Using Hybrid Genetic Algorithms to Solve Discrete Location Allocation Problems with Rectilinear Distance

This paper presents a hybrid genetic algorithm to solve the uncapacitated location allocation problems as a combinatorial optimization problem. The proposed method incorporates a modified K-means algorithm that clusters the customers into groups based on the rectilinear distance, and then the initial population of solutions is calculated according to the derived centers of clusters. The hybrid genetic algorithm consists of two evolutionary processes synchronized with each other. It uses the elite strategy, local search, multi-start mechanism and distance-based mutation to efficiently obtain the optimal or approximated solutions. In order to verify the effectiveness of the proposed method, four GAs -KGA, HGA, LMGA and TGA, are built and compared. Furthermore, using six commonly used test problems, the performance of the proposed method is evaluated relative to the Simulated Annealing method [4] and the SOFMGR method [14] as benchmarks. Experimental results show that the proposed method is excellent in quality of solutions and speed of computation.