连续变量问题全局优化的模拟退火法

本文针对过程系统连续变量优化问题中普遍存在的多峰现象, 探索了应用模拟退火法求解其全局最优解。文中根据连续变量问题的特性, 提出了一种相邻状态的产生函数和迭代方案, 并分析了模拟退火过程的起始温度、终止温度以及降温速度等参数对优化计算的影响, 给出了这些参数的适宜区域, 通过三个例题的计算, 将模拟退火法与传统优化方法一梯度法进行了对比分析, 结果表明该法能够有效地解决传统的确定型优化方法所不能奏效的全局优化问题。

A Simulated Annealing Method for Global Optimization to the Continuous Variable Problems

In this paper,aiming at multipeak phenomenon which gen-erally exists in the process system optimization problems with continuous variables,a simulated annealing approach is introduced to search the global optimal solution. According to the property of the continuous variable prob-lem,a function of producing adjacent state is proposed ,and effects of initial temperature,terminal temperatnre as well as annealing schedule to the opti-mal calcuation are analysed,and their suitable scopes are given. The results of three examples and the comparison with the solutions by the gradient-based mathods show the simulated anne aling approach can efficiently solve the global optimal problem which is difficultly dealt with by the traditional deterministic optimal methods.