永磁同步電機在線參數辨識方法研究

石建飛 戈寶軍 呂艷玲 韓繼超

摘 要：針對在永磁同步電機參數辨識過程中，由于“數據飽和”和噪聲影響，導致傳統的遞推最小二乘法存在參數估計誤差大和收斂慢的問題。利用改進的遞推最小二乘法提高參數辨識的精度和收斂速度，以滿足伺服系統在不同工況下動態性能。首先，結合永磁同步電機數學模型，設計了一種折息遞推最小二乘辨識算法，通過在傳統的最小二乘法中引入“折息因子”增強了算法的靈活性。然后，通過對存在白噪聲干擾的永磁同步電機模型進行辨識算法的動態仿真。最后，利用搭建的實驗測試平臺進行算法的實驗驗證。仿真和實驗結果表明本文提出的折息遞推最小二乘算法，在參數辨識過程中降低了舊數據對辨識結果的影響，增強了算法對噪聲干擾的魯棒性，提高參數辨識結果的準確性和實時性。

關鍵詞：永磁同步電機；參數辨識；折息遞推最小二乘；數據飽和

Abstract： In the process of parameter identification of permanent magnet synchronous motor， due to the influence of data saturation and noise， the traditional recursive least squares has the problems of high error and slow convergence in the parameter estimation. Using the improved recursive least squares algorithm can improve the identification accuracy and rate of convergence， thus meet the dynamic performance of servo system under different working conditions. First of all， combined with the mathematical model of the permanent magnet synchronous motor， a discount recursive least squares identification algorithm is designed， and the flexibility of the algorithm is enhanced by introducing the "discount factor" in the traditional recursive least square. Then， Dynamic simulation of identification algorithm was finished of the motor with white noise model. Finally， experiments were carried out using the experimental test platform. The simulation and experimental results show that the discount recursive least squares algorithm effectively reduce the influence of old data on the identification results and enhances the robustness to noise interference， and improves the accuracy of parameters identification and real time.

Keywords： permanent magnet synchronous motors； parameter identification； discount recursive least square； data saturation