Development of a methodology for agent based schedule optimization in construction considering resource constrained process duration

在本项目中，引入了一种基于离散事件代理的新方法来提高建筑项目的生产率。考虑到所有的约束条件，根据不同的目标生成优化的项目时间表。建设项目的特点是参与者数量庞大，内部和外部相互依赖。考虑到所有约束条件，人工调度达到了极限。因此，要么将较大的缓冲时间添加到项目计划中，要么不考虑基本约束。在合理的时间内，几乎不可能精确地解决这个np困难的问题，即在一般规模的项目计划中最小化施工完工时间——现有的方法通常只考虑可能的优化的单个片段。因此，创建了一个多主体系统（MAS），该系统专门用于根据各自的目标生成和优化项目进度。除了确定最短工程工期目标外，还应以最少的资源部署按时完成所有建设任务，以及根据使用的特点、数量和时间分配资源所产生的最低成本。流程代理在一个中心市场（协商协议）上请求建议，所有代理可以在考虑现有约束的同时就资源分配进行协商。根据优化的主题，将可用资源分配到关键工序，并分配相同的资源，以确保理想的施工流程。智能、自主代理的应用揭示了项目调度优化的进一步潜力。在现有的启发式方法中，资源是在完成先前的任务后立即分配的，因此不会为可能更相关的过程保留资源。分析了可变资源分配对过程持续时间和各自运营工作链的影响，以及对整个项目的影响。建筑工程受到各种各样的影响，这些影响是难以预测的，但对工程计划的质量有负面影响。因此，时间表需要根据建筑工程的当前进度，通过整合和重新优化时间表来进行迭代修订。在多代理系统中，各种自治实体的合作和协调允许管理所有约束和考虑相关目标的资源分配。结果是在动态多代理框架中应用基于知识的实体的方法，以确定包含相关约束和资源特定过程持续时间的优化建设项目时间表。

Within this project a new approach based on discrete event agents is introduced to increase productivity of construction projects. Considering all constraints optimized project schedules are generated depending on various objectives.Construction projects are characterized by a huge number of participants and a variety of internal and external interdependencies. Considering all constraints manual scheduling reaches its limits. Hence either large buffer times are added to the project plan or essential constraints are not taken into account.An exact solution to this NP-hard problem of minimizing construction makespan for common sized project schedules is hardly possible within a reasonable amount of time - existing approaches usually consider only single fragments of the optimization possible. Therefore a Multi-Agent-System (MAS) is created, which is specialized in generation and optimization of project schedules due to the respective objectives. Besides the determination of minimum project duration objectives are minimal resource deployment to complete all construction tasks in time as well as minimal costs incurred by resource assignment depending on characteristics, numbers and time of utilization.Process agents request proposals on a central market place (negotiation protocol), where all agents may negotiate on resource allocation whilst considering existing constraints. Depending on the subject of optimization available resources are allocated to critical processes and identical resources are distributed to ensure ideal construction flow.The application of intelligent, autonomous agents reveals further potential for the optimization of project scheduling. In existing heuristic approaches resources are allocated immediately after the completion of prior tasks and will therefore not be reserved for potentially more relevant processes. Moreover the influences of variable resource allocation on process duration and respective operational work chains as well as corresponding consequences on the overall project are analyzed.Construction works are subject to various influences, which are difficult to anticipate but have negative impact on the quality of project plans however. Hence schedules require iterative revision by integrating and re-optimizing the schedule based on the current progress of construction works.Cooperation and coordination of the various autonomous entities in a Multi-Agent-System allows the management of all constraints and the allocation of resources considering relevant objectives. Result is a methodology for the application of knowledge-based entities in a dynamic Multi-Agent-framework to determine optimized construction project schedules comprising relevant constraints and resource specific process durations.