基于证据理论的矿井小构造空间分析模型研究

发布时间：2018-06-05 05:49

本文选题：DS证据理论 + 基本概率赋值函数　；参考：《中国矿业大学》2017年硕士论文

【摘要】：矿井小构造不仅会严重影响煤矿安全生产,也是绝大部分煤矿地质灾害发生的诱发因素。而煤矿生产安全事故会给煤矿企业带来巨大的经济损失和人员伤亡。因此,加强对矿井小构造的分析,全面掌控矿井小构造的空间分布情况,实现小构造空间分析的定量化、精细化、实用化,对煤矿生产和灾害防治具有重要的理论意义和实用价值。论文针对目前矿井小构造分析预测方法单一、应用领域局限、实用性差等缺陷,根据矿井小构造空间分析问题的特点,综合提取趋势面分析法、曲面磨光法、定量评价等分析结果的构造特征,构建基于特征提取的矿井小构造基本概率赋值函数,并在此基础上构建矿井小构造空间分析识别框架,建立基于证据理论的矿井小构造空间分析模型。最后采用交叉检验法和实例证明模型结果的可靠性,并分析了引起模型不确定性的原因及相应对策。主要取得如下成果:(1)基于特征提取的基本概率赋值函数构造方法。根据矿井构造定量分析结果所呈现的构造空间分布基本特征,研究了提取这些特征的方法并将其转化为证据理论领域的基本概率,构造了基于特征提取的基本概率赋值方法。该方法是针对矿井构造空间特征专门定制的概率赋值方法,保证了矿井小构造空间多源信息在转化为概率时的完整性和精确性,从而确保模型应用的可靠性。(2)基于证据理论的矿井小构造空间分析模型。根据上述基本概率赋值函数所提取的多源信息,研究并构建了矿井小构造空间分析问题的识别框架,分析证据组合规则及证据冲突处理机制,在此基础上建立基于证据理论的矿井小构造空间分析模型,为矿井小构造空间分析奠定理论和技术基础。(3)模型可靠性分析。根据模型在基本概率赋值函数构造、模型构建和实例分析阶段可能产生的误差因素,通过实例应用和交叉验证法对这些过程进行误差分析,评价算法的可靠性,探索通过控制影响因素来实现改善和优化模型应用效果的对策。同时将模型的流程和算法编程实现,形成完整、实用的分析模型软件。
[Abstract]:Mine small structure will not only seriously affect the safety of coal production, but also the most of coal mine geological hazards induced factors. And the coal mine production safety accident will bring the huge economic loss and the personnel casualty to the coal mine enterprise. Therefore, strengthen the analysis of mine small structure, control the spatial distribution of mine small structure in an all-round way, realize the quantification, refinement and practicality of small structure space analysis, It has important theoretical significance and practical value for coal mine production and disaster prevention. Aiming at the shortcomings of the present mine small structure analysis and prediction method, such as single method, limited application field, poor practicability and so on, according to the characteristics of spatial analysis of mine small structure, the paper synthetically extracts trend surface analysis method, surface polishing method, etc. Based on the quantitative evaluation of the structural characteristics of the analysis results, the basic probability assignment function of mine small structures based on feature extraction is constructed, and the spatial analysis and identification framework of mine small structures is constructed. The spatial analysis model of mine small structure based on evidence theory is established. Finally, the reliability of the model results is proved by the cross test method and an example, and the causes of the uncertainty of the model and the corresponding countermeasures are analyzed. The main achievements are as follows: 1) the basic probability assignment function construction method based on feature extraction. According to the basic characteristics of spatial distribution of structures presented by quantitative analysis of mine structures, the methods of extracting these features are studied and transformed into the basic probabilities in the field of evidence theory, and a basic probability assignment method based on feature extraction is constructed. This method is a probability assignment method specially tailored to the spatial features of mine structures, which ensures the integrity and accuracy of the multi-source information in the small structure space when it is converted to probability. Thus ensuring the reliability of the model application. (2) the spatial analysis model of mine small structure based on evidence theory. According to the multi-source information extracted from the above basic probability assignment function, the identification framework of spatial analysis of mine small structures is studied and constructed, and the rule of evidence combination and the mechanism of dealing with evidence conflict are analyzed. On this basis, the spatial analysis model of mine small structure based on evidence theory is established, which lays a theoretical and technical foundation for the reliability analysis of mine small structure spatial analysis. According to the error factors in the basic probability assignment function construction, model construction and case analysis, the error analysis of these processes is carried out through case application and cross-validation, and the reliability of the algorithm is evaluated. This paper explores the countermeasures to improve and optimize the application effect of the model by controlling the influencing factors. At the same time, the process and algorithm of the model are programmed to form a complete and practical analysis model software.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD82

【参考文献】