four key areas: 1. The evolution of cloud, distributed computing, and software; 2. The development environments for control and integration; 3. AI operation management, including real-time software and digital services; and 4. Autonomous systems and the expansion of the workforce [7]. Frost and Sullivan defined the differences between the two, emphasizing the provision of customer services, high customization, responsiveness, distributed supply chains, interactive products, and the return of human labor to factories. From Industry 4.0 to the upcoming Industry 5.0 and from IoT formed by industrial integration to AI formed by big data utilization, it is evident that the future will see the combination of IoT and AI [8].

In the development process of Industry 4.0, most manufacturing industries and smart factories have introduced Programmable Logic Controllers (PLCs) as a critical component to enhance their competitiveness, industrialization, innovation, and infrastructure, as shown in Figure 1. From Figure 1, it is evident that the PLC serves as a crucial link for the exchange of information between sensors and other industrial systems. From the replacement of traditional automation circuits to the integration of numerous sensors, global industrial development has accelerated rapidly. Through the PLC, external information from sensors is captured and directed towards the development of a new industrial paradigm in the Internet of Things (IoT) and smart factories. The PLC has become an indispensable component for the sustainable development of future industries [9]. This study sets the development background from Industry 4.0 to Industry 5.0 through expert interviews to understand the current state and future trends of PLCs. Furthermore, using fuzzy Delphi method analysis, the consensus among experts on the current state and future trends can be determined, facilitating the identification of possible development directions [10]. Given the infinite possibilities and limited resources in trend development, the Fuzzy Analytic Hierarchy Process further analyzes the priority order of achieving consensus trends [11,12]. The application of technology roadmaps, with their overall and continuous characteristics, can effectively present possible development directions of current and future trends, aiding in their research, development, and promotion in related industries [13,14].

Figure 1. System hierarchy diagram.

The specific research objectives of this study are as follows: 1. To present the development trend map of PLC hardware and software through a technology roadmap. 2. To propose overall development trend suggestions for PLCs, providing a reference for industrial development and research.

2. Literature and Theoretical Discussion

2.1. Programmable Logic Controllers (PLCs)

Programmable Logic Controllers (PLCs) are essential for the industrial development of automation, robotics, and smart technologies [15,16,17,18]. Their hardware is divided into five units: Power Module, Central Processing Unit (CPU), Memory Unit, Input/Output Unit, and Communication Module [19,20]. Thiago Alves and colleagues (2018) noted that PLCs exhibit strong performance in the initial phase [21]. Arquimedes Canedo and colleagues (2014) highlighted that scan cycle time and high-speed communication are among the most critical indicators in PLC-based control systems [22]. Laurence Crestani Tasca and colleagues (2020) mentioned that due to the simplistic needs of early automation systems, PLCs, as the primary and extensively used central controllers, could not meet the more diverse and computationally intensive requirements of current automation systems, leading to a reduction in scan cycles to enhance performance [23]. Daniil Chivilikhin and colleagues (2020) pointed out the delays and high costs many manufacturers face when transitioning to Industry 4.0 [24]. Xia Mao and colleagues (2022) observed that PLCs are widely used in Industrial Control Systems (ICSs), with many systems recently shifting to cloud-based control due to the development of Industry 4.0 [25]. Scott C. Rowe and colleagues (2022) stated that automation courses are a professional branch for students in mechanical, electrical, and mechatronic engineering, covering topics like robotics, pneumatics, and PLCs to find the best automation design to improve system efficiency [17]. Martin A. Sehr and colleagues (2021) mentioned that Industry 4.0, the era of the Internet of Things and big data, places PLCs at the boundary between the digital world and the physical world as mature factory automation controllers and industrial process design platforms. Yet, researchers have limited understanding of them, thus necessitating an analysis of PLCs [26]. Vicente Esteve and colleagues (2024) proposed defining the monitoring process for industrial induction surface hardening applications, emphasizing the integration of sensors and current measurement technologies with control systems. They suggest using Programmable Logic Controllers (PLCs) to support integrated monitoring systems and Field-Programmable Gate Arrays (FPGAs) to achieve hybrid acquisition and instrumentation systems [27]. Dong Ji and colleagues (2024) developed an automatic tomato transplanter using Programmable Logic Controllers (PLCs) in combination with sensors, mechanical transmission structures, pneumatic technology, and stepper motors [28].

Reviewing the aforementioned literature, it is evident that recent studies on Programmable Logic Controllers (PLCs) have mainly focused on performance enhancement, communication security, and integrated applications across various systems. However, research based on market demand and industrial development is relatively lacking, and there has not been sufficient exploration of whether the trends of PLCs meet the required development needs. It is imperative to consider the future development trends of the Central Processing Unit (CPU), Memory Unit, Communication Module, and programming software in Programmable Logic Controllers [29,30,31,32]. This study acknowledges the vast array of PLC brands and models. Hence, it specifically focuses on the hardware of the Mitsubishi Programmable Logic Controller FX Series Model FX5U, as delineated in Table 1 [33,34]. The software and programming are primarily centered around GX Works3.

Table 1. Evolution of the FX Series.

Model	Function
FX1N	Input unit: DC; Output unit: Relay and Transistor
	Power Supply Unit: AC 100 V~240 V
	Communication Module: RS-232C and RS-422/485
	Memory Unit: 8 K SETP
	Central Processing Unit: Basic Order 0.55~0.7 μs; Application Order 2~100 μs
FX2N	Input unit: DC; Output unit: Relay, Transistor, and SSR
	Power Supply Unit: AC 100 V~240 V
	Communication Module: RS-232C, RS-422/485, and CC-Link
	Memory Unit: 8 K STEP/16K STEP
	Central Processing Unit: Basic Order 0.08 μs; Application Order 1.52~100 μs
FX3U	Input unit: DC; Output unit: Relay and Transistor
	Power Supply Unit: AC 100 V~240 V
	Communication Module: RS-232C, RS-422/485, CC-Link, and Ethernet
	Memory Unit: 64 K SETP
	Central Processing Unit: Basic Order 0.065 μs; Application Order 0.642~100 μs
FX5U	Input unit: DC; Output unit: Relay and Transistor
	Power Supply Unit: AC 100 V~240 V
	Communication Module: Built-in RS-485 and Ethernet
	Memory Unit: 64 K SETP
	Central Processing Unit: Basic Order 34 ns; Application Order 34 ns

2.2. Technology Roadmap

Technology roadmaps serve as essential frameworks for companies to strategize their future layout analysis and planning over the next few years or even up to ten years, including aspects such as product planning, allocation of resources, and choosing directions for technological advancements. The earliest technology roadmap was developed by Motorola in the late 1970s, aimed at planning the development of automotive transistor radios [35].

Including technological evolution and market entry timing, this prior preparation facilitates resource allocation and technological development planning, allowing the company’s limited resources to be maximized. It is evident that technology roadmaps possess qualities of futurism, comprehensiveness, and communicability [36]. Ummaraporn Pora and colleagues (2022) highlighted the importance of creating technology roadmaps for organizations to grasp future development directions amidst rapid environmental changes and the emergence of new technologies [37]. Maicon Gouvêa de Oliveira and colleagues (2022) suggested that gathering information through personal interviews and then synthesizing it for expert team discussions (although it increases the preliminary workload) enhances the effectiveness of information collection for roadmap creation [38]. Hong Miao and colleagues (2022) emphasized understanding the entire innovation development process of a new technology’s market entry as a critical aspect of strategic planning and development decisions, with technology roadmaps being seen as the most effective method [39]. Masayoshi Watanabe and colleagues (2022) noted the collaboration between Japan’s Ministry of Economy, Trade and Industry, and the New Energy and Industrial Technology Development Organization to develop and apply the Kaizen method for evaluating, verifying, and revising technology roadmaps for the implementation of industrial science and technology policies [40]. Mintak Han and Youngjung Geum (2022) pointed out the significance of data in the business environment, suggesting the integration of data into strategic management through a technology roadmap to enhance corporate competitiveness [41].

Technology roadmaps provide a wider and more far-reaching view of the selected area, acting as a comprehensive route from the current state to future market dynamics, knowledge acquisition, and envisioning within the domain [37,39]. This involves examining the interactions among the subject, environment, and target, as shown in Figure 2, to analyze technological trajectories and resource allocations for the coming years or decades [36]. The intention of this research is to depict the present conditions and anticipated trends by crafting a technology roadmap.

Figure 2. Diagram detailing the interplay between technology maps, subjects, targets, and environmental contexts.

2.3. Expert Interview

This method, also known as the interview method or in-depth interviews, aims to deepen the interviewee’s understanding of the topic through various means of communication and to obtain genuine and relevant information. It is a type of qualitative research characterized by authenticity, interactivity, representativeness, and subjectivity. The approach can be categorized based on the number of participants for individual interviews or group interviews, the mode of contact in face-to-face, telephone, or video interviews, the frequency of single (cross-sectional) and multiple (longitudinal) interviews, and the structure of the interviews (e.g., structured, semi-structured, and unstructured interviews) [42], as shown in Table 2.

Table 2. Types of expert interviews.

Interview Structure Types	Description
Structured Interviews	Interviews that are prepared with a detailed framework and content beforehand and tailored to the theme exhibit a higher degree of structure. Pros: They enable respondents to grasp the information desired by the interviewer more clearly, making the process of compiling the data from the interview more straightforward. Cons: The approach suffers from inflexibility, often resulting in a rigid process.
Semi-Structured Interviews	Interviews that are conducted with a pre-planned structure and content while allowing for flexibility and spontaneous questions during the process fall into a moderately structured category. Pros: They enable respondents to better understand the interviewer’s desired information, offering more flexibility in the interview process and preventing rigidity while reflecting the true nature of the topic more accurately. Cons: Compared to structured interviews, the data are not as straightforward to organize.
Unstructured Interviews	Interviews that are pre-planned with a detailed structure and content tailored to the subject while accommodating flexibility and spontaneous queries during the session possess a moderate degree of structure. Pros: They provide interviewees with a better understanding of the information that the interviewer seeks, ensure more flexibility throughout the interview process to prevent stiffness, and capture the true nature of the topic more accurately. Cons: Compared to structured interviews, the data may not be as easy to organize.

Trevor Hogan and colleagues (2016) noted that interviews are commonly used for post-event data collection and understanding situations, which can be structured, semi-structured, or exploratory, mainly focusing on result processing [42]. Mohammed Sayagh and colleagues (2020) highlighted that conducting semi-structured interviews with experts under defined objectives can identify necessary adjustments and potential challenges, preventing openness from becoming unmanageable, followed by creating surveys for analysis to find possible solutions [43]. Pramod Chundury and colleagues (2022) suggested collecting data on how visually impaired individuals perceive spatial environments without relying on visual senses through semi-structured interviews and then analyzing them to produce a feasibility report on accessible visualization facilities [44]. Therefore, this study employs this method to understand future trends of Programmable Logic Controllers according to experts and to establish evaluation items, facilitating the creation of subsequent fuzzy Delphi method questionnaires.

2.4. Fuzzy Delphi Method

This method involves a group that communicates not through face-to-face meetings but by completing questionnaires on specific topics, which are then distributed to selected experts for their input. The feedback is collected and synthesized to achieve the highest level of consensus, serving as an analytical and forecasting tool. To prevent bias, the questionnaire cycle is typically repeated multiple times rather than only once [45,46]. Geet Parekh and colleagues (2018) pointed out that cybersecurity is a critical factor in national competitiveness. To uncover the fundamental concepts that beginners need to grasp in cybersecurity and the core ideas that professionals should understand, they used the Delphi technique for analysis [47]. Whisper Maisiri and colleagues (2021) observed that Industry 4.0 significantly impacts the technologies, system workflows, and competency demands in the manufacturing sector. They used the Delphi technique to identify the key skills needed in the manufacturing industry in the context of Industry 4.0 [48]. The Delphi method typically involves more than three rounds of surveys to ensure thorough understanding. In many situations, due to ambiguous and unclear boundaries that make precise answers elusive, Akira Ishikawa and colleagues introduced an innovative analytical and forecasting tool in 1993, combining the Max–Min rule of fuzzy theory and fuzzy integration with the traditional Delphi method, giving rise to the fuzzy Delphi method. This aimed to reduce the time costs and potential inaccuracies inherent in conducting several rounds of traditional Delphi surveys [49]. Jao-Hong Cheng (2001) used Likert’s five-point scale and weighted scores, combined with the method of triangular fuzzy numbers, to establish both optimistic and conservative triangular models. By examining the overlap between these dual triangular models and employing the gray area test method to calculate consensus [50], Jeng Tsang-Bin (2001) posited that the absence of a gray area should also be considered a valid expression of expert and scholarly opinions. Consequently, he revised the algorithm to allow for the direct provision of interval quantitative values for each evaluation criterion by experts, with the interval’s minimum value representing a conservative estimate and the maximum value representing an optimistic estimate. He then established both optimistic and conservative triangular models. Unlike previous methods, Deng suggested that a lack of overlap between these two triangular models indicates a consensus on the quantified intervals provided by the experts. The consensus value is calculated as the arithmetic mean of the geometric means of the two models when there is no overlap, or, if there is an overlap, by calculating the maximum quantification score within the overlapping area [51]. This study uses the fuzzy Delphi method to analyze whether the opinions expressed by experts in the questionnaires reach a consensus.

2.5. Fuzzy Analytic Hierarchy Process

T.L. Saaty and L.G. Vargas developed the Analytic Hierarchy Process (AHP) in 1979, a technique designed to select the optimal decision from various decision-making criteria by structuring the issue at hand, assigning weights to different elements, and evaluating the consistency of the decisions made [52,53]. However, the survey process is susceptible to variations in interpretation. To minimize these semantic discrepancies, P.J.M. van Laarhoven and W. Pedrycz introduced the integration of AHP with fuzzy logic (Fuzzy AHP) in 1983, aiming to tackle the challenges of multiple decision criteria and the biases resulting from ambiguous thought processes [54]. Buckley, J. J. proposed the Fuzzy Analytic Hierarchy Process (FAHP) in 1985, merging the principles of the Analytic Hierarchy Process with fuzzy logic, aiming to diminish the impact of semantic discrepancies, thereby being more closely aligned with the true expressions and intentions of participants [55]. R.L.N. Murty and colleagues (2020) used the Fuzzy Analytic Hierarchy Process to pinpoint essential decision-making criteria, thereby facilitating the formulation of policies geared towards the advancement of small businesses [56]. Zhiqiang Geng and colleagues (2016) observed that the environment for industrial production operations has become significantly more complex than before and introduced a novel approach for fuzzy process capability analysis that integrates the kernel function with the Fuzzy Analytic Hierarchy Process [57]. Changiz Valmohammadi and colleagues (2021) outlined a strategy aimed at improving the performance, quality, and responsiveness to market demands of car manufacturers in Iran. They employed the Fuzzy Analytic Hierarchy Process to determine the importance and prioritization of these projects and utilized the Fuzzy Technique for Order Preference by Similarity to Ideal Solution (FTOPSIS), along with Fuzzy Gray Relational Analysis, to identify six viable projects to tackle the challenges identified [58]. In summary, the Fuzzy Analytic Hierarchy Process has been frequently used in recent years to analyze the weight and priority sequence of multiple decision criteria. Compared to the Analytic Hierarchy Process, it is more closely aligned with the intentions of experts [59,60,61,62]. Therefore, this study uses this method to analyze the weight of plans that have reached a consensus through the fuzzy Delphi technique and sequence them in order of priority, facilitating the scheduling of the technology roadmap.

2.6. Sustainable Development Goals

In 2015, the United Nations announced the “2030 Sustainable Development Goals” (SDGs), comprising 17 goals. Among them, Goal 8 (“Decent Work and Economic Growth”) and Goal 9 (“Industry, Innovation, and Infrastructure”) are closely related to the development process of Industry 4.0 [4,5]. In addition to enhancing industrial competitiveness, flexibility, efficiency, and quality through technological advancements, industrial manufacturing also emphasizes the sustainability of overall development to reach the 2030 Sustainable Development Goals [6]. Innovation is the foundation for increasing economic competitiveness and sustainable development. Under the concept of sustainability, product development, service method changes, and process transformations should be viewed as an integrated whole [63]. Maria Carmela Macrì and colleagues (2024) mentioned that working conditions and wages are core issues in the European technical industry environment, and the fairness of these working conditions is also included in the sustainability indicators [64]. Ester Lorente and colleagues (2024) proposed a study on the sustainability of a multimodal transport system combining ride-sharing services with public transportation networks [65]. Wenjun Li and colleagues (2024) proposed strategies to enhance SEG performance through digital transformation [66]. In summary, product innovation and infrastructure can change the working environment and enhance economic growth. Therefore, this study, based on the premise of sustainable development goals, explores the development trends of Programmable Logic Controllers in the transition from Industry 4.0 to Industry 5.0 to meet the needs of industrial development.

3. Methods

3.1. Expert Interviews

This research utilized semi-structured, one-off, and in-depth personal interviews with experts. To ensure a balanced collection of information, the interviews involved experts from the field of automation control and professors who teach related subjects in academia, with a total of 20 experts being invited. The interview process is shown in the Figure 3.

Figure 3. Expert interview process.

3.2. Fuzzy Delphi Method

This study references the concept of the fuzzy Delphi method with dual triangular fuzzy numbers proposed by scholars such as Jao-Hong Cheng (2001) and Jeng Tsang-Bin (2001) [50,51]. Dual triangular fuzzy numbers have the following advantages on the physical level: effectively handling uncertainty and fuzziness in data, simple structure and ease of use, and the robustness of the model. Practically, dual triangular fuzzy numbers offer high flexibility, can be applied in various fields, and have significant advantages in decision support, being easy to integrate and capable of reflecting the opinions of multiple experts. Based on these advantages, this study employed dual triangular fuzzy numbers for analysis, compiled information from expert interviews, and created questionnaires and distributed them online to the invited 20 experts and scholars for investigation, ensuring the validity and timely collection of the survey.

Step 1: Gather and analyze data from expert interviews, create frameworks, and design questionnaires based on evaluation metrics. Experts were asked to independently assign optimistic and conservative values to each metric on a scale of 1 to 10. The research aimed to invite 20 experts from the fields of automatic control and educational scholarship for the survey.

Step 2: Compile the statistics obtained from experts for each indicator in Step 1, including the calculation of the maximum, minimum, and geometric mean of the optimistic and conservative values.

Step 3: Define and establish a dual triangular fuzzy model. Optimistic Triangular Model (O): This model consolidates the opinions of various experts on a specific topic to provide the best possible quantitative values. The maximum quantitative value is taken as the maximum value, the minimum value is taken as the minimum value, and the geometric mean is taken as the middle value (minimum value O_L, geometric mean O_M, and maximum value O_U). Conservative Triangular Model (C): This model consolidates the opinions of various experts on a specific topic to provide the worst possible quantitative values. The maximum quantitative value is taken as the maximum value, the minimum value is taken as the minimum value, and the geometric mean is taken as the middle value (minimum value C_L, geometric mean C_M, and maximum value C_U). As shown in the Figure 4.

Figure 4. Optimistic and conservative triangular fuzzy models.

Step 4: Examine the model formed in the third phase, which can be categorized into three distinct cases.

Case 1: When the bimodal fuzzy model has no overlapping gray area ${(\mathrm{C}}_{\mathrm{U}}\le {\mathrm{O}}_{\mathrm{L}})$, it indicates that experts have reached a consensus on the trend of opinions for the evaluation indicator i. Therefore, the expert consensus position G_i for the evaluation indicator i can be represented by the arithmetic mean of the geometric means of the bimodal fuzzy model.

$${\mathrm{G}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{C}}_{\mathrm{M}}+{\mathrm{O}}_{\mathrm{M}}}{2}}$$

(1)

Case 2: When the bimodal fuzzy model overlaps ${(\mathrm{C}}_{\mathrm{U}}>{\mathrm{O}}_{\mathrm{L}})$ and there is a minimal overlapping gray zone ${{(\mathrm{Z}}_{\mathrm{i}}=\mathrm{C}}_{\mathrm{U}}-{\mathrm{O}}_{\mathrm{L}}>0$ and ${{\mathrm{M}}_{\mathrm{i}}=\mathrm{O}}_{\mathrm{M}}-{\mathrm{C}}_{\mathrm{M}}>{\mathrm{Z}}_{\mathrm{i}})$, it signifies that while there is no distinct consensus among experts on the evaluation indicator i, the views of the experts who provide extremely conservative maximum and optimistic minimum values do not differ greatly from the other experts. Thus, the consensus value G_i for the evaluation indicator i can be determined by applying an intersection operation on the fuzzy relationships within the gray zone of the bimodal fuzzy model, leading to the identification of a relevant set and the calculation of the quantified score based on the highest degree of membership in that set:

$${\mathrm{G}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{C}}_{\mathrm{U}}\times {\mathrm{O}}_{\mathrm{M}}-{\mathrm{O}}_{\mathrm{L}}\times {\mathrm{C}}_{\mathrm{M}}}{{\mathrm{C}}_{\mathrm{U}}+{\mathrm{O}}_{\mathrm{M}}-{\mathrm{O}}_{\mathrm{L}-}{\mathrm{C}}_{\mathrm{M}}}}$$

(2)

Case 3: When the bimodal fuzzy model has an intersection ${(\mathrm{C}}_{\mathrm{U}}>{\mathrm{O}}_{\mathrm{L}})$ and a larger overlapping gray area ${{(\mathrm{Z}}_{\mathrm{i}}=\mathrm{C}}_{\mathrm{U}}-{\mathrm{O}}_{\mathrm{L}}>0$ and ${{\mathrm{M}}_{\mathrm{i}}=\mathrm{O}}_{\mathrm{M}}-{\mathrm{C}}_{\mathrm{M}}<{\mathrm{Z}}_{\mathrm{i}})$ it indicates that there is a divergence in the opinions of experts. This is because the views of experts providing extremely conservative maximum values and optimistic minimum values differ significantly from those of other experts. A further round of questionnaires is necessary, and if the divergence remains significant, it suggests that there is no consensus on the evaluation indicator, and it is recommended that it is removed.

Step 5: Based on the model established in Step 4, calculate the G_i values for each evaluation indicator.

This research is focused on forecasting the trends of Programmable Logic Controllers and involves experts from related fields for both interviews and questionnaire surveys, ensuring the reliability of the data. The fuzzy Delphi method was employed for analysis and prediction, serving as an effective tool to interpret the views and agreement levels of experts regarding each assessment indicator, typically with settings between 6 and 7 [51,52], Therefore, in this study, we set the following criteria: the evaluation indicator Gi values must not be less than 6, and the test values must be positive. Indicators meeting these thresholds indicated a consensus among experts, while those not meeting the thresholds indicated a lack of consensus and were not included.

3.3. Fuzzy Analytic Hierarchy Process

In this research, the procedure for fuzzy hierarchical analysis proposed by Shiang-Lin Lin (2015) was adopted [62]. Utilizing the outcomes of the fuzzy Delphi method survey, trends that reached a consensus were compiled and incorporated into a fuzzy hierarchical analysis questionnaire. This process allowed for a detailed examination of the significance level of each trend, aiding in the development of a technology map in later stages.

Step 1: Following the aggregation of the fuzzy Delphi technique questionnaire to identify consensus-based evaluation items, a comparative questionnaire among these items was developed. This employed a left-to-right format for juxtaposing the items, with each one being rated on a 9-point scale to determine their level of importance, as illustrated in Table 3. For this part of this study, 20 experts in automatic control and educational fields, knowledgeable in the fuzzy Delphi technique, were invited to partake in this survey.

Table 3. Nine-point scale.

Semantics	Magnitude of Degree
Absolutely Essential	9
Absolutely Essential to Extremely Important	8
Extremely Important	7
Extremely Important to Important	6
Important	5
Important to Somewhat Important	4
Somewhat Important	3
Somewhat Important to Equally Important	2
Equally Important	1

Step 2: Compile questionnaires from various experts to establish a positive and negative value matrix A of n evaluation items and calculate the maximum eigenvalue (${\mathsf{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$) and eigenvector to determine the weights of each item. Saaty (1979) [52] mentioned that to ensure consistency before and after decision-making, it is important to examine the rationality of the comparison process among evaluation items. It is recommended that the consistency index (C.I.) is less than or equal to 0.1.

$$\mathrm{A}=\left[\begin{array}{ccccc}1& {\mathrm{a}}_{12}& .& .& {\mathrm{a}}_{1\mathrm{n}}\\ 1/{\mathrm{a}}_{12}& 1& ⋮& ⋮& ⋮\\ ⋮& .& 1& ⋮& ⋮\\ .& .& ⋮& 1& ⋮\\ 1/{\mathrm{a}}_{1\mathrm{n}}& .& .& .& 1\end{array}\right]$$

(3)

$$\mathrm{C}.\mathrm{I}.={\displaystyle \frac{{\mathsf{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}-\mathrm{n}}{\mathrm{n}-1}}$$

(4)

Furthermore, Saaty (1979) [52] pointed out that when the number of evaluation items in a study increases to more than three, leading to increased complexity, it becomes challenging to assess consistency. As a solution, he proposed the use of a random index (R.I.) to modify the variations in C.I. values caused by the varying number of evaluation items. This adjustment results in the consistency ratio (C.R.), for which he also recommended a value of no more than 0.1.

$$\mathrm{C}.\mathrm{R}.={\displaystyle \frac{\mathrm{C}.\mathrm{I}.}{\mathrm{R}.\mathrm{I}.}}$$

(5)

When both C.I. and C.R. are less than or equal to 0.1, it signifies that the consistency test has been passed.

Step 3: If the consistency test values are passed, calculate the strength of the j-th evaluation item by the i-th expert or scholar using Table 4 to perform triangular fuzzification (L_ij, M_ij, R_ij) for each evaluation item, where L represents the left boundary value, M represents the middle boundary value, and R represents the right boundary value, and establish the fuzzy positive and negative pairwise matrix FA.

$$\mathrm{FA}=\left[\begin{array}{ccccccccc}1& 1& 1& .& .& .& {\mathrm{a}}_{1\mathrm{n}\mathrm{L}}& {\mathrm{a}}_{1\mathrm{n}\mathrm{M}}& {\mathrm{a}}_{1\mathrm{n}\mathrm{R}}\\ ⋮& ⋮& ⋮& 1& 1& 1& ⋮& ⋮& ⋮\\ 1/{\mathrm{a}}_{1\mathrm{n}\mathrm{L}}& 1/{\mathrm{a}}_{1\mathrm{n}\mathrm{M}}& 1/{\mathrm{a}}_{1\mathrm{n}\mathrm{R}}& .& .& .& 1& 1& 1\end{array}\right]$$

(6)

Table 4. Triangular semantic fuzziness.

Semantics	Magnitude of Degree	Left (Lij, Mij, Rij)	Right (Lij, Mij, Rij)
Absolutely Essential	9	(8, 9, 9)	(1/9, 1/9, 1/8)
Absolutely Essential to Extremely Important	8	(7, 8, 9)	(1/9, 1/8, 1/7)
Extremely Important	7	(6, 7, 8)	(1/8, 1/7, 1/6)
Extremely Important to Important	6	(5, 6, 7)	(1/7, 1/6, 1/5)
Important	5	(4, 5, 6)	(1/6, 1/5, 1/4)
Important to Somewhat Important	4	(3, 4, 5)	(1/5, 1/4, 1/3)
Somewhat Important	3	(2, 3, 4)	(1/4, 1/3, 1/2)
Somewhat Important to Equally Important	2	(1, 2, 3)	(1/3, 1/2, 1,)
Equally Important	1	(1, 1, 1)	(1, 1, 1)

Determine the weights for the left boundary value (GL_ij), middle boundary value (GM_ij), and right boundary value (GR_ij) for each evaluation criterion by applying the geometric mean to the left, middle, and right boundary values and then individually normalize these weights as DL, DM, and DR, respectively.

$$\mathrm{D}\mathrm{L}={\displaystyle \frac{{\mathrm{G}\mathrm{L}}_{\mathrm{i}\mathrm{j}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{G}\mathrm{R}}_{\mathrm{i}\mathrm{j}}}}$$

(7)

$$\mathrm{D}\mathrm{M}={\displaystyle \frac{{\mathrm{G}\mathrm{M}}_{\mathrm{i}\mathrm{j}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{G}\mathrm{M}}_{\mathrm{i}\mathrm{j}}}}$$

(8)

$$\mathrm{D}\mathrm{R}={\displaystyle \frac{{\mathrm{G}\mathrm{R}}_{\mathrm{i}\mathrm{j}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{G}\mathrm{L}}_{\mathrm{i}\mathrm{j}}}}$$

(9)

Ultimately, the centroid of a triangle method is employed to solve the fuzzy calculations and determine the weight values (W_ij) of each evaluation item. To ensure that the sum of the weight values (W_ij) for all evaluation items equals 1, standard normalization is required to calculate the final weight values (DW_ij) of each evaluation item.

$${\mathrm{W}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{(\mathrm{D}\mathrm{M}-\mathrm{D}\mathrm{L})+(\mathrm{D}\mathrm{R}-\mathrm{D}\mathrm{L})}{3}}+\mathrm{D}\mathrm{L}$$

(10)

$${\mathrm{D}\mathrm{W}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{W}}_{\mathrm{i}\mathrm{j}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}\mathrm{j}}}}$$

(11)

Step 4: When multiple hierarchical levels are present, calculating the overall global weights requires chaining the weights together. For instance, if the weight of the N-th evaluation criterion in the first level is DW_N, the weight of the O-th criterion in the second level beneath the first one is DW_NO, and the weight of the P-th criterion in the third level under the second one is DW_NOP. Then, to determine the global weight DW_P of the P-th criterion in the third level, these weights must be sequentially combined.

$${\mathrm{D}\mathrm{W}}_{\mathrm{P}}={\mathrm{D}\mathrm{W}}_{\mathrm{N}}\times {\mathrm{D}\mathrm{W}}_{\mathrm{N}\mathrm{O}}\times {\mathrm{D}\mathrm{W}}_{\mathrm{N}\mathrm{O}\mathrm{P}}$$

(12)

3.4. Research Dimensions

This study proposes two principal frameworks: software and hardware. The software framework is further categorized into software interfaces and syntax, while the hardware is broken down into Power Module Units, Central Processing Units, Communication Modules, Memory Units, and Input/Output units. The order of future development will aim to gain a deeper understanding of the developmental priorities of each component within the software and hardware sectors. As shown in the Figure 5.

Figure 5. Research dimensions.

4. Discussion

4.1. Construction of Questionnaire and Evaluation Items

In this study, we invited 13 experts from the automation industry and 7 from the academic field, totaling 20 experts, for interviews and surveys. The basic information about the experts is shown in Table 5.

Table 5. Expert Profile Information.

No.	Education Level	Occupational Titles	Area	Age
Expert 1	Bachelor	Engineer	Automation	12
Expert 2	Master	Company Owner	Automation	22
Expert 3	Bachelor	Engineer	Automation	8
Expert 4	Bachelor	Company Owner	Automation	5
Expert 5	Bachelor	Engineer	Automation	5
Expert6	Bachelor	Engineer	Automation	4.5
Expert 7	Bachelor	Engineer	Automation	25
Expert 8	Master	General Manager	Automation	25
Expert 9	Master	Engineer	Automation	11
Expert 10	Master	Engineer	Automation	16
Expert 11	Bachelor	Engineer	Automation	6
Expert 12	Bachelor	Sales	Automation	7
Expert13	Bachelor	Engineer	Automation	1
Expert 14	Master	Teacher	Academic	12
Expert 15	Master	Teacher	Academic	12
Expert 16	Master	Assistant Trainer	Academic	1
Expert 17	Master	Teacher	Academic	17
Expert 18	Master	Teacher	Academic	10
Expert 19	Master	Teacher	Academic	17
Expert 20	Master	Teacher	Academic	20

After the first interview with 20 experts regarding the current situation and future trends of Programmable Logic Controllers (FX5U) and programming software (GX Works3), open-ended in-depth interviews were conducted with 7 of these experts. The key excerpts are as follows:

Expert 2: Significant advancements have been made in Mitsubishi’s Programmable Logic Controllers with the introduction of the FX5U model, marking a departure from previous versions. It is recommended to concentrate on the FX5U for detailed discussions. The Power Module and Input/Output units of this model are already mature and do not necessitate further debate. However, for IoT and AI capabilities, the controller must interact with a third-party server, and external network connectivity can enhance its usability. Therefore, it is vital to improve the functionalities of the Communication Module. As for the programming languages, there are no alternatives beyond the five standard languages already in use. Conversations with international engineers have revealed that Programmable Logic Controllers are less open and more challenging to integrate with other chip controllers, hindering their evolution.

Expert 7: Presently, the FX5U features one RS485 and one Ethernet port. Transitioning to a dual Ethernet port setup and standardizing the Communication Module would greatly simplify operations for electrical control engineers. Although the current program and expandable memory capacities are sufficient, increasing them or enabling user-installed upgrades would reduce limitations. In response to industrial advancements, the trend across various brands is to adopt ST (Structured Text) syntax for programming.

Expert 8: For the FX5U, the Communication Module and syntax warrant closer examination. While it is possible to link it to IoT or AI through third-party servers for other aspects, the current multitude and complexity of Communication Modules complicate integration efforts. The standardization of Ethernet for module expansion could substantially boost integration. The majority of the industry employs ST (Structured Text) syntax, which poses a significant barrier to entry for electrical control engineers due to its lack of resemblance to the multifaceted and compatible chip programming languages. It would be more beneficial for Programmable Logic Controllers to integrate other languages directly, ensuring compatibility, rather than through external connections.

Expert 10: The editing interface of GXWorks3 shows a trend towards visualization, but the current level of visualization is relatively low. The Communication Modules, including RS232, RS485, RTU/TCP, and Ethernet, among others, are quite complex. The FX5U model has one channel each for RS485 and Ethernet. According to client feedback, having dual Ethernet channels would be more advantageous for applications. Additionally, the instruction processing time is 34 ns, which is the current technological limit for processing speed. Surpassing this would be beneficial for industrial development.

Experts 17, 19, and 20: At the forefront of industrial wiring skills and craftsmanship competitions, Programmable Logic Controllers (PLCs) constitute a significant element. Apart from the programming itself, ensuring connectivity with other equipment is crucial. Currently equipped with one RS485 and one Ethernet channel, the Communication Modules could be more effective if upgraded to a dual Ethernet channel or enabled for WiFi connectivity, substantially improving system cohesiveness. While traditional programming languages like Ladder and SFC have been adequate, they show limitations when addressing the complexities of today’s industrial challenges, thus shifting towards ST syntax. However, this shift makes the learning process harder for those who are new to industrial wiring and PLCs. Introducing a sixth language could significantly boost programming efficiency.

After conducting in-depth interviews with 7 experts and consulting another 13 experts, the research dimensions proposed in this study were revised, as shown in Figure 6.

Figure 6. Research dimensions (revised).

In the context of IoT and AI, this study investigates trends in interface visualization and syntax for software interfaces, particularly the application and future trends of programming syntax in GX Works3, with suggestions for new programming languages to be added. On the hardware side, the focus is on trends and computational speeds for Central Processing Units, trends and channel specifics for Communication Modules, and trends, capacity, and expansion methods for Memory Units. This study excludes Power Supply and Input/Output modules due to their technological maturity, which eliminates the need for further discussion. A draft questionnaire was prepared following the interview insights, as detailed in Table 6.

Table 6. Fuzzy Delphi method (preliminary draft).

NO.	Dimensions	Items
1	Editing interface	In the context of evolving from Industry 4.0 to Industry 5.0, converting the Works3 interface to a visual format will facilitate more intuitive operations of the editing interface, including but not limited to component settings and definitions as well as network configurations.
2		In the context of evolving from Industry 4.0 to Industry 5.0, rendering the Works3 interface and external configurations in a visual format will enhance the clarity of various feature settings, including but not limited to component settings and definitions and network configurations.
3	Syntax	In the evolution of Industry 4.0 to Industry 5.0, where ST syntax is increasingly the preferred choice for editing software, the inclusion of a sixth programming language, in addition to the five mandated by IEC61131-3, could facilitate the progression and deployment of Programmable Logic Controllers. Potential candidates for this additional language are JAVA, LabVIEW, Python, Scratch, and more.
4	Central Processing Unit	The current instruction processing time of the FX5U is 34 ns. To meet future demands, this processing time needs to be further reduced, at least below 34 ns, to support the computational speeds required by current industrial development.
5	Memory Unit	At present, the memory of the FX5U can be expanded to 16 GB, but considering future industrial progress, this expansion limit of 16GB will be inadequate. A minimum of 16GB will be necessary for adequate usage.
6	Communication Module	To enhance the integration and application of Communication Modules in Programmable Logic Controllers, the adoption of Ethernet as the main interface, with the capability of dual channels, is suggested.
7		Enabling Programmable Logic Controllers to directly interface with external networks enhances their application from Industry 4.0 to Industry 5.0.
8		The integration of wireless network functionalities into Programmable Logic Controllers, acting as pathways for internal and external networks, significantly benefits their deployment in the progression of Industry 4.0 to Industry 5.0.

In a second round of consultations with 20 experts, discussions were centered on the initial draft of the fuzzy Delphi method questionnaire. The main points extracted are summarized below, with expansions, subdivisions, and amendments as shown in Table 7.

Table 7. Fuzzy Delphi method (final draft).

NO.	Dimensions	Items
1	Editing interface	Within the context of transitioning from Industry 4.0 to Industry 5.0, GX Works3 is now capable of designing IoT functionalities.
2		Within the context of evolving from Industry 4.0 to Industry 5.0, GX Works3 can now be linked with AI functionalities.
3		In the context of the transition from Industry 4.0 to Industry 5.0, GX Works3 has become visualized to some extent.
4		Within the context of moving from Industry 4.0 to Industry 5.0, Works3’s local/global tag settings have been made visual, facilitating a more intuitive operation of the editing interface and a clear indication of external connection statuses.
5		In the shift from Industry 4.0 to Industry 5.0, the merging of Works3’s local/global tag settings with module configuration diagrams into a visualized operation format will improve the editing interface’s intuitiveness and explicitly indicate the external connection conditions.
6		In the shift from Industry 4.0 to Industry 5.0, adopting a visual approach to configure the current connection target interface will improve the editing interface’s intuitiveness and explicitly display the external connection conditions.
7	Syntax	The ability to use the five programming languages outlined in IEC61131-3 in an intercompatible manner boosts programming efficiency.
8		GX Works3 has achieved compatibility among the five programming languages outlined in the IEC61131-3 standard.
9		In the transition from Industry 4.0 to Industry 5.0, the five programming languages specified by IEC61131-3 are already sufficient for developing Internet of Things (IoT) and AI-related applications.
10		Currently, programming software syntax is predominantly moving towards Structured Text (ST) as the mainstream, and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—JAVA could be beneficial for the development and application of Programmable Logic Controllers.
11		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—Scratch—could be beneficial for the development and application of Programmable Logic Controllers.
12		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—Python—could be beneficial for the development and application of Programmable Logic Controllers.
13		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—LabVIEW—could be beneficial for the development and application of Programmable Logic Controllers.
14	Central Processing Unit	The instruction processing time of the FX5U (34 ns per basic or application instruction) is still sufficient to handle the demands of the development of Industry 4.0 and Industry 5.0.
15		The instruction processing speed of the FX5U has reached its current limit.
16		If the instruction processing time of the FX5U could be reduced to below 34 ns, it would be more advantageous for the current stage of industrial development.
17	Memory Unit	According to the official FX5U user manual, the program capacity is 64 k steps (128 k bytes), which can be expanded to 128 k steps (256 k bytes) through software adjustments. This capacity is still adequate for current industrial development needs.
18		According to the official FX5U user manual, the program capacity is initially set to 64 k steps (128 k bytes), with an option to extend it to 128 k steps (256 k bytes) through software adjustments. Enhancing the capacity further to 256 k steps (512 k bytes) or even to 512 k steps (1024 k bytes) and beyond would be advantageous for industrial development.
19		According to the official FX5U user manual, the program capacity is 64 k steps (128 k bytes), with an option to expand it to 128 k steps (256 k bytes) through software adjustments. If the program capacity could be further customized by installing an M.2 specification SSD, it would benefit industrial development.
20		Considering future industrial development, the current expansion capacity of 16 G via an external SD card is still sufficient for use.
21		With future industrial development in mind, the present memory expansion capability of 16 G with an additional external SD card is restricted. A capacity of at least 32 G would better serve industrial progress.
22	Communication Module	As we transition from Industry 4.0 to Industry 5.0, the diversity of communication protocols in Programmable Logic Controllers complicates their integration and usage.
23		The FX5U comes equipped with two types of built-in Communication Modules: RS-485 and Ethernet, each with one channel, which are still capable of supporting the needs of the transition from Industry 4.0 to Industry 5.0.
24		The inability of Programmable Logic Controllers’ (PLCs) Communication Modules to directly connect to external networks limits their development with regard to the Internet of Things (IoT) or AI technologies.
25		For the Communication Modules of Programmable Logic Controllers to be integrated and used more efficiently, it is advisable to focus on Ethernet as the main interface and ensure that it has dual channels.
26		Programmable Logic Controllers being able to directly connect to external networks will facilitate more effective utilization in the transition from Industry 4.0 to Industry 5.0.
27		Adding wireless networking capabilities to Programmable Logic Controllers as a conduit for both internal and external networks will enhance their effectiveness in the advancement from Industry 4.0 to Industry 5.0.

• The current situation of each evaluation item can be included for our understanding.
• For the language component, breaking down the listed four languages into more detailed categories will help determine which one is better suited.
• In addition to augmenting memory capacity, incorporating program storage capacity will render a more holistic overview.

4.2. Analysis and Statistical Treatment of Questionnaires Based on the Fuzzy Delphi Method

In this research, we conducted interviews with experts to produce a report on the status quo and trends concerning Programmable Logic Controllers and the GX Works 3 Ver.1 software interface and syntax, gathering expert consensus levels on different assessment items, which are compiled in Table 8.

Table 8. Statistical analysis by the fuzzy Delphi method.

No.	Dimensions	Items	Mi − Zi	Gi
1	Editing interface	Within the context of transitioning from Industry 4.0 to Industry 5.0, GX Works3 is now capable of designing IoT functionalities.	4.735738195	6.292731183
2		Within the context of evolving from Industry 4.0 to Industry 5.0, GX Works3 can now be linked with AI functionalities.	1.97475615	5.741630396
3		In the context of the transition from Industry 4.0 to Industry 5.0, GX Works3 has become visualized to some extent.	3.106189417	6.416851659
4		Within the context of evolving from Industry 4.0 to Industry 5.0, Works3’s local/global tag settings have been made visual, facilitating a more intuitive operation of the editing interface and a clear indication of external connection statuses.	3.904208183	6.485504145
5		Within the shift from Industry 4.0 to Industry 5.0, the merging of Works3’s local/global tag settings with module configuration diagrams into a visualized operation format will improve the editing interface’s intuitiveness and explicitly indicate the external connection conditions.	3.377671789	6.660339948
6		Within the shift from Industry 4.0 to Industry 5.0, adopting a visual approach to configure the current connection target interface will improve the editing interface’s intuitiveness and explicitly display the external connection conditions.	0.345067163	6.608379288
7	Syntax	The ability to use the five programming languages outlined in IEC61131-3 in an intercompatible manner boosts programming efficiency.	3.897292402	7.28009312
8		GX Works3 has achieved compatibility among the five programming languages outlined in the IEC61131-3 standard.	−0.720744393	−1
9		Within the context of transitioning from Industry 4.0 to Industry 5.0, the five programming languages specified by IEC61131-3 are already sufficient for developing Internet of Things (IoT) and AI-related applications.	3.338869803	6.635872918
10		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—JAVA—could be beneficial for the development and application of Programmable Logic Controllers.	2.814563418	7.386339754
11		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—Scratch—could be beneficial for the development and application of Programmable Logic Controllers.	0.416741102	6.651351122
12		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—Python—could be beneficial for the development and application of Programmable Logic Controllers.	1.659331811	6.309990766
13		Currently, programming software syntax is predominantly moving towards Structured Text (ST), and in the context of the transition from Industry 4.0 to Industry 5.0, in addition to the five programming languages specified by IEC61131-3, adding a sixth language—LabVIEW—could be beneficial for the development and application of Programmable Logic Controllers.	4.066785535	6.587055934
14	Central Processing Unit	The instruction processing time of the FX5U (34 ns per basic or application instruction) is still sufficient to handle the demands of the development of Industry 4.0 and Industry 5.0.	0.409215202	6.062800936
15		The instruction processing speed of the FX5U has reached its current limit.	0.503762727	5.70916773
16		If the instruction processing time of the FX5U could be reduced to below 34 ns, it would be more advantageous for the current stage of industrial development.	0.635850947	6.575920192
17	Memory Unit	According to the official FX5U user manual, the program capacity is 64 k steps (128 k bytes), which can be expanded to 128 k steps (256 k bytes) through software adjustments. This capacity is still adequate for current industrial development needs.	1.094935386	6.944851383
18		According to the official FX5U user manual, the program capacity is initially set to 64 k steps (128 k bytes), with an option to extend it to 128 k steps (256 k bytes) through software adjustments. Enhancing the capacity further to 256 k steps (512 k bytes) or even to 512 k steps (1024 k bytes) and beyond would be advantageous for industrial development.	2.881839877	6.61704979
19		According to the official FX5U user manual, the program capacity is 64 k steps (128 k bytes), with an option to expand it to 128 k steps (256 k bytes) through software adjustments. If the program capacity could be further customized by installing an M.2 specification SSD, it would benefit industrial development.	1.526296649	6.633019294
20		Considering future industrial developments, the current expansion capacity of 16 G via an external SD card is still sufficient for use.	2.408071964	6.407809115
21		With future industrial development in mind, the present memory expansion capability of 16 G with an additional external SD card is restricted. A capacity of at least 32 G would better serve industrial progress.	0.10419119	6.562052711
22	Communication Module	As we transition from Industry 4.0 to Industry 5.0, the diversity of communication protocols in Programmable Logic Controllers complicates their integration and usage.	1.446388291	6.109844903
23		The FX5U comes equipped with two types of built-in Communication Modules: RS-485 and Ethernet, each with one channel, which are still capable of supporting the needs of the transition from Industry 4.0 to Industry 5.0.	1.301260626	5.78391427
24		The inability of Programmable Logic Controllers’ (PLCs) Communication Modules to directly connect to external networks limits their development with regard to the Internet of Things (IoT) or AI technologies.	2.510772534	7.353863624
25		For the Communication Modules of Programmable Logic Controllers to be integrated and used more efficiently, it is advisable to focus on Ethernet as the main interface and ensure that it has dual channels.	2.143723647	8.209240923
26		Programmable Logic Controllers being able to directly connect to external networks will facilitate more effective utilization in the transition from Industry 4.0 to Industry 5.0.	3.206562981	7.410600633
27		Adding wireless networking capabilities to Programmable Logic Controllers as a conduit for both internal and external networks will enhance their effectiveness in the advancement from Industry 4.0 to Industry 5.0.	4.735738195	6.292731183

According to the research methods and standards set in the previous section, the test values must be positive, and consensus among experts is considered to be achieved if the consensus value reaches 6. From the table above, it can be seen that the expert consensus values for evaluation items 2, 8, 15, and 23 did not reach 6, indicating that consensus among experts on these criteria was not achieved. For the remaining evaluation items, the expert consensus values all exceeded 6, and the test values were positive, indicating that consensus was reached.

Items 1 to 6 in the evaluation were designed to assess the software interface’s present conditions and prospects. From the analysis conducted, it was determined that, within the present phase of industrial growth, the design of IoT solutions using GX Works3 is achievable. However, a connection with AI features is not yet operational. The interface was partially visualized, and visual-based operations like setting regional or global tags and specifying connection goals could significantly improve it.

Items 7 to 13 were designed to gather insights into the programming syntax of Programmable Logic Controllers, focusing on their present conditions and future directions. This study found that the compatible use of the five syntaxes defined by the IEC61131-3 standard could significantly improve programming efficiency. Currently, programming for IoT and AI integration is feasible. Yet, GX Works 3 Ver.1 does not support the interoperable use of these IEC61131-3 standard syntaxes. Additionally, the introduction of a sixth programming language is considered advantageous for PLC development and usage, including options like JAVA, Scratch, Python, LabVIEW, and more.

Items 13 to 15 were designed to understand the status quo and future directions of CPUs. It was determined from the research that the FX5U operates with a computational time of 34 ns, which meets the demands of the present stage of industrial evolution; yet, the computational time could be shortened further.

For items 18 to 21, which are focused on assessing the memory unit’s present conditions and future directions, it was found that the FX5U’s programming capacity meets the current needs, and its data storage capacity can currently be expanded up to 16 GB. Given the rising complexity of system software and the demands for managing large datasets, enhancing the programming capacity by either doubling the existing capacity or enabling users to expand it by purchasing additional hard disks would significantly improve utilization.

For items 22 to 27, which are focused on the status quo and future directions of communication units, it was discovered that the diverse communication protocols of Programmable Logic Controllers complicate their integration. At this stage, the Communication Modules in the FX5U are insufficiently fast for industrial applications. Switching to dual Ethernet channels could greatly improve integration and usage. Moreover, since PLCs act as the initial interface with the physical layer but cannot directly link to external networks without a third-party server, their potential for IoT and AI applications is restricted. Enabling direct external network connections and adding wireless networking capabilities would markedly enhance their usability.

4.3. Analysis and Statistical Treatment of Questionnaires Based on the Fuzzy Hierarchical Analysis

All units of the Programmable Logic Controller (PLC) are important, but due to the limitations of time, technology, and resources, there is an inevitable priority in development trends. Therefore, this study uses the Fuzzy Analytic Hierarchy Process to understand the importance of the evaluation items that reached a consensus in the previous section and ranks them according to the size of their weights, as compiled in Table 9.

Table 9. Statistical analysis by the fuzzy hierarchical analysis.

Items	DW	NO.
The FX5U Communication Module is modified to primarily use Ethernet with dual channels.	0.119	1
The FX5U directly connects to external networks and adds a wireless network as the channel for both internal and external networks.	0.104	2
The FX5U’s command processing time has been decreased to below 34 ns.	0.091	3
Python as the sixth syntax option for GX Works3.	0.091	4
Visualization of local/global tag settings and configuration of the current connection target in GX Works3.	0.083	5
The program capacity of the FX5U can be increased to more than double its original capacity.	0.08	6
The program capacity of the FX5U can be altered based on user needs by installing an M.2 specification SSD.	0.069	7
LabVIEW as the sixth syntax option for GX Works3.	0.068	8
The memory capacity of the FX5U can be increased to more than double its original capacity.	0.061	9
JAVA as the sixth syntax option for GX Works3.	0.058	10
Scratch as the sixth syntax option for GX Works3.	0.054	11

C.I. = 0.081, C.R. = 0.053.

Based on the research methods and design proposed in the previous section, the consistency index (C.I.) and consistency ratio (C.R.) of the Fuzzy Analytic Hierarchy Process questionnaire must both be less than or equal to 0.1 for the questionnaire results to be considered valid and reasonable. The analysis of the questionnaire revealed a C.I. value of 0.081 and a C.R. value of 0.053, both of which are below 0.1, thereby passing the consistency test.

From Table 9, it is clear that the suggested order of future development trends for the FX5U and GX Works 3 Ver.1prioritizes upgrading Communication Modules to dual Ethernet channels and enabling external network connections, followed by enhancing wireless network capabilities. Subsequently, increasing the Central Processing Unit’s (CPU) computational speed, incorporating Python syntax, and improving the editing interface for local/global tag settings and the visualization of current connection targets are of secondary importance. The final priorities include integrating LabVIEW, expanding memory capacity, and adding JAVA and Scratch.

4.4. Technology Roadmap

Based on the outcomes of the fuzzy Delphi method and Fuzzy Analytic Hierarchy Process, this research has developed a technology and trend development map for Programmable Logic Controllers, presented in Figure 7.

Figure 7. Technology roadmap of Programmable Logic Controllers.

Figure 7 shows the present and future viable trends for the FX5U and GX Works3 Programmable Logic Controllers.

Their importance is determined through the consolidation of expert and scholar opinions and ranked according to their weight. Due to the limitations of this study, it is not possible to access more confidential information from manufacturers. Therefore, by considering multiple factors such as the market and resources, in conjunction with their importance, the most suitable planning scheme for the manufacturer itself can be obtained.

Thus, with manufacturing costs in mind, we recommend first focusing on enhancements that increase overall efficiency in the next-generation models, such as the visualization of the editing interface and dual Ethernet channels. Following this, a gradual shift towards reducing the time it takes to process instructions, enhancing program and memory capacities, and extending connectivity through direct external network links and extra wireless network channels is recommended. The inclusion of a sixth programming language can be pursued concurrently.

With time efficiency in mind, we recommend prioritizing the development of dual Ethernet channels and the reduction of instruction processing times in the next-generation models. Following this, there should be a gradual move towards enhancing program and memory capacity. Concurrently, efforts towards visualizing the editing interface, adding a sixth programming language, and extending direct external network connections and wireless network capabilities should also be planned.

5. Conclusions

Programmable Logic Controllers (PLCs) are an indispensable part of current and future industrial development, especially in smart factories, smart home technology, automated production lines, and even mechanical manufacturing machines. This study aims to present the trend map of PLC hardware and software development through a technical map and propose overall development trend suggestions for PLCs, providing industry players with references for sustainable development, enhancing market competitiveness, research, and development, in line with the needs of industrial sustainable development.

Through the application of the Delphi method and Fuzzy Analytic Hierarchy Process, the calculated weights indicate that the most important priorities include converting Communication Module channels to dual Ethernet channels, enabling external network connections, and adding wireless networks. The next priority is increasing the processing speed of the central processor, adding Python syntax, setting regional/global tags in the editing interface, visualizing the current connection target, and increasing the program capacity. Finally, LabVIEW, memory expansion, JAVA, and Scratch are considered.

The final technical map suggests that from a manufacturing cost perspective, it is recommended to first plan the next-generation models, starting with improvements that can enhance overall efficiency, such as visualizing the editing interface and dual Ethernet channels. Gradually, efforts should be made towards reducing instruction processing time, increasing program and memory capacity, and adding direct external network connections and wireless network channels. From a time cost perspective, it is recommended to first plan the next-generation models, starting with dual Ethernet channels and reducing instruction processing time, and then gradually work towards increasing the program and memory capacity, visualizing the editing interface, and planning for external network connections and wireless network channels. Since adding the sixth programming language takes longer, it is suggested to handle this simultaneously during the model development process.

These recommendations are provided for reference to relevant industry players in their development and research. Only by continuously innovating products, evaluating social and environmental needs, enhancing product competitiveness, and providing suitable work and economic growth while fulfilling social responsibilities can this study contribute to sustainable development.