A Novel Approach to Optimal Input Signal Design for Data Driven Control

In the last few years, data driven control is growing very rapidly owing that it does not require identification process of the plant. The data driven controller can be designed directly by the input and output measurement data. This paper studies how to design the optimal input signal for data driven control, and proposes a new input signal optimal correlation function. The optimal input signal is designed such that the expected value of the considered control is reduced. Here some results and conditions which are tools to improve the convergence of classical prediction error method are extended into data driven control. In addition, the candidate attraction domain of the objective function is proposed, and a convergence condition is derived, which can ensure that a given set is a candidate attraction domain. In general, the results and conditions corresponding to the optimal input signal for data driven control are different from classical model based methods. Finally, simulation examples verify the effectiveness and feasibility of the proposed strategy.