Optimizing Parameters of Direct Adaptive Neural Sliding Mode Controller for Coupled Tank System Using MDE Optimization Algorithm

This paper proposes an MDE (Modified Differential Evolution) Optimization algorithm to optimize the parameters of a Direct Adaptive Neural Sliding Mode Controller (DANSMC) for water level control in a coupled tank system (CTS). The controlled object is nonlinear, with time delays, and subject to internal and external uncertainties such as sensor measurement noise, variations in outlet valve opening, etc. The control system consists of a direct adaptive control component estimated online through an RBF neural network and a sliding mode control component tasked with compensating for estimation errors from the RBF network to ensure system stability, following Lyapunov theory. The MDE algorithm optimizes the parameters of both the sliding mode controller and the direct adaptive controller. To demonstrate the effectiveness of the proposed DANSMC-MDE algorithm, simulations compare it with several other optimization algorithms under the same initial conditions and with the same initial fitness function values to ensure fair comparisons between algorithms. Additionally, the control algorithm with the proposed optimized parameters is compared with the traditional direct adaptive neural sliding mode control (DANSMC) algorithm, the conventional SMC algorithm, and the PID control algorithm.