基于RBF神经网络非线性预测模型的开关磁阻电机自适应PID控制

RBF ANN Nonlinear Prediction Model Based Adaptive PID Control of Switched Reluctance Motor

开关磁阻电机的非线性和变参数特性使得采用传统的PID控制很难取得较好的控制效果。人工神经网络在一定的条件下可以任意精度逼近任意非线性函数且具有较强的自学习、自适应、自组织能力。故将其与传统的PID控制相结合构成神经网络自适应PID控制策略,应用于非线性严重的开关磁阻电机,可实现对开关磁阻电机的高性能控制。同时,神经网络所具有的非线性变换特性和高度的并行运算能力使得其适合建立非线性预测模型进行参数预测。通过对被控系统参数的预测,可提高系统的动态响应性能。该文采用两个神经网络-BP神经网络和RBF神经网络来分别构成神经网络NNC和神经网络NNI。神经网络NNC进行自适应PID参数调节；神经网络NNI用来建立非线性预测模型进行参数预测。为进一步加快神经网络的学习收敛速度,该文采用变学习速率的神经网络学习算法,学习速率随收敛过程误差的大小而自适应地进行调整,这可大大加快神经网络学习训练的收敛速度,进一步提高系统动态响应速度。实验结果表明,系统的动态响应快,超调小,稳态精度高,鲁棒性强,有较强的抗扰动能力,具有较好的控制效果。

The inherent nonlinearity of switched reluctance motor（SRM） makes it hard to get a good performance by using the conventional proportional-integral-derivative（PID） controller to the speed control of SRM. This paper develops a radial basis function（RBF） artificial neural network（ANN） nonlinear prediction model based adaptive PID controUer for SRM. ANN, under certain condition, can approximate any nonlinear function with arbitrary precision. It also has a strong ability of adaptive, serf-learning and self-organization. So, combining it With the conventional PID controller, an ANN based adaptive PID controller can be constructed. Appling it to the speed control of nonlinear SRM, a good control performance can be gotten. At the same time, the nonlinear mapping property and high parallel process ability of ANN make it suitable to be applied to establish nonlinear prediction model to perform parameter prediction. In this paper, two ANN-NNC and NNI are employed. One is back propagation（BP） ANN with sigmoid activation function. Another is an ANN with RBF activation function. The former is used to adaptively adjust the parameters of the PID controller on line. The later is used to establish nonlinear prediction model performing parameter prediction. To improve the convergence speed of ANN, an adaptive learning algorithm is adopted in this paper that is to adjust the learning rate according to the error. This can increase the convergence speed of ANN and make the system response quick. The experimental results demonstrate that a high control performance is achieved. The system responds quickly with little overshoot. The steady-state error is zero. The system is robust to load torque disturbance.