To increase performance and accuracy of Genetic Algorithm (GA) optimisation methods it is proposed to embed a local search algorithm into the GA. The specific local search algorithm is described and tested with the multi-maxima Rastrigin's function. It is shown that the addition of the local search algorithm indeed increases both performance and accuracy of the GA, as long as the GA is capable of hitting the neighbourhood of the global minimum. The modified GA is also tested on an industry-size system, where it also increases performance - which is important, since industry size systems tend to require considerable computing resources per Monte Carlo Simulation run.
展开▼
