萤火虫算法优化SVM的高压断路器故障诊断方法

Fault diagnosis method of high-voltage circuit breaker based on SVM optimized by firefly algorithm

为提高支持向量机识别高压断路器故障诊断率,提出采用萤火虫算法优化支持向量机的方法。采用EMD分解振动信号,选取几乎能包含原始信号全部特征信息的前7阶IMF分量,提取其样本熵作为特征输入向量,利用萤火虫算法寻优SVM的关键参数c和g,优化支持向量机模型,诊断高压断路器的不同状态。结果表明,与GA、GWO、GSA算法相比,萤火虫算法优化支持向量机能快速准确地识别高压断路器的状态,故障诊断率达到97.5%。

This paper proposes a method of firefly algorithm to optimize support vector machine, aimed at improving the fault diagnosis rate of high-voltage circuit breaker by using SVM. The study includes decomposing the vibration signal by EMD;selecting the first seven order IMF components almost containing all the feature information of the original signals;extracting the sample entropy as the feature input vecto;optimizing the key parameters c and g of SVM by the firefly algorithm and inputting the extracted feature vector into the trained firefly optimized support vector machine model to diagnose different states of the circuit breaker. The results show that compared with GA, GWO and GSA algorithms, firefly algorithm optimized support vector can identify the status of high-voltage circuit breakers quickly and accurately, and the fault diagnosis rate reaches 97.5%.