An improved particle swarm optimization algorithm for winner determination in multi-attribute combinatorial reverse auction

An electronic reverse auction system with one buyer and multiple suppliers is considered in this paper. The buyer procures multi-items from potential suppliers with unconstrained capacity and the suppliers bid competitively on combinations of items in the system. As an important decision problem from the buyer's perspective, a winner determination problem (WDP) of multi-items single-unit combinatorial reverse auction with multi-attributes of each item is described and a bi-objective programming model that minimizes the total procurement cost and maximizes the total score of the winning suppliers based on multi-attributes of each item is established. According to the characteristics of the model, an equivalent single-objective programming model is obtained. However, as the problem is NP-hard, an improved particle swarm optimization (IPSO) algorithm embedded with the quantum-inspired evolutionary and the asynchronous time-varying learning strategies is proposed. Also, a heuristic search algorithm is applied to repair the infeasible solutions in the process of IPSO. Experimental results show the effectiveness of the improved algorithm.