Abstract
Genetic algorithms are widely used in various mathematical and real world problems. They are approximate metaheuristic algorithms, commonly used for solving NP-hard problems in combinatorial optimisation. Industrial scheduling is one of the classical NP-hard problems. We analyze three classical industrial scheduling problems: job-shop, flow-shop and open-shop. Canonical genetic algorithm is applied for those problems varying its parameters. We analyze some aspects of parameters such as selecting optimal parameters of algorithm, influence on algorithm performance. Finally, three strategies of algorithm – combination of parameters and new conceptualmodel of genetic algorithm are proposed.
Highlights
Canonical genetic algorithm is applied for those problems varying its parameters
We analyze some aspects of parameters
Three strategies of algorithm – combination of parameters and new conceptual model of genetic algorithm are proposed
Summary
Viename iš klasikiniu kombinatorinio optimizavimo apibrežimu [1] teigiama, kad kombinatoriniame optimizavime ieškoma objekto iš baigtines ar galimai baigtines aibes. Šis objektas paprastai yra sveikasis skaicius, poaibis, permutacija ar grafinestruktura, t.y. turime kombinatorinio optimizavimo problema P = (S, f ); kintamuju aibe X = {x1, . Dn; apribojimus tarp kintamuju; tikslo funkcija f , kuria reikia minimizuoti (maksimizuoti), kur f : D1 ×. . .×Dn → R+; visu galimu priskyrimu aibe S = {s = {(x1, v1), . (xn, vn)|vi ∈ Di , s tenkina visus apribojimus}. Tradiciniai lokalios paieškos metodai dauguma atveju yra neefektyvus, todel taikomi metaeuristiniai algoritmai Metaeuristiniai algoritmai yra apytiksliai ir paprastai nedeterminuoti. I klausima, kuria strategija taikyti gana vaizdžiai atsako viena iš „No free lunch“ teoremu [2], kuri teigia, kad vidutinis visu paieškos algoritmu visoms problemoms našumas yra lygus. Ideja yra ta, jog reikia naudoti tinkama algoritma tinkamai problemai
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
