基于因子分析及卡尔曼滤波的瓦斯涌出量预测研究

发布时间：2019-01-04 21:49

【摘要】：矿井瓦斯灾害威胁着煤矿安全生产,掌握瓦斯涌出量预测方法、实现矿井瓦斯涌出量准确预测是研究矿井瓦斯涌出量规律及瓦斯地质规律的基础,对矿井瓦斯灾害防治及井下人员生命保障有重要意义。本文以兖矿集团新疆某瓦斯矿井为试验研究对象,从矿井地质条件及开采条件两个方面着手,研究瓦斯涌出量与其影响因素之间相互作用关系,探究得到瓦斯涌出量影响因素具有影响因素众多、作用程度不尽相同;与瓦斯涌出量之间存在复杂非线性关系;随时间、工作面推进发生变化等特征。针对瓦斯涌出量影响因素众多且作用程度各异这一特征,提出基于因子分析法的瓦斯涌出量预测指标选取方法,通过对瓦斯涌出量影响因素原始变量提取有效相同公共因子的方法,减少了原始变量之间信息重叠性,实现了原始变量的优化降维,得到了瓦斯涌出量预测指标。针对瓦斯涌出量预测指标与瓦斯涌出量之间非线性及其自身时变性的特征,构建了 BP神经网络与卡尔曼滤波耦合的瓦斯涌出量预测模型。通过BP神经网络不仅实现了瓦斯涌出量预测指标的非线性映射识别,还为卡尔曼滤波理论递推方程组提供了状态变量。当预测指标随工作面推进发生变化后,BP神经网络能有效将指标信息变化有效地识别反应到卡尔曼滤波的状态变量上,实现了瓦斯涌出量的动态预测。通过研究总结因子分析法、BP神经网络卡尔曼滤波等理论在矿井瓦斯涌出量预测中的应用,以MATLAB软件为开发平台,图形用户界面GUI为软件开发工具,设计开发了基于因子分析及卡尔曼滤波的瓦斯涌出量预测软件。软件有效集成了因子分析法选取预测指标模块以及BP神经网络与卡尔曼滤波耦合的预测模型等模块。通过应用实例分析表明,该软件具有操作方便、界面友好、预测精度高的特点,能够满足矿井瓦斯涌出量预测的实际要求。
[Abstract]:Mine gas disaster threatens coal mine safety production. It is the foundation of studying mine gas emission law and gas geology law to grasp the prediction method of gas emission quantity and to realize accurate prediction of mine gas emission quantity. It is of great significance to the prevention and control of mine gas disaster and the life support of underground personnel. In this paper, a gas mine of Yankuang Group is taken as the experimental research object. From the two aspects of mine geological conditions and mining conditions, the interaction between gas emission and its influencing factors is studied. To explore the influence factors of gas emission quantity has many factors, the action degree is not the same; There is a complex nonlinear relationship between gas emission and the change of working face propulsion with time. In view of the fact that there are many factors affecting the quantity of gas emission and the degree of action is different, a method based on factor analysis is put forward to select the prediction index of gas emission quantity. By extracting the effective same common factors from the original variables, the information overlap between the original variables is reduced, the dimension reduction of the original variables is realized, and the prediction index of the gas emission quantity is obtained. Aiming at the nonlinearity between gas emission prediction index and gas emission and its own time-varying characteristics, a gas emission prediction model coupled with BP neural network and Kalman filter is constructed. The BP neural network not only realizes the nonlinear mapping identification of gas emission prediction index, but also provides the state variable for the recursive equations of Kalman filter theory. When the prediction index changes with the advance of the working face, the BP neural network can effectively identify the change of the index information to the state variables of Kalman filter, and realize the dynamic prediction of gas emission. The application of factor analysis, BP neural network Kalman filter and so on in the prediction of mine gas emission is studied. The MATLAB software is used as the development platform, and the graphical user interface (GUI) is used as the software development tool. The software of gas emission prediction based on factor analysis and Kalman filter is designed and developed. The software effectively integrates the module of selecting prediction index by factor analysis method and the prediction model of coupling BP neural network and Kalman filter. The application examples show that the software has the characteristics of convenient operation, friendly interface and high prediction precision, and it can meet the actual requirements of the prediction of mine gas emission.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD712.5

【参考文献】