Intelligent Controller Emulating Human Control Abilities

本计画提出仿人智能控制（HSIC）理论，其模拟人类的控制能力，并能有效控制参数未知或不确定的非线性系统。以单电机驱动的小车-单摆系统为目标系统进行仿人智能控制。本文首先通过计算机仿真研究了仿人智能控制理论的机理、性能以及与其他控制方法的区别，证实了仿人智能控制理论在模拟人的手动控制方面具有良好的性能。然后，我们采用了一个双连杆臂系统，发展了一个新的HSIC理论。操作员在手臂的第二关节上施加一个力矩来控制手臂的位置，利用操作员的控制输出对仿人智能控制理论进行了修正。最后，基于所提出的仿人智能控制理论，对小车摆系统进行了设计和控制。整个控制过程包括起摆和稳定两个阶段。在摆起过程中，采用非对称Bang-Bang控制为系统收集能量。在镇定过程中，采用了多变量PD正反馈协调解耦控制。HSIC控制器通过在线特性辨识实现控制模式的切换。在此基础上设计的实时智能控制系统成功地实现了小车-单摆系统的起摆和稳定控制。实验证明，该控制系统在摆长变化、摆质量变化、摆形状呈“L”形变化的情况下仍能保持稳定。本课题证明了仿人智能控制理论可以在不知道小车-单摆系统数学模型的情况下，对非线性、多变量、强耦合系统进行有效的控制，如小车-单摆系统的起摆和镇定。

This project proposes a Human Simulating Intelligent Control (HSIC) theory which simulates human control abilities and enables effective control of a nonlinear system whose parameters are unknown or uncertain. A car-pendulum system moving on a restricted track and driven by a single motor is selected as a target system of the HSIC. At first, we investigated the mechanism of the former HSIC theory, it's performance and the difference between the HSIC theory and the other control methods by the computer simulation, then we confirmed that the HSIC theory shows high performance to simulate the manual control of human. Then, we employed a two-link arm system to develop a new HSIC theory. A human operator applies a torque to the second joint of the arm to controlled the position of the arm. The HSIC theory was revised by using the control output of the human operator. Finally, based on the proposed HSIC theory, the cart-pendulum system was designed and controlled. The whole control process consists of two steps : swinging-up and stabilizing. In swinging-up process, asymmetrical Bang-Bang control is employed to gather energy for the system. In stabilizing process, multi-variable PD control with positive feedback coordinating de-coupling is employed. The HSIC controller switches the control modes by means of online characteristic identifying. A real-time intelligent control system, which is designed based on our studies, successfully accomplished swing-up and stabilizing control of the car-pendulum system. The experiments prove that the control system is very stable even though the length of the pendulum varies or mass of it varies or the shape changes like "L". This project proves that the HSIC theory can effectively control the nonlinear, multi-variable and strong-coupling system such as swinging-up and stabilizing a cart-pendulum system without knowing its mathematical model.