Compound compensator based on cerebellar model articulation controller for surplus torque in passive electric loading system

Because of the coupling of torque control and position control in a passive electric loading system, the movement of the object will bring torque control error that is called surplus torque and will deteriorate the dynamic performance of torque control if not compensated. Aiming at the characteristic of surplus torque, a compound compensation strategy based on cerebellar model articulation controller (CMAC) network has been put forward. Feed forward compensation is the effective method to restrain surplus torque. The keys are the selection of input signal to compensation controller and the modeling of system. Appropriate input signal will help to reduce the sensitivity of the compensation controller to noise and ensure the physical realization and compensation effect. CMAC neural network can study and approach the discrepancy between model and actual system, and ensure the sameness of compensation and actual disturbance channel. Simulation and test results have illustrated the compensation effect.