Mapping the Dynamics Behind Breakthrough Innovations in China’s Energy Sector: The Evolution of Research Foci and Collaborative Networks

Abstract

The energy sector profoundly influences both economic growth and ecosystems, making it pivotal to human development. Substantial evidence confirms that scientific and technological innovations, particularly breakthrough achievements, are key drivers of sustainable development in this sector. Consequently, comprehending the evolutionary trajectory of such breakthroughs is crucial for policymakers and practitioners to refine their strategic approaches. To analyze the evolution of energy-related breakthrough innovations, this study leverages a dataset of 552 projects awarded China’s State Science and Technology Advancement Prize. By employing large language models and word cloud analysis, we trace the shifting research priorities of the 552 projects to delineate the pathway of scientific and technological development in the energy sector. Furthermore, we utilize social network analysis to reveal the evolving collaboration patterns underlying these innovations. Our findings indicate that, consistent with China’s current energy consumption structure, most innovations remain concentrated in fossil fuels. However, a clear trend emerges: research is focusing less on fossil fuels and more on clean energy and high efficiency equipment. Regarding collaborative innovation, the pattern in China differs from that of countries like the United States; in China, universities, rather than enterprises, occupy the central position in innovation networks. The insights from this study can assist government officials in designing effective policies to support breakthrough innovations across different types of research entities. Moreover, research entities can adjust their collaboration strategies based on our findings to enhance innovation performance.

1. Introduction

Energy is an essential resource for human survival and economic development; thus, the energy sector has a significant impact on society [1]. Following the industrial and information revolutions, increasing energy consumption has become a challenge to sustainable development worldwide, resulting in multiple environmental problems including air pollution, solid waste pollution, and global warming [2,3]. To reduce the adverse impact of energy consumption, novel technologies—including advanced fossil fuel extraction and clean utilization, renewable and clean energy, smart energy systems, and carbon neutral technology—have been adopted to reshape the energy system [1,4,5,6]. Substantial evidence confirms that innovations in science and technology, especially breakthrough innovations, have played a significant role in fulfilling the sustainable development goal of the energy sector [7,8,9]. For example, ultra-high-voltage power transmission technology has enabled China to efficiently transmit electricity from power-rich regions to those with shortages, altering the geographical distribution of power generation and easing the environmental burden caused by thermal power [10].

Breakthrough innovation refers to fundamentally novel advancements that disrupt existing social or economic systems, with a significant impact on the demand or supply side of the industry [11]. Given the leading role of breakthrough innovations in driving the energy sector towards sustainable development, it is crucial to reveal the basic mechanisms behind them [12]. Since the evolutionary pathway of energy-related breakthrough innovations can indicate the future trajectory of global energy industrial transition, exploring the evolutionary characteristics of these innovations in large energy-consuming countries can help policymakers make appropriate adjustments to the global energy system [13]. Studies examined the evolutionary pathway of energy-related innovations from the perspectives of R&D investment [14], supply chain reconstruction [15], spatial and temporal heterogeneity [16], and energy policy [17]. Most of these studies are based on economic data, publications, or patents. However, very few studies focus on breakthrough innovations in the energy sector. Economic data reflect the input and output of innovation but do not directly map the pathway of breakthrough innovation. Patent (or publication) -based studies often treat innovation outputs as homogenous, failing to discriminate between incremental improvements and true breakthroughs within a large dataset. An abundance of low-quality patents (or publications) may dilute the visibility of genuine breakthrough innovations. Compared with incremental improvements, breakthroughs exert a deeper impact on human development and scientific progress [18]. For instance, a breakthrough in controlled nuclear fusion technology would profoundly transform humanity’s reliance on fossil fuels [19]. Therefore, these genuine breakthrough innovations deserve further investigation.

To bridge this research gap, this study utilizes a dataset of the State Science and Technology Advancement Prize (SSTAP)—representing China’s most pivotal breakthrough achievements—to trace the evolutionary trajectory of innovations within China’s energy sector. As one of the largest energy-consuming countries, China has become a key player in global energy-related innovations, especially breakthrough innovations [20]. China’s public and private sector investment in energy-related R&D occupies a large proportion of the total global investment [5]. Therefore, China’s innovation can, to some extent, reflect the global level of innovation. China’s experience in promoting energy-related breakthrough innovations is valuable for countries facing similar environmental issues.

In this study, a hybrid research framework is designed to capture the evolutionary trajectory of breakthrough innovations in China’s energy sector (BICES). Large language model (LLM) and text analysis technologies are employed to trace the evolving research foci of BICES. Understanding the shifting research priorities of breakthrough innovations enables us to map the trend of science and technology development in China’s energy sector. Furthermore, using social network analysis (SNA), this study investigates the collaboration networks (CNs) of the 552 SSTAP projects. Given the multi-stakeholder nature of the energy value chain and the complexity of the energy system, breakthrough innovation typically involves inter-organizational collaboration across enterprises, universities, research institutes, and government [21]. It is increasingly challenging for organizations to achieve breakthrough innovations using only their limited resources and expertise [22]. Therefore, inter-organizational collaboration is an important characteristic of breakthrough innovations. By examining the CNs of SSTAP projects, this study reveals the hidden mechanisms of collaboration patterns underlying breakthrough innovations.

This study contributes to innovation management scholarship in the energy sector. Unlike prior studies that relied on patents or publications, we leverage a unique dataset from SSTAP to focus specifically on breakthrough innovations in China’s energy sector. Our findings offer policymakers in-depth insights for steering the energy system transition. Furthermore, by employing the dual lenses of research focus and collaboration to trace the evolution of these innovations, we introduce a valuable organizational perspective to the analysis. A key finding is that China’s innovation collaboration model differs from that of the United States, with universities playing a more central role than enterprises. This finding enriches the understanding of cross-organizational collaboration dynamics across diverse institutional contexts. Finally, we provide suggestions for research organizations to optimize their collaboration patterns to enhance innovation efficiency, as well as for government officials to align energy policies with future science and technology trends.

2. Literature Review

2.1. Evolutionary Trajectory of Innovations in the Energy Field

As global competition intensifies, continuous innovation has become a fundamental requirement for organizations and nations to maintain their competitive edge in a rapidly evolving landscape [23]. Innovation refers to the iterative process of translating ideas into practice through design, testing, application, and refinement [11]. This process enables the development of novel technologies, products, services, and/or business models that support sustainable development [1,24]. A growing body of evidence confirms that innovation—particularly breakthrough innovation—plays a decisive role in shaping human society. As a critical component of societal infrastructure, the energy sector is profoundly influenced by innovation [25]. To a large extent, progress in the energy industry can be understood as a process of technological evolution that continually expands the types of energy available for human use, enhances energy efficiency, and reduces both costs and negative environmental impacts [8,26]. Therefore, understanding the evolution of energy-related innovations is crucial for policymakers and practitioners to refine future market strategies and policy frameworks.

A growing body of scholarship has sought to uncover the underlying patterns of innovation evolution using diverse methodological approaches. Plakitkin et al. [27] employed neural network analysis to trace transformation pathways in global energy systems, encompassing demographic, technological, energy, transport, and communication dimensions. Their research posited that future energy innovation would increasingly concentrate on outer space exploration, robotics development, and artificial intelligence capabilities. Through a systematic analysis of 694 academic publications and 122 patents, Hernández-Alvarado et al. [28] document the technological evolution of renewable energy innovations, noting a remarkable surge in research on biocomposite research that signals this innovation’s growing importance for industry practitioners. Ullah et al. [29] conducted a comprehensive literature analysis mapping developmental trajectories of energy harvesting technologies, revealing the significant potential of low-power energy harvesting systems to deliver sustainable and reliable energy solutions for various electronic applications. In a bibliometric study of 495 articles from Web of Science, Al-Awamleh et al. [30] tracked knowledge development and technological change in energy management, identifying social demand as a primary driver of innovation pathways. From an R&D investment perspective, Zhang et al. [14] analyzed the evolution of global public energy innovation, demonstrating a strategic shift in research focus from fossil-based to low-carbon technologies. Meanwhile, Lee and Lee [17] leveraged patent data to construct six patent maps tracing innovation pathways in the United States, with their analysis underscoring the critical role of effective collaboration in driving innovation.

While the aforementioned studies have substantially advanced our understanding of innovation evolution in the energy sector, few studies have specifically examined breakthrough innovations, which are distinguished by their transformative impact across industries and disciplines. Given that innovation quality surpasses mere quantitative measures [31], breakthrough innovations are poised to play a pivotal role in steering the energy sector toward sustainable development. Moreover, breakthrough innovations differ fundamentally from incremental innovations in their underlying characteristics [32]. For instance, while a clear correlation exists between R&D investment and innovation output, the relationship between R&D spending and breakthrough innovation is notably non-linear [32]. Thus, dedicated research is warranted to investigate the evolutionary trajectory of breakthrough innovations within the energy domain.

2.2. Role of Collaboration in Innovation

Due to the complexity of the energy industry, energy-related innovation typically requires the integration of multidisciplinary knowledge, including expertise in mining, electricity, advanced materials, engineering, environmental science, and system design [7]. Given this breadth of necessary knowledge, organizations often find it difficult to achieve breakthrough innovation when they rely solely on their own resources and capabilities. Thus, as Anadon and Holdren [7] observed, partnerships among universities, research institutes, industry, and governments play a crucial role in fostering such innovation. Drawing on social capital theory and the resource-based view, Lin and Zhu [33] further argue that effective collaboration enhances knowledge-sharing efficiency and mitigates risks in the process of technological innovation. Moreover, substantial evidence indicates that organizations engaged in extensive collaboration tend to innovate more frequently and generate more original outcomes than those that do not [34].

Consequently, scholars in the energy field have conducted extensive studies on collaboration patterns in energy-related innovation. Using spatial Durbin and Super-SBM models, Ren et al. [35] investigated how urban collaboration networks shape the innovation efficiency of new energy vehicles, revealing a positive relationship between network size and innovation performance. In a complementary study, Ba et al. [36] identified an inverted U-shaped effect of geographical distance among collaborators on innovation outcomes. Leveraging a dataset of 14,506 observations from 2007 to 2018, Uyar et al. [37] employed a country–industry–year fixed-effects regression and confirmed that government–enterprise collaboration significantly facilitates energy innovation. Separately, applying the Complex Product System innovation management framework, Badi and Pryke [38] evaluated collaboration quality in Sustainable Energy Innovation within Private Finance Initiative projects, emphasizing that effective stakeholder communication critically shapes innovation performance.

Although prior studies have explored collaboration patterns in innovation, identified factors influencing collaboration, and assessed collaboration’s impact on innovation outcomes, few studies have specifically investigated the dynamic evolution of CNs in the context of breakthrough innovation. Due to the temporally extended nature of breakthrough innovations, their CNs are inherently dynamic, marked by the continuous formation and dissolution of ties among diverse organizations. Further research is needed to advance our understanding of how these networks evolve in the context of energy-related breakthrough innovations.

3. Materials and Methods

3.1. Data Collection

Previous studies have predominantly relied on publications and patents as proxies for mapping research frontiers and collaborative patterns in science and technology innovations. However, due to the long-term nature of breakthrough innovations, publications and patents fall short of capturing innovation progress in its entirety. Additionally, the lack of effective measures for evaluating the value of publications and patents presents a significant challenge to identifying breakthrough innovations from vast volumes of research works. Therefore, data related to publications and patents are not used in this study. Instead, we employ a dataset of SSTAP, representing China’s most pivotal breakthrough innovations, to trace the evolutionary trajectory of research and collaboration within China’s energy sector.

SSTAP is one of the highest honors conferred by the State Council of China, recognizing individuals and organizations that have made exceptional contributions to scientific and technological progress [39,40]. Established in 1999, three SSTAP prizes—Special, First, and Second Prize—were awarded annually from 2000 to 2020. Since 2021, the award cycle has been adjusted to biennial intervals, with modifications to both the number of awards and evaluation criteria implemented by the central government of China. Affected by this reform, SSTAP prizes have only been presented once since 2021, which was in 2023. Due to structural differences in the award system before and after 2021, the 2023 SSTAP projects were excluded from this study.

We meticulously examine 3438 SSTAP projects awarded between 2000 and 2020, among which 552 are identified as energy-related innovations. We first collect the list of SSTAP projects from the official website of China’s Ministry of Science and Technology. Based on official project descriptions and categorizations, we then identify 552 BICES projects (energy-related SSTAP projects) from the SSTAP list through a comprehensive review. Each selected project must demonstrate a clear connection to energy extraction, transportation, utilization, or associated pollution challenges. Relevant information for these projects—including project titles, award years, award levels, key participants, participating organizations, brief introductions, and research fields—is compiled into a structured dataset.

3.2. Research Focus Trajectory Analysis

First of all, descriptive statistical analysis is performed to illustrate the temporal distribution of SSTAP projects across all fields and specifically within the energy sector. Furthermore, we utilize LLM and text analysis methods to track evolving research priorities in the 552 BICES projects. Specifically, we employ DeepSeek (V3.1), an LLM developed by Chinese researchers, to extract research themes from each BICES project as descriptive tags (e.g., Electricity, Coal, Clean Manufacturing, Waste Treatment). Basic information on the 552 projects is compiled into a vector database integrated with the LLM. The framework of the instruction engineering for the LLM is illustrated in the Supplementary Material. In line with the research design of Eloundou et al. [41], two members of our research team independently review the assigned tags to ensure they accurately reflect the thematic focus of each BICES project.

Next, by analyzing the frequency and annual distribution of tags, we delineate the evolution of research foci in BICES. To clarify the development trajectory of energy innovations, we apply visualization methods—specifically, VOSviewer v1.6.20 (Visualization of Similarities) and word clouds—to make a fine-grained examination of the awarded projects. VOSviewer, a tool for constructing and visualizing bibliometric networks, is capable of processing large datasets and generating insightful maps such as co-citation and co-occurrence networks. It is particularly valued for its ability to produce rich, interpretable visual representations [42]. Moreover, word clouds offer a visual summary of prominent terms, where the size of each word corresponds to its frequency. In this study, word clouds help illustrate prominent application trends in BICES, providing an accessible overview that supports further thematic analysis.

3.3. CN Analysis

To examine the evolution of collaboration patterns underlying breakthrough innovations, CNs of BICES projects are constructed and analyzed using SNA. The CN in this study consists of nodes representing awarded organizations and edges representing collaborative ties between them. To support SNA-based statistical examination, 1288 participating organizations are systematically coded. By referencing project information, we identify the role of each participant, allowing for classification into four organizational types: university, research institute, enterprise, and government agency. To ensure accuracy in the network analysis, all organization names are standardized to their most recent official designations (e.g., “Huainan Institute of Technology” is updated to “Anhui University of Technology”). Additionally, subsidiary institutions are merged into their parent organizations to avoid fragmentation (e.g., “Xi’an Jiaotong University Institute of Power & Electronics” is consolidated under “Xi’an Jiaotong University”).

The raw data of BICES projects do not include information on the strength of inter-organizational relationships. Therefore, each collaboration (i.e., co-authorship in one project) is assumed to carry equal weight, and an undirected network is used. Unlike previous studies that examine network states at discrete time intervals, this study adopts a more comprehensive approach, in which an overview of the collaborative network spanning two decades is followed by a detailed year-by-year analysis to trace the evolution of innovation collaborations. The overall network structure reveals organizational linkages and growth mechanisms, while annual analysis of collaboration networks (CNs)—from the perspective of participant positions and ties—helps identify organizations that play dominant roles in BICES. In line with the methodology of Han et al. [40], key network metrics—including network density, network centrality, and node-level relationship strength—are applied to characterize the collaborative properties of each CN. Ucinet v6.415, a specialized software for social network analysis, is used to map the CNs from 2000 to 2020, generating discrete visualizations [43].

4. Results

4.1. Descriptive Statistics

Based on annual SSTAP data, Figure 1 presents the temporal distribution of awards across all fields and specifically for BICES. The number of BICES awards shows fluctuations that closely mirror the overall trend in SSTAP awards. A noticeable and consistent upward trend in BICES awards emerged around 2006 and continued for approximately five years. This increase can be largely attributed to a series of environmental policies implemented by the Chinese government in 2006, such as the Renewable Energy Law of the People’s Republic of China, as well as stricter emission standards—including those targeting air pollutants like SO₂ [44]. These regulations show the growing importance of environmental protection among both policymakers and the public during this period. The heightened awareness directly stimulated innovation in cleaner energy and accelerated the development of renewable energy technologies throughout China.

Figure 2 shows the annual number of BICES awards, and the distribution of Special, First, and Second Prizes as a proportion of the annual BICES total. Over the 21-year period from 2000 to 2020, BICES awards consistently accounted for 12% to 20% of all SSTAP awards, which span all fields of science and technology. Notably, four projects critical to China’s energy system construction were awarded the prestigious Special Prize, all of which involved extensive collaboration among multiple institutions and researchers. These four projects are: (1) Efficient Exploration and Development Technologies for Sustained Stable Production of Over 40 Million Tons at the High Water-Cut Stage of the Daqing Oilfield (2010); (2) Safe and Efficient Development Technology and Industrial Application of a Giant, Ultra-Deep, High-Sulfur Gas Field (2012); (3) Key Technologies, Complete Equipment, and Engineering Application of Ultra-High Voltage AC Power Transmission (2012); and (4) Ultra-High Voltage ±800 kV Direct Current Transmission Project (2017). In terms of thematic focus, two awards recognized advancements in fossil fuels (oil and gas), and two were conferred for breakthroughs in electric power, specifically ultra-high-voltage transmission.

Although most BICES projects were awarded the Second Prize, the proportion of First Prizes is more indicative of the shifting focus of BICES. A turning point in China’s BICES capabilities occurred around 2006. Following this point, the proportion of BICES projects winning the First Prize increased and has since been maintained at an overall rate of approximately 40%. This trend is attributable to factors such as energy structure transformation, supportive government policies, and growing demands for environmental conservation.

4.2. Evolution of Research Foci in BICES

4.2.1. General Characteristics of Research Foci in BICES

In this study, thematic tags are assigned to each of the 552 BICES projects to characterize their research focus. The co-occurrence relationships among these tags are first visualized to provide an overarching perspective. Figure 3 and Figure 4 outline the general research themes of BICES over the 21-year period, with “coal,” “oil and petroleum,” “electricity,” “clean energy,” and “energy efficiency” emerging as dominant domains within China’s BICES landscape. This observation aligns with the findings of Grubler et al. [5], who reported that studies related to fossil fuel technologies account for over 50% of energy supply technology research, followed by electricity and clean energy. Furthermore, Figure 4 reveals a strong correlation between innovations related to coal and electricity, which is consistent with the actual situation of coal consumption and electricity production in China. Specifically, thermal power (predominantly coal-based) accounted for approximately 70% of electricity generation in 2024, reflecting substantial coal consumption.

This dominance of coal is rooted in its abundance, which has long secured its central role in China’s energy mix. However, the severe environmental pollution associated with coal combustion has prompted the central government to promote technologies for cleaner and more efficient energy use, stimulating innovation in the clean energy sector. This policy emphasis is reflected not only in the “coal” domain but also in the advancing research on “clean and efficient energy technologies and equipment.”

Concurrently, electricity, as the primary carrier of energy, has experienced sustained demand growth driven by rapid economic development and urbanization over the past two decades. Rising living standards have further amplified electricity consumption, necessitating shifts in consumption patterns and imposing higher requirements for power system security and reliability. Consequently, this growing demand has been a key driver of innovation in power technologies, as exemplified by the fact that two of the four Special Prizes in the BICES projects were awarded to breakthroughs in ultra-high voltage transmission technology.

4.2.2. Evolving Research Foci of BICES

Figure 5 presents a series of word clouds tracking the evolution of research foci in the BICES. An analysis of tags from 552 projects reveals a clear declining trend in the prominence of fossil fuels, particularly coal, from 2000 to 2020. This trajectory aligns with China’s commitment to environmental goals and economic transition. In line with making emission reduction a national strategy, long-term targets were established to curb coal consumption. These measures, reinforced by market-based and administrative policies, successfully reduced coal’s share in the primary energy mix and led to an absolute decline in consumption after 2013. This shift was accelerated by the 14th Five-Year Plan, which transitioned the economic agenda from “high-speed growth” to “high-quality development,” creating strategic opportunities for energy restructuring [45]. Consequently, not only coal but also other fossil fuels (e.g., oil/petroleum) showed a clear decline.

In contrast, electricity maintained a significant role throughout the period. As discussed previously, growing residential consumption and higher demands on grid reliability have led to stricter requirements for power quality. Thus, innovation in the power sector has focused not only on upgrading conventional technologies but also on adapting systems to integrate new energy sources.

While the government began prioritizing renewable energy in 2005, Figure 5 shows that clean energy technologies did not dominate the innovation landscape during the past two decades. This finding may reflect the early stage of China’s energy transition. Nonetheless, due to their diverse and environmentally friendly advantages, clean energy technologies hold significant potential to become a key driver of future economic growth.

4.3. CN Evolution of BICES Projects

4.3.1. Characteristics of Organizations Participating in BICES Projects

To investigate the evolving collaboration patterns behind breakthrough innovations, we constructed and analyzed the CNs of BICES projects using SNA. In these networks, nodes represent awarded organizations and edges denote collaborative ties between them. Following Wang et al. [22], the awarded organizations were classified into four types: universities, research institutes, enterprises, and government agencies. Among the 1288 organizations involved in the 552 projects, 882 were enterprises, 243 were research institutes, 120 were universities, and 43 were government agencies. Figure 6 shows the participation proportions of these four organizational types over the 21-year period (2000–2020). Despite some yearly fluctuations, the overall participation levels remained relatively stable. Enterprises constituted about half of all participants, while universities and research institutes each accounted for approximately 20%. It is worth noting that although research institutes outnumbered universities, their participation rates were similar. This suggests that certain universities participated in multiple BICES, leading to their repeated inclusion in the count. Government agencies, which served as bridges within the collaboration network, were the least numerous and showed a declining trend decreasing from about 10% in the early years to roughly 4% in later stages. This decline may indicate that, over time, the collaborative relationships initially facilitated by government agencies became self-sustaining, enabling the other three organization types to collaborate more spontaneously.

Table 1. Ranking of organizations participating in BICES projects.

Rank	Organization Code	Frequency	Organization
1	U002	63	China University of Petroleum
2	U001	62	China University of Mining and Technology
3	U003	36	Tsinghua University
4	R001	22	China Electric Power Research Institute
5	U004	20	Zhejiang University
6	U005	19	Tianjin University
7	R002	16	Research Institute of Petroleum Exploration & Development
8	E001	14	Daqing Oilfield of CNPC
9	U006	14	Xi’an Jiaotong University
10	E002	13	Sinopec Engineering Incorporation
11	U007	13	China University of Geosciences
12	U012	13	Southwest Petroleum University
13	R003	12	Sinopec Research Institute of Petroleum Processing
14	R005	12	China Electric Power Research Institute
15	U013	12	North China Electric Power University
16	R004	10	Sinopec Petroleum Exploration and Production Research Institute
17	U010	10	East China University of Science and Technology
18	U008	10	Shandong University of Science and Technology
19	U011	10	Huazhong University of Science and Technology
20	U009	10	University of Science and Technology Beijing

Note: U = university, R = research institute, E = enterprise

ranks organizations by their total participation frequency from 2000 to 2020, listing those with a frequency of 10 or higher. Although enterprises account for the largest aggregate number of participations, universities rank highest, indicating a higher concentration of innovative contributions among a few leading universities. The top three participants are China University of Mining and Technology, China University of Petroleum, and Tsinghua University. The first two of these are recognized for their strong research capabilities in the national energy technology sector, while Tsinghua University is a comprehensive, prestigious institution excelling across multiple disciplines.

Among the organizations listed in Table 1, the only two enterprises featured are Daqing Oilfield of CNPC and Sinopec Engineering Incorporation, both of which are state-owned enterprises (SOEs). Typically, China’s SOEs, which were established earlier and possess more substantial resources than other enterprises, have a strong presence in the traditional fossil energy technology sector. In contrast, private enterprises are absent from the ranking. This absence can be attributed to their more recent emergence and focus on the new energy technology sector, where they are still in a relatively early stage of development.

4.3.2. Visualization of CNs from 2000 to 2020

To reveal the evolving collaborative landscape of the BICES, discrete CNs from 2000 to 2020 are visualized using Ucinet, as illustrated in Figure 7. Several developmental trends are evident: during the focal period, the network continuously expanded in scope, the number of innovative organizations increased steadily, and connections grew denser over time. In the first five years (2000–2004), the CNs were relatively sparse. In subsequent years, newly joined organizations established connections with incumbents through specific projects, and these cooperative ties gradually strengthened. The period from 2010 to 2012 marked the network’s largest scale, with several award-winning projects being jointly proposed by diverse participants. Participation peaked in 2010, which was also notable for the energy field receiving the top honor—the Special Prize—three times. This flourishing period for BICES can be attributed to growing energy innovation demands driven by economic growth and urbanization, as well as a series of policies promoting energy structural reform. From 2013 to 2020, the structure of the now well-established network gradually stabilized.

The size of each node in Figure 7 corresponds to its degree centrality, revealing that universities, SOEs, and research institutes—which are highly ranked in Table 1—gradually solidified their roles as stable core innovators as the network evolved. Simultaneously, repeated interorganizational collaboration became increasingly common and prominent over the 21-year period, as indicated by the growing thickness of the edges, which represent tie strength. The overall proliferation of network connections reflects a significant intensification of collaborative innovation activity among BICES participants.

To track changes in the network’s scale, we used Ucinet to quantify the number of nodes and ties over time. As shown in Figure 8, the numbers of both nodes (organizations) and ties (collaborations) in the BICES network grew consistently from 2000 to 2012, evolving in tandem. The expansion of participation positively influenced the network by increasing opportunities for interaction, knowledge exchange, and access to resources. It also enlarged the pool of potential partners, facilitating the formation of new collaborative relationships. The steady rise in ties indicates a strengthening of collaborative activity, resulting in a more interconnected and robust network structure during this period.

4.3.3. Network Characteristics and Key Nodes of CNs

This study employs key SNA metrics—network density, betweenness centrality, degree centrality, and closeness centrality—to characterize the CNs and identify pivotal organizations. Network density, defined as the ratio of actual connections to the theoretical maximum, reflects the overall interconnectedness among nodes [46]. As shown in Figure 9, the density of the BICES networks exhibited a gradual increase from 2000 to 2020. This upward trend indicates a rise in collaborative activity and, from a technological innovation perspective, suggests that the denser connectivity facilitated the establishment of collaboration mechanisms and enhanced knowledge diffusion through more direct and indirect pathways.

This study leverages centrality indicators to pinpoint pivotal organizations in the CNs of BICES.

Table 2. Top 5 organizations ranked by centrality.

Rank	Ranked by Betweenness Centrality	Organization Name	Ranked by Degree Centrality	Organization Name	Ranked by Closeness Centrality	Organization Name
1	U017	Chongqing University	U003	Tsinghua University	U017	Chongqing University
2	U002	China University of Petroleum	R005	China Electric Power Research Institute	U003	Tsinghua University
3	U011	Huazhong University of Science and Technology	U002	China University of Petroleum	E135	China Power Engineering Consulting Group Southwest Electric Power Design Institute Co., Ltd.
4	U003	Tsinghua University	R001	China Southern Power Grid CSG Electric Power Research Institute	U013	North China Electric Power University
5	U001	China University of Mining and Technology	E135	China Power Engineering Consulting Group Southwest Electric Power Design Institute Co., Ltd.	U002	China University of Petroleum

Note: U = university, R = research institute, E = enterprise.

ranks the top five organizations based on these indicators. Specifically, betweenness centrality denotes a node’s function as a network bridge [46]. The high betweenness centrality values observed for Chongqing University, China University of Petroleum, Huazhong University of Science and Technology, Tsinghua University, and China University of Mining and Technology suggest that they establish pervasive network connections and command significant influence over information and resource channels. Consequently, these organizations can more swiftly and efficiently acquire, assimilate, and integrate external resources. Their capacity to synthesize diverse strengths also positions them to pioneer new BICES initiatives.

Degree centrality reflects a node’s level of connectivity to all other nodes in a network [46]. Organizations with high degree centrality can acquire diverse and critical resources more efficiently, allowing them to play a leading role in CNs. As evidenced in Table 2, key leaders in the CNs of BICES projects include Tsinghua University, the Electric Power Research Institute, China University of Petroleum, China Southern Power Grid (CSG) Electric Power Research Institute, and China Power Engineering Consulting Group Southwest Electric Power Design Institute Co., Ltd.

Closeness centrality quantifies a node’s central position within a network by calculating the average shortest path distance to all other reachable nodes. Therefore, nodes with higher closeness centrality experience lower collaboration costs and, as a result, are more likely to attract partnerships from other organizations in the network. In this study, such cost-effective collaborators include Chongqing University, Tsinghua University, China Power Engineering Consulting Group Southwest Electric Power Design Institute Co., Ltd., North China Electric Power University, and China Power Engineering Consulting Group Southwest Electric Power Design Institute Co., Ltd.

To compare the centrality of different organization types,

Table 3. Different types of organizations ranked by centrality.

Rank	Degree Centrality				Betweenness Centrality				Closeness Centrality
	U	R	E	G	U	R	E	G	U	R	E	G
Top 10	5	3	2	0	6	3	1	0	4	2	4	0
Top 20	6	4	10	0	12	4	4	0	5	4	11	0
Top 30	6	5	19	0	19	5	6	0	5	6	19	0

Note: U = university, R = research institute, E = enterprise, G = government agency.

lists the top 10, 20, and 30 organizations ranked by centrality. A clear disparity emerges between the actual proportion of organization types in the network and their representation in the rankings. For instance, although enterprises constitute 68% of all nodes, they are underrepresented in the top tiers. Conversely, universities, which represent only 9% of the network, hold a disproportionately high number of top positions. This is particularly evident in the betweenness centrality rankings, where universities demonstrate an absolute advantage, comprising more than half of the top 10 and top 20 organizations (6/10 and 12/20, respectively). Similarly, universities and research institutes dominate the top 10 in terms of both degree and closeness centrality. Overall, while enterprises maintain a numerical presence—constituting about one third of the top 30—their influence is overshadowed by the dominant performance of universities and research institutes in the highest centrality ranks.

5. Discussion

5.1. Implications of Research Focus Evolution

The convergence of environmental and economic factors has ushered in an era of industrial transformation within the energy sector, making achieving balance between economic growth and long-term sustainability a pivotal challenge [1]. Sustainable progress in this field is fundamentally driven by continuous innovation in energy science and technology, which motivates diverse organizations to pursue technological breakthroughs. Given the multi-stakeholder nature of the energy value chain, project-based collaboration has become essential [47]. The evolution of the modern energy industry underscores the central role of collaborative innovation, where success in a market-driven environment increasingly depends on effective partnerships among universities, research institutes, enterprises, and government agencies [22]. Consequently, CNs have gained importance in shaping knowledge creation and dissemination within this domain.

This study examines the evolution of BICES by analyzing research foci and CNs. It draws on data from 552 energy-related SSTAP projects spanning from 2000 to 2020, utilizing project attributes and participant organizations to trace longitudinal developments across different phases. Aligning with Grubler et al. [5], we find that nearly half of BICES focus on novel technologies related to fossil energy, followed by electronics and clean energy. Previous studies, such as those by Fu et al. [48] and Shi et al. [49], argue that BICES in China should prioritize low-carbon technologies and smart power systems. However, due to inertia in energy consumption patterns, technologies associated with coal and oil continue to play a leading role in BICES, revealing the significant influence of market forces on science and technology innovation. Thus, the evolution of BICES appears to be shaped by the broader economic structure and market demand.

The word clouds presented in this study illustrate developmental trends in BICES, indicating a gradual decline in innovations related to fossil fuels such as coal and oil. This shift may be attributed to China’s current environmental and economic policies, which emphasize the quality of development over mere speed. Accordingly, technologies related to clean energy and energy efficiency improvement are playing an increasingly prominent role in BICES. In the future, China’s energy structure is expected to undergo a significant transformation, driven by breakthrough innovations in the energy sector.

5.2. Implications of CNs for BICES

The CNs of BICES integrate diverse participants from universities, research institutes, enterprises, and government agencies. Analysis of network attributes indicates a continuing expansion in the scale of these CNs, with ties among organizations growing stronger and more frequent. In this evolving context, organizations should take proactive steps to enhance their competitiveness and consolidate their network positions. Selecting suitable innovation partners and building robust collaborative relationships are crucial steps in this process. It is advisable for organizations to invest in R&D and to collaborate with high-centrality organizations (e.g., top universities) that possess substantial expertise and experience. Repeated cooperation among entities has been observed, suggesting that sustained interaction builds trust, facilitates mutual learning, and strengthens shared vision within groups.

Although enterprises constitute the largest number of participants, prominent universities and research institutes dominate the top rankings in terms of network centrality, occupying core positions and maintaining extensive connections with other members. Research-intensive universities (e.g., China University of Mining and Technology, China University of Petroleum, and Tsinghua University) play indispensable roles in BICES due to their strong research capabilities in the energy sector. These institutions hold comprehensive informational advantages and exert considerable influence over BICES initiatives. They also tend to gain greater trust and endorsement from other actors, positioning them as leaders in establishing energy technology standards and technical norms. An organization’s structural position within the network thus contributes significantly to its sustained importance throughout the evolution of BICES.

In addition to leading universities and research institutions, large SOEs with monopolistic positions in the energy industry also occupy elevated network positions. However, the majority of SOEs face challenges such as information redundancy, which test their managerial capacity. At the same time, the current BICES model has certain limitations. Although enterprises account for more than half of all participants, few rank among the top ten in centrality. While enterprises are expected to drive technological innovation, their capacity for independent innovation remains relatively weak. The shortage of innovation-leading enterprises leading innovation may be attributed to China’s predominantly top-down approach to BICES, which may constrain enterprise-led innovation, particularly among SOEs. Thus, policymakers should adjust their strategies to encourage enterprises’ independent innovation and strengthen collaborative R&D to enhance their innovation capabilities. Attracting experienced energy technology experts to emerging energy enterprises and providing financial and market incentives can effectively stimulate innovation [50].

CNs also vary significantly among enterprises with different ownership structures. While SOEs benefit from close government ties that facilitate access to funding and policy protection, this may reduce their incentive to innovate. Conversely, although private enterprises (PEs) are active in new energy technology sectors, they have not yet achieved significant network positions, and they lack clear advantages in obtaining government subsidies compared to SOEs. Nevertheless, in countries such as the United States, as government involvement in applied energy technology research declines, PEs assume greater influence [26,51]. Therefore, this study recommends research organizations strengthening energy innovation collaboration with PEs.

Based on comprehensive network analysis, we can conclude that current BICES efforts rely heavily on industry–university–research collaboration. While these partnerships appear as synergies among enterprises, universities, and research institutes, they essentially reflect organizational complementarities. The ultimate aim of collaboration is to advance BICES and translate research results into practical outcomes. The industry–university–research mechanism should be improved to gradually build a national BICES innovation system. Practical steps may include establishing cooperative centers and enabling staff exchanges between entities. The government should also facilitate communication among organizations to build trust and cooperation. Supportive policies and legal frameworks aligned with BICES strategies are essential to guide, regulate, and protect research and investment activities. The following steps are recommended. First, stable financial support should be ensured for key organizations within the CN. In particular, increased investment in critical technology R&D can further propel technological advancement, and breakthroughs in these areas could significantly elevate energy innovation outcomes. Second, within China’s research funding system, government-commissioned energy technology projects mainly target universities and research institutes while underutilizing the potential of enterprises. Therefore, national science and technology programs should more actively encourage enterprises’ independent innovation. Third, robust intellectual property protection policies are needed, as intellectual property is not only an outcome of BICES participation but also a core competitive asset for enterprises. Finally, building a shared vision among different types of organizations remains essential. Research on innovation systems confirms that successful innovation requires aligning the diverse roles of participants toward common goals.

6. Conclusions

Innovation, and particularly breakthrough innovation in science and technology, plays a pivotal role in balancing economic and environmental objectives within the energy sector. To map the research foci and collaborative patterns of BICES, this study analyzes a dataset of 552 energy-related SSTAP projects spanning the period from 2000 to 2020. A hybrid research framework is designed to identify evolving research themes across these projects and examine the dynamics of their underlying CNs. The methodology involves four main steps. First, an LLM is employed to generate descriptive tags characterizing the research focus of each project. Second, statistical analysis and word cloud visualization are applied to track trends in research focus evolution. Third, SNA is used to visualize and assess the structural features of CNs. Finally, evidence-based suggestions are proposed to strengthen collaborative mechanisms within BICES.

The findings reveal that innovations related to fossil energy continue to dominate BICES due to China’s existing energy consumption structure. However, driven by the positive effects of energy transition policies, a noticeable shift in research focus is occurring, from fossil-based systems toward clean and high-efficiency energy technologies. In terms of collaborative patterns within BICES, universities occupy most of the central positions in the CNs. Although enterprises constitute the largest group of participants, the majority—and particularly PEs—exert limited influence on innovation outcomes. Therefore, targeted efforts are needed to strengthen the innovation capacity of PEs.

Previous studies have relied on publications or patents to identify research foci and collaborative patterns in energy-related innovations. However, such approaches have often failed to fully capture the evolving landscape of breakthrough innovations, given the extended time horizon typically required for such advancements. In contrast, this study employs a dataset of SSTAP projects to trace the evolution of breakthrough innovations in China’s energy sector over time, thereby contributing to the theoretical understanding of breakthrough innovation pathways. Furthermore, using SNA, we examine the relative importance of different types of organizations within CNs. In a departure from patterns observed in countries such as the United States, universities (rather than enterprises) occupy the most central positions in the CNs of BICES. This finding deepens our understanding of the collaborative mechanisms underlying breakthrough innovations in varying national contexts. Finally, this study offers suggestions for policy and practice aimed at enhancing the efficiency of inter-organizational collaboration and strengthening industry–university–research mechanisms within BICES.