Research On Identification Algorithm In Permanent Magnet Synchronous Motor Parameter

Because of the simple structure, small size, light weight and low loss, et al, Permanent magnet synchronous motor(PMSM) can be applied to High-performance drive system and other industrial areas with the rich storage of rare earth permanent magnetic material in china.As a strong coupling, nonlinearity and time-varying dynamical system, the parameters of PMSM can be easily influenced by temperature, flux saturation and stator current, which results in not only low operation reliability but also high difficulties of control system. However, the realization of a high-performance PMSM control system relies heavily on precise PMSM parameters. Therefore, aiming to improve the performance of PMSM, it is a prerequisite to accurately identify the real-time motor parameters.Firstly, this paper introduces the basic structure, mathematical model and the current parameter identification technologies of PMSM, transforming the mathematical model from static three-phase coordinates to rotary two-phase d-q coordinates due to the vector transformation. Some other theories such as the basic vector control principle and the implementation of space vector pulse width modulation technology are also introduced, and the reason for the application of id=0 control method has been explained after compared with several common decoupling methods.In order to solve the slow speed and height error of traditional particle swarm optimization(PSO) and least square method(LSM) when dealing with motor multi-parameters off-line identification problem, Coral Reefs Algorithm(CRO) has been applied, which has been validated to be effective in terms of estimate speed and estimate accuracy when compared with other methods. Meanwhile, Gaussian Mutation and Cauchy Mutation have been simultaneously introduced to CRA to avoid trapping in local optimum, and the experimental results indicate that this approach outperforms other methods on the aspects of convergence rate and accuracy.According to the problem that Forgetting Factor Recursive Least Squares(FRLS) method with forgetting factor can be easily affected by the size of forgetting factor and unstable results, it has been combined with multi-innovation algorithm to improve the performance. Considering that the message length is influenced by convergence rate and identification accuracy, it is appropriately determined by a set of experiments. Experiment results based on both the constant parameter and step parameter transformation show that this approach represents better stability and faster convergence rate, especially superior tracking performance under unchangeable parameters.