CN-122021284-A - Intelligent simulation diagnosis method and system for power faults

CN122021284ACN 122021284 ACN122021284 ACN 122021284ACN-122021284-A

Abstract

The invention relates to the technical field of power system simulation and discloses an intelligent simulation diagnosis method and system for power faults, wherein the method comprises the steps of collecting an operation data sequence, adjusting parameters of a pre-established digital twin model of a power system through a self-adaptive parameter estimation algorithm, and generating a dynamic simulation model; sequentially injecting a plurality of preset fault types, running simulation to generate full-time sequence simulation data sequences corresponding to each preset fault type, extracting a plurality of time sequence feature vectors, respectively calculating the matching degree between the real-time running data sequences and the full-time sequence simulation data sequences of each preset fault type on the plurality of time sequence feature vectors to obtain a plurality of matching degree indexes, fusing the plurality of matching degree indexes according to preset weights to generate comprehensive similarity evaluation values, inputting the comprehensive similarity evaluation values into a fault classification model, and outputting diagnosis results. The invention can improve the adaptability and the accuracy of the power failure diagnosis.

Inventors

YANG HAI
LI PEISHUANG
HUANG FUQIANG
HOU XIAOHU
XIAO JIANGTAO
TIAN PENGYUN
WANG DAN
ZHOU GUOJUN
Liao Zhengyang
ZHANG DIYA

Assignees

三峡金沙江川云水电开发有限公司

Dates

Publication Date: 20260512
Application Date: 20260123

Claims (10)

1. An intelligent simulation diagnosis method for power faults is characterized by comprising the following steps: Collecting a real-time operation data sequence of the power system in a preset time window, and adjusting parameters of a digital twin model of the pre-established power system by combining the real-time operation data sequence and a self-adaptive parameter estimation algorithm to generate a calibrated dynamic simulation model; sequentially injecting a plurality of preset fault types into the dynamic simulation model, running simulation to generate a full-time sequence simulation data sequence corresponding to each preset fault type, and extracting a plurality of time sequence feature vectors from the full-time sequence simulation data sequence; Respectively calculating the matching degree of the real-time operation data sequence and the full-time simulation data sequence of each preset fault type on a plurality of time sequence feature vectors to obtain a plurality of matching degree indexes; and fusing the multiple matching degree indexes according to preset weights to generate a comprehensive similarity evaluation value, inputting the comprehensive similarity evaluation value into a pre-trained fault classification model, and outputting an actual fault type diagnosis result of the power system.
2. The intelligent simulation and diagnosis method for power failure according to claim 1, wherein the collecting the real-time operation data sequence of the power system within the preset time window comprises: synchronously collecting real-time operation data sequences from a plurality of monitoring points in a preset time window to form an original operation data set, wherein the real-time operation data sequences comprise a voltage data sequence, a current data sequence and a frequency data sequence; Performing real-time quality evaluation and screening on an original operation data set, wherein the quality evaluation comprises data integrity verification, validity verification and consistency verification; and filling and correcting abnormal data points which do not pass the verification based on the historical contemporaneous data of the same monitoring point or the real-time data of adjacent monitoring points to generate a real-time operation data sequence.
3. The intelligent simulation and diagnosis method of power failure according to claim 1, wherein the step of combining the real-time operation data sequence and the adaptive parameter estimation algorithm to adjust parameters of a pre-established digital twin model of the power system to generate a calibrated dynamic simulation model comprises the steps of: Inputting a real-time operation data sequence into a pre-established digital twin model of the electric power system, and dynamically identifying and adjusting equivalent parameters of key elements in the digital twin model of the electric power system based on an online recursive parameter identification algorithm, wherein the key elements comprise a generator, a circuit and a load; Operating a digital twin model of the power system by using the adjusted parameters to generate a corresponding simulation output sequence; Calculating normalized root mean square error of the simulation output sequence and the real-time operation data sequence in the same time period, and if the error is larger than a preset convergence threshold value, repeatedly executing the two steps to perform iterative calibration; and when the error is smaller than or equal to the convergence threshold, locking the model under the current parameters into a calibrated dynamic simulation model.
4. The intelligent simulation diagnosis method of power failure according to claim 3, wherein the dynamically identifying and adjusting equivalent parameters of key elements in a digital twin model of a power system based on an online recursive parameter identification algorithm comprises: adopting a recursive least square method with forgetting factors as an online recursive parameter identification algorithm; Intercepting a real-time operation data sequence by a sliding time window with a preset length, and constructing a parameter identification input and output data pair of a key element; Aiming at each key element to be identified, calculating the parameter variability of the key element according to the deviation of the historical operation data and the simulation data; according to the parameter variation degree, dynamically determining a key element subset which needs to be subjected to parameter updating in the current time window; And performing recursive parameter identification and adjustment on elements in the key element subset only, and updating equivalent parameters in the digital twin model of the power system.
5. The intelligent simulation and diagnosis method of power failure according to claim 1, wherein the step of sequentially injecting a plurality of preset failure types into the dynamic simulation model and running simulation to generate a full-time-series simulation data sequence corresponding to each preset failure type comprises the steps of: Configuring a plurality of preset fault types to be injected on the calibrated dynamic simulation model, and setting fault parameters, injection time and duration of each fault type; Initializing the simulation environment of the calibrated dynamic simulation model according to the running state of the real-time running data sequence at the tail end of the preset time window; sequentially executing injection and simulation of each preset fault type according to a set simulation step length, and synchronously recording simulation data from a steady state before a fault, fault occurrence, fault duration to a fault clearing and recovering whole process; and integrating and time-aligning the whole process simulation data corresponding to each preset fault type to generate a whole-time sequence simulation data sequence.
6. The intelligent power failure simulation diagnostic method of claim 1, wherein the extracting a plurality of timing feature vectors from the full-timing simulation data sequence comprises: Sliding window segmentation is carried out on the full-time sequence simulation data sequence to obtain a plurality of simulation data subsequences; for each simulation data subsequence, respectively calculating a plurality of initial characteristic indexes on a time domain, a frequency domain and a time-frequency domain; Arranging the same initial characteristic indexes of all simulation data subsequences under the same fault type in time sequence to form a time sequence change curve of the characteristic indexes; selecting a plurality of curves with highest association degree with a typical fault mode from all time sequence change curves, and taking corresponding characteristic indexes as basic characteristics; Based on the base characteristics, performing dimension reduction and reconstruction by a principal component analysis method, and generating a plurality of time sequence characteristic vectors corresponding to each preset fault type.
7. The intelligent simulation diagnosis method of power failure according to claim 1, wherein the calculating the matching degree between the real-time operation data sequence and the full-time simulation data sequence of each preset failure type on a plurality of time sequence feature vectors to obtain a plurality of matching degree indexes includes: Respectively extracting a plurality of time sequence feature vectors corresponding to the real-time operation data sequence and the full-time sequence simulation data sequence aiming at each preset fault type; For each time sequence feature vector, calculating the minimum regular distance between the feature vector time sequence of the real-time operation data sequence and the corresponding feature vector time sequence of the full-time simulation data sequence by adopting a dynamic time alignment algorithm; Mapping the minimum regular distance into a value between 0 and 1 through a preset normalization function, and taking the value as a single matching degree under the time sequence feature vector; and summarizing the single matching degree of each preset fault type on all time sequence feature vectors to form a plurality of matching degree indexes of the fault type.
8. The intelligent simulation and diagnosis method of power failure according to claim 1, wherein the fusing the plurality of matching degree indexes according to a preset weight to generate the comprehensive similarity evaluation value comprises: Obtaining a plurality of matching degree indexes corresponding to each preset fault type; Constructing a matching degree index matrix, wherein rows correspond to different preset fault types, columns correspond to different time sequence feature vectors, and matrix elements are corresponding matching degree indexes; aiming at each time sequence feature vector, calculating the distinguishing degree and the dispersion degree of the corresponding matching degree index among all preset fault types; Based on an entropy weight method, dynamically calculating the fusion weight of each time sequence feature vector according to the distinction degree and the dispersion degree, and taking the fusion weight as a preset weight; And carrying out weighted summation on the matching degree indexes of each row in the matching degree index matrix by utilizing the calculated fusion weight, and generating a comprehensive similarity evaluation value corresponding to each preset fault type.
9. The intelligent simulation diagnosis method of power failure according to claim 1, wherein the inputting the comprehensive similarity evaluation value into the pre-trained failure classification model outputs an actual failure type diagnosis result of the power system, comprises: Constructing a pre-trained fault classification model based on a deep neural network, wherein the number of nodes of an input layer is the same as the dimension of the comprehensive similarity evaluation value, and the number of nodes of an output layer is the same as the number of preset fault types; generating a comprehensive similarity evaluation value sample set of different fault types and corresponding real fault type labels by using simulation data under a historical multi-fault scene, and training and cross-verifying a fault classification model; inputting the comprehensive similarity evaluation value into a pre-trained fault classification model, and outputting probability distribution corresponding to each preset fault type through a softmax layer of the model; Selecting the fault type with the highest probability value as a candidate diagnosis result, checking whether the probability value exceeds a preset credibility threshold value, outputting the candidate diagnosis result as an actual fault type diagnosis result of the power system if the probability value exceeds the preset credibility threshold value, and outputting an indication signal of unknown or compound faults if the probability value does not exceed the preset credibility threshold value.
10. An intelligent simulation and diagnosis system for power faults is characterized by comprising: The dynamic simulation model construction module is configured to acquire a real-time operation data sequence of the power system in a preset time window, and adjust parameters of a pre-established digital twin model of the power system by combining the real-time operation data sequence and a self-adaptive parameter estimation algorithm to generate a calibrated dynamic simulation model; the time sequence feature vector extraction module is configured to sequentially inject a plurality of preset fault types into the dynamic simulation model, run simulation to generate a full-time sequence simulation data sequence corresponding to each preset fault type, and extract a plurality of time sequence feature vectors from the full-time sequence simulation data sequence; The matching degree index calculation module is configured to calculate the matching degree between the real-time operation data sequence and the full-time simulation data sequence of each preset fault type on a plurality of time sequence feature vectors respectively to obtain a plurality of matching degree indexes; the fault type diagnosis result output module is configured to fuse a plurality of matching degree indexes according to preset weights, generate a comprehensive similarity evaluation value, input the comprehensive similarity evaluation value into a pre-trained fault classification model and output an actual fault type diagnosis result of the power system.

Description

Intelligent simulation diagnosis method and system for power faults Technical Field The invention relates to the technical field of power system simulation, in particular to an intelligent simulation diagnosis method and system for power faults. Background The operation stability of the electric power system as a core infrastructure for energy supply is directly related to industrial production continuity and social civil security. With the increase of new energy grid connection proportion and the trend of complex power grid topological structure, the occurrence probability of various faults such as short circuit faults, equipment abnormality and the like is obviously increased, if the fault type is not accurately diagnosed in time, the cascade reaction is extremely easy to be caused, large-area power failure accidents are caused, and huge economic losses are caused. Therefore, the rapid and accurate diagnosis of the power faults is realized, and the method is a key technical requirement for guaranteeing the safe and reliable operation of the power system. The traditional power fault diagnosis method mainly relies on manual experience and local monitoring data to judge, and has the problems of low diagnosis efficiency, strong subjectivity, high misjudgment rate and the like, and is difficult to adapt to the dynamic operation scene of a complex power grid. In recent years, a diagnosis technology combining data driving and simulation modeling is gradually developed, but the diagnosis technology still has a plurality of defects that a simulation model cannot be matched with the running state change of a power system in real time due to factors such as load fluctuation, equipment aging and the like, so that deviation exists between simulation data and actual running data, the diagnosis accuracy is affected, the fault characteristics lack of mining on the time sequence relevance of the whole period of a fault, and the fault evolution rule is difficult to comprehensively describe. Therefore, there is a need to design a solution that overcomes the above drawbacks. Disclosure of Invention In order to solve the problems of poor adaptability, insufficient precision and the like in the traditional power failure diagnosis, the invention provides an intelligent simulation diagnosis method and system for power failure, which can improve the adaptability and the precision of the power failure diagnosis. The technical scheme adopted by the invention is as follows: an intelligent simulation diagnosis method for power faults comprises the following steps: Collecting a real-time operation data sequence of the power system in a preset time window, and adjusting parameters of a digital twin model of the pre-established power system by combining the real-time operation data sequence and a self-adaptive parameter estimation algorithm to generate a calibrated dynamic simulation model; sequentially injecting a plurality of preset fault types into the dynamic simulation model, running simulation to generate a full-time sequence simulation data sequence corresponding to each preset fault type, and extracting a plurality of time sequence feature vectors from the full-time sequence simulation data sequence; Respectively calculating the matching degree of the real-time operation data sequence and the full-time simulation data sequence of each preset fault type on a plurality of time sequence feature vectors to obtain a plurality of matching degree indexes; and fusing the multiple matching degree indexes according to preset weights to generate a comprehensive similarity evaluation value, inputting the comprehensive similarity evaluation value into a pre-trained fault classification model, and outputting an actual fault type diagnosis result of the power system. Further, the collecting the real-time operation data sequence of the power system in the preset time window includes: synchronously collecting real-time operation data sequences from a plurality of monitoring points in a preset time window to form an original operation data set, wherein the real-time operation data sequences comprise a voltage data sequence, a current data sequence and a frequency data sequence; Performing real-time quality evaluation and screening on an original operation data set, wherein the quality evaluation comprises data integrity verification, validity verification and consistency verification; and filling and correcting abnormal data points which do not pass the verification based on the historical contemporaneous data of the same monitoring point or the real-time data of adjacent monitoring points to generate a real-time operation data sequence. Further, the step of combining the real-time operation data sequence and the adaptive parameter estimation algorithm to adjust parameters of a pre-established digital twin model of the power system to generate a calibrated dynamic simulation model comprises the following steps: Inputting a real-time operation dat