求解一类不可微多目标优化问题的社会认知算法  

Social cognitive optimization algorithm for class of non-differentiable multi-objective optimization problems

作　　者：雍龙泉[1] 

机构地区：[1]陕西理工学院数学系,陕西汉中723001

出　　处：《计算机应用研究》2010年第11期4128-4129,4139,共3页Application Research of Computers

基　　金：陕西省教育厅自然科学研究项目(09JK381)

摘　　要：针对一类不可微多目标优化问题,给出了一个新的算法——极大熵社会认知算法。利用极大熵方法将带有约束的不可微多目标优化问题转化为无约束单目标优化问题,然后利用社会认知算法对其进行求解。该算法是基于社会认知理论,通过一系列的学习代理来模拟人类的社会性和智能性从而完成对目标的优化。利用两个测试算例对其进行测试并与其他算法进行比较,计算结果表明,该算法在求解的准确性和有效性方面均优于其他算法。To solve a class of non-differentiable multi-objective optimization problems,this paper proposed a new method called maximum-entropy social cognitive optimization algorithm.First,used the maximum-entropy function,transformed the constrained non-differentiable multi-objective optimization problem to the approximation unconstrained differentiable optimization problem,then used the social cognitive optimization algorithm to solve this problem.The algorithm was based on social cognitive theory,through a series of learning agents to simulate human social and intelligent thereby completing the optimization of the target.Used two examples to demonstrate the validity of the method and compared the results with the ones of other methods.It shows that the proposed method is more accurate and effective.

关 键 词：社会认知算法 极大熵方法 不可微多目标优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]