关键词: 空间受限, 资源约束型项目调度, 材料采购, 集成优化模型, 双层启发式算法

Abstract:

Due to the large quantity and high strength of material demand， prefabricated building projects require a large amount of material storage space， but the storage space at the construction site is usually extremely limited. The camped storage space greatly deteriorates the contradiction between the large demand for construction materials and the small storage space of materials. Meanwhile， the limited storage space not only sharply restricts the parallel execution of activities， but also greatly increases the order times of materials， thus worsening the project performance. In order to solve this practical dilemma， the integrated resource constrained project scheduling and material ordering problem with limited storage space （RCPSMOP-LSS） is studied.Considering the renewable resources， precedence constraints， non-renewable resource constraints， limited storage constraints and dynamic inventory updating formula， an integrated optimization model of resource-constrained project scheduling and material ordering with limited storage space is proposed to minimize the total project cost consisting of the material ordering cost， inventory cost and the cost associated with the completion time. In order to obtain a project scheduling plan and a material ordering plan， the integrated model contains two types of decision variables： finish time of each activity for project scheduling part and quantity of each material at each time period for material ordering part， which greatly increased the complexity of the model.In order to solve the model， the properties of the model are firstly analyzed. It is found that the model is strongly NP-hard and needs to be solved in two stages. Then， a double-layer heuristic algorithm is developed. To be more specific， an improved genetic algorithm is firstly designed on the outer layer to obtain the project scheduling plan. Besides， a novel chromosome representation， and mutation operator is developed according to characteristics of solution space. Then， the activity schedule obtained in outer layer is considered as an input of the inner algorithm. By analyzing the nature of the problem， the material ordering subproblem is modeled as shortest path model， which considerably reduces solving difficulty. Furthermore， an exact algorithm is presented on inner layer to obtain the optimal materials ordering plan under the specific activity schedule.To evaluate the effectiveness of the proposed model and algorithm， large scale numerical experiments are carried out based on the instances generated by ProGen and selected from PSPLIB. The orthogonal experiment method is designed to determine the appropriate parameter sets for the proposed algorithm. Besides， to prove the validity of the integrated problem， a comparative experiment of decentralized decision-making and the proposed integrated model is set up， and the experimental result shows that integrated model can reduce the project cost by 12%. Furthermore， the comparisons with the Cplex software show that the proposed algorithm has better computational efficiency.From the perspective of limited storage space， the overall cost of project is optimized by integrating project scheduling and material ordering， which provides project managers with more comprehensive decision support on construction projects with congested site space. Moreover， reference value is provided for the research on project scheduling with limited space.

Key words: limited space, resource-constrained project scheduling, material ordering, integrated optimization model, double-layer heuristic algorithm