In railway operation, unsafe events such as faults may occur, and a large number of unsafe event records are generated in the process of unsafe events’ recording and reporting. Unsafe events have been described in unstructured natural language, which often has inconsistent structure and complex sources, involving multiple railway specialties, with multisource, heterogeneous, and unstructured characteristics. In practical application, the efficiency of processing is extremely low, leading to potentially unsafe management utilization. Based on the data on unsafe events, this paper utilizes big data processing technology, conducts association rules mining and association degree analysis, extracts the word segmentation, and obtains the feature vector of unsafe fault event data. At the same time, the unsafe event data analysis model is constructed in combination with regular expression and pattern matching technology. This paper establishes the matching model of high-speed railway derailment-based external environment risk factors and applies it to the occurrence of unsafe events. This model could be utilized to analyze and excavate the link between external environment risk factors and the occurrence of unsafe events and carry out the automatic extraction of characteristic information such as risk possibility and consequence severity; hence, it has potential for identifying, with enhanced accuracy, high-risk factors that may lead to high-speed railway derailment. Based on this study, we could make full use of the unsafe event data, forecast the risk trend, and discover the law of high-speed railway derailment. This study introduces a viable approach to analyzing the unsafe event data, forecasting risk trend, and conceptualizing high-speed railway derailment. It could also enforce the accurate quantification of high-speed railway safety situation, refine the risk index and conduct in-depth analysis combined with the model, and effectively support the digitalization and intellectualization of high-speed railway operation safety.

中文翻译：

---

基于非结构化数据的高铁脱轨外部环境风险因素相关性分析

在铁路运营中，可能会发生故障等不安全事件，在不安全事件的记录和上报过程中会产生大量的不安全事件记录。非结构化自然语言描述了不安全事件，其往往结构不一致，来源复杂，涉及多个铁路专业，具有多源、异构和非结构化特征。在实际应用中，处理效率极低，导致潜在的管理利用不安全。本文基于不安全事件数据，利用大数据处理技术，进行关联规则挖掘和关联度分析，提取分词，得到不安全故障事件数据的特征向量。同时，结合正则表达式和模式匹配技术构建不安全事件数据分析模型。本文建立了基于高铁脱轨的外部环境风险因素匹配模型，并将其应用于不安全事件的发生。该模型可用于分析和挖掘外部环境风险因素与不安全事件发生之间的联系，进行风险可能性、后果严重程度等特征信息的自动提取；因此，它有可能以更高的精度识别可能导致高速铁路脱轨的高风险因素。基于此研究，可以充分利用不安全事件数据，预测风险趋势，发现高铁脱轨规律。本研究介绍了一种分析不安全事件数据、预测风险趋势和概念化高速铁路脱轨的可行方法。还可以加强对高铁安全形势的准确量化，细化风险指标，结合模型进行深入分析，有效支撑高铁运营安全数字化、智能化。