Abstract
Removing impurities from forages is a key step in fine feed process, in the process of sorting weeds and foreign objects from the forages, affected by the working environment of the mechanical arm, dust will cause nonlinear friction in the mechanical arm joints, besides, the different sorting objects cause the problem of uncertain loads, these factors affect the normal operation of the mechanical arm badly. In order to track impurities accurately by end effector of mechanical arm, Kane modeling method was adopted and different control algorithms were analyzed in this paper. Firstly, we designed a conventional PID control algorithm, and analyzed its feasibility condition based on Routh criterion. Secondly, to weaken the effect caused by nonlinear friction in mechanical arm joints and uncertain loads in end effector, we combined Radial Basis Function Neural Network(RBFNN) and Expansion State Observer(ESO) with sliding mode control, proposed an adaptive sliding mode controller based on fuzzy PD compensation(FASMC), numerical simulation demonstrated that the controller can reject the unknown effects of the system and achieved a very good control performances.