Research Of Neural Network Friction Compensation For Servo Systems

There is an increasing number of applications in high-precision motion control systems. But it's very difficult to assure such accuracy due to many factors affecting the precision of motion, including frictions and disturbances in the drive system. The standard proportional-integral-derivative(PID) type servo control algorithms are not capable of satisfying the desired precision under the influence of frictions and disturbances. So, on the other hand, in order to satisfy the desired accuracy, the strongpoint of adaptive neural control is utilized to compensate for the effects of frictions and disturbances in this paper.Firstly, an adaptive tracking control design for robotic servo system combing augmented neural network with fuzzy tuning PD control is proposed. Augmented neural network with accurate approximation capability is employed to approximate the unknown dynamic friction of the robotic servo system. Depending on the finite number of basis functions results in inevitable approximation errors. Therefor, a robust control law is provided to guarantee the stability of the closed-loop robotic servo system than can be proved by Lyapunov thoery. The effectiveness of the augmented neural network-based control approach is illustrated by simulation results.In the later part of this paper, an adaptive backstepping control for turn-table servo system using wavelet network is proposed to compensate for uncertaintied of the system, including friction and disturbance. Meanwhile, the adaptive H_∞control problem based on the neural network technique is studied in this part. By combining the backstepping technique with H_∞control design, an adaptive neural controller is synthesized to attenuate the effect of approximation errors and guarantee H_∞tracking performance for the closed-loop turn-table system. In the work, the structure of property of the system is employed to synthesize the controller such that the singularity problem of the controller usually encountered in feedback linearization design is avoided. Simulation results show the H_∞performance of the closed-loop system and the effectiveness of friction compensation.