An Evolutionary Algorithm Applied to Inventory Control for Natural Gasoline

Abstract

A metaheuristic technique for controller design is explored and its performance is assessed for a real-world case. An evolutionary algorithm for self-tuning systems is proposed to tune the optimal parameter settings in order to achieve the best overall performance to control accurately the volume of natural gasoline in a storage tank. For high-level optimization a flexible iterative technique based on Genetic Algorithms is presented. The algorithm is stable and its architecture is simple, thus becoming easy to implement. Not only heuristics that produce superior solutions are taken into account, but also the computational speed is contemplated as a key factor. The experimental evaluation yields satisfactory fitness values and relatively little computational effort is required.