Non-iterative data-driven controller tuning using the correlation approach

摘要: Data-driven controller tuning for the model-reference control problem is investigated. A new controller-tuning scheme linear time-invariant single-input single-output systems proposed. The method, which based on correlation approach, uses a single set of input/output data taken in open-loop or closed-loop operation. specific choice instrumental variables makes criterion an approximation criterion. and parameters are asymptotically not affected by noise. Although finite length biased, bias analysis shows that it generally improves robustness. effectiveness proposed method illustrated via simulation example.