Abstract

This work formalizes a multi-objective evolutionary approach for the segmentation issue according to Piecewise Linear Representation. It consists in the approximation of a given digital curve by a set of linear models minimizing the representation error and the number of such models. This solution allows the final user to decide from the best array of best found solutions considering the different objectives jointly. The proposed approach eliminates the difficult a-priori parameter choices in order to satisfy the user restrictions (the solution choice is performed a-posteriori, from the obtained array of solutions) and allows the algorithm to be run a single time (since the whole Pareto front is obtained with a single run and different solutions may be chosen at different times from that Pareto front in order to satisfy different requirements). This solution will be applied to Petroleum Industry in particular to the problem of identifying resources from extraction areas in order to optimize their operational costs and production capacity.