Parameter Identification For Industrial Robot Based On Artificial Bee Colony Algorithm

Industrial robots have become more and more indispensable in many fields due to their productivity. The ever increasing quality standards impose higher requirements on accuracy of a robot. At present, most domestic robots still use PID motion controllers, which has not considered the complex nonlinear dynamics of a robot, resulting in the overload of the actuators, vibrations etc. In order to improve the performances of motion control, model-based motion controllers should be applied. Dynamic parameters are needed to design the model-based controllers, which is usually obtained from experimental identification.Firstly, this paper established the dynamic model of an industrial robot using Newton-Euler method, and the linear form of the equation was then deduced. On this basis, several factors which effects the dynamics of industrial robots were discussed and the effect of joint frictions was then compensated. After that, the model selection and optimizations of excitation trajectory were done. This paper also provided several pretreatment method for data processing which improves the accuracy of parameter identification.Secondly, some traditional identification algorithms were listed as well as their shortcomings. Then Artificial-Bee-Colony(ABC in short) algorithm was proposed to identify the dynamic parameters of the robots. The basic steps of system identification using ABC were listed afterwards.Finally, satisfactory identification result of an actual robot was obtained using ABC algorithm. Fluctuation and error peaks on predictive torque curves could be observed from the result, which is mainly caused by compensating joint frictions using traditional friction model. The accuracy of identification result could be remarkably improved via compensating friction model with higher accuracy and better performance in low-speed condition. Error peaks on predictive torque curves were also restrained by compensating this model. On this basis, the model-based feed forward controller was designed. The tracing accuracy was remarkably improved with the help of feed forward controller.