Trang 1Defence Science Journal, Vol. 54, No. 3, July 2004, pp. 395-405© 2004, DESIDOC Hybrid Genetic-simulated Annealing Algorithm for OptimalWeapon Allocation in Multilayer Defence Scenario Sanjay BishtInstitute for Systems Studies & Analyses, Delhi-llQ 054 ABSTRACT Simulated annealing is one of the several heuristic optimisation techniques, that has beenstudied in the past to determine the most effective mix of weapons and their allocation to enemytargets in a multilayer defence scenario. Simulated annealing is a general stochastic searchalgorithm. It is usually employed as an optimisation method to find a near-optimal solution forhard combinatorial optimisation problems, but it is very difficult to give the accuracy of thesolution found. To find a better solution, aji often used strategy is to run the algorithm byapplying the existing best solution from the population space as the initial starting point. Givingmany passes of genetic algorithm can generate the best start-point solution. This paper describesa new hybrid optimisation method, named genetic-simulated annealing, that combines the globalcrossover operators from genetic algorithm and the local stochastic hill-climbing features fromsimulated annealing, to arrive at an improved solution with reduced computational time. Thebasic idea is to use the genetic operators of genetic algorithm to quickly converge the searchto a near-global minima/maxima, that will further be refined to a near-optimum solution by simulatedanneling using annealing process. The new hybrid algorithm has been applied to optimal weaponallocation in multilayer defence scenario problem to arrive at a better solution than producedby genetic algorithm or simulated annealing alone
đang được dịch, vui lòng đợi..
