基于BP神经网络和最小二乘支持向量机的灰熔点预测和对比

doi:10.12096/j.2096-4528.pgt.20003

摘要/Abstract

摘要：

为了预测燃煤锅炉受热面的结渣情况，以灰成分金属氧化物、煤灰SO₃含量以及结渣评判指标为自变量，灰熔点变形温度(deformation temperature，DT)和软化温度(softening temperature，ST)为因变量，建立了BP神经网络(BP neural network，BPNN)和最小二乘支持向量机(least squares support vector machine，LSSVM)的灰熔点预测模型。回归分析和误差分析结果表明：针对样本量多的DT预测过程，2种模型精度接近，预测结果置信度均达到95%，相关系数均约为0.92，平均相对误差均约为3.4%；针对样本量较少的ST预测过程，LSSVM模型预测效果较优，相关系数为0.950 52，高于BPNN模型的0.904 26，平均相对误差为4.98%，并且大误差点个数少于BPNN模型。因此，LSSVM模型能够更准确预测飞灰的DT和ST。

关键词: BP神经网络(BPNN), 最小二乘支持向量机(LSSVM), 灰熔点, 灰成分, 结渣评判指标

Abstract:

To predict the slagging on heating surface of coal-fired boilers, BP neural network (BPNN) and least squares support vector machine (LSSVM) prediction models were established to predict ash fusion temperature, deformation temperature (DT) and softening temperature (ST). The models take ash metal oxide, SO₃ content of ash and slagging evaluation indexes as independent variables, and take DT and ST as dependent variables. Regression analysis and error analysis show that when predicting DT with a large number of samples, the prediction accuracy of the two models is similar, and the confidence of prediction is over 95%. The correlation coefficients are both about 0.92, and the average relative errors are about 3.4%. When predicting ST with less samples, LSSVM model is better with a correlation coefficient of 0.950 52, which is higher than 0.904 26 of BPNN model. The average relative error is 4.98%, and the number of large error points is less than the BPNN model. Therefore, LSSVM model can predict DT and ST of fly ash more accurately.

Key words: BP neural network (BPNN), least squares support vector machine (LSSVM), ash fusion point, ash composition, slagging evaluation index

中图分类号:

TK 223

时浩, 肖海平, 刘彦鹏. 基于BP神经网络和最小二乘支持向量机的灰熔点预测和对比[J]. 发电技术, 2022, 43(1): 139-146.

Hao SHI, Haiping XIAO, Yanpeng LIU. Prediction and Comparison of Ash Fusion Temperatures Based on BP Neural Network and Least Squares Support Vector Machine[J]. Power Generation Technology, 2022, 43(1): 139-146.

图/表 11

参考文献 16

1	VASSILEV S V， KITANO K， TAKEDA S，et al ．Influence of mineral and chemical composition of coal ashes on their fusibility[J]．Fuel，1995，45(1)：27-51． doi:10.1016/0378-3820(95)00032-3
2	VASSILEV S V， KITANO K， VASSILEVA C G ．Relations between ash yield and chemical and mineral composition of coals[J]．Fuel，1997，76 (1)：3-8． doi:10.1016/s0016-2361(96)00181-0
3	操岳峰，李寒旭，胡洋，等．煤灰熔点数学模型的研究[J]．煤炭与化工，2017，40(10)：108-111．
	CAO Y F， LI H X， HU Y，et al ．Study on mathematical model of coal ash fusion temperature[J]．Coal and Chemical Industry，2017，40(10)：108-111．
4	CHEHREH C S， MAKAREMI S ．Explaining the relationship between common coal analyses and Afghan coal parameters using statistical modeling methods[J]．Fuel Processing Technology，2013，110：79-85． doi:10.1016/j.fuproc.2012.11.005
5	林德平．支持向量机在预测配煤灰熔点中的应用[J]．电力科学与工程，2015，31(8)：66-70． doi:10.3969/j.issn.1672-0792.2015.08.012
	LIN D P ．The application of support vector machine in predicting ash fusion temperature of blended coals[J]．Electric Power Science and Engineering，2015，31(8)：66-70． doi:10.3969/j.issn.1672-0792.2015.08.012
6	李清毅，周昊，林阿平，等．基于网格搜索和支持向量机的灰熔点预测[J]．浙江大学学报(工学版)，2011，45(12)：2181-2187． doi:10.3785/j.issn.1008-973X.2011.12.017
	LI Q Y， ZHOU H， LIN E P，et al ．Prediction of ash fusion temperature based on grid search and support vector machine[J]．Journal of Zhejiang University (Engineering Science)，2011，45(12)：2181-2187． doi:10.3785/j.issn.1008-973X.2011.12.017
7	赵显桥，吴胜杰，何国亮，等．支持向量机灰熔点预测模型研究[J]．热能动力工程，2011，26(4)：436-439． doi:10.1007/s12583-011-0163-z
	ZHAO X Q， WU S J， HE G L，et al ．Study of a support vector machine-based model for predicting melting points of ash[J]．Journal of Engineering for Thermal Energy and Power，2011，26(4)：436-439． doi:10.1007/s12583-011-0163-z
8	LIANG W， WANG G W， NING X J，et al ．Application of BP neural network to the prediction of coal ash melting characteristic temperature[J]．Fuel，2020，260：116324． doi:10.1016/j.fuel.2019.116324
9	蒋绍坚，付国富，黄靓云，等．基于Elman神经网络的烟煤与生物质混烧灰熔点预测[J]．中南大学学报(自然科学版)，2016，47(12)：4240-4247． doi:10.11817/j.issn.1672-7207.2016.12.036
	JIANG S J， FU G F， HUANG L Y，et al ．Forecasting on ash fusion temperatures of bituminous coal and biomass co-firing based on Elman neural network[J]．Journal of Central South University (Science and Technology)，2016，47(12)：4240-4247． doi:10.11817/j.issn.1672-7207.2016.12.036
10	ZHANG X， ZHANG H， NA Y ．Transformation of sodium during the ashing of Zhundong coal[J]．Procedia Engineering，2015，102：305-314． doi:10.1016/j.proeng.2015.01.147
11	WU H， YRJAS P， VAINIKKA P，et al ．Sulfation of alkali halides in a bench-scale bubbling fluidized bed reactor[J]．Fuel，2016，177：173-179． doi:10.1016/j.fuel.2016.03.005
12	刘彦鹏．蚁群优化算法的理论研究及其应用[D]．杭州：浙江大学，2007．
	LIU Y P ．Research on ant colony optimization and its application[D]．Hangzhou：Zhejiang University，2007．
13	王陆顺．基于神经网络的混煤灰熔点研究[D]．上海：上海交通大学，2016． doi:10.1007/s12204-016-1724-1
	WANG L S ．Study on the ash fusion temperature of blended coals based on neural network[D]．Shanghai：Shanghai Jiaotong University，2016． doi:10.1007/s12204-016-1724-1
14	李青．基于神经网络煤的发热量和灰熔点预测[D]．北京：华北电力大学，2012．
	LI Q ．Predicting calorific value and ash fusion temperature of coal based on neural network[D]．Beijing：North China Electric Power University，2012．
15	李颖．基于非线性理论的动力配煤模型的研究[D]．杭州：浙江大学，2002．
	LI Y ．Research on the model of power coal blending based on nonlinear theory[D]．Hangzhou：Zhejiang University，2002．
16	陈裕辉．基于支持向量机煤灰熔融特性及低温受热面积灰监测[D]．北京：华北电力大学，2019． doi:10.1016/j.epsr.2019.01.030
	CHEN Y H ．Ash fusion characteristics and fouling monitoring of low temperature heating surface based on support vector machine[D]．Beijing：North China Electric Power University，2019． doi:10.1016/j.epsr.2019.01.030

序号	DT/℃	ST/℃	w(SiO₂)/%	w(Al₂O₃)/%	w(Fe₂O₃)/%	w(TiO₂)/%	w(CaO)/%	w(MgO)/%	w(Na₂O)/%	w(K₂O)/%	w(SO₃)/%
1	1 250	1 268	49.09	16.59	11.11	0.81	7.58	1.90	0.44	1.40	7.26
2	1 206	1 210	32.65	12.07	9.36	0.49	23.86	3.38	0.84	0.71	13.22
3	1 278	1 314	61.42	16.92	7.52	4.76	0.71	1.24	0.48	0.09	4.68
4	1 472	1 485	48.19	25.80	4.32	1.28	6.90	1.02	0.50	0.81	5.25
5	1 476	1 486	50.51	21.42	4.29	1.33	4.50	0.64	0.50	0.73	4.76
6	1 270	1 320	56.29	23.66	8.86	0.58	5.10	0.96	0.37	1.03	2.40
7	1 457	1 471	52.56	21.51	4.48	1.35	5.14	0.70	0.49	0.78	4.92

序号	DT/℃	ST/℃	w(SiO₂)/%	w(Al₂O₃)/%	w(Fe₂O₃)/%	w(TiO₂)/%	w(CaO)/%	w(MgO)/%	w(Na₂O)/%	w(K₂O)/%	w(SO₃)/%
1	1 250	1 268	49.09	16.59	11.11	0.81	7.58	1.90	0.44	1.40	7.26
2	1 206	1 210	32.65	12.07	9.36	0.49	23.86	3.38	0.84	0.71	13.22
3	1 278	1 314	61.42	16.92	7.52	4.76	0.71	1.24	0.48	0.09	4.68
4	1 472	1 485	48.19	25.80	4.32	1.28	6.90	1.02	0.50	0.81	5.25
5	1 476	1 486	50.51	21.42	4.29	1.33	4.50	0.64	0.50	0.73	4.76
6	1 270	1 320	56.29	23.66	8.86	0.58	5.10	0.96	0.37	1.03	2.40
7	1 457	1 471	52.56	21.51	4.48	1.35	5.14	0.70	0.49	0.78	4.92

序号	DT/℃	w(SiO₂)/%	w(Al₂O₃)/%	w(Fe₂O₃)/%	w(TiO₂)/%	w(CaO)/%	w(MgO)/%	w(Na₂O)/%	w(K₂O)/%	w(SO₃)/%
1	1 310	53.26	33.22	4.04	1.24	3.04	1.90	0.21	0.71	1.10
2	1 290	53.26	32.62	4.48	1.28	3.27	1.68	0.20	0.72	1.52
3	1 173	44.43	14.30	10.00	0.89	17.16	2.03	0.82	1.52	9.56
4	1 179	43.87	16.33	9.41	0.73	15.74	2.10	1.19	1.16	9.44
…	…	…	…	…	…	…	…	…	…	…
88	1 140	38.60	11.40	13.53	0.71	19.15	2.61	0.89	1.00	11.27
89	1 170	49.18	18.60	9.13	0.96	15.57	0.84	0.80	1.24	2.65
90	1 460	54.66	28.65	6.70	0.98	4.16	0.52	0.69	1.86	0.85
91	1 280	56.25	25.02	5.20	1.00	5.04	1.56	0.54	1.74	2.32

序号	DT/℃	w(SiO₂)/%	w(Al₂O₃)/%	w(Fe₂O₃)/%	w(TiO₂)/%	w(CaO)/%	w(MgO)/%	w(Na₂O)/%	w(K₂O)/%	w(SO₃)/%
1	1 310	53.26	33.22	4.04	1.24	3.04	1.90	0.21	0.71	1.10
2	1 290	53.26	32.62	4.48	1.28	3.27	1.68	0.20	0.72	1.52
3	1 173	44.43	14.30	10.00	0.89	17.16	2.03	0.82	1.52	9.56
4	1 179	43.87	16.33	9.41	0.73	15.74	2.10	1.19	1.16	9.44
…	…	…	…	…	…	…	…	…	…	…
88	1 140	38.60	11.40	13.53	0.71	19.15	2.61	0.89	1.00	11.27
89	1 170	49.18	18.60	9.13	0.96	15.57	0.84	0.80	1.24	2.65
90	1 460	54.66	28.65	6.70	0.98	4.16	0.52	0.69	1.86	0.85
91	1 280	56.25	25.02	5.20	1.00	5.04	1.56	0.54	1.74	2.32

序号	DT/℃	碱酸比	硅铝比	沾污指数	硅铝和
1	1 310	0.11	1.60	0.02	86.48
2	1 290	0.12	1.63	0.02	85.88
3	1 173	0.53	3.11	0.43	58.73
4	1 179	0.49	2.69	0.58	60.20
…	…	…	…	…	…
88	1 140	0.73	3.39	1.39	50.00
89	1 170	0.40	2.64	0.82	67.78
90	1 460	0.17	1.91	0.43	83.31
91	1 280	0.17	2.25	0.39	81.27