An Improvement of Compact Genetic Algorithm and Its Application to Water Distribution Systems Optimization

This paper presents two different techniques in order to improve of Compact Genetic Algorithms (cGA).The first technique proposes a modification in updating method of probability vector of existing compact GA, while the second technique proposes an elitist method which can avoid oscillating in results of program. This method can also result in a better quality of solutions and a higher rate of convergence than do the existing compact GA. The speed-up obtained by each of the methods is predicted and verified with testing results of difficult well-known optimization problem. The results indicae that the use of elitist method and especially modified probability update method, remarkably increases the rate of convergence and reliability in predicting globally optimal problems and reduces the convergence time and requred number of function evaluation. As well as the two above-mentioned techniques, there has seen relations between populationsize and number of tournaments which demonstrated the optimum rate of convergence.