Sciences & technologie. B, Sciences de l’ingénieur
Volume 0, Numéro 35, Pages 9-13

A Neural Network Controller Optimised With Multi Objective Genetic Algorithms For A Laboratory Anti-lock Braking System

Authors : Lamamra K . Belarbi K . Bosche J . El Hajjaji A .


In this work, we consider the design of a neural network controller for the ABS laboratory system witch’s highly non linear. The objective is to control the wheel slip. The controller is designed off-line using a multi-objective optimisation process solved using a multi objective genetic algorithms. The objective of the design process is to find a satisfactory controller with a reasonable structure. The structure is defined as the number of input variables and the number of neurons in the hidden layer. Thus the multi objective genetic algorithms has to minimize three objectives: the number of neurons in the hidden layer, the error which is the difference between the wheel slip reference and the real wheel slip and the third objective is the number and type of inputs to the network. The results of simulation are encouraging.


ABS systems, Neural Network, Multi Objective Genetic Algorithms