Genetic Algorithm for Optimization of Heating Networks

An algorithm for redistribution of flows in an oriented network is proposed, the maximum flow of which is sought by the Push-Relabel method in the Maple system of computer mathematics. At each stage of optimization, the most effective options are identified, the combination of which gives the next generation of genes. It is shown that highly ramified networks are most sensitive to the distribution of the capacity of individual branches. For example, the maximum flows in networks with two and three parallel branches of the same bandwidth are compared. It is shown which parameters of the algorithm have the greatest influence on the result.