Orthogonal And Hybrid Of Chemical Reaction Optimization Algorithm And Its Application Research

Optimizat ion proble ms genera lly exist in sc ient ific research and engineering projects.When so lving the comp lex proble ms wit h high dime ns ions,the conve nt iona l optimizat io n methods are unable to sat isfy the require ments about accuracy of solut io ns and convergence speed.Intelligent optimizat ion algor ithms are designed by simulat ing the natura l phenome na or the princip le of organis m,they have good adaptive capacity to enviro nment t hough the y are derived from rando m searches.Owe to the characters like global,paralle l optimizat io n of high perfor mance,no proble m specific infor mat ion and robustness,they are wide ly used in computer scie nce,productio n mana ge ment,transportatio n proble m and other important fie lds.Domest ic and foreign researchers have paid muc h attent io n to the m.More and more mode rn optimizat io n algor it hms emerge,the improve me nt and applicat ion of the existed algor it hms are becoming a research focus.How to design a precise,fast converged and robust intelligent algorithm is still the researc h emphasis.In this thesis,some introduct ion work about the optimizat ion prob lems and algor it hms is done,the research ma inly focuses on chemica l reaction optimizat ion algor it hm.We proposed two improved algor ithm,OCRO and HGCRO.OCRO has been tested on so lving global numer ical optimizat ion proble ms,and HGCRO has been applied to solve the heterogeneous task scheduling proble ms.The ma in contents of this paper include:(1)Review o f opt imizatio n proble ms and algorithms.The research background,significa nce and status of the m are summar ized.We not only ana lyzed the drawbacks of conve nt iona l optimizat ion met hods,but a lso introduced the theoretica l princ ip le of genetic algorit hm and chemical reaction optimization algorithm in detail.(2)Analyzed the opt imizing process of che mical reactio n optimizat ion a lgor it hm,and designed an improved met hod aimed at its poor global search abilit y.The orthogona l exper imenta l design method is robust,whic h possesses systemat ic reasoning ability.Thus orthogona l crossover operator is emp lo yed to enhance the global search mec hanis m of c he mical reactio n optimizat ion a lgorit hm,so that the algor it hm can conver ge quick ly to the particular promis ing re gions where the global optima are more likely to reside.Meanwhile,two local search operators in origina l chemica l react ion opt imizat ion algorit hm still serve as loca l searches,whic h is he lpful for the algori thm to find the global optima quickly.(3)Based on the analys is about che mical reaction optimizat ion a lgor it hm,another improved method HGCRO is proposed.It combines the global search ability o f genet ic a lgor it hm wit h local search ability o f che mica l reaction opt imizat ion algor it hm.The hybrid algorithm's effic ienc y is ver ified by the successful applicat ion to heterogeneous task scheduling problems.