Predictive learning control and application to servo system of DCmotor

摘要

A novel algorithm for learning control is proposed andinvestigated experimentally. It is assumed that the desired output isperiodic and that its period is known. At each sampling time, thecontrol input is modified so as to minimize the quadratic criterion onpredicted future errors. Future errors are estimated from the followingdata: past errors in the last attempt, present error, control input upto the present time, and the step-response data of the system measuredpreviously. Experimental results on the DC servo motor showed thattracking errors were reduced to ±1 pulse after only fourattempts. It is concluded that the proposed method is very useful forcontrollers of machine tools and industrial robots because the algorithmcan be easily implemented on a personal computer or a microprocessor,the learning converges quickly, and no identification is required otherthan measuring its step response