双策略耦合优化的含瓦斯煤破裂过程信号辨识

doi:10.16265/j.cnki.issn1003-3033.2022.10.1868

摘要/Abstract

摘要：

为解决含瓦斯煤破裂过程信号特征的辨识问题,以双向长短时记忆网络(BiLSTM)为基分类器,提出一种基于AdaBoost算法与哈里斯鹰优化(HHO)算法双策略耦合优化的辨识模型。首先针对AdaBoost算法中错误样本占比随迭代不断增加影响最终强分类器辨识效果的问题,引入权重参数,以改变弱分类器权重,提高辨识精度;然后为确定最优的模型参数,结合HHO,优化辨识参数与权重参数,优化过程中HHO与改进的AdaBoost算法产生耦合作用,使得辨识模型的准确性和稳定性达到最优,平均准确率为91.36%,标准差缩小至0.017 4。研究结果表明:双策略耦合优化HHO-AdaBoost-BiLSTM含瓦斯煤体破裂过程信号特征辨识模型准确性更高,稳定性更强。

关键词: 含瓦斯煤破裂, 信号特征辨识, 双策略, 耦合优化, AdaBoost算法, 哈里斯鹰优化(HHO)算法

Abstract:

In order to address identification problems of signal characteristics of fracture in gas bearing coal during fracture process, with Bi-directional Long Short-Term Memory network (BiLSTM) as base classifier, a dual strategy coupling optimization identification model for such signal characteristics was proposed based on AdaBoost algorithm and HHO algorithm. Firstly, in view of the problem that proportion of error samples in AdaBoost algorithm increased with iteration, which had an impact on results of final strong classifiers, weight parameters were introduced and weights of weak classifiers were changed so as to improve identification. Then, to determine optimal model parameters, identification parameters and weight parameters were optimized based on HHO, and during optimization process, HHO and improved AdaBoost algorithm produced a coupling effect, which made the identification model's accuracy and stability reach optimal level, resulting in an average accuracy of 91.36% and standard deviation being reduced to 0.017 4. The results show that the dual strategy coupling optimization model of HHO-AdaBoost-BiLSTM identification of signal characteristics of fracture in gas bearing coal has higher accuracy and stability.

Key words: fracture in gas bearing coal, identification of signal characteristics, dual strategy, coupling optimization, AdaBoost algorithm, Harris hawk optimization (HHO) algorithm

付华, 赵俊程, 刘昊, 刘雨竹, 卢万杰. 双策略耦合优化的含瓦斯煤破裂过程信号辨识[J]. 中国安全科学学报, 2022, 32(10): 40-47.

FU Hua, ZHAO Juncheng, LIU Hao, LIU Yuzhu, LU Wanjie. Signal identification of fracture in gas bearing coal based on dual strategy coupling optimization[J]. China Safety Science Journal, 2022, 32(10): 40-47.

图/表 10

表1

图1

表2

图2

图3

图4

图5

图6

表3

表4

参考文献 25

[1]	CHU Yapei, SUN Haitao, ZHANG Dongming. Experimental study on evolution in the characteristics of permeability, deformation, and energy of coal containing gas under triaxial cyclic loading-unloading[J]. Energy Science & Engineering, 2019, 7(5):2112-2123.
[2]	罗浩, 潘一山, 赵扬锋, 等. 含瓦斯煤体加载破坏声-电前兆信息试验研究[J]. 煤炭学报, 2015, 40(3):548-554.
	LUO Hao, PAN Yishan, ZHAO Yangfeng, et al. Experimental study on acousto-charge precursory information of coal containing gas during loading failure process[J]. Journal of China Coal Society, 2015, 40(3):548-554.
[3]	李忠辉, 刘永杰, 王恩元, 等. 含瓦斯煤体受载破坏的电位信号灰色突变特征研究[J]. 煤炭科学技术, 2015, 43(10):18-22, 132.
	LI Zhonghui, LIU Yongjie, WANG Enyuan, et al. Research on grey-catastrophe characteristics of electric potential signal during gas-bearing coal under loaded breaking[J]. Coal Science and Technology, 2015, 43(10):18-22, 132.
[4]	高保彬, 吕蓬勃, 郭放. 不同瓦斯压力下煤岩力学性质及声发射特性研究[J]. 煤炭科学技术, 2018, 46(1):112-119, 149.
	GAO Baobin, LYU Pengbo, GUO Fang. Study on mechanical properties and acoustic emission characteristics of coal at different gas pressure[J]. Coal Science and Technology, 2018, 46(1):112-119, 149.
[5]	陈亮, 王恩元. 含瓦斯煤受载破坏瓦斯涌出的前兆特征研究[J]. 中国安全科学学报, 2020, 30(12):79-84. doi: 10.16265/j.cnki.issn 1003-3033.2020.12.011
	CHEN Liang, WANG Enyuan. Study on precursory characteristics of gas emission from damaged coal containing gas[J]. China Safety Science Journal, 2020, 30(12):79-84. doi: 10.16265/j.cnki.issn 1003-3033.2020.12.011
[6]	王雨虹, 孙福成, 付华, 等. 基于优化的量子门节点神经网络的煤与瓦斯突出预测[J]. 信息与控制, 2020, 49(2):249-256. doi: 10.13976/j.cnki.xk.2020.9204
	WANG Yuhong, SUN Fucheng, FU Hua, et al. Prediction of coal and gas outburst based on optimized quantum gated neural networks[J]. Information and Control, 2020, 49(2):249-256. doi: 10.13976/j.cnki.xk.2020.9204
[7]	毕娟, 李希建, 陈刘瑜. 预测冲击地压危险性等级R型因子Fisher判别[J]. 中国安全科学学报, 2019, 29(12):103-109. doi: 10.16265/j.cnki.issn1003-3033.2019.12.017
	BI Juan, LI Xijian, CHEN Liuyu. R-factor Fisher discrimination for rock burst hazard level prediction[J]. China Safety Science Journal, 2019, 29(12):103-109. doi: 10.16265/j.cnki.issn1003-3033.2019.12.017
[8]	POGIATZIS A, SAMAKOVITIS G. Using BiLSTM networks for context-aware deep sensitivity labelling on conversational data[J]. Applied Sciences, 2020, 10(24): DOI: 10.3390/APP10248924. doi: 10.3390/APP10248924
[9]	徐超, 叶宁, 徐康, 等. 融合MAML与BiLSTM的微博负面情感多分类方法[J]. 计算机工程与应用, 2022, 58(5):179-185. doi: 10.3778/j.issn.1002-8331.2009-0337
	XU Chao, YE Ning, XU Kang, et al. Multiple classification method for micro-blog negative emotions integrating MAML and BiLSTM[J]. Computer Engineering and Applications, 2022, 58(5):179-185. doi: 10.3778/j.issn.1002-8331.2009-0337
[10]	HEIDARI A A, MIRJALILI S, FARIS H, et al. Harris hawks optimization: agorithm and applications[J]. Future Generation Computer Systems, 2019, 97:849-872. doi: 10.1016/j.future.2019.02.028
[11]	PHONG T N, DUONG H H, ABOLFAZL J, et al. Groundwater potential mapping combining artificial neural network and real AdaBoost ensemble technique: the Daknong province case-study, Vietnam[J]. International Journal of Environmental Research and Public Health, 2020, 17(7):2473-2473. doi: 10.3390/ijerph17072473
[12]	温廷新, 孙雪, 孔祥博, 等. 基于PSOBP-AdaBoost模型的瓦斯涌出量分源预测研究[J]. 中国安全科学学报, 2016, 26(5):94-98.
	WEN Tingxin, SUN Xue, KONG Xiangbo, et al. Research on prediction of gas emission quantity with sub sources basing on PSOBP-AdaBoost[J]. China Safety Science Journal, 2016, 26(5):94-98.
[13]	任克强, 高晓林, 谢斌. 基于AFSA和PSO融合优化的AdaBoost人脸检测算法[J]. 小型微型计算机系统, 2016, 37(4):861-865.
	REN Keqiang, GAO Xiaolin, XIE Bin. AdaBoost face detection algorithm based on fusion optimization of AFSA and PSO[J]. Journal of Chinese Computer Systems, 2016, 37(4):861-865.
[14]	张均, 叶庆卫. 基于PSO的改进AdaBoost人脸检测算法[J]. 计算机应用, 2020, 40(增1):61-64.
	ZHANG Jun, YE Qingwei. Improved AdaBoost face detection algorithm based on particle swarm optimization[J]. Journal of Computer Applications, 2020, 40(S1):61-64.
[15]	LYDIE M M A, USHAD S A, YUVRAJ S. Time series modelling, NARX neural network and hybrid KPCA-SVR approach to forecast the foreign exchange market in Mauritius[J]. African Journal of Economic and Management Studies, 2020, 12(1):18-54. doi: 10.1108/AJEMS-04-2019-0161
[16]	MEKRUKSAVANICH S, JITPATTANAKUL A. LSTM networks using smartphone data for sensor-based human activity recognition in smart homes[J]. Sensors, 2021, 21(5):DOI: 10.3390/s21051636. doi: 10.3390/s21051636
[17]	郭雨鑫, 刘升, 高文欣, 等. 精英反向学习与黄金正弦优化的HHO算法[J]. 计算机工程与应用, 2022, 58(10):153-161. doi: 10.3778/j.issn.1002-8331.2011-0321
	GUO Yuxin, LIU Sheng, GAO Wenxin, et al. Elite opposition-based learning golden-sine Harris hawks optimization[J]. Computer Engineering and Applications, 2022, 58(10): 153-161. doi: 10.3778/j.issn.1002-8331.2011-0321
[18]	ABOOZAR T, GEORGINA C, MCGINNITY T M. AdaBoost-CNN: an adaptive boosting algorithm for convolutional neural networks to classify multi-class imbalanced datasets using transfer learning[J]. Neurocomputing, 2020, 404: 351-366. doi: 10.1016/j.neucom.2020.03.064
[19]	朱亮, 徐华, 崔鑫. 基于基分类器系数和多样性的改进AdaBoost算法[J]. 计算机应用, 2021, 41(8): 2225-2231. doi: 10.11772/j.issn.1001-9081.2020101584
	ZHU Liang, XU Hua, CUI Xin. Improved AdaBoost algorithm based on classifier coefficient and diversity[J]. Journal of Computer Applications, 2021, 41(8): 2225-2231. doi: 10.11772/j.issn.1001-9081.2020101584
[20]	郑晓亮, 来文豪, 薛生. MI和SVM算法在煤与瓦斯突出预测中的应用[J]. 中国安全科学学报, 2021, 31(1):75-80. doi: 10.16265/j.cnki.issn 1003-3033.2021.01.011
	ZHENG Xiaoliang, LAI Wenhao, XUE Sheng. Application of MI and SVM in coal and gas outburst prediction[J]. China Safety Science Journal, 2021, 31(1):75-80. doi: 10.16265/j.cnki.issn 1003-3033.2021.01.011
[21]	牟多铎, 刘磊. ELM与SVM在高光谱遥感图像监督分类中的比较研究[J]. 遥感技术与应用, 2019, 34(1):115-124.
	MOU Duoduo, LIU Lei. Comparative study of ELM and SVM in hyperspectral image supervision classification[J]. Remote Sensing Technology and Application, 2019, 34(1):115-124.
[22]	蔺瑞管, 王华伟, 车畅畅, 等. 基于LSTM分类器的航空发动机预测性维护模型[J]. 系统工程与电子技术, 2022, 44(3):1052-1059.
	LIN Ruiguan, WANG Huawei, CHE Changchang, et al. Predictive maintenance model of aeroengine based on LSTM classifier[J]. Systems Engineering and Electronics, 2022, 44(3):1052-1059.
[23]	刘可真, 苟家萁, 骆钊, 等. 基于粒子群优化-长短期记忆网络模型的变压器油中溶解气体浓度预测方法[J]. 电网技术, 2020, 44(7):2778-2785.
	LIU Kezhen, GOU Jiaqi, LUO Zhao, et al. Prediction of dissolved gas concentration in transformer oil based on PSO-LSTM model[J]. Power System Technology, 2020, 44(7):2778-2785.
[24]	王立辉, 杨辉斌, 王银堂, 等. 基于GWO-LSTM的丹江口水库入库径流预测[J]. 水利水运工程学报, 2021(6):51-59.
	WANG Lihui, YANG Huibin, WANG Yintang, et al. Prediction of inflow to the Danjiangkou reservoir based on GWO-LSTM[J]. Hydro-Science and Engineering, 2021(6):51-59.
[25]	谢磊, 丁达理, 魏政磊, 等. AdaBoost-PSO-LSTM网络实时预测机动轨迹[J]. 系统工程与电子技术, 2021, 43(6):1651-1658.
	XIE Lei, DING Dali, WEI Zhenglei, et al. Real time prediction of maneuver trajectory for AdaBoost-PSO-LSTM network[J]. Systems Engineering and Electronics, 2021, 43(6):1651-1658.

样本	P₁/ m	P₂/ MPa	P₃	P₄	P₅/ 次	P₆/ mV	P₇/ (kg·m^-1)	P₈/ MPa	P₉/ (m³·t^-1)	P₁₀/ (L·min^-1)	P₁₁/ (L·min^-1)	P₁₂/ (m³·min^-1)	P₁₃	风险等级
1	425	12.9	0.58	3	885	224	8.88	2.80	10.24	8	2.1	6.3	1	高
2	744	15.2	0.37	3	735	167	8.38	1.24	9.01	8	2.9	5.7	2	高
样本	P₁/ m	P₂/ MPa	P₃	P₄	P₅/ 次	P₆/ mV	P₇/ (kg·m^-1)	P₈/ MPa	P₉/ (m³·t^-1)	P₁₀/ (L·min^-1)	P₁₁/ (L·min^-1)	P₁₂/ (m³·min^-1)	P₁₃	风险等级
3	512	5.7	0.54	1	120	12	8.30	0.44	4.61	7	2.3	2.2	1	低
4	484	16.6	0.52	5	673	138	5.06	1.28	8.26	6	3.1	3.9	2	中
5	397	11.1	0.61	1	332	142	4.24	1.19	9.05	5	3.5	2.9	1	中
6	462	12.4	0.16	3	357	129	3.42	1.20	10.27	18	3.3	6.9	1	中
7	750	7.9	0.35	1	146	142	5.19	0.74	5.21	8	3.6	1.0	1	低
8	400	3.9	0.62	1	223	33	5.91	0.62	2.01	6	4.2	1.8	3	低
9	446	13.2	0.23	3	649	241	8.51	0.94	13.06	6	3.4	7.9	1	中
10	621	7.7	0.31	5	579	163	5.02	2.76	10.03	20	2.6	2.6	3	中

指标	填补前		填补后
指标	均值	标准差	均值	标准差
P₃	0.44	0.15	0.42	0.17
P₅	5.85	2.48	5.90	2.43
P₁₀	2.67	0.50	2.63	0.57

模型	平均准确率/%	标准差
SVM	66.64	0.065 3
ELM	71.32	0.056 9
LSTM	74.48	0.050 6
BiLSTM	77.92	0.045 2

辨识模型	平均准确率/%	标准差
PSO-BiLSTM	81.68	0.042 7
GWO-BiLSTM	85.84	0.040 7
HHO-BiLSTM	87.76	0.037 8
原始AdaBoost-BiLSTM	86.08	0.027 7
原始AdaBoost-HHO-BiLSTM	85.92	0.026 2
HHO-AdaBoost-BiLSTM	91.68	0.018 7