Automatic Synthesis of Control System using Learning Method

改善位置伺服系统的瞬态特性的方法，在理论和实践上都有研究，本文提出了一种从系统的响应特性而不是传递函数特性来改善位置伺服系统瞬态特性的新方法。也就是说，即使在传递函数未知的伺服机构的情况下，通过引入学习的概念，通过不断地观察系统调节过程中的响应，证明了可以实现在快速响应方面具有出色性能的理想瞬态响应。利用阶跃响应的过冲值或下冲值对速度反馈回路中的Bang-Bang元件的高度进行修正，最终得到有限时间响应。此外，提出了一种新的控制算法（我们称之为滞后控制）来消除输入信号的影响。该算法在二阶系统模型的伺服系统中引入速度滞环反馈，实现了不受输入影响的Posicast控制原理。最后，在起重机控制系统中进行了实验，验证了所提出的方法能够产生具有良好响应特性的简单控制结构。

Methods of improving transientcharacteristics of positioning servomechanism have been often studied both from theritical and practical aspects.In this work, the author proposes a new method to improve the characteristics by focusing on the response rather than the transfer function of the systems. Namely, even in the case of the servomechanism with unknown transfer function, by way of introducing the concept of learning, by constantly looking at the response in adjusting process of the system, it was proved possible to realize an ideal transient response with outstanding performance in terms of rapid response.Specifically, the finite time response is to be ultimately obtained by correcting the height of bang-bang element in the velocity velocity feedback loop by the overshoot valeu or the undershoot valeu of the step response.Furthermore, a new control algorithm (we call hysterisis control) is proposed to eliminate the influence of the input signal. In this algorithm, a velocity feedback through a hysterisis element is introduced in a servo system modeled by a second order system" to accomplish the Posicast control principle without influence of the input . signal.Finally, some experiments were done in a crane control system to confirm that the proposed method generates a simple control structure with good response features.