Genetic Algorithm in Optimal Control of Nonstationary Pipeline Flows



The problems of optimal control of nonstationary liquid-gas mixture flows in pipelines have a vast range from optimal model-parameter calibration, through minimization of pressure oscillations, up to optimal pipeline design. The mathematical model of simulation leads to highly irregular optimization goal functions, so a global optimum finding and gradient free optimization method must be applied. Such a method is the genetic algorithm (GA) created by Goldberg. In this paper the questions of variable coding, of choice of GA operator variations and of different probability values, as applied to the optimal control of nonstationary pipeline flows are discussed. Also an application of GA to an optimal model parameter calibration problem is presented. Using GA the friction factors in different sections of the pipeline that is the water conducting system of the hydroelectric power plant "Vinodol" are found that min

Copyright 2013 - Department of Fluid Mechanics and Computational Engineering