气动人工肌肉驱动的仿人体上臂结构设计及控制研究

发布时间：2018-11-13 09:49

【摘要】：气动人工肌肉是一种拉伸型的气动执行器,类似于生物肌肉,具有功率/质量比高、响应速度快、噪声小等优点,具有广泛的应用前景。由于其只能提供轴向的拉力,所以大多被使用在两根气动肌肉对拉产生一个转动自由度的结构中。本文利用气动人工肌肉类似人体肌肉的力学特性,将其运用于人体中大转动范围、高灵活性的肩部关节结构设计中。同时对仿人体上臂机构进行优化以获取更优的构型,并对该机构的控制算法进行研究,具体工作如下:首先,分析了人体上臂中骨骼、肌肉及其运动过程中肌肉对骨骼的作用机理,选定使用串并联结构作为仿人体上臂的基本结构,然后运用拓扑结构分析方法判定机构的合理性,在确定好结构之后,对该机构进行三维模型设计。其次,对气动人工肌肉驱动的仿人体上臂机构的运动学和动力学进行分析。由于串并联机构的特殊性,在求解运动学逆解时,机构的位姿具有一定的耦合性,本文解算出机构的末端位姿耦合方程。运用凯恩方法建立机构的动力学方程。由于仿人体上臂机构是冗余机器人,所以在求解它的动力学方程时会存在多解情况。本文使用线性规划来解其动力学方程来获得运动控制时机构中人工肌肉的输出力。同时分析了仿生机构的奇异位形,并且通过运动学逆解和正解混合使用计算该机构的工作空间。然后,采用加权法将仿人体上臂结构的多目标优化转变成单一目标优化,并对权重系数进行归一化处理,运用粒子群算法进行优化。为了提高优化效率,使用自适应调整权重系数的方式,实时调整优化过程中优化函数的权重系数。对比归一化处理和自适应权重两种方法的优化结果可知,加入自适应的优化效率更高,优化结果更好。最后,提出气动人工肌肉的名义收缩力的概念。根据单根肌肉测试数据,计算并绘制名义输出力和收缩率的曲线图,然后进行数据拟合以获得单根气动人工肌肉的数学模型。搭建机构控制试验平台,并编写控制程序。设计闭环PID控制和模糊PID控制两种控制器,然后分别进行两种运动的试验研究。根据这两种控制方法的试验结果对比可知,加入了模糊控制算法的运动具有更好的控制精度。
[Abstract]:Pneumatic artificial muscle is a kind of stretch pneumatic actuator, which is similar to biological muscle. It has the advantages of high power / mass ratio, fast response speed, low noise and so on. Because it can only provide axial tension, it is mostly used in a structure where two pneumatic muscles produce a rotational degree of freedom to pull. Based on the mechanical properties of pneumatic artificial muscles similar to human muscles, this paper applies them to the structural design of shoulder joints with large rotation range and high flexibility. At the same time, the mechanism of imitation human upper arm is optimized to obtain better configuration, and the control algorithm of the mechanism is studied. The specific work is as follows: firstly, the skeleton in the upper arm of human body is analyzed. The mechanism of muscle action on bone during muscle and its movement, the series-parallel structure is chosen as the basic structure of human upper arm, and then the rationality of the mechanism is determined by using topological structure analysis method. The three-dimensional model of the mechanism is designed. Secondly, the kinematics and dynamics of the arm mechanism driven by pneumatic artificial muscle are analyzed. Because of the particularity of the series-parallel mechanism, the position and pose of the mechanism have some coupling property when solving the inverse kinematics solution. In this paper, the coupling equation of the end position and attitude of the mechanism is solved. Kane's method is used to establish the dynamic equation of the mechanism. Since the upper arm mechanism of human body is redundant robot, there are many solutions in solving its dynamic equation. In this paper, linear programming is used to solve its dynamic equation to obtain the output force of artificial muscle in motion control. At the same time, the singular configuration of the bionic mechanism is analyzed, and the workspace of the mechanism is calculated by mixing the kinematics inverse solution and the positive solution. Then, the weighted method is used to transform the multi-objective optimization of the human upper arm structure into a single objective optimization, and the weight coefficient is normalized and the particle swarm optimization algorithm is used to optimize. In order to improve the optimization efficiency, the weight coefficient of the optimization function is adjusted in real time by adjusting the weight coefficient adaptively. Comparing the results of normalized processing and adaptive weight, we can see that the efficiency of adding adaptive optimization is higher and the result of optimization is better. Finally, the concept of nominal contractility of pneumatic artificial muscle is proposed. According to the test data of a single muscle, the curves of nominal output force and contraction rate are calculated and drawn, and then the mathematical model of a single pneumatic artificial muscle is obtained by fitting the data. Set up the control test platform, and write the control program. The closed-loop PID control and fuzzy PID control are designed. According to the experimental results of these two control methods, it can be seen that the motion with fuzzy control algorithm has better control accuracy.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：R318.1;TP242

【参考文献】