Design of robust autotuning control system based on model validation

点击翻译按钮获取中文摘要

1) We propose a data-driven design method of PID controllers for a mixed sensitivity control problem based on unfalsified control concept. This method does not require constructing a mathematical model but only requires the input and output response of the plant of a feedback system excited by sinusoidal reference inputs for many frequencies. A falsified region which satisfies a necessary condition on the PID gains is drawn on the parameter plane for each frequency.2) We propose two adaptive identification methods of the critical point, and examine the property by water level-control experiments. The critical gain is identified by using the self-oscillation of the feedback system composed of the plant, a saturation function and a tuning gain where the gain is decreased gradually based the saturation information. This method can be also applied to multi-input multi-output plants. Further, inorder to identify the critical frequency, a new tuning rule which use an adaptive frequency estimator is proposed. These two methods works well for computer simulation, but we found by experiments that the convergence speed is much slower than the simulation result for practical use.3) We examine nonlinear control problems such as antiwindup control for input saturation plant, backward driving control problem of articulated vehicles, and hovering control of twin rotor helicopter model in order to obtain a fundamental knowledge to such a complicated systems and to study such systems by unfalsified control concept in the future. The usefulness is examined by simulation and experiments.