电液伺服系统摩擦参数辨识与补偿控制  

Friction Parameter Identification and Compensation Control of Electrohydraulic Servo System

作　　者：冯浩 姜金叶 宋倩玉 殷晨波[3] 俞宏福 曹东辉 FENG Hao;JIANG Jinye;SONG Qianyu;YIN Chenbo;YU Hongfu;CAO Donghui(School of Artificial Intelligence,Nanjing University of Information Science&Technology Nanjing,210044,China;School of Computing,Nanjing University of Information Science&Technology Nanjing,210044,China;United Institute of Excavator Key Technology,Nanjing Tech University Nanjing,211816,China;SANY Group Co.,Ltd.Kunshan,215300,China)

机构地区：[1]南京信息工程大学人工智能学院,南京210044 [2]南京信息工程大学计算机学院,南京210044 [3]南京工业大学挖掘机关键技术联合研究所,南京211816 [4]三一重机有限公司,昆山215300

出　　处：《振动．测试与诊断》2024年第5期922-927,1037,1038,共8页Journal of Vibration,Measurement & Diagnosis

基　　金：国家自然科学基金资助项目(52105064);江苏省基础研究计划自然科学基金面上资助项目(BK20221342);国家重点研发计划资助项目(2021YFB2011904);南京信息工程大学科研启动经费资助项目(2021R042);江苏省研究生实践创新计划资助项目(SJCX23_0401)。

摘　　要：针对非线性摩擦造成挖掘机器人电液伺服系统稳态误差和低速爬行的问题,需要精准辨识摩擦以进行摩擦补偿。首先,将Karnopp速度阈值理论引入经典Stribeck模型,建立挖掘机器人电液伺服系统非线性摩擦模型,并根据非对称液压缸力平衡方程建立目标函数,测量不同恒定速度下有杆腔和无杆腔的压力计算实际摩擦力;其次,优化遗传算法的适应度、交叉概率和变异概率,分别采用改进遗传算法和基本遗传算法对摩擦模型中4个未知参数进行辨识;最后,结合前馈补偿控制器进行正弦轨迹跟踪实验。结果表明:所提出的改进遗传算法辨识精度最高,相较于基本遗传算法,模型误差减少了34%;2种摩擦模型下的正弦轨迹跟踪误差分别为26 mm和59 mm,验证了所提出的摩擦模型在提升挖掘机器人性能上的优越性。Nonlinear friction force will cause steady-state error and low-speed crawling phenomenon of robotic excavators,accurate identification of friction is an important premise of friction compensation.Karnopp velocity threshold theory is introduced into the classical Stribeck model to establish the nonlinear friction model of the ro‑botic excavator.On this basis,combined with the force balance equation of the asymmetric hydraulic cylinder,the objective function in the identification algorithm is established.The actual friction force is calculated by measuring the pressures of the rod and rodless chambers of the hydraulic cylinder at different constant speeds.Furthermore,the fitness function,the crossover probability and mutation probability of the genetic algorithm are im‑proved.Four unknown parameters in the friction model are identified by improved genetic algorithm and basic genetic algorithm respectively.The friction model identified by two different algorithms are compared with the measured friction force.The results show that the nonlinear friction model which is identified by the improved genetic algorithm has the highest accuracy,and the prediction error of the nonlinear friction force is reduced about 34%.Finally,sinusoidal trajectory tracking experiments are carried with the feedforward compensator.The sinusoidal trajectory tracking errors of the two friction models are 26 mm and 59 mm respectively,which verifies the advantages of the proposed friction model in improving the performance of the robotic excavator.

关 键 词：电液伺服系统 挖掘机器人 摩擦辨识 遗传算法 

分 类 号：TH117[机械工程—机械设计及理论] TP27[自动化与计算机技术—检测技术与自动化装置]