Personnel comprehensive ability quantitative evaluation method, system and equipment of new energy station and medium

Abstract

The invention provides a personnel comprehensive ability quantitative evaluation method, system and device for a new energy station and a medium, and belongs to the technical field of new energy stations. Four types of first-level indexes of duty performance, technology, learning and reward and punishment and twelve second-level indexes are defined; connecting multiple systems to acquire data and performing structured storage; carrying out abnormal value elimination, missing value filling and Min-Max standardization preprocessing; extracting a core feature with a weight greater than or equal to 0.05 by adopting a random forest; determining a first-level index weight coefficient through SVM (Support Vector Machine) model training; calculating a comprehensive score and outputting a quantitative score, a radar map and a structured portrait; and finally, a result is fed back to a management system and is applied to work order assignment, training matching and other scenes. Scientific evaluation of personnel ability from subjective qualitative to objective quantitative is realized, the weight is dynamically optimized through machine learning, the method adapts to operation and maintenance characteristics of a new energy station, and the evaluation accuracy is improved; the visualization result visually presents the personnel state, and the operation and maintenance efficiency and safety of the station are effectively improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of new energy station, and particularly relates to a personnel comprehensive capability quantitative evaluation method, system, device and medium for a new energy station. BACKGROUND

[0002] The new energy industry is a key field for achieving the goal of "carbon peak and carbon neutrality", and has formed a diversified development pattern mainly based on wind power and photovoltaic power. New energy stations are transforming from manual operation and maintenance to intelligent and unmanned operation and maintenance, and can be applied to unmanned inspection and maintenance. The industry also increasingly requires the capabilities of operation and maintenance personnel, who not only need to master traditional operation and maintenance skills, but also need to have new capabilities such as data analysis, remote monitoring, and intelligent device operation.

[0003] In related technologies, the evaluation method for operation and maintenance personnel follows the attendance rate, basic operation proficiency, and team cooperation of personnel in industrial enterprises, and does not include the dimensions of operation and maintenance response time, multi-system linkage fault work order processing capability, and wind power / photovoltaic certification in the evaluation, and only focuses on basic indicators.

[0004] Related technologies use expert subjective setting of fixed weights, and do not consider the dynamic changes of business priorities in new energy stations. Once the weights are set unchanged, it is unable to respond to business iteration, such as the need to improve the technical score weight related to the fault solution capability of the energy storage battery after the addition of an energy storage system in the station, and it is unable to meet the evaluation requirements and the operation and maintenance requirements on site. SUMMARY

[0005] The application provides a personnel comprehensive capability quantitative evaluation method for a new energy station, which realizes multi-dimensional quantitative evaluation of personnel capability and improves the objectivity and accuracy of the evaluation.

[0006] The method comprises the following steps: S101: defining a new energy station personnel comprehensive capability evaluation index system comprising four first-level indexes of duty performance score, technical score, learning score, and punishment record, and twelve corresponding second-level indexes; S102: collecting various types of data related to the index system of S101 by interfacing with a personnel management system, and storing the data in the format of employee ID-index category-index subclass-time-value; S103: performing outlier rejection filling, missing value filling, and Min-Max standardization on the stored data to generate a monthly raw data matrix; S104: extracting key features based on the raw data matrix, and selecting and retaining core features with a weight greater than or equal to 0.05 to form a feature vector; S105: training an evaluation model using the feature vector to determine the weight coefficients of the four first-level indexes in S101; S106: Calculate the personnel comprehensive score according to the weight coefficient and the standardized score of each first-level index; S107: Output the quantitative score, the ability radar chart and the structured personnel portrait as the evaluation result based on the comprehensive score; S108: Feedback the evaluation result to the personnel management system and apply it to the actual management scenarios such as operation and maintenance work order assignment and short board training matching.

[0007] Preferably, step S102 specifically includes the following manner: Determine the list of systems to be connected and the specific data items to be provided, and establish the mapping relationship between the data items and the evaluation index system; Configure the data interface permission of the personnel management system, send data extraction requests at a predetermined period, and obtain technical and reward and punishment type raw data and store them temporarily; Configure the API interface parameters of the operation and maintenance management system and the training management system, obtain the performance and learning type raw data through API calling and store them temporarily; Based on the pre-set mapping relationship table, classify all the temporarily stored raw data into the corresponding four first-level indexes and twelve second-level indexes according to their contents; Format convert the classified data according to the unified field structure, and write it into the database for storage.

[0008] Preferably, step S104 specifically includes the following manner: S1041: Extract the twelve second-level index standardized data of all employees from the pre-processed monthly raw data matrix, construct a feature matrix, and form a label vector combined with the expert score results; S1042: Set the number of decision trees, maximum depth, node splitting standard and minimum sample size of leaf nodes and other key parameters of the random forest model; S1043: Train the random forest model using the feature matrix and the label vector, and calculate the average importance weight of each second-level index feature; S1044: Compare the calculated average importance weight of each feature with the pre-set threshold value, and select the core second-level index feature; S1045: Extract the standardized data corresponding to the selected core feature from the monthly raw data matrix to generate a core feature vector.

[0009] Preferably, step S1043 specifically includes the following manner: Classify the twelve second-level indexes according to the four first-level indexes they belong to, and establish a four-index group mapping table for performance, technology, learning and reward and punishment; Randomly select two different index groups for each decision tree, and only extract candidate splitting features from the second-level indexes in the selected groups for node splitting; Record the original contribution of the features used for node splitting, and conditionally adjust the contribution based on the number of employee samples in the splitting node; Calculate the statistical range of the contribution of each secondary indicator in all decision trees, and remove abnormal contribution data that exceed the reasonable range; The processed contribution data are summed and multiplied by the preset weighting coefficient of the corresponding indicator group to obtain the final total contribution value of each secondary indicator.

[0010] Preferably, step S105 specifically includes the following methods: The core features are grouped according to the first-level indicator categories from the core feature vector, forming four feature subsets: performance, technology, learning, and rewards and punishments. Configure the kernel function type and parameter value range of the support vector machine model to prepare for model training; A multi-round cross-validation method is adopted, and the support vector machine model is trained and its parameters are optimized using core feature vectors and label vectors; The optimized model is used to extract the core feature weights, which are then aggregated and normalized to obtain the weight coefficients of the four primary indicators. Periodically merge newly added data and re-execute the training process to generate updated primary indicator weight coefficients to replace the original coefficients.

[0011] Preferably, step S106 specifically includes the following methods: The standardized scores of four primary indicators for each employee are extracted from the preprocessed monthly raw data matrix to form a data table containing employee identification, time period, and scores for each indicator. Obtain the weight coefficients of the four primary indicators in the trained evaluation model. After verifying that the sum of the weight coefficients is 100%, associate them with the data table to generate an extended data table containing the weight coefficients. For each record in the extended data table, the standardized score of each primary indicator is multiplied by the corresponding weight coefficient and then summed to calculate the employee's monthly comprehensive score. The monthly comprehensive scores of employees are summarized periodically, and the quarterly and annual comprehensive scores are calculated using the arithmetic mean method. Set up comprehensive score verification rules to check whether the monthly, quarterly, and annual comprehensive scores are within the range of zero to one hundred. Mark and recalculate abnormal scores, and store them in the database after confirming their validity.

[0012] Preferably, step S107 specifically includes the following methods: From the overall employee score calculated by S106, extract the overall score value of each employee, and combine it with the employee ID and evaluation period information to generate a quantitative score record; Standardize the scores of the four primary indicators in the quantitative score record, convert them into coordinate data for a radar chart, and generate the radar chart data structure. Based on the primary indicator scores and preset label rules in the radar chart data structure, capability labels are assigned to employees to generate structured personnel profiles. Check the accuracy of the quantitative score records, radar chart data structure, and structured personnel profile calculations and their compliance with the rules, and make corrections accordingly; The revised quantitative score records, radar chart data structure, and structured personnel profiles are stored in the database and linked to the basic employee information.

[0013] This application also provides a quantitative evaluation system for the comprehensive capabilities of personnel at new energy power stations, the system comprising: The indicator definition module is used to define a comprehensive evaluation indicator system for personnel at new energy power stations, which includes four primary indicators: performance score, technical score, learning score, and reward and punishment records, as well as twelve corresponding secondary indicators. The data acquisition and storage module is used to collect various types of data related to the indicator system by connecting to the personnel management system, and store the data in the format of employee ID-indicator category-indicator subcategory-time-value. The data preprocessing module is used to perform outlier removal and filling, missing value filling, and Min-Max standardization on the stored data to generate a monthly raw data matrix; The feature extraction module extracts key features based on the original data matrix, and selects and retains core features with a weight ≥ 0.05 to form feature vectors; The model training module is used to train the evaluation model using feature vectors and determine the weight coefficients of the four primary indicators. The comprehensive score calculation module is used to calculate the comprehensive score of personnel based on the weighting coefficients and the standardized scores of each primary indicator. The evaluation results generation module outputs quantitative scores, capability radar charts, and structured personnel profiles as evaluation results based on the comprehensive score. The results feedback module is used to feed the evaluation results back to the personnel management system, and is applied to actual management scenarios such as operation and maintenance work order assignment and gap training matching.

[0014] According to another embodiment of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the method for quantitatively evaluating the comprehensive capabilities of personnel at the new energy power station.

[0015] According to another embodiment of this application, a storage medium is also provided, on which a computer program is stored, wherein when the computer program is executed by a processor, the steps of the method for quantitatively evaluating the comprehensive capabilities of personnel at the new energy power station are implemented.

[0016] As can be seen from the above technical solutions, the present invention has the following advantages: This invention provides a quantitative evaluation method for the comprehensive capabilities of personnel in new energy power stations. By defining unique indicators specific to new energy power stations, the evaluation accurately captures the main capabilities of maintenance personnel, rather than relying solely on non-critical dimensions such as attendance rate and basic skills. It integrates with multiple systems, storing data in an employee ID-indicator-time-value format, reducing the workload of data collection and preprocessing. Employing different processing strategies, compared to general preprocessing, reduces data distortion, making the standardized monthly raw data matrix more closely reflect actual maintenance scenarios. The method uses an SVM model to determine the weights of primary indicators and updates these weights quarterly with new data, addressing the problem of fixed weights in existing technologies that cannot adapt to the iterative development of new energy power station operations, ensuring that weights always reflect current maintenance priorities. It outputs quantitative scores, capability radar charts, and structured personnel profiles to meet diverse management needs. Quantitative scores are used for personnel ranking and performance evaluation, while radar charts help managers quickly identify key personnel and provide a basis for targeted management. Attached Figure Description

[0017] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 Flowchart of a quantitative evaluation method for the comprehensive capabilities of personnel at new energy power plants; Figure 2 A schematic diagram of a quantitative evaluation system for the comprehensive capabilities of personnel at new energy power plants; Figure 3 This is a schematic diagram of an electronic device. Detailed Implementation

[0019] This invention provides a quantitative evaluation method for the comprehensive capabilities of personnel at new energy power stations. By integrating multi-dimensional data such as performance checklist scores, on-site inspection scores, safety scores, technical scores, training and examination scores, and reward and punishment records, it constructs a comprehensive evaluation index suitable for personnel management scenarios at new energy power stations. This index includes four primary indicators and twelve secondary indicators. Random forest feature extraction is used, and an SVM model is constructed to output evaluation coefficients and scores. Personnel capabilities and weaknesses are visualized using radar charts and personnel profiles, providing data support for matching maintenance work orders with personnel and optimizing training courses. As the system accumulates data and technology iterates, the model can be continuously trained and the evaluation coefficients optimized. This method can be widely applied to scenarios such as recruitment and selection of personnel at new energy power stations, skills training planning, project team building, and performance appraisal optimization.

[0020] The following will describe in detail the method for quantitatively evaluating the comprehensive capabilities of personnel in new energy power stations involved in this application. Specific details such as particular system structures and technologies are presented for illustrative purposes rather than limiting, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details.

[0021] It should be understood that, when used in this specification, terms include indicating the presence of a described feature, integral, step, operation, element, and / or component, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The terms include, encompass, have, and variations thereof mean including but not limited to, unless otherwise specifically emphasized.

[0022] The statements such as "one embodiment" or "some embodiments" described in this application mean that one or more embodiments of this application include the specific features, structures, or characteristics described in that embodiment. Therefore, the statements such as "in one embodiment," "in some embodiments," "in other embodiments," and "in still other embodiments" in this application do not necessarily refer to the same embodiment, but rather mean one or more, but not all, embodiments, unless otherwise specifically emphasized.

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Please see Figure 1 The diagram shows a flowchart of a method for quantitatively evaluating the comprehensive capabilities of personnel at a new energy power station, as described in a specific embodiment. The method includes: S101: Defines a comprehensive evaluation index system for personnel at new energy power stations, which includes four primary indicators: performance evaluation, technical evaluation, learning evaluation, and reward and punishment records, as well as twelve corresponding secondary indicators.

[0025] In some embodiments, when four primary indicators and subordinate secondary indicators are defined, for the performance evaluation, the operation and maintenance response time in the secondary indicators is set to ≤10 minutes as the full score benchmark, and the complexity of operation and maintenance work orders is divided into 4 levels according to the fault type and corresponding to different scores.

[0026] In the technical evaluation, the secondary indicators of professional titles and certificate levels are linked to maintenance engineers, while the fault resolution success rate focuses on the quality of equipment repairs. In the learning evaluation, the score for practical application of new technologies is linked to training courses on wind turbine pitch system maintenance and photovoltaic module fault diagnosis. In the reward and punishment record, penalties for violations are differentiated by severity, with a focus on violations of safety regulations.

[0027] This embodiment breaks down the evaluation dimensions and transforms each dimension into a quantifiable secondary indicator, forming an indicator system that covers the entire operation and maintenance process, ensuring that the indicators are directly related to the actual business scenarios of the site.

[0028] S102: By connecting to the personnel management system, collect various types of data related to the indicator system described in S101, and store the data in the format of employee ID-indicator category-indicator subcategory-time-value.

[0029] In some embodiments, the system acquires the planned and actual number of inspection tasks, the order acceptance and closure times of maintenance work orders, and work order fault descriptions. The training management system provides attendance records for quarterly technical sharing sessions and individual assessment scores for new energy technology training. All collected data is organized in a fixed format: employee ID - indicator category (e.g., performance rating) - indicator subcategory (e.g., maintenance response time) - time (accurate to monthly) - value (e.g., 8 minutes, 90 points). The numerical fields must match the quantification rules of the secondary indicators.

[0030] Optionally, the format can be organized into a structured category such as employee ID - performance score - operation and maintenance response time - accurate to the month - 8 minutes or 90 points, to ensure clear data ownership and uniform format.

[0031] S103: Perform outlier removal and filling, missing value filling, and Min-Max standardization on the stored data to generate the monthly raw data matrix.

[0032] In some embodiments, for outlier handling, if the maintenance response time is greater than 60 minutes, it is removed, and the arithmetic mean of the employee's three most recent non-outlier response times is calculated to fill the missing value. For missing value handling, when the completion rate / closure rate of performance data is missing, the corresponding indicator data of all employees in the same position during the same period is extracted and the average is calculated to fill the missing value. For learning data, courses for which the employee did not participate are filled with the passing grade. When the complexity of a work order cannot be determined, it is defaulted to the general level. Data standardization uses the Min-Max formula, mapping all secondary indicator data to the [0, 100] range.

[0033] For example, when the minimum operation and maintenance response time is 5 minutes and the maximum is 60 minutes, the standardized score of an employee's 7-minute response time is (60-7) / (60-5)×100=96.36 points; the final monthly raw data matrix uses employee ID + month as the key fields, with each row corresponding to the data of one employee in the current month, and each column corresponding to the standardized score of one secondary indicator, which is associated with the primary indicator category to ensure that the data meets the input requirements for subsequent feature extraction and model training.

[0034] S104: Extract key features based on the original data matrix, and select and retain core features with a weight ≥ 0.05 to form a feature vector.

[0035] In some embodiments, from the monthly raw data matrix, four primary indicators are decomposed into 12 secondary indicators, and a feature matrix is ​​constructed with employee samples as rows and the 12 secondary indicators as columns. Each employee is independently scored based on their fault resolution efficiency and work order processing quality, and the arithmetic mean of three scores is taken as the employee's label y. When training the random forest model, 100 decision trees are set, with a maximum depth of 8 layers, and the node splitting criterion is the Gini coefficient. The importance weights of the 12 secondary indicators are calculated based on the contribution of each decision tree to feature splitting. Finally, features with weights ≥ 0.05 are retained, generating feature vectors containing only core features, thus improving model training efficiency and the accuracy of evaluation results.

[0036] S105: Use eigenvectors to train the evaluation model and determine the weight coefficients of the four primary indicators in S101.

[0037] In some embodiments, an SVM model is selected, using the RBF kernel function. The penalty coefficient C is set to a range of [1, 100], and the kernel parameter γ is set to a range of [0.01, 10]. Parameters are optimized using 5-fold cross-validation. Optionally, the dataset is divided into 5 mutually exclusive subsets: 4 for training and 1 for validation, repeated 5 times. Parameter combinations with an average prediction error ≤ 5% are selected, such as C = 10 and γ = 0.1. After training, the core features are categorized according to their respective primary indicators. The sum of the weights of all core features under each primary indicator is calculated as the initial weights. Then, normalization is applied to ensure that the sum of the weight coefficients of the four primary indicators is 1, such as 0.31 for performance, 0.36 for technical skills, 0.23 for learning, and 0.1 for rewards and punishments. Core feature data and expert scores of newly added employees are collected quarterly. The model is retrained and the weight coefficients are adjusted to ensure that the weights adapt to business changes.

[0038] Compared to the linear model, the SVM with the RBF kernel function is more suitable for the complex relationships between the capability dimensions of personnel in new energy power stations. The 5-fold cross-validation ensures that the model parameters are optimal, and the dynamic update of weights avoids the evaluation lag caused by fixed weights, so that the indicator weights always conform to the actual priority of power station operation and maintenance.

[0039] S106: Calculate the comprehensive score of personnel based on the weighting coefficients and the standardized scores of each primary indicator.

[0040] In some embodiments, when calculating the monthly comprehensive score, the standardized scores of the four primary indicators for the employee in that month are extracted. Optionally, these are: performance score (85 points), technical score (90 points), learning score (75 points), and rewards / penalties score (90 points). The comprehensive score is calculated as follows: Comprehensive Score = Performance Score × Performance Weight + Technical Score × Technical Weight + Learning Score × Learning Weight + Rewards / Punishments Score × Rewards / Punishments Weight, and is rounded to one decimal place.

[0041] It should be noted that the quarterly comprehensive score is the arithmetic mean of the scores of the three months in that quarter, and the annual comprehensive score is the arithmetic mean of the scores of the four quarters in that year, both rounded to one decimal place. After the calculation is completed, the score is checked to see if it is within the range of 0-100. If it is outside the range, return to S103 to re-check the preprocessed data and recalculate until the score is within a reasonable range, to ensure that the comprehensive score is true and reliable.

[0042] S107: Based on the comprehensive score, output quantitative scores, capability radar charts, and structured personnel profiles as evaluation results.

[0043] In some embodiments, when quantifying scores, the output is organized by employee ID, time period, overall score, and score level. When creating the capability radar chart, four primary indicators are used as coordinate axes, with each employee's corresponding period's primary indicator score as the coordinate point. These are connected by lines to form a closed polygon. Monthly radar charts are distinguished by blue, quarterly by red, and annual by black. The structured personnel profile includes four dimensions: employee ID, name, position, the highest-scoring primary indicator and average score for a preset time period, the lowest-scoring primary indicator and average score for a preset time period, and the rate of change of the overall score for a preset time period. A rate of change >5% indicates an increase, -5% to 5% indicates stability, and <-5% indicates a decrease. In this way, the quantified scores meet the needs for rapid ranking and comparison, the radar chart enables managers to identify employee strengths and weaknesses, and the structured personnel profile comprehensively reflects the current state and development direction of employee capabilities.

[0044] S108: Feedback the evaluation results to the personnel management system and apply them to actual management scenarios such as work order assignment and gap training matching.

[0045] In some embodiments, when assigning maintenance work orders, the complexity of the work order is identified, and employees with a comprehensive score of ≥80 are given priority for assignment. When matching training for areas of weakness, if an employee's learning score is <70, new energy technology-specific courses are selected from the training course library and pushed to the employee's training task list. A monthly evaluation result application report is generated, which counts the percentage of high-scoring employees handling complex work orders and the training participation rate of employees with learning weaknesses. By deeply integrating the evaluation with actual maintenance management, high-scoring employees handling complex work orders can reduce equipment downtime due to failures, and precise training for employees with learning weaknesses can improve the overall capabilities of the maintenance team, forming a management closed loop and promoting a two-way improvement in the efficiency of new energy power plant maintenance and personnel capabilities.

[0046] In one embodiment of the present invention, based on step S102, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner. Step S102 specifically includes the following methods: S1021: Determine the list of systems to be connected and the corresponding data collection items. Based on the determined evaluation index system, establish the mapping relationship between the evaluation index and the system data items, clarify the data content and scope that each system needs to provide, and form the basis for the direction of data collection.

[0047] In some embodiments, the list includes a self-developed new energy power station personnel management system, a new energy power station operation and maintenance management system, and a new energy power station training management system. Data collection items are as follows: The personnel management system needs to provide technical data including professional titles and certificate levels, and reward and punishment data including safety production commendations and violation penalties; the operation and maintenance management system needs to provide performance data including inspection task completion rate, operation and maintenance response time, operation and maintenance work order closure rate, operation and maintenance work order complexity, and safety data; the training management system needs to provide learning data including new energy technology training assessment results, technical sharing session participation frequency, and new technology application practice scores.

[0048] S1022: Configure the access permissions of the personnel management system, set the data extraction cycle to monthly, extract raw data through this interface according to the technical and reward / punishment data collection items determined in S1021, and temporarily store the extracted raw data.

[0049] S1023: Submit an API access permission application to the New Energy Power Station Operation and Maintenance Management System and the New Energy Power Station Training Management System to obtain the API key and interface documentation; configure the API request parameters according to the interface documentation, including the employee ID range, data time range, and data item identifiers corresponding to each secondary indicator; establish an API data transmission channel with the Operation and Maintenance Management System and the Training Management System, send API requests according to the configured parameters, receive the original data returned by each system in the categories of duty performance, learning, and safety, and temporarily store the received original data.

[0050] In some embodiments, the request header, request employee ID range, time range, etc. are assembled according to the interface documentation requirements to generate a request; after sending the request, the response data returned by the system is received, and the response data is verified to include the employee ID, data time, and indicator value. If they are missing, the request is resent; if they are complete, the response data is retained.

[0051] S1024: Create a data classification rule table in the temporary database. This table records the correspondence between four primary indicators and twelve secondary indicators, as well as the mapping relationship between each secondary indicator and the data item identifiers of each system. Traverse the temporary data in the intermediate database, read the data item identifier field of each data entry, match it with the data item identifiers in the classification rule table, and classify the successfully matched data into the secondary indicator category under the corresponding primary indicator. The four primary indicators are performance evaluation, technical evaluation, learning evaluation, and reward / punishment records.

[0052] In some embodiments, the data classification and matching method is as follows: A three-level mapping rule base of data item identifier - secondary indicator - primary indicator is established; when traversing the temporarily stored data, the data item identifier of each data item is read and precisely matched with the data item identifier in the mapping rule base. Upon successful matching, the corresponding secondary and primary indicators are automatically associated, completing the data classification. This automatically associates raw data from different systems with the primary and secondary indicators in the evaluation indicator system, achieving a precise correspondence between data and indicators.

[0053] S1025: Define the field structure of the data, including employee ID, indicator category, indicator subcategory, time, and value; convert the secondary indicator data categorized in S1024 according to the above field structure. Specifically, convert the time field to year-month format and the value field to floating-point type; create a structured database table and write the converted data into the database table for storage.

[0054] The main category of indicators can be character type, with values ​​limited to performance evaluation, technical evaluation, learning evaluation, and reward and punishment records. The sub-category of indicators can be character type, with values ​​limited to twelve secondary indicator names.

[0055] In some embodiments, based on preset structured field specifications, the non-uniform format data after classification is standardized to establish a unified data storage method, ensuring that the data format meets the input requirements of subsequent data preprocessing stages and improving the overall efficiency of subsequent data processing stages.

[0056] In one embodiment of the present invention, based on step S104, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner. Step S104 specifically includes the following methods: S1041: Extract the standardized data of twelve secondary indicators for all employees from the monthly raw data matrix formed in S103. Construct a feature matrix X with the employee samples as rows and the twelve secondary indicators as columns. Independently score the comprehensive ability of each employee sample in the feature matrix X, and take the arithmetic mean of the scores of three experts as the label y of the employee sample to form a label vector.

[0057] S1042: Configure the parameters of the random forest model, where the number of decision trees is set to 100, the maximum depth of the decision trees is set to 8 layers, the node splitting criterion is the Gini coefficient, and the minimum number of samples in the leaf nodes is set to 2 to limit the complexity of a single decision tree.

[0058] In some embodiments, considering the multi-dimensional and easily overfitted characteristics of new energy power station indicators, parameters such as the number of decision trees, depth, splitting criteria, and number of leaf node samples in the random forest model are set to limit the model complexity. 100 decision trees can reduce the overfitting risk of a single decision tree through ensemble learning; a maximum depth of 8 layers can avoid the model overlearning data noise; and a minimum number of leaf node samples of 2 ensures that the leaf nodes have a certain degree of representativeness, achieving a balance between fitting ability and generalization ability.

[0059] S1043: Input the feature matrix X and label vector y constructed in S1041 into the configured random forest model and start model training; during the training process, record the contribution of each secondary indicator feature to the improvement of node purity when each decision tree splits, and calculate the total contribution value of each secondary indicator feature in 100 decision trees based on the contribution value. Divide the total contribution value by 100 to obtain the average importance weight of each secondary indicator feature.

[0060] In some embodiments, the method for calculating the importance weight of a feature can be based on the following: the importance of a feature in a single decision tree = Σ(the sample proportion of the split node of the feature × (the Gini coefficient of the node before split - the mean Gini coefficient of the child nodes after split)).

[0061] The average importance weight of a feature = the sum of the importance of the feature in 100 decision trees / 100; In this embodiment, a random forest model is trained, the Gini coefficient is used to measure the purity of the nodes, and the contribution of each feature to the improvement of node purity during the splitting process is calculated to obtain the importance weight of each feature.

[0062] S1044: Set the feature importance weight threshold to 0.05. Compare the average importance weight of the twelve secondary indicator features calculated in S1043 with the threshold. Retain secondary indicator features with an average importance weight greater than or equal to 0.05 and remove secondary indicator features with an average importance weight less than 0.05.

[0063] In some embodiments, a fixed feature importance weight threshold is set, the average importance weight of each feature is compared with the threshold, features with the required weight are retained, and redundant features with the required weight are removed, thereby reducing the computational load of the model and improving the training efficiency and accuracy of the model.

[0064] S1045: Extract standardized data corresponding to the core secondary indicator features retained in S1044 from the monthly raw data matrix of S103, classify them according to employee ID and month, and form a core feature vector for each employee sample consisting of standardized data of core features.

[0065] In some embodiments, core feature data is extracted from the preprocessed monthly raw data matrix and organized into vector form according to the employee sample dimension, forming an input data format suitable for subsequent model training. The core feature vector contains only key influencing factor data, which allows subsequent model training to focus more on indicators that play a major role in the comprehensive ability of personnel, thereby improving the relevance of model training.

[0066] In some specific embodiments, step S1043 specifically includes the following methods: S10431: Based on the primary indicator classification defined in S101, the twelve secondary indicators are divided into performance indicator groups, technical indicator groups, learning indicator groups, and reward and punishment indicator groups according to their respective primary indicators.

[0067] The performance-related indicator group includes four secondary indicators: inspection task completion rate, operation and maintenance response time, operation and maintenance work order closure rate, and operation and maintenance work order complexity. The technical indicator group includes two secondary indicators: professional title and certificate level score and fault resolution success rate. The learning indicator group includes three secondary indicators: new energy technology training and assessment results, frequency of participation in technical sharing sessions, and score of new technology application practice. The reward and punishment indicator group includes three secondary indicators: number of safety production commendations, number and severity of violation penalties, and emergency response rewards. An indicator group mapping table is established.

[0068] In some embodiments, based on the requirements for evaluating the capabilities of personnel at new energy power stations, the indicators are grouped according to their impact dimensions and importance levels on operation and maintenance services. This allows the subsequent decision tree training process to focus specifically on the key capability dimensions of the power station, avoiding the dilution of the contribution of key dimensions caused by the indiscriminate participation of indicators in general random forest training.

[0069] S10432: For each decision tree to be trained, before starting the node split, two different indicator groups are randomly selected from the indicator grouping mapping table. Only from the secondary indicators within the two selected indicator groups, 3-5 secondary indicators are randomly selected as candidate split features for the decision tree. The candidate features are restricted to come only from the selected indicator groups.

[0070] In some embodiments, during the initialization of each decision tree, a random number generator selects two different groups from four indicator groups. For example, the first tree selects a performance category plus a technical category, and the second tree selects a technical category plus a learning category. From each selected group, 1-3 secondary indicators are selected using sampling without replacement to form the candidate feature pool for that decision tree. Features are selected only from this pool for node splitting. This ensures that the feature selection for each tree covers at least two capability dimensions, meeting the evaluation requirements for personnel at new energy power stations to possess multi-dimensional capabilities.

[0071] S10433: When a node of a decision tree completes a split based on a candidate splitting feature, record the original contribution of that feature and the number of employee samples contained in that splitting node. If the number of employee samples in the splitting node is less than 5, then the original contribution of the feature is corrected and recorded in the feature contribution record table of that decision tree.

[0072] In some embodiments, after a node splits, the number of employee samples contained in the split node is first counted. If the node sample size is less than 5, the corrected contribution is calculated as: Corrected Contribution = Original Contribution × (Node Sample Size / Total Sample Size). If the node sample size is ≥ 5, the original contribution is directly retained. The original contribution is the difference between the node purity before splitting and the average purity of the child nodes after splitting, with purity measured by the Gini coefficient. This improves the reliability of the contribution data and makes the feature importance weights more closely reflect the actual ability distribution of most employees in new energy power plants.

[0073] S10434: After all 100 decision trees have been trained, for each secondary indicator, extract its contribution data in each of the 100 decision trees, calculate the standard deviation and mean of the contribution data of the indicator, and obtain the fluctuation coefficient; if the contribution of the indicator in a certain decision tree exceeds the range of mean - 2 × standard deviation to mean + 2 × standard deviation, then remove the contribution data of the indicator in that decision tree.

[0074] In some embodiments, by calculating the fluctuation of the contribution of the same indicator in different decision trees, extreme contribution data caused by abnormal tree structure can be identified and eliminated, ensuring that the contribution data participating in the calculation of the total contribution value has consistency and stability.

[0075] S10435: Accumulate the contribution data of each secondary indicator after removing abnormal data to obtain the initial total contribution value of the indicator; according to the operation and maintenance priority of new energy power stations, set a weighting coefficient of 0.35 for the duty performance indicator group, 0.35 for the technical indicator group, 0.2 for the learning indicator group, and 0.1 for the reward and punishment indicator group. Multiply the initial total contribution value of each secondary indicator by the weighting coefficient of its respective indicator group to obtain the final total contribution value.

[0076] In some embodiments, the contribution data after removing abnormal data is first accumulated to obtain the initial total contribution value; then, according to the preset group weighting coefficient, the final total contribution value is calculated as 8.55 × 0.35 ≈ 3.0; the weighting coefficient is set with reference to the statistical data of new energy power station operation and maintenance accidents.

[0077] It can be seen that by combining the priority of new energy power station operation and maintenance business, different weighting coefficients are set for different indicator groups, so that the final total contribution value of the core indicator group can be reasonably improved, and the contribution value of the auxiliary indicator group can be appropriately adjusted, ensuring that the total contribution value can reflect the importance level of the indicators in actual operation and maintenance.

[0078] In one embodiment of the present invention, based on step S105, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner. Step S105 specifically includes the following methods: S1051: Extract the primary indicator category to which each core feature belongs from the core feature vector generated in S104. Group the core feature vectors according to the primary indicator category to form a performance feature subset, a technical feature subset, a learning feature subset, and a reward and punishment feature subset. Each subset contains all core secondary indicator data under the corresponding primary indicator.

[0079] In some embodiments, based on the attribution relationship between core features and primary indicators, the core feature vector is split into four feature subsets corresponding to primary indicators, clarifying the correspondence between each subset and the primary indicators, providing a basis for subsequent weight aggregation, and ensuring that the weights of the primary indicators calculated subsequently can accurately reflect the overall influence of their subordinate core features.

[0080] S1052: Configure the Support Vector Machine (SVM) model parameters, select the RBF kernel function as the model kernel function, set the penalty coefficient C to a value range of [1, 100], and the kernel function parameter γ to a value range of [0.01, 10]. The model output is the predicted score of the person's comprehensive ability.

[0081] In some embodiments, the RBF kernel function is selected to handle the nonlinear relationships between features by setting the range of values ​​for the penalty coefficient C and the kernel parameter γ. The RBF kernel function can effectively capture the complex nonlinear correlations between the capability characteristics of personnel in new energy power stations, and setting the parameter range can avoid overfitting or underfitting the model caused by extreme parameter values.

[0082] S1053: Using the label vector y from S1041 as the target value and the core feature vector X' as the input value, the SVM model is trained and its parameters are optimized using the 5-fold cross-validation method. The dataset is randomly divided into 5 mutually exclusive subsets, and 4 subsets are selected as the training set and 1 subset is selected as the validation set. The training is repeated 5 times. After each training, the prediction error of the validation set is calculated, and the set C and γ with the smallest average prediction error is selected as the optimal parameters.

[0083] In some embodiments, a 5-fold cross-validation algorithm is involved, which divides the dataset into 5 equal subsets, uses 4 subsets for training and 1 subset for validation each time, and calculates the average error of 5 validations. Cross-validation reduces the impact of dataset partitioning bias on parameter selection, and selects the parameter combination with the smallest average prediction error as the optimal parameters.

[0084] S1054: Retrain the core feature vector using the optimized SVM model and extract the weights of each core feature output by the model; sum the weights of all core features under the same primary indicator to obtain the initial weight of that primary indicator; normalize the initial weights of the four primary indicators so that the sum of the weights is 1, and obtain the weight coefficients of the four primary indicators of performance, technology, learning, and rewards and punishments.

[0085] In some embodiments, core feature weights are extracted using an SVM model. Feature weights under the same primary metric are aggregated into the initial weights of that metric, and then normalized to obtain weight coefficients with a sum of 1. Normalization ensures the comparability of weights and guarantees the rationality of subsequent comprehensive score calculations.

[0086] S1055: Set up a dynamic update method for weight coefficients. Collect core feature data and expert rating labels of newly added employees every quarter. Merge the new data with historical data to form an updated dataset. Repeat steps S1052 to S1054 to retrain the SVM model, generate updated first-level indicator weight coefficients, and replace the original weight coefficients.

[0087] In some embodiments, a quarterly update cycle is set, and the model is retrained after fusing new data with historical data to generate updated weight coefficients. This allows the weights to adapt to changes in the competency evaluation needs of personnel at new energy power plants. As employee data accumulates, the weight coefficients can be dynamically adjusted to reflect the latest competency evaluation focus, improving the adaptability of the evaluation method to business changes.

[0088] In one embodiment of the present invention, based on step S106, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner. Step S106 specifically includes the following methods: S1061: From the monthly raw data matrix formed in S103, extract the standardized scores of four primary indicators: performance score, technical score, learning score, and reward and punishment record, according to employee ID and month. Each employee corresponds to a set of standardized scores of the four primary indicators each month, forming a data table of employee ID-month-performance score-technical score-learning score-reward and punishment score.

[0089] In some embodiments, standardized scores of four primary indicators are extracted from the preprocessed monthly raw data matrix, and organized into a structured table by employee and month, specifying the scores of each employee for each indicator each month, providing basic data for the calculation of the comprehensive score.

[0090] S1062: Retrieve the weight coefficients of the four primary indicators (performance weight, technical weight, learning weight, and reward / punishment weight) determined in S105, and verify whether the sum of the weight coefficients is 1. If it is not 1, return to S105 to recalculate; if it is 1, associate the weight coefficients with the structured data table in S1061 to form an extended data table containing the weight coefficients.

[0091] In some embodiments, the weight coefficient verification formula is: Sum of weight coefficients = Performance weight + Technical weight + Learning weight + Reward / penalty weight. The validity of the weight coefficients is verified by summing these coefficients, ensuring that the weight settings meet normalization requirements and guaranteeing the accuracy of the overall score calculation.

[0092] S1063: For each record in the extended data table, calculate the employee's comprehensive score for the month. The calculation method is to multiply the performance score by the performance weight, add the technical score by the technical weight, add the learning score by the learning weight, and add the reward and punishment score by the reward and punishment weight to obtain the monthly comprehensive score, which is rounded to one decimal place.

[0093] In some embodiments, the monthly comprehensive score is calculated as follows: Monthly Comprehensive Score = Performance Score × Performance Weight + Technical Score × Technical Weight + Learning Score × Learning Weight + Reward / Punishment Score × Reward / Punishment Weight; The standardized scores of the four primary indicators are integrated into a single comprehensive score by weighted summation, reflecting the relative importance of each indicator in the evaluation.

[0094] S1064: Periodically summarize the monthly comprehensive scores of the same employee. The quarterly comprehensive score is the arithmetic mean of the monthly comprehensive scores of the three months in that quarter, and the annual comprehensive score is the arithmetic mean of the comprehensive scores of the four quarters in that year. All scores are rounded to one decimal place.

[0095] In some embodiments, the quarterly composite score = (the composite score of the current month + the composite score of the next month + the composite score of the third month) / 3.

[0096] Annual overall score = (quarterly overall score + next quarterly overall score + third quarterly overall score + fourth quarterly overall score) / 4. Monthly scores are averaged to form quarterly and annual scores, reflecting the periodic performance of employees and meeting the management needs of the site for periodic evaluation of personnel capabilities.

[0097] S1065: Set the comprehensive score verification rules to verify whether each employee's monthly, quarterly, and annual comprehensive scores are within the range of 0-100. If they are outside the range, they are marked as abnormal scores and returned to S1063 for recalculation; if they are within the range, the scores are confirmed to be valid and stored in the evaluation results database.

[0098] In some embodiments, abnormal scores are identified by interval verification (0-100), and abnormal results are recalculated and corrected to ensure that the final score conforms to the value range of standardized data.

[0099] In one embodiment of the present invention, based on step S107, the following will provide a possible embodiment and its specific implementation will be described in a non-limiting manner. Step S107 specifically includes the following methods: S1071: From the final quantitative score obtained in S106, extract the comprehensive score for each employee, associate it with the employee ID and evaluation period, retain one decimal place, and generate a quantitative score record containing the employee ID, evaluation period, and comprehensive score. The evaluation period is monthly / quarterly / annual.

[0100] S1072: Extract the standardized scores of the four primary indicators, and use the performance score, technical score, learning score, and reward and punishment record as the four coordinate axes of the radar chart. Map the score of each axis to the radius of the radar chart to generate a radar chart data structure containing the scores of each axis and the coordinates of the vertex.

[0101] S1073: Based on the standardized scores and core feature data of each primary indicator, set structured profile labeling rules, such as marking a performance score below 60 as needing to improve performance ability, a technical score above 85 as having strong technical ability, and a learning score below 70 as having insufficient learning initiative, to generate a structured personnel profile containing employee ID, capability strength label, and capability weakness label.

[0102] S1074: Verify the sum of the standardized scores of each primary indicator multiplied by the corresponding weight coefficients in the comprehensive score calculation, verify whether the radar chart data is consistent with the standardized scores of the primary indicators, check whether the profile labels conform to the preset rules, and correct any calculation errors or mismatched labels in the output.

[0103] S1075: Store quantitative score records, radar chart data, and structured personnel profiles into the personnel competency evaluation results database, link them to the employee basic information table and evaluation cycle, and ensure that the results are traceable and can be queried later.

[0104] In some embodiments, three types of results—quantified score records, radar charts, and structured personnel profiles—are used to meet different management scenarios. Radar charts can intuitively present an employee's strengths and weaknesses across four competency dimensions. For example, if an employee scores 90 in technical skills but 65 in learning skills, this quickly identifies a strong technical skill but weak learning ability. By verifying the overall score calculation logic, radar chart data matching, and profile label compliance, errors in the application of weighting coefficients during overall score calculation and discrepancies between profile labels and scores can be corrected promptly, ensuring that employee profile labels accurately reflect their abilities and improving the credibility of the evaluation results.

[0105] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0106] The following are embodiments of the quantitative evaluation system for the comprehensive capabilities of personnel at new energy power stations provided in this disclosure. This system and the quantitative evaluation method for the comprehensive capabilities of personnel at new energy power stations described in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the quantitative evaluation system for the comprehensive capabilities of personnel at new energy power stations, please refer to the embodiments of the quantitative evaluation method for the comprehensive capabilities of personnel at new energy power stations described above.

[0107] like Figure 2 As shown, the system includes: The indicator definition module 201 is used to define a comprehensive evaluation indicator system for the capabilities of personnel at new energy power stations, which includes four primary indicators: performance evaluation, technical evaluation, learning evaluation, and reward and punishment records, as well as twelve corresponding secondary indicators. The data acquisition and storage module 202 is used to collect various types of data related to the indicator system by connecting to the personnel management system, and store the data in the format of employee ID-indicator category-indicator subcategory-time-value. Data preprocessing module 203 is used to perform outlier removal and filling, missing value filling and Min-Max standardization on the stored data to generate a monthly raw data matrix; The feature extraction module 204 extracts key features based on the original data matrix and selects and retains core features with a weight ≥ 0.05 to form a feature vector; Model training module 205 is used to train the evaluation model using feature vectors and determine the weight coefficients of the four primary indicators. The comprehensive score calculation module 206 is used to calculate the comprehensive score of personnel based on the weighting coefficients and the standardized scores of each primary indicator. The evaluation result generation module 207 outputs quantitative scores, capability radar charts, and structured personnel profiles as evaluation results based on the comprehensive score. The results feedback module 208 is used to feed the evaluation results back to the personnel management system, and is applied to actual management scenarios such as operation and maintenance work order assignment and shortcoming training matching.

[0108] like Figure 3 As shown, this application also provides an electronic device, including a display module 103, a memory 102, a processor 101, and a computer program stored in the memory and executable on the processor 101. When the processor 101 executes the program, it implements the steps of a method for quantitatively evaluating the comprehensive capabilities of personnel at new energy power stations.

[0109] In embodiments of the present invention, electronic devices include, but are not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments described and / or claimed herein.

[0110] In this embodiment, processor 101 may be implemented using at least one of an application-specific integrated circuit, a programmable logic device, a field-programmable gate array, a processor, a controller, a microcontroller, a microprocessor, or an electronic unit designed to perform the functions described herein. In some cases, such an implementation may be implemented within a controller. For software implementation, implementations such as processes or functions may be implemented with separate software modules that allow the performance of at least one function or operation. Software code may be implemented by a software application (or program) written in any suitable programming language, and the software code may be stored in memory and executed by the controller.

[0111] The display module 103 is used to display information input by the user or information provided to the user. The display module 103 may include a display panel, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like.

[0112] The memory 102 can be used to store software programs and various data. The memory 102 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0113] This application also provides a storage medium storing a computer program thereon, wherein when the computer program is executed by a processor, it implements the steps of the quantitative evaluation method for the comprehensive capabilities of personnel at the new energy power station.

[0114] The storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0115] In a storage medium, a readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. This propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0116] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for quantitatively evaluating the comprehensive capabilities of personnel at new energy power stations, characterized in that the method... Comprise: S101: define the new energy station personnel comprehensive ability evaluation index system including four first-level indexes of duty performance score, technical score, learning score and reward and punishment record and corresponding twelve second-level indexes; S102: collect various data related to the index system of S101 by interfacing with the personnel management system, and store the data in the format of employee ID-index category-index subclass-time-value; S103: perform outlier rejection filling, missing value filling and Min-Max standardization on the stored data to generate a monthly original data matrix; S104: extract key features based on the original data matrix, filter and retain core features with a weight greater than or equal to 0.05 to form a feature vector; S105: use the feature vector to train the evaluation model to determine the weight coefficients of the four first-level indexes in S101; S106: calculate the personnel comprehensive score according to the weight coefficients and the standardized scores of each first-level index; S107: based on the comprehensive score, output the quantitative score, ability radar chart and structured personnel portrait as the evaluation result; S108: feedback the evaluation result to the personnel management system and apply it to the actual management scenarios of operation and maintenance work order assignment and short board training matching.

2. The method of claim 1, wherein the method further comprises: Step S102 specifically includes the following modes: Determine the list of systems to be interfaced and the specific data items to be provided, establish the mapping relationship between data items and evaluation index system; Configure the data interface permissions of the personnel management system, send data extraction requests at a predetermined period, and obtain technical and reward original data and store them temporarily; Configure the API interface parameters of the operation and maintenance management system and the training management system, obtain duty and learning original data through API call and store them temporarily; Based on the pre-set mapping relationship table, classify all the temporarily stored original data into the corresponding four first-level indexes and twelve second-level indexes according to their content; Format conversion is performed on the classified data according to the unified field structure, and the data is written into the database for storage.

3. The method of claim 1, wherein the method further comprises: Step S104 specifically includes the following modes: S1041: extract the twelve second-level index standardized data of all employees from the preprocessed monthly original data matrix, construct a feature matrix, and form a label vector combined with expert scoring results; S1042: set the number of decision trees, maximum depth, node splitting standard and minimum number of leaf nodes of the random forest model; S1043: train the random forest model using the feature matrix and label vector, and calculate the average importance weight of each second-level index feature; S1044: compare the calculated average importance weight of each feature with the pre-set threshold to select the core second-level index feature; S1045: extract the standardized data corresponding to the selected core feature from the monthly original data matrix to generate a core feature vector.

4. The method of claim 1, wherein the method further comprises: Step S1043 specifically includes the following modes: Classify the twelve second-level indexes according to the four first-level indexes they belong to, and establish a four-index group mapping table for duty, technology, learning and reward; Randomly select two different index groups for each decision tree, and only extract candidate splitting features from the second-level indexes in the selected groups for node splitting; The original contribution degree of the features used for node splitting is recorded, and the contribution degree is conditionally corrected according to the number of employee samples of the split node; The statistical range of the contribution degree of each secondary index in all decision trees is calculated, and abnormal contribution degree data beyond the reasonable range is removed; The processed contribution degree data is accumulated and multiplied by the preset weighting coefficient of the corresponding index group to obtain the final total contribution value of each secondary index.

5. The method of claim 1, wherein the method further comprises: Step S105 specifically includes the following modes: Group the core features according to the primary index categories to form four feature subsets of performance, technology, learning, and rewards and punishments; Configure the kernel function type and parameter value range of the support vector machine model to prepare for model training; Use the multi-round cross-validation method to train and optimize the support vector machine model using the core feature vector and the label vector; Use the optimized model to extract the core feature weight, aggregate and normalize the data to obtain the weight coefficients of the four primary indexes; Periodically merge the new data and re-execute the training process to generate updated primary index weight coefficients to replace the original coefficients.

6. The method of claim 1, wherein the method further comprises: Step S106 specifically includes the following modes: Extract the standardized scores of the four primary indexes of each employee from the preprocessed monthly raw data matrix to form a data table containing employee identification, time period, and scores; Obtain the four primary index weight coefficients in the trained evaluation model, verify that the sum of the weight coefficients is one hundred, and then associate them with the data table to generate an extended data table containing weight coefficients; For each record in the extended data table, multiply each primary index standardized score by the corresponding weight coefficient and sum them up to calculate the employee monthly comprehensive score; Periodically aggregate the continuous monthly comprehensive scores of the employees to calculate the quarterly and annual comprehensive scores through the arithmetic mean method; Set comprehensive score verification rules to check whether the monthly, quarterly, and annual comprehensive scores are within the range of zero to one hundred, mark and recalculate the abnormal scores, and store them in the database after verification.

7. The method of claim 1, wherein the method further comprises: Step S107 specifically includes the following modes: From the employee comprehensive scores calculated in S106, extract the comprehensive score values of each employee, combine them with the employee ID and evaluation period information, and generate a quantitative score record; Use the four primary index standardized scores in the quantitative score record to convert them into coordinate data for a radar chart to generate a radar chart data structure; According to the primary index scores in the radar chart data structure and the preset label rules, assign the employee a capability label to generate a structured personnel profile; Check the calculation accuracy and rule compliance of the quantitative score record, radar chart data structure, and structured personnel profile, and make corrections; Store the corrected quantitative score record, radar chart data structure, and structured personnel profile in the database and associate them with the employee basic information.

8. A personnel comprehensive ability quantitative evaluation system for a new energy station, characterized in that, The system is used to implement the personnel comprehensive capability quantification evaluation method of the new energy station as claimed in any one of claims 1 to 7; The system comprises: The index definition module is used to define a new energy station personnel comprehensive ability evaluation index system including four first-level indexes of duty performance score, technology score, learning score and reward and punishment record and corresponding twelve second-level indexes. The data acquisition and storage module is used to acquire various data related to the index system through the personnel management system and store the data in the format of employee ID-index category-index subclass-time-value. The data preprocessing module is used to perform outlier elimination and filling, missing value filling and Min-Max standardization on the stored data to generate a monthly original data matrix. The feature extraction module extracts key features based on the original data matrix, filters and retains core features with a weight greater than or equal to 0.05 to form a feature vector. The model training module is used to train the evaluation model using the feature vector to determine the weight coefficients of the four first-level indexes. The comprehensive score calculation module is used to calculate the personnel comprehensive score according to the weight coefficients and the standardized score of each first-level index. The evaluation result generation module outputs the quantitative score, the ability radar chart and the structured personnel portrait as the evaluation result based on the comprehensive score. The result feedback module is used to feed back the evaluation result to the personnel management system and apply it to actual management scenarios such as operation and maintenance work order assignment and short board training matching.

9. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to realize the steps of the new energy station personnel comprehensive ability quantitative evaluation method according to any one of claims 1 to 7.

10. A storage medium having stored thereon a computer program, characterized in that The computer program is executed by the processor to realize the steps of the new energy station personnel comprehensive ability quantitative evaluation method according to any one of claims 1 to 7.

Images