Genetic Diversity in the Multiobjective Optimization of Paths in Graphs

摘要

Existing systems that allow Users to plan the route, usually do not support the multiple criteria during the search process. In the sense of the multi-criteria optimization, such a situation involves the search for the Pareto-optimal solutions, but the present services use a weighted sum of the supported criteria. For solving Multiobjective Shortest Path problem we incorporate genetic algorithms with modified genetic operators, what allows the reduction of the search space. In this paper we compare genetic diversity in the algorithms which incorporate our method. Conducted research shown that proposed modifications allowed to obtain better diversity without either changing parameters or apply some rules to the algorithm.