Disturbances, inaccurate parameters, and unknown model functions (e.g., the friction model), making MPC implementation challenging. The neural network (NN) method has been broadly applied as a sturdy tool in coping with uncertainties and unknown model functions. For instance, the NN approach has been employed to approximate the friction model for implementing friction compensation [25,26], and it was applied to estimate the unknown model parameters and uncertainties for achieving adaptive control [279]. Nonetheless, for the best of our expertise, handful of performs make use of the NN strategy to approximate the FJ robot technique dynamics. Lately, the study of merging MPC and NN strategies has enhanced [30], in which the NN methods are utilized to cope with the difficulty of modeling technique dynamics. For example, in [31], a deep recurrent neural network (RNN) MPC architecture has been established to slice foods. In [324], deep NN was applied to approximate soft robot dynamics for implementing MPC. Besides, NN has been utilized to approximate the MPC laws in [359]. In robot program, the optimization problem of MPC continues to be the challenges as a result of nonlinear dynamic model along with other non-convex constraints [40]. In addition, the robot program frequently suffers the deadloack trouble, which has been well investigated in [413]. A suitable method for solving the nonlinear MPC (NMPC) is differential evolution optimization (DEO). DEO is really a heuristic strategy ML-SA1 custom synthesis proposed by [44], which can be effective for solving numerical optimization issues. DEO has been developed as a stochastic parallel direct search system, and there are various studies on parallel DEO [458]. The DEO algorithm has the benefit of getting a global optimization method that is certainly easy to know and implement, and has sturdy robustness and fewer parameters to become adjusted. Resulting from its positive aspects, DEO has been extensively investigated [49] and effectively applied in diverse fields, like robot manipulator systems [50], mobile robots [513], autonomous automobiles [54], spectrum sensing systems [55], and permanent magnet synchronous motor systems [56]. Even though the integrating MPC and NN procedures made good benefits in robot applications, handful of researches are focusing around the position control with the FJ robot. On 1 hand, the FJ robot method dynamics are tough to get. Contrarily, the optimization process of NMPC is usually a nonlinear programming Sobetirome custom synthesis challenge, which is tough to resolve. Within this study, we present an RNN and DEO primarily based NMPC method for position control of a single-link FJ robot. The RNN is employed to approximate the method dynamics, as well as the DEO algorithm is applied to solve the NMPC controller. The crucial contributions of this research are summarized as follows: Very first, an RNN and DEO primarily based NMPC method is proposed for the position handle of a single-link FJ robot. The merit of this course of action is the fact that not only is the manage precision satisfied, but additionally the overshoots as well as the residual vibration is nicely suppressed. To overcome the difficulty of modeling, a basic three-layer RNN with leaky rectified linear units as an activation function (ReLU-RNN) is established to approximate the FJ robot dynamic model with satisfactory precision. Then, as outlined by the RNN predictive model and MPC strategy, an RNN and DEO based NMPC controller is developed, in which the DEO algorithm is applied to resolve the controller. Finally, to demonstrate the efficiency and efficiency of this method, some numerical simulation comparisons among our technique.