基于BP神经网络的煤与瓦斯突出声发射监测仪的设计

发布时间：2018-06-01 01:02

本文选题：煤与瓦斯突出 + 声发射　；参考：《太原理工大学》2013年硕士论文

【摘要】：煤炭作为我国的第一能源,随着经济的发展需求量与日俱增,伴随而来的煤矿安全问题也不容忽视。煤与瓦斯突出是一种严重的矿井自然灾害,它虽具有突发性,但在突出前均有前兆显现,其中声发射就是前兆之一。针对我国当前的煤矿开采现状,本文在声发射理论的基础上设计了一个煤与瓦斯突出的声发射监测仪,并通过仿真、调试,最终达到了较为理想的预期效果。本文主要从以下三个方面对监测仪进行了设计。在煤矿井下,突出声发射信号众多但无明显规律可循,而人工神经网络具有强大的非线性处理能力,它可以不用研究大量监测数据之间的复杂关系,仅仅通过对输入、输出的记忆学习便可以找出相应的非线性映射关系,这正适合于声发射数据的动态预测。本文系统研究了BP神经网络在煤与瓦斯突出声发射预测的基本原理,并论证了该设计的可行性。在声发射和BP神经网络的理论基础上,设计了声发射检测仪的总体框架。根据煤与瓦斯突出预测快速、及时、准确的要求,设计了一种基于DSP+单片机的双CPU并行处理、计算的监测系统,巧妙了运用了DSP对声发射信号强大的数据处理能力和单片机的人机交互功能。从硬件结构出发,根据突出声发射数据监测的要求详细设计了系统中各个功能模块。接着以硬件结构为基础,对每一个模块按其功能进行了软件编程,并进行了调试。最后在MATLAB软件中利用神经网络工具箱搭建了声发射信号BP神经网络处理的模型,并对相应的数据进行了学习训练,最后完成了仿真,并运用遗传算法对BP神经网络进行了改进,其结果显示,该方法能比较好的预测煤与瓦斯突出的危险性。
[Abstract]:Coal as the first energy in China, with the increasing demand for economic development, the coal mine safety problems can not be ignored. Coal and gas outburst is a serious mine natural disaster. Although it is sudden, it has the precursors before the outburst, and the acoustic emission is one of the precursors. In this paper, an acoustic emission monitor for coal and gas outburst is designed on the basis of acoustic emission theory, and the desired results are achieved through simulation and debugging. This paper mainly designs the monitor from the following three aspects.
In the coal mine, the outburst acoustic emission signals are numerous but have no obvious rules to follow, and the artificial neural network has a strong non-linear processing ability. It can not study the complex relationship between the large number of monitoring data, and only through the memory learning of input and output, it can find the corresponding nonlinear mapping relation, which is suitable for sound hair. The basic principle of BP neural network in predicting the acoustic emission of coal and gas outburst is systematically studied, and the feasibility of the design is demonstrated.
Based on the theory of acoustic emission and BP neural network, the overall framework of acoustic emission detector is designed. According to the fast, timely and accurate requirements of coal and gas outburst prediction, a dual CPU parallel processing based on DSP+ single chip computer is designed and the monitoring system is calculated. The powerful data processing capability of DSP to acoustic emission signals is skillfully carried out. The man-machine interactive function of the single chip microcomputer.
From the hardware structure, each function module in the system is designed in detail according to the requirements of the outburst of acoustic emission data monitoring. Then, based on the hardware structure, each module is programmed according to its function, and the debugging is carried out.
Finally, in MATLAB software, a neural network toolbox is used to build a model of acoustic emission signal BP neural network processing, and the corresponding data are studied and trained. Finally, the simulation is completed, and the genetic algorithm is used to improve the BP neural network. The results show that the method can predict the danger of coal and gas outburst. Sex.
【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TD713

【参考文献】