基于模拟退火的改进群搜索优化算法

ccording to the problem that the standard Group Search Optimizer algorithm(GSO) easily traps in a local optimum and has low effieicency,an improved group search optimizer algorithm with Metroplis rule which is based on GSO is presented in this paper.The major improvement is that the Metroplis rule is introduced in the searching mode of producer which is in the process of seaching optimal value.In order to strengthening the global search capability and then jumping out of local extrame point,the Metroplis rule can jump out of the local optimum by accepting a certain probalility inferior solution.Meanwhile,predictive model is accepted by the producer,which makes the producer no longer blindly update and improves the performance of optimization.A set of benchmark functions are employed to evaluate the improved algorithm.No matter in the low and high dimensional case,the optimization of this improved algorithm has better global convergence than PGSO algorithm,GSO algorithm and GA algorithm.The algorithm is applied to group animation and provides new ideas and methods for the realiztion of group animation.