摘要
双余度永磁无刷直流电机具有可靠性高、体积小、重量轻等优点,但绕组之间的耦合性对故障诊断提出了更高要求。针对双余度永磁无刷直流电机常见的绕组故障,包括绕组开路和绕组短路,选择相电流作为故障分析信号,通过拆分定子槽,改变控制电路的方式,建立了电机的绕组故障有限元仿真模型。根据故障信号和小波函数的特点,分别采用Daubechies3和coif5小波函数对故障信号进行特征提取。结果表明:在小波分解高频部分的第2层,信号有明显突变,并由此确定coif5小波函数进行故障特征检测。采用coif5小波函数对相电流d2分解系数进行了能量特征提取,得到了各相短路时的故障特征向量。建立了基于PNN神经网络的故障诊断模型,对故障样本进行了诊断,诊断结果准确率100%,验证了所用方法的有效可行。
Abstract ： Dual-redundancy permanent magnet brushless DC motor has the advantages of high reliability, small vol- ume, light weight. But the two windings cause fault diagnosis to be difficult. According to the winding faults, inclu- ding winding open-circuit and winding short circuit, phase current is chosen to be the fault analysis signal. Based on the method of splitting the stator slot and changing the control circuit, the motor winding fault simulation finite element model is established. According to the characteristic of fault signal and the wavelet functions, this paper u- ses daubechies3 and coil5 wavelet function for fault signal feature extraction. The results and their analysis show preliminarily that the signal has a significant change in the second layer of high frequency part and that the coif5 wavelet function is better. The d2 decomposition coefficients of phase current were features extracted by coil5 wave- let function and the fault feature vector is obtained. The fault diagnosis model is established based on PNN neural network. The fault samples were detected by the model. The diagnosis accuracy is 100% and it proves that the method is effective and feasible.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2014年第1期93-97,共5页
Journal of Northwestern Polytechnical University
基金
陕西省自然科学基金(2013JQ7035)
航空科学基金(2013ZC53045)资助
关键词
无刷直流电机
电枢绕组
故障分析
特征提取
brushless DC motors, electric windings, failure analysis, feature extraction
