基于神经网络的风力发电机变频器故障诊断研究  

Fault Diagnosis of Wind Power Generator Frequency Converter Based on Neural Network

作　　者：张涛[1] 刘玲[2] 张新燕[1] 王维庆[1] 胡立锦[1] 

机构地区：[1]新疆大学电气工程学院,乌鲁木齐830047 [2]广东电网公司佛山供电局,广东佛山528000

出　　处：《华东电力》2013年第6期1161-1165,共5页East China Electric Power

基　　金：国家自然科学基金资助项目(50867004);新疆大学博士基金项目(BS100122)

摘　　要：为了准确分辨风力发电机变频器故障类型,针对直驱式风力发电机变频器,给出了一种基于神经网络的风力发电机变频器故障诊断方法。分析了直驱式风力发电机变频器的结构、故障原理和故障类型,得到了一系列不同故障状态下直驱式风力发电机的输出电压波形变化情况,并对设计好的BP神经网络进行训练和测试,基于单纯神经网络训练时间长、预测结果不太理想等因素考虑,尝试采用遗传算法优化神经网络权值、阀值参数。仿真结果表明优化后的神经网络训练时间更短,且有更高的预测精度。In order to accurately identify the fault type of the frequency converter of the wind power generator, this pa- per, based on neural network, presents a frequency converter fault diagnosis method specifically for the direct-drive wind power generator. It analyzes the structure, the fault principle and the fault type of the frequency converter, ob- taining a series of the output voltage waveform changes in different fault conditions. Then, the designed BP neural network is trained and tested. Considering the factors such as the long neural network training time and imperfect fore- cast result, this research tries to use the genetic algorithm to optimize the weights and the threshold parameter of neu- ral networks. The simulation results show that the optimized neural network has＂ shorter training time and the higher prediction precision.

关 键 词：风力发电 故障诊断 神经网络 遗传算法 MATLAB 仿真 

分 类 号：TM614[电气工程—电力系统及自动化]