An Experimental Study of Compensation Algorithms for Backlash Nonlinearity of Servo Systems

To research the method effectively eliminating the limit cycle caused by the backlash characteristics of the servo system, a digital servo system, including the position measuring unit before and after the clearance, was designed to test and compare the three different compensation methods: the direct, BP neural networks and switch compensation. The experimental results showed that all of the three methods are effectual to eliminate the limit cycle. The direct compensation has less accuracy, and need combine with the accuracy compensation to achieve the better result; the switch compensation has the highest accuracy; the accuracy of the BP neural networks compensation is between them.