Novel electric power system academic hotspot analysis and prediction method based on literature metrology method and expert system

Abstract

The invention discloses a novel electric power system academic hotspot analysis and prediction method based on a literature metrology method and an expert system, and belongs to the technical field of academic hotspot analysis and prediction. Comprising the following steps of 1, summarizing retrieval fields according to characteristics of the novel power system and related research reports, performing retrieval through a CNKI database, and collecting required data; 2, importing the data collected in the step 1 into a literature metrology analysis tool CiteSpace for clustering analysis, and summarizing a keyword cluster in the research field based on a clustering result; 3, researching year-by-year word frequency evolution characteristics of the keyword clusters through a CNKI database so as to perform year-by-year knowledge tracking of the novel power system; 4, carrying out the analysis of related government data of the novel power system, the analysis of keyword outburst and the analysis of time zone distribution; and 5, establishing an academic hotspot analysis and prediction model based on an expert system, and predicting the academic hotspots by taking an analysis result in the step 4 as model input data.

Description

Technical Field

[0001] The present invention relates to the technical field of academic hot spot analysis and prediction, and in particular to a novel power system academic hot spot analysis and prediction method based on bibliometrics and expert system. Background Art

[0002] As the global demand for energy continues to rise, energy shortages have become a major challenge that social and economic development must face. Building a new power system is the basis for meeting this challenge. Its core purpose is to ensure the security of energy and power supply, and its primary task is to meet the social and economic demand for high-quality electricity. The new power system is committed to creating a supply and demand system characterized by a high proportion of new energy applications, which requires close coordination and flexible interaction among power sources, power grids, loads and energy storage. A strong, intelligent and flexible power grid is crucial as a hub platform, while scientific and technological innovation and institutional innovation are the fundamental driving force and guarantee for the development of the power system in the new era. The new power system is not only a key component of the new energy system, but also a key way to solve the global energy crisis.

[0003] Bibliometrics is a discipline that uses mathematical and statistical methods to study the generation, structure and transmission of literature. It focuses on the quantitative characteristics of scientific and technological literature, such as the number of documents, the number of citations, the distribution of authors, the frequency of keyword occurrence, etc., and uses these characteristics to reveal the regularity and trend of scientific development. In the implementation process, bibliometrics usually involves data mining and analysis of a large number of documents, using quantitative methods to evaluate the impact of research and predict the development trend of disciplines. Expert system is a computer program that simulates the decision-making ability of human experts. It collects and integrates a large amount of professional knowledge and uses rules and reasoning mechanisms to deal with complex decision-making problems. Expert systems usually include a knowledge base containing rules and facts provided by domain experts, and a reasoning engine for applying these rules to solve problems or provide advice.

[0004] As an important medium for the dissemination of scientific and technological theories, scientific papers show the evolution trend of technical theories in specific fields, point out research paths for researchers, and give birth to a revolution in technological innovation. Therefore, a new power system academic hotspot analysis and prediction method based on bibliometrics and expert system is needed to reveal the scientific research dynamics in a certain field. Summary of the invention

[0005] The purpose of the present invention is to propose a new power system academic hot spot analysis and prediction method based on bibliometrics and expert system, which includes the following steps:

[0006] Step 1: Summarize the search fields based on the characteristics of the new power system and related research reports, and search through the CNKI database to collect the required data;

[0007] Step 2: Import the data collected in step 1 into the bibliometric analysis tool CiteSpace for cluster analysis, and summarize the keyword clusters in the research field based on the clustering results;

[0008] Step 3: Use the CNKI database to study the frequency evolution characteristics of keyword clusters year by year, so as to track the knowledge of new power systems year by year;

[0009] Step 4: Conduct policy data analysis, keyword emergence analysis and time zone distribution analysis related to the new power system;

[0010] Step 5: Establish an academic hotspot analysis and prediction model based on an expert system, use the analysis results of step 4 as model input data, and predict academic hotspots.

[0011] The beneficial effects of the present invention are:

[0012] The present invention forms a set of standard electric power science and technology literature analysis methods with "new power system" as the theme and bibliometrics and expert system as the core, which can explore the research trend of new power system and guide scientific and technological innovation in the power field. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a flow chart of a novel power system academic hot spot analysis and prediction method based on bibliometrics and expert system of the present invention;

[0014] Figure 2 It is the interface of the bibliometric analysis tool CiteSpace;

[0015] Figure 3 This is a diagram of the visualized network after clustering;

[0016] Figure 4 This is a heat map of keyword tracking changes for the theme "auxiliary services";

[0017] Figure 5 It is a heat map of keyword tracking changes for the theme "virtual power plant";

[0018] Figure 6 This is a heat map of keyword tracking changes with the theme of "low carbon";

[0019] Figure 7 It is a heat map of keyword tracking changes in the theme "electricity market";

[0020] Figure 8 This is a heat map of keyword tracking changes for the theme "smart grid"

[0021] Fig. 9 This is a heat map of keyword tracking changes for the theme "Optimized Scheduling";

[0022] Fig.10 This is a heat map of keyword tracking changes for the theme "energy storage";

[0023] Fig.11 It is a heat map of keyword tracking changes for the theme "demand response";

[0024] Fig.12 It is a keyword emergence map;

[0025] Fig.13 It is the keyword time zone distribution map. DETAILED DESCRIPTION

[0026] The present invention proposes a novel power system academic hot spot analysis and prediction method based on bibliometrics and expert system. The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0027] Figure 1 This is a flow chart of the novel power system academic hot spot analysis and prediction method based on bibliometrics and expert system of the present invention, which is as follows:

[0028] (1) Field definition and research dataset construction

[0029] Based on the characteristics of the new power system and the summary of relevant research reports on the new power system, you can select keywords such as "new power system", "carbon neutrality", "new energy", "pumped storage", "energy storage technology", "virtual power plant", "low carbon", "demand response", "power market", and "renewable energy" as search fields to search for literature in the field of new power systems through the CNKI database.

[0030] First, go to the CNKI official website, use the "Advanced Search" function in the CNKI search bar, and select the "Professional Search" tab. Professional search is used by library and information professionals for novelty search, information analysis, etc., and uses operators and search terms to construct search formulas for search.

[0031] Secondly, determine the search scope. Considering that the new power system is a new research field, its related research literature is lagging, so the literature in recent years can be searched as the research subject, and the subject column is selected as power industry.

[0032] Finally, literature search was performed based on the search code. Literature data was retrieved based on the search instructions to construct the data set required for the research. Through CNKI professional search, after selecting the subject and year, the literature data in Refworks format were exported respectively.

[0033] (2) Constructing keyword clusters based on CiteSpace cluster analysis

[0034] Clustering is the process of dividing a collection of physical or abstract objects into multiple categories. The elements within the groups formed by clustering have high similarities, while the elements between different groups have high differences. Figure 2 This is the interface of the bibliometric analysis tool CiteSpace. This embodiment uses the clustering algorithm of CiteSpace to divide the literature in the field of new power systems into several categories, and then analyzes and summarizes them into keyword clusters. The algorithm used by CiteSpace is mainly a clustering algorithm based on co-word analysis. Its core principle is to reveal the research trends and theme evolution in scientific literature by identifying and analyzing the co-occurrence relationship of keywords or topics in the literature:

[0035] 1) Co-occurrence analysis

[0036] a. Keyword co-occurrence: CiteSpace counts the keywords in the documents and then analyzes the co-occurrence of these keywords in the documents. If two keywords appear in multiple documents at the same time, we say that there is a co-occurrence relationship between them.

[0037] b. Time series co-occurrence: CiteSpace can also analyze the co-occurrence relationship of keywords in time series, that is, which keywords appear together in different time stages. This helps to reveal the evolution trend of research topics over time.

[0038] 2) Similarity measurement

[0039] a. Cosine similarity: CiteSpace uses cosine similarity to measure the similarity between documents. Cosine similarity is calculated based on the cosine value of the angle between the vectors. If the angle between the keyword vectors of two documents is smaller, their similarity is higher.

[0040] 3) Hierarchical clustering algorithm

[0041] a. Agglomerative hierarchical clustering: CiteSpace uses an agglomerative hierarchical clustering algorithm to construct a cluster tree. This algorithm starts by treating each document as a separate cluster, and then gradually merges clusters with high similarity. Clusters with high similarity will be merged into larger clusters.

[0042] b. Distance threshold: During the clustering process, CiteSpace sets a distance threshold. Only documents with a similarity higher than this threshold will be merged into the same cluster.

[0043] 4) Clustering stability evaluation

[0044] a. Modularity: CiteSpace uses modularity to evaluate the stability of clustering. Modularity is an indicator that measures the tightness of connections within a cluster and the sparseness of connections between clusters. The higher the value, the better the clustering effect.

[0045] 5) Visualization

[0046] a. Network diagram: CiteSpace can generate a network diagram to show the connection relationship and clustering results between documents in an intuitive way, such as Figure 3 shown.

[0047] Based on the clustering results, the keywords that are not clear enough about "new power system" in the cluster are excluded, and combined with the existing research reports on new power system and the experience of experts in the field, 8 keyword clusters about "new power system" research are summarized. Since each keyword cluster can highlight some research topics under the scope of new power system, see Table 1.

[0048] Table 1 Keyword clusters of new power system (frequency greater than 20)

[0049]

[0050]

[0051] (3) New power system knowledge tracking year by year

[0052] Through the CNKI database, the frequency evolution characteristics of keyword clusters are studied year by year, so as to track the knowledge of new power systems year by year. Figures 4 to 11 As shown, it shows the keyword tracking change heat map of 8 different topics.

[0053] As shown in the figure, the horizontal axis represents the year, covering 2020-2023; the vertical axis represents the keywords in the subject term; in the heat map, the numbers represent the number of responses to the keywords in each year, and the response indicates the direction of research on the keyword at this time; the color depth of each square expresses the response situation, as shown in the legend on the right, the darker the color, the more responses, and vice versa.

[0054] (4) Analysis of policy data related to new power systems, keyword emergence analysis and time zone distribution analysis

[0055] 1) Keyword emergence analysis and time zone distribution analysis

[0056] Keyword emergence refers to high-frequency keywords that appear in a certain period. The emergence graph composed of these keywords can, to a certain extent, reflect the research hotspots and development trends of a specific period. If a large number of new keywords and content related to research in this field appear in a certain stage, it is helpful to grasp and understand the latest trends of academic research and scientifically predict the future development direction. The "emergent word detection method" is used to determine the research frontier. The principle is to count the keywords with high emergence rates on the basis of the keyword co-occurrence map, and determine the research frontier based on the change history of its emergence change line graph. Among them, the line graph has an obvious upward and downward trend. Keywords that suddenly change in a certain period of time are keywords with high emergence. This indicates that the attention of the keyword has obviously increased or decreased, and the result obtained is the research frontier in this period.

[0057] By using the "Citation Burst" function in CiteSpace to find the emergent words in the literature and sort them in order, we can get the historical stage diagram of 26 highly emergent keywords in articles related to new power systems. The red area indicates the time period when the emergent words appear, and the time stage diagram of each highly emergent keyword can be clearly seen. Fig.12 shown.

[0058] 2) Time zone distribution analysis

[0059] In order to explore the distribution characteristics of keywords in the time dimension, grasp the dynamic change process of keywords as a whole, and understand the evolution of new power system research, we can use the timeline view of keyword co-occurrence. The time zone distribution function (Timeline View) of CiteSpace can show the change trend and evolution process of specific keywords or topics over a period of time, such as Fig.13 In the drawn keyword timeline view, the main keywords of each time slice are displayed respectively, and the co-occurrence of each keyword is indicated by connecting lines, indicating that there is a temporal inheritance relationship between related research, so as to explore the update changes and development trends of research hotspots in this field in recent years.

[0060] 3) Analysis of policy data on new power systems

[0061] Policy documents play a significant leading role in the development of new power systems. Taking "carbon neutrality" as an example, keywords such as "carbon reduction", "carbon neutrality" and "carbon peak" emerged 21 years later.

[0062] (5) Constructing academic hotspot analysis and prediction model based on expert system

[0063] Using expert systems to predict knowledge in the field of "new power systems":

[0064] Data collection and preprocessing module:

[0065] Data collection means collecting academic papers, reports, books and other documentary materials in the field of document management over the years, using text mining technology to extract keywords, themes and concepts in the documents, and then cleaning, deduplicating, and tagging the extracted keywords.

[0066] Knowledge base construction:

[0067] Collect expert knowledge in the field of library and document management, including historical research trends, theoretical frameworks, important concepts, etc., as well as organize research keywords and related literature over the years to build a time series database.

[0068] Keyword analysis module:

[0069] Conduct frequency analysis on keywords and count the frequency of occurrence of each keyword in different years; analyze the co-occurrence relationship between keywords and build a co-occurrence network; and identify the changing trends of keywords over time through time series analysis.

[0070] Inference Engine:

[0071] Based on expert knowledge and historical data, we build a reasoning rule library. We apply logical reasoning, pattern recognition and other methods to reason based on keyword analysis results and build a reasoning mechanism.

[0072] As one of the cores of the expert system, the inference engine has the basic function of calling a reasonable knowledge framework from the knowledge base to solve the problem. Whether the knowledge base has complete professional and domain knowledge and whether the inference engine can make reasonable selections and calls to this knowledge are two important aspects that determine the quality of the expert system. Case-based reasoning system, rule-based reasoning system and pattern-based reasoning system are the three commonly used reasoning mechanisms. Considering the characteristics of the above three reasoning systems, this embodiment adopts a rule-based reasoning system.

[0073] Rule-based reasoning means that after mastering the expert domain knowledge, the relevant professional knowledge and computational models are converted into rule form and saved in the knowledge base. All saved rules are represented by the production knowledge representation rule method. The production knowledge representation rule method has its unique advantages in representing knowledge with causal relationships. The basic form of production is:

[0074] if P then Q

[0075] Among them, P represents the premise of the production, that is, the precondition, which is used to indicate whether the production rule can be used. Q represents the result or operation process of the production, which is used to indicate what kind of result is produced or what kind of operation should be performed when the condition of the rule is met. The meaning of the entire production expression is: if the condition P is met, the operation specified in Q can be performed.

[0076] Organizing the description of knowledge sets or computational models and putting them into the knowledge base is the basis for the expert system based on rule reasoning to solve problems. What rules need to be established in the rule base, whether the knowledge is complete, and the method of rule retrieval are the key issues that need to be considered in order to better perform reasoning and express the domain knowledge in a reasonable way. The strategy of the reasoning machine is used to decide which rule to use next. The reasoning process and the reasoning of the behavior process are two important steps in the reasoning process of the reasoning machine. The reasoning process can well understand the results of the reasoning, and its characteristics are similar to the thinking process of people. Therefore, in the system, its knowledge base and reasoning machine can be independently updated and improved, similar to the characteristics of modular design, so as to facilitate the upgrade and parameter modification of the system. The rule-based reasoning method is the most widely used reasoning method in expert system reasoning.

[0077] The inference engine rules established in this embodiment are as follows:

[0078] a. Based on the keyword tracking results of each topic, and taking into account the number of responses and the magnitude of change, we can filter out the directions that experts and scholars focus on. The filtering rule is: if the frequency of this keyword shows an upward trend with the increase of the year, then the keyword is listed as a key keyword; if the frequency of this keyword remains stable as the year increases, and the value is large, then the keyword is listed as a key keyword. The above rules are expressed in production form, written as:

[0079] if: The frequency of a keyword increases with the increase of years.

[0080] then: This keyword is the key word

[0081] if: The frequency of a keyword remains stable as the year goes by

[0082] and the keyword frequency value is large

[0083] then: This keyword is the key word

[0084] b. The performance and evolution of each keyword with respect to the time series can be obtained from the keyword time zone distribution map and emergence map of the new power system keywords. The time zone distribution map usually shows the mention frequency and importance of different keywords in different time periods through time series analysis, thereby revealing the hot topics and their changes in research and practice. Through keyword mutation analysis, hot topics and key technologies in new power systems can be discovered, thereby identifying new trends in the fields of research and practice. Based on the properties and characteristics of the two graphs, the following two rules can be listed: In the keyword time zone distribution map, if a keyword appears in a relatively new year and its frequency on the graph is better than other keywords of the same year, then the keyword is listed as a key keyword; in the keyword emergence map, if a keyword appears in a relatively new year, then the keyword is listed as a key keyword. The above rules are expressed in production form, written as:

[0085] if: The year in which a keyword appears in the time zone distribution diagram is relatively new

[0086] and frequency performs better than other keywords in the same year

[0087] then: This keyword is the key word

[0088] if: The year in which a keyword appears in the emergent graph is relatively recent

[0089] then: This keyword is the key word

[0090] c. Considering that policy documents play a significant leading role in the development of new power systems, this embodiment considers extracting keywords from policy documents as a reference for future research directions. The rules are listed as follows: If a keyword is pointed out as a future hot spot in the policy documents of the past two years, then the keyword is an emerging word. The above rules are expressed in production form, written as:

[0091] if: a keyword is a hot topic pointed out in policy documents in the past two years

[0092] then: This keyword is the key word

[0093] Based on the above three rules, the keywords are screened and classified to predict the future research hotspots of new power systems, thus realizing the academic hotspot analysis and prediction function.

[0094] The present invention forms a set of standard electric power science and technology literature analysis methods with "new power system" as the theme and bibliometrics and expert system as the core, which can explore the research trend of new power system and guide scientific and technological innovation in the power field.

Claims

1. A new method for analyzing and predicting academic hot spots in power systems based on bibliometrics and expert systems, characterized by: The following steps are involved: Step 1: Summarize the search fields based on the characteristics of the new power system and related research reports, and search through the CNKI database to collect the required data; Step 2: Import the data collected in step 1 into the bibliometric analysis tool CiteSpace for cluster analysis, and summarize the keyword clusters in the research field based on the clustering results; Step 3: Use the CNKI database to study the frequency evolution characteristics of keyword clusters year by year, so as to track the knowledge of new power systems year by year; Step 4: Conduct policy data analysis, keyword emergence analysis and time zone distribution analysis related to the new power system; Step 5: Establish an academic hotspot analysis and prediction model based on an expert system, use the analysis results of step 4 as model input data, and predict academic hotspots.

2. According to the new power system academic hot spot analysis and prediction method based on bibliometrics and expert system according to claim 1, it is characterized by: The search fields in step 1 include new power system, carbon neutrality, new energy, pumped storage, energy storage technology, virtual power plant, low carbon, demand response, electricity market, and renewable energy.

Images