Virtual Population and Acceleration Techniques for Evolutionary Power Flow Calculation in Power Systems

This paper first introduces the concept of virtual population for the formation of high quality chromosomes or individuals in a population. It then describes the numerical acceleration and analytical acceleration techniques for the creation of the virtual population. The new concept and the developed acceleration techniques are embedded into the standard GA algorithm to form the Accelerated Genetic Algorithm (AGA). The power and usefulness of the new concept and techniques are demonstrated through the applications of AGA to solving the Branin RCOS, De Jong 1 and Colville problems. The load flow problem in power systems is then introduced. The new techniques developed are incorporated in a constrained genetic algorithm based load flow algorithm. The enhanced algorithms are then applied to solving the load flow problem of the Klos-Kerner power system under very heavy-load condition.