基于本体论的铁路风险关联知识发现研究

发布时间：2019-01-27 19:26

【摘要】：在传统的重载铁路运输风险预警预控研究中,往往注重风险与事故之间的线性因果关联,然而现实中事故致因机理是风险之间相互影响、聚合、升级最终形成的复杂情境,为了有效的减少事故发生频率、对风险源进行前期预控,有必要研究风险源之间的关联特性与关联模式。本文针对重载铁路的事故特点,结合文本挖掘与数据挖掘,基于本体论的知识发现和知识推理方法研究风险源关联特性,为风险源的前期预控提供精准判断。论文的主要研究内容如下:(1)铁路风险本体构建:基于历年的事故报告、描绘事故情境与事故致因机理,通过解析不同事故情境下各类(环境、设备、人员等)风险源的致因机理,以事故-风险源等知识重用为目的,构建初始的重载铁路风险本体模型。(2)铁路风险分析:针对重载铁路的风险隐患数据库(半结构数据与文本数据),以文本分析的手段提取半结构化数据中风险源关键词,提取可能影响事故发生的风险因素,并验证风险因素对事故发生的影响。将通过验证的风险因素添加到重载铁路风险本体中。(3)基于隐患数据库的数据挖掘:依据铁路事故升级具有滞后性特征,在传统的Apriori关联规则算法的基础上增加时序分析能力,使用改进的算法对国内某重载铁路公司的风险隐患数据进行挖掘分析,挖掘风险因素之间的关联模式,进一步解析铁路事故产生的潜藏致因机理,扩充调整基于事故分析的重载铁路风险本体模型。(4)铁路风险本体生成:提出一种面向重载铁路风险数据的重载铁路领域本体的半自动构建方法进行本体构建。半自动的构建方法,保障本体知识推理效率,以及构建本体的可靠性和领域知识的准确性和专业性,为基于结构化和半结构化数据源进行本体学习提供可行的方法。(5)铁路风险本体学习:基于风险关联知识,研究重载铁路风险本体知识推理,提出基于本体论的重载铁路知识建模方法;提出的铁路风险关联知识推理机制,促进了重载铁路风险知识应用,进而实现事故的提前预警准确性以及事中处理的有效性,同时也完成铁路风险知识的共享和重用,实现铁路风险关联知识的快速获取和维护进程。
[Abstract]:In the traditional research of early-warning and pre-control of heavy haul railway transportation risk, the linear causal relationship between risk and accident is often emphasized. However, in reality, the mechanism of accident cause is the interaction, aggregation and upgrading of the complex situation. In order to effectively reduce the frequency of accidents and pre-control the risk sources, it is necessary to study the correlation characteristics and patterns between the risk sources. According to the accident characteristics of heavy haul railway, combining text mining and data mining, this paper studies the risk source association characteristics based on ontology knowledge discovery and knowledge reasoning method, which provides accurate judgment for pre-control of risk source. The main contents of this paper are as follows: (1) the construction of railway risk ontology: based on the accident reports over the years, describe the accident situation and accident cause mechanism, analyze the various types (environment, equipment) under different accident situations, The causative mechanism of risk sources is aimed at the reuse of knowledge such as accident and risk sources, Constructing the initial heavy haul railway risk ontology model. (2) Railway risk analysis: aiming at the risk hidden trouble database (semi-structural data and text data) of heavy haul railway, the key words of risk source in semi-structured data are extracted by text analysis. Extract the risk factors that may affect the accident and verify the impact of the risk factors on the accident. The verified risk factors are added to the heavy haul railway risk ontology. (3) data mining based on hidden danger database: according to the characteristics of lag in railway accident upgrading, On the basis of the traditional Apriori association rules algorithm, the ability of timing analysis is added, and the improved algorithm is used to mine and analyze the hidden risk data of a domestic heavy-haul railway company, and to mine the association patterns between risk factors. To further analyze the underlying cause mechanism of railway accidents, Expand and adjust the heavy-haul railway risk ontology model based on accident analysis. (4) Railway risk ontology generation: a semi-automatic ontology construction method for heavy-haul railway domain ontology for heavy haul railway risk data is proposed. The semi-automatic construction method ensures the efficiency of ontology knowledge reasoning, the reliability of ontology construction and the accuracy and professionalism of domain knowledge. It provides a feasible method for ontology learning based on structured and semi-structured data sources. (5) Railway risk ontology learning: based on risk correlation knowledge, the heavy haul railway risk ontology knowledge reasoning is studied. An ontology based modeling method for heavy haul railway knowledge is proposed. The proposed reasoning mechanism of railway risk association knowledge promotes the application of heavy-haul railway risk knowledge, and realizes the accuracy of early warning of accidents and the effectiveness of in-process handling. At the same time, it also accomplishes the sharing and reuse of railway risk knowledge. To realize the rapid acquisition and maintenance of railway risk related knowledge.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U298;TP311.13

【参考文献】