基于BP神经网络的大功率直流电机故障诊断研究  

Research on Fault Diagnosis of High Power DC Motors Based on BP Neural Network

作　　者：成振华[1] 樊利民[1] 

机构地区：[1]华南理工大学电力学院,广东广州510641

出　　处：《电气自动化》2017年第3期4-5,15,共3页Electrical Automation

摘　　要：提出了一种用人工神经网络实现直流电机故障诊的方法。推导了故障诊断所需的特征参量。在特征参量的基础上改变了传统的电机故障分类方法,将电机故障分为电枢故障、励磁故障、换向器故障和机械故障4类。以MATLAB仿真数据为基础,训练了一个可行的神经网络。测试结果表明,对电机故障诊断的正确率较高,可行性较强,建立的故障诊断模型有效地实现了特征参量提取和故障映射的功能。This paper presents a method of fault diagnosis which uses artificial neural network （ ANN ） to realize fault diagnosis for DC motors. Characteristic parameters needed for fault diagnosis are derived. On that basis, the traditional motor fault classification method is changed, namely, motor faults are divided into 4 types： armature fault, excitation fault, commutator fault and mechanical fault. Based on MATLAB simulation data, a feasible neural network is trained. The test result shows that this approach has high accuracy and feasibility for motor fault diagnosis. The fault diagnosis model established can effectively realize extraction of characteristic parameters and fault mapping.

关 键 词：直流电机 故障诊断 BP神经网络 故障分类 训练样本 

分 类 号：TM331[电气工程—电机]