Spatiotemporal multi-step prediction of hydraulic support pressure based on LSTM-Informer model
-
摘要: 目前多步液压支架压力预测大多为单步液压支架压力的累计预测，单步累计次数越多，累计误差就越大，影响预测精度。针对该问题，提出了一种基于长短时记忆（LSTM）−Informer模型的液压支架压力时空多步长预测方法。采用卡尔曼滤波消除液压支架压力数据中的振动噪声后，在工作面端部和中部各选取相邻的5台液压支架压力数据建立2个时空数据集（数据集1和数据集2），并对时空数据进行标准化预处理。将时空数据输入LSTM模型提取时空特征，并将提取的时空特征输入Informer模型的编码器，经过位置编码后利用多头概率稀疏自注意力来关注压力序列的变化特征，经过最大池化和一维卷积消除最终输出特征图的冗余组合。利用多头概率稀疏自注意力来关注压力序列的变化特征，将Informer模型的解码器改为全连接层，得到液压支架压力的预测结果。实验结果表明：与基于门控循环单元（GRU）、LSTM和Informer模型的预测方法相比， 基于LSTM−Informer模型的预测方法在预测6，12，24步长液压支架压力时的均方根误差（RMSE）和平均绝对误差（MAE）均最小；其中基于数据集1预测的6步长液压支架压力的RMSE分别降低了41.63%，49.74%，11.85%，MAE分别降低了41.75%，50.00%，12.00%；基于数据集2预测的6步长液压支架压力的RMSE分别降低了48.15%，59.86%，19.88%，MAE分别降低了49.87%，54.90%，13.16%。
-
关键词:
- 液压支架压力 /
- 多步长液压支架压力预测 /
- LSTM−Informer模型 /
- 时间相关性 /
- 卡尔曼滤波
Abstract: Currently, most multi-step hydraulic support pressure predictions are cumulative predictions of single step hydraulic support pressure. The more times a single step accumulates, the greater the cumulative error, which affects the prediction precision. In order to solve the above problems, a spatiotemporal multi-step prediction method of hydraulic support pressure based on long short term memory (LSTM)-Informer model is proposed. After using Kalman filtering to eliminate vibration noise in hydraulic support pressure data, two spatiotemporal datasets (Dataset 1 and Dataset 2) are established by selecting 5 adjacent hydraulic support pressure data at the end and middle of the working face. The spatiotemporal data is standardized and preprocessed. The method inputs spatiotemporal data into the LSTM model to extract spatiotemporal features, and inputs the extracted spatiotemporal features into the encoder of the Informer model. After position encoding, the method outputs multi head probability sparse self attention to focus on the changing features of the pressure sequence. After maximum pooling and one-dimensional convolution, the method eliminates the redundant combination of output feature map. By utilizing multi head probability sparse self attention to further focus on pressure sequence features, the decoder of the Informer model is changed to a fully connected layer to obtain the prediction results of hydraulic support pressure. The experimental results show that compared with prediction methods based on gated recurrent unit (GRU), LSTM, and Informer models, prediction methods based on LSTM-Informer model has the smallest root mean square error (RMSE) and mean absolute error (MAE) in predicting hydraulic support pressure at 6, 12, and 24 step sizes. The RMSE of the 6-step hydraulic support pressure predicted based on dataset 1 decreases by 41.63%, 49.74%, and 11.85%, and the MAE decreases by 41.75%, 50.00%, and 12.00%, respectively. The RMSE of the 6-step hydraulic support pressure predicted based on dataset 2 decreases by 48.15%, 59.86%, and 19.88%, and MAE decreases by 49.87%, 54.90%, and 13.16%, respectively. -
图 1 液压支架压力数据经过卡尔曼滤波的前后对比
Figure 1. Comparison of pressure data of hydraulic support before and after Kalman filtering
图 4 基于LSTM−Informer模型的液压支架压力预测流程
Figure 4. Hydraulic support pressure prediction process based on the LSTM-Informer model
表 1 基于数据集1的预测结果对比
Table 1. Comparison of the prediction results based on dataset 1
模型 RMSE MAE 6步长 12步长 24步长 6步长 12步长 24步长 GRU 0.663 0.932 1.220 0.491 0.683 0.935 LSTM 0.770 0.987 1.232 0.572 0.735 0.943 Informer 0.439 0.758 1.117 0.325 0.559 0.845 LSTM−Informer 0.387 0.599 0.897 0.286 0.451 0.710 
下载: 导出CSV
表 2 基于数据集2的预测结果对比
Table 2. Comparison of prediction results based on dataset 2
模型 RMSE MAE 6步长 12步长 24步长 6步长 12步长 24步长 GRU 0.513 0.671 1.039 0.395 0.518 0.795 LSTM 0.563 0.697 1.065 0.439 0.541 0.814 Informer 0.332 0.582 0.840 0.228 0.405 0.625 LSTM−Informer 0.266 0.457 0.737 0.198 0.353 0.572
下载: 导出CSV
-
[1] WANG Tong,WANG Qingwei,SHAO Longyi,et al. Current status of the research on coal geology in China[J]. Acta Geologica Sinica(English Edition),2016,90(4):1284-1297. doi: 10.1111/1755-6724.12770 [2] LIU Yue. Research on the operating mechanism of China's coal market[J]. Agro Food Industry Hi Tech,2017,28(1):3075-3077. [3] LIN Jiang,FRIDLEY D,LU Hongyou,et al. Has coal use peaked in China:near-term trends in China's coal consumption[J]. Energy Policy,2018,123:208-214. doi: 10.1016/j.enpol.2018.08.058 [4] TANG Xu,JIN Yi,MCLELLAN B C,et al. China's coal consumption declining-impermanent or permanent?[J]. Resources,Conservation & Recycling,2018,129:307-313. [5] ZHANG Yujiang,FENG Guorui,ZHANG Min,et al. Residual coal exploitation and its impact on sustainable development of the coal industry in China[J]. Energy Policy,2016,96:534-541. doi: 10.1016/j.enpol.2016.06.033 [6] 张吉雄,张强,巨峰,等. 深部煤炭资源采选充绿色化开采理论与技术[J]. 煤炭学报,2018,43(2):377-389.ZAHNG Jixiong,ZHANG Qiang,JU Feng,et al. Theory and technique of greening mining integrating mining,separating and backfilling in deep coal resources[J]. Journal of China Coal Society,2018,43(2):377-389. [7] 王世斌,侯恩科,王双明,等. 煤炭安全智能开采地质保障系统软件开发与应用[J]. 煤炭科学技术,2022,50(7):13-24.WANG Shibin,HOU Enke,WANG Shuangming,et al. Development and application of geological guarantee system software for safe and intelligent coal mining[J]. Coal Science and Technology,2022,50(7):13-24. [8] 程建远,朱梦博,王云宏,等. 煤炭智能精准开采工作面地质模型梯级构建及其关键技术[J]. 煤炭学报,2019,44(8):2285-2295.CHENG Jianyuan,ZHU Mengbo,WANG Yunhong,et al. Cascade construction of geological model of longwall panel for intelligent precision coal mining and its key technology[J]. Journal of China Coal Society,2019,44(8):2285-2295. [9] YANG Li,BIRHANE G E,ZHU Junqi,et al. Mining employees safety and the application of information technology in coal mining:review[J]. Frontiers in Public Health,2021,9. DOI: 10.3389/FPUBH.2021.709987. [10] 赵毅鑫,杨志良,马斌杰,等. 基于深度学习的大采高工作面矿压预测分析及模型泛化[J]. 煤炭学报,2020,45(1):54-65.ZHAO Yixin,YANG Zhiliang,MA Binjie,et al. Deep learning prediction and model generalization of ground pressure for deep longwall face with large mining height[J]. Journal of China Coal Society,2020,45(1):54-65. [11] 李泽萌. 基于LSTM的采煤工作面矿压预测方法研究[D]. 西安:西安科技大学,2020.LI Zemeng. Research on the prediction method of mining pressure in coal mining face based on LSTM[D]. Xi'an:Xi'an University of Science and Technology,2020. [12] 曾庆田,吕珍珍,石永奎,等. 基于Prophet+LSTM模型的煤矿井下工作面矿压预测研究[J]. 煤炭科学技术,2021,49(7):16-23.ZENG Qingtian,LYU Zhenzhen,SHI Yongkui,et al. Research on prediction of underground coal mining face pressure based on Prophet+LSTM model[J]. Coal Science and Technology,2021,49(7):16-23. [13] 李泽西. 基于可变时序移位Transformer−LSTM的集成学习矿压预测方法[J]. 工矿自动化,2023,49(7):92-98.LI Zexi. Ensemble learning mine pressure prediction method based on variable time series shift Transformer-LSTM[J]. Journal of Mine Automation,2023,49(7):92-98. [14] LIU Yaping,DONG Lihong,YE Ou. Mine pressure prediction model of fully mechanized mining face based on Improved Transformer[C]. IEEE International Conference on Signal Processing,Communications and Computing,Xi'an,2022:1-5. [15] DONG Jianjun,XIE Zhengquan,JIANG Hao,et al. Multiple regression method for working face mining pressure prediction based on hydraulic support monitoring dataset[J]. Frontiers in Earth Science,2023. DOI: 10.3389/FEART.2023.1114033. [16] 向玲,刘佳宁,苏浩,等. 基于CEEMDAN二次分解和LSTM的风速多步预测研究[J]. 太阳能学报,2022,43(8):334-339.XIANG Ling,LIU Jianing,SU Hao,et al. Research on multi-step wind speed forecast based on CEEMDAN secondary decomposition and LSTM[J]. Acta Energiae Solaris Sinica,2022,43(8):334-339. [17] 潘超,李润宇,王典,等. 基于风速时空关联的多步预测方法[J]. 太阳能学报,2022,43(2):458-464.PAN Chao,LI Runyu,WANG Dian,et al. Multi-step wind speed prediction method based on wind speed spatial-time correlation[J]. Acta Energiae Solaris Sinica,2022,43(2):458-464. [18] JINAH K,TAEKYUNG K,JOON-GYU R,et al. Spatiotemporal graph neural network for multivariate multi-step ahead time-series forecasting of sea temperature[J]. Engineering Applications of Artificial Intelligence,2023,126. DOI: 10.1016/J.ENGAPPAI.2023.106854. [19] 严恭敏,邓瑀. 传统组合导航中的实用Kalman滤波技术评述[J]. 导航定位与授时,2020,7(2):50-64.YAN Gongmin,DENG Yu. Review on practical Kalman filtering techniques in traditional integrated navigation system[J]. Navigation Positioning and Timing,2020,7(2):50-64. [20] ZHOU Haoyi,ZHANG Shanghang,PENG Jieqi,et al. Informer:beyond efficient transformer for long sequence time-series forecasting[C]. AAAI Conference on Artificial Intelligence,Vancouve,2021:11106-11115. [21] VASWANI A,SHAZEER N,PARMAR N,et al. Attention is all you need[C]. The 31st International Conference on Neural Information Processing Systems,Long Beach,2017:5998-6008. -
点击查看大图
图(4) / 表(2)
计量
- 文章访问数: 266
- HTML全文浏览量: 55
- PDF下载量: 44
- 被引次数: 0
