Simulation of Automatic Fault Diagnosis of Inverter Unit of Frequency Converter
In the switching process of the converter,overcurrent impact may cause the voltage fluctuation of the power grid and frequent tripping of the inverter unit.In order to reduce the fault risk of the frequency regulator,an au-tomatic fault diagnosis method for the inverter unit of frequency converter was proposed.Firstly,the spectrum residual prediction equation was constructed by comparing the voltage signals in a normal state with the voltage signals after a fault.Then,the least square estimation method was adopted to update the residual prediction value of the inverter sig-nal spectrum.Moreover,the fault characteristics of the inverter were extracted by windowed Fourier transform and Hartley algorithm.Finally,the fault characteristics were input into the BP neural network for training.The output value is the fault diagnosis result.Experimental results show that the function loss rate of the proposed method decrea-ses to zero when the number of iterations reaches 200,and the fault diagnosis accuracy is close to 100%.
Frequency converterInverter unitFault diagnosisFeature extractionNeural network
NETL
NSTL
万方数据
