Abstract: Aiming at solving the problem to match orders with inventory surplus in an iron and steel plant, an optimal multi-objective 0-1 programming model is established to maximize the utilization of material surplus on inventory and minimize the matching cost orders. Two objective functions are incorporated by fuzzy decision-making approach and the model is solved by improved PBIL (Population-based increased learning). Natural number encoding is used to represent the result of orders matching based on the model's characteristic with the impractical chromosomes repaired in terms of learning probability. Then, the computation of a practical instance and a comparison of the computational result with the result by genetic algorithm further demonstrate that the model and the algorithm are the ideal way to solve the optimal problem of order matching.