摘要
在工业过程的异步电机故障诊断问题的研究中,由于实际运行中异步电机受工作环境的影响以致故障情况复杂,产生不正常信息。传统神经网络故障诊断方法难以对变化环境下电机产生的状态信息进行准确分类,且其训练速度慢,泛化能力低,难以满足实际应用要求。为实现快速可靠的故障诊断,提出了小波包-在线贯序极限学习机故障诊断方法。通过在线贯序极限学习机对小波包分析提取出的电流特征向量进行再学习,可自适应工作环境识别故障。仿真结果表明,改进方法相比传统神经网络能更快速、准确的判断故障类型和故障位置,故障诊断提供了科学依据。
The paper studied the diagnosis problems of asynchronous motors in industrial production process. To make the diagnosis fast and reliable, we proposed a diagnosis method based on wavelet packet-extreme learning machine online. We can use online sequential learning machine to further study the current feature vector extracted from wavelet packet, which is able to identify the failure according to the working environment. Simulation results show that, compared to the traditional neural networks, the new method is more quickly and accurately for the diagnosis of asynchronous motors.
出处
《计算机仿真》
CSCD
北大核心
2016年第11期418-421,442,共5页
Computer Simulation
基金
"十二五"国家科技支撑重点支撑项目(2011BAF18B01)
关键词
故障诊断
异步电机
小波包
极限学习机
Fault diagnosis
Asynchronous motor
Wavelet packet
Extreme learning machine