Sensitivity based Evolutionary algorithms for Reactive Power Dispatch

This paper presents a new idea with in the evolutionary algorithms for solving the optimal reactive power dispatch problem. Loss sensitivity approach is developed and is used in each of the evolutionary algorithms such as differential evolution (DE), particle swarm optimization (PSO) and genetic algorithm (GA). The objectives are to minimize real power loss and to improve the voltage profile of a given interconnected power system. A term ¿Loss sensitivity¿ at each bus is defined which depends on variation of real power loss with respect to the voltage at that bus. Based on the values of the loss sensitivity, corrective action is taken by adding shunt capacitor at the weak buses identified by weak bus analysis, by controlling reactive generations at the generator buses by judging the sensitivity at these buses, and also tap changing positions are controlled if the tap changing transformers are in between the loss sensitive buses. Then, evolutionary algorithms are run to make one iteration complete. The whole process is allowed to continue for the specified number of iteration. The solutions obtained by this method is compared with the solutions obtained by each of these evolutionary algorithm separately and also with their hybrids with simulated annealing (SA) and with simple quadratic programming (SQP). From the comparisons authors have shown how sensitivity based Evolutionary technique can be a very faithful new tool for the reactive power dispatch.