Three pseudo-utility ratio-inspired particle swarm optimization with local search for multidimensional knapsack problem

In this study, a three-ratio self-adaptive check and repair operator-inspired particle swarm optimization (3R-SACRO-PSO) with neighborhood local search is developed to solve the multidimensional knapsack problem (MKP). The proposed 3R-SACRO-PSO systematically alters substitute pseudo-utility ratios as the PSO method is executed. In addition, a local search scheme is introduced to improve solution quality. The proposed 3R-SACRO-PSO algorithm is tested using 168 different widely used benchmarks from the OR-Library to demonstrate and validate its performance. The control parameters for the performance test are determined through the Taguchi method. Experimental results parallel those of other PSO algorithms, and statistical test results show that the quality and efficiency of the proposed 3R-SACRO are better than those of the two-ratio SACRO method. Moreover, the proposed 3R-SACRO-PSO is on par with state-of-the-art PSO approaches. Thus, introducing the third pseudo-utility ratio into SACRO improves the performance of SACRO-based PSO. The neighborhood local search scheme further improves the solution quality in handling MKPs.