关键词: 重复性建设项目, 资源转移成本, 序贯博弈, 非支配排序, 改进遗传算法

Abstract: A vast majority of infrastructure construction projects are called repetitive construction projects (RCP) because of their wide distribution and high repeatability. Due to dispersed locations, a large number of resources involved, the difficulty with transportation, and the time and cost of resources transfer between units will significantly affect the results of RCP. However, current studies seldom consider the cost of resources transfer, and the way to determine the path of resources transfer is relatively simple. Especially most of RCP regard the working group as a basic resources allocation unit, which fails to carry out plans or pays high idle costs.
This paper studies a resources-constrained multi-mode repetitive construction project scheduling problem with an integrated resources transfer cost and tries to find a combination of construction sequence and resources transfer path to minimize the total cost. The idea of resources allocation based on the working group is changed. The resources transfer cost and idle cost caused by resources redundancy are fully considered to satisfy the resources demand of different units. The quantity and path of resources transfer are determined. An improved genetic algorithm (AEIAGA) is designed to solve the problem based on the self-adaptive elite retention and mutation strategy and the crossover strategy based on the iteration number and fitness value.
In the simulation analysis, taking the lower part of a bridge as an example, we find that: 1.The scheme that does not consider the actual resources transfer cost is quite different from the actual situation, and the plan needs to be significantly adjusted to meet the actual project needs. Compared with scenario 1 and scenario 2, the scheduling results of most of the units are adjusted to meet the actual project requirements. The overall construction period changes from 87 days to 131 days, increasing by 44 days or 50.57%, and the fee increases from 803,758 yuan to 922,544 yuan, by 118,786 yuan or 14.78%. 2.The introduction of resources transfer cost will have a more significant impact on the construction sequence, and the consequences are more inclined to transfer resources between units with a lower transfer cost. In scenario 1, the sequence of resources transfers is relatively chaotic and there is a large amount of cross-cell transfer behavior because of the small impact of resources transfers; in scenario 3, considering the influence of resources transfer, most of resources transfer processes take place between adjacent units. 3.Considering the actual cost of resources transfer, a hybrid strategy can effectively reduce the project time and total cost. Compared with scenario 2, the number of resources transfers in scenario 3 is reduced from 40 to 35, resulting in a reduction of 36-day by 27.48%, and a reduction of 111,130 yuan by 12.05% in cost, so the overall scheme has been significantly optimized. 4. The improvement effect of the genetic algorithm in this paper is remarkable, especially in solving large-scale problems, so the algorithm has obvious advantages. When dealing with large-scale problems, the difference between AEIAGA and AEIAGA begins to widen, and the mean value of the solution is 1.039 times and 1.028 times of AEIAGA, respectively.

Key words: repetitive construction project, resources transfer time cost, sequential game, non-dominant sort, improved genetic algorithm