Fault Detection And Fault Tolerant Control Design Of PV Inverter Circuit

As the core part of photovoltaic power generation system, PV inverter circuit connects the PV components and the loads, and plays an important role in DC to AC conversion. Its stability and reliability directly affects the work performance of the system. Its faults will lead to the damage of the other part of system, and even the power grid and electricity equipment. Therefore, the research of the fault detection and fault tolerant control is of great significance.However the research on the reliability of inverter circuit are mainly concentrated in the field of motor drive. The detection precision, the detectable fault types and the fault tolerant control effect cannot meet the PV inverter circuit whose structure and the function are more complicated. Therefore, using the three-phase and six-switch inverter circuit as an example and aiming at the combined power tube faults, this paper proposes a improved fault detection and fault tolerant control method, the research work mainly contains the following several aspects:1.The control method of PV inverter main circuit is designed, the common fault types of PV inverter main circuit are analyzed and coded, the reliable fault model is established;2.The fault characteristics extracting method based on wavelet packet energy analysis technology is proposed. The energy of the three phase output current signals is extracted. And then the wavelet packet energy spectrum is obtained. The method improves a lot on the basic of traditional wavelet decomposition, can decompose the high frequency wavelet decomposition coefficient and fully characterize the information of the whole frequency band;3.The BP neural network is trained by the fault characteristic samples, which set up the corresponding relation of fault characteristics and fault types. Using the network as the fault classifier to deal with fault characteristic values, can get the corresponding error codes. Adding the current signal average as a fault characteristic, can avoid the in-phase bridge fault misjudgment and make the detection more accurate. Besides, the additional momentum method and auto-adjustive learning rate method are used to optimize the network algorithm and improve the network convergence;4.The fault tolerant structure and control strategy of PV inverter circuit is designedby using the hardware redundancy. And the fault tolerant control strategy of single bridge and the two bridges faults are explored;The experiment results show that: the PV inverter main circuit fault detection and fault tolerant control design proposed by this paper is feasible, which can accurately detect the combined power tube fault and achieve the fault tolerant control of single bridge and the two bridges faults. This method can be further developed to online detection, applied to the motor drive, power supply design, and other fields, has the certain theory significance.