混合布谷鸟算法在高压断路器故障诊断上的应用

Application of Hybrid Cuckoo Algorithm in High Voltage Circuit Breaker Fault Diagnosis

为了更准确、快速地对高压断路器故障进行分类、诊断,提出一种基于混合布谷鸟算法优化的最小二乘支持向量机(LSSVM)的故障诊断方法。首先提取分合闸线圈的时间和电流特征量得到特征向量,再利用模拟退火算法(SA)与布谷鸟算法(CS)结合形成的混合布谷鸟算法(CS-SA),对支持向量机进行寻优,旨在得到具有最优参数支持向量机分类模型,提高诊断结果的准确性。最后,利用收集到的数据对该算法进行诊断验证,结果表明利用混合布谷鸟算法优化后的LS-SVM得到的分类模型比常用的粒子群算法、遗传算法、标准布谷鸟算法优化得到的模型准确率更高。

A diagnostic method based on least squares support vector machine （SVM） optimized by Hybrid cuckoo algorithm is proposed to accurately and quickly classification, diagnose faults in high voltage circuit breakers. First, the eigenvectors are obtained by extracting the time and current characteristic of the tripping and closing coil. Second, optimizing the support vector machine（SVM） use Hybrid cuckoo algorithm（CS-SA） which synthetized by Simulated annealing algorithm （SA） and cuckoo algofithm（CS）to get the support vector machine classification model with optimal parameter and improve the accuracy of diagnostic results. Finally, using the collected data to verify this algorithm, and, the experimentation shows that the accuracy rate of classification model obtained by LS-SVM optimized by CS-SA will be higher than optimized by particle swarm optimization,genetic algorithm or normal cuckoo algorithm.