Three Level Inverter Based On Neural Network Fault Diagnosis Research

Multilevel inverter has the advantages of overcoming large switch stress and low DC bus voltage utilization compared with the traditional inverter, which is especially suitable for the application of high voltage and high power occasions. The background, highlighting advantages and hardware topology of multilevel inverter are firstly introduced with its development course of the control strategy summarized. Then, it is pointed out that the multilevel inverter fault diagnosis technology is very immature currently, and revealed that difficulties in researches mainly concentrating on four aspects, that is modeling and manufacturing costs, midpoint potential fluctuations, dead time compensation and narrow pulse and passive soft switching by introducing three major fault diagnosis methods of power electronic devices based on signal processing, neural network and support vector machine. To solve this historic topology problem, fault diagnosis problem of multilevel inverter, it is shown that emphasis should be placed on the mechanism of topology and control strategy, etc. By elaborating five trends of multilevel inverter technology in the future in detail, it is indicated that multilevel inverter has broad application prospect.As one kind of multi-level inverters, three-level inverter is widely used in the high-power and high-voltage applications. However, the increasing of the power devices improves the system’s fault rate. How to ensure safe operation of the system has become an important research question. Using three-level diode-clamped inverter as an example, the paper analyzes operation mode and behavior patters of inverter under open-circuit faults under situation, and presents a technique of fault detection and diagnosis for open-circuit fault of power device in three-level diode-clamped inverter based on neural network.Firstly, in terms of three-level inverter’s topology and work principle, fault types of inverter’s main circuit are analyzed in this thesis. The open circuit faults of power devices in main circuit are separated into nine types. Simulation analysis of these faults is implemented in MATLAB software. And establish the table of fault diagnosis. Secondly, using Fourier transform methods to extract various fault waveform of each harmonic components as characteristic parameters of fault diagnosis, spectrum characteristics and fault waveform analysis, are used for fault codes. Fault diagnosis based on neural network model is established. Diagnosis method of neural networks is combined with the wavelet analysis for fault diagnosis rules. Finally, built three level invertors to test the platform, the experimental result confirms this method to have the very good classified effect.And finally, the design was based on the DSP three-level inverter fault diagnosis experiment platform, build a three level of the inverter main circuit. Design the DSP peripheral circuit, including signal buffer protection circuit, driver circuit, voltage and current sensing circuit auxiliary power supply write the three level inverter SVPWM signal process and fault diagnosis procedures, etc.The experimental results show that the proposed three level inverter based on neural network fault diagnosis method has certain feasibility and practical application value.