Study Of On-line Parameter Identification For Surface Mounted PMSM

Permanent magnet synchronous motors (PMSM) are now widely employed inhigh-performance drive systems, due to their simple structure, small size, light weight,low loss and high efficiency. The requirement of systems’ performance becomes moreand more strict. The exact parameters are indispensable for implement ofhigh-performance drive system. However, the parameters of the motor vary nonlinearlyin dependency on operating conditions, which reduces the dynamic performance of thesystem. As a result, on-line parameters identification of PMSM has become a hot spotof research. The main challenges for the on-line parameter identification are the noprecision identification results, complex of identification algorithm and decoupling forthe multi parameters identification. First of all, the paper introduces the mathematicalmodel of PMSM, theory of coordinate transformation and vector-control of PMSM; avector-control of PMSM which adopts coupling terms in current loop is introduced.The system can calculate precise voltages and currents under different operatingconditions through adopting coupling terms. Then, the factors of parameters variationand its effects are analyzed. After that, three common on-line parameter identificationalgorithms are introduced: least squares method, model reference adaptive method, andextended Kalman filter method, and their advantages and disadvantages are analyzed.At last, to reduce the effect of the parametric variation, a new on-line identificationmethod is proposed based on the analysis of the relationship between the effects ofparametric variation and current loop PI regulators. This method utilizes outputs of theregulators to identify the parameters under d qsynchronous rotational coordinates.Simulation of on-line parameters identification for surface mounted PMSM isestablished in the Matlab/Simulink platform. Simulation results demonstrate thevalidity of the proposed method, and the parameters can convergence to actual valuerapidly. After the system is stable, outputs of the regulators convergence to zero, so thedynamic performance can be improved.And experiment is established based onTMS320F2812which introduces the proposed algorithm by programming. Theexperiment results also demonstrate the validity of the proposed method.