基于随机搜索策略确定介质间边界的非线性规划方法  

Nonlinear Planning Algorithm Used in Boundary Determination Based on Stochastic Seeking Strategy

作　　者：王一清[1,2] 黄惟一[1] 宋爱国[1] 

机构地区：[1]东南大学仪器科学与工程系 [2]徐州供电公司调度中心,徐州221000

出　　处：《电子测量与仪器学报》2005年第1期61-65,共5页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金资助项目 (No .6 9875 0 0 4 )

摘　　要：利用非线性规划方法对不同属性介质间的边界确定问题建立了相应的数学模型 ,为了克服目标函数的Hesse阵不能满足半正定条件 ,不能用凸规划的方法进行直接求解 ,本文在迭代寻优的过程中引入了混沌机制 ,提出了一种基于混沌搜索策略的混合优化方法。借助于混沌系统的遍历性和非线性规划方法的选择性达到了快速整体寻优的目的。该方法能有效地实现不同介质之间的边界辨识 ,从而可以解决诸如空洞探测等无损检测问题。相较于此前提出的方法 ,本文的算法不需要计算目标函数的导数 ,而且对迭代初值也没有特别的限制 ,从解空间的任意点出发均能收敛到全局最优解 ,避免了常规进化算法中容易出现的停滞现象 ,强化了发现全局最优解的能力 ,因此具有更强的鲁棒性和更灵活的计算尺度。试验结果表明 ,该算法是一种有效的随机优化算法。Building a model to get the boundary among different media, this paper mixed nonlinear planning technique(NLP) with the chaos seeking strategy and presented a hybrid optimal method, by which the obstacle that convex programming unavailable for Hesse matrix of target function does not satisfy with the condition of semi-positive definite can be removed. Both the ergodic property of chaos system and selectivity of NLP make the proposed method recognize the media boundary effectively and solve the nondestructive testing. Compared with previous ways, the proposed algorithm finds the optimal solution rapidly without calculating derived function and has no limitation of initial value selection, as avoid the stagnation phenomena in regular way and strengthen the ability of gaining the optimal solution, so it has better performance on robustness and flexibility. The experiment results show that the hybrid algorithm is an effective stochastic optimal method.

关 键 词：非线性规划 全局最优解 随机搜索 半正定 凸规划 迭代 收敛 进化算法 鲁棒性 混沌搜索 

分 类 号：TH132[机械工程—机械制造及自动化] O242.23[理学—计算数学]