Structure and algorithm of hybrid control in cerebella model articulation controller for electric loading system
新型小脑模型关联控制器复合控制在电动加载系统中的结构及算法

The disturbance of surplus torque and the nonlinearity deteriorate the precision of torque tracking in the electric loading system. This undesired effect can not be removed satisfactorily by the conventional control. To deal with this problem, after analyzing the effect of surplus torque on the electric loading system, we propose a hybrid control based on the novel cerebella model articulation controller(CMAC) and investigate its structure and algorithm. This control adopts the desired output and the actual output of the electric loading system as the incentive signals of CMAC, and treats the error between the desired output and the actual output as the training signal for the CMAC structure. A non-uniform quantization scheme is then proposed according to the characteristics of the incentive signals. In the conventional CMAC, errors are equally distributed into memory cells; while the novel CMAC allocates errors according to the Gaussian weighting coefficients. Simulation results show that the proposed hybrid controller effectively suppresses the disturbance of surplus torque and the nonlinearity such as friction, improves the control precision of the electric loading system and enhances the control stability of the system.