Power grid artificial intelligence achievement excellence evaluation method, system, device and medium

Abstract

The present application relates to a kind of power grid artificial intelligence achievement excellence evaluation method, system, equipment and medium, the method includes based on the evaluation demand of built power grid artificial intelligence achievement, corresponding evaluation index system is built;Obtain the index data matrix of each built power grid artificial intelligence achievement in running cycle after pre-processing;According to index data matrix and evaluation index system, evaluation model is built, and evaluation model is optimized and solved, and the optimal projection direction is obtained;According to the optimal projection direction, the comprehensive excellence index of each built power grid artificial intelligence achievement is determined, and diagnosis analysis result is determined based on comprehensive excellence index and optimal projection direction.The present application has the effect of improving evaluation accuracy.

Description

Technical Field

[0001] This invention belongs to the technical field of power grid artificial intelligence achievement evaluation, and in particular relates to a method, system, equipment and medium for evaluating the excellence of power grid artificial intelligence achievements. Background Technology

[0002] Currently, with the continuous application of artificial intelligence (AI) technology in power grid operations, the AI ​​achievements already established in the power grid are gradually covering multiple aspects such as dispatching and operation, equipment maintenance, business support, and scenario applications. These achievements will continuously generate data related to technical performance, scenario adaptation, business effectiveness, and application status throughout their operational cycle. Power grid business scenarios are characterized by complex operating conditions, diverse evaluation dimensions, limited sample size, and significant differences in data characteristics. This necessitates that the evaluation of power grid AI achievements not only consider the operational scenario but also take into account the continuity and comprehensiveness of the operational period.

[0003] Existing evaluation methods for AI achievements in power grids typically employ a comprehensive assessment based on several performance indicators, involving quantitative comparisons and result determinations. However, this approach is susceptible to the influence of how the evaluation dimensions are handled in practice. This can lead to difficulties in accurately reflecting the differences between multi-dimensional indicators and the overall performance in actual operation, resulting in discrepancies between the evaluation results and the true excellence of the achievements in the power grid scenario, and impacting the effectiveness of subsequent diagnostic analyses. Summary of the Invention

[0004] The purpose of this invention is to provide a method, system, device and medium for evaluating the excellence of artificial intelligence achievements in power grids, so as to solve the technical problem that existing evaluation methods for artificial intelligence achievements in power grids are difficult to objectively reflect the actual operational effectiveness of the achievements due to evaluation distortion.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for evaluating the excellence of artificial intelligence achievements in power grids, the method comprising: Based on the evaluation needs of the existing power grid artificial intelligence achievements, a corresponding evaluation index system is constructed. Obtain the preprocessed indicator data matrix of each of the aforementioned artificial intelligence achievements of the built power grid within the operating cycle; An evaluation model is constructed based on the indicator data matrix and the evaluation indicator system, and the evaluation model is optimized and solved to obtain the optimal projection direction; The comprehensive excellence index of each of the existing power grid artificial intelligence achievements is determined based on the optimal projection direction, and the diagnostic analysis results are determined based on the comprehensive excellence index and the optimal projection direction.

[0006] By adopting the above technical solutions and constructing corresponding evaluation index systems based on the evaluation needs of existing power grid AI achievements, the excellence evaluation can be centered around the actual application scenarios and business concerns of the achievements, thereby establishing unified and targeted evaluation standards for different achievements. By obtaining the pre-processed index data matrix of each existing power grid AI achievement during its operating cycle, index data from different sources, with different dimensions, and different value directions can be transformed into a unified and comparable data foundation, thereby improving the standardization and comparability of subsequent evaluation calculations. By constructing an evaluation model based on the index data matrix and the evaluation index system, and optimizing and solving the evaluation model to obtain the optimal projection direction, the directional features that best reflect the differences and comprehensive performance of achievements can be extracted from multi-dimensional index data, thereby improving the objectivity and accuracy of the excellence evaluation. By determining the comprehensive excellence index of each existing power grid AI achievement based on the optimal projection direction, and determining the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction, the weaknesses and improvement directions of the achievements can be identified while quantitatively characterizing the overall level of the achievements, thus providing a basis for the continuous optimization and promotion of the achievements.

[0007] In one example, the present invention can be further configured as follows: Based on the evaluation requirements of the existing power grid artificial intelligence achievements, a corresponding evaluation index system is constructed, including: Determine the target layer, criterion layer, element layer, and indicator layer of the evaluation index system; The target layer is set as the power grid artificial intelligence achievement excellence index; The criteria layer is set as technological innovation capability, scenario expansion capability, business support capability, and convenient application capability; Based on the evaluation requirements corresponding to the criterion layer, the element layer and the indicator layer are refined and configured to obtain the evaluation indicator system.

[0008] By adopting the above technical solution, and by defining the target layer, criterion layer, element layer, and indicator layer of the evaluation index system, and setting the target layer as the power grid artificial intelligence achievement excellence index, and setting the criterion layer as technological innovation capability, scenario expansion capability, business support capability, and convenient application capability, the achievement can be systematically characterized from three levels: overall goal, core dimension, and specific indicator. This improves the coverage, completeness, and hierarchical clarity of the evaluation index system for the comprehensive performance of power grid artificial intelligence achievements.

[0009] In one example, the present invention can be further configured as follows: obtaining the preprocessed indicator data matrix of each of the existing power grid artificial intelligence achievements within the operating cycle includes: Obtain the data of each indicator in the evaluation indicator system corresponding to the artificial intelligence achievements of the existing power grids during the operating cycle; Based on the evaluation index system, the data of each index are organized to determine the correspondence between the indexes, thus obtaining the original evaluation data set. The original evaluation data set is dimensionless to generate preprocessed index data corresponding to each of the artificial intelligence achievements of the built power grid. The indicator data matrix is ​​constructed based on the preprocessed indicator data.

[0010] By adopting the above technical solution, and by acquiring the data of each indicator in the evaluation indicator system corresponding to the artificial intelligence achievements of each built power grid during the operation cycle, and by organizing the correspondence between indicators, performing dimensionless processing and matrix construction, the scattered original indicator data can be transformed into a unified indicator data matrix, thereby providing reliable data support for the construction of subsequent evaluation models and unified calculations.

[0011] In one example, the present invention can be further configured as follows: the step of constructing an evaluation model based on the indicator data matrix and the evaluation indicator system, and optimizing the evaluation model to obtain the optimal projection direction, includes: Based on the aforementioned indicator data matrix, construct the projection value calculation relationship corresponding to each of the existing power grid artificial intelligence achievements; Based on the relationship between the projection values, a projection index function is constructed to characterize the projection effect, thus obtaining the evaluation model; The evaluation model is subjected to directional optimization processing to obtain the optimal projection direction.

[0012] By adopting the above technical solution, the projection value calculation relationship corresponding to each completed power grid artificial intelligence achievement is constructed based on the index data matrix, and a projection index function for characterizing the projection effect is constructed according to the projection value calculation relationship. Then, the optimal projection direction is obtained by performing direction optimization processing on the evaluation model. This can transform high-dimensional index data into an evaluation model that can highlight the differences in achievement characteristics, thereby improving the ability to express and effectively distinguish the comprehensive performance of the achievement in a reduced dimension.

[0013] In one example, the present invention can be further configured as follows: the directional optimization process of the evaluation model to obtain the optimal projection direction includes: The candidate directions are encoded to obtain multiple individual directions to be solved; Based on the projection index function, an adaptive evaluation is performed on each individual in the direction to be solved, and the corresponding evaluation results are obtained; Based on the evaluation results, each individual in the direction to be solved is iteratively updated to obtain the individual in the target direction. The optimal projection direction is determined based on the target direction individual.

[0014] By adopting the above technical solution, multiple individual directions to be solved are obtained by encoding the candidate directions. Adaptive evaluation and iterative updates are performed based on the projection index function to finally determine the optimal projection direction. This can gradually filter out better solutions from multiple candidate directions, thereby improving the search efficiency and solution stability of the optimal projection direction determination process.

[0015] In one example, the present invention can be further configured as follows: determining the comprehensive excellence index of each of the existing power grid artificial intelligence achievements according to the optimal projection direction, and determining the diagnostic analysis result based on the comprehensive excellence index and the optimal projection direction, includes: The comprehensive projection value corresponding to each of the artificial intelligence achievements of the existing power grid is determined according to the optimal projection direction. The comprehensive excellence index corresponding to each of the completed power grid artificial intelligence achievements is determined based on the comprehensive projection value. The optimal projection direction is grouped according to the criteria layer in the evaluation index system to obtain the dimensional analysis results corresponding to each criteria layer. Based on the comprehensive excellence index and the dimensional analysis results, the diagnostic analysis results corresponding to each of the established power grid artificial intelligence achievements are determined.

[0016] By adopting the above technical solution, the comprehensive projection value corresponding to each completed power grid artificial intelligence achievement is determined according to the optimal projection direction, and the comprehensive excellence index is determined according to the comprehensive projection value. Then, the optimal projection direction is grouped according to the criteria layer in the evaluation index system to obtain the dimensional analysis results, and then the diagnostic analysis results are determined. This can achieve the combination of overall achievement level evaluation and dimensional contribution analysis, thereby improving the interpretability of the evaluation results and the diagnostic pertinence.

[0017] In one example, the present invention can be further configured as follows: determining the diagnostic analysis results corresponding to each of the existing power grid artificial intelligence achievements based on the comprehensive excellence index and the dimensional analysis results includes: Based on the comprehensive excellence index, the target has been established as an achievement in power grid artificial intelligence. Based on the dimensional analysis results, the weak dimensions corresponding to the existing power grid artificial intelligence achievements of the target are determined. Based on the aforementioned weak dimensions, a contribution analysis is performed on the indicator layer of the evaluation indicator system to determine the key indicators; Based on the key indicators, corresponding optimization suggestions are generated, and the weak dimension, the key indicators, and the optimization suggestions are combined as the diagnostic analysis result.

[0018] By adopting the above technical solution, the target already built power grid artificial intelligence achievements are determined based on the comprehensive excellence index, and the weak dimensions are identified based on the dimensional analysis results. Then, contribution analysis is performed on the indicator layer in the evaluation indicator system to identify key indicators. Based on the key indicators, corresponding optimization suggestions are generated, and the weak dimensions, key indicators, and optimization suggestions are combined as diagnostic analysis results. This allows for the layer-by-layer location and attribution of the sources of the shortcomings of the target achievements, thereby providing more operational guidance for the functional iteration, performance optimization, and scenario adaptation improvement of the achievements.

[0019] In a second aspect, the present invention provides a system for evaluating the excellence of artificial intelligence achievements in power grids, the system comprising: The system construction module is used to build a corresponding evaluation index system based on the evaluation needs of the existing power grid artificial intelligence achievements; The data matrix module is used to obtain the preprocessed indicator data matrix of the artificial intelligence achievements of each built power grid during the operation cycle; The model solving module is used to construct an evaluation model based on the indicator data matrix and the evaluation indicator system, and to optimize and solve the evaluation model to obtain the optimal projection direction. The results diagnosis module is used to determine the comprehensive excellence index of each completed power grid artificial intelligence achievement based on the optimal projection direction, and to determine the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction.

[0020] By adopting the above technical solutions and constructing corresponding evaluation index systems based on the evaluation needs of existing power grid AI achievements, the excellence evaluation can be centered around the actual application scenarios and business concerns of the achievements, thereby establishing unified and targeted evaluation standards for different achievements. By obtaining the pre-processed index data matrix of each existing power grid AI achievement during its operating cycle, index data from different sources, with different dimensions, and different value directions can be transformed into a unified and comparable data foundation, thereby improving the standardization and comparability of subsequent evaluation calculations. By constructing an evaluation model based on the index data matrix and the evaluation index system, and optimizing and solving the evaluation model to obtain the optimal projection direction, the directional features that best reflect the differences and comprehensive performance of achievements can be extracted from multi-dimensional index data, thereby improving the objectivity and accuracy of the excellence evaluation. By determining the comprehensive excellence index of each existing power grid AI achievement based on the optimal projection direction, and determining the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction, the weaknesses and improvement directions of the achievements can be identified while quantitatively characterizing the overall level of the achievements, thus providing a basis for the continuous optimization and promotion of the achievements.

[0021] In a third aspect, the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the aforementioned method for evaluating the excellence of power grid artificial intelligence achievements.

[0022] In a fourth aspect, the present invention provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the aforementioned method for evaluating the excellence of artificial intelligence achievements in power grids. Attached Figure Description

[0023] The accompanying drawings, which form part of this specification, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of a method for evaluating the excellence of artificial intelligence achievements in power grids, as described in an embodiment of the present invention. Figure 2 This is a structural block diagram of the power grid artificial intelligence achievement excellence evaluation system in an embodiment of the present invention; Figure 3 This is a structural block diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0024] The present invention will now be described in detail with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0025] The following detailed description is exemplary and intended to provide further detailed explanation of the invention. Unless otherwise specified, all technical terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this invention is for describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention.

[0026] Example 1 like Figure 1 As shown, this invention discloses a method for evaluating the excellence of artificial intelligence achievements in power grids, specifically including the following steps: S10: Based on the evaluation needs of the existing power grid artificial intelligence achievements, construct a corresponding evaluation index system.

[0027] In this embodiment, the completed power grid artificial intelligence achievements refer to artificial intelligence application achievements that have been developed, deployed, and put into actual operation in power grid business scenarios, and can continuously generate technical performance data, scenario adaptation data, business support data, and application operation data for excellence evaluation during the operation cycle.

[0028] Specifically, focusing on business scenarios where the achievements of artificial intelligence in the power grid need to be continuously evaluated, compared horizontally, and tracked vertically during their operational cycle, we first sort out the evaluation focus points of the achievements in terms of technical performance, scenario adaptation, business implementation, and application operation and maintenance. Then, we organize the various evaluation focus points into an evaluation framework that can support subsequent data collection, evaluation model construction, comprehensive excellence index calculation, and diagnostic analysis, so that different achievements can enter a unified evaluation process under the same evaluation objectives and the same data standards, thereby providing a foundation for the subsequent formation of comparable, traceable, and diagnosable excellence evaluation results.

[0029] S20: Obtain the preprocessed indicator data matrix of artificial intelligence achievements of each built power grid during the operation cycle.

[0030] Specifically, based on the established evaluation framework, the technical performance data, business indicator data, scenario adaptation data, and application operation data generated by the artificial intelligence achievements of each completed power grid during their operation cycle are aggregated. The indicator data from different sources, with different dimensions and different value directions are converted into a matrix data carrier that can be uniformly expressed, so that each row corresponds to an evaluated achievement and each column corresponds to an evaluation indicator. After the data is normalized, an indicator data matrix that can be directly called by the subsequent evaluation model is formed.

[0031] S30: Construct an evaluation model based on the indicator data matrix and evaluation indicator system, and optimize the evaluation model to obtain the optimal projection direction.

[0032] Specifically, the indicator data matrix is ​​used as the input of high-dimensional evaluation data, and the evaluation indicator system is used as the basis for indicator organization. The differences in the results in the multi-dimensional indicator space are transformed into comparable expressions in the low-dimensional space. An evaluation model that can simultaneously characterize the overall dispersion and local aggregation is constructed. Then, through the direction optimization process, the target direction that best reveals the original high-dimensional data structure characteristics and the differences in results is searched, so that the subsequent comprehensive evaluation is based on the projection direction that can fully reflect the true excellence of the results.

[0033] S40: Determine the comprehensive excellence index of each completed power grid artificial intelligence achievement based on the optimal projection direction, and determine the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction.

[0034] Specifically, the optimal projection direction is used for the comprehensive evaluation calculation of the artificial intelligence achievements of each completed power grid, resulting in a comprehensive excellence index that can be compared on a unified scale. Furthermore, the distribution of the optimal projection direction across various evaluation dimensions is combined to perform a diagnostic decomposition of the comprehensive excellence index, forming diagnostic analysis results oriented towards achievement comparison, shortcoming identification, and optimization and improvement. This ensures that the evaluation output not only reflects the overall level of the achievements but also supports subsequent iterative optimization.

[0035] In one embodiment, in step S10, based on the evaluation requirements of the existing power grid artificial intelligence achievements, a corresponding evaluation index system is constructed, including: S11: Determine the target layer, criterion layer, element layer, and indicator layer of the evaluation indicator system.

[0036] Specifically, based on the need to consider the hierarchical characteristics of overall evaluation, dimensional evaluation, element evaluation, and indicator evaluation in the evaluation of the excellence of power grid artificial intelligence achievements, the evaluation indicator system is divided into four levels: target layer, criterion layer, element layer, and indicator layer. The target layer is used to carry the overall evaluation output, the criterion layer is used to carry the core evaluation dimensions of the comprehensive performance of the achievements, the element layer is used to further decompose each core dimension, and the indicator layer is used to carry the three-level quantifiable indicators that can be directly obtained, directly calculated, and directly compared. This establishes a multi-layered evaluation structure that unfolds layer by layer from top to bottom and converges layer by layer from bottom to top.

[0037] S12: Set the target layer as the power grid artificial intelligence achievement excellence index.

[0038] Specifically, the target layer is uniformly set as the power grid artificial intelligence achievement excellence index, and this index is used as the final output to uniformly represent the comprehensive performance of individual achievements. This allows multi-dimensional information such as technical performance, scenario adaptation, business effect, and application operation and maintenance to be converged under the same evaluation target. At the same time, this excellence index is used as the result expression after the subsequent comprehensive projection value normalization processing, so as to make intuitive comparison and ranking among different achievements.

[0039] S13: Set the criteria layer as technological innovation capability, scenario expansion capability, business support capability, and convenient application capability.

[0040] Specifically, the criteria layer is divided into four evaluation dimensions: technological innovation capability, scenario expansion capability, business support capability, and convenient application capability. Among them, technological innovation capability is used to characterize the technological advancement, integration capability, and architectural scalability of the results; scenario expansion capability is used to characterize the generalization and reusability potential of the results in different scenarios; business support capability is used to characterize the quality improvement and efficiency enhancement effect of the results on actual business; and convenient application capability is used to characterize the ease of use, response timeliness, and operation and maintenance support level of the results, so that the evaluation process can focus on the real business value of the power grid artificial intelligence results.

[0041] S14: Based on the evaluation requirements corresponding to the criteria layer, the element layer and indicator layer are refined and configured to obtain the evaluation indicator system.

[0042] Specifically, under the dimension of technological innovation capability, indicators such as multimodal fusion capability score, complex task decomposition and execution capability score, cross-network and provincial reuse frequency, and data feedback capability score are configured. Under the dimension of scenario expansion capability, indicators such as adaptation and transformation cost, scenario generalization and adaptation score, and promotion and development potential score are configured. Under the dimension of business support capability, indicators such as task execution accuracy, percentage improvement in business efficiency, strategy compliance score, and understanding feedback mechanism score are configured. Under the dimension of convenient application capability, indicators such as mean time between failures (MTBF), fault recovery capability score, end-to-end latency of critical tasks, and response processing timeliness score are configured. These indicators are then assigned to the corresponding element layer and indicator layer, ultimately forming an evaluation indicator system that can cover the comprehensive performance of the results during the operational period.

[0043] In one embodiment, step S20, namely obtaining the preprocessed index data matrix of each existing power grid's artificial intelligence achievements within its operating cycle, includes: S21: Obtain the data of each indicator in the evaluation indicator system corresponding to the artificial intelligence achievements of each completed power grid during the operation cycle.

[0044] Specifically, according to the established evaluation index system, data items corresponding to each evaluation index are extracted from the operation data, business data, scenario data and application data continuously generated by the artificial intelligence achievements of each completed power grid during the operation cycle. This ensures that each achievement forms an original value that can participate in subsequent evaluation calculations for each evaluation index, thereby establishing a data correspondence between achievements and indicators.

[0045] S22: Organize the data of each indicator according to the evaluation index system to obtain the original evaluation data set.

[0046] Specifically, the original indicator data corresponding to each result are organized according to the organizational relationship of the target layer, criterion layer, element layer and indicator layer in the evaluation indicator system. First, the indicator data collected separately under the same result are merged into a unified record. Then, the data of different results on the same indicator are aligned according to the unified field order. Finally, a set of original evaluation data that can directly enter the matrix construction stage is obtained with the result as the object and the indicator as the dimension.

[0047] S23: Perform dimensionless processing on the original evaluation data set to generate preprocessed index data corresponding to the artificial intelligence achievements of each completed power grid.

[0048] Specifically, the extreme value method is used to perform dimensionless processing on each indicator in the original evaluation dataset, and a formula is used for positive indicators. For contrarian indicators, a formula is used. , where x ij Let max(x) represent the original value of the i-th completed power grid artificial intelligence achievement on the j-th indicator.j ) represents the maximum value of the j-th indicator among all results, min(x j ) represents the minimum value of the j-th indicator among all results, y ij This represents the index value after dimensionless processing. This processing eliminates the differences in dimensions and values ​​between different indicators, generating preprocessed index data corresponding to each result.

[0049] S24: Construct an indicator data matrix based on the preprocessed indicator data.

[0050] Specifically, the preprocessed indicator data corresponding to all results are arranged and combined according to the result number and indicator number to construct an indicator data matrix. Where m represents the number of evaluated results, n represents the number of tertiary indicators, the i-th row of the matrix represents the preprocessing result of the i-th result on all tertiary indicators, and the j-th column of the matrix represents the preprocessing result of the j-th tertiary indicator on all results. Thus, the indicator data matrix that is uniformly called by the subsequent evaluation model is obtained.

[0051] In one embodiment, step S30, namely, constructing an evaluation model based on the indicator data matrix and the evaluation indicator system, and optimizing the evaluation model to obtain the optimal projection direction, includes: S31: Construct the projection value calculation relationship for each completed power grid artificial intelligence achievement based on the indicator data matrix.

[0052] Specifically, each outcome in the indicator data matrix is ​​treated as an n-dimensional evaluation vector, and an n-dimensional unit projection direction vector is set. , where a j This represents the component of the j-th index in the projection direction, and a one-dimensional projection value calculation relationship is established based on this. , where z i Let y represent the one-dimensional projection value of the i-th result along the projection direction a. ij This represents the preprocessed value of the i-th result on the j-th indicator, m represents the number of results, and n represents the number of third-level indicators. This calculation relationship maps the high-dimensional indicator information to the subsequent comparable one-dimensional projection space.

[0053] S32: Construct a projection index function to characterize the projection effect based on the relationship between projection values, and obtain the evaluation model.

[0054] Specifically, after obtaining the one-dimensional projection values ​​of all results, a projection index function is constructed. S z D represents the standard deviation of the projected values. z The local density of the projected values ​​is represented by the formula. calculate, The average value represents all projected values; local density is expressed by the formula... Calculation, where The distance between the projected values ​​of the i-th and k-th results is represented by R, which represents the density window width parameter, usually taken as 0.1Sz. u(t) represents the unit step function, which takes a value of 1 when t≥0 and a value of 0 when t<0. By coupling the standard deviation and local density into the same objective function, an evaluation model that can simultaneously reflect global differences and local commonalities is obtained.

[0055] S33: Perform orientation optimization on the evaluation model to obtain the optimal projection orientation.

[0056] Specifically, with the maximum value of the projection index function Q(a) as the optimization objective, the projection direction in the evaluation model is iteratively searched so that the projection direction continuously converges towards the direction that best reveals the characteristics of the original data structure under the premise of satisfying the unit vector constraint. Thus, the optimal projection direction a* corresponding to the maximum projection index function value is output, and this optimal projection direction is used as the common basis for subsequent comprehensive projection value calculation, excellence index normalization processing, dimensional analysis, and diagnostic analysis.

[0057] In one embodiment, step S33, namely, performing orientation optimization on the evaluation model to obtain the optimal projection orientation, includes: S331: Encode the candidate directions to obtain multiple individual directions to be solved.

[0058] Specifically, the projection direction vector Candidate directions are encoded as chromosomes and represented using real-number encoding. Multiple initial projection directions are then randomly generated as individual directions to be solved, each satisfying the following conditions: The unit vector constraint enables the subsequent direction optimization process to be carried out within a unified feasible region.

[0059] S332: Based on the projection index function, perform an adaptive evaluation on each individual in the direction to be solved, and obtain the corresponding evaluation results.

[0060] Specifically, each individual direction to be solved is substituted into the projection index function. The corresponding function value is calculated and used as the individual fitness assessment result. The larger the function value, the more the current direction can make the projection value have a larger overall dispersion and better local clustering characteristics. This forms a set of fitness assessment results for all individuals in the directions to be solved.

[0061] S333: Based on the evaluation results, iteratively update the individuals in each direction to be solved to obtain the individuals in the target direction.

[0062] Specifically, based on the results of the fitness assessment, an iterative update process of individual selection, individual crossover, and individual mutation is performed. In the selection phase, a roulette wheel method is used to increase the probability of high-fitness individuals being retained. In the crossover phase, arithmetic crossover is used to generate new offspring individuals in the target direction. In the mutation phase, chromosomes are perturbed with a small probability to increase population diversity. The iteration ends when the maximum number of iterations is reached or the fitness change is less than a preset threshold, and individuals in the target direction are retained from the final population.

[0063] S334: Determine the optimal projection direction based on the target direction individual.

[0064] Specifically, the target direction individuals that are ultimately retained are taken as the optimal direction output, and the direction components corresponding to each index in the target direction individuals are determined as the component values ​​of the optimal projection direction a*, so that the optimal projection direction can be directly substituted into the comprehensive projection value calculation formula and the subsequent dimensional analysis formula.

[0065] In one embodiment, step S40 involves determining the comprehensive excellence index of each completed power grid AI achievement based on the optimal projection direction, and determining the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction, including: S41: Determine the comprehensive projection value corresponding to the artificial intelligence achievements of each existing power grid based on the optimal projection direction.

[0066] Specifically, the optimal projection direction Substituting these values ​​into the projection value calculation relationship, we obtain the comprehensive projection value of each result. , where z i This represents the comprehensive projection value of the i-th completed power grid AI achievement in the optimal projection direction. y represents the directional component of the optimal projection direction on the j-th index. ij This represents the preprocessed value of the i-th result on the j-th indicator. The multidimensional information of each result on all indicators is compressed into a single comprehensive representation value through the comprehensive projection value.

[0067] S42: Determine the comprehensive excellence index corresponding to the artificial intelligence achievements of each completed power grid based on the comprehensive projection value.

[0068] Specifically, the comprehensive projection value of all results is normalized to obtain the comprehensive excellence index. GEM i This represents the overall excellence index of the i-th achievement. Let z represent the comprehensive projection value of the i-th result, max(z*) represent the maximum value among all comprehensive projection values ​​of all results, and min(z*) represent the minimum value among all comprehensive projection values ​​of all results. The comprehensive excellence index obtained after this normalization process is between 0 and 1. The larger the value, the higher the comprehensive excellence of the corresponding result.

[0069] S43: Group the optimal projection directions according to the criteria layer in the evaluation index system to obtain the dimensional analysis results corresponding to each criteria layer.

[0070] Specifically, the optimal projection direction is divided into four groups according to the criteria layer: technological innovation capability, scenario expansion capability, business support capability, and convenient application capability. The sum of the directional components of each group is calculated to obtain the result. Where T, S, B, and C represent the sets of indicators belonging to technological innovation capability, scenario expansion capability, business support capability, and convenient application capability, respectively, and W T W S W B W C These represent the projected weights corresponding to the four criterion layers, respectively. The grouping results are used to form dimensional analysis results that reflect the relative importance of each evaluation dimension.

[0071] S44: Determine the diagnostic analysis results corresponding to the artificial intelligence achievements of each completed power grid based on the comprehensive excellence index and dimensional analysis results.

[0072] Specifically, after obtaining the comprehensive excellence index and dimensional analysis results, the comprehensive excellence index is used to lock in the overall performance level for the diagnostic task of a single achievement or multiple achievements, and the dimensional analysis results are used to break down the contribution of different evaluation dimensions. Based on this, diagnostic analysis results are formed for the weak areas, key influencing indicators and improvement directions of the achievements, so that the evaluation output is extended from a simple ranking result to an interpretable diagnostic result.

[0073] In one embodiment, step S44, which involves determining the diagnostic analysis results corresponding to each completed power grid artificial intelligence achievement based on the comprehensive excellence index and dimensional analysis results, includes: S441: Based on the comprehensive excellence index, the target has been established as an artificial intelligence achievement for the power grid.

[0074] Specifically, the index value of a single achievement is read from the comprehensive excellence index sequence corresponding to all completed power grid artificial intelligence achievements, and the achievement is selected as the object of subsequent diagnostic analysis, so that subsequent identification of weak dimensions, analysis of key indicators and generation of optimization suggestions are all centered around the same goal of completed power grid artificial intelligence achievements.

[0075] S442: Based on the dimensional analysis results, determine the weak dimensions corresponding to the artificial intelligence achievements of the target power grid.

[0076] Specifically, first calculate the projection components of the target existing power grid artificial intelligence achievements on four evaluation dimensions, namely... , where z iT z iS ziB z iC The dimensions of the target outcome are represented by their respective projection components in terms of technological innovation capability, scenario expansion capability, business support capability, and convenient application capability. Then, the projection components of each dimension are compared with the average or optimal value of similar outcomes to determine the dimension with the largest deviation as the weak dimension corresponding to the target outcome.

[0077] S443: Conduct contribution analysis on the indicator layer in the evaluation indicator system based on the weak dimensions to determine the key indicators.

[0078] Specifically, after identifying the weak dimension, the contribution of each tertiary indicator under that weak dimension is further analyzed, and the difference between the indicator j with a significantly low contribution and the optimal value is calculated. The difference is expressed using the formula... Calculate, where δ ij This represents the gap value of the i-th result on the j-th indicator. y represents the directional component of the optimal projection direction on the j-th index. 0j Let y represent the optimal value corresponding to the j-th indicator. ij This represents the preprocessing value of the i-th result on the j-th indicator. The larger the difference, the greater the impact of the indicator on the overall score. Based on this, the key indicators that need to be optimized first are determined.

[0079] S444: Generate corresponding optimization suggestions based on key indicators, and combine the weak dimensions, key indicators and optimization suggestions as the diagnostic analysis results.

[0080] Specifically, based on the weaknesses corresponding to key indicators, a pre-set optimization knowledge base is invoked to generate targeted optimization suggestions. When the understanding feedback mechanism score is low, suggestions are given to optimize the human-computer interaction interface design, establish a closed-loop user feedback mechanism, and improve the accuracy of command intent understanding and response adaptation capabilities. When the cost of model fine-tuning is high, suggestions are given to introduce lightweight processing such as model quantization and knowledge distillation to reduce the consumption of computing resources for model iteration and adaptation. When the cost of adaptation and transformation is high, suggestions are given to adopt a modular architecture design, decouple functional components from business scenarios, and reduce the overhead of secondary development and adaptation in different scenario deployments. When the end-to-end latency of critical tasks is high, suggestions are given to optimize model inference efficiency, implement model structure simplification and operator fusion, or upgrade computing resource configuration to meet the real-time requirements of business. Finally, the weak dimensions, key indicators, and optimization suggestions are combined into diagnostic analysis results.

[0081] Example 2 like Figure 2 As shown, based on the same inventive concept as the above embodiments, the present invention also provides a power grid artificial intelligence achievement excellence evaluation system, comprising: The system construction module is used to build a corresponding evaluation index system based on the evaluation needs of the existing power grid artificial intelligence achievements; The data matrix module is used to obtain the preprocessed indicator data matrix of the artificial intelligence achievements of each built power grid during the operation cycle; The model solving module is used to construct an evaluation model based on the indicator data matrix and the evaluation indicator system, and to optimize and solve the evaluation model to obtain the optimal projection direction. The results diagnosis module is used to determine the comprehensive excellence index of each completed power grid artificial intelligence achievement based on the optimal projection direction, and to determine the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction.

[0082] Optionally, the system building modules include: The hierarchy determination submodule is used to determine the target layer, criterion layer, element layer, and indicator layer of the evaluation indicator system. The target setting submodule is used to set the target layer as the power grid artificial intelligence achievement excellence index; The criteria setting submodule is used to set the criteria layer as technological innovation capability, scenario expansion capability, business support capability, and convenient application capability; The detailed configuration submodule is used to refine the configuration of the element layer and the indicator layer according to the evaluation requirements corresponding to the criterion layer, so as to obtain the evaluation indicator system.

[0083] Optionally, the data matrix module includes: The data acquisition submodule is used to acquire the data of each indicator in the evaluation indicator system corresponding to the artificial intelligence achievements of each completed power grid during the operating cycle; The relationship organization submodule is used to organize the corresponding relationships of each indicator data according to the evaluation indicator system to obtain the original evaluation data set. The dimensionless submodule is used to perform dimensionless processing on the original evaluation data set to generate preprocessed index data corresponding to the artificial intelligence achievements of each built power grid. The matrix construction submodule is used to construct an indicator data matrix based on the preprocessed indicator data.

[0084] Optionally, the model solver module includes: The relationship construction submodule is used to construct the projection value calculation relationship corresponding to each completed power grid artificial intelligence achievement based on the indicator data matrix. The function construction submodule is used to construct a projection index function to characterize the projection effect based on the projection value calculation relationship, and obtain the evaluation model; The orientation optimization submodule is used to perform orientation optimization on the evaluation model to obtain the optimal projection orientation.

[0085] Optional, the direction optimization submodule includes: The encoding processing unit is used to encode the candidate directions to obtain multiple individual directions to be solved; The adaptive evaluation unit is used to perform adaptive evaluation on each individual in the direction to be solved based on the projection index function, and obtain the corresponding evaluation results; The iterative update unit is used to iteratively update the individuals in each direction to be solved based on the evaluation results, so as to obtain the individuals in the target direction. The direction determination unit is used to determine the optimal projection direction based on the target direction individual.

[0086] Optionally, the result diagnostic module includes: The projection calculation submodule is used to determine the comprehensive projection value corresponding to the artificial intelligence results of each existing power grid based on the optimal projection direction; The index determination submodule is used to determine the comprehensive excellence index corresponding to each artificial intelligence achievement of the existing power grid based on the comprehensive projection value. The grouping analysis submodule is used to group the optimal projection direction according to the criteria layer in the evaluation index system, and obtain the dimensional analysis results corresponding to each criteria layer. The results determination submodule is used to determine the diagnostic analysis results corresponding to each completed power grid artificial intelligence achievement based on the comprehensive excellence index and dimensional analysis results.

[0087] Optionally, the result determination submodule includes: The target determination unit is used to determine the completed artificial intelligence achievements of the power grid based on the comprehensive excellence index. The dimension determination unit is used to determine the weak dimensions corresponding to the AI ​​achievements of the target power grid based on the dimensional analysis results. The contribution analysis unit is used to analyze the contribution of the indicator layer in the evaluation indicator system based on the weakness dimension, and to identify the key indicators. The suggestion generation unit is used to generate corresponding optimization suggestions based on key indicators, and combines the weak dimensions, key indicators and optimization suggestions as the diagnostic analysis results.

[0088] Example 3 like Figure 3 As shown, the present invention also provides an electronic device 100 for implementing a method for evaluating the excellence of artificial intelligence achievements in power grids; The electronic device 100 includes a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on at least one processor 102, and at least one communication bus 104.

[0089] The memory 101 can be used to store computer program 103. The processor 102 implements the steps of the power grid artificial intelligence achievement excellence evaluation method of Embodiment 1 by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101.

[0090] The memory 101 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created based on the use of the electronic device 100 (such as audio data), etc. In addition, the memory 101 may include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other non-volatile solid-state storage device.

[0091] At least one processor 102 may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 102 may be a microprocessor or any conventional processor. Processor 102 is the control center of electronic device 100, connecting various parts of electronic device 100 via various interfaces and lines.

[0092] The memory 101 in the electronic device 100 stores multiple instructions to implement a method for evaluating the excellence of artificial intelligence achievements in power grids, and the processor 102 can execute multiple instructions to achieve the following: Based on the evaluation needs of the existing power grid artificial intelligence achievements, a corresponding evaluation index system is constructed. Obtain the preprocessed indicator data matrix of artificial intelligence achievements of each built power grid during its operating cycle; An evaluation model is constructed based on the indicator data matrix and the evaluation indicator system, and the evaluation model is optimized and solved to obtain the optimal projection direction; The comprehensive excellence index of artificial intelligence achievements of each completed power grid is determined based on the optimal projection direction, and the diagnostic analysis results are determined based on the comprehensive excellence index and the optimal projection direction.

[0093] Example 4 If the modules / units integrated in the electronic device 100 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or system capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, and read-only memory (ROM).

[0094] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0095] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A system that specifies functions in one or more boxes.

[0096] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including an instruction set implemented in a process. Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0097] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0098] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for evaluating the excellence of artificial intelligence achievements in power grids, characterized in that, The method includes: Based on the evaluation needs of the existing power grid artificial intelligence achievements, a corresponding evaluation index system is constructed. Obtain the preprocessed indicator data matrix of each of the aforementioned artificial intelligence achievements of the built power grid within the operating cycle; An evaluation model is constructed based on the indicator data matrix and the evaluation indicator system, and the evaluation model is optimized and solved to obtain the optimal projection direction; The comprehensive excellence index of each of the existing power grid artificial intelligence achievements is determined based on the optimal projection direction, and the diagnostic analysis results are determined based on the comprehensive excellence index and the optimal projection direction.

2. The method for evaluating the excellence of power grid artificial intelligence achievements according to claim 1, characterized in that, Based on the evaluation requirements of the existing power grid artificial intelligence achievements, a corresponding evaluation index system is constructed, including: Determine the target layer, criterion layer, element layer, and indicator layer of the evaluation index system; The target layer is set as the power grid artificial intelligence achievement excellence index; The criteria layer is set as technological innovation capability, scenario expansion capability, business support capability, and convenient application capability; Based on the evaluation requirements corresponding to the criterion layer, the element layer and the indicator layer are refined and configured to obtain the evaluation indicator system.

3. The method for evaluating the excellence of power grid artificial intelligence achievements according to claim 1, characterized in that, The process of obtaining the preprocessed indicator data matrix of each of the existing power grid artificial intelligence achievements within its operating cycle includes: Obtain the data of each indicator in the evaluation indicator system corresponding to the artificial intelligence achievements of the existing power grids during the operating cycle; Based on the evaluation index system, the data of each index are organized to determine the correspondence between the indexes, thus obtaining the original evaluation data set. The original evaluation data set is dimensionless to generate preprocessed index data corresponding to each of the artificial intelligence achievements of the built power grid. The indicator data matrix is ​​constructed based on the preprocessed indicator data.

4. The method for evaluating the excellence of power grid artificial intelligence achievements according to claim 1, characterized in that, The step of constructing an evaluation model based on the indicator data matrix and the evaluation indicator system, and optimizing the evaluation model to obtain the optimal projection direction, includes: Based on the aforementioned indicator data matrix, construct the projection value calculation relationship corresponding to each of the existing power grid artificial intelligence achievements; Based on the relationship between the projection values, a projection index function is constructed to characterize the projection effect, thus obtaining the evaluation model; The evaluation model is subjected to directional optimization processing to obtain the optimal projection direction.

5. The method for evaluating the excellence of power grid artificial intelligence achievements according to claim 4, characterized in that, The process of performing direction optimization on the evaluation model to obtain the optimal projection direction includes: The candidate directions are encoded to obtain multiple individual directions to be solved; Based on the projection index function, an adaptive evaluation is performed on each individual in the direction to be solved, and the corresponding evaluation results are obtained; Based on the evaluation results, each individual in the direction to be solved is iteratively updated to obtain the individual in the target direction. The optimal projection direction is determined based on the target direction individual.

6. The method for evaluating the excellence of power grid artificial intelligence achievements according to claim 1, characterized in that, The process of determining the comprehensive excellence index of each of the existing power grid artificial intelligence achievements based on the optimal projection direction, and determining the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction, includes: The comprehensive projection value corresponding to each of the artificial intelligence achievements of the existing power grid is determined according to the optimal projection direction. The comprehensive excellence index corresponding to each of the completed power grid artificial intelligence achievements is determined based on the comprehensive projection value. The optimal projection direction is grouped according to the criteria layer in the evaluation index system to obtain the dimensional analysis results corresponding to each criteria layer. Based on the comprehensive excellence index and the dimensional analysis results, the diagnostic analysis results corresponding to each of the established power grid artificial intelligence achievements are determined.

7. The method for evaluating the excellence of power grid artificial intelligence achievements according to claim 6, characterized in that, The step of determining the diagnostic analysis results corresponding to each of the existing power grid artificial intelligence achievements based on the comprehensive excellence index and the dimensional analysis results includes: Based on the comprehensive excellence index, the target has been established as an achievement in power grid artificial intelligence. Based on the dimensional analysis results, the weak dimensions corresponding to the existing power grid artificial intelligence achievements of the target are determined. Based on the aforementioned weak dimensions, a contribution analysis is performed on the indicator layer of the evaluation indicator system to determine the key indicators; Based on the key indicators, corresponding optimization suggestions are generated, and the weak dimension, the key indicators, and the optimization suggestions are combined as the diagnostic analysis result.

8. A system for evaluating the excellence of artificial intelligence achievements in power grids, characterized in that, The system includes: The system construction module is used to build a corresponding evaluation index system based on the evaluation needs of the existing power grid artificial intelligence achievements; The data matrix module is used to obtain the preprocessed indicator data matrix of the artificial intelligence achievements of each built power grid during the operation cycle; The model solving module is used to construct an evaluation model based on the indicator data matrix and the evaluation indicator system, and to optimize and solve the evaluation model to obtain the optimal projection direction. The results diagnosis module is used to determine the comprehensive excellence index of each completed power grid artificial intelligence achievement based on the optimal projection direction, and to determine the diagnostic analysis results based on the comprehensive excellence index and the optimal projection direction.

9. An electronic device, characterized in that, It includes a processor and a memory, the processor being used to execute a computer program stored in the memory to implement the power grid artificial intelligence achievement excellence evaluation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one instruction, which, when executed by a processor, implements the power grid artificial intelligence achievement excellence evaluation method as described in any one of claims 1 to 7.

Images