Research On The Ivrerter Fauit Detection Of Motor Drive System Based On Current

With the three-phase motor drive system is widely used in the electric vehicles,aerospace and other fields fields, its reliability becomes more and more important.Three-phase motor drive system is generally composed of motor, inverter, motor controller and sensors, the inverter as the weak link which easily occurs fault, its reliability is very important for the safe operation of the whole drive system. By using real-time fault detection of the inverter power devices and processing in a timely and oppropriate manner, the reliability of three-phase motor drive system can be effectively improved.Therefore, we study deeply the fault detection for the three-phase inverter of motor drive system in this paper.Firstly, we make the comparative analysis on the existing fault detection methods which are expert system detection method, the voltage-based approach for open-circuit fault detection and the current-based approach for open-circuit fault detection. Among the current-based approachs for open-circuit fault detection, the normalized current method can only detect the open-circuit fault of single power switch. In order to solve this problem, a new open-circuit fault detection method of the three-phase inverter based on current is proposed in this paper, and then building Simulink model to simulate it. Simulation results show that: the proposed open-circuit fault detection method of inverter can detect and locale all 27 kinds of open-circuit faults under the condition of without increasing the auxiliary detection variables.Secondly, we make the quantitative analysis on the tradeoff between detection time and accuracy of inverter open-circuit fault detection methods. Through a large number of simulation experiments for the open-circuit fault detection method proposed in this paper under the condition of different motor speed, load and detection threshold values,we statistics a series of datas. Using the correlation analysis of statistical analysis methods to analysis of the datas’ correlativity, it is concluded that there is a strong relationship between the detection threshold values with detection time and accuracy. Establishing the regression model which can represent the relationship between detection threshold values with detection time and accuracy through specific model and support vector machine provide an effective prediction ofthe threshold values selection.Finally, we establish a dSPACE model of inverter and three-phasea synchronous motor and design the hardware and software structure of motor controller, and combine them into hardware in loop experiment platform to verity the proposed open-circuit fault detection method of the three-phase inverter based on current. The experimental results show that: the proposed open-circuit fault detection method of inverter can on-line and real-time detect and locale all 27 kinds of open-circuit faults, and it is convenient and can achieves the desired effect that using the fitted function expression to calculate the preset threshold values for balance the detection time and accuracy.