基于改进型BP算法的煤矿瓦斯预测与仿真研究

Research on Prediction and Simulation of Coal Mine Gas Based on Improved BP Algorithm

有效控制煤矿井下瓦斯体积分数是保证财产和人员安全的重要前提。现阶段各大煤矿企业采用的瓦斯体积分数测量方法存在一定缺陷,未能实现动态性和实时性,导致瓦斯爆炸事故时有发生。为了科学、准确地控制煤层瓦斯含量,深入分析了影响瓦斯体积分数的因素,提出了基于BP神经网络结构的预测模型方案。为了优化预测模型,从引入陡度因子、自适应调整学习速率、改进误差函数等方面入手,不断提高预测精度,继而采取网络训练和仿真的模型来验证其准确度,为煤矿安全生产提供有效借鉴。

The effective control of coal mine gas volume fraction is an important prerequisite to ensure the property and personal safety. There are some defects in the gas volume fraction measurement methods of coal enterprises at the present stage, it is impossible to realize dynamic and real-time, resulting in gas explosion accidents which have occurred. In order to scientifically and accurately control the coal seam gas content, the factors of influencing the gas volume fraction are analyzed indepth, and the gas volume fraction prediction model is put forward. In order to optimize the prediction model, starting from the introduction of steepness factor, adaptive learning rate, improvement of the error function and so on, the prediction accuracy is continuously improved, and the network training and simulation model are taken to verify its accuracy, the effective reference is provided for coal mine safety production.