信息融合技术在矿井底板突水预测中的应用研究

发布时间：2018-05-31 22:15

本文选题：突水预测 + 信息融合　；参考：《辽宁工程技术大学》2013年硕士论文

【摘要】：煤矿水害事故的频发，给煤矿工人的生命安全带来隐患的同时煤矿的经济效益也因水害受到了极大的威胁。其中，造成最严重危害的当属矿井底板突水事故。由于我国煤矿特殊的水文地质条件,造成底板突水的影响因素多而且关系复杂，许多传统的矿井底板突水预测方法已经不能够在实际中达到理想的预测效果。面对仍不完善的矿井底板突水预测法，急需新的技术理论融入其中，本文就此提出了一种双层信息融合技术，将其应用于矿井底板突水预测中，并设计了实现突水预测的方案。首先，介绍了底板突水预测和信息融合技术的研究现状；其次论述了多源信息融合的原理、层次结构和融合算法等，接着主要研究了基于RBF神经网络的信息融合方法，将影响矿井底板突水的多个传感器采集的数据经过处理后作为RBF神经网络的输入，通过分析选择量子粒子群智能算法对网络各个参数值进行优化，建立了基于RBF神经网络特征层信息融合的矿井底板突水预测方法。单一采用特征层信息融合的结果存在一定的不稳定性，为了提高突水预测的可靠性，引入了D-S证据理论进行决策层信息融合，把每一个神经网络的输出归一化处理后转变为各个证据体的基本概率分配函数，再利用融合规则对各个证据体进行融合，根据决策规则判决，得出最后决策结果。本文将信息融合技术引入矿井底板突水预测中，采用了双层信息融合算法，建立了矿井底板突水预测的一般框架。经实验分析，双层信息融合方法所计算出的预测结果准确性高、不确定性低，在矿井底板预测领域有很好的应用前景。
[Abstract]:The frequent occurrence of mine water hazard accidents brings hidden trouble to the life safety of coal miners. Meanwhile, the economic benefits of coal mines are also greatly threatened by water hazards. Among them, the most serious harm is the mine floor water inrush accident. Because of the special hydrogeological conditions of coal mine in our country, there are many factors affecting the water inrush from the floor and the relationship is complex. Many traditional prediction methods of water inrush from the floor of the mine can no longer achieve the ideal prediction effect in practice. In the face of the imperfect prediction method of mine floor water inrush, it is urgent to incorporate new technical theory into it. In this paper, a two-layer information fusion technique is put forward, which is applied to the prediction of mine floor water inrush, and a scheme to realize the prediction of water inrush is designed. Firstly, this paper introduces the research status of water inrush prediction and information fusion technology of bottom plate, then discusses the principle, hierarchical structure and fusion algorithm of multi-source information fusion, and then mainly studies the information fusion method based on RBF neural network. The data collected by several sensors which affect the water inrush of the mine floor are processed as the input of the RBF neural network, and the parameters of the network are optimized by analyzing and selecting the quantum particle swarm intelligence algorithm. A prediction method of mine floor water inrush based on RBF neural network feature layer information fusion is established. In order to improve the reliability of water inrush prediction, the D-S evidence theory is introduced to fuse the information of decision level in order to improve the reliability of water inrush prediction. The output of each neural network is normalized and transformed into the basic probability distribution function of each evidence body, and then the fusion rules are used to fuse each evidence body, and the final decision result is obtained according to the decision rule. In this paper, the information fusion technology is introduced into the prediction of water inrush from mine floor, and a general frame of prediction of water inrush from mine floor is established by using the two-layer information fusion algorithm. The experimental results show that the prediction results calculated by the two-layer information fusion method have high accuracy and low uncertainty. It has a good application prospect in the field of mine floor prediction.
【学位授予单位】：辽宁工程技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TD745;TP202

【参考文献】