基于随机配置网络的井下供给风量建模

王前进; 杨春雨; 马小平; 张春富; 彭思敏

doi:10.16383/j.aas.c190602

基于随机配置网络的井下供给风量建模

doi: 10.16383/j.aas.c190602

1.
盐城工学院电气工程学院盐城 224051
2.
中国矿业大学信息与控制工程学院徐州 221116

基金项目: 国家自然科学基金(61873272, 61603392), 江苏省自然科学基金(BK20191043), 江苏省“双创团队” 项目(2017), 盐城工学院校级科研项目(xjr2019018)资助

详细信息

作者简介:
王前进：盐城工学院电气工程学院讲师. 2018年获得中国矿业大学博士学位. 主要研究方向为数据驱动建模与控制, 机器学习算法. E-mail: wangqianjinabc@163.com

杨春雨：中国矿业大学信息与控制工程学院教授. 2009年获得东北大学博士学位. 主要研究方向为广义系统和鲁棒控制. E-mail: chunyuyang@cumt.edu.cn

马小平：中国矿业大学信息与控制工程学院教授. 主要研究方向为过程控制, 网络控制系统, 故障诊断. 本文通信作者. E-mail: xpma@cumt.edu.cn

张春富：盐城工学院电气工程学院副教授. 2007年获得哈尔滨工业大学博士学位. 主要研究方向为自动化测控. E-mail: zhangchunfu@hit.edu.cn

彭思敏：盐城工学院副教授, IEEE高级会员. 2013年获得上海交通大学电力系统及其自动化专业博士学位. 主要研究方向为新能源发电, 电池储能系统及微电网控制技术. E-mail: psmsteven@163.com

计量
- 文章访问数: 1714
- HTML全文浏览量: 256
- PDF下载量: 168
- 被引次数: 0
出版历程
- 收稿日期: 2019-08-24
- 录用日期: 2019-11-16
- 网络出版日期: 2021-06-30
- 刊出日期: 2021-08-20

Underground Airflow Quantity Modeling Based on SCN

1.
School of Electrical Engineering, Yancheng Institute of Technology, Yancheng 224051
2.
School of Information and Control Engineering, China University of Mining and Technology, Xuzhou 221116

Funds: Supported by National Natural Science Foundation of China (61873272, 61603392), Natural Science Foundation of Jiangsu Province (BK20191043), Jiangsu Dual Creative Teams Programme Project (2017), Funding for School-Level Research Projects of Yancheng Institute of Technology (xjr2019018)

More Information

Author Bio:
WANG Qian-Jin　Lecturer at the School of Electrical Engineering, Yancheng Institute of Technology. He received his Ph. D. degree from China University of Mining and Technology in 2018. His research interest covers data-driven modeling and control, and machine learning algorithm

YANG Chun-Yu　Professor at the School of Information and Control Engineering, China University of Mining and Technology. He received his Ph. D. degree from Northeastern University in 2009. His research interest covers descriptor systems and robust control

MA Xiao-Ping　Professor at the School of Information and Control Engineering, China University of Mining and Technology. His research interest covers process control, networked control system, and fault detection. Corresponding author of this paper

ZHANG Chun-Fu　Associate professor at the School of Electrical Engineering, Yancheng Institute of Technology. He received his Ph. D. degree from Harbin Institute of Technology in 2007. His research interest covers automatic test and control

PENG Si-Min　 Associate professor at Yancheng Institute of Technology, IEEE senior member. He received his Ph. D. degree in electric power system and its automation from Shanghai Jiao Tong University in 2013. His research interest covers control of renewable sources generation, battery energy storage system and microgrid

摘要

摘要:
主通风机切换过程中, 取压风量测量作为监测井下供给风量的主要手段, 是矿井主扇通风系统安全、稳定与经济运行的重要保障. 然而, 由于取压孔极易出现堵塞现象, 需要频繁维护, 导致无法实时测量井下供给风量, 难以实现主通风机切换过程的闭环优化控制. 同时, 随着隐含层节点数的增加, 基于随机配置网络(Stochastic configuration network, SCN)的估计模型存在过拟合和泛化能力差的缺点. 为了解决上述问题, 结合正则化(Regularization, R)技术, 本文提出一种新型的改进SCN算法, 即RSC算法, 用于井下供给风量的建模. 基准回归分析和工业实验表明: 与SCN方法相比, 建立的RSC模型具有较高的模型精度和较好的泛化性能.
- 主通风机 /
- 切换过程 /
- 井下供给风量 /
- 随机配置网络 /
- 正则化
Abstract:
In a main fan switchover process (MFSP), airflow quantity measuring by pressure drop across a duct is the main means for monitoring the distribution of underground airflow quantity (UAQ). It ensures the safe, stable and economic operation of the mine main fan ventilation system. However, as the pressure port is prone to blocking and requires frequent maintenance, it is impossible to realize the real-time measurement of UAQ and closed-loop optimal control of MFSP. Meanwhile, with the increase of number of hidden nodes, the estimation model based on stochastic configuration network (SCN) has the disadvantages of overfitting and poor generalization ability. To solve the above problems, this paper develops a novel improved SCN algorithm by incorporating the regularization (R) technique called RSC algorithm. Benchmark regression analysis and industrial tests show that compared with SCN, the established RSC model possesses higher model accuracy and better generalization ability.
- Main fan /
- switchover process /
- underground airflow quantity /
- stochastic configuration network /
- regularization

HTML全文

图 1 主通风机切换过程示意图

Fig. 1 Diagram of a main fan switchover process

下载: 全尺寸图片幻灯片

图 2 不同$C$下RSC-II和SCN-III算法对函数近似的训练与测试精度结果对比

Fig. 2 The result comparison of training and testing accuracy of RSC-II and SCN-III on function approximation with varying $C$

下载: 全尺寸图片幻灯片

图 3 (a)包含5 %异常值的1 000个函数近似训练样本和目标函数; (b)两种算法对测试数据的近似性能

Fig. 3 (a) 1 000 training samples containing 5 % outliers for function approximation and target function; (b) Approximation performance on the test dataset by two learning algorithms

下载: 全尺寸图片幻灯片

图 4 平均两种算法在测试数据集上的测试RMSE

Fig. 4 Average testing RMSE of the two algorithms on the test dataset

下载: 全尺寸图片幻灯片

图 5 不同$L_{\max}$下RSC-II和SCN-III算法对基准数据集的测试RMSE对比

Fig. 5 Test RMSE comparison of the RSC-II and SCN-III algorithms on four benchmark datasets with different $L_{\max}$

下载: 全尺寸图片幻灯片

图 6 平均两种算法在实际MFSP数据集上的测试RMSE

Fig. 6 Average testing RMSE of the two algorithms on the actual MFSP dataset

下载: 全尺寸图片幻灯片

图 7 $L_{\max} = 50$所对应的MFSP数据集的测试性能: (a) SCN-III; (b) RSC-II

Fig. 7 Test performance at $L_{\max} = 50$ on the actual MFSP dataset: (a) SCN-III; (b) RSC-II

下载: 全尺寸图片幻灯片

表 1 主通风机切换过程相关变量

Table 1 Related variables in the MFSP

变量	符号	单位
井下供给风量	$Q_{\rm 0}$	${\rm m^3/s}$
井下风阻	$R_{\rm 0}$	${\rm kg/m^7}$
一号垂直风门风量	$Q_{\rm 1c}$	${\rm m^3/s}$
一号垂直风门风阻	$R_{\rm 1c}$	${ \rm kg/m^7}$
一号水平风门风量	$Q_{\rm 1s}$	${\rm m^3/s}$
一号水平风门风阻	$R_{\rm 1s}$	${\rm kg/m^7}$
一号主通风机风量	$Q_{\rm 1m}$	${\rm m^3/s}$
一号主通风机压头	$H_{\rm 1d}$	${\rm Pa}$
二号垂直风门风量	$Q_{\rm 2c}$	${\rm m^3/s}$
二号垂直风门风阻	$R_{\rm 2c}$	${\rm kg/m^7}$
二号水平风门风量	$Q_{\rm 2s}$	${\rm m^3/s}$
二号水平风门风阻	$R_{\rm 2s}$	${\rm kg/m^7}$
二号主通风机风量	$Q_{\rm 2m}$	${\rm m^3/s}$
二号主通风机压头	$H_{\rm 2d}$	${\rm Pa}$

下载: 导出CSV

表 2 函数近似的性能比较

Table 2 Performance comparisons on the function approximation

算法	不同$L_{\max}$所对应的测试性能 (Mean, STD)
算法	50	70	90	110	130	150	170	190
SCN-III	0.0315, 0.0032	0.0329, 0.0033	0.0388, 0.0048	0.0396, 0.0032	0.0459, 0.0043	0.0512, 0.0058	0.0526, 0.0048	0.0557, 0.0095
RSC-II	0.0229, 0.0019	0.0209, 0.0013	0.0209, 0.0008	0.0209, 0.0003	0.0210, 0.0005	0.0211, 0.0004	0.0213, 0.0004	0.0218, 0.0003

下载: 导出CSV

表 3 回归数据集与参数设置

Table 3 Specifications of benchmark problems and some parameter settings

数据集	属性	输出	训练数据	测试数据	${\lambda_{\min}}$	${\Delta\lambda}$
Wine Quality	12	1	3918	980	1	0.1
Concrete	9	1	772	258	1	0.1
Yacht	7	1	227	81	1	1
Airfoil Self-noise	6	1	1352	151	1	0.1

下载: 导出CSV

表 4 对基准数据集的性能对比

Table 4 Performance comparisons on benchmark datasets

数据集	算法	不同$L_{\max}$所对应的测试性能 (Mean, STD)
数据集	算法	50	70	90	110	130	150	170	190
(a)	SCN-III	0.2476, 0.0233	0.2546, 0.0531	0.2543, 0.0492	0.2559, 0.0464	0.2908, 0.0612	0.2997, 0.0841	0.3480, 0.1484	0.3934, 0.1886
(a)	RSC-II	0.2232, 0.0011	0.2229, 0.0014	0.2229, 0.0016	0.2225, 0.0016	0.2231, 0.0017	0.2235, 0.0022	0.2236, 0.0021	0.2238, 0.0021
(b)	SCN-III	0.4017, 0.1060	0.5209, 0.1587	0.7120, 0.2247	0.7739, 0.2010	0.8640, 0.2324	1.3580, 0.7945	1.4025, 0.4526	1.6700, 0.5415
(b)	RSC-II	0.2248, 0.0068	0.2277, 0.0059	0.2295, 0.0094	0.2295, 0.0076	0.2323, 0.0089	0.2341, 0.0090	0.2384, 0.0063	0.2416, 0.0056
(c)	SCN-III	0.3167, 0.2289	0.3528, 0.1436	0.3820, 0.1781	0.6620, 0.2900	0.9062, 0.5600	1.2226, 0.3571	2.7450, 1.1718	4.0242, 1.0377
(c)	RSC-II	0.1308, 0.0209	0.1216, 0.0107	0.1197, 0.0152	0.1168, 0.0111	0.1164, 0.0118	0.1113, 0.0117	0.1056, 0.0112	0.1066, 0.0113
(d)	SCN-III	0.2607, 0.0313	0.3489, 0.1007	0.4332, 0.1219	0.6615, 0.2487	1.2668, 0.4213	1.4295, 0.5883	1.8264, 0.5194	2.0539, 0.9744
(d)	RSC-II	0.2281, 0.0091	0.2378, 0.0147	0.2332, 0.0153	0.2395, 0.0214	0.2443, 0.0279	0.2615, 0.0402	0.2437, 0.0234	0.2848, 0.0512

下载: 导出CSV

表 5 在实际MFSP数据集上的性能对比

Table 5 Performance comparisons on the actual MFSP dataset

算法	不同$L_{\max}$所对应的测试性能 (Mean, STD)
算法	50	110	170	230	270	330	370	410
SCN-III	0.0287, 0.0029	0.0494, 0.0039	0.0623, 0.0069	0.0758, 0.0063	0.0837, 0.0059	0.0983, 0.0107	0.1033, 0.0096	0.1146, 0.0095
RSC-II	0.0269, 0.0021	0.0316, 0.0017	0.0372, 0.0035	0.0411, 0.0046	0.0424, 0.0047	0.0454, 0.0066	0.0460, 0.0054	0.0509, 0.0077

下载: 导出CSV

参考文献(24)

[1]	杨校卫, 韩传功, 何胜良, 等. 一种煤矿老通风系统不停风倒机装置, 中国专利CN208456655U, 2019年2月1日
[2]	Chatterjee Arnab, Zhang Li-Jun, Xia Xiao-Hua. Optimization of mine ventilation fan speeds according to ventilation on demand and time of use tariff. Applied Energy, 2015, 146: 65-73 doi: 10.1016/j.apenergy.2015.01.134
[3]	Wu X Z, Ma X P, Ren Z H. Study on coal mine main fan automatic switchover aiming at ventilation unceasing and its numerical simulation. In: Proceedings of the 2nd International Asia Conference on Informatics in Control, Automation and Robotics. Wuhan, China: IEEE, 2010. 2: 227−230
[4]	Ge Heng-Qing, Ma Xiao-Ping, Wu Xin-Zhong, Liang Chun. Study on mine main fan switchover aiming at invariant ventilation based on nonlinear constrained programming. International Journal of Digital Content Technology and Its Applications, 2012, 6(17): 496-505 doi: 10.4156/jdcta.vol6.issue17.54
[5]	王前进, 代伟, 杨春雨, 马小平. 煤矿主通风机切换系统建模与分析. 煤炭学报, 2018, 43(S2): 606-614. DOI: 10.13225/j.cnki.jccs.2017.1509 Wang Qian-Jin, Dai Wei, Yang Chun-Yu, Ma Xiao-Ping. Modeling and analysis of coal mine main fan switchover system. Journal of China Coal Society, 2018, 43(S2): 606-614. DOI: 10.13225/j.cnki.jccs.2017.1509
[6]	Wang Qian-Jin, Ma Xiao-Ping, Yang Chun-Yu, Dai Wei. Modeling and control of mine main fan switchover system. ISA Transactions, 2019, 85: 189-199 doi: 10.1016/j.isatra.2018.10.024
[7]	任子晖, 王翠, 倪婷婷, 成江洋. 基于神经网络不停风倒机风量变化的研究. 煤炭技术, 2016, 35(11): 229-231 Ren Zi-Hui, Wang Cui, Ni Ting-Ting, Cheng Jiang-Yang. Research on air volume of reversing without stopping ventilation based on neural network. Coal Technology, 2016, 35(11): 229-231
[8]	Zou Wei-Dong, Xia Yuan-Qing, Li Hui-Fang. Fault diagnosis of tennessee-eastman process using orthogonal incremental extreme learning machine based on driving amount. IEEE Transactions on Cybernetics, 2018, 48(12): 3403-3410 doi: 10.1109/TCYB.2018.2830338
[9]	韩敏, 李德才. 基于替代函数及贝叶斯框架的1范数ELM算法. 自动化学报, 2011, 37(11): 1344-1350 Han Min, Li De-Cai. An norm 1 regularization term ELM algorithm based on surrogate function and bayesian framework. Acta Automatica Sinica, 2011, 37(11): 1344-1350
[10]	邹伟东, 夏元清. 基于压缩动量项的增量型ELM虚拟机能耗预测. 自动化学报, 2019, 45(7): 1290-1297 Zou Wei-Dong, Xia Yuan-Qing. Virtual machine power prediction using incremental extreme learning machine based on compression driving amount. Acta Automatica Sinica, 2019, 45(7): 1290-1297
[11]	陈晓云, 廖梦真. 基于稀疏和近邻保持的极限学习机降维. 自动化学报, 2019, 45(2): 325-333 Chen Xiao-Yun, Liao Meng-Zhen. Dimensionality reduction with extreme learning machine based on sparsity and neighborhood preserving. Acta Automatica Sinica, 2019, 45(2): 325-333
[12]	Li Ming, Wang Dian-Hui. Insights into randomized algorithms for neural networks: Practical issues and common pitfalls. Information Sciences, 2017, 382: 170-178
[13]	Wang Dian-Hui, Li Ming. Stochastic configuration networks: Fundamentals and algorithms. IEEE Transactions on Cybernetics, 2017, 47(10): 3466-3479 doi: 10.1109/TCYB.2017.2734043
[14]	王前进, 马小平, 张守田. PLC软冗余在通风机监控系统中的应用. 工矿自动化, 2014, 40(1): 93-96 Wang Qian-Jin, Ma Xiao-Ping, Zhang Shou-Tian. Application of PLC soft redundancy in monitoring and control system for ventilator. Industry and Mine Automation, 2014, 40(1): 93-96
[15]	Ge H Q, Ma X P, Wu X Z, Zhang J. Fuzzy PID control of mine main fan switchover aiming at invariant ventilation. In: Proceedings of the 2011 International Conference on Intelligence Science and Information Engineering. Wuhan, China: IEEE, 2011. 325−328
[16]	Wang Qian-Jin, Dai Wei, Ma Xiao-Ping, Yang Chun-Yu. Multiple models and neural networks based adaptive PID decoupling control of mine main fan switchover system. Iet Control Theory & Applications, 2018, 12(4): 446-455
[17]	Ge Heng-Qing, Xu Guang, Huang Jin-Xin, Ma Xiao-Ping. A mine main fans switchover system with lower air flow volatility based on improved particle swarm optimization algorithm. Advances in Mechanical Engineering, 2019, 11(3): 1687814019829281
[18]	L. Ray Allen, Couse Derek. Cement plant fan efficiency upgrades. IEEE Transactions on Industry Applications, 2017, 53(2): 1562-1568 doi: 10.1109/TIA.2016.2631526
[19]	Papar R, Szady A, Huffer W D, Martin V, McKane A. Increasing Energy Efficiency of Mine Ventilation Systems. Industrial Energy Analysis, Lawrence Berkeley National Laboratory, University of California, 1999.
[20]	Zhu Song, Chang Wen-Ting, Liu Dan. Almost sure exponential stability of stochastic fluid networks with nonlinear control. International Journal of Control, 2018, 91(2): 400-410 doi: 10.1080/00207179.2017.1282626
[21]	Camilleri Robert, Howey David A., McCulloch Malcolm D.. Predicting the temperature and flow distribution in a direct oil-cooled electrical machine with segmented stator. IEEE Transactions on Industrial Electronics, 2016, 63(1): 82-91 doi: 10.1109/TIE.2015.2465902
[22]	Abareshi Maryam, Hosseini Seyed Mahmood, Sani Ahmad Aftabi. A simple iterative method for water distribution network analysis. Applied Mathematical Modelling, 2017, 52: 274-287 doi: 10.1016/j.apm.2017.07.053
[23]	Gorban Alexander N., Tyukin Ivan Yu., Prokhorov Danil V., Sofeikov Konstantin I.. Approximation with random bases: Pro et Contra. Information Sciences, 2016, 364: 129-145
[24]	Lichman M. UCI machine learning repository [Online] available: http://archive.ics.uci.edu/ml, January 1, 2013