基于神经网络算法的电感参数辨识在弱磁控制中的研究

Research of Inductance Parameter Identification Based on Neural Network Algorithm in Field-weakening

准确的数学模型和电机参数是内置式永磁同步电机调速系统高性能控制的基础,其中电感参数对电机的稳态和动态运行性能影响较大。由于电机d、q轴电感会随着磁路饱和程度的不同而发生改变,进而会降低其弱磁控制的有效性。考虑到定子电流引起的磁路饱和及交叉饱和效应的影响,基于模型参考自适应的理论基础,论文提出了加入神经网络算法对d、q轴电感进行辨识的算法,并将辨识结果应用到电机弱磁控制中,进而提高参数辨识的鲁棒性。仿真结果验证了该算法的有效性,在不影响电机系统正常运行的同时,实现电感参数的快速辨识,且具有较好的辨识精度。

The high performance control of PMSM speed control system is based on accurate mathematical model and motor parameters.The inductance parameters have a great influence on the steady and dynamic performance of the motor.The inductance of the(IPMSM)d,q axis of the built-in permanent magnet synchronous motor will change with the saturation degree of the magnetic circuit,which will reduce the effectiveness of the field-weakening.Therefore,considering the influence of magnetic circuit saturation and cross saturation effect caused by stator current,a corresponding identification algorithm of d,q-axis inductor is proposed.Based on the model reference adaptive theory,neural network algorithm is added to the on-line identification algorithm of d,q-axis inductance,and the identification results are applied to the field-weakening to improve the robustness of parameter identification.The effectiveness of the algorithm is verified by simulation results.With no influence on the normal operation of the motor system,the fast identification of the parameters is realized,and the identification accuracy is good.