迭代学习控制器参数的数据驱动自适应整定方法

Data-Driven Adaptive Tuning of Iterative Learning Control

针对PID型迭代学习控制(iterative learning control,ILC)方法,提出了两种数据驱动自适应整定(data-driven adaptive tuning,DDAT)方法。首先采用紧格式迭代动态线性化(compact form iterative dynamic linearization,CFIDL)方法将原始的非线性系统转化为等价的线性数据模型,设计了一个目标函数来动态地调整PID型ILC的学习增益。其次,通过对设计的目标函数进行优化,提出了一种基于CFIDL的DDAT方法。该方法只使用实际的I/O数据,而不需要任何机理模型信息。进一步,引入偏格式迭代动态线性化(partial form iterative dynamic linearization,PFIDL)方法对结果进行扩展,提出了一种基于PFIDL的DDAT方法。所提出的两种DDAT方法都可以提高PID型ILC对不确定性的鲁棒性。最后,通过仿真验证了两种方法的有效性。

In this paper,we propose two data-driven adaptive tuning(DDAT)approaches of PID-type ILC.First,we use a compact form iterative dynamic linearization(CFIDL)method to transfer the original nonlinear system into a equivalent linear data model,and we design an objective function to dynamically tune the learning gains of ILC law.Then,by optimizing the designed objective function,a CFIDL based DDAT method is proposed.This DDAT method only uses the real I/O data and doesn't need to know any mathematical model information.On this basis,we introduce a partial form iterative dynamic linearization(PFIDL)method to extend the research results,and propose a PFIDL based DDAT method.Both the proposed DDAT methods can help the PID-type ILC have a better robustness against to the uncertainties.Finally,the effectiveness of the two proposed DDAT-based ILC methods is verified by the simulations.