Partner Selection Model on Bidding Alliance for BOT Projects Based on Modified Auto-adapted Ant Colony Algorithm

Bidding alliance partner selection is a key problem for bidding alliance formation, and it has the characteristics of multi-objective decision-making. In order to overcome the traditional partner evaluation´s weakness, such as single goal oriented or a single process oriented, 0-1 integer programming model for selecting multi-partner in multi-projects was constructed to minimize the comprehensive evaluation value, which was based on cost, financing, risks, and quality indexes. According to characteristics of this problem, principle of ant colony algorithm was analyzed. Based on basic ant colony algorithm, transition probability was improved and information coefficient was auto-adapted, which better realized the global optimization and fast convergence. Finally the numerical analysis showed this method was more accurate and could get better solution compared to genetic algorithm and the basic ant colony algorithm. This method could solve multi-objective decision-making problem, such as alliance partner selection.