基于摩擦模糊建模与补偿的机器人低速控制

Slow motion control of serial robots with friction compensation based on fuzzy logic system

针对摩擦引起机器人低速时运动性能恶化、作业精确度变差的问题,提出考虑负载力矩影响的摩擦模糊建模方法及模糊自适应鲁棒控制策略。通过摩擦测量实验,提取关节摩擦随负载力矩变化的特征,并提出一种扩展摩擦模型以描述关节摩擦特性。为克服固定补偿难以处理摩擦不确定性的弱点,引入模糊逻辑系统逼近摩擦现象,并实现模型的线性化以设计自适应学习机制。在此基础上,设计模糊自适应鲁棒控制算法,该算法采用前馈的方式补偿关节摩擦的影响,自适应项实现参数的在线调整,并根据系统中不确定性的界,设计鲁棒控制项以保证系统的鲁棒性。对比实验结果表明,所提出的摩擦模型与控制算法使关节最大跟踪误差可控制在0.005°以内,与未考虑负载力矩影响的控制器相比,平均跟踪误差和最大跟踪误差分别降低了25%、14.55%。

Due to the friction, serial robots may appear control precision deterioration at low speed. To deal with this problem, fuzzy modeling of friction was proposed, which considered the impact of load torque, and a fuzzy adaptive robust controller was constructed. Firstly, based on joint friction measure- ment, variation characteristics between friction and load torque were analyzed. Then, an extended friction model was proposed. Since fixed compensation can not handle friction uncertainty, fuzzy logic system was involved to approximate the friction, and linear form of the fuzzy model was derived for adaptive algorithm design. Second, the fuzzy adaptive robust controller was designed. In this control scheme, a feed-forward term was involved for friction compensation, an adaptive term was adopted to adjust pa- rameters on-line, and a robust term was designed using uncertainty bounds to ensure robustness. Finally, experimental results illustrate that adopting the proposed friction model and controller, the maximum tracking error in joint space is less than 0.005 degree. Compared with the controller without considering load torque effect, the average error is reduced by 25%, and 14.55% for maximum error.