The learning space for executing general motions of aflexible joint manipulator is quite large and the dynamics are, ingeneral, nonlinear, time-varying, and complex. The objective ofthis paper is to design a nonlinear system based on the fuzzyneural network control using supervised training, into executingreference trajectories by a flexible joint manipulator. Thestructure identifications of controller networks are performed byusing the Adaptive Neural Fuzzy Inference System (ANFIS), withnew parameters and weight coefficients automatically adaptedand adjusted, in order to decrease position tracking errors. Inorder to adapt and reduce the number of undefined parametersin the network, a new technique is used. Reported research worksuse the Euler method for the resolution of the arm's dynamicfunction, in this paper, a more exact method was used,represented by the Fourth-Order Runge-Kutta (RK4) method. Acomparative study has been carried out between these twomethods in order to prove the effectiveness of the later. Finally, inorder to test the robustness of the proposed approach, it was alsoinvestigated considering parameter variations. The trackingspeed of the model on the system control accuracy was alsoanalyzed. The simulation results show that the proposedapproach has a good tracking effect.
展开▼
