Study on Multi- Knapsack Problem Based on Improved Artificial Fish School Algorithm

Multi-knapsack problem (MKP) is one of the typical NP problem in optimization field, because the computing complexities of the traditional algorithms are high or they have slow convergence speeds, this paper proposes an improved artificial fish school algorithm (IAFSA) for MKP. Firstly, the integer coding was used in the input ways of knapsacks, secondly, IAFSA adopted the strategy of "random repair" to repair infeasible artificial fish coding and inadequate artificial fish coding, and the coding after taking behaviors were improved and repaired, finally, IAFSA was verified by an example. Experimental results show that the advantage of IAFSA over other AFSA algorithms mainly lies in its faster convergence speed, stronger robustness, and its convergence to the optimal solution with greater probability.