离散制造全流程在线动态调度数据

该数据集面向离散制造全流程在线动态调度任务建设，聚焦智能制造系统中复杂生产环境的实时响应与优化决策需求，针对调度算法性能对比、智能调度系统训练等研究与工程应用的数据支撑缺口，填补了动态调度场景下高质量方案与评估数据的空白，对提升调度算法优化效率、推动智能调度技术落地、保障生产连续性与资源利用率具有重要意义，可广泛服务于学术研究、工业技术研发及制造企业生产调度优化实践。
数据集并非来源于真实工厂运行日志，而是通过高保真仿真环境结合项目自主研发的三类核心优化算法生成 —— 基于多规则搜索算法、基于双向贪婪搜索算法、基于代理模型搜索算法。建模过程充分考虑任务动态到达、设备故障与维护中断、资源竞争与优先级调整、分布式车间协同等离散制造典型动态特性，确保数据的场景适配性与实用性。
数据集核心内容为三类典型调度场景的高质量调度方案与性能评估数据：分布式车间制造任务动态调度质量数据、车间不间断任务动态排班质量数据、分布式车间维保任务动态调度质量数据。数据包含调度方案解（记录任务分配状态、开始时间、资源编号等信息，维度由问题规模决定）与调度质量评估数据（含最大完工时间、总延迟、资源利用率等目标函数值），以 float32 精度的纯文本格式存储，采用简洁的文件结构，包含 bestSolution.txt（存储单目标最优调度方案集合）与 Pareto_Front.txt（存储多目标帕累托最优解集）两个核心文件，便于高效读取与批量处理。
数据体量方面，数据集总大小为 130KB，包含多组单目标最优解与多目标非支配解，覆盖不同调度场景与问题规模，数据维度可变且针对性强，既能支撑调度算法性能对比与可视化展示，也能满足智能调度系统训练与验证的需求。
该数据集完全公开共享，支持 C++、Python、MATLAB 等主流编程环境及文本解析工具读取，无特殊环境要求，为离散制造全流程在线动态调度研究提供了高质量、结构化的算法生成数据支撑，助力相关技术在生产调度实践中的优化应用。

This dataset is developed for online dynamic scheduling tasks throughout the entire discrete manufacturing process, focusing on the requirements of real-time response and optimal decision-making in complex production environments of intelligent manufacturing systems. Aiming at the data support gap in research and engineering applications such as scheduling algorithm performance comparison and intelligent scheduling system training, it fills the vacancy of high-quality scheduling schemes and evaluation data in dynamic scheduling scenarios. It is of great significance for improving the optimization efficiency of scheduling algorithms, promoting the practical application of intelligent scheduling technologies, and ensuring production continuity and resource utilization rate, and can widely serve academic research, industrial technology R&D, and production scheduling optimization practices of manufacturing enterprises.

This dataset is not derived from real factory operation logs, but is generated through a high-fidelity simulation environment combined with three core optimization algorithms independently developed by the project: multi-rule search algorithm, bidirectional greedy search algorithm, and surrogate model-based search algorithm. The modeling process fully considers typical dynamic characteristics of discrete manufacturing, such as dynamic task arrival, equipment failure and maintenance interruptions, resource competition and priority adjustment, and distributed workshop collaboration, to ensure the scenario adaptability and practicality of the data.

The core content of the dataset includes high-quality scheduling schemes and performance evaluation data for three typical scheduling scenarios: dynamic scheduling quality data of manufacturing tasks in distributed workshops, dynamic scheduling quality data of uninterrupted tasks in workshops, and dynamic scheduling quality data of maintenance tasks in distributed workshops. The data contains scheduling solution sets (recording information such as task allocation status, start time, resource number, etc., with dimensions determined by the problem scale) and scheduling quality evaluation data (including objective function values such as makespan, total tardiness, and resource utilization rate). It is stored in pure text format with float32 precision, and adopts a concise file structure, including two core files: bestSolution.txt (storing the set of single-objective optimal scheduling schemes) and Pareto_Front.txt (storing the multi-objective Pareto optimal solution set), which facilitates efficient reading and batch processing.

In terms of data volume, the total size of the dataset is 130 KB, containing multiple sets of single-objective optimal solutions and multi-objective non-dominated solutions, covering different scheduling scenarios and problem scales. The data dimensions are variable and highly targeted, which can not only support scheduling algorithm performance comparison and visualization display, but also meet the requirements of intelligent scheduling system training and verification.

This dataset is fully open and shared, and supports reading in mainstream programming environments such as C++, Python, MATLAB and text analysis tools, without special environment requirements. It provides high-quality, structured algorithm-generated data support for research on online dynamic scheduling throughout the entire discrete manufacturing process, and facilitates the optimized application of related technologies in production scheduling practices.