Modified SBX and adaptive mutation for real world single objective optimization

Real-world optimization problems often involve highly non-linear objectives and constraints. From an application point of view, it is usually desirable that the global optimum be achieved in such cases. Among selection, crossover and mutation operators of a genetic algorithm, the last two are responsible for search and diversity maintenance. By improving these operators, the efficiency of GAs can be improved. In this paper, we solve the problems specified in "CEC 2011 Competition on Testing Evolution Algorithms on Real World Optimization Problems" using a variation of the Simulated Binary Crossover (SBX) which adaptively shifts between parent-centric and mean-centric recombinations. The shift occurs automatically during program execution through the use of current population statistics and is expected to improve the performance of GA. Further, we also employ a self-adaptive mutation strategy developed earlier