摘要
煤矿井下交通系统的安全与稳定,是煤矿产业顺利发展的必要前提,同时,矿用车辆的轨迹预测又是煤矿井下交通系统的重中之重。针对井下环境错综复杂,交通流量大等难题,构建了一种基于注意力机制与双向长短期记忆网络(Attention-BiLSTM)的轨迹预测模型,利用GPS车辆历史轨迹数据,实现了对未来时刻车辆运行轨迹的预测。首先,对数据进行预处理并优化模型,然后,将所提模型与RNN、GRU、标准LSTM等基准模型进行对比实验。结果表明,本文提出的Attention-BiLSTM模型预测准确率为96.8%,且其平均位移误差显著低于对比模型,验证了该模型在井下复杂环境中的有效性与优越性。
The safety and stability of the underground traffic system in coal mines are essential prerequisites for the smooth development of the coal mining industry.Meanwhile,the trajectory prediction of mine vehicles is of paramount importance in the underground traffic system.In response to the complex underground environment and heavy traffic flow,a trajectory prediction model based on the attention mechanism and bidirectional long short-term memory network(Attention-BiLSTM)is constructed.By using the historical GPS vehicle trajectory data,the model predicts the future movement trajectories of vehicles.Firstly,the data is preprocessed and the model is optimized.Then,the proposed model is compared with benchmark models such as RNN,GRU,and standard LSTM through experiments.The results show that the prediction accuracy of the proposed Attention-BiLSTM model is above 96.8%,and its average displacement error is significantly lower than that of the comparison models,verifying the effectiveness and superiority of the model in the complex underground environment.
作者
孟广瑞
刘伟
孙洪涛
周晓东
Meng Guangrui;Liu Wei;Sun Hongtao;Zhou Xiaodong(Intelligent Technology Center,Shendong Coal Branch,China Shenhua Energy Co.,Ltd.,Yulin 719300,China;Coal Preparation Plant HVAC Design Institute,Beijing Huayu Engineering Co.,Ltd.,China Coal Technology and Engineering Group,Beijing 100120,China;College of Instrument Science and Electrical Engineering,Jilin University,Changchun 130061,China)
出处
《煤炭技术》
2026年第1期145-151,共7页
Coal Technology
关键词
煤矿井下交通
车辆轨迹预测
深度学习
长短期记忆网络
注意力机制
双向循环神经网络
underground coal-mine traffic
vehicle trajectory prediction
deep learning
long shortterm memory network
attention mechanism
bidirectional recurrent neural network
