Data-driven Feedforward Controller Tuning Method with Convex Optimization and Its Application

Abstract

This paper discusses data-driven parameter tuning of feedfoward controller. In particular, the goal of this paper is to propose a method whose cost function becomes convex for any controller structure. By focusing on Estimated Response Iterative Tuning (ERIT) and equation-error form identification, this paper proposes a method whose cost function becomes a quadratic function. This paper also proposes an online tuning method of feedforward controller. This online tuning method employs Recursive Least Squares (RLS) algorithm to solve the quadratic function. The effectiveness of the proposed method is demonstrated through practical experiments.