Real-coded genetic algorithm enhanced with self adaptation for solving economic dispatch problem with prohibited operating zone

This paper presents the application of self adaptation phenomenon in real-coded genetic algorithms (GA) for the solution of economic dispatch (ED) problems, taking into account the nonlinear generator characteristics such as prohibited operating zones and ramp-rate limits. The self adaptation is achieved by means of tournament selection along with simulated binary crossover (SBX). This selection process has a power exploration capability by creating tournaments between two parents. The better parents are chosen and placed in the mating pool leading to better convergence and reduced computational burden. In SBX, a child string is created from a probability distribution which depends on the location of the parent and thus leads to small probability of creating child far away from the parents. The SBX is advantageous in ED problems where the true optimum is not known beforehand. In these problems, SBX increases the probability of solutions towards the global optimum. The ED problem with disjoint subset (prohibited operating zones) is solved by self adaptive real-coded GA (SARGA). To improve the performance of this algorithm, penalty parameter-less constraint handling scheme is employed to handle power balance, prohibited operating zones and ramp-rate limits. Further, Karush-Kuhn-Tucker (KKT) conditions are applied to the dispatch results verifying optimality. Also, the KKT error metric based termination criterion is effectively implemented for successful termination of the algorithm. Simulation results on the ED problems of varying complexity and dimensionality suggest that the SARGA performs better in terms of solution quality and consistency due to local search ability around optimal solution. Moreover the optimal power dispatch obtained by these algorithms is superior to previous reported results.