Improved Hybrid Differential Evolution-Estimation of Distribution Algorithm with Feasibility Rules for NLP/MINLP Engineering Optimization Problems  

基于差分进化和分布估计的改进混合算法在NLP及MINLP工程优化问题中的应用(英文)

作　　者：摆亮 王钧炎 江永亨 黄德先 

机构地区：[1]Department of Automation,Tsinghua University [2]National Laboratory for Information Science and Technology,Tsinghua University [3]Marvell Technology (Shanghai) Ltd,Shanghai 201203,China

出　　处：《Chinese Journal of Chemical Engineering》2012年第6期1074-1080,共7页中国化学工程学报（英文版）

基　　金：Supported by the National Basic Research Program of China (2012CB720500);the National Natural Science Foundation of China (60974008)

摘　　要：In this paper, an improved hybrid differential evolution-estimation of distribution algorithm (IHDE-EDA) is proposed for nonlinear programming (NLP) and mixed integer nonlinear programming (MINLP) models in engineering optimization fields. In order to improve the global searching ability and convergence speed, IHDE-EDA takes full advantage of differential information and global statistical information extracted respectively from differential evolution algorithm and annealing mechanism-embedded estimation of distribution algorithm. Moreover, the feasibility rules are used to handle constraints, which do not require additional parameters and can guide the population to the feasible region quickly. The effectiveness of hybridization mechanism of IHDE-EDA is first discussed, and then simulation and comparison based on three benchmark problems demonstrate the efficiency, accuracy and robustness of IHDE-EDA. Finally, optimization on an industrial-size scheduling of two-pipeline crude oil blending problem shows the practical applicability of IHDE-EDA.In this paper, an improved hybrid differential evolution-estimation of distribution algorithm （IHDE-EDA） is proposed for nonlinear programming （NLP） and mixed integer nonlinear programming （MINLP） models in engineering optimization fields. In order to improve the global searching ability and convergence speed, IHDE-EDA takes full advantage of differential information and global statistical information extracted respectively from differential evolution algorithm and annealing mechanism-embedded estimation of distribution algorithm. Moreover, the feasibility rules are used to handle constraints, which do not require additional parameters and can guide the population to the feasible region quickly. The effectiveness of hybridization mechanism of IHDE-EDA is first discussed, and then simulation and comparison based on three benchmark problems demonstrate the efficiency, accuracy and robustness of IHDE-EDA. Finally, optimization on an industrial-size scheduling of two-pipeline crude oil blending problem shows the practical applicability of IHDE-EDA.

关 键 词：differential evolution estimation of distribution hybrid evolution mixed-coding feasibility rules 

分 类 号：TB114.1[理学—运筹学与控制论]