Faouzi Bouani

Multi-criteria optimization in nonlinear predictive control

The multi-criteria predictive control of nonlinear dynamical systems based on Artificial Neural Networks (ANNs) and genetic algorithms (GAs) are considered. The (ANNs) are used to determine process models at each operating level; the control action is provided by minimizing a set of control objective which is function of the future prediction output and the future control actions in tacking in account constraints in input signal. An aggregative method based on the Non-dominated Sorting Genetic Algorithm (NSGA) is applied to solve the multi-criteria optimization problem. The results obtained with the proposed control scheme are compared in simulation to those obtained with the multi-model control approach.

A comparative study on the channel modeling using feedforward and recurrent neural network structures

In digital mobile communication, the non-stationary channel linear modeling become insufficient for channel nonlinear variations. The objective of this work is to select a suitable neural structure for the channel modeling. We present the advantages of a new neural structure, which is the modified Elman network (MEN), applied to digital communication problems such us the channel modeling. By comparison with the multilayer perceptron (MLP), we deduce that the MEN structure has proved the same results with MLP but involve much less computational cost.

Model Predictive Control of fractional systems using numerical approximation

This paper focuses on Model Predictive Control (MPC) of fractional systems based on numerical approximation as an internal model which is used to predict the plant future dynamic behavior. Simulation results are presented to show that the use of numerical approximation achieves better control performances compared to Oustaloup approximation.

Equalization with decision delay estimation using recurrent neural networks

In this paper, we propose a new recursive classifier based on a recurrent neural network. A supervised algorithm is employed to estimate the classifier parameters. The proposed classifier is used to form a non-linear Decision Feedback Equalizer (DFE) for communication channels. A new procedure allowing the estimation of the decision delay is also presented so that the classifier parameters and the decision delay are estimated at the same time. This new DFE leads to suitable equalization performances even in presence of non-linear and non-minimum phase channels.

Implementation of SISO robust predictive control to a three tanks system

This work presents an application of linear and nonlinear robust predictive control onto a three tanks system. The design of the linear solution is based on Single-Input Single-Output Controlled Auto Regressive Integrated Moving Average (CARIMA) model and the nonlinear controller considers Nonlinear Auto Regressive with eXogenous output (NARX) model. Parametric uncertainties and polytopic uncertainties are adopted in order to take into account the uncertain behavior of the system. Based on worst case strategy, the control law is obtained by the resolution of a min-max optimization problem. However, the performance criterion to be optimized is non-convex. To overcome this problem, non-determinist and determinist global optimization algorithms are proposed namely Genetic Algorithms (GA) and Generalized Geometric Programming (GGP). Experiment results onto a three tanks system are given to illustrate the effectiveness of the developed strategies.

Robust predictive controller based on an uncertain neural networks model

This paper deals with the robust predictive control of nonlinear systems. The behavior of the nonlinear system is described by an uncertainty Feedforward neural networks model, i.e. each output layers parameter is uncertain. The control problem is formulated as a minimax optimization one which is a non convex problem. The performances of the proposed controller are illustrated and compared to a classical PI controller by a simulation example.

MIMO nonlinear control based on Divided Difference Filters

Considering the superiority of Divided Difference Filters (DDF) in state estimation of nonlinear systems versus conventional Extended Kalman Filter (EKF), DDF which are derivate-free Kalman filtering approach are exploited in state feedback control based on proportional integral (PI) controller. The proposed combination is applied to a Multi-Input Multi-Output three-tank system. The efficiency of the DDF algorithms in state estimation of the nonlinear system is reflected in the control law synthesized by the PI controller.

Three tanks level estimation using Divided Difference Filters

This paper deals with state estimation of nonlinear systems using Divided Difference Filters (DDF) which are derivate-free Kalman filtering approach. The performances in convergence, robustness and numerical stability of DDF, in state estimation of nonlinear systems are illustrated by an application to a three-tank system. The DDF are easy to implement because they require only the state and the output models and no partial derivatives are looked-for.

Constrained Nonlinear Multi-objective Predictive Control

This paper describes constrained multi objective predictive control of nonlinear systems. A nonlinear model based on the artificial neural networks (ANNs) is used to characterize the process at each operating point. The control law is provided by minimizing a set of control objective which is function of the future prediction output and the future control actions. Three aggregative methods are used to compute the control law. The first and the second methods are based on genetic algorithms (GAs) and the third method is found on the combination between the weighted sum method and the ellipsoid algorithm. The proposed control scheme is applied to a numerical example to illustrate the performance of the proposed predictive controller

Design of GPC closed-loop performances using multi-objective optimisation

In this paper, a strategy for automatic tuning of predictive controller synthesis parameters based on multi-objective optimisation is proposed. This strategy allows computation of the predictive controller synthesis parameters (the prediction horizon, the control horizon and the cost weighting factor) by minimising a set of closed-loop performances (the overshoot, the variance of the control and the settling time). Two simulation examples are presented to illustrate the performance of this strategy in the adjustment of generalised predictive control parameters.