Clustered genetic algorithm to solve multidimensional knapsack problem

Genetic Algorithm (GA) has emerged as a powerful tool to discover optimal for multidimensional knapsack problem (MDKP). Multidimensional knapsack problem has recognized as NP-hard problem whose applications in many areas like project selection, capital budgeting, loading problems, cutting stock etc. Attempts has made to develop cluster genetic algorithm (CGA) via modified selection and modified crossover operators of GA. Clustered genetic algorithm consist of (1) fuzzy roulette wheel selection for individual selection to form the mating pool (2) A different kind of crossover operator which uses hierarchical clustering method to form two clusters from individuals of mating pool. CGA performance has examined against GA with respect to 30 benchmark problems for multidimensional knapsack. Experimental results show that CGA has significant improvement over GA in relation to discover optimal and CPU running time. The data set for MDKP is available at http://people.brunel.ac.uk/mas-tjjb/jeb/orlib/files/mknap2.txt.