Distributed Event-driven Scheduling Benchmark (DESBench)
收藏资源简介:
DESBench是由浙江大学研究团队创建的分布式事件驱动调度基准数据集,旨在为多智能体系统在复杂工业调度场景中的协调机制提供结构化评估平台。该数据集构建于共享离散事件仿真环境之上,模拟了具有层次结构、部分可观测性和动态耦合约束的工厂调度场景,涵盖了作业、机器、缓冲区、运输路径等多种实体及资源约束。数据集的创建过程基于离散事件驱动模型,通过定义任务实例、协调范式和评估指标,系统化地生成了用于评估集中式、分层式、异构式和整体式四种典型协调范式的仿真环境。该数据集主要应用于多智能体协调研究领域,旨在深入探究不同协调范式在动态演化环境中的性能表现与内在权衡,为解决工业调度中因资源耦合和延迟反馈所引发的复杂协调问题提供实证基础。
DESBench is a distributed event-driven scheduling benchmark dataset developed by the research team from Zhejiang University, aiming to provide a structured evaluation platform for coordination mechanisms of multi-agent systems in complex industrial scheduling scenarios. Built on a shared discrete-event simulation environment, this dataset simulates factory scheduling scenarios with hierarchical structures, partial observability, and dynamic coupling constraints, covering various entities and resource constraints such as jobs, machines, buffers, and transportation routes. Constructed based on a discrete-event-driven model, the dataset systematically generates simulation environments for evaluating four typical coordination paradigms—centralized, hierarchical, heterogeneous, and holistic—by defining task instances, coordination paradigms, and evaluation metrics. This dataset is primarily applied in the field of multi-agent coordination research, aiming to deeply explore the performance and inherent trade-offs of different coordination paradigms in dynamically evolving environments, and provide empirical foundations for solving complex coordination problems caused by resource coupling and delayed feedback in industrial scheduling.

- 1When Does Hierarchy Help? Benchmarking Agent Coordination in Event-Driven Industrial Scheduling浙江大学 · 2026年



