Performance-Driven Adaptive Output Feedback Control with Direct Design of PFC

Abstract

<div class=""abs_img""> <img src=""[disp_template_path]/JRM/abst-image/00270005/01.jpg"" width=""300"" /> Block diagram of proposed control</div> An adaptive control system is configured simply by output feedback when the controlled system is almost strictly positive real (ASPR). ASPR conditions are, however, very severe restrictions for actual systems. The parallel feedforward compensator (PFC), which is designed for making augmented controlled system ASPR, has been proposed to solve this problem. We propose a performance-driven adaptive output feedback control system with a PFC designed via direct design. Our proposed adaptive control system is to be used for systems whose properties change during operation. Our PFC’s direct design uses the system’s input/output data and readjusts output feedback gain based on a performance index. The effectiveness of the proposed method is confirmed through experiments using liquid level control for the two-tank process system. </span>

Adaptive Output Feedback Control Design with Adaptive Parallel Feedforward Compensator and Feedforward Input via Kernel Method for Nonlinear System

“…It should be noted that the key idea for the adaptive PFC design scheme is the following property of the linear operator on the ideal PFC output given in (15).…”

“…[14], a robust and model-based design scheme of PFC for systems with unmodeled dynamics including a time-delay system has been provided. Recently, adaptive and/or data driven PFC design schemes have been proposed, and thus automatic design strategies of PFC have been developed [15,16]. However, most of them were just for linear systems.…”

Design and Experimental Evaluation of a Discrete‐time ASPR‐based Adaptive Output Feedback Control System Using FRIT

“…The ASPR augmented system in discrete-time can not satisfy the imposed conditions unless a direct feedthrough term exists in parallel with the system. Therefore, one can-not utilize control signal (11) directly because e a (k) is not available in the current step owing to the causality problem. Thus, the equivalent output feedback control signal is generated by…”

“…Thus, for the above adaptive control system with aforementioned assumptions and controller, the following theorem holds. Theorem: Under Assumptions 1 and 2, the use of control input (11) with the adaptive algorithm (13)-(14) will guaranteed the ultimate boundedness of all signals.…”

“…Mizumoto and Fujimoto presented the idea that a feedforward input obtained by adaptive predictive control could compensate for the steady state error. Several other extended works have been published based on this idea. In these publications, the adaptive predictive control output was obtained based on neural networks or generalized predictive control (GPC), which implied that the control structure was complex, and learning cost is considerable large.…”