摘要
顶板压力预测大多都使用1台液压支架的压力数据,缺乏相邻支架的压力相关性,基于相邻多台液压支架信息融合的LSTM-TCN模型预测一支架下一时刻的压力值。实验结果表明,该模型相比LSTM和TCN单支架预测模型,RMSE压力误差分别减小了12.3%和13.4%,MAE压力误差减小了10.3%和26.6%。
Most of the top plate pressure prediction uses the pressure data of one hydraulic bracket,and the pressure correlation of adjacent brackets is lacking,and predicts the pressure value of a bracket at the next moment based on the LSTM-TCN model of the information fusion of multiple adjacent hydraulic supports.Experimental results show that compared with LSTM and TCN singlebracket prediction models,the RMSE error value is reduced by 12.3%and 13.4%,the MAE error value is reduced by 10.3%and 26.6%.
作者
余琼芳
王联港
杨艺
YU Qiongfang;WANG Liangang;YANG Yi(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454000,China;Postdoctoral Research Station,Beijing Research Institute,Dalian University of Technology,Beijing 100000,China)
出处
《煤炭技术》
CAS
北大核心
2023年第6期5-9,共5页
Coal Technology
基金
国家自然科学基金资助项目(61601172)
中国博士后科学基金资助项目(2018M641287)。
关键词
顶板事故
顶板压力
液压支架
滑动均值滤波
top plate accident
top plate pressure
hydraulic support
sliding mean filtering