IMPROVING THE EFFICIENCY OF SIMULATED ANNEALING OPTIMIZATIONTHROUGH DETECTION OF PRODUCTIVE SEARCH

摘要

The popularity of simulated annealing for engineering designapplications has grown in recent years, increasing the needfor new techniques that improve algorithm performance. Simulatedannealing is a time-consuming, iteration-intensive algorithm.One area of algorithm enhancement with high potentialimpact is the development of methods for improving the algorithmby reducing the amount of wasted or non-productivesearch. This paper presents an approach to detection of productivesearch based on statistical process control (SPC) concepts.The proposed Detection of Productive Search (DPS) annealingschedule is compared to three other viable schedules using a100-city traveling salesman problem. The DPS schedule producesresults on par with the best from the more traditionalschedules but does so with significantly fewer iterations.