BP神经网络在煤矿井下涌水量预测中的应用
Application of BP neural network to forecasting and analysis of coal mine underground water inflow
权文斌 1路文文1
作者信息
- 1. 陕西陕煤黄陵矿业有限公司一号煤矿,陕西 延安 727307
- 折叠
摘要
矿井涌水量受到多种因素的共同影响,具有非线性和高度复杂性的特点.根据黄陵一号煤矿井下涌水量影响因素及2014-2018 年的涌水量数据,采用2 种不同的输入神经元的方法创建神经网络预测模型,用已知数据对创建好的模型进行训练,得到拟合精度较好的模型,并用得到的神经网络模型对涌水量进行预测,最后与实际值进行比较.结果表明,2 种神经网络模型的预测结果精度都较好,但预测精度有差异,用涌水量影响因素为输入神经元的模型在短期预测精度上低于涌水量组合作为输入神经元的模型;而在长期预测方面,涌水量影响因素为输入神经元的模型预测精度高于涌水量组合作为输入神经元的模型.
Abstract
The problem of mine water inflow is affected by many factors,which is non-linear and highly complex.According to the influential factors of underground water inflow and the data of water inflow from 2014 to 2018 in Huangling No.1 Coal Mine,two different methods of input neuron are used to create the neural network prediction model.The model is trained with known data,and the model with better fitting accuracy is obtained.The model is used to predict the water inflow,and fi-nally compared with the actual value.The results show that the prediction accuracy of the two neural network models is good,but the prediction accuracy is different.In terms of accuracy,the model with water inflow impact factor as input neuron is lower than that with water inflow combination as input neuron in short period.In long period however,the model with wa-ter inflow impact factor as input neuron is higher than that with water inflow combination as input neuron regarding the accu-racy.
关键词
矿井涌水量/BP神经网络/迭代训练/拟合精度:模型预测Key words
mine water inflow/BP neural network/iterative training/fitting accuracy:model prediction引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据