机器人末端臂惯性参数辨识的人工神经网络方法  

Application of ANN Method in Inertial Parameter Identification of the End Effector of Robot

作　　者：陈恩伟[1] 刘正士[1] 干方建[2] 

机构地区：[1]合肥工业大学,合肥230009 [2]中国科学技术大学,合肥230026

出　　处：《中国机械工程》2006年第3期268-271,共4页China Mechanical Engineering

基　　金：安徽省自然科学基金资助项目(00043310)

摘　　要：分析了机器人操作臂末端连杆惯性参数辨识的原理及数学模型,提出了一种与传统神经网络问题不同的惯性参数辨识方法,使神经网络的结构与权值具有明确的物理意义,解决了获取样本难的问题。探讨了人工神经网络在系统参数辨识应用中的一般规律及优越性。惯性参数辨识方法简单、直观、运算量小,可应用基于传感器信号的机器人末端惯性参数的在线辨识。在PUMA562机器人上,用实验验证了惯性参数辨识方法的可行性与有效性。Algorithm and mathematical model of identification ior the inertial parameters oi the end--effector of manipulator in a robot were analyzed. A new method of identifying inertial parameters which was different from traditional artificial neural network problems was developed, hence the physical meaning can be found from the structure and the weights of the networks. The difficulty of obtaining samples was resolved. General rules and advantages of application of artificial neural network in parameter identification of the system were investigated. This method is simple, direct and less in operation, which can be applied in on line identification of inertial parameters of the end--effector of a robot based on the signals of sensor. Feasibility and effectiveness of the method are proved by the results of experiment in a PUMA562 robot.

关 键 词：机器人 惯性参数 人工神经网络 参数辨识 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]