摘要
对煤矿井下瓦斯浓度的预测一直以来是矿井安全进行早期预警和管理的关键问题。为了进一步提高煤矿瓦斯浓度预测的准确度,提出了一种基于深度学习的方法,称为AR-LSTM,它包括残差神经网络(ResNet)、长短时记忆(LSTM)网络和基于注意力的网络,用于煤矿井下瓦斯浓度的预测。AR-LSTM不仅使用瓦斯浓度这一变量,同时将采集的温度、风速和一氧化碳浓度作为输入。因此,在AR-LSTM模型中,ResNet-LSTM网络学习多变量时间序列数据的时序相关性和相互依赖性,注意力机制用于捕捉过去不同时间步的特征状态对未来瓦斯浓度的重要性程度。基于注意力的层可以自动加权过去的特征状态以提高预测准确性,使用煤矿地区的瓦斯浓度数据进行预测,并将其与3种基准方法进行比较。为了比较每种方法的整体性能,实验中使用了均方根误差E_(RMS)、平均绝对误差E_(MA)和决定系数R^(2)。实验结果表明,AR-LSTM模型能够以最高性能处理煤矿瓦斯浓度的预测问题,并且可以实现1步或多步提前预测。
The prediction of gas concentration in underground coal mines has always been an essential issue of early warning and management for mine safety.In order to further improve the accuracy of gas concentration prediction in coal mines,a method based on deep learning is proposed,referred to as AR-LSTM.This method includes Residual Neural Network(ResNet),Long Short-Term Memory(LSTM)network,and Attention-based network for predicting gas concentration in underground coal mines.AR-LSTM not only uses gas concentration as a variable but also takes temperature,wind speed,and carbon monoxide concentration collected as inputs.Therefore,in the AR-LSTM model,the Resnet-LSTM network learns the temporal correlation and interdependence of multivariate time series data,while the attention mechanism captures the importance of past feature states at different time steps for the future gas concentration.The attention-based layer automatically weights past feature states to enhance prediction accuracy.The model is evaluated using gas concentration data from coal mining areas and compared with three benchmark methods.To assess the overall performance of each method,root mean square error E_(RMS),mean absolute error E_(MA),and coefficient of determination R^(2) are used in the experiments.The results demonstrate that the AR-LSTM model performs at the highest level in handling the prediction of gas concentration in coal mines,enabling both single-step and multi-step ahead forecasting.
作者
张玲
杨超宇
ZHANG Ling;YANG Chaoyu(Department of Economics and Management,Anhui University of Science and Technology,Huainan 232000,China;Department of Artificial Intelligence,Anhui University of Science and Technology,Huainan 232000,China)
出处
《煤炭技术》
CAS
2024年第8期208-213,共6页
Coal Technology
基金
国家自然科学基金项目(61873004)。
作者简介
张玲(1999-),女,安徽安庆人,硕士研究生,研究方向:数据分析与数据挖掘,电子信箱:1309455849@qq.com;通信作者:杨超宇(1981-),安徽淮南人,教授,博士,研究方向:煤矿安全信息化技术应用.
