【作者】 刘东；

【导师】 陈宏钧；

【作者基本信息】 哈尔滨工业大学 ， 电气工程， 2010， 硕士

【摘要】 欠驱动机械系统是指独立控制输入少于系统自由度的机械装置。这类机械系统由于驱动装置的减少,特别适合应用在对重量和能源敏感的领域,如航空航天,机器人等。然而,驱动装置的减少也给欠驱动系统的控制增加了困难。Acrobot是欠驱动控制研究的基准对象之一,对Acrobot的控制研究有助于解决欠驱动领域的控制难题。本文以Acrobot为控制对象,研究行为约束下其在大范围内的起摆控制问题,旨在使Acrobot起摆到竖直向上的平衡位置时能达到较小的速度,同时缩短Acrobot起摆时间,使其接近于理想的最优起摆时间。行为约束是指Acrobot实物系统中力矩和角度的限制。在这种约束条件下的研究更有实际意义。针对Acrobot实物系统,本文建立了区别于理想模型的Acrobot实物系统数学模型,并通过仿真实验验证了模型的正确性。实物系统的动力学方程中增加了和实际因素相关的参数项,如摩擦力,带轮等,这些因素会对系统产生不同程度的影响。另外,本文采用遗传算法获取了Acrobot实物系统精确的参数,从而保证了控制器的设计。文中采用砰砰控制规律获取了系统的最优起摆时间,然而这种意义下的时间并不能完成起摆,它只提供了一个时间基准。以此时间为依据,文中设计了两种快速起摆的控制器:一种是采用部分反馈线性化并用遗传算法优化其参数的控制器,一种是基于遗传算法的起摆控制器。基于部分反馈线性化的起摆控制器采用能量泵升的策略,使连杆二跟踪一个动态的参考轨迹,从而保证Acrobot起摆。文中改进了参考轨迹函数,从理论上保证了两杆可同时达到最高点,并以时间为指标应用遗传算法优化了控制器的参数,最终达到快速起摆。基于遗传算法的控制器没有采用能量泵升的思想,而是模拟自然界中生物的遗传,进化操作,采用优胜劣汰机制最终使Acrobot起摆。这种方法在保证Acrobot快速起摆的同时,还可以使Acrobot达到最高点时的速度较小。文中最后对实验平台的驱动电路部分做了改进,提高了电机的输出力矩。在Simulink7.2环境下建立了基于RTW的半实物仿真平台,并在此基础上做了实验,验证了控制算法的有效性。更多还原

【Abstract】 Underactuated mechanical system is a kind of system with fewer independent control inputs than degrees of freedom. Because of the reduction of drivers, it particularly is suitable for fields which are sensitive to weight and energy , such as aerospace, robotics. However, the reduction of drivers increases the difficulty of controling system. As a typical underactuated system, studing on the Acrobot is helpful to solve the problem in unedractuated system.In this paper, the aim is to move the Acrobot from its stable downward position to its unstable inverted position. When Acrobot is attached to upright equilibrium, it has a smaller rate. At the same time, the swing up time is reduced as short as possible.In actual system, the torque and angle is restricted. It is more practical studying on this system.The physical system is different from the ideal model. Mathematical model of physical Acrobot is established in this paper. The simulation results verify the validity of the model. Practical factors related to the paremeters is added to the kinetic equation, such as friction, pulleys, etc.. These factors will produce different effects on the system. In addition, for parameter identification of the Acrobot, genetic algorithm is used in the paper, that ensures the design of swing up controller. Bang-Bang control law is used to obtain the optimal swing up time, but it can not complete the heigth task, it only provides a time basis. Based on the time, two fast swing up controllers are designed. One is using partial feedback linearized, whose parameters of controller is optimized by genetic algorithm. The other controller is based on genetic algorithm. In the first controller based on partial feedback linearization, an energy pumping strategy was used to solving the swing up control problem. The second link tracks a dynamic reference trajectory, that ensures the problem of swing-up. In this paper, a new reference function is designed such that the second link may reach upright position simultaneously when the first link reaches upright position. Besides, genetic algorithm is used to optimize the parameters of controller such that the time of swing up can be shorten. The latter based on genetic algorithm does not use the idear of energy pumping strategy. It simulates the biological behavior of genetic and evolutionary operation in the nature. With the mechanism of survival of the fittest, the Acrobot can swing up gradually. In the scheme, the swing up time can be reduced as short as possible. What is more, it can make the rate of Acrobot smaller when Acrobot arrive to the tiptop.At last, the experimental platform is improved. The circuit is designed to enhance the motor output. The simulation platform is established based on the RTW in the Simulink7.2. The experimental results verify the control algorithm.更多还原