Abstract:

Large engineering projects are often managed by project groups. There is a logical relationship between subprojects， and the project information is fully shared. At the same time， it is necessary to consider the overall optimization objectives of the project group and the optimization objectives of each subproject. Therefore， the project group scheduling problem is a special multi project scheduling optimization problem with practical application scenarios. When considering multi skilled human resources， the difference in the skill level of global resources and the change in the use of local resources will lead to the change of activity duration. Large projects tend to be completed and put into use as soon as possible， that is， project group managers usually need to consider how to reduce the project group duration； When the sub project is responsible for its own profits and losses within the budget， it often reduces the cost of the sub project as much as possible to protect its own interests， that is， the sub project manager usually needs to consider how to reduce the cost of the sub project. Therefore， the global and local objectives are to minimize the total duration of the project group and the cost of each sub project.To solve this problem， a global and local two-stage hierarchical scheduling optimization model is established. Under the global and local resource constraints， the project group manager establishes a global scheduling optimization model to minimize the total duration of the project group. Under the global resource constraints and sub project duration constraints given by the project group manager， the sub project managers establish a local scheduling optimization model with the goal of minimizing the sub project cost. To solve this problem， a two-stage solution mechanism based on idle global resource reallocation is proposed， and an adaptive improved multi population genetic algorithm is designed to solve the global scheduling problem.The experimental research was carried out based on Ran Gen randomly generated project group example set. According to the problem size， the example is divided into three problem subsets， and each problem subset contains two groups of project group examples generated by 27 different parameter combinations. The experimental results show that under the same problem scale， the stronger the global resource conflict intensity and the greater the network density of the project group， the longer the total duration of the project group， and the greater the local resource demand intensity， the greater the average cost of the sub project. The experiment further proves that the idle global resource reallocation mechanism designed in this paper can reasonably allocate the global multi skilled human resources and effectively reduce the cost of sub projects on the premise of ensuring the total duration of the project group. In addition， the idle resource reallocation strategy can provide a reference for project group managers when allocating project group resources.The research gap of project group scheduling problem is made up for considering multi skilled human resources， and the door for further research on project group scheduling problem considering human resources balance and multi skilled human resources project group scheduling problem under uncertainty is opened. Due to the multi skill heterogeneity of human resources and the different needs of activities， the working hours of employees tend to be unbalanced， which is not conducive to the use and management of human resources. Therefore， in the follow-up study， the work time balance will be considered for further research. At the same time， more efficient meta heuristic algorithm or artificial intelligence algorithm is explored to solve the problem， so as to further improve the optimization effect of project group scheduling problem.

Key words: project group scheduling, multi-skilled human resources, variable duration, multi population genetic algorithm