Nuclear power plant operator simulator training demand analysis method and computer device

Abstract

The invention provides a nuclear power plant operator simulator training demand analysis method and a computer device, the nuclear power plant operator simulator training demand analysis method comprises the following steps: collecting and managing operator simulator training demands by using a discrete partition mode to obtain demand data, different collection modes are adopted for different demand sources in the discrete partition mode; word embedding preprocessing is conducted on the demand data according to feature dimensions, and the feature dimensions comprise events, regulations, knowledge points, basic skills, special requirements and operation key points; and generating a training demand result through clustering analysis. According to the training demand analysis method based on basic feature extraction and analysis, the problems that in the existing manual management process, the number of demands for simulator training of operators is large, a large amount of manual investment is needed, and omission or errors may exist are solved.

Description

Technical Field

[0001] The present application belongs to the field of nuclear power technology, and specifically relates to a method and computer device for analyzing the training needs of a simulator for operators of a nuclear power plant. Background Art

[0002] Simulator training for nuclear power plant operators is a core method for improving the skills of operators. Operators control the operation of the reactor on a full-range simulator according to targeted simulator scenarios developed by instructors, so as to train operators' ability to control the reactor in a safe state under normal unit operating conditions, expected operating events, design basis accidents, and beyond-design basis accidents. During the training, simulator instructors observe the operators' response operations and correct the operators' operations based on the observations, so that they can take the best intervention operations under specific conditions to ensure that the unit is in a safe and controlled state.

[0003] There are many systems and equipment in nuclear power plants, and there are nearly a thousand events (equipment failures or accidents) that can be set. In simulator training, it is necessary to select appropriate events from nearly a thousand events to form complex scenarios to train operators to control the unit under complex working conditions. Only instructors with rich experience in unit operation and simulator training can effectively sort out training needs, design training goals, and select targeted events to develop high-standard simulator training scenarios.

[0004] The development of simulator training scenarios requires rigorous training needs analysis. There are many types of needs for simulator training for operators, including laws and regulations, industry standards, rotation training procedures, experience feedback, evaluation reports, training / production observations, training summaries, crew changes, etc. At the same time, the methods of obtaining needs are also different, including questionnaires, production documents, meeting actions, training reports, manual conversions, etc., which require a lot of human resources to organize and analyze. In addition, possible omissions or errors in the manual management process will have a certain impact on the high-quality training. Summary of the invention

[0005] In view of this, the embodiments of the present application are committed to providing a method and computer device for analyzing the training needs of nuclear power plant operator simulators. Through a training needs analysis method based on basic feature extraction and analysis, the problems of the existing manual management process in which the needs for operator simulator training are diverse, require a large amount of manual input, and may result in omissions or errors are solved.

[0006] The first aspect of the present application provides a method for analyzing the training needs of a nuclear power plant operator simulator, comprising: collecting and managing the operator simulator training needs using a discrete partitioning mode to obtain demand data, wherein different collection methods are used for different demand sources in the discrete partitioning mode; performing word embedding preprocessing on the demand data according to feature dimensions, wherein the feature dimensions include events, procedures, knowledge points, basic skills, special requirements and operating points; and generating training demand results through cluster analysis.

[0007] In a specific embodiment of the present application, the above-mentioned discrete partitioning mode is used to collect and manage the operator simulator training needs to obtain demand data, including: partitioning matrix according to the type of demand data, encoding each type of data, defining the demand source type as XI, to form a partition vector of the demand data, A = [X1, X2, X3...Xn], n is the type of source; for each partition vector XI, there are multiple data, the generated row vector is expanded to form a discrete data matrix vector D,

[0008] n is the type of demand data source, and m is the number of each type of data in the demand data source.

[0009] In a specific embodiment of the present application, the above-mentioned word embedding preprocessing of the demand data according to the feature dimension includes: establishing a feature matrix, defining attribute features yi for each data item, forming a feature matrix T, and using feature vectors to clean and assign values ​​to the demand source; according to the recommended sentence description of the feature dimension, using word embedding technology, the feature matrix T is vectorized to establish a feature vector database; after data collection, each data item in the discrete data matrix vector D is also word embedded to obtain the feature vector of the demand data, and the cosine similarity is calculated with each feature vector in the feature vector database. When the cosine similarity reaches 0.95, a feature is configured for the data of the discrete data matrix vector D, and the value is assigned to 1. The necessity is statistically calculated as an independent feature after data analysis.

[0010] In a specific embodiment of the present application, the above-mentioned recommendation sentence description according to the feature dimension uses word embedding technology to convert the feature matrix T into a vector and establish a feature vector database, including: for the event feature dimension of the demand data, the event requirements for training are divided into three categories: normal operating conditions, expected operating event work and accident conditions, and the features of the text description are embedded to establish an event vector database; for the procedure feature dimension of the demand data, the name and code of the procedure are embedded according to the sorted out list of current existing procedures to establish a procedure vector database; for the knowledge point feature dimension of the demand data, the knowledge and skills of the operators are sorted out to establish a database, the names of key knowledge points are embedded to establish a knowledge point vector database; for the basic skills feature dimension of the demand data, the basic skills and detailed dimensions are embedded to establish a basic skills vector database; for the special requirements feature dimension of the demand data, the special requirements are embedded to establish a special requirements vector database; for the operation key point feature dimension of the demand data, the names of the operation key points are embedded to establish an operation key point vector database.

[0011] In a specific embodiment of the present application, the above-mentioned training demand generation result through cluster analysis includes: the discrete data matrix vector D is nested with the attribute feature matrix T to obtain the training demand result matrix R, and all the data in R are represented by the assignment of 0 or 1 to the feature dimension.

[0012] n is the type of demand data source, m is the number of various types of data in the demand data source, a is the feature dimension category, and b is the subdivision dimension of each feature dimension; organize the data of the demand result matrix R to obtain the feature distribution of the demand data, and replace the data code with the feature dimension; use the business association relationship between the feature dimensions to check and eliminate duplicate results of each feature dimension; according to the business rule logical relationship between various feature dimensions, for any characteristic working condition, it may be that the assigned event and the procedure are both 1, and there is an envelope relationship between the two. In this case, the event dimension of the data is set to 1, and the procedure dimension is retained as 0; the knowledge points and the knowledge and skills of basic skills are also 1, and there is a preferred relationship between the two. In this case, the knowledge and skills dimension of basic skills is set to 0, and the knowledge point dimension is retained as 1. The data that must be followed in the design events and procedure training, laws and regulations, and curriculum development management requirements in the data source is assigned a value of 1; the necessity is calculated based on the results of clustering and deduplication. Necessity is defined as p, the number of times the feature dimension appears is k, and according to the formula Calculate the necessity, where R is the correction coefficient, which is a real number greater than 1, and make adjustments based on the actual effect; sort out and count the data in the demand result matrix that are not assigned 0, calculate the necessity for the same dimensions that appear repeatedly, arrange them according to the necessity of the characteristic dimensions of the demand data, and obtain a statistical table of the demand analysis results. Finally, obtain the demand analysis results for operator simulator training, and include the analysis results in the course training outline to guide the development of training lesson plans.

[0013] In a specific embodiment of the present application, different sources of demand include operator capability self-assessment, training assessment results, internal and external corrective actions, company management expectations, supervision requirements, industry recommended standards, internal and external experience feedback, design changes, working condition rotation training and procedural training requirements.

[0014] In a specific embodiment of the present application, different collection methods include questionnaire surveys, external system input, training report conversion and self-built databases. The questionnaire survey method mainly collects data on operator self-assessment of capabilities and company management expectations. The external system input mainly collects data on internal and external experience feedback and design changes. The training report conversion mainly converts training evaluation results. The self-built database mainly collects data on internal and external corrective actions, supervision requirements, industry recommended standards, working condition rotation training and procedural training requirements.

[0015] The second aspect of the present application provides a computer device, which includes a processor and a memory. The processor is used to execute the training needs analysis method for simulator training of nuclear power plant operators provided in the first aspect of the present application. The memory is used to store executable instructions of the processor.

[0016] The third aspect of the present application provides a computer-readable storage medium on which computer executable instructions are stored. When the executable instructions are executed by a processor, the method for analyzing the training needs of simulators for nuclear power plant operators provided in the first aspect of the present application is implemented.

[0017] The fourth aspect of the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the nuclear power plant operator simulator training needs analysis method provided in the first aspect of the present application.

[0018] The beneficial effects of the technical solution of this application are: through the collection of operator simulator training needs and the discrete partition induction method, the problems of insufficient sources of operator simulator training needs, messy data, and unclear sources are solved, and the effectiveness and precise management of training needs are achieved, providing clear goals and directions for the development of operator simulator training courses. The operator simulator training needs preprocessing method effectively solves the problems of unclear classification of characteristic dimensions of training needs, inaccurate classification categories, and poor correlation between the front and back. By constructing a characteristic dimension matrix, the core elements of operator simulator training are refined, and the pertinence, effectiveness, and accuracy of operator simulator training lesson plan development are improved. The intelligent analysis of operator simulator training needs data solves the problems of poor matching between needs and training content, insufficient closeness, and insufficient guidance for course development. After data analysis and induction, the training goals are more comprehensive and accurate, and the management is more convenient and meticulous. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Shown is a flow chart of a method for analyzing simulator training needs for nuclear power plant operators provided in one embodiment of the present application.

[0020] Figure 2 Shown is a flow chart of a method for analyzing simulator training needs for nuclear power plant operators provided in another embodiment of the present application.

[0021] Figure 3 Shown is a schematic diagram of the source of operator simulator training demand data in a nuclear power plant operator simulator training demand analysis method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0023] At least one embodiment of the present application provides a method for analyzing the training needs of a nuclear power plant operator simulator. The method for analyzing the training needs of a nuclear power plant operator simulator includes the following steps.

[0024] S1: Collect and manage the simulator training needs of operators using the discrete partition mode to obtain demand data. Different collection methods are used for different demand sources in the discrete partition mode.

[0025] S2: Perform word embedding preprocessing on demand data according to feature dimensions. Feature dimensions include events, procedures, knowledge points, basic skills, special requirements, and key points of operation.

[0026] S3: Generate results of training needs through cluster analysis.

[0027] According to the technical solution provided in the embodiment of the present application, the operator simulator training demand collection and discrete partition induction method (corresponding to S1) is used to solve the problems of insufficient sources of operator simulator training demand, messy data, and unclear sources, so as to achieve accurate management of the effectiveness of training demand and provide clear goals and directions for the development of operator simulator training courses. The operator simulator training demand preprocessing method (corresponding to S2) effectively solves the problems of unclear classification of characteristic dimensions, inaccurate classification categories, and poor correlation between the training demand. By constructing a characteristic dimension matrix, the core elements of operator simulator training are refined, and the pertinence, effectiveness, and accuracy of the operator simulator training lesson plan development are improved. The intelligent analysis of operator simulator training demand data (corresponding to S3) solves the problems of poor matching between demand and training content, insufficient closeness, and insufficient guidance for course development. After data analysis and induction, the training goals are more comprehensive and accurate, and the management is more convenient and meticulous.

[0028] The results of the training needs can be used to guide the precise development of simulator training lesson plans and meet the needs of high-quality training for operators. The training needs analysis method based on basic feature extraction and analysis successfully solved the identification and analysis of operator simulator training objectives and provided a systematic training idea for nuclear power plant operator simulator training.

[0029] For example, in S1 above, the requirements and sources involved in improving the skill level of nuclear power plant operators are analyzed, and the classification of demand data is sorted out, such as: self-assessment and training evaluation of trainees, design events and procedures training, industry good practices and experience feedback, laws and regulations, management expectations, curriculum development management requirements, unit design changes, etc. Set up different areas to collect data, classify and organize according to the source of demand, and ensure the management of various data demand data in multi-dimensional and discrete modes.

[0030] In at least one embodiment of the present application, S1.1 and S1.2 are a specific implementation of the above S1.

[0031] S1.1: Create a partition matrix based on the type of demand data and encode each type of data. Define the type of demand source as XI to form a partition vector for demand data. A = [X1, X2, X3...Xn], where n is the type of source.

[0032] S1.2: For each partition vector XI, there are multiple data. The generated row vector is expanded to form a discrete data matrix vector D.

[0033] n is the type of demand data source, and m is the number of each type of data in the demand data source.

[0034] It should be noted that the discrete data matrix vector D can also be called a data source matrix or a demand source matrix.

[0035] In the embodiment of the present application, a detailed demand source partition matrix is ​​formed through steps S1.1 and S1.2, and each discrete data generates an independent code to facilitate statistical quantity and management, and establish a data-based foundation for subsequent analysis.

[0036] For example, in S2 above, the training needs analysis needs to be as comprehensive and sufficient as possible in data preprocessing and assignment. Information preprocessing and assignment are the core methods of this module. After data collection, word embedding technology is used to preprocess and assign values ​​according to feature dimensions. Each feature dimension is subdivided into multiple categories. For example, basic skills are divided into monitoring, control, conservatism, team, knowledge, and behavioral norms. The six core elements of operator capabilities are used as one of the important data in the entire demand analysis framework. Other classification dimensions are also expanded to enrich the assignment characteristics of needs.

[0037] In at least one embodiment of the present application, S2.1 and S2.3 are a specific implementation of the above S2.

[0038] S2.1: Establish a feature matrix, define the attribute feature yi for each data item, form a feature matrix T, and use the feature vector to clean and assign values ​​to the demand source.

[0039] a is the feature dimension category, and b is the subdivision dimension of each feature dimension.

[0040] S2.2: According to the recommended sentence description of the feature dimension, use word embedding technology to transform the feature matrix T into a vector and establish a feature vector database.

[0041] S2.3: After data collection, each item in the discrete data matrix vector D is also word embedded to obtain the feature vector of the demand data, and the cosine similarity is calculated with each feature vector in the feature vector database. When the cosine similarity reaches 0.95, a feature is configured for the data of the discrete data matrix vector D and assigned a value of 1. Necessity is used as an independent feature for statistical calculation after data analysis.

[0042] It should be noted that the cosine similarity can be verified based on test results, with 1 being a complete match.

[0043] In at least one embodiment of the present application, S3.1 and S3.2 are a specific implementation of the above S3.

[0044] S3.1: Nest the discrete data matrix vector D and the attribute feature matrix T to obtain the training demand result matrix R. All the data in R are represented by the assigned value 0 or 1 of the feature dimension.

[0045] n is the type of demand data source, m is the number of various types of data in the demand data source, a is the feature dimension category, and b is the subdivision dimension of each feature dimension.

[0046] S3.2: Sort the data of the demand result matrix R, obtain the characteristic distribution of the demand data, and replace the data code with the characteristic dimension.

[0047] For example, the feature distribution of the demand data is shown in Table 1. The replacement data codes are the feature dimensions as shown in Table 2, where 0 indicates that the identification is empty and 1 indicates that there is a clear feature dimension mapping.

[0048] Table 1. Distribution of demand data characteristics

[0049]

[0050] Table 2 Replace data codes with feature dimensions

[0051]

[0052] Because the feature dimensions of demand data are diverse and there is an envelope relationship between the features, after data preprocessing, each demand data may match multiple assignments, but the training objectives achieved by different feature dimensions are consistent. For example, the data X12 assignment y14 event condition is false safety note, and the data X13 assignment y23 procedure is I31 procedure. The file used during the false safety note event training is the I31 procedure. The training objectives of the two are consistent and there is no need to repeat the training. Similarly, there are correlations between other dimensions. Therefore, it is necessary to use the business correlations between feature dimensions to check and eliminate duplicate results of each feature dimension through clustering statistics.

[0053] S3.3: Use the business association relationship between feature dimensions to perform clustering statistics on the results of each feature dimension to check and eliminate duplication.

[0054] The business association rules between feature dimensions are shown as follows:

[0055]

[0056] The priority order is event > procedure > knowledge point > basic skills > special requirements > key points of operation.

[0057] S3.4: According to the business rule logical relationship between various feature dimensions, for any feature working condition, the assigned event and the procedure may both be 1, and there is an envelope relationship between the two. In this case, the event dimension of the data is set to 1, and the procedure dimension is retained as 0; the knowledge point and the basic skills are both 1, and there is a preferred relationship between the two. In this case, the basic skills dimension is set to 0, and the knowledge point dimension is retained as 1. The data that must be followed in the design events and procedure training, laws and regulations, and curriculum development management requirements in the data source is assigned a value of 1.

[0058] In this way, S3.4 is used to ensure the uniqueness and accuracy of the demand result dimension and reduce the probability of repeated development.

[0059] S3.5: Calculate the necessity based on the result of clustering deduplication. Define the necessity as p, the number of times the feature dimension appears as k, and according to the formula Calculate the necessity. R is the correction coefficient, which is a real number greater than 1 and is adjusted according to the actual effect.

[0060] For example, a real number greater than 1 is 1.2.

[0061] S3.6: Sort and count the data with values ​​other than 0 in the demand result matrix, calculate the necessity of the same dimensions that appear repeatedly, arrange them according to the necessity of the characteristic dimensions of the demand data, and obtain a statistical table of the demand analysis results. Finally, the demand analysis results of the operator simulator training are obtained, and the analysis results are included in the course training outline to guide the development of training lesson plans.

[0062] Table 1 Demand analysis results statistics

[0063]

[0064] In the above embodiment, according to the envelope and exclusion rules between the data feature dimensions, the data assigned values ​​other than 0 in the demand result matrix are checked for duplication, eliminated, and summarized, and arranged according to the demand attribute dimensions to obtain the demand analysis results. The priority of the feature dimensions is event>procedure>knowledge point>basic skills>special requirements>operation points. If the event and procedure are both 1, and there is an envelope relationship between the two, then the procedure dimension of the demand data is set to 0, and the event dimension is retained as 1; if the knowledge point and the knowledge and skills of the basic skills are both 1, and there is a mutually exclusive relationship between the two, then the knowledge and skills dimension of the basic skills is set to 0, and the knowledge point dimension is retained as 1. Finally, the necessity is calculated according to the results of clustering and deduplication after classification. Necessity is defined as p, the number of occurrences of the feature dimension is k, and the necessity is calculated according to the following formula: Where R is the correction coefficient, which is 1.2 (a real number greater than 1) and adjusted according to the actual effect). Sort by necessity from large to small to obtain the results of the needs analysis and guide the development of training lesson plans.

[0065] Next, combine Figure 2 , the simulator training needs analysis method for nuclear power plant operators is illustrated with examples. Figure 2 In the figure, 010 is the classification of the sources of simulator training requirements for operators, 020 is the method for collecting training requirements, 030 is the establishment of feature dimension attributes and feature vector libraries, 040 is the encoding of the demand data and the similarity matching and assignment in the feature dimension library, and 050 is the horizontal and vertical duplication checking, disambiguation and necessity statistics of the assignment results of the demand data to obtain the demand analysis results. It should be noted that 030 is also the establishment of the training demand feature dimension library. 040 is also the preprocessing and assignment of simulator training demand data. 050 is also the analysis process of simulator training demand data, including duplication checking, disambiguation and induction after data assignment, and result necessity analysis and sorting (corresponding to the above S3).

[0066] In the above S1, the needs for simulator training for operators are sorted out and classified. According to the job competency requirements of operators and the industry characteristics and regulatory characteristics of nuclear power plants, in some embodiments, reference is made to Figure 2 and Figure 3 , and divides the sources of demand into 10 parts. The different sources of demand include 0101 operator capability self-assessment, 0102 training assessment results, 0103 internal and external corrective actions, 0104 company management expectations, 0105 supervision requirements, 0106 industry recommended standards, 0107 internal and external experience feedback, 0108 design changes, 0109 working condition training and 0110 procedure training requirements.

[0067] The training evaluation results can also be called the training summary report. The procedure training requirements can also be called the procedure rotation training system.

[0068] For the self-assessment of the 0101 operator's ability: a self-assessment template is designed based on the operator's job ability requirements, including routine operations, transient faults, accident operations, basic skills weaknesses, knowledge point weaknesses, etc. These data from the self-assessment of trainees are very valuable and reflect the trainees' self-needs. At the same time, the demand data in this part has the highest quality and can be designed according to the items in the feature dimension. The word embedding matching degree is the highest and it is easy to obtain and analyze.

[0069] Regarding the 0102 training evaluation results: the weaknesses of individuals and teams identified during the training process mainly include three aspects: weak basic skills dimension, weak operation points and weak events, and the handling evaluation score of this event is calculated based on the evaluation of each item. The report is sorted based on the average score of each observation point of each event, and the 10 items with the lowest average score of each type of event, the 10 items with the lowest score in the observation operation points, and the three items with the lowest score in the basic skills dimension and the three items with the lowest secondary dimension in each basic skill are taken, totaling 9 items (the data range and quantity are adjusted according to actual needs) as part of the demand data. This part reflects the overall weaknesses of the trainees during the entire training process. The statistical results are carried out according to the characteristic dimensions, and the statistical results are most accurate.

[0070] For 0103 internal and external corrective actions: collect requirements for internal and external corrective actions, including actions related to operator training generated when various assessment activities are carried out inside and outside the nuclear power plant, collect these actions, clarify the format (training requirements + frequency) and manage them in the information system, and carry out relevant training regularly according to the action requirements. The action clearly defines the requirements and time frequency for implementation, and the description of the requirements should be as close as possible to the description of the feature dimension to facilitate data coding and embedding assignment. The frequency of demand implementation determines whether it appears in this demand analysis. Once it appears, the necessity is assigned a value of 1.

[0071] For 0104 company management expectations: operator training must be supported by the company's leadership and management, and management expectations of operators and management should be collected regularly, including meeting instructions, observation guidance and other requirements, and expectations should be implemented in training events and observation points to strengthen the implementation of management expectations. Management expectations should be pre-processed when entering the information system, and described according to the recommended statements of the feature dimensions to facilitate data coding, embedding and assignment.

[0072] Regarding the 0105 supervision requirements: Nuclear power plant operator training must comply with national laws, regulations, and the rules and regulations of the competent and regulatory units. Collect the requirements of the superior units for operator training and implement them in specific training lesson plans to avoid the risk of violations of power plant operator training. Clarify the format (training requirements + frequency) and manage it in the information system. Carry out relevant training regularly according to the superior supervision requirements. The description of the requirements should be as close as possible to the description of the feature dimension to facilitate data coding and embedding. The frequency of demand development determines whether it appears in this demand analysis. Once it appears, the necessity is assigned a value of 1.

[0073] For the 0106 industry recommended standards: The World Association of Nuclear Operators (WANO) and the China Nuclear Energy Association have produced a number of nuclear power guidelines and standards after years of accumulation and summary, providing standard references and good practices for nuclear power operation. During the training of operators, good practices should be absorbed and added to the training lesson plan, and the advanced experience of the industry should be fully learned. The format (training requirements + frequency) should be clarified during entry and managed in the information system. Relevant training should be carried out regularly according to the requirements of the industry association, and the description should be as close as possible to the description of the feature dimension to facilitate data encoding, embedding and assignment.

[0074] Regarding 0107 internal and external experience feedback: In the nuclear power industry, experience feedback must be conducted frequently, and continuous summarization and improvement must be made to prevent the recurrence of incidents. Experience feedback from inside and outside the nuclear power plant should be collected, and various event conditions in training should be effectively combined with experience feedback. While improving the technical capabilities of operators, rapid experience feedback training should be carried out to improve operators' safety awareness and risk awareness. The format (training requirements + frequency) should be clarified during input and managed in the information system. The description should be as close as possible to the description of the feature dimension (event or operation point dimension) to facilitate data coding, embedding and assignment.

[0075] For 0108 design changes: collect design change data of nuclear power plant simulator reference units, sort out changes that require simulator training according to the impact of the design changes, and upgrade the training lesson plan according to the changes. When other information systems use design change data, specify the format and require the description to be as close to the description of the feature dimension (event or knowledge point dimension) as possible to facilitate data encoding, embedding and assignment.

[0076] For 0109 working condition rotation training: collect various working conditions of safety analysis in the nuclear power plant FSAR (Final Safety Analysis Report), establish a rotation training mechanism for various working conditions, and ensure that nuclear power plant operators can deal with various working conditions in the design proficiently. When entering, clarify the format (training requirements + frequency) and manage it in the information system. Carry out relevant training regularly according to FSAR requirements, and require the description to be as close as possible to the description of the feature dimension (event or procedure dimension) to facilitate data coding, embedding and assignment.

[0077] Requirements for 0110 regulation training: When the regulations are upgraded or newly released, the standards for simulator training should be established to improve the proficiency of operators in using the regulations and effectively respond to design conditions. When entering the information system, the format should be clarified and the description should be as close as possible to the description of the feature dimension (regulation or knowledge point dimension) to facilitate data coding, embedding and assignment.

[0078] Different collection methods are used to obtain information according to different demand sources. Figure 2 and Figure 3, different collection methods mainly include 0201 questionnaire survey, 0202 external system input, 0203 training report conversion and 0204 self-built database. 0201 Questionnaire survey mainly collects data on 0101 operator capability self-assessment and 0104 company management expectations. 0202 External system input mainly collects data on 0107 internal and external experience feedback and 0108 design changes. 0203 Training report conversion mainly converts 0102 training evaluation results. 0204 Self-built database mainly collects data on 0103 internal and external corrective actions, 0105 supervision requirements, 0106 industry recommended standards, 0109 working condition rotation training and 0110 procedure training requirements.

[0079] For the 0201 questionnaire survey method: all operators and management of the operation department are required to conduct self-assessment according to job requirements. The survey content is designed according to job requirements and statistically summarized after collection.

[0080] For 0202 external system input: form an interface with other self-built information software within the company, extract key information from other information software, and introduce it into operator training. Mainly for the collection of experience feedback 0107 and 0108, considering the construction of other information software in the later stage, expand the number of interfaces according to actual conditions. Clarify the format during information interaction and describe the description close to the feature dimension to facilitate data encoding, embedding and assignment.

[0081] For the conversion of 0203 training reports: the company will regularly summarize and evaluate the training and capabilities of operators. Through these reports, the training needs 0102 of operators are extracted, and the weak dimensions and weak events are collected and summarized into the training needs data matrix for analysis. When collecting data, the format should be clear and the description should be close to the description of the characteristic dimension to facilitate data coding, embedding and assignment.

[0082] For 0204 self-built database: for sources 0103, 0105, 0106, 0109, and 0110, which have fixed numbers, small amounts of data, and low update rates, self-built information databases and established regular management systems to supplement and improve the sources of training needs to ensure the authenticity and reliability of data. Since the contents of the self-built database are clear in quantity and objectives, instructors compile the demand database in the characteristic dimension format, which can be more efficiently applied to training needs analysis during data entry.

[0083] 030 is to establish the characteristic dimension attribute and characteristic vector library, and it can also be used to build the characteristic dimension database of demand data. The characteristic dimension database is the key data of this method, and how to match the characteristic dimension with the demand is the core of this method. Through the accumulation and research of industry experience, the six key elements of lesson plan development are discovered: events, procedures, knowledge points, basic skills, special requirements and key points of operation. On this basis, the relationship between training needs and training lesson plans is constructed, and the relationship between training needs and training lesson plans is opened up to solve the problem of insufficient accuracy, pertinence and comprehensiveness of the business source of the training lesson plan development for nuclear power plant operators.

[0084] In at least one embodiment of the present application, S2.2.1 to S2.2.6 are a specific implementation of the above S2.2.

[0085] S2.2.1: Based on the event feature dimension of the demand data, the event requirements for training are divided into three categories: normal operating conditions, expected operating event work, and accident conditions. The event vector database is established by embedding the features of the text description. (Corresponding to 0301)

[0086] It should be noted that each of the three categories includes dozens to hundreds of types.

[0087] S2.2.2: For the procedure feature dimension of the demand data, embed the name and code of the procedure according to the list of existing procedures, and establish a procedure vector database. (Corresponding to 0302)

[0088] It should be noted that operators need to follow various procedures to intervene in the unit and are very familiar with the procedures, especially when training is required after the procedures are upgraded or added.

[0089] S2.2.3: Based on the knowledge point feature dimensions of the demand data, organize the knowledge and skills of the operators, establish a database, embed the names of key knowledge points, and establish a knowledge point vector database.

[0090] (corresponding to 0303)

[0091] For example, the operator's knowledge and skills include understanding of basic principles, understanding of system equipment characteristics, familiarity with management documents, and application of technical specifications.

[0092] S2.2.4: Based on the basic skill feature dimensions of the demand data, embed the basic skills and detailed dimensions to establish a basic skill vector database. (Corresponding to 0304)

[0093] It should be noted that in industry standards, the basic skills of operators refer to the various dimensions of ability demonstrated by operators in the process of controlling the unit, which are mainly divided into monitoring, control, conservation, teamwork, knowledge, and behavioral norms. In external evaluations or industry standards, basic functional capabilities are generally used to evaluate the performance level of operators. Therefore, these six basic skills need to be observed and evaluated during training to improve the basic functional capabilities of operators. Each training scenario focuses on different basic skills. By identifying the basic weaknesses of operators, clear guidance can be provided for the setting of scenarios.

[0094] S2.2.5: Based on the special requirement feature dimensions of demand data, embed special requirements and establish a special requirement vector database. (Corresponding to 0305)

[0095] It should be noted that during the training, the instructors play various roles to respond to the operators' instructions, including fault introduction methods, telephone interference, wrong instructions, misleading information, etc., and meet the training requirements through special operations. In some corrective actions and management expectations, there will be special requirements for training to improve the operators' ability to deal with abnormalities.

[0096] S2.2.6: Based on the characteristic dimension of the key operation points of the demand data, embed the names of the key operation points and establish a key operation point vector database. (Corresponding to 0306)

[0097] It should be noted that the simulator training process for operators requires a large number of response actions, each of which is an observation point. The difficulty of these observation points varies, and some require multiple exercises to be proficient. The training process is to discover these weaknesses and improve them, so it is very necessary to include these operation points in the training requirements for analysis and collation.

[0098] 040 Feature vector matching and assignment of demand data. Embed the demand data to obtain the vector library of demand data, compare the similarity of the demand data vector library with the feature dimension vector item by item, assign 1 if the conditions are met, and assign 0 if the requirements are not met, and obtain a feature dimension vector matrix with the feature dimension as the column vector and 0 and 1 as the row vector.

[0099] At least one embodiment of the present application further provides a computer device, which includes a processor and a memory. The processor is used to execute the nuclear power plant operator simulator training needs analysis method provided in any of the above embodiments of the present application. The memory is used to store executable instructions of the processor.

[0100] At least one embodiment of the present application further provides a computer-readable storage medium on which computer executable instructions are stored. When the executable instructions are executed by a processor, the method for analyzing the training needs of a simulator for nuclear power plant operators provided in any of the above embodiments of the present application is implemented.

[0101] The fourth aspect of the present application provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the nuclear power plant operator simulator training needs analysis method provided in any of the above embodiments of the present application.

[0102] It should be noted that the combination of the various technical features in the embodiments of the present application is not limited to the combination described in the embodiments of the present application or the combination described in the specific embodiments, and all technical features described in the present application can be freely combined or combined in any way unless there is a contradiction between them.

[0103] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the term "comprising" only indicates that the steps and elements that have been clearly identified are included, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0104] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for analyzing the training needs of simulators for operators in nuclear power plants, characterized in that: include: The discrete partition mode is used to collect and manage the simulator training needs of operators to obtain demand data. Different collection methods are used for different demand sources in the discrete partition mode. The demand data is preprocessed by embedding words according to the feature dimensions, where the feature dimensions include events, procedures, knowledge points, basic skills, special requirements, and key points of operation; Generate results of training needs through cluster analysis.

2. The method for analyzing training needs of simulator for nuclear power plant operators according to claim 1, characterized in that: The discrete partition mode is used to collect and manage the simulator training needs of operators to obtain demand data, including: Create a partition matrix based on the type of demand data, encode each type of data, and define the demand source type as XI to form a partition vector of the demand data, A = [X1, X2, X3...Xn], where n is the type of source; For each partition vector XI, there are multiple data. The generated row vector is expanded to form a discrete data matrix vector D. n is the type of demand data source, and m is the number of each type of data in the demand data source.

3. The method for analyzing training needs of simulator for nuclear power plant operators according to claim 1, characterized in that: The word embedding preprocessing of the demand data according to the feature dimension includes: Establish a feature matrix, define attribute features yi for each data item, form a feature matrix T, and use feature vectors to clean and assign values ​​to demand sources; According to the description of the recommendation sentence in the feature dimension, the feature matrix T is transformed into a vector using word embedding technology to establish a feature vector database; After data collection, each item in the discrete data matrix vector D is also word embedded to obtain the feature vector of the demand data, and the cosine similarity is calculated with each feature vector in the feature vector database. When the cosine similarity reaches 0.95, a feature is configured for the data of the discrete data matrix vector D and assigned a value of 1. Necessity is used as an independent feature for statistical calculation after data analysis.

4. The method for analyzing training needs of simulator for nuclear power plant operators according to claim 3 is characterized in that: The described recommendation sentence description according to the feature dimension uses word embedding technology to transform the feature matrix T into a vector and establish a feature vector database, including: According to the event feature dimension of demand data, the event requirements of training are divided into three categories: normal operating conditions, expected operating event work and accident conditions. The features of text description are embedded to establish an event vector database. In view of the procedure feature dimension of the demand data, according to the sorted list of existing procedures, the name and code of the procedure are embedded to establish a procedure vector database; According to the knowledge point feature dimension of the demand data, the knowledge and skills of the operators are sorted out, a database is established, the names of key knowledge points are embedded, and a knowledge point vector database is established; Based on the basic skill feature dimensions of the demand data, the basic skills and detailed dimensions are embedded to establish a basic skill vector database; According to the special requirement feature dimension of demand data, special requirements are embedded and a special requirement vector database is established; According to the characteristic dimension of key operation points of demand data, the names of key operation points are embedded and a key operation point vector database is established.

5. The method for analyzing training needs of simulator for nuclear power plant operators according to claim 1, characterized in that: The results of generating training needs through cluster analysis include: The discrete data matrix vector D is nested with the attribute feature matrix T to obtain the training demand result matrix R. All the data in R are represented by the value 0 or 1 of the feature dimension. n is the type of demand data source, m is the number of data of each type in the demand data source, a is the feature dimension category, and b is the subdivision dimension of each feature dimension; Arrange the data of the demand result matrix R, obtain the characteristic distribution of the demand data, and replace the data code with the characteristic dimension; Use the business association relationship between feature dimensions to perform duplicate checking and deduplication on the result clustering statistics of each feature dimension; According to the business rule logical relationship between various feature dimensions, for any feature working condition, the assigned event and the procedure may both be 1, and there is an envelope relationship between the two. In this case, the event dimension of the data is set to 1, and the procedure dimension is retained as 0; the knowledge point and the basic skills are both 1, and there is a preferred relationship between the two. In this case, the basic skills dimension is set to 0, and the knowledge point dimension is retained as 1. The data that must be followed in the design events and procedure training, laws and regulations, and curriculum development management requirements in the data source is assigned a value of 1; The necessity is calculated based on the result of clustering deduplication. The necessity is defined as p, and the number of occurrences of the feature dimension is k. According to the formula Calculate the necessity, where R is the correction coefficient, which is a real number greater than 1, and make adjustments based on the actual effect; sort out and count the data in the demand result matrix that are not assigned 0, calculate the necessity for the same dimensions that appear repeatedly, arrange them according to the necessity of the characteristic dimensions of the demand data, and obtain a statistical table of the demand analysis results. Finally, obtain the demand analysis results for operator simulator training, and include the analysis results in the course training outline to guide the development of training lesson plans.

6. The method for analyzing training needs of simulator for nuclear power plant operators according to any one of claims 1 to 5, characterized in that: Different sources of demand include operator self-assessment of capabilities, training assessment results, internal and external corrective actions, company management expectations, supervision requirements, industry recommended standards, internal and external experience feedback, design changes, working condition rotation training and procedural training requirements.

7. The method for analyzing training needs of simulator for nuclear power plant operators according to claim 6, characterized in that: Different collection methods include questionnaire surveys, external system input, training report conversion and self-built databases. The questionnaire survey method mainly collects data on operator self-assessment of capabilities and company management expectations. External system input mainly collects data on internal and external experience feedback and design changes. Training report conversion mainly converts training evaluation results. Self-built databases mainly collect data on internal and external corrective actions, supervision requirements, industry recommended standards, working condition rotation training and procedural training requirements.

8. A computer device, characterized in that: include: A processor, configured to execute the method for analyzing training needs of a nuclear power plant operator simulator according to any one of claims 1 to 7; as well as A memory is used to store executable instructions of the processor.

9. A computer-readable storage medium having computer executable instructions stored thereon, characterized in that: When the executable instructions are executed by the processor, the method for analyzing training needs of a nuclear power plant operator simulator according to any one of claims 1 to 7 is implemented.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the method for analyzing training needs of a nuclear power plant operator simulator according to any one of claims 1 to 7 is implemented.

Images