供电系统中高压断路器故障优化诊断仿真

Simulation of High Voltage Circuit Breaker Fault Diagnosis Optimization in Power Supply System

提高高压断路器的故障诊断准确率对降低生产成本、提高运行可靠性具有重要作用。由于高压断路器故障种类繁多、监测信号有限,导致高压断路器故障诊断准确率不高。为了提高高压断路器故障诊断准确率,提出了基于概率神经网络算法的高压断路器故障诊断方法。在高压断路器振动信号能量进行量化的基础上,构建了概率神经网络诊断模型,利用样本数据集进行训练,获得最优的高压断路器故障诊断模型,最后利用测试集对模型进行检验。实验证明,利用概率神经网络进行高压断路器故障诊断具有训练过程简单、诊断速度快、准确率高和训练样本追加能力强等优点。与标准BP网络相比,概率神经网络可有效地提高高压断路器的故障诊断准确率。

ABSTRACT ： It＇s important to improve the high voltage circuit breaker fault diagnosis accuracy to reduce the produc- tion cost and improve its operation reliability. Because of a wide variety of fault types, it is difficult to accurately i- dentify the source of failure： In order to improve the high voltage circuit breaker fault diagnosis accuracy, based on probabilistic neural network, a fault diagnosis method of high voltage circuit breaker of the algorithm is proposed in the paper. Firstly, the energy of vibration signal energy in high voltage circuit breaker was quantify, and then the probability neural network diagnosis model was built. The sample data set was used for training to obtain the optimal high voltage circuit breaker fault diagnosis model. Finally, the test set was used to test the model. The experimental results show that the probabilistic neural network for fault diagnosis of high voltage circuit breaker has the advantages of easy training, fast diagnosis, high fault diagnosis accuracy. Compared with BP network, the probabilistic neural network can effectively increase the high voltage circuit breaker fault diagnosis accuracy.