Parameter Identification And Adaptive Control Of Flexible Joint

The negative impact of joint flexible is significant especially under high acceleration conditions. As the improving of rigid joint control theory, the flexibility of manipulator joint has become main constraints to the performance of manipulator control system. Besides, with the generation of safety robot, control problem of robot with intrinsic flexible joint draws lots of attention. Under the situation mentioned above, our research focuses on the parameter identification and adaptive control of flexible joint.A layered control theory based on elastic decupling is raised according to the strong coupling and under-actuation of flexible joint. By introducing a non-interference system function estimation method, a new kind of simple structured adaptive decoupling control strategy is constructed in this paper. Meantime, another adaptive decoupling control method based on Radial Basis Function Neural Network is also introduced to verify the efficiency of the new proposed control strategy. Finally, the numerical simulation of the flexible joint control is finished.The traditional system parameter identification method based on direct flexible joint dynamic model is easily disturbed by high-order derivative noise caused by the indirect measurement of angle acceleration. To solve this problem, an indirect system parameter identification method based on the same non-interference system function estimation is raised in this paper. By taking the estimated system function as a prior model, no angle acceleration measurement is needed. Therefore, the new raised system identification method is more robust. What’s more, since this method only needs sensor that already existed in the control system, there is no need to attach extra sensors for parameter identification which convenient for engineering applicationA flexible joint platform is constructed for the demand of experimental research. Based on this platform, sine wave trajectory tracking experiments under different frequencies, different amplitude and different load are implemented, which verified the efficiency of our proposed adaptive decoupling control theory. Robustness of the proposed control theory was proved by trajectory tracking experiment with obvious system model error exist. At last, the proposed non-interference system function estimation method and related system identification method were also verified by corresponding experiments.