煤矿员工安全行为评价及预警研究

发布时间：2018-08-22 17:46

【摘要】：众所周知,能源问题关乎国民经济命脉。国家"十三五"规划指出打造绿色煤炭能源势在必行。然而,据《中国煤炭工业年鉴》数据显示,近二十年的煤矿事故占全国工矿企业事故的25%左右,死亡人数占40%左右。不言而喻,造成这些事故的原因不能仅仅从单一层面加以断定,地方政府监督不严、企业经济利益驱使、矿工安全意识低下、安全技能欠缺、作业环境差、安全管理混乱等因素均在特定水平上导致矿难发生,但归根结底源于员工的行为。因此,针对复杂环境下的煤矿员工安全行为影响因素进行有效识别和预控成为政府及企业决策者亟需解决的难题。本文以煤矿一线员工安全行为为研究对象,基于煤矿安全生产的复杂性和系统性,研读国内外文献,剖析2001年至2016年典型煤矿事故案例,归纳整合员工安全行为影响因素,借助典型事故分析、实地调研以及问卷调查、行为事件访谈等,验证所提取影响因子的可靠性和科学性。在甄别煤矿员工安全行为影响指标的基础上,对指标进行优选及分析,量化指标层级结构,进而构建切实有效的煤矿员工安全行为评价指标体系。采用信息熵法,辨析计算煤矿员工安全行为各指标权重。接着借助5P神经网络的自学习、自适应能力,通过对淮南矿业集团、河南平煤矿业集团下辖的10个已知样本的学习,获取专家思维,采用训练好的网络仿真尚未测度的样本,有效缩减了人因在安全评价中影响程度;此外,通过训练好的网络还可以求得各指标相应的权重大小,进而根据权重值明晰指标对煤矿员工安全行为的影响程度。在此基础上,进一步明确煤矿员工安全行为预警机制,即:预警指标选取、预警体系构成、单一指标预警区间确定以及综合指标预警区间确定等。以此为基础,选用5P神经网络同遗传算法改进后的5P神经网络进行对比分析,结果表明:GA-5P的收敛速度与测算精度更加准确、有效。最后,依据预警分析与规避对策,以期实现煤矿安全预警管理模式的良好运行及员工安全行为的有效综合管控。
[Abstract]:As we all know, the energy problem concerns the lifeblood of the national economy. National "13 th five-year plan points out to build green coal energy is imperative." However, according to the data of China Coal Industry Yearbook, coal mine accidents account for about 25% of the accidents in China's industrial and mining enterprises in the past 20 years, and the death toll accounts for about 40%. It goes without saying that the causes of these accidents cannot be determined from a single level. Local governments are not strictly supervised, enterprises are driven by economic interests, miners have low awareness of safety, lack of safety skills, and poor working environment. Confusion in safety management and other factors all lead to mine accidents at certain level, but ultimately result from the behavior of employees. Therefore, it is a difficult problem for government and enterprise decision-makers to effectively identify and control the influencing factors of coal mine employees' safety behavior in complex environment. Based on the complexity and systematization of coal mine safety, this paper analyzes the typical coal mine accident cases from 2001 to 2016, and summarizes the influencing factors of integrating the safety behavior of the workers, based on the complexity and systematization of coal mine safety production. By means of typical accident analysis, field investigation, questionnaire investigation and behavior event interview, the reliability and scientificity of the factors extracted were verified. On the basis of discriminating the influence index of coal mine employee safety behavior, the index is selected and analyzed, and the index hierarchy structure is quantified, and an effective evaluation index system of coal mine employee safety behavior is constructed. The information entropy method is used to analyze and calculate the weights of coal mine employees' safety behavior. Then, with the help of self-learning and adaptive ability of 5p neural network, through the learning of 10 known samples under Huainan Mining Group and Henan Pingmei Mining Group, the expert thinking is obtained, and the trained network is used to simulate the unmeasured samples. The influence degree of human factor in safety evaluation is reduced effectively, in addition, the corresponding weight of each index can be obtained by the trained network, and then the influence degree of safety behavior of coal mine employees can be determined according to the weight value. On this basis, it is further clear that the early warning mechanism of coal mine employees' safety behavior, namely: early warning index selection, early warning system composition, single index early warning interval determination and comprehensive index early warning interval determination and so on. On this basis, 5p neural network is compared with the improved 5P neural network based on genetic algorithm. The results show that the convergence rate and the accuracy of the calculation are more accurate and effective. Finally, according to the early warning analysis and the circumvention countermeasure, the author hopes to realize the good operation of the coal mine safety early warning management mode and the effective comprehensive control of the safety behavior of the staff.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD79;F426.21

【参考文献】