Contribution to intelligent control

Abstract

The main purpose of this work is to develop efficient, simple, and robust control algorithms for a large class of nonlinear systems. The idea is to improve the performance of the well known and popular controllers, namely the PID controller and the predictive control, using artificial intelligence tools, such as neural networks, fuzzy logic and meta heuristic optimization methods. The study of meta heuristic optimization methods has allowed to determine the appropriate optimization method that can be used in real time control applications and give good performance. In fact, this study has also allowed proposing an improvement to the teaching learning based optimization algorithm. Furthermore, the carried out research work has allowed proposing several control algorithms, namely the adaptive neural network PID controller, the adaptive Fourier series neural network PID controller, and the neural network model predictive control using the teaching learning based optimization method. In order to improve the convergence rate of the teaching learning based optimization algorithm, a new strategy to the selecting process of the students' pairs, based on the grade of each student during the optimization process, is proposed. The convergence rate and the efficiency of the modified algorithm are assessed by considering several well-known benchmark functions. This algorithm is used to solve the optimization problem of nonlinear predictive control. In the proposed adaptive neural network PID controller, a multilayer percepron neural network is used to online determine the gain values of the conventional PID controller. The adaptation algorithm is developed using the back propagation method. The proposed controller is analyzed and compared with several different controllers through computer simulation and experimental study ontroller. In this work, due to its simple architecture and very attractive proprieties, the ourier series neural network is used to online adjust the parameters of the PID controller. To assess the effectiveness of the proposed controller, the control of a 3-DOF robot arm manipulator is considered and a comparative study, using several control algorithms, is arried out. The third work concerns the constrained nonlinear predictive control using neural etworks and teaching learning based optimization. In this work, a feed forward multilayer eural network is used to predict the future outputs of the system, and the optimization roblem of predictive control is resolved using different versions of the teaching learning ased optimization strategy; namely the TLBO algorithm, the Improved TLBO (ITLBO) and ne enhanced TLBO (ETLBO). To demonstrate the effectiveness of the proposed control Igorithms, the control of the model of the continuous stirred tank rector, and the 2-DOF manipulator robot model, is considered and a comparative study, using several control Igorithms, is carried out