大数据驱动下基于灾民特性的重大自然灾害应急物资精准配置研究

The purpose of this research is to explore the theory and method, which is under the driving force of big data, of the accurate allocation of large-scales natural disasters emergency materials, based on the characteristics of victims. Its characteristic is to breakthrough the limitations of the previous studies on characteristic of victims; Adopting the advantages of big data in terms of effective information extraction and information analysis; to improve the accuracy for the demand prediction made by the victims and the motivation to solve the disaster; to investigate the data mining algorithm of non-deterministic disaster information and the optimization of the busy route selection method in term of emergency material. Consequently solve the problem that emergency materials allocation dose not victim demand. In turn to allocate emergency materials accurately, this provides the basis for directional distribution of emergency suppliers. The research content includes: Explore measurement indicators and analysis method for the characteristics of victims in large-scale natural disasters; Online disaster events real-time detection and deep digging method for disaster information; Forecast emergency materials’ urgent demand in the characteristics of victims and demand digging method under big data-driven; Build busy route scheduling optimization model in the characteristics of victims under the driving force of big data technology; Establish emergency materials accurate allocation mode in the characteristics of victims by the application of big data technology; Propose the strategic selection and policy designation for emergency materials accurate allocation in the characteristics of victims by applying big data. This study has original theoretical significance on establishing and improving knowledge system of large-scale natural disasters emergency materials allocation. The results from this research can contribute to the life-cycle management of disaster emergency events, offering support for large-scale natural disasters precision governance and target policy in government level.

本研究目的是探索大数据驱动下基于灾民特性的重大自然灾害应急物资精准配置的理论与方法，其特色在于：突破以往对灾民特性研究不足的局限，运用大数据高效的信息抽取和分析技术，提高灾民需求预测的精准性和治理的主动性；探究非确定性灾情信息的数据挖掘算法和应急物资精准占线调度优化路径选择方法，有效解决应急物资配置与灾民需求不匹配问题，为实现应急物资的精准配置提供依据。主要研究内容包括：重大自然灾害灾民特性的测度指标和分析方法；在线灾害事件爆发检测和灾情信息深度挖掘方法；大数据驱动下基于灾民特性的应急物资紧急需求预测和深度挖掘方法、应急物资精准占线调度优化模型、应急物资精准配置模式、策略选择和政策设计。本研究对于建立和完善重大自然灾害应急物资精准配置的知识体系具有理论原创意义，可应用于灾害应急管理的整个生命周期之内，为将来上升到国家层面的重大自然灾害政府精准治理和靶向施策提供支持。

本研究旨在探索大数据驱动下基于灾民特性的重大自然灾害应急物资精准配置的理论与方法，主要研究内容包括：重大自然灾害灾民特性的测度指标和分析方法；在线灾害事件爆发检测和灾情信息深度挖掘方法；大数据驱动下基于灾民特性的应急物资紧急需求预测和深度挖掘方法、应急物资精准占线调度优化模型、应急物资精准配置模式、策略选择和政策设计。该项目共发表论文21篇，申请专利2项，IEEE行业标准1项，其中在SCI Q1期刊《Risk Analysis》发表1篇。该项目的研究结果如下：①构建了重大自然灾害灾民特性的测度指标。②提出了重大自然灾害灾民特性的分析方法。③在应急物资需求预测方面，针对震灾初期需求信息紧迫且难以获取的实际情况，提出一种新的更适用于实际灾情需求的预测方法和模型。④在应急物资调度优化决策研究方面，探讨了基于灾情信息特征的两大类调度优化决策问题，即完备灾情信息和非决策灾情信息情况下的调度优化决策。应用机器学习和面向数据库相结合的数据挖掘方法，构建考虑灾民特性的基于模糊灾情信息的调度优化决策模型，改进了经典Apriori算法。⑤提出了基于确定性供需信息的兼顾效率与公平的应急物资多周期分配优化模型。⑥提出了基于不确定灾情信息的应急物资多周期分配优化模型。⑦提出从“多周期整体优化”视角，分析应急物资多周期分配优化模型的有效性，探究多周期分配优化的提升路径。⑧设计大数据驱动下基于灾民特性的重大自然灾害应急物资精准配置模式，提出重大自然灾害应急物资精准配置的策略选择及政策设计。本研究科学意义在于：突破以往对灾民特性研究不足的局限，提高灾民需求预测的精准性和治理的主动性；探究非确定性灾情信息的数据挖掘算法和应急物资精准占线调度优化路径选择方法，有效解决应急物资配置与灾民需求不匹配问题，为实现应急物资的精准配置提供依据。对于建立和完善重大自然灾害应急物资精准配置的知识体系具有理论原创意义。