摘要
电网故障中继电保护和断路器的拒动、误动以及信息上传过程中的丢失、畸变等问题使快速、准确的故障诊断仍是一个难题。神经网络方法虽已应用,但神经网络容易陷入局部极小值,针对此情况,提出了基于小波神经网络和遗传算法相结合的故障诊断方法。用遗传算法学习小波神经网络的权值、尺度函数、结构,可以确定用于故障诊断的最优小波神经网络。并对算例进行了仿真,仿真结果表明优化的故障诊断系统优于BP算法的诊断系统,提高了故障诊断精度。
In the power grid fault protection ,relay protection and breaker tripping ,misoperation ,and information in the upload process lost ,distortion and other problems make fast and accurate fault diagnosis is still a problem .Although the neural network method is applied ,the neural network is easy to fall into local minimum .In this case ,the combination of fault diagnosis method based on wavelet neural networks and genetic algorithms were proposed .Gwnetic algorithms learn wavelet neural network weights ,scaling function ,structure to determine the optimal wavelet neural network .Computer simulation was done in an example .The simulation results show that the optimized fault diagnosis system is superior to BP algorithm of fault diagnosis system ,improve the accuracy of fault diagnosis .
出处
《石油化工高等学校学报》
CAS
2013年第6期78-82,共5页
Journal of Petrochemical Universities
基金
国家青年自然科学基金资助项目(51207069)
关键词
电网
故障诊断
小波神经网络
Ppower grid
Fault diagnosis
Wavelet neural network
