基于小波分析的高压开关柜局部放电监测仿真

Simulation of Partial Discharge Monitoring for High Voltage Switchgear Based on Wavelet Analysis

传统开关柜局部放电信号波检测算法易受外部电磁干扰,具有一定的局限性。为解决由于噪声干扰而导致局部放电信号检测准确率低,时效性差的问题,提出一种基于小波分析与特征分析相融合的放电信号优化算法。算法首选采用小波变换算法对局部放电信号进行分解重构,解决STFT变换中频率分辨率低的问题;然后采用软阈值处理算法降低PD模型的噪声,提高波形的平滑度,去除外部电磁干扰;接着基于Pearson相关分析法优化高纬度特征数据,提高模型构建效率;最优采用十折交叉验证法优化SVM的系统参数,构建出SVM-PD放电信号分类模型。消融实验结果表明,三组优化模块的引入能显著提高分类模型的综合性能,降低模型的构建时间;对比实验结果表明,较基线算法相比,新建的模型R提升了2.41%,F1平均提高了1.32%,迁移性上升了1.43%,较文献[9]相比,SVM-PD模型准确率提高了2.36%。因此,本文基于小波分析方法构建的SVM-PD放电信号分类模型在解决外部电磁干扰的同时,提高了信号识别的准确率与时效性。

The traditional partial discharge signal detection algorithm for switchgear is susceptible to external elec⁃tromagnetic interference and has certain limitations.In order to solve the problem of low accuracy and poor timeliness of partial discharge signal detection caused by noise interference,a discharge signal optimization algorithm based on wavelet analysis and feature analysis is proposed.Firstly,the wavelet transform algorithm was used to decompose and reconstruct the partial discharge signal to solve the problem of low frequency resolution in STFT transform.Then,the soft threshold processing algorithm was used to reduce the noise of PD model,improve the smoothness of waveform and remove the external electromagnetic interference.Then,the Pearson correlation analysis method was used to opti⁃mize the high-latitude characteristic data to improve the efficiency of model construction;The system parameters of SVM were optimized by the best ten fold cross validation method.The classification model of SVM-PD discharge sig⁃nal was constructed.The results of ablation experiments show that the introduction of three groups of optimization modules can significantly improve the comprehensive performance of the classification model and reduce the construc⁃tion time of the model;The comparative experimental results show that compared with the baseline algorithm,the R of the proposed model increases by 2.41%,the F1 increases by 1.32%on average,and the migration increases by 1.43%.Compared with literature[9],the accuracy of the SVM-PD model increases by 2.36%.Therefore,the SVM-PD discharge signal classification model based on wavelet analysis method in this paper not only solves the external electromagnetic interference,but also improves the accuracy and timeliness of signal recognition.