Research On Online Rotor Resistance Identification Of AC Induction Motor For Electric Vehicle

The drive performance of AC induction motor (ACIM) is highly improved by vector control. As a key part of vector control, filed orientation is very important which is dependent on motor parameter, especially rotor resistance. However, rotor resistance can change rapidly due to temperature rise of the motor and the change of environment temperature, which should be resolved by online identification. In this paper, the traditional MRAS based rotor resistance identification method is analyzed first. On that bias, a new rotor resistance identification method which is free of stator resistance is proposed. Hope the paper will make contribution for improving the performance of vector control.This paper explains the theory of vector control. The equivalent circuits of one phase under no-load and rotor-locked tests and the motor parameters calculation equations are deduced by the T equivalent circuit. Thus, motor parameters like stator and rotor resistances, self-inductances can be determined offline.Traditional rotor resistance identification method is realized with MRAS. Flux observer of voltage model is used as the reference model and that of current model is chosen as the adjustable model. The adaptive system is constructed by PI controller. Phase voltage and current distortions of the observers are eliminated by dead time compensation method. Integration saturation of the voltage model observer is resolved by discrete high pass filter and amplitude compensation factor is introduced through theory analysis. The reliable observers are verified by simulation and experimental results.The decreased accuracy of the traditional MRAS based rotor resistance identification method caused by stator resistance is analyzd. It can be determined that the magnetic component and rotor flux will change a lot when rotor resistance is inaccurate. Through the change of reactive power equation under different reference formats, the real magnetic component and rotor flux can be obtained. Utilizing the calculated real rotor flux as the reference model and rotor flux observer of current model as the adjustable model, a new MRAS based rotor resistance identification method can be proposed. The calculation of real rotor flux is free of stator resistance, thus, the proposed method is not influenced by stator resistance. Simulation and experiment are constructed and their results verified the validity of proposed method.