The Research On Fault Features Extraction Of Mine Frequency Converter

Frequency conversion velocity modulation as the most important speed control way of AC motor develops more and more quickly. Its control performance and energy-efficient is remarkable, and it widely used in various fields. However, contradictions are everywhere, frequency converter compared with other equipment that often is one kind or another of failure. The main reason is that certain components of the main circuit malfunction which cause the entire system is not functioning. According to valid number statistics,80% of the control system can’t work normally because of some components with varying degrees of fault has occurred, so there is no doubt that studying failure problems in this important part of the frequency converter’s main circuit is of great significance.By using the experiment to obtain frequency converter’s fault data has the certain difficulty. At first, the on-line simulation model of frequency conversion velocity modulation system was build by using MATLAB in this paper. And on this base, aim at the popular voltage source frequency converter in the market, the paper build its simulation model and simulate some kinds of frequency converter’s main circuit faults that can be met usually based on SIMULINK. For the inverter section of frequency converter’s main circuit is the one of the most critical part, therefore we selected it as a research object. And its output voltage waveform as a data source of frequency converter’s fault features. According to the analysis of the characteristics of the data achieved the purpose of fault diagnosis.The paper dealt with signal of voltage waveform of inverter circuit under each fault and extracted fault characteristics respectively by using the current mainstream of two kinds of signal processing methods. In the end, the two sets of characteristics array became the input of BP neural network respectively. Of the two kinds of methods of fault feature extraction were compared, and study the following conclusions:At first, the mean amplitude of Intrinsic Mode Function components of each voltage waveform by the HHT transform was regarded as characteristic vector. Then BP neural network identified the fault location of inverter circuit. The test results of simulation show that this method can diagnose the inverter circuit’s fault of frequency converter. It has a faster convergence speed and higher accuracy of diagnosis compared with wavelet transform.