改进模拟退火算法在模块划分中的研究及应用

Research and Application of Improved Simulated Annealing Algorithms in Module Identification

模块划分是产品模块化设计的关键技术之一。目前大多采用非数值方法划分模块,数值划分方法主要是使用模拟退火算法或遗传算法。模拟退火算法虽可以一次性得到模块划分最优方案,但是操作困难,效率不高。而遗传算法容易陷入局部最优解。该文在模拟退火算法的基础上,融入遗传算法的种群思想,提出了基于改进模拟退火算法的模块划分方法,研究了其实现的关键技术,并通过VC++6.0将其实现。通过具体的模块划分实例,证实了该方法的高效性和易操作性。

Module identification is one of the key technique for product modularization design. While non-numerical value module identification method is more common, the method of numerical value mainly uses simulated annealing algorithms or genetic algorithms. Using simulated annealing algorithms can get the best solution of module identification at one time, but it is hard to operate and has low efficiency, using genetic algorithms easy gets into local optimization. This paper puts forward module identification way which is based on improved simulate annealing algorithms and researches the key techniques, blends the genetic algorithms＇ population idea. This method is realized with Visual C＋＋ 6.0. Using one example, this method is validated easy operation and high efficiency.