An improved method for solving the large-scale multidimensional 0–1 knapsack problem

In order to improve the searching performance and the efficiency for solving the large-scale multidimensional 0–1 knapsack problem, an improved method based on genetic algorithm(IGA)which combines with pattern substitution and greedy algorithm is designed. The algorithm makes pattern substitution that the excellent genes replaces some relatively worse ones, then sorts those individuals through the improved greedy thought, finally optimizes the solution by maximum repair strategy after a series of genetic operation such as crossover and mutation. The experiment results show that IGA improves the ability of searching optimization and convergence speed compared with simple genetic algorithm and greedy method - genetic algorithm. On the problem of solving large-scale, IGA significantly improves the precision and stability compared with greedy algorithm.