Research On Multi-fault Diagnosis Technology In Photovoltaic Grid-connected Inverter System Based On Artificial Neural Network

As the hub of photovoltaic(PV)power generation system,the operation of PV grid-connected inverter has significant impact on the reliability of the entire PV power generation system.However,in the actual photovoltaic power station,and the internal equipment has to withstand high electrical stress and thermal stress,as well as the influence of the disturbances in grid and DC side,which will lead to the increase of the fault rate of the inverter system.Due to the many types of faults in photovoltaic grid-connected inverter systems and the damage caused by the faults is serious,and normal single fault diagnosis method cannot deal with multiple fault problems in the photovoltaic grid-connected inverter system.Therefore,the design of a universal inverter multi-fault diagnosis system plays a crucial role in enhancing the reliability of photovoltaic power generation system and the generation capacity of photovoltaic power station.In this paper,a universal multi-fault diagnosis system for the three-phase half-bridge inverter is designed.Firstly,the topological structure and control method of the three-phase half-bridge inverter are briefly introduced,and the main faults in the inverter system are classified from different levels.At the same time,the formation and evolution mechanism of the open circuit fault of the switching device and the DC undervoltage fault are analyzed.Secondly,the common feature extraction methods and artificial neural network model are briefly introduced.The technical difficulties in the methods of fault feature extraction and recognition are analyzed,and the key technologies in inverter fault diagnosis are introduced in detail.Then,the universal framework of multi-fault diagnosis is built,and the inverter multi-fault diagnosis system is analyzed through the process of fault diagnosis based on the requirements of data volume and time which is fault detection,fault classification,fault location,and fault isolation.And a fault diagnosis model is built based on the MATLAB platform,the simulation model is introduced through the two aspects of software and hardware.Finally,the switching devices open circuit fault and DC undervoltage fault are simulated by RT-LAB simulation platform,and the designed multi-fault diagnosis system is used for experimental verification.The experimental results prove that the proposed scheme can diagnose multi-fault system in real time and is practical and portable.