摘要
为提高10kV高压断路器潜伏性故障声信号识别稳定性与准确性,提升电力系统传输的稳定性,提取融合断路器故障声信号的幅值与MFCC特征,利用PSO优化的SVM分类器,构建出KSST-PSVM断路器故障识别模型。模型首先对声信号进行Mel滤波、离散化处理,降低信号噪点,提高系统计算的时效性;接着利用KS算法在时域上提取声信号幅值特征,同时采用STFT算法在频域上提取声信号MFCC特征,并对双特征进行数据集融合构建;最后利用PSO算法优化SVM的上、下壁弹性变量参数,并利用优化的PSVM算法,完成断路器故障检测识别。消融仿真结果显示,优化策略的叠加有利于故障识别准确性的持续提升,且当优化策略叠加达到三组时,模型的性能参数达到峰值;对比仿真结果表明,上述算法较其它基线算法相比,P、R、F1参数分别平均提升了3.31%、3.74%和3.05%,即本文算法的稳定性、准确性与综合性,均有极大提升。综上,本文提出的KSST-PSVM 10kV高压断路器潜伏性故障声信号识别算法提高了识别检测的有效性,在电力系统仿真中由较重要的意义。
In order to improve the stability and accuracy of acoustic signal recognition of 10 kV high voltage circuit breaker latent fault and enhance the stability of power system transmission,this paper extracts the amplitude and MFCC features of circuit breaker fault acoustic signal,and uses SVM classifier optimized by PSO to construct KSSTPSVM circuit breaker fault recognition model.Firstly,Mel filtering and discretization are performed on the acoustic signal to reduce the noise of the signal and improve the timeliness of system calculation;secondly,the KS algorithm is used to extract the amplitude feature of the acoustic signal in the time domain,and the STFT algorithm is used to extract the MFCC feature of the acoustic signal in the frequency domain,and the data set fusion construction is performed on the double features;Finally,the PSO algorithm is used to optimize the parameters of the upper and lower wall elastic variables of SVM,and the optimized PSVM algorithm is used to complete the circuit breaker fault detection and identification.The results of ablation simulation experiments show that the superposition of optimization strategies is conducive to the continuous improvement of fault identification accuracy,and when the superposition of optimization strategies reaches three groups,the performance parameters of the model reach the peak;Compared with other baseline algorithms,the simulation results show that the parameters of P,R and F,of the proposed algorithm are improved by 3.31%,3.74%and 3.05%,respectively,which means that the stability,accuracy and comprehensive-ness of the proposed algorithm are greatly improved.To sum up,the KSST-PSVM 10kV high voltage circuit breaker latent fault acoustic signal identification algorithm proposed in this paper improves the effectiveness of identification and detection,and has more important significance in power system simulation.
作者
王德全
王海波
李勇群
庄斐
WANG De-quan;WANG Hai-bo;LI Yong-qun;ZHUANG Fei(China National Grid Jiangsu Electric Power Co.,Ltd..Huai'an Power Supply Branch,Jiangsu Huai'an 223002;Nanjing University,Nanjing Jiangsu 210023,China)
出处
《计算机仿真》
2025年第4期93-98,共6页
Computer Simulation
基金
国网江苏省电力有限公司科技项目(SGJSWA00KJJ S2311151)。
关键词
故障识别
特征融合
智能识别
Fault recognition
Feature fusion
Intelligent identification
