论文部分内容阅读
传统的基于机理或局部线性化模型的控制策略不足以解决越来越复杂的控制问题,而神经网络用于控制也存在泛化能力差等缺陷,因此本文提出一种将被控对象已知机理和RBF神经网络结合起来卖现逆模控制的方法,一方面能发挥神经网络非线性逼近的强大功能,另一方面利用被控对象已知机理信息指导神经网络的收敛方向,改进神经网络的泛化能力。由此方法设计的逆模控制器,在保证控制精度的前提下,速度远快于标准径向基神经网络逆模控制器,且对扰动、时延、非线性及对象参数的摄动有较强的适应能力,具有良好的控制品质。
The traditional control strategy based on mechanism or local linear model is not enough to solve the more and more complex control problems. However, the neural network has some defects such as poor generalization ability. Therefore, Combining RBF neural network with RBF neural network, the method can not only exert the powerful function of nonlinear approximation of neural network, but also use the known mechanism information of the controlled object to guide the convergence direction of neural network and improve the generalization of neural network Ability. The inverse mode controller designed by this method is much faster than the standard RBF neural network inverse model controller under the premise of ensuring the control accuracy, and has more perturbation, delay, nonlinear and perturbation of the object parameters Strong adaptability, with good control quality.