Research On Short Circuit Fault Diagnosis Of A Three-Phase Voltage Source Inverter

With the development of power electronic technology,three-phase voltage source inverters are becoming more and more important in the fields of transmission systems,new energy generation,electric transportation and smart grids due to their good control performance,and have now become an industry standard.The inverter is a key equipment in the motor speed control system,which plays important role in the transformation and control of electrical energy.The switching devices of the main circuit in the inverter realize the control function through the frequent on/off of the drive signal,so it is the weakest link that is most prone to failure.The main fault types are open circuit or short circuit of switching elements,with short circuit fault being one of the most serious cases.There are many reasons for short circuit fault,such as aging,overvoltage,thermal breakdown,wrong drive signal,etc.The short circuit occurs for a very short time(usually about 10μs)and is extremely destructive.If it is not detected and removed in time,it is easy to cause serious consequences such as equipment damage,system shutdown and even personnel safety.Therefore,the study of inverter short circuit fault diagnosis is of great significance to the safe operation and stability of the system.This paper takes the TP-VSI in the variable frequency speed regulation system of the mine hoist as the object,and studies the short-circuit fault diagnosis of the switch tube in the inverter.The structure and principle of TP-VSI are analyzed,and its short-circuit failure mechanism is studied.Focus on the short-circuit fault mechanism of the switching tubes in the inverter and carry out the simulation of single and double short-circuit faults of the switching tubes.On this basis,the TP-VSI short-circuit fault diagnosis method based on normalized average current Park vector transform is proposed.The short-circuit fault of the inverter switching tube is diagnosed and precisely located according to the current trajectory graph traits and the normalized current amplitude.The normalized currents are obtained by normalizing the current under different operating conditions to reduce the influence of load transformation.Compared with the knowledge-based method,this method does not need to acquire a large amount of fault data for training and learning,simplifies the diagnosis process and computation,and improves the efficiency of TP-VSI short-circuit fault diagnosis.Based on the fault mechanism and simulation analysis,the TP-VSI short-circuit fault diagnosis methods based on BP and Rough Set Genetic Algorithm-Bayes Net(RS-GABN)are proposed,respectively.The BP-based fault diagnosis method uses fast Fourier transform to extract the feature values under different faults,constructs a fault feature vector,and establishes a 3-layer BP neural network to form a one-to-one correspondence between fault features and short-circuit faults for fault diagnosis to achieve precise location of short-circuit faults.The RS-GA-BN based fault diagnosis method uses the attribute simplification and global optimization features of RS-GA to perform discrete simplification of the fault features with high importance obtained from the samples,which avoids the redundancy of fault attributes and the complexity of data processing.The BN fault diagnosis model with probabilistic inference is constructed and optimized for short-circuit fault diagnosis based on the diagnosis time and diagnosis accuracy,which improves the fault diagnosis efficiency and realizes the TP-VSI switching tubes shortcircuit fault diagnosis.A TP-VSI short-circuit fault experimental platform was constructed to verify the normalized average current Park vector transform,BP and RS-GA-BN algorithms.The construction of TP-VSI short circuit is studied from three aspects: the construction of the experimental platform,the acquisition of the waveform and the verification of the diagnosis algorithm,including the selection of the experimental device,the setting of the short circuit mode and the setting of the current limiting protection circuit.The three aspects of the experimental platform construction,waveform acquisition and diagnostic algorithm verification are studied respectively,including the selection of experimental devices,the setting of short-circuit mode and the setting of current-limiting protection circuit.According to the obtained short-circuit fault waveforms,the corresponding fault characteristics are extracted for the verification of the diagnostic algorithm.The experiment realizes the experimental analysis and algorithm verification of TP-VSI shortcircuit fault,which proves the feasibility and reliability of the fault diagnosis method.48 pictures,16 tables,100 references.