Study On IGBT Open-circuit Fault Diagnosis Of Three-level Active Power Filter

With the massive applications of nonlinear devices,the massive injection of harmonic and reactive power seriously threatens the safe operation of the power grid and the normal use of power equipment.The active power filter(APF)has been widely used for its excellent harmonic compensation effect,especially the APF of three-level main circuit.Compared with the traditional two-level circuit,the number of the switching devices in the three-level main circuit is increased so that the output voltage waveform is closer to the command signal,but the reliability of the circuit is accordingly reduced.Therefore,this thesis studies the open-circuit fault diagnosis of IGBT in neutral point clamped(NPC)three-level active power filter.A model of NPC three-level APF is designed and constructed to provide an experimental platform for the fault diagnosis,based on the analysis of the open-circuit fault type and operation mechanism of IGBT in APF.First of all,aiming at the problem of the selection of testing signal for open-circuit fault diagnosis of IGBT in APF,the testing signal in the circuit is analyzed,and the test method of measuring the voltage across the clamped diode bridge arm is proposed.During the experiment,the current of inverter output side,voltage of DC side,current of power side are selected respectively as the test signal for analysis initially.It is found that when the harmonic source changes,the fault types can not be well distinguished by the above three test methods.But the proposed measuring method can effectively distinguish the different fault types without other auxiliary test signal.Secondly,aiming at the one-sidedness of fault information extraction based on single feature,a multi-feature fusion method based on energy characteristics,uncertainty and complexity is proposed to extract the open-circuit fault feature of IGBT in active power filter.During this process,in order to avoid the over-fitting caused by the increase of the dimension of the characteristic parameter,the principal component analysis method is used to reduce the dimension of the extracted features.And the adaptive experiment of feature extraction on the three-phase rectifier bridge harmonic source in load mutation and triggering angle change is made.The experimental results show that the proposed multi-feature fusion extraction method overcomes the one-sidedness of single feature and has good distinguishing performance at various working conditions.In order to eliminate the influence of noise on the feature extraction method,the noisy signal is preprocessed by the compound denoising method combining median filter and wavelet threshold.The results show that this method can effectively remove the noise components and the impact of noise on feature extraction methods is eliminated.Finally,aiming at the problem that the recognition rate by using traditional algorithm to optimize parameter set of SVM does not meet the requirement,the parameter set optimization algorithm of the fault diagnosis method is studied.The double subgroups fruit fly algorithm is adopted to optimize the parameter set,and the recognition rate of fault diagnosis is improved.So the IGBT open-circuit fault diagnosis method of three-level active power filter is completed.