Research On Fault Diagnosis And Fault Tolerant Control Of Three-Level ANPC Inverter

Multilevel inverter has many advantages,such as less harmonics,high stability and so on.It has been widely used in high power and high voltage applications.Compared with two-level inverter,multi-level inverter topology is more complex,any power device failure may cause the inverter can not work normally,or even secondary failure,resulting in huge losses.Therefore,it is of great significance to study fault diagnosis and fault tolerance of multilevel inverter.Aiming at the problem that the state of power devices is difficult to detect,this paper studies the fault diagnosis of multilevel inverter based on the research of many scholars at home and abroad.Taking the three-level active neutral point clamped(ANPC)inverter as an example,this paper studies the fault diagnosis of power devices and fault tolerance of single device.The main contents of this paper are as follows:First of all,the topology of three-level ANPC inverter is studied,the working principle of the inverter is introduced,and its commutation path and modulation algorithm are analyzed.The faults of inverter are divided,and two kinds of faults are summarized.On this basis,through MATLAB Simulink simulation,the voltage signals of upper,middle and lower bridge arms at each time are extracted as detection signals,and the fault characteristics are determined by analyzing the voltage waveform.Then,according to the fault characteristics of three-level ANPC inverter,the fault diagnosis is divided into two parts,one is fault feature extraction,the other is fault diagnosis algorithm.In this paper,aiming at fault feature extraction,the energy spectrum entropy method is proposed to process the voltage signal of bridge arm,which effectively solves the problem that the original feature signal is not suitable for direct input into neural network,and can effectively save the original feature of fault.An optimized wavelet neural network is proposed for fault diagnosis algorithm,which overcomes the shortcomings of traditional neural network,such as slow training speed and low accuracy.In order to study the fault tolerance of single device in different position,two methods of changing modulation strategy and topology structure can be used to achieve better fault tolerance control,and the feasibility of two fault tolerance control methods is verified by MATLAB Simulink simulation.Finally,a three-level inverter experiment platform based on DSP + FPGA is designed,and the inverter fault diagnosis is completed.Firstly,the voltage of upper,middle and lower bridge arm of three-level ANPC inverter is collected as the measurement signal,and the energy spectrum entropy is used to extract the characteristics of bridge arm voltage signal,establish the fault eigenvector,and divide the eigenvector into training data and test data;secondly,the wavelet neural network is optimized through the training data,the fault diagnosis model is established,and the performance of the model is verified by the test data Finally,the feasibility of the fault diagnosis model is verified by the experimental platform.The experimental results show that compared with other algorithms,the algorithm in this paper has the advantages of fast speed and high accuracy of fault feature identification,and it is suitable for real-time fault diagnosis of three-level ANPC inverter.