良好非线性模型辨识及其内模控制

摘要

针对良好非线性模型及其线性化补偿器模型难以建立的问题,结合神经网络的万能逼近性,提出一种模型辨识和内模控制方法.通过建立系统整体的目标函数,利用传统的BP学习算法,通过优化该目标函数得到良好非线性模型及其线性化补偿器,并给出在适当约束条件下的良好非线性模型及其线性化补偿器惟一性的证明.为提高系统鲁棒性,减小模型误差和外部扰动等不确定性,针对补偿后的伪线性系统设计非线性内模控制系统.仿真结果表明,通过优化该目标函数可以得到精确的的辨识模型和线性化补偿器,能有效地对良好非线性模型实现线性化;对补偿后的伪线性系统设计的内模控制器具有较强的鲁棒性,控制系统能精确地跟踪参考信号.%A novel model identification and internal control method are proposed with the universal approxomation of neural networks to solve the hardness of establish of nice nonlinear model and its linearizing compensator. The system objective function is founded and optimized to capture the nicely nonlinear model and its linearizing compensator, utilizing traditional Back-Propagation algorithm, whose uniqueness is approved under some proper conditions. To improve the systerm robustness and reduce the uncertainty including model error and external disturbance,a nonlinear internal control system is designed on the compensated pseudo-linear system.The results show that the identified model and linearizing compensator are precise by optimizing the system objective function and the nicely nonlinear model can be linearized well. The internal controller designed on the compensated pseudo-linear system has good ability in robustness and the control system can track the reference signal accurately.