基于多模型智能组合算法的锅炉炉膛温度建模

doi:10.11949/0438-1157.20190037

摘要/Abstract

摘要：

炉膛温度是表征锅炉燃烧状态的重要参数，但是影响炉膛温度的参数多、机理复杂，导致难以建立准确的预测模型。针对这一问题，提出一种多模型智能组合算法(multi-model intelligent combination algorithm, MICA)实现对炉膛温度的建模预测。首先，对实际运行生产数据进行小波降噪，并结合机理分析和分类回归树（classification and regression tree，CART）算法选取预测模型输入参数。然后，通过多种数据驱动方法构建锅炉炉膛温度预测模型。最后，采用决策树C4.5算法建立多模型智能组合预测模型。基于实际生产数据的实验结果表明，所提出算法能够建立准确的炉膛温度预测模型。

关键词: 数据驱动, 炉膛温度, 小波, 智能组合, 算法, 模型, 预测

Abstract:

Furnace temperature is an important parameter which can reflect boiler combustion status. However, furnace temperature is affected by many parameters and the mechanism is complicated. Therefore, it is difficult to establish accurate prediction model. To solve this problem, a multi-model intelligent combination algorithm (MICA) is proposed to construct an accurate prediction model. First, the actual production data is pre-processed by wavelet denoising algorithm, and the model input variables were selected based on classification and regression trees algorithm and mechanism analysis. Then, several furnace temperature prediction models are constructed by many data-driven algorithms. Finally, a C4.5 algorithm is applied to combine these models into a multi-model intelligent combination model. The experimental results illustrate that the proposed algorithm can construct an accurate furnace temperature prediction model through actual operating data.

Key words: data-driven, furnace temperature, wavelet, intelligent combination, algorithm, model, prediction

中图分类号:

TK 227

唐振浩,张宝凯,曹生现,王恭,赵波. 基于多模型智能组合算法的锅炉炉膛温度建模[J]. 化工学报, 2019, 70(S2): 301-310.

Zhenhao TANG,Baokai ZHANG,Shengxian CAO,Gong WANG,Bo ZHAO. Furnace temperature modeling based on multi-model intelligent combination algorithm[J]. CIESC Journal, 2019, 70(S2): 301-310.

图/表 13

参考文献 30

1	ZhaoH, WangP H, PengX, et al. Constrained optimization of combustion at a coal-fired utility boiler using hybrid particle swarm optimization with invasive weed[C]//International Conference on Energy and Environment Technology. IEEE, 2009: 564-567.
2	LukaszS, KonradW, KonradS, et al. Optimization of combustion process in coal-fired power plant with utilization of acoustic system for in-furnace temperature measurement[J]. Applied Thermal Engineering, 2017, 123: 711-720.
3	HasanO, CalinZ, IbrahimD. Transient modeling of a gas-liquid piston-cylinder mechanism for low temperature energy conversion applications[J]. International Journal of Thermal Sciences, 2017, 111: 525-532.
4	蔡金锭, 鄢仁武. ARMA双谱分析与离散隐马尔可夫模型在电力电子电路故障诊断中的应用[J]. 中国电机工程学报, 2010, 30(24): 54-60.
	CaiJ D, YanR W. Fault diagnosis of power electronic circuit applying ARMA bispectrum and discrete hidden Markov model[J]. Proceedings of the CSEE, 2010, 30(24): 54-60.
5	HisashiT, YoshiyasuT, SeishoS. Using the ensemble Kalman filter for electricity load forecasting and analysis[J]. Energy, 2016, 104: 184-198.
6	曹鹏飞, 罗雄麟. 化工过程软测量建模方法研究进展[J]. 化工学报, 2013, 64(3): 788-800.
	CaoP F, LuoX L. Modeling of soft sensor for chemical process[J]. CIESC Journal, 2013, 64(3): 788-800.
7	钱晓山, 阳春华, 徐丽莎. 基于改进差分进化和最小二乘支持向量机的铝酸钠溶液浓度软测量[J]. 化工学报, 2013, 64(5): 1704-1709.
	QianX S, YangC H, XuL S. Soft sensor of sodium aluminate solution concentration based on improved differential evolution algorithm and LSSVM[J]. CIESC Journal, 2013, 64(5): 1704-1709.
8	岳毅宏, 韩文秀, 张伟波. 基于关联度的混沌序列局域加权线性回归预测法[J]. 中国电机工程学报, 2004, 24(11): 19-22.
	YueY H, HanW X, ZhangW B. Local adding-weight linear regression forecasting method of chaotic series based on degree of incidence[J]. Proceedings of the CSEE, 2004, 24(11): 19-22.
9	李强, 徐建政. 基于主观贝叶斯方法的电力系统故障诊断[J]. 电力系统自动化, 2007, 31(15): 46-50.
	LiQ, XuJ Z. Power system fault diagnosis based on subjective Bayesian approach[J]. Automation of Electric Power Systems, 2007, 31(15): 46-50.
10	唐志杰, 唐朝晖, 朱红求. 一种基于多模型融合软测量建模方法[J]. 化工学报, 2011, 62(8): 2248-2252.
	TangZ J, TangZ H, ZhuH Q. A multi-model fusion soft sensor modeling method[J]. CIESC Journal, 2011, 62(8): 2248-2252.
11	王锋, 李宏光, 臧灏. 基于Logistic和ARMA模型的过程报警预测[J]. 化工学报, 2012, 63(9): 2941-2947.
	WangF, LiH G, ZangH. Process alarm prognosis based on Logistic and ARMA models[J]. CIESC Journal, 2012, 63(9): 2941-2947.
12	朱鹏飞, 夏陆岳, 潘海天. 基于改进Kalman滤波算法的多模型融合建模方法[J]. 化工学报, 2015, 66(4): 1388-1394.
	ZhuP F, XiaL Y, PanH T. Multi-model fusion modeling method based on improved Kalman filtering algorithm[J]. CIESC Journal, 2015, 66(4): 1388-1394.
13	闫哲, 张卜升, 刘永忠. 基于Kalman滤波的换热器非线性状态参数校正[J]. 化工学报, 2012, 63(2): 523-529.
	YanZ, ZhangB S, LiuY Z. Data reconciliation for nonlinear state parameters of heat exchangers using Kalman filtering[J]. CIESC Journal, 2012, 63(2): 523-529.
14	顾燕萍, 赵文杰, 吴占松. 基于最小二乘支持向量机的电站锅炉燃烧优化[J]. 中国电机工程学报, 2010, 30(17): 91-97.
	GuY P, ZhaoW J, WuZ S. Combustion optimization for utility boiler based on least square-support vector machine[J]. Proceedings of the CSEE, 2010, 30(17): 91-97.
15	TangZ H, ZhangH Y, CheP, et al. Data analytics based dual-optimized adaptive model predictive control for the power plant boiler[J]. Mathematical Problems in Engineering, 2017, 2017: 1-9.
16	王占能, 徐祖华, 赵均, 等. 基于负荷划分数据和支持向量机的火电厂燃烧过程建模[J]. 化工学报, 2013, 64(12): 4496-4502.
	WangZ N, XuZ H, ZhaoJ, et al. Coal-fired power plant boiler combustion process modeling based on support vector machine and load data division[J]. CIESC Journal, 2013, 64(12): 4496-4502.
17	韩改堂, 乔俊飞, 韩红桂. 基于递归模糊神经网络的污水处理控制方法[J]. 化工学报, 2016, 67(3): 954-959.
	HanG T, QiaoJ F, HanH G. Wastewater treatment control method based on recurrent fuzzy neural network[J]. CIESC Journal, 2016, 67(3): 954-959.
18	罗滇生, 姚建刚, 何洪英, 等. 基于自适应滚动优化的电力负荷多模型组合预测系统的研究与开发[J]. 中国电机工程学报, 2003, 23(5): 59-62.
	LuoD S, YaoJ G, HeH Y, et al. Research and development of multi-model combining load forecasting system based on self-adaptive rolling optimization[J]. Proceedings of the CSEE, 2003, 23(5): 59-62.
19	CheJ X. Optimal sub-models selection algorithm for combination forecasting model[J]. Neurocomputing, 2015, 151: 364-375.
20	HaiX V, DurlofskyL J. Regularized kernel PCA for the efficient parameterization of complex geological models[J]. Journal of Computational Physics, 2016, 322: 859-881.
21	童楚东, 史旭华. 基于互信息的PCA方法及其在过程监测中的应用[J]. 化工学报, 2015, 66(10): 4101-4106.
	TongC D, ShiX H. Mutual information based PCA algorithm with application in process monitoring[J]. CIESC Journal, 2015, 66(10): 4101-4106.
22	朱群雄, 陈希, 贺彦林, 等. 基于PCA-DEA的乙烯装置能效分析[J]. 化工学报, 2015, 66(1): 278-283.
	ZhuQ X, ChenX, HeY L, et al. Energy efficiency analysis for ethylene plant based on PCA-DEA[J]. CIESC Journal, 2015, 66(1): 278-283.
23	唐贵基, 邓飞跃, 张超, 等. 基于倒谱预白化和奇异值分解的滚动轴承故障特征提取方法[J]. 中国电机工程学报, 2014, 34(35): 6355-6361.
	TangG J, DengF Y, ZhangC, et al. Extraction method of rolling bearing fault feature based on cepstrum pre-whitening and singular value decomposition[J]. Proceedings of the CSEE, 2014, 34(35): 6355-6361.
24	唐炬, 董玉林, 樊雷, 等. 基于Hankel矩阵的复小波-奇异值分解法提取局部放电特征信息[J]. 中国电机工程学报, 2015, 35(7): 1808-1817.
	TangJ, DongY L, FanL, et al. Feature information extraction of partial discharge signal with complex wavelet transform and singular value decomposition based on Hankel matrix[J]. Proceedings of the CSEE, 2015, 35(7): 1808-1817.
25	黄南天, 彭华, 蔡国伟, 等. 电能质量复合扰动特征选择与最优决策树构建[J]. 中国电机工程学报, 2017, 37(3): 776-786.
	HuangN T, PengH, CaiG W, et al. Feature selection and optimal decision tree construction of complex power quality disturbances[J]. Proceedings of the CSEE, 2017, 37(3): 776-786.
26	MonalisaM, SantanuS, PradyutB, et al. Efficient classification of ventricular arrhythmias using feature selection and C4.5 classifier[J]. Biomedical Signal Processing and Control, 2018, 44: 200-208.
27	董明, 屈彦明, 周孟戈. 基于组合决策树的油浸式电力变压器故障诊断[J]. 中国电机工程学报, 2005, 25(16): 35-41.
	DongM, QuY M, ZhouM G. Fault diagnosis of oil-immersed power transformer using combinatorial decision tree[J]. Proceedings of the CSEE, 2005, 25(16): 35-41.
28	栗然, 刘宇, 黎静华, 等. 基于改进决策树算法的日特征负荷预测研究[J]. 中国电机工程学报, 2005, 25(23): 36-41.
	LiR, LiuY, LiJ H, et al. Study on the daily characteristic load forecasting based on the optimized algorithm of decision tree[J]. Proceedings of the CSEE, 2005, 25(23): 36-41.
29	秦品乐, 林焰, 陈明. 基于平移不变小波阈值算法的经验模态分解方法[J]. 仪器仪表学报, 2008, 29(12): 2637-2641.
	QinP L, LinY, ChenM. Empirical mode decomposition method based on wavelet with translation invariance algorithm[J]. Chinese Journal of Scientific Instrument, 2008, 29(12): 2637-2641.
30	MeiL X, MinH, HongF L. Wavelet-denoising multiple echo state networks for multivariate time series prediction[J]. Information Sciences, 2018, 465: 439-458.

序号	变量名	单位	变化范围	标签
1	烟气含氧量	%	[2.431—2.892]	F₁
2	主蒸汽流量	t/h	[183.078—251.084]	F₂
3	主蒸汽压力	MPa	[12.976—13.316]	SP
4	主蒸汽温度	℃	[533.611—542.525]	ST
5	机组负荷	MW	[109.863—145.578]	P_a
6	再热器压力	MPa	[1.290—1.813]	RP
7	燃料量	t/h	[132.382—155.044]	F₃
8	送风量	t/h	[143.255—176.080]	F₄
9	给水流量	t/h	[361.375—505.756]	F₅
10	炉膛负压	Pa	[-62.732—-20.063]	FP
11	过热减温水流量	t/h	[2.205—10.462]	F₆
12	锅炉蒸发量	t/h	[373.350—512.889]	F₇
13	再热蒸汽温度	℃	[525.254—542.707]	RT
14	再热减温水流量	t/h	[3.958—7.734]	F₈
15	排烟温度	℃	[161.903—171.807]	GT
16	送风挡板开度	%	[3.399—40.828]	F₉

序号	变量名	单位	变化范围	标签
1	烟气含氧量	%	[2.431—2.892]	F₁
2	主蒸汽流量	t/h	[183.078—251.084]	F₂
3	主蒸汽压力	MPa	[12.976—13.316]	SP
4	主蒸汽温度	℃	[533.611—542.525]	ST
5	机组负荷	MW	[109.863—145.578]	P_a
6	再热器压力	MPa	[1.290—1.813]	RP
7	燃料量	t/h	[132.382—155.044]	F₃
8	送风量	t/h	[143.255—176.080]	F₄
9	给水流量	t/h	[361.375—505.756]	F₅
10	炉膛负压	Pa	[-62.732—-20.063]	FP
11	过热减温水流量	t/h	[2.205—10.462]	F₆
12	锅炉蒸发量	t/h	[373.350—512.889]	F₇
13	再热蒸汽温度	℃	[525.254—542.707]	RT
14	再热减温水流量	t/h	[3.958—7.734]	F₈
15	排烟温度	℃	[161.903—171.807]	GT
16	送风挡板开度	%	[3.399—40.828]	F₉

序号	测量值	预测结果				类别
序号	测量值	BP	RBF	MLP	LSSVM	类别
1	803.9	802.4	803.8	802.3	803.5	2
2	805.1	804.9	803.6	803.9	804.5	1
3	805.6	804.2	804.9	804.1	805.0	4
4	803.2	802.4	802.6	801.4	803.0	4
?	?	?	?	?	?	?
298	796.2	795.7	796.6	796.3	796.4	3
299	796.4	795.1	795.9	795.8	795.6	2
300	796.8	796.2	795.5	796.4	796.6	4

序号	测量值	预测结果				类别
序号	测量值	BP	RBF	MLP	LSSVM	类别
1	803.9	802.4	803.8	802.3	803.5	2
2	805.1	804.9	803.6	803.9	804.5	1
3	805.6	804.2	804.9	804.1	805.0	4
4	803.2	802.4	802.6	801.4	803.0	4
?	?	?	?	?	?	?
298	796.2	795.7	796.6	796.3	796.4	3
299	796.4	795.1	795.9	795.8	795.6	2
300	796.8	796.2	795.5	796.4	796.6	4

时间	标签	输入变量	输出变量	训练集	测试集	验证集	采用间隔	备注（温度变化范围）
2016\03\01 08:00~03\02 02:18	T₁	10	1	700	300	100	60s	757℃上升至835℃
2016\03\03 10:00~03\04 06:18	T₂	10	1	700	300	100	60s	754~805℃范围内变化
2016\03\10 07:00~03\11 01:18	T₃	10	1	700	300	100	60s	845℃下降至779℃