Research On Fault Diagnosis Methods Of Three-level Grid-connected Inverter In Micro-grid

Micro-grid can effectively solve the large-scale access of the distributed generation(DG)to the power grid.With the large number of applications of micro-grid,the reliability of micro-grid system has received more and more attention.In the micro-grid system,the inverter is an important connection device between the distributed power supply and the power grid.The multi-level inverter has the advantages of higher working efficiency,lower harmonic distortion in the output and lower voltage stress in power switches.However,multi-level inverter is prone to faults because of low reliability,which affects the normal operation of the micro-grid system.Therefore,it is important to study the fault detection and diagnosis methods of multi-level inverters in micro-grid.This thesis analyses the working principle of Neutral Point Clamped(NPC)three-level inverter,summarizes the fault types of NPC three-level inverter,studies the control mode of inverter in micro-grid,establishes the fault diagnosis model of three-level grid-connected inverter in micro-grid,and verifies the correctness of the model through simulation experiments.The fault feature extraction methods and fault identification methods involved in the fault diagnosis of inverter have been analyzed in recent years.In this thesis,wavelet packet transform and support vector machine,signal sparse representation and support vector machine are proposed.The wavelet packet transform and signal sparse representation are fault feature extraction methods.The feature information of the three-phase inductor current signals of the three-level grid-connected inverter in micro-grid are extracted,and the feature vectors of different fault types are composed.Then the support vector machine is used for fault identification.The data samples are divided into training samples and test samples.The training samples are used to train the support vector machine to obtain the multi-class support vector machine classifier.The test samples are used to predict and analyze the effect of fault diagnosis.The results show that the two fault diagnosis methods can well diagnose the fault of the three-level grid-connected inverter in micro-grid,and the signal sparse representation method has higher diagnostic accuracy.Therefore,the signal sparse representation method is more reasonable and desirable.The fault diagnosis model of two three-level grid-connected inverters parallel system in micro-grid is established.When the overall control strategy of the micro-grid is peer-to-peer control and the control strategy of each distributed power is PQ control,the independent operation of each inverter module can be obtained by analyzing the operating characteristics and fault performance.The fault diagnosis can be performed separately for each inverter module,and the diagnosis result is consistent with the single grid-connected inverter,which can greatly reduce the complexity of fault diagnosis.Therefore,the fault diagnosis of single three-level grid-connected inverter can be extended to the parallel system of multiple three-level grid-connected inverters in micro-grid to improve the efficiency of fault diagnosis.