An intelligent evaluation method and system for integrated training results based on personalized learning curve

Abstract

The present invention relates to the field of integrated control training and evaluation technology, and discloses a method and system for intelligent evaluation of integrated control training results based on personalized learning curves, including: collecting training data of controllers, constructing training intelligent evaluation indicators; performing data processing, calculating indicator vectors corresponding to evaluation indicators; constructing a training evaluation model through a multi-layer perceptron, capturing the relationship between deviation and score, and generating a training result score vector; determining the weight of the evaluation indicator, and calculating the training result; constructing a corresponding learning growth curve, identifying weak knowledge points in training, and formulating targeted plans for future training. The present invention can improve the accuracy and reliability of evaluation results, can comprehensively analyze the ability change trend of controllers, accurately identify their learning achievements and ability shortcomings at different training stages, provide a scientific basis for personalized training design and training program optimization, and thus improve the adaptability and comprehensive operational capabilities of controllers.

Description

Technical Field

[0001] The present invention relates to the technical field of regulation-integrated training evaluation, and in particular to a regulation-integrated training result intelligent evaluation method and system based on personalized learning curves. Background Art

[0002] With the rapid development of the power system, a large number of renewable energy sources have been connected to the grid, significantly increasing the proportion of clean energy in the grid. However, this has also brought huge challenges to the safety and stability of the grid. The volatility and intermittency of renewable energy power generation have increased the uncertainty of grid operation, especially in extreme weather or large load fluctuations. This has put higher demands on the professionalism and flexibility of the integrated control of the power system. As the core of grid operation, the professional skills and emergency response capabilities of grid controllers are directly related to the safe and stable operation of the grid.

[0003] Currently, traditional grid controller training is mostly assessed through simulation scenarios and exams. These evaluation methods primarily rely on static indicators or single exam scores, making it difficult to fully reflect a controller's competency development and overall performance. This static evaluation method fails to dynamically integrate a controller's historical training data with their real-time operational records. This results in an incomplete assessment of skill growth trends, weakness analysis, and overall competency. This inability to provide a scientific basis for training optimization and fails to meet the demands for personalized development of controllers in complex grid environments. Furthermore, existing evaluation systems lack the effective dynamic integration of historical training records and real-time operational data. Existing technologies are generally unable to effectively capture the evolution of controller competency over time, particularly regarding the accumulated competency gained through multiple training sessions and actual operations, resulting in weak correlation. For example, it is unable to identify a controller's competency improvements or weaknesses based on their performance at different training stages, nor can it fully account for the impact of previous training on their current performance. This evaluation method, lacking dynamic tracking and correlation analysis, fails to provide a scientific basis for subsequent training and individual competency development, nor does it offer precise guidance for optimizing training programs. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed. Therefore, the present invention provides an intelligent evaluation method for training results based on personalized learning curve integration to solve the above problems.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] In a first aspect, the present invention provides an intelligent evaluation method for integrated control training results based on a personalized learning curve, comprising:

[0007] Collect controller training data and build intelligent evaluation indicators for integrated grid control training results;

[0008] Performing data processing on the training data to calculate an indicator vector corresponding to the evaluation indicator;

[0009] Based on the indicator vector, a training evaluation model is constructed by a multi-layer perceptron to capture the relationship between the deviation and the score, and generate a training result score vector;

[0010] Based on the training result score vector, a weight vector is generated, the weight of the evaluation index is determined, and the training result is calculated;

[0011] Based on the training results, a corresponding learning growth curve is constructed to identify weak knowledge points in the training and develop targeted plans for future training.

[0012] As a preferred solution of the intelligent evaluation method for integrated control training results based on personalized learning curves described in the present invention, the training data of controllers is collected and the intelligent evaluation indicators for integrated power grid control training results are constructed, including:

[0013] First-level indicators, including task completion, emergency response, operational standardization, and learning growth;

[0014] Secondary indicators are comprehensive performance data based on primary indicator data, including comprehensive scores, performance rankings, and expertise analysis.

[0015] As a preferred solution of the intelligent evaluation method for integrated control training results based on personalized learning curves described in the present invention, wherein: data processing is performed on the training data, and the indicator vector corresponding to the evaluation indicator is calculated, which includes:

[0016] Performing data standardization on the training data to remove the data dimension;

[0017] Based on the processed data, an indicator vector is constructed, where the indicator vector includes a training result vector, a benchmark reference vector, and a deviation vector.

[0018] As a preferred solution of the intelligent evaluation method for integrated training results based on personalized learning curves described in the present invention, a training evaluation model is constructed by a multi-layer perceptron to capture the relationship between deviation and score, and generate a training result score vector including:

[0019] The multilayer perceptron includes an input layer, an output layer, and a hidden layer. The input layer receives input data, the output layer outputs the training evaluation model, and the hidden layer constructs the training evaluation model through weighted connections of neurons.

[0020] The deviation vector is input into the training and evaluation model, and the weighted sum is calculated for each neuron in each layer. A nonlinear activation function is introduced, with the leaky rectified linear unit selected as the activation function for the hidden layer and the normalized exponential function selected as the activation function for the output layer.

[0021] Output the secondary indicator score vector, and obtain the primary indicator score vector based on the corresponding relationship between the secondary indicator and the primary indicator.

[0022] As a preferred solution of the intelligent evaluation method for integrated training results based on personalized learning curves of the present invention, determining the weights of the evaluation indicators and calculating the training results include:

[0023] Calculate the probability distribution of each indicator and get the probability vector , expressed as:

[0024] ,

[0025] in, Indicates the The score of the indicator, Indicates the The score of the indicator, 、 Indicates the index of the indicator, n is a natural number, indicating the number of different indicators;

[0026] Calculate the information entropy of each indicator by entropy weight method , expressed as:

[0027] ,

[0028] in, is the entropy normalization coefficient, 0≤ ≤1;

[0029] Calculate the weight of each indicator according to the information entropy to obtain the weight vector , and meet , expressed as:

[0030] ,

[0031] Based on the weight vector, the training result is calculated as:

[0032] ,

[0033] in, Indicates the The training score, Indicates the The first-level indicator score of the training session, Indicates the The transpose of the weight vector for the training step.

[0034] As a preferred solution of the intelligent evaluation method for integrated training results based on personalized learning curves described in the present invention, constructing the corresponding learning growth curve includes:

[0035] Through the time series attention dynamic weighting algorithm, the corresponding learning growth curve is constructed, the historical training results are input into the time series vector, the time series features are constructed, and the training results are combined with the time series features;

[0036] Generate query, key, and value vectors through linear transformation of features at each time point, and calculate attention scores based on query vector and key vector to obtain the importance of each time point;

[0037] Based on the importance of each time point, the learning value of each time point is calculated to generate a learning growth curve.

[0038] As a preferred embodiment of the intelligent evaluation method for integrated training results based on personalized learning curves of the present invention, the learning value of each time point is calculated based on the importance of each time point, and the generation of the learning growth curve includes:

[0039] Perform weighted summation of the value vector according to the corresponding attention weight to generate the first The learning value at a time point is expressed as:

[0040] ,

[0041] in, For time point Time point The attention weight, represents the jth element in the learning value vector, and T represents the total number of time points in the time period;

[0042] Recursively generate the entire learning curve, calculate the learning value at each time point, and generate a learning growth curve, which is expressed as:

[0043] ,

[0044] The time decay factor is introduced to enhance the weight difference of time distance, which is expressed as:

[0045] ,

[0046] in, Indicates a time point Time point The attention score, represents the query vector, represents the key vector, is the time decay coefficient.

[0047] In a second aspect, the present invention provides an intelligent evaluation system for integrated training results based on personalized learning curves, comprising:

[0048] The collection module is used to collect controller training data and build intelligent evaluation indicators for the results of integrated grid control training;

[0049] A first calculation module is used to process the training data and calculate an indicator vector corresponding to the evaluation indicator;

[0050] A model building module is used to build a training evaluation model based on the indicator vector through a multi-layer perceptron, capture the relationship between the deviation and the score, and generate a training result score vector;

[0051] A second calculation module is used to generate a weight vector based on the training result score vector, determine the weight of the evaluation index, and calculate the training result;

[0052] The output module is used to construct a corresponding learning growth curve based on the training results, identify weak knowledge points in the training, and formulate targeted plans for future training.

[0053] In a third aspect, the present invention provides an electronic device, comprising:

[0054] memory and processor;

[0055] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the intelligent evaluation method for integrated regulation and control training results based on personalized learning curves are implemented.

[0056] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the intelligent evaluation method for integrated regulation and control training results based on personalized learning curves.

[0057] Compared with the existing technology, the beneficial effects of the present invention are as follows: the present invention improves the accuracy and reliability of the evaluation results by combining historical training data with real-time operation records and introducing the dynamic empowerment of the temporal attention mechanism. It can comprehensively analyze the ability change trend of the controller, accurately identify his learning achievements and ability shortcomings at different training stages, and provide a scientific basis for personalized training design and training program optimization, thereby improving the controller's adaptability and comprehensive operational capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0059] Figure 1 This is a schematic diagram of the overall process of an intelligent evaluation method for integrated training results based on personalized learning curves according to one embodiment of the present invention;

[0060] Figure 2 A schematic diagram of a training evaluation model for an intelligent evaluation method for integrated training results based on personalized learning curves according to an embodiment of the present invention;

[0061] Figure 3 A schematic diagram of weights for generating evaluation indicators for an intelligent evaluation method for integrated training results based on personalized learning curves according to an embodiment of the present invention. DETAILED DESCRIPTION

[0062] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0063] Reference Figure 1-Figure 3 , which is an embodiment of the present invention, provides an intelligent evaluation method for integrated training results based on personalized learning curves, including:

[0064] S100 collects controller training data and constructs intelligent evaluation indicators for integrated grid control training results;

[0065] S200, performing data processing on the training data and calculating the index vector corresponding to the evaluation index;

[0066] S300, based on the indicator vector, builds a training evaluation model through a multi-layer perceptron to capture the relationship between deviation and score and generate a training result score vector;

[0067] S400, generating a weight vector based on the training result score vector, determining the weight of the evaluation index, and calculating the training result;

[0068] S500, based on training results, builds corresponding learning growth curves, identifies weak knowledge points in training, and develops targeted plans for future training.

[0069] It should be noted that the present invention dynamically integrates the historical and real-time data of controllers. The system can generate personalized learning curves, comprehensively display the trend of ability changes, provide scientific guidance for the optimized design of controller training, and help the integrated power grid control to cope with more complex operational challenges in the future.

[0070] In a preferred embodiment, collecting controller training data and constructing intelligent evaluation indicators for integrated grid control training results include:

[0071] Level 1 indicators include task completion, emergency response, operational standardization, and learning growth;

[0072] Secondary indicators are comprehensive performance data based on primary indicator data, including comprehensive scores, performance rankings, and expertise analysis.

[0073] Specifically, in order to comprehensively evaluate the training results of controllers, this evaluation system designs a two-tier evaluation index. The first-level index includes task completion, emergency response, operation standardization, learning growth, etc., and the second-level index includes comprehensive performance. Among them, task completion is mainly measured by task completion rate and task accuracy, which is the performance of controllers in terms of the number and accuracy of completed tasks. Emergency response focuses on alarm response time, fault location accuracy and emergency strategy execution efficiency to evaluate the controller's rapid response and decision-making ability to deal with emergencies. Operation standardization reflects the controller's operation standardization level and resource utilization efficiency by examining the correctness of operation steps and the rationality of resource allocation. Learning growth is based on data analysis of historical performance improvement rate, learning curve and skill proficiency, and dynamically displays the controller's ability change trend and learning progress. Finally, comprehensive performance integrates all indicator data, and comprehensively reflects the controller's overall ability and outstanding advantages through comprehensive scoring, performance ranking and expertise analysis.

[0074] In a preferred embodiment, processing the training data to calculate the indicator vector corresponding to the evaluation indicator includes:

[0075] Standardize the training data, remove the data dimension, and normalize it;

[0076] Based on the processed data, an indicator vector is constructed, which includes a training result vector, a benchmark reference vector, and a deviation vector.

[0077] In an optional embodiment, the data is dimensionlessly processed and normalized, and expressed as:

[0078] ,

[0079] in, is the normalized data, is the data to be normalized, is the minimum value in the data set, is the maximum value in the data set.

[0080] In an optional embodiment, constructing an indicator vector based on the processed data includes:

[0081] Training result vector It is composed of the actual training data of the corresponding elements of the secondary indicators, expressed as:

[0082] ,

[0083] Among them, R represents the training result vector, r n Represents the actual training data of the nth element;

[0084] Baseline reference vector It is composed of the standard values ​​of the corresponding elements of the secondary indicators, expressed as:

[0085] ,

[0086] Where S represents the reference vector, s n Indicates the standard value of the nth element;

[0087] Deviation vector The deviation between the training result and the standard value is calculated element by element, and the resulting deviation vector is expressed as:

[0088] ,

[0089] Take the absolute value of the deviation to reflect the degree of deviation and get the deviation vector, which is expressed as:

[0090] ,

[0091] In a preferred embodiment, a training evaluation model is constructed by a multi-layer perceptron to capture the relationship between deviation and score, and generating a training result score vector includes:

[0092] The multilayer perceptron consists of an input layer, an output layer, and a hidden layer. The input layer receives input data, the output layer outputs the training and evaluation model, and the hidden layer constructs the training and evaluation model through weighted connections of neurons.

[0093] The deviation vector is input into the training and evaluation model, and the weighted sum is calculated for each neuron in each layer. A nonlinear activation function is introduced. The hidden layer uses the leaky rectified linear unit (Leaky ReLU) as the activation function, and the output layer uses the normalized exponential function (Softmax) as the activation function.

[0094] Output the secondary indicator score vector, and according to the corresponding relationship between the secondary indicators and the primary indicators, obtain the primary indicator score vector.

[0095] In an optional embodiment, constructing a training evaluation model using a multi-layer perceptron includes:

[0096] Input bias vector in training and evaluating model , calculate the weighted sum for each neuron in each layer, expressed as:

[0097] ,

[0098] in, is the weight between the input feature and the neuron, is the bias of the neuron, The first Features.

[0099] A nonlinear activation function is introduced. The hidden layer uses the leaky rectified linear unit (Leaky ReLU) as the activation function, and the output layer uses the normalized exponential function (Softmax) as the activation function, which can be expressed as:

[0100] ,

[0101] in, is a small constant, usually 0.01, Represents the elements in the output vector;

[0102] ,

[0103] in, is the first elements, the sum of the probabilities of the output elements is 1.

[0104] Output secondary index score vector ,in, Indicates the output The corresponding relationship between the secondary indicators and the primary indicators is calculated as follows:

[0105] ,

[0106] in, , α, β, ..., μ represent the weight coefficients of the secondary indicators corresponding to the primary indicators, Indicates the First-level indicator scores;

[0107] Then, the first-level index score vector is obtained, which is expressed as:

[0108] ,

[0109] in, Indicates the A first-level indicator score.

[0110] In a preferred embodiment, determining the weights of the evaluation indicators and calculating the training results include:

[0111] The entropy weight method is used to generate a weight vector and determine the weights of different primary evaluation indicators. The entropy weight method determines the importance of different indicators by calculating the information entropy value of each indicator and uses it as its weight to finally form a weight vector, as follows:

[0112] Calculate the probability distribution of each indicator and get the probability vector , expressed as:

[0113] ,

[0114] in, Indicates the probability vector, Indicates the The score of the indicator, Indicates the The score of the indicator, 、 Indicates the index of the indicator, n is a natural number, indicating the number of different indicators;

[0115] Calculate the information entropy of each indicator by entropy weight method , expressed as:

[0116] ,

[0117] in, is the entropy normalization coefficient, ensuring 0≤ ≤1, Indicates the The information entropy of an indicator;

[0118] Calculate the weight of each indicator according to the information entropy and get the weight vector , and meet , expressed as:

[0119] ,

[0120] in, Indicates the The information entropy of each indicator;

[0121] Based on the weight vector, the training results are calculated as:

[0122] ,

[0123] in, Indicates the The training score, Indicates the The first-level indicator score of the training session, Indicates the The transpose of the weight vector for the training step.

[0124] In a preferred embodiment, constructing a corresponding learning growth curve includes:

[0125] Through the time series attention dynamic weighting algorithm, the corresponding learning growth curve is constructed, the historical training results are input into the time series vector, the time series features are constructed, and the training results are combined with the time series features;

[0126] Generate query, key, and value vectors through linear transformation of features at each time point, and calculate attention scores based on query vector and key vector to obtain the importance of each time point;

[0127] Based on the importance of each time point, the learning value of each time point is calculated to generate a learning growth curve.

[0128] In an optional implementation, a corresponding learning growth curve is constructed based on a temporal attention dynamic weighting method, including:

[0129] Input historical training results into the time series vector ,in, is the total number of time points of the training results, For the The results of the training;

[0130] Construct time series features and embed the time features based on position encoding, which can be expressed as:

[0131] ,

[0132] in, is the embedding dimension, For time point Position encoding features, is the current dimension index;

[0133] Combine the training results with the time features and express it as:

[0134] ,

[0135] in, express Combination of time point training results and time features;

[0136] Calculate the attention weights and generate query, key, and value vectors for each time point through linear changes, which are expressed as:

[0137] ,

[0138] ,

[0139] ,

[0140] in, , , is a trainable parameter matrix, and the attention score is calculated by the query vector and the key vector to measure the importance of each time point, which is expressed as:

[0141] ,

[0142] ,

[0143] in, For time point Time point The attention weight, is the scaling factor to avoid gradient instability caused by being too large. Indicates a time point Time point Attention score.

[0144] Perform weighted summation of the value vector according to the corresponding attention weight to generate the first The learning value at a time point is expressed as:

[0145] ,

[0146] in, For time point Time point The attention weight, Represents the jth element in the learning value vector, the closer Time point , attention weight The larger it is, the more T represents the total number of time points in the time period;

[0147] Recursively generate the entire learning curve, calculate the learning value at each time point, and generate a learning growth curve, which is expressed as:

[0148] ,

[0149] The time decay factor is introduced to enhance the weight difference of time distance, which is expressed as:

[0150] ,

[0151] in, Indicates a time point Time point The attention score, represents the query vector, represents the key vector, is the time attenuation coefficient, which controls the rate of decrease of the time weight.

[0152] It should be noted that the present invention combines actual control scenarios to construct a two-layer personalized evaluation index for integrated control training, realizes the intelligent generation of controller training result evaluation based on the entropy weight perception method, and uses the temporal attention mechanism to draw the controller's personalized learning curve based on the controller's historical evaluation data to realize intelligent evaluation of training results.

[0153] By combining historical training data with real-time operation records and introducing the dynamic empowerment of the temporal attention mechanism, the accuracy and reliability of the evaluation results are improved. This enables a comprehensive analysis of the controller's ability change trends, and accurately identifies their learning outcomes and ability shortcomings at different training stages. This provides a scientific basis for personalized training design and training program optimization, thereby improving the controller's adaptability and comprehensive operational capabilities.

[0154] The above is a schematic scheme of a method for intelligently evaluating training results of integrated control based on personalized learning curves in this embodiment. It should be noted that the technical scheme of the intelligent evaluation system for training results of integrated control based on personalized learning curves and the technical scheme of the intelligent evaluation method for training results of integrated control based on personalized learning curves are based on the same concept. For details not described in detail in the technical scheme of the intelligent evaluation system for training results of integrated control based on personalized learning curves in this embodiment, please refer to the description of the technical scheme of the intelligent evaluation method for training results of integrated control based on personalized learning curves.

[0155] The intelligent evaluation system for integrated training results based on personalized learning curves in this embodiment includes:

[0156] The collection module is used to collect controller training data and build intelligent evaluation indicators for the results of integrated grid control training;

[0157] A first calculation module is used to process the training data and calculate the indicator vector corresponding to the evaluation indicator;

[0158] A model building module is used to build a training evaluation model based on the indicator vector through a multi-layer perceptron, capture the relationship between deviation and score, and generate a training result score vector;

[0159] The second calculation module is used to generate a weight vector based on the training result score vector, determine the weight of the evaluation index, and calculate the training result;

[0160] The output module is used to build a corresponding learning growth curve based on the training results, identify weak knowledge points in the training, and develop targeted plans for future training.

[0161] This embodiment further provides an electronic device suitable for intelligent evaluation of training results based on integrated regulation and control based on personalized learning curves, including:

[0162] Memory and processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the intelligent evaluation method for integrated training results based on personalized learning curves as proposed in the above embodiment.

[0163] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent evaluation method for integrated training results based on personalized learning curves as proposed in the above embodiment.

[0164] The storage medium proposed in this embodiment and the intelligent evaluation method for integrated training results based on personalized learning curve proposed in the above embodiment belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0165] Through the above description of the embodiments, those skilled in the art will clearly understand that the present invention can be implemented using software and necessary general-purpose hardware. Of course, it can also be implemented using hardware, but in many cases the former is a more preferred embodiment. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk, or optical disk, and includes a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods of various embodiments of the present invention.

[0166] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent evaluation method for integrated control training results based on personalized learning curves, characterized in that: include: Collect controller training data and build intelligent evaluation indicators for integrated grid control training results; Performing data processing on the training data to calculate an indicator vector corresponding to the evaluation indicator; Based on the indicator vector, a training evaluation model is constructed by a multi-layer perceptron to capture the relationship between the deviation and the score, and generate a training result score vector; Based on the training result score vector, a weight vector is generated, the weight of the evaluation index is determined, and the training result is calculated; Based on the training results, a corresponding learning growth curve is constructed to identify weak points in the training and develop targeted plans for future training; Collecting controller training data and constructing intelligent evaluation indicators for integrated grid control training results include: First-level indicators include task completion, emergency response, operational standardization, and learning growth; Secondary indicators, which are comprehensive performance data based on primary indicator data, including comprehensive scores, performance rankings, and expertise analysis; Performing data processing on the training data and calculating the indicator vector corresponding to the evaluation indicator includes: Performing data standardization on the training data to remove the data dimension; Based on the processed data, an indicator vector is constructed. The indicator vector includes a training result vector, a benchmark reference vector, and a deviation vector. The training result vector R is composed of the actual training data of the corresponding elements of the secondary indicators, the benchmark reference vector S is composed of the standard values ​​of the corresponding elements of the secondary indicators, and the deviation vector L is calculated by element-by-element to obtain the deviation between the training result and the standard value. The deviation vector takes the absolute value of the deviation; A training evaluation model is constructed through a multi-layer perceptron to capture the relationship between bias and score, and generate a training result score vector including: The multilayer perceptron includes an input layer, an output layer, and a hidden layer. The input layer receives input data, the output layer outputs the training evaluation model, and the hidden layer constructs the training evaluation model through weighted connections of neurons. The deviation vector is input into the training and evaluation model, and the weighted sum is calculated for each neuron in each layer. A nonlinear activation function is introduced, with the leaky rectified linear unit selected as the activation function for the hidden layer and the normalized exponential function selected as the activation function for the output layer. Output secondary index score vector E = [e1, e2, ..., e n ], where e i It represents the output of the i-th secondary indicator score. According to the corresponding relationship between the secondary indicators and the primary indicators, the following formula is obtained: f j =αe1+βe2+…+μe n Among them, α+β+…+μ=1, α, β,…, μ represent the weight coefficients of the secondary indicators corresponding to the primary indicators, f j represents the jth first-level indicator score; Then, the first-level index score vector is obtained, which is expressed as: F=[f1,f2,…,f n ] Among them, f i represents the i-th first-level indicator score; Building a corresponding learning growth curve includes: Through the time series attention dynamic weighting algorithm, the corresponding learning growth curve is constructed, the historical training results are input into the time series vector, the time series features are constructed, and the training results are combined with the time series features; Generate query, key, and value vectors through linear transformation of features at each time point, and calculate attention scores based on query vector and key vector to obtain the importance of each time point; Based on the importance of each time point, calculating the learning value of each time point and generating a learning growth curve; Based on the importance of each time point, calculating the learning value of each time point and generating a learning growth curve includes: The value vector is weighted and summed according to the corresponding attention weight to generate the learning value at the t-th time point, which is expressed as: Among them, α t,j is the attention weight of time point j to time point t, V j represents the jth element in the learning value vector, and T represents the total number of time points in the time period; Recursively generate the entire learning curve, calculate the learning value at each time point, and generate a learning growth curve, which is expressed as: Y=[y1,y2,…,y T ] The time decay factor is introduced to enhance the weight difference of time distance, which is expressed as: Among them, Score t,j represents the attention score of time point j to time point t, Q t represents the query vector, K t represents the bond vector, and γ is the time decay coefficient.

2. The intelligent evaluation method for integrated training results based on personalized learning curve according to claim 1, characterized in that: Determining the weights of the evaluation indicators and calculating the training results include: Calculate the probability distribution of each indicator and get the probability vector P = [p1, p2, ..., p n ], expressed as: Among them, p i represents the i-th probability vector, z i represents the score of the i-th indicator, z j represents the score of the jth indicator, i,j represents the index of the indicator, and n is a natural number representing the number of different indicators; The information entropy H of each indicator is calculated by the entropy weight method i , expressed as: Among them, k is the entropy normalization coefficient, 0≤H i ≤1, H i represents the information entropy of the i-th indicator; The weight of each indicator is calculated according to the information entropy to obtain the weight vector W = [w1, w2, ..., w n ], and meet Expressed as: Among them, H j represents the information entropy of the jth indicator; Based on the weight vector, the training result is calculated as: Among them, x n Indicates the score of the nth training, F n represents the first-level indicator score of the nth training, Represents the transpose of the weight vector for the nth training pass.

3. An intelligent evaluation system for integrated training results based on personalized learning curves, applying the method according to any one of claims 1 to 2, characterized in that: include, The collection module is used to collect controller training data and build intelligent evaluation indicators for the results of integrated grid control training; A first calculation module is used to process the training data and calculate an indicator vector corresponding to the evaluation indicator; A model building module is used to build a training evaluation model based on the indicator vector through a multi-layer perceptron, capture the relationship between the deviation and the score, and generate a training result score vector; A second calculation module is used to generate a weight vector based on the training result score vector, determine the weight of the evaluation index, and calculate the training result; The output module is used to construct a corresponding learning growth curve based on the training results, identify weak knowledge points in the training, and formulate targeted plans for future training.

4. An electronic device, characterized in that: include: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 2 are implemented.

5. A computer-readable storage medium, characterized in that It stores computer-executable instructions, which, when executed by a processor, implement the steps of the method according to any one of claims 1 to 2.

Images