论文部分内容阅读
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。
Effective control of coal mine gas volume fraction is an important prerequisite to ensure the safety of property and personnel. At this stage, the measurement methods of gas volume fraction adopted by major coal mining enterprises have certain defects, failing to achieve dynamic and real-time performance, and causing gas explosion accidents occasionally. In order to control the gas content of coal seam scientifically and accurately, the factors influencing the volume fraction of gas are deeply analyzed, and the prediction model scheme based on BP neural network structure is proposed. In order to optimize the forecasting model, the steepness factor, the adaptive adjustment of the learning rate and the improvement of the error function are introduced to improve the prediction accuracy. The model of network training and simulation is adopted to verify the accuracy and provide an effective reference for the coal mine safety production .