Programação de horários usando um algoritmo genético especializado

• Ano de defesa: 2014
• Autor(a) principal: Atala, Ali Veggi [UNESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual Paulista (Unesp)
• Palavras-chave em Português: Programação heuristica; Algoritmos genéticos; GRASP (Sistema operacional de computador); Programação linear; Heuristic programming
• Link de acesso: http://hdl.handle.net/11449/111140

The problem with optimum schedule programming consists of programming class-events for a given number of classrooms and students, with the goal of satisfying certain factibility conditions. The problem is represented by a full mixed linear programming model and has been solved by use of a genetic Chu-Beasley algorithm, that presents population homogenization avoidance features, allowing for best solutions, modified in three core points: (i) initial population generation; (ii) local improvement phase; (iii) diversity increase. Additionally, a constructive algorithm for the initial phase of initial population generation and local search is presented, this allows for correction of possible unsuitabilities and improves population quality, such in initial phase as in local improvements. Population diversity and updating control is done according to parameters that assure each new individual has a different genes number in relation to the other individuals of the population, avoiding homogenization. The proposed method is applied for test cases of specialized literature, and data for the tests are presented by International Timebling Competition