Nonlinear Neural Control Strategies versus Conventional Control — Case Study and Performance Comparison

The main objective of this paper is to design and implement in MATLAB Simulink R2023b programming environment an intelligent nonlinear neural control strategy for a full-state feedback linearization nonlinear plant model which belongs to a particular class of second degree of linearization. The key idea consists of the use the input-output dataset measurements of a conventional proportional-integral-derivative controller connected in a closed loop control structure with the nonlinear plant. It works in real time based on the online acquisition of the input-output dataset that is processed by a combination of shallow or deep learning neural network structures for its mapping. For “proof concept” and simulation purposes, a model of a shunt-connected dc motor is under investigation as a case study. The effectiveness of the proposed algorithm is demonstrated through an intensive number of simulations conducted on MATLAB Simulink programming platform. For performance analysis comparison, a benchmark is constructed based on the statistic indicators calculated for three control strategies, very useful to reveal the improvements.