近年来，随着社会经济的快速发展和城镇化建设的持续推进，城市消防救援需求逐年攀升，处置任务也由单一火灾扑救向灭火、救援与救助并重的“全灾种、大应急”方向转变。在这样的时代背景下，从“自然—人为”两个方面的诸多维度着手，探索消防警情在时间和空间上的分布规律及其影响因素，更精细地识别消防安全风险，实现消防资源的优化配置，是统筹开展高效应急救援工作的科学基础。 本文基于火警和救援救助等细分城市消防警情历史数据，采用机器学习与时空统计分析方法，从多个尺度揭示了各类消防警情的时空异质分布规律，建立了精细化的消防警情与自然气象—居民活动等多源因素的关联关系，发展了城市详细分区的消防风险评估模型，并进一步提出了区域内消防救援站的多点拓扑选址方法。主要研究内容和成果如下： （1）运用多尺度的时空分析方法，揭示了消防警情的演变规律 基于广度性质的气象因素，构建了累积效应指标，通过我国3个大型城市的实证对比分析，发现了气象因素对消防警情的累积影响规律；建立了多尺度地理加权模型，发现居民活动对消防警情的影响具有时空异质性；构建了预测消防警情的经验公式，发现影响消防警情的多参数气象组合影响因素。 （2）融合多维自然—人为因素，建立了其与消防警情之间的关联关系 提出了基于气象修正的消防警情时序模型，从时间维度提高了消防警情发生数量预测的准确性；基于POI要素构建了城市活力多样性指标，基于访客流量要素构建了人群活力波动性指标，将二者空间加权融合，构建了城市综合活力指标；提出了自然—社会因素的多维时空多源异构数据融合方法，进一步从时间和空间两个维度发展了城市详细分区的消防风险评估方法。 （3）结合城市路网整合度计算，提出了消防救援站多点拓扑选址方法 考虑早晚高峰和平峰的影响，改进了消防救援站覆盖面积计算方法；统筹考虑火警与救援救助需求，改进了消防救援站选址依据；基于路网整合度提出了多点拓扑选址方法，给出了客观约束条件下的最优可行解集合，改进了传统位置—分配模型；提出了战勤保障消防救援站的选址方法。 上述融合了多尺度—多维度的精细化城市消防警情及影响因素分析，是对现有大尺度单维度城市和区域消防警情分析工作的重要发展，适应我国城市化进程中消防救援多样化复杂化的演变趋势，符合消防能力建设从增量向提质转变的方向，能够为我国城市消防救援能力规划、建设与更新等提供依据。

With rapid socio-economic development and urbanization, the demand for urban emergency services has increased annually. In addition, there has been a shift from firefighting to a "multi-hazard, comprehensive emergency" direction with equal emphasis on firefighting and technical rescue. Therefore, effective emergency services must be established by scientifically investigating the distribution of emergency services and their spatio-temporal influencing factors, identifying fire safety risks comprehensively, and ensuring the optimal allocation of natural and artificial firefighting resources. Based on historical data on subdivided urban emergency services, such as fire suppression and technical rescue, this study reveals the heterogeneous distribution of various emergency services across multiple scales using machine learning and spatio-temporal statistical analysis methods. In addition, this study establishes correlations between emergency services and multi-source factors, such as natural weather and resident activities, develops a fire risk assessment model for detailed urban zoning, and proposes a multipoint topological site selection method for local fire stations. The main contributions of this study are as follows: (1) The multi-scale spatio-temporal analysis approach reveals the influence of meteorological factors on emergency service evolution. A cumulative effect model was developed based on the breadth of meteorological factors and investigated the effect of meteorological factors on emergency services through a comparative empirical analysis of three megacities in China. For this, a multi-scale geographically weighted model was designed, and spatio-temporal heterogeneity of social activities on emergency services was characterized. An empirical formula for predicting the number of emergency services was constructed, and meteorological factors affecting these services were identified as a combination of temperature and water vapor pressure. (2) Multi-dimensional natural and human factors were incorporated and correlated with emergency services. The proposed time-series model of emergency services based on meteorological modifications improved the emergency service prediction accuracy in the temporal dimension. The urban vitality diversity and crowd vitality volatility indices based on the point of interest (POI) and visitor throughput factors was constructed respectively. The two indices were spatially weighted to construct a comprehensive urban vitality index. Moreover, a multi-dimensional spatio-temporal heterogeneous method for fusing data from natural-social factors and a fire risk assessment method for detailed urban zoning in spatio-temporal dimensions were developed. (3) Combined with the urban road network closeness, a multipoint topology method for fire station planning is proposed. The fire station coverage area was calculated considering the influences of morning and evening traffic peaks and flat traffic peaks. The fire station planning principle was modified considering firefighting and technical rescue demands. A multipoint topology planning method is proposed based on the road network proximity. Moreover, an optimal set of feasible solutions under objective constraints enhanced the traditional location-allocation model. Finally, a planning method for a fire station providing logistic support is proposed. Such an advanced analysis of urban emergency services and their influencing factors, integrating multi-scale and multi-dimensional factors, is important for understanding large-scale single-dimensional urban and regional emergency services. Moreover, it adapts to the evolving trend of diversification and complexity of emergency services in Chinese urbanization and changes in the fire safety standards from incremental to qualitative. Furthermore, it provides a basis for planning, constructing, and updating urban emergency service capacity in China.