Application of a Compact Genetic Algorithm to Pipe Network Optimization Problems

Abstract

Abstract. This paper presents the application of a compact Genetic Algorithm (cGA) to pipe network optimization problems. A compact genetic algorithm is proposed to reduce the storage and computational requirements of population-based genetic algorithms. A compact GA acts like a standard GA, with a binary chromosome and uniform crossover, but does not use a population. Instead, the cGA represents a virtual population for a binary GA by a vector of probabilities representing the chance that the optimal solution has a one at each bit position. The application of the cGA to pipe network optimization problems is considered in this paper and the results are presented for two benchmark examples and compared with existing solutions in the literature. The results show the ability of the cGA to locate the optimal solution of problems, considered with a computational e ort, comparable to improved population-based GAs and with much fewer storage requirements.