1, a model-based diagnosis method. The basis of this diagnosis method is mathematical modeling, that is to say, the application of mathematical model in motor. Among them, an observer must be used. The information observed by the observer is compared with the actual measurement information of the current sensor, so as to judge the fault. The full-order adaptive observer is used to generate the residual, and the fault of the current sensor is judged according to the residual and the given threshold.
2. Signal-based diagnosis method. This diagnosis method is to diagnose whether the fault occurs by measuring the signal and distinguishing the signal characteristics. When the current sensor fails, different signal characteristics will be displayed and recorded. The characteristics of fault signals are different from those of normal systems. According to the past experience, we can accurately grasp the fault location, identify the fault and solve it. When there is no obstacle to smooth operation, the fault location variables of each phase will approach a fixed value. However, after the fault of the phase current sensor, the value will be obviously different from the other two phases, thus locating the fault.
3. Knowledge-based fault diagnosis method. The basis and foundation of this diagnosis method are slightly different from the former two methods, which require real-time data and historical data, and can only be diagnosed when they are both available. This kind of diagnosis is still very extensive in practical application.