基于小波神经网络的感应电机直接转矩控制

Direct Torque Control of Induction Motor Based on Wavelet Neural Network

为了改善感应电机传统直接转矩控制(DTC)的性能缺陷,特别是低速状态下定子磁链畸变、定子电流谐波大、电磁转矩脉动大的缺点,提出一种基于小波神经网络(WNN)的新型非线性自回归移动平均模型(NARMA)。该模型根据H.Akaike的最终预测误差(FPE)准则确定WNN模型中所需的最佳小波个数。小波神经网络有很强的自学习能力,经过训练可很好地识别DTC系统,基于WNN的NARMA速度控制器可代替PI控制器控制感应电机的转矩。理论分析和仿真表明,该方法是非常有效的。

In order to amend the performance deficiencies of the traditional direct torque control（ DTC）,especially at low speeds,where there are some shortcomings such as stator flux distortion,large stator current harmonics and electromagnetic torque ripple,a novel model denoted as nonlinear autoregressive moving average（ NARMA）model based on wavelet neural networks（ WNN） is presented. According to the Akaike＇s final prediction error（ FPE） criterion,the optimum number of wavelets to be used in the WNN model is selected. The WNN has so strong learning ability as to be trained well to identify DTC system. The WNN controller with the structure of NARMA can be utilized as speed controller to control the torque of the induction motor instead of PI controller. Theoretic analysis and simulation show that the novel method is highly effective.