矿用变频器的故障特征提取方法研究

发布时间：2018-12-26 07:49

【摘要】：变频调速技术作为交流电动机的最重要的调速方式发展的越来越迅速,其控制性能和节能效果显著,广泛的应用于各个领域。然而矛盾无处不在,和其他机器设备相比,变频器也会常常出现这样或那样的故障。其中造成整个系统不能正常运转的主要原因就是主电路某些元器件发生了故障。据有效数字统计,80%的控制系统无法正常工作都是因为部分元器件发生了不同程度的故障,因此研究变频器中主电路这一重要组成部分的故障问题,毫无疑问有着十分重要的意义。因利用实验办法来获取变频器故障数据有一定的困难,所以本文首先运用MATLAB软件对变频调速系统进行在线模拟仿真,指向目前市场上使用最广泛的电压型变频器,利用SIMULINK建立其主电路部分的仿真模型,在线模拟实现了实际情况中几种经常发生的变频器主电路故障波形。因变频器主电路中的逆变部分是其最关键部分之一,故本文选定其作为研究对象,将其所输出的电压波形作为变频器故障特征信息的数据来源,根据分析这些数据的特点来实现对变频器进行故障诊断的目的。本文利用目前主流的两种信号处理方法分别对逆变电路各故障下的电压波形进行信号处理和故障特征提取,最终将两组特征数组分别作为BP (Back Propagation)神经网络的输入,对这两种故障特征提取方法进行研究比较,得出以下结论：由希尔伯特-黄变换法得到各电压波形的本征模态函数分量的振幅均值来作为特征向量,再经BP神经网络实现逆变电路故障定位,仿真试验结果表明该方法可诊断变频器逆变电路故障,与小波变换相比较,有更快的收敛速度和更高的诊断准确度。
[Abstract]:As the most important speed regulation method of AC motor, frequency conversion technology is developing more and more rapidly, its control performance and energy saving effect are remarkable, and it is widely used in various fields. However, contradictions are ubiquitous, and frequency converters often fail in one way or another compared with other machines. The main reason for the failure of the whole system is the malfunction of some components of the main circuit. According to the effective statistics, 80% of the control system can not work properly because part of the components have different degrees of failure, so the main circuit in frequency converter is studied. There is no doubt of great significance. Because it is difficult to obtain the frequency converter fault data by using the experimental method, this paper first uses the MATLAB software to carry on the on-line simulation to the frequency conversion speed regulation system, pointing to the most widely used voltage type frequency converter in the current market. The simulation model of its main circuit is established by using SIMULINK, and several frequently occurring fault waveforms of inverter main circuit are realized by on-line simulation. Because the inverter is one of the most important parts in the main circuit of the inverter, this paper selects the inverter as the research object and uses the output voltage waveform as the data source of the fault characteristic information of the inverter. According to the analysis of the characteristics of these data to achieve the purpose of frequency converter fault diagnosis. In this paper, two main signal processing methods are used to process the voltage waveform and extract the fault feature of the inverter circuit respectively. Finally, the two sets of characteristic arrays are used as the input of the BP (Back Propagation) neural network. By comparing these two fault feature extraction methods, the following conclusions are drawn: the amplitude mean of eigenmode function components of each voltage waveform is obtained by Hilbert-Huang transform method as the eigenvector. Then the fault location of inverter circuit is realized by BP neural network. The simulation results show that this method can diagnose inverter circuit fault. Compared with wavelet transform, this method has faster convergence speed and higher diagnostic accuracy.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TD605;TN773

【相似文献】