Electric power material detection task generation method and device

Abstract

The invention discloses an electric power material detection task generation method and device, and relates to the technical field of electric power material detection, and the method comprises the specific steps: giving out a plurality of target features based on the analysis of to-be-detected electric power material information, processing the target features through an electric power material clustering classification model, and obtaining an electric power material detection task; detection classification of the electric power materials is given; obtaining a candidate detection item set in combination with the detection classification of the electric power materials and a pre-constructed detection demand model; based on a pre-established detection method library, a multi-objective optimization function is used, and through iterative optimization of a genetic algorithm, an optimal detection item set is given; and matching a corresponding detection method for each detection item in the optimal detection item set by combining real-time detection resources and time cost and adopting a dynamic fitness field matching algorithm, and generating a detection task list. According to the invention, efficient utilization of detection resources is realized, and the detection precision, the detection reliability and the cost effectiveness of electric power materials are considered.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power material detection, and in particular relates to a method and device for generating electric power material detection tasks. Background Art

[0002] As a key link in ensuring the health of power equipment, power material testing faces the problems of heavy testing tasks, uneven resource allocation, and difficulty in balancing testing efficiency and accuracy.

[0003] Traditional inspection models rely on manual experience to develop inspection plans, which can be highly subjective, lack coverage, and waste resources. Currently, methods for inspecting power supplies rely on static rules to implement inspection and tracking. For example, patent CN114092047A describes a system and method for tracing power supply inspection information. This system, based on blockchain technology, comprises a data layer, a network layer, a consensus layer, a contract layer, and an application layer. To trace power supplies, this system first obtains inspection information, verifies and encrypts it to generate a genesis block, then obtains lifecycle information, verifies and encrypts it to generate other blocks. During the block generation process, a block finality mechanism in the contract layer prevents malicious tampering. Finally, based on the genesis block and other blocks, the entire lifecycle of the power supply can be traced. The design of system and human identification IDs ensures that the physical power supply's offline ID matches the online system ID, facilitating direct access to online data by offline personnel.

[0004] Furthermore, power material inspection involves multiple types of equipment, multiple inspection items, and complex scenarios, placing higher demands on the dynamic management of inspection cycles, indicator thresholds, and association rules. Relying on manual experience or static rules presents the following drawbacks: First, historical inspection data is underutilized, lacking in-depth analysis and standardized processing of multiple types of reports, resulting in data fragmentation and inefficient extraction of key information. Second, the construction of inspection demand models is crude, failing to fully consider material status characteristics, historical abnormal data, and rule correlations, which can easily lead to over-inspection or missed inspections. Third, inspection item screening lacks a multi-objective optimization mechanism, making it difficult to balance comprehensiveness, resource adaptability, and time efficiency, resulting in insufficiently rational inspection plans. Finally, the matching of inspection methods and resources relies on fixed rules, failing to dynamically adapt to real-time resource status and lacking flexible adjustment strategies in the event of resource conflicts. Furthermore, traditional methods cannot automatically generate inspection task lists, requiring manual information integration, which is inefficient and prone to errors.

[0005] Therefore, there is an urgent need for an intelligent and systematic task generation method to achieve the optimal configuration of testing resources and the precise matching of testing needs. Summary of the Invention

[0006] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide a task generation method and device for power material detection. By constructing a semantic tensor extraction model for power materials, in-depth analysis and standardized storage of historical detection reports are achieved; combined with the detection demand model, clustering classification algorithm and multi-objective optimization function, the detection demand is accurately predicted and the optimal detection items are screened; the dynamic fitness field algorithm is used to match the detection method and resources to generate an executable detection task list.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: On the one hand, a method for generating tasks for power material detection, the specific steps of the evaluation method are as follows: Based on the analysis of the information of the power materials to be tested, multiple target features are given, and the target features are processed through the power material clustering classification model to give the detection classification of power materials; Combining the inspection classification of power materials and the pre-built inspection demand model, a set of candidate inspection items is obtained; Based on the pre-established detection method library, the optimal detection item set is given by using multi-objective optimization function and iterative optimization of genetic algorithm; Combining real-time detection resources and time costs, and adopting a dynamic fitness field matching algorithm, each detection item in the optimal detection item set is matched with a corresponding detection method to generate a detection task list.

[0008] Furthermore, the target features include material type, core performance parameters, and historical detection anomaly rate. The power material clustering classification model is used to detect and classify power materials, specifically including: Obtain multiple target features and fuse them into cluster feature vectors; Through the power material clustering classification model and the manifold clustering algorithm, the manifold clustering distance is obtained; Based on the manifold clustering distance, the power materials to be tested are clustered and classified, and the detection classification of the corresponding power materials to be tested is given.

[0009] Furthermore, based on the manifold clustering distance, the power materials to be tested are clustered and classified, and the detection classification of the corresponding power materials to be tested is given, specifically including: Select clustering algorithm based on manifold cluster distance; Determine the number of clusters by combining domain knowledge and effectiveness indicators; Perform clustering using the selected clustering algorithm to divide the power materials to be tested into several clusters; Count the characteristic distribution of each cluster and obtain the risk characteristics of each cluster; According to the risk characteristics of the cluster, the risk label is obtained, and the risk label and material type are integrated to give the detection classification of the corresponding power material to be detected.

[0010] Furthermore, the construction of the detection demand model specifically includes: Obtain historical inspection reports and inspection requirement rule sets for multiple types of power materials; The power material semantic tensor extraction model is used to parse historical inspection reports of various types of power materials, extract target information, and build a standardized information database; Combine the standardized information database and the detection requirement rule set to build a detection demand model.

[0011] Furthermore, the target information includes material type, test items, and test result data; Combining the standardized information base and the detection requirement rule set, a detection requirement model is constructed, specifically including: Obtain material type, test items, and test result data from the standardized information database, convert the test requirement rule set into a rule field, form a structured data set, and divide it into a training set and a validation set; Obtain the initial detection demand model, and give the rule field gravity, historical data potential, and rule deviation penalty based on the training set; Combining the rule field gravity, historical data potential, and rule deviation penalties, we can obtain the required field strength for each test item. According to the required detection field strength, a set of candidate detection items is given; The parameters of the initial detection requirement model are optimized through training, and verified and adjusted using the validation set to obtain the detection requirement model.

[0012] Furthermore, the construction of the detection method library specifically includes: A detection method set, which includes multiple detection methods and detection items corresponding to each detection method; Detection attribute data, including resource consumption parameters and performance indicator data. Resource consumption parameters include detection equipment occupancy, manpower requirements, and time cost. Performance indicator data includes detection accuracy, applicable scenarios, and historical anomaly correlation. Dynamic detection adaptation rules, which include detection priority strategy and detection substitution strategy.

[0013] Furthermore, the construction of the multi-objective optimization function specifically includes: Based on the standardized information base and the set of testing requirement rules, a quality risk set is given, and the coverage of the candidate testing item set on the quality risk is determined; Combined with the detection resource upper limit, the adaptability of the candidate detection project set and the available resources is obtained; Combined with the time cost upper limit, the time efficiency of the candidate detection item set is obtained; Combining coverage, adaptability, and time efficiency, a multi-objective optimization function is constructed, and constraints are set. The constraints include mandatory inspection item constraints, resource upper limit constraints, and time upper limit constraints.

[0014] Furthermore, a dynamic adaptive field matching algorithm is used to match a corresponding detection method for each detection item in the optimal detection item set, specifically including: For each detection item in the optimal detection item set, obtain the detection efficiency, matching degree with real-time resources, and task urgency of each detection method associated with the detection item; The dynamic adaptability of each detection method is obtained by combining its detection efficiency, matching degree with real-time resources, and task urgency. The dynamic adaptability of each detection method is compared, and the detection method with the largest dynamic adaptability is selected as the detection method matching the current detection item, until all detection items in the optimal detection item set are matched.

[0015] Furthermore, it also includes: Generate a limited number of effective combination detection methods based on the detection methods in the detection method library; The dynamic adaptability of each combined detection method is obtained by combining its detection efficiency, matching degree with real-time resources, task urgency, tunneling probability, and combined gain. Compare the dynamic adaptability of each combination detection method, and select the combination detection method with the largest dynamic adaptability as the combination detection method that matches the current detection item; Among them, the dynamic adaptability of the combined detection method is specifically expressed as: ; Among them, D j is the dynamic fitness of the jth combination detection method, m is the total number of combination detection methods, Eff(M j ) is a combined detection method M j The detection efficiency of Res(M j ,Z) is the combined detection method M j The matching degree with real-time resource Z, Urgent(T) is the urgency of the task, T represents the remaining time of the task, P tunnel is the tunneling probability, is the combined gain; The tunneling probability is specifically expressed as: ; Among them, h is the resource adaptation constant, is the resource equivalent quality, V barrier is the resource barrier, and D is the resource substitution distance; Resource barriers are specifically expressed as: ; in, is the weight coefficient; is the deviation in detection accuracy between the combined detection method and the replaced detection method, is the resource consumption cost ratio of the combined detection method to the replaced detection method, T win is the time window pressure; Combined Gain Specifically expressed as: ;

[0016] Where n is the number of detection methods that make up the combined detection method; Con k is the confidence of the k-th detection method, is the order penalty factor, and OP is the execution order penalty.

[0017] In another aspect, a device for generating a power material detection task adopts the above-mentioned method for generating a power material detection task, and the device includes: The detection clustering classification module is used to analyze the information of the power materials to be detected, give multiple target features, process the target features through the power material clustering classification model, and give the detection classification of the power materials; The test item acquisition module is used to obtain a set of candidate test items by combining the test classification of power materials and the pre-built test demand model; The detection project optimization module is used to provide the optimal detection project set based on the pre-established detection method library, using multi-objective optimization functions and iterative optimization using genetic algorithms; The detection method matching module is used to combine real-time detection resources and time costs, and adopt a dynamic fitness field matching algorithm to match the corresponding detection method for each detection item in the optimal detection item set and generate a detection task list.

[0018] The present invention provides a method and device for generating tasks for power material detection, which have at least the following beneficial effects: (1) By constructing a semantic tensor extraction model for power materials, this paper achieves in-depth analysis of historical inspection reports and standardized data storage, improving the efficiency and accuracy of data processing. Through semantically enhanced tensors and contextual attention mechanisms, it accurately extracts key information on material types, inspection items, and results, providing a data foundation for the construction of subsequent inspection demand models. In addition, the inspection demand model achieves accurate prediction and dynamic adjustment of inspection demand through the comprehensive calculation of rule field gravity, historical data potential, and rule deviation penalties, effectively avoiding over-inspection or missed inspections, and improving the scientific nature and pertinence of inspection work.

[0019] (2) The present invention selects material type, core performance parameters and historical detection anomaly rate as feature vectors and uses the manifold clustering algorithm for quantitative classification, thereby achieving refined management of power materials. In terms of detection project optimization, this method takes detection comprehensiveness, resource adaptability and time efficiency as the goals, constructs a multi-objective optimization function, and uses genetic algorithms for iterative optimization to output the optimal detection project set. This not only ensures the coverage and effectiveness of the detection projects, but also improves the utilization efficiency of detection resources and reduces detection costs. The introduction of the dynamic adaptability field detection method matching algorithm further enhances the matching accuracy between the detection method and resources, providing a strong guarantee for the safe operation of power materials.

[0020] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A flow chart of a method for generating a power material detection task provided by the present invention; Figure 2 This is a flow chart of detecting and classifying electric power materials using an electric power material clustering classification model according to an embodiment of the present invention; Figure 3 This is a diagram of the power material clustering classification model architecture according to one embodiment of the present invention; Figure 4 This is a flow chart of detecting and classifying based on an example of manifold clustering according to an embodiment of the present invention; Figure 5 This is a diagram of the power material semantic tensor extraction model architecture according to one embodiment of the present invention; Figure 6 Constructing a flow chart for a detection requirement model of a certain embodiment of the present invention; Figure 7 Constructing a flow chart for a multi-objective optimization function according to an embodiment of the present invention; Figure 8 This is a schematic diagram of a module of a power material detection task generation device provided by the present invention. DETAILED DESCRIPTION

[0022] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods. Obviously, the embodiments described 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 technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0023] The terms used in the embodiments of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The singular forms "a," "an," "the," and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, and unless the context clearly indicates otherwise, "a plurality" generally includes at least two.

[0024] It should also be noted that the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a product or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such product or device. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the product or device comprising the element.

[0025] like Figure 1 As shown, the present invention provides a method for generating a power material detection task, the specific steps of which include: Based on the analysis of the information of the power materials to be tested, multiple target features are given, and the target features are processed through the power material clustering classification model to give the detection classification of power materials; Combining the inspection classification of power materials and the pre-built inspection demand model, a set of candidate inspection items is obtained; Based on the pre-established detection method library, the optimal detection item set is given by using multi-objective optimization function and iterative optimization of genetic algorithm; Combining real-time detection resources and time costs, and adopting a dynamic fitness field matching algorithm, each detection item in the optimal detection item set is matched with a corresponding detection method to generate a detection task list.

[0026] The system analyzes the information of the power materials to be tested, extracting target features such as material type, core performance parameters, and historical detection anomaly rates. This target feature is then processed using a power material clustering classification model to accurately classify the power materials to be tested, dividing them into different risk levels for testing. The classification results are more closely aligned with the actual status of the materials. By constructing a multi-objective optimization function, it balances comprehensiveness of testing, resource adaptability, and time efficiency, enabling efficient and reasonable selection of required testing items. The dynamic adaptability field matching algorithm adapts to real-time resource changes, optimizes resource utilization, and quickly and efficiently matches the optimal testing method for each test item.

[0027] Based on the analysis of the power material information to be tested, multiple target features are given, including: Collect multi-source information on the power materials to be tested, including basic attributes (such as material type, rated parameters, and years of operation), historical test data (such as historical test items, test results, and historical abnormality rates), and standard requirements (such as mandatory test items and risk association rules in power industry standards). Select features that reflect material risks and meet standard requirements from multiple sources of information to avoid redundancy. For example, select risk-related features, type attribute-related features, and standard requirement-related features as target features. The selected target features are used as input to the power material clustering classification model.

[0028] like Figure 2 As shown in the figure, the target features are processed by the power material clustering classification model to give the detection classification of power materials, including: Obtain multiple target features and fuse them into cluster feature vectors; Through the power material clustering classification model and the manifold clustering algorithm, the manifold clustering distance is obtained; Based on the manifold clustering distance, the power materials to be tested are clustered and classified, and the detection classification of the corresponding power materials to be tested is given.

[0029] In this example, when testing and classifying the power materials to be tested, the power material clustering classification model used is the manifold clustering model enhanced by deep learning, such as Figure 3 As shown, it includes feature preprocessing layer, feature fusion layer, manifold embedding layer, manifold distance calculation layer and clustering classification layer: The feature preprocessing layer standardizes and normalizes target features such as material type, core performance parameters, and historical detection anomaly rates. For example, it encodes the material type, standardizes the rated voltage value in the core performance parameters, and normalizes the historical detection anomaly rate. The feature fusion layer combines the preprocessed target features into a cluster feature vector; The manifold embedding layer uses the isometric mapping algorithm (ISOMAP) or the locally linear embedding algorithm (LLE) to map the concatenated high-dimensional cluster feature vectors into a low-dimensional manifold space. This can preserve the nonlinear structure of high-dimensional data and avoid the loss of key information when mapping high-dimensional data into a low-dimensional space. The manifold distance calculation layer calculates the Riemannian metric in the manifold space and incorporates the risk correction term to obtain the final manifold cluster distance. The Riemannian metric reflects the original geometric differences of the features, and the risk correction term is used to control the impact of historical anomalies on the manifold cluster distance, making it easier for high-risk materials (such as cables with high historical anomaly rates) to be clustered together. The clustering classification layer uses hierarchical clustering or K-means clustering methods to classify according to the final manifold clustering distance, and gives the detection classification of the power materials to be tested, providing a basis for the subsequent acquisition and optimization of detection items (such as adding detection items for high-risk categories).

[0030] Furthermore, the specific expression for obtaining the manifold clustering distance is: ; Among them, d pq is the manifold clustering distance between materials p and q, M is the characteristic manifold space of power materials, Metric(F p , F q ) is the clustering feature vector F in the power material feature manifold space p With F q The Riemann metric of is the risk modifier, is the risk sensitivity coefficient, and Risk(p,q) is the historical abnormal correlation between materials p and q.

[0031] It can be understood that the characteristic manifold space of power materials is a low-dimensional curved space mapped by high-dimensional clustering feature vectors, which is learned from high-dimensional clustering feature vectors through manifold learning algorithms such as ISOMAP and LLE; the Riemann metric is used to measure the similarity between clustering feature vectors, which is the shortest path length between clustering feature vectors in the manifold space. Its calculation depends on the geometric structure of the manifold space. Common calculation methods include geodesic distance, local linear reconstruction distance, probability distribution distance, etc.; the risk correction term consists of two parts: risk sensitivity coefficient and historical anomaly correlation, which is used to adjust the manifold clustering distance between materials, so that materials with high historical anomaly correlation (such as transformers with "oil deterioration" and "partial discharge" anomalies) are clustered. The distance is smaller, so they are more likely to be classified into the same category; the risk sensitivity coefficient can be set according to domain knowledge or adjusted experimentally to control the intensity of risk correction; the historical anomaly correlation is used to measure the degree of correlation between abnormal situations that occurred in the historical detection process of different materials, reflecting the similarity between materials in abnormal types (such as "oil quality deterioration" and "winding insulation reduction") or abnormal frequencies. It is calculated from the historical detection data in the standardized information database. The calculation method that can be used is "the overlap of the historical abnormal types of the two materials × the weighted sum of the number of abnormalities" (for example, the overlap of the "partial discharge" abnormality type of materials p and q is 0.8, and the weighted sum of the number of abnormalities is 0.9, then Risk(p,q)=0.8*0.9=0.72).

[0032] In one embodiment, if Figure 4 As shown in the figure, based on the manifold clustering distance, the power materials to be tested are clustered and classified, and the detection classification of the corresponding power materials to be tested is given, which specifically includes: Select clustering algorithm based on manifold cluster distance; Determine the number of clusters by combining domain knowledge and effectiveness indicators; Perform clustering using the selected clustering algorithm to divide the power materials to be tested into several clusters; Count the characteristic distribution of each cluster and obtain the risk characteristics of each cluster; According to the risk characteristics of the cluster, the risk label is obtained, and the risk label and material type are integrated to give the detection classification of the corresponding power material to be detected.

[0033] Specifically, the manifold clustering distance combines feature similarity (Riemannian metric) and risk correlation (historical anomaly correlation), and selects algorithms that can handle arbitrary distance metrics, such as hierarchical clustering and spectral clustering; Determine the number of clusters by combining domain knowledge (e.g., risk classification for power operation and maintenance) and clustering effectiveness metrics (e.g., silhouette coefficient, elbow method, etc.); Using the selected clustering algorithm and manifold clustering distance as the metric, the power materials to be tested are merged into several clusters. Taking hierarchical clustering as an example, this process specifically includes: (1) Initialize each power material to be detected as an independent cluster; (2) Obtain the average value of the manifold clustering distances between all points in cluster A and all points in cluster B as the inter-cluster distance; (3) Each time, the two clusters with the smallest inter-cluster distance are merged (for example, the manifold clustering distance between "oil-immersed transformers that have been in operation for 10 years" and "oil-immersed transformers that have been in operation for 8 years" is the smallest and they are merged into one cluster); (4) Repeat steps (2) and (3) until the number of clusters is reduced to the determined number of clusters; Count the characteristic distribution of each cluster (such as material type, historical abnormality rate, operating years, etc.), understand the risk meaning of the cluster, and obtain the risk characteristics of the cluster (such as historical abnormality rate, operating status, etc.); According to the risk characteristics of the cluster, a risk label (such as high risk, regular, low risk) is assigned to each cluster. The risk label is integrated with the material type to give the detection classification of the corresponding power material to be tested.

[0034] In one embodiment, the construction of the detection demand model specifically includes: Obtain historical inspection reports and inspection requirement rule sets for multiple types of power materials; The power material semantic tensor extraction model is used to parse historical inspection reports of various types of power materials, extract target information, and build a standardized information database; Combine the standardized information database and the detection requirement rule set to build a detection demand model.

[0035] The detection requirement rule set includes: List of mandatory inspection items for various types of power materials, such as the mandatory inspection items for 10kV cross-linked polyethylene cables include insulation resistance test and voltage withstand test; The test index thresholds for each mandatory inspection item include qualified thresholds and warning thresholds. For example, the qualified threshold for cable insulation resistance testing is ≥1000MΩ, and the warning threshold is ≥800MΩ. The inspection cycles for different types of power materials in different scenarios, including pre-commissioning inspection, daily operation inspection, and post-fault repair inspection, ensure timely inspection and avoid "over-inspection" or "missed inspection"; Association rules between test items: when the test result of a certain test item meets specific conditions, other test items need to be linked together. For example, if the "insulation resistance test result of a cable is less than the warning threshold (800MΩ)", the "partial discharge test" will be triggered to further identify insulation defects. Special inspection requirements for special types of electrical materials, such as adding "temperature rise monitoring" to high-load (voltage ≥ 90% rated voltage) transformers to prevent overheating failures.

[0036] The detection requirement rule set provides rule constraints for the detection demand model, provides constraint conditions for detection project optimization, and provides a selection basis for detection method matching.

[0037] In this example, the power material semantic tensor extraction model can be based on a neural network model such as BERT, integrated with the power knowledge graph through pre-training. Figure 5 As shown in the figure, the power material semantic tensor extraction model consists of a word embedding layer, a semantic enhancement layer, a contextual attention layer, a professional feature normalization layer, and a semantic fusion layer. For historical inspection reports, preprocessing is performed, followed by word segmentation to generate a word embedding matrix. Semantics are then enhanced using a knowledge graph, contextual attention weights are obtained, features are extracted and normalized, and finally, this information is fused to produce a semantic vector for each report. Target information, such as material type, inspection items, and inspection results, is then extracted from the semantic vector.

[0038] The parsing process of the power material semantic tensor extraction model for historical inspection reports includes: The word embedding layer inputs the word sequence after word segmentation of the historical detection report, uses a pre-trained word embedding model (such as Word2Vec, BERT) to convert the word sequence into a low-dimensional vector, and generates a word embedding matrix. This converts the unstructured text words into computable semantic vectors. The semantic enhancement layer inputs the word embedding matrix, queries the triples associated with the vocabulary through the pre-built power knowledge graph, calculates the semantic enhancement tensor of the vocabulary, and enhances the semantics of the vocabulary in the power field; The contextual attention layer inputs the current sentence topic and the global detection context, uses the attention mechanism to calculate the sentence weight, highlights the key sentences related to the detection in the historical detection report, and filters out irrelevant information (such as the report title, the compiler, etc.); The professional feature normalization layer inputs power-related professional features from sentences, extracts feature values ​​and units using regular expressions or a rule engine, and then normalizes them into a standardized format. This converts unstructured detection results (such as natural language descriptions) into computable structured data, laying the foundation for the subsequent construction of a standardized information database. The semantic fusion layer fuses the outputs of the above layers into a report-level semantic extraction vector, which is the global semantic representation of the historical inspection report and contains all key information such as "material type", "inspection items", and "inspection results".

[0039] Specifically, the semantic extraction vector of electric power materials in this example is specifically expressed as: ; Among them, S i is the report-level semantic extraction vector of the i-th detection report, n is the total number of sentences in the i-th report, E ij is the word embedding matrix of the jth sentence in the i-th report, KG(W ij ) is the electric power knowledge graph for vocabulary W ij The semantic enhancement tensor of Attn(T ij , C j ) is the attention weight based on the context, T ij is the current sentence topic, C j is the global detection context, Norm(F ij ) is the normalized function of the electric power professional characteristics, F ij is the professional feature vector of the j-th sentence.

[0040] Because the report-level semantic extraction vector integrates all key information, the target information can be subsequently extracted through a classification model or entity extraction model, such as: For material types, use a text classification model (such as CNN, BERT, etc.) to classify the report-level semantic extraction vector and output the material type; For detection items, a named entity recognition (NER) model is used to extract from the report-level semantic extraction vector; For detection results, a rule engine or sequence labeling model is used to extract from the report-level semantic extraction vector.

[0041] Gather the above target information and build a standardized information database.

[0042] This example uses the layered fusion architecture of the power material semantic tensor extraction model to achieve in-depth analysis of historical inspection reports and target information extraction, and ultimately builds a standardized information database, providing reliable data support for the construction of the inspection demand model.

[0043] Furthermore, if Figure 6 As shown in the figure, a testing requirement model is constructed by combining the standardized information library and the testing requirement rule set, which specifically includes: Obtain material type, test items, and test result data from the standardized information database, convert the test requirement rule set into a rule field, form a structured data set, and divide it into a training set and a validation set; Obtain the initial detection demand model, and give the rule field gravity, historical data potential, and rule deviation penalty based on the training set; Combining the rule field gravity, historical data potential, and rule deviation penalties, we can obtain the required field strength for each test item. According to the required detection field strength, a set of candidate detection items is given; The parameters of the initial detection requirement model are optimized through training, and verified and adjusted using the validation set to obtain the detection requirement model.

[0044] In specific implementation, the process of obtaining a structured data set includes: Convert the unstructured rules of the detection requirement rule set into a quantifiable rule field matrix, such as using binary labels for mandatory inspection items (1 = mandatory, 0 = not mandatory), numerical representations of detection thresholds and detection cycles, and conditional weights for association rules to represent the probability of associated items. The material data (material type, test items, and test result data) in the standardized information database are associated with the rule field matrix to form a structured data set, which is then divided into a training set and a validation set in a ratio of 7:3 or 8:2 for model training and evaluation.

[0045] The initial detection demand model can be a fusion model based on a neural network, including an input layer, a feature calculation layer, a hidden layer, and an output layer: The input layer is used to receive material status characteristics (such as material type, core performance parameters), historical detection data (such as historical anomaly rate, fault records), rule matching data (such as the deviation between the current detection result and the standard rule), etc. The feature calculation layer is used to obtain the rule field gravity, historical data potential energy and rule deviation penalty; The hidden layer is used to integrate the rule field gravity, historical data potential, and rule deviation penalty to obtain the detection required field strength; The output layer generates a set of candidate detection items based on the detection requirement field strength.

[0046] The rule field gravity indicates the degree of fit between the current status of an electrical material (e.g., "oil-immersed transformer") and the rule field (e.g., "mandatory oil quality inspection"). The higher the degree of fit, the more the material complies with the rule requirements and the clearer the inspection requirements. Specifically, it is expressed as: ; Where, is the regular field gravity of the k-th type of material, S ck is the material status characteristic, Rule field mapping for detection requirement sets; is the matching function used to calculate S ck and Similarity of W g is the graviton matrix, which represents the association weights between rules.

[0047] It is understandable that the material status characteristics come from the standardized information database, such as "oil-immersed transformer", "rated capacity 500kVA", etc.; the rule field mapping is the quantified result of the detection requirement rule set; the matching function can be in the form of Boolean matching, calculation of weighted similarity, etc., or it can be the matching probability output by the model through training; the graviton matrix is ​​the correlation strength matrix between the rules in the rule field.

[0048] The historical data potential energy indicates the degree of risk accumulation of historical abnormal data of power materials. The more historical anomalies, the higher the risk potential energy and the more urgent the detection need. Specifically expressed as: ; Where, is the potential energy of historical data, I c For the abnormal characteristics of a certain material, The historical abnormal potential field of the same material; Density(I c ) is the feature density, which indicates the frequency of occurrence of a certain abnormal feature; is the manifold distance, which indicates the similarity between the current material characteristics and the historical abnormal potential energy field. The smaller the distance, the higher the risk.

[0049] It is understandable that abnormal characteristics come from the standardized information database, such as "historical abnormality rate 5%", "number of failures 3 times", etc.; the historical abnormal potential energy field of similar materials comes from the historical abnormal data of the standardized information database, such as "abnormality rate distribution of oil-immersed transformers in the past 5 years".

[0050] The rule deviation penalty indicates the degree of deviation between the current test results of the power material and the rule threshold. The greater the deviation, the higher the penalty and the more urgent the test needs. Specifically expressed as: ; Where, is the rule deviation penalty, which means that the penalty is triggered when the rule deviates from the boundary. ck is the matching degree of the current rule, is the standard rule boundary; is the tolerance range, which indicates the deviation allowed by the rules; is the rule penalty coefficient, which is determined by the importance of the rule.

[0051] It is understandable that the current rule matching degree comes from the current test results of the standardized information library, such as "oil quality test result 0.05mg / L"; the standard rule boundary comes from the threshold of the test requirement rule set, such as "qualified threshold 0.1mg / L".

[0052] The inspection demand field strength is the weighted sum of the rule field attraction, historical data potential, and rule deviation penalty, indicating the urgency of the inspection demand for a certain inspection item for power supplies. The higher the field strength, the more urgent the inspection demand. Specifically, it is expressed as: ; Among them, D k is the required field strength for the detection of the kth type of materials, It is the weight coefficient, which can be set flexibly.

[0053] Generate a set of candidate detection items based on the required detection field strength, including: Obtain a list of mandatory inspection items for this type of material from the inspection requirement rule set (e.g., "Required inspection items for oil-immersed transformers: oil quality testing, winding insulation testing") and include it in the candidate inspection item set; According to the set detection requirement field strength threshold, the detection items whose detection requirement field strength exceeds the threshold are screened out and included in the candidate detection item set; According to the association rules in the test requirement rule set (such as "when the test results of a certain test item meet specific conditions, other items need to be linked"), when the test requirement field strength of a test item is too high, a linked test item is added and included in the candidate test item set; The duplicate detection items in the candidate detection item set are removed to obtain the final candidate detection item set.

[0054] It is understandable that, according to the set detection requirement field strength threshold, the detection requirement field strength can be divided into "high field strength", "medium field strength" and "low field strength", and the threshold can be an empirical value or an experimental value.

[0055] Using training data, model parameters are optimized to ensure that the model's predicted inspection requirement intensity matches actual inspection requirements (e.g., historical inspection items for this type of material). Model performance is then evaluated using validation data, comparing the model's predicted inspection requirements with actual inspection requirements. Metrics such as precision, recall, and F1-score are calculated. Model parameters are adjusted based on these feedback to improve model accuracy and generalization. After training and validation, an optimized inspection requirement model is obtained. The model inputs are material data from a standardized information database (e.g., "oil-immersed transformer, oil quality test result 0.05 mg / L") and the rule fields of the inspection requirement rule set (e.g., "mandatory inspection flag 1, threshold 0.1 mg / L"). The output is the inspection requirement intensity and a set of candidate inspection items.

[0056] The demand detection model constructed in this example ensures that no mandatory inspection items are missed through the gravity of the rule field, balances the inspection frequency through the potential energy of historical data and the penalty for rule deviation, and dynamically adjusts candidate items through association rules and material status. This achieves accurate prediction of inspection needs. The resulting set of candidate inspection items forms the basis for subsequent inspection item optimization.

[0057] In one embodiment, the construction of the detection method library includes: A detection method set, which includes multiple detection methods and detection items corresponding to each detection method; Detection attribute data, including resource consumption parameters and performance indicator data. Resource consumption parameters include detection equipment occupancy, manpower requirements, and time cost. Performance indicator data includes detection accuracy, applicable scenarios, and historical anomaly correlation. Dynamic detection adaptation rules, which include detection priority strategy and detection substitution strategy.

[0058] In this example, the detection method library associates at least one detection method with each detection item to adapt to the needs of different scenarios and avoid the limitations of a single method. The quantification of resource consumption parameters and performance indicators makes resource requirements calculable and resource conflicts early warning, thereby improving detection accuracy and pertinence. The setting of dynamic detection adaptation rules can cope with resource fluctuations and emergencies, enhancing detection flexibility.

[0059] In one embodiment, if Figure 7 As shown in Figure 2, the construction of the multi-objective optimization function specifically includes: Based on the standardized information base and the set of testing requirement rules, a quality risk set is given, and the coverage of the candidate testing item set on the quality risk is determined; Combined with the detection resource upper limit, the adaptability of the candidate detection project set and the available resources is obtained; Combined with the time cost upper limit, the time efficiency of the candidate detection item set is obtained; Combining coverage, adaptability, and time efficiency, a multi-objective optimization function is constructed, and constraints are set. The constraints include mandatory inspection item constraints, resource upper limit constraints, and time upper limit constraints.

[0060] It can be understood that the historical abnormal data of the historical inspection reports in the standardized information database, the mandatory inspection items in the inspection requirement rule set and the risk association rules are extracted to form a quality risk set.

[0061] Coverage is the coverage ratio of the candidate test item set to the quality risk set, which is specifically expressed as: ; Where, Coverage(X r , R r ) is the candidate detection item set X r Quality risk R r The coverage of; n is the total number of risks; is the weight of the k-th quality risk, which is determined by the severity of the risk and can be either an empirical value or an experimental value; Represents the indicator function. If the detection item Coverage risk k ,but ,otherwise ; represents the sum of weighted risks of the candidate detection item set; Represents the total weighted risk of the quality risk set.

[0062] The adaptability is used to quantify the degree of match between the resource consumption (equipment, manpower) of the candidate detection project set and the available resources. The higher the adaptability, the higher the resource utilization. Specifically expressed as: ; In the formula, Resource(X r , R) is the candidate detection item set X r Compatibility with available resources R; is the total number of test items in the candidate test item set; Indicates the test items Resource consumption; represents the total resource consumption of the candidate detection item set; R total The upper limit of available resources.

[0063] Time efficiency is used to quantify the degree of matching between the time consumption of the candidate detection item set and the time limit. The higher the efficiency, the less time consumption. Specifically expressed as: ; Where, Time(X r,T) is the candidate detection item set X r Time efficiency; is the total number of test items in the candidate test item set; This is the first test item Time cost; represents the total time cost of the candidate detection item set; T max The time limit.

[0064] The coverage, adaptability, and time efficiency are weighted and combined to form an objective function for optimization. The objective weights are adjusted to reflect the importance of different factors, and constraints are set. Specifically, it is expressed as: ; ; Where F represents the multi-objective function, is the target weight, and , , B r It is a set of mandatory inspection items for Class R materials.

[0065] In this example, a multi-objective optimization function is constructed and constraints are set. "Coverage" is used to ensure that the inspection items cover the main quality risks. "Adaptability" and "time efficiency" are used to ensure that the inspection items do not exceed the resource and time limits. The importance of the objectives is adjusted through weights, taking into account both the comprehensiveness of inspection and the resource and time efficiency.

[0066] After constructing a multi-objective optimization function and setting constraints, we use genetic algorithms to iteratively optimize and obtain the optimal set of detection items. The specific process includes: (1) A binary string is used to represent the set of candidate test items. Each binary bit corresponds to a test item. 1 indicates that the item is selected, and 0 indicates that it is not selected. The binary bit corresponding to the mandatory test item is fixed to 1 to ensure that the mandatory test item is not missed. The binary bit corresponding to the non-mandatory test item is randomly set to 0 or 1 to allow flexible selection.

[0067] (2) Randomly generate a certain number of binary strings that conform to the encoding rules of step (1) to represent the candidate item set.

[0068] (3) For each candidate project set, calculate the multi-objective optimization function value and eliminate the candidate project sets that violate the constraints.

[0069] (4) Using a competition selection method or other selection methods, select a certain number of candidate project sets with high multi-objective optimization function values.

[0070] (5) Cross the candidate item sets in step (4), that is, randomly select several different candidate item sets, swap the same positions of the binary strings corresponding to different candidate item sets, generate new binary strings, that is, new candidate item sets, and eliminate the candidate item sets that violate the constraints.

[0071] (6) Mutate the candidate item set obtained in step (5), that is, randomly select several candidate item sets, randomly select the binary bit corresponding to a non-mandatory item, and flip (1→0, 0→1) the binary bit. Eliminate the candidate item sets that violate the constraints.

[0072] (7) Repeat steps (3) to (6) until the maximum number of iterations is reached or the maximum value of the multi-objective optimization function changes less than a certain threshold (convergence).

[0073] (8) Output the candidate item set with the largest multi-objective optimization function value as the optimal detection item set.

[0074] In this example, through iterative optimization using a genetic algorithm, we obtained an optimal set of testing items that covers key quality risks, achieves high resource adaptability, and achieves high time efficiency. The entire process balances comprehensive testing, resource adaptability, and time efficiency, addressing the challenges of traditional testing item selection—high subjectivity, waste of resources, and insufficient coverage—and ensures accurate, efficient, and executable testing.

[0075] In one embodiment, a dynamic adaptive field matching algorithm is used to match a corresponding detection method for each detection item in the optimal detection item set, specifically including: For each detection item in the optimal detection item set, obtain the detection efficiency, matching degree with real-time resources, and task urgency of each detection method associated with the detection item; The dynamic adaptability of each detection method is obtained by combining its detection efficiency, matching degree with real-time resources, and task urgency. The dynamic adaptability of each detection method is compared, and the detection method with the largest dynamic adaptability is selected as the detection method matching the current detection item, until all detection items in the optimal detection item set are matched.

[0076] It is understood that the detection method library includes the detection methods associated with each detection item, as well as the performance indicators, resource consumption and dynamic adaptation rules of each detection method. Real-time resource data can be obtained in real time through means such as dynamic resource detection technology.

[0077] For each test item in the optimal test item set (such as "insulation resistance test" and "voltage withstand test"), all associated test methods are retrieved from the test method library, along with three key parameters for each test method: test efficiency, resource matching, and task urgency. Test efficiency refers to the performance of a test method (such as test accuracy and match to applicable scenarios). Higher values ​​indicate a more effective method. Resource matching refers to the degree to which the resource consumption (equipment, manpower, and time) of a test method matches real-time resources. Higher values ​​indicate higher resource utilization. Task urgency refers to the urgency of the test task (the less time remaining, the higher the urgency). Higher values ​​increase the weighting of the method's "time efficiency."

[0078] In this example, there is no specific limitation on the calculation method of detection efficiency, resource matching and task urgency.

[0079] The dynamic adaptability of each detection method is obtained by combining its detection efficiency, matching degree with real-time resources, and task urgency. Specifically, it is expressed as: ; Among them, D j is the dynamic fitness of the j-th detection method, M j is the jth detection method, m is the total number of detection methods associated with the detection item, Eff(M j ) is the detection efficiency of the jth detection method, Res(M j ,Z) is the matching degree between the jth detection method and the real-time resource Z, Urgent(T) is the task urgency, and T represents the remaining time of the task.

[0080] After obtaining the dynamic adaptability of all detection methods associated with the current detection item, the detection method with the largest dynamic adaptability is selected as the detection method matching the current detection item.

[0081] The following is an example of matching the detection method for each test item in the optimal test item set ("Oil Quality Test" and "Winding Insulation Resistance Test") using the dynamic adaptability field matching algorithm: For the "Oil Quality Detection" project, it is associated with the "Spectral Analysis Method" and "Chromatographic Analysis Method", and the detection efficiency, matching degree with real-time resources, and task urgency of the two detection methods are calculated respectively, and the dynamic adaptability is obtained respectively. Assuming that the dynamic adaptability is 0.85 and 0.78 respectively, the "Spectral Analysis Method" with the largest dynamic adaptability of 0.85 is selected; for the "Winding Insulation Resistance Test" project, it is associated with the "Megohmmeter Method" and "Insulation Resistance Tester Method", and the detection efficiency, matching degree with real-time resources, and task urgency of the two detection methods are calculated respectively, and the dynamic adaptability is obtained respectively. Assuming that the dynamic adaptability is 0.90 and 0.82 respectively, the "Megohmmeter Method" with the largest dynamic adaptability of 0.90 is selected.

[0082] After completing the matching of detection methods for all detection items in the optimal detection item set, the execution information (such as detection personnel grouping, equipment usage time, detection sequence, etc.) is supplemented to generate the final detection task list.

[0083] This method organically combines data, standards, and resources through core models and algorithms, avoiding the drawbacks of manually setting tasks, achieving full-process optimization of detection tasks from data to execution, and improving the accuracy and efficiency of detection.

[0084] In another embodiment, when a resource conflict is detected, a combination detection method may be generated and its dynamic adaptability may be obtained. By comparing the dynamic adaptability, a combination detection method may be selected, specifically including: (1) Genetic algorithms or other heuristic search algorithms are used to generate a limited number of effective combination detection methods based on the detection methods in the detection method library. The combination detection method is composed of multiple detection methods in the detection method library according to a specific logical relationship, and inherits their resource consumption parameters and performance indicator data.

[0085] If the ultrasonic instrument required for the "Ultrasonic Test Method" in the "Partial Discharge Detection" is occupied and unavailable for the next X hours, a combined test method of "First perform a High-Frequency Current Transformer (HFCT) test, then perform a Transient Earth Voltage (TEV) test" will be generated to replace the "Ultrasonic Test Method."

[0086] (2) Obtain the dynamic fitness of each combination detection method. The dynamic fitness is specifically expressed as: ; Among them, D j is the dynamic fitness of the jth combination detection method, m is the total number of combination detection methods, Eff(M j ) is a combined detection method M j The detection efficiency of Res(M j ,Z) is the combined detection method M jThe matching degree with real-time resource Z, Urgent(T) is the urgency of the task, T represents the remaining time of the task, P tunnel is the tunneling probability, is the combined gain.

[0087] Tunneling probability P tunnel , represents the probability of using an alternative solution (combined detection method) to replace the original solution (detection method in the detection method library) to overcome the resource barrier when there is a resource conflict. It is obtained based on the resource adaptation constant, resource equivalent quality, resource barrier and resource replacement distance. Specifically expressed as: ; Where h is the resource adaptation constant, which can be an empirical value. In this example, it is set to 0.2~0.5; is the resource equivalent quality, which is determined by equipment scarcity and manpower redundancy; V barrier is the resource barrier; D is the resource substitution distance, which indicates the difference between the substitution resource and the original resource (range 0~1). The greater the difference, the greater D.

[0088] Resource Barrier V barrier The detection accuracy deviation, resource consumption cost ratio and time window pressure of the combined detection method and the replaced detection method are obtained. The calculation formula is: ; in, is the weight coefficient; is the deviation in detection accuracy between the combined detection method and the replaced detection method, is the resource consumption cost ratio of the combined detection method to the replaced detection method, T win It is the time window pressure, which is determined by the ratio of the remaining time of the task to the required time.

[0089] Combined Gain , used to compensate for the confidence loss of the combined detection method due to substitution, is obtained based on the confidence and order penalty of each detection method that constitutes the combined detection method. Its calculation formula is: ; Where n is the number of detection methods that make up the combined detection method; Con k is the confidence of the k-th detection method; is the order penalty factor, and OP is the execution order penalty. Both can be flexibly set based on experience or experimentation. The confidence of a combined detection method is derived from the product of the confidences of each detection method and is typically lower than that of a single detection method. Sequential execution incurs an additional time penalty (the more steps, the higher the probability of error), so this is adjusted using exponential decay. The confidence of a detection method can be obtained from a detection method library. For example, the performance indicator data corresponding to the detection method (detection accuracy, applicable scenarios, historical anomaly correlation, etc.) is normalized and weighted to derive the confidence, which is used to quantify the reliability of the detection method.

[0090] (3) Compare the dynamic adaptability of each combination detection method and select the combination detection method with the largest dynamic adaptability as the combination detection method that matches the current detection item.

[0091] In this example, a combined detection method was generated to transform resource conflicts into executable alternatives, avoiding long waits for detection tasks due to insufficient detection resources and ensuring the timely completion of urgent tasks. By using tunneling probabilities to obtain alternative resources for alternative detection, and then using combined gains to compensate for the loss in confidence, this approach balances detection accuracy, reliability, and cost-effectiveness.

[0092] On the other hand, Figure 8 As shown, the present invention provides a device for generating a power material detection task, which adopts the above-mentioned power material detection task generation method, and the device includes: The detection clustering classification module is used to analyze the information of the power materials to be detected, give multiple target features, process the target features through the power material clustering classification model, and give the detection classification of the power materials; The test item acquisition module is used to obtain a set of candidate test items by combining the test classification of power materials and the pre-built test demand model; The detection project optimization module is used to provide the optimal detection project set based on the pre-established detection method library, using multi-objective optimization functions and iterative optimization using genetic algorithms; The detection method matching module is used to combine real-time detection resources and time costs, and adopt a dynamic fitness field matching algorithm to match the corresponding detection method for each detection item in the optimal detection item set and generate a detection task list.

[0093] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the invention. Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the invention. Thus, the present invention is intended to include such changes and modifications as fall within the scope of the claims and their equivalents.

Claims

1. A method for generating a power material detection task, characterized in that: The specific steps include: Based on the analysis of the information of the power materials to be tested, multiple target features are given, and the target features are processed through the power material clustering classification model to give the detection classification of power materials; Combining the inspection classification of power materials and the pre-built inspection demand model, a set of candidate inspection items is obtained; Based on the pre-established detection method library, the optimal detection item set is given by using multi-objective optimization function and iterative optimization of genetic algorithm; Combining real-time detection resources and time costs, and adopting a dynamic fitness field matching algorithm, each detection item in the optimal detection item set is matched with a corresponding detection method to generate a detection task list.

2. The method for generating electric power material detection tasks according to claim 1, characterized in that: Target features include material type, core performance parameters, and historical detection anomaly rates. These features are processed using a power material clustering classification model to provide a detection classification for power materials, specifically: Obtain multiple target features and fuse them into cluster feature vectors; Through the power material clustering classification model and the manifold clustering algorithm, the manifold clustering distance is obtained; Based on the manifold clustering distance, the power materials to be tested are clustered and classified, and the detection classification of the corresponding power materials to be tested is given.

3. The method for generating electric power material detection tasks according to claim 2, characterized in that: Based on the manifold clustering distance, the power materials to be tested are clustered and classified, and the detection classification of the corresponding power materials to be tested is given, including: Select clustering algorithm based on manifold cluster distance; Determine the number of clusters by combining domain knowledge and effectiveness indicators; Based on the selected clustering algorithm and taking the manifold clustering distance as the metric, the power materials to be tested are merged into several clusters; Count the characteristic distribution of each cluster and obtain the risk characteristics of each cluster; According to the risk characteristics of each cluster, a risk label is given, and the risk label and material type are integrated to give the detection classification of the corresponding power materials to be detected.

4. The method for generating electric power material detection tasks according to claim 1, characterized in that: The construction of the detection demand model includes: Obtain historical inspection reports and inspection requirement rule sets for multiple types of power materials; The power material semantic tensor extraction model is used to parse historical inspection reports of various types of power materials, extract target information, and build a standardized information database; Combine the standardized information database and the detection requirement rule set to build a detection demand model.

5. The method for generating electric power material detection tasks according to claim 4, characterized in that: Target information includes material type, test items, and test result data; Combining the standardized information base and the detection requirement rule set, a detection requirement model is constructed, specifically including: Obtain material type, test items, and test result data from the standardized information database, convert the test requirement rule set into a rule field, form a structured data set, and divide it into a training set and a validation set; Obtain the initial detection demand model, and give the rule field gravity, historical data potential, and rule deviation penalty based on the training set; Combining the rule field gravity, historical data potential, and rule deviation penalties, we can obtain the required field strength for each test item. According to the required detection field strength, a set of candidate detection items is given; The parameters of the initial detection requirement model are optimized through training, and verified and adjusted using the validation set to obtain the detection requirement model.

6. The method for generating electric power material detection tasks according to claim 1, characterized in that: The construction of the detection method library specifically includes: A detection method set, which includes multiple detection methods and detection items corresponding to each detection method; Detection attribute data, including resource consumption parameters and performance indicator data. Resource consumption parameters include detection equipment occupancy, manpower requirements, and time cost. Performance indicator data includes detection accuracy, applicable scenarios, and historical anomaly correlation. Dynamic detection adaptation rules, which include detection priority strategy and detection substitution strategy.

7. The method for generating electric power material detection tasks according to claim 6, characterized in that: The construction of multi-objective optimization function includes: Based on the standardized information base and the set of testing requirement rules, a quality risk set is given, and the coverage of the candidate testing item set on the quality risk is determined; Combined with the detection resource upper limit, the adaptability of the candidate detection project set and the available resources is obtained; Combined with the time cost upper limit, the time efficiency of the candidate detection item set is obtained; Combining coverage, adaptability, and time efficiency, a multi-objective optimization function is constructed, and constraints are set. The constraints include mandatory inspection item constraints, resource upper limit constraints, and time upper limit constraints.

8. The method for generating electric power material detection tasks according to claim 7, characterized in that: A dynamic adaptation field matching algorithm is used to match the corresponding detection method for each detection item in the optimal detection item set, specifically including: For each detection item in the optimal detection item set, obtain the detection efficiency, matching degree with real-time resources, and task urgency of each detection method associated with the detection item; The dynamic adaptability of each detection method is obtained by combining its detection efficiency, matching degree with real-time resources, and task urgency. The dynamic adaptability of each detection method is compared, and the detection method with the largest dynamic adaptability is selected as the detection method matching the current detection item, until all detection items in the optimal detection item set are matched.

9. The method for generating electric power material detection tasks according to claim 8, characterized in that: Also includes: Generate a limited number of effective combination detection methods based on the detection methods in the detection method library; The dynamic adaptability of each combined detection method is obtained by combining its detection efficiency, matching degree with real-time resources, task urgency, tunneling probability, and combined gain. Compare the dynamic adaptability of each combination detection method, and select the combination detection method with the largest dynamic adaptability as the combination detection method that matches the current detection item; Among them, the dynamic adaptability of the combined detection method is specifically expressed as: ； Among them, D j is the dynamic fitness of the jth combination detection method, m is the total number of combination detection methods, Eff(M j ) is a combined detection method M j The detection efficiency of Res(M j ,Z) is the combined detection method M j The matching degree with real-time resource Z, Urgent(T) is the urgency of the task, T represents the remaining time of the task, P tunnel is the tunneling probability, is the combined gain; The tunneling probability is specifically expressed as: ； Among them, h is the resource adaptation constant, is the resource equivalent quality, V barrier is the resource barrier, and D is the resource substitution distance; Resource barriers are specifically expressed as: ； in, is the weight coefficient; is the deviation in detection accuracy between the combined detection method and the replaced detection method, is the resource consumption cost ratio of the combined detection method to the replaced detection method, T win is the time window pressure; Combined Gain Specifically expressed as: ； Where n is the number of detection methods that make up the combined detection method; Con k is the confidence of the k-th detection method, is the order penalty factor, and OP is the execution order penalty.

10. A device for generating a power material detection task, using the method for generating a power material detection task according to any one of claims 1 to 9, characterized in that: include: The detection clustering classification module is used to analyze the information of the power materials to be detected, give multiple target features, process the target features through the power material clustering classification model, and give the detection classification of the power materials; The test item acquisition module is used to obtain a set of candidate test items by combining the test classification of power materials and the pre-built test demand model; The detection project optimization module is used to provide the optimal detection project set based on the pre-established detection method library, using multi-objective optimization functions and iterative optimization using genetic algorithms; The detection method matching module is used to combine real-time detection resources and time costs, and adopt a dynamic fitness field matching algorithm to match the corresponding detection method for each detection item in the optimal detection item set and generate a detection task list.

Images