Genetic Algorithms and Hydropower System Design

Abstract

This paper presents research concerning the integration of the genetic algorithm with various elements of hydropower system design. A brief description of the successful application of the genetic algorithm to the shape optimization of a hydraulic/pneumatic chamber and the concrete shell of an arch dam is presented. The paper then concentrates on the initial integration of a structured genetic algorithm with the preliminary stages of large‐scale hydropower system design. The potential of concurrent, multilevel adaptive search across a hierarchical design structure described by both discrete and continuous variable parameters is investigated. The structure comprises the main elements of a large‐scale hydropower system with five potential sites. The objective is to provide the engineer with a selection of optimal system configurations within those sites that display the greatest potential. Problems associated with search across such a discontinuous domain include high levels of parameter/gene redundancy and a high‐probability of premature convergence due to the presence of a large number of discrete design switches. Appropriate mutation regimes are introduced to ensure a diverse search across the design hierarchy during the early generations of an evolutionary search.