Abstract:

In order to apply the fireworks algorithm to discrete optimization and solve the multi-dimensional knapsack problem effectively, an binary opposite backward learning fireworks algorithm is designed. Firstly, on the basis of defining binary string distance and binary convert operator, fireworks explosion and mutation explosion are discretized to build the binary fireworks algorithm. Secondly, incomplete binary backward operator is designed to build the binary opposite backward learning fireworks algorithm, and its convergence is proved. Finally, compared with several typical evolutionary algorithms, simulation on 10 typical benchmark instances is analyzed. Results show that the binary opposite backward learning based fireworks algorithm has excellent performance on convergence rate, optimization accuracy and robustness.