Search

CN-122022586-A - Artificial intelligence-based electric power engineering electromechanical equipment decoration test quality assessment method

CN122022586ACN 122022586 ACN122022586 ACN 122022586ACN-122022586-A

Abstract

The invention discloses an artificial intelligence-based method for evaluating the quality of a decoration test of an electric power engineering electromechanical device, which relates to the technical field of electric power device evaluation, and solves the technical problems that the accuracy of data history and node matching are not considered, the influence of low-quality data is easy, the evaluation efficiency is low, an evaluation result is easy to be disjointed with the actual state of the device, and the quality evaluation result is inaccurate. By establishing a dynamic confidence analysis model, calculating confidence weights by combining data history accuracy and node matching, the weighted fusion of data features is realized, the wiring process evaluation model of high-voltage, medium-voltage and low-voltage equipment is analyzed by combining full life cycle data of the equipment, and the evaluation accuracy is improved.

Inventors

  • WANG YING
  • WANG YIBIN

Assignees

  • 宁波安兴电力建设有限公司

Dates

Publication Date
20260512
Application Date
20260202

Claims (10)

  1. 1. The method for evaluating the quality of the decoration test of the electric engineering electromechanical equipment based on the artificial intelligence is characterized by comprising the following steps: distributing distributed data nodes in a transformer substation or an overhaul area of an electric power engineering electromechanical equipment management area, classifying the electric power engineering electromechanical equipment in a management area according to voltage levels by the distributed data nodes, and collecting state quantity monitoring data, power failure overhaul data and manual analysis data of various equipment; the distributed data nodes perform preliminary detection on data according to a preset data early warning threshold value and then transmit the data to a cloud data center; the cloud data center establishes a data association matching degree formula by adopting a cosine similarity algorithm, analyzes the matching degree of data uploaded by the distributed data nodes, establishes a data matching degree map, and analyzes the matching degree of the corresponding data nodes; According to the historical accuracy of state quantity monitoring data, outage overhaul data and manual analysis data of the maintenance test of the management slice equipment, a dynamic confidence analysis model is established by combining the matching of corresponding data nodes, the extracted data features are subjected to weighted fusion, and the fusion quality score is analyzed; Establishing a multi-mode equipment life cycle prediction model by analyzing an operation environment correction factor, a load characteristic correction factor and a maintenance level correction factor; and building a decoration test quality assessment model according to life cycle prediction results of the equipment with different voltage levels, and the weighted and fused data characteristics and the fusion quality scores, and respectively analyzing the comprehensive health degree of the equipment with different voltage levels.
  2. 2. The method for evaluating the quality of a decoration test of an electric power engineering electromechanical device based on artificial intelligence according to claim 1, wherein distributed data nodes are deployed in a transformer substation or a maintenance area of an electric power engineering electromechanical device management area, and the distributed data nodes classify the electric power engineering electromechanical devices in a management area according to voltage levels, comprising the following steps: Deploying edge nodes in an equipment concentration area of a transformer substation or an overhaul area; core nodes are deployed in a control room of a transformer substation or an overhaul area and are responsible for summarizing data acquired by the edge nodes and are connected with a cloud data center in an opposite way; Classifying equipment with the working voltage of 110kV and above of the electric engineering electromechanical equipment in the management area as high-voltage equipment, classifying equipment with the working voltage of 10 kV-35 kV as medium-voltage equipment, classifying equipment with the working voltage of below 1kV as low-voltage equipment; each device is assigned a unique ID and is associated with a voltage class, device type, and location information as a device tag.
  3. 3. The artificial intelligence-based electric power engineering electromechanical equipment decoration test quality assessment method according to claim 1, wherein the construction of a data association matching degree formula by adopting a cosine similarity algorithm comprises the following steps: extracting data feature vectors of high-voltage equipment, medium-voltage equipment and low-voltage equipment, and collecting state quantity monitoring data, power failure maintenance data and manual analysis data; setting state quantity monitoring data, outage overhaul data and manual analysis data as state quantity monitoring data nodes, outage overhaul data nodes, manual analysis data nodes and environment monitoring data nodes; Cosine similarity algorithm is adopted for the data feature vectors of the power outage overhaul data node, the manual analysis data node and the environment monitoring data node, and cosine similarity among the data feature vectors of the data nodes is calculated and used as the matching degree of the data nodes; For the data of the state quantity monitoring data node, generating corresponding time sequence data according to the timestamp of the data detection, and calculating cosine similarity of the data of the state quantity monitoring data node by adopting a dynamic time warping algorithm to serve as matching degree of the corresponding data node: The DTW-CosSim (S1, S2) is cosine similarity of state quantity monitoring data node data, S1 and S2 are two time sequences, i and j are indexes for identifying element positions in the time sequences S1 and S2, P is an optimal alignment path found through a dynamic time warping algorithm, L is an alignment path length, ti is a time point corresponding to an ith element in the time sequence S1, tj is a time point corresponding to a jth element in the time sequence S2, S1 (ti) is a feature vector of the time sequence S1 at a time point ti, S2 (tj) is a feature vector of the time sequence S2 at a time point tj, and CosSim (S1 (ti), S2 (tj)) is cosine similarity of the feature vector S1 (ti) and the feature vector S2 (tj).
  4. 4. The artificial intelligence-based power engineering electromechanical equipment decoration test quality assessment method according to claim 1, wherein the steps of analyzing the matching degree of the data uploaded by the distributed data nodes, establishing a data matching degree map, and analyzing the matching degree of the corresponding data nodes are as follows: for all data nodes of the same equipment, calculating a consistency index CIn=1-sigma S/mu S, wherein mu S is the average value of the similarity among all nodes, and sigma S is the standard deviation; Carrying out data consistency early warning on equipment with a consistency index smaller than a threshold value; If the matching degree between the data nodes is larger than the matching threshold, matching edges are established between the corresponding data nodes, and the weight of the matching edges is the matching degree; wherein the matching threshold is set according to the voltage level, the matching threshold of the high-voltage equipment is set to 0.85, the matching threshold of the medium-voltage equipment is set to 0.80, and the matching threshold of the low-voltage equipment is set to 0.75; and establishing a data matching degree map, and calculating the degree centrality of each node as the matching property of the corresponding data node.
  5. 5. The artificial intelligence-based power engineering electromechanical equipment decoration test quality assessment method according to claim 1, wherein the dynamic confidence analysis model is established by combining the matching of corresponding data nodes according to the historical accuracy of state quantity monitoring data, outage maintenance data and manual analysis data of the management area equipment decoration test, and the extracted data features are subjected to weighted fusion, and the method comprises the following steps: Acquiring a mean value of historical data of the kth data node feature vector corresponding to the working condition, extracting a feature vector of the mean value of the working condition, and calculating cosine similarity between the extracted feature vector of the working condition at the moment t and the feature vector of the working condition at the moment t to obtain working condition correlation of the kth data node feature vector at the moment t; according to the historical accuracy of state quantity monitoring data, outage maintenance data and manual analysis data of the maintenance test of the management slice equipment, analyzing the dynamic confidence coefficient: Confidence k (t)=α k *A k +β k *F k (t)+γ k *C k (t)+δ k *P k ; Wherein Confidence k (t) is the dynamic confidence of the kth data node feature vector at time t; a k is the accuracy of historical data of state quantity monitoring data, power failure maintenance data and manual analysis data; F k (t) is a data freshness factor, F k (t) = T is the current time, tup is the time of the latest update of the data, Is an attenuation coefficient associated with the kth data node feature vector; C k (t) is the working condition correlation of the kth data node feature vector at the moment t; P k is the matching property corresponding to the characteristic vector of the kth data node; And according to the dynamic confidence coefficient of the feature vector of the kth data node of the same equipment at the time t, carrying out weighted fusion on the feature vector of each data node of the same equipment at the time t to obtain the weighted fused data feature.
  6. 6. The artificial intelligence based method for evaluating the quality of a decoration test of an electrical engineering machine of an electrical engineering according to claim 1, wherein analyzing the fusion quality score comprises the steps of: According to the dynamic confidence corresponding to the data source of the data characteristics, and combining the matching of the corresponding data nodes, establishing an analysis fusion quality score analysis model: qfused (t) is the fusion quality score of the feature vectors of all data nodes of the same equipment at the moment t; Quality evaluation values for the kth data node feature vector at time t; The matching of the characteristic vector of the kth data node at the moment t; is the weight coefficient of the characteristic vector of the kth data node at the time t.
  7. 7. The artificial intelligence based power engineering machinery electrical equipment decoration test quality assessment method according to claim 1, wherein the analysis of the operating environment correction factor, the load characteristic correction factor and the maintenance level correction factor comprises the following steps: the calculation formula of the operation environment correction factor is as follows: Wherein Kenv is an operation environment correction factor, βf is a weight coefficient related to the f-th environmental parameter, and is used for measuring the importance degree of different environmental parameters on the life cycle of the equipment, f epsilon (1, 2, the..Y), Y is the total number of the environmental parameters, ef is the value of the f-th environmental parameter, ef and ref are the reference values of the f-th environmental parameter; the calculation formula of the load characteristic correction factor is: ; Wherein Kload is a load characteristic correction factor, lavg is an average load of the equipment, lrated is a rated load of the equipment, gamma 1 and gamma 2 are coefficients related to the load, lfluctuation is a load fluctuation quantity of the equipment, and changes of the load in the running process of the equipment are reflected; The calculation formula of the maintenance level correction factor is as follows: Kma is maintenance level correction factor, which considers the influence of equipment maintenance level on the life cycle of the equipment and is used for correcting the life cycle of the equipment, maintenanceScore (t) is maintenance score when the equipment is maintained at the latest time before the moment t, and the maintenance score is given by maintenance personnel when the equipment is maintained; μ is a time-dependent coefficient, μ=the time difference between the latest maintenance time and the time T, T is the maintenance time period set by the device, scoremax is the maximum maintenance score in the device history data.
  8. 8. The artificial intelligence based method for evaluating the quality of a decoration test of an electrical engineering machine and electrical equipment according to claim 7, wherein the method for establishing a life cycle prediction model of the multi-mode equipment comprises the following steps: Carrying out normalization processing on the operation environment correction factors Kenv and Kload serving as load characteristic correction factors Kload and maintaining a horizontal correction factor Kma, and carrying out weighted summation; the formula for establishing the life cycle prediction model of the multi-mode equipment is as follows: ; wherein LIFECYCLEPREDICTED is a predicted multi-mode device life cycle, LIFECYCLEDESIGN is a device design life cycle, and is a service life specified when the device leaves the factory, kav is a normalized running environment correction factor Kenv, kload is a load characteristic correction factor Kload, and a result of weighted summation of a maintenance level correction factor Kma is maintained.
  9. 9. The artificial intelligence-based method for evaluating the quality of a decoration test of an electric engineering electromechanical device according to claim 1, wherein the method for evaluating the quality of the decoration test is characterized by establishing a model for evaluating the quality of the decoration test according to life cycle prediction results of devices with different voltage levels and weighted fusion data characteristics, and respectively analyzing the comprehensive health degree of the devices with different voltage levels, and comprises the following steps: Calculating life cycle prediction results of equipment with different voltage levels, and weighting and fusing data characteristics and fusing quality scores through state quantity monitoring data, outage maintenance data and manual analysis data of the equipment with different voltage levels in the historical data; Calculating the ratio of the actual times of faults in the actual residual service life of the equipment to the times of faults in the whole service life of the equipment to obtain an operation health coefficient, and carrying out weighted summation on the operation cycle coefficient and the operation health coefficient to obtain the comprehensive health of the equipment; Respectively training GRU-LSTM neural network models through life cycle prediction results of equipment with different voltage levels, weighted fusion data characteristics and fusion quality scores, and marking the output comprehensive health degree to obtain a support repair quality assessment model for the equipment with different voltage levels, wherein the health degree score of a key part of the equipment is increased for the output data of the high-voltage equipment; and analyzing the comprehensive health degree of the current detection equipment through the acceptance and maintenance test quality evaluation models aiming at the equipment with different voltage levels.
  10. 10. The artificial intelligence based power engineering electromechanical equipment decoration test quality assessment method according to claim 9, wherein adding a equipment key component health score to high voltage equipment output data comprises: Setting an additional score of the health degree of the key component for the high-voltage equipment, and calculating the health degree score of the key component of the equipment: Wherein H is the health degree score of key parts of the equipment, rn is the number of key parts of the high-voltage equipment, The loss factor for the p-th critical component, For the accumulated run time of the p-th critical component, The service life of the design for the p-th key component.

Description

Artificial intelligence-based electric power engineering electromechanical equipment decoration test quality assessment method Technical Field The invention belongs to the field of power equipment evaluation, and particularly relates to an artificial intelligence-based power engineering mechanical and electrical equipment decoration test quality evaluation method. Background The electric power industry is the basic industry of national economy, and the safe and stable operation of an electric power system is directly related to the normal order of social production and life. The electric engineering electromechanical equipment is used as a core component of the electric power system, and the quality level of the electric power engineering electromechanical equipment is a key for guaranteeing the reliable operation of the electric power system. With the rapid development of the electric power industry in China, the power grid scale is continuously enlarged, the voltage level is continuously improved, the types of the electric power engineering electromechanical equipment are increasingly abundant, the structure is increasingly complex, and the requirements on the operation and maintenance management of the electric power engineering electromechanical equipment are also continuously improved. The decoration test is used as a core link in the whole life cycle management of the electric engineering electromechanical equipment, and the quality of the decoration test directly determines the operation reliability, service life and safety performance of the equipment. Meanwhile, in recent years, the rapid development of artificial intelligence technology provides a brand new technical path for the quality evaluation of the decoration test of the electromechanical equipment of the power engineering. The artificial intelligence technology has strong data processing, feature extraction, pattern recognition and predictive analysis capabilities, and can effectively break pain points of the traditional evaluation method. Based on the method, an artificial intelligence-based electric engineering mechanical and electrical equipment decoration test quality evaluation method is generated. The traditional method for evaluating the quality of the decoration test of the electromechanical equipment of the electric power engineering directly adopts original data to analyze, is not considered in terms of data history accuracy and node matching, is easily influenced by low-quality data, is characterized in that in the traditional evaluation method, data such as state quantity detection, power failure overhaul and manual analysis are stored in a polydisperse mode, and are required to be manually arranged and associated, so that the efficiency is low, and the traditional evaluation method is less considered in terms of the influence of multidimensional factors such as an operation environment, load characteristics and maintenance level on the life cycle of the equipment, and the evaluation result is disjointed from the actual state of the equipment. The process standard, fault mode and quality influence factor of the equipment with different voltage grades are greatly different, the traditional evaluation method mostly adopts unified standard, and the pertinence is insufficient, so that the quality evaluation result is inaccurate. Disclosure of Invention The invention aims to at least solve one of the technical problems in the prior art, and therefore, the invention provides an artificial intelligence-based electric engineering mechanical and electrical equipment decoration test quality assessment method, which is used for solving the technical problems that the historical accuracy and node matching of data are not considered, the influence of low quality data is easy, the assessment efficiency is low, the assessment result is easy to be disjointed with the actual state of equipment, and the quality assessment result is inaccurate. In order to solve the above problems, a first aspect of the present invention provides an artificial intelligence-based quality evaluation method for a decoration test of an electrical engineering electromechanical device, comprising the steps of: distributing distributed data nodes in a transformer substation or an overhaul area of an electric power engineering electromechanical equipment management area, classifying the electric power engineering electromechanical equipment in a management area according to voltage levels by the distributed data nodes, and collecting state quantity monitoring data, power failure overhaul data and manual analysis data of various equipment; the distributed data nodes perform preliminary detection on data according to a preset data early warning threshold value and then transmit the data to a cloud data center; the cloud data center establishes a data association matching degree formula by adopting a cosine similarity algorithm, analyzes the matching degree of data uploaded