Online Measurement of Outlet Temperature of Gasifier Based on Data Driven
-
摘要: 为实时监控气流床气化炉的气化温度和运行状态，收集气化炉激冷系统和反应系统等系统的可测量数据，采用理论计算模型和遗传算法改进的BP(GABP)神经网络模型对气化炉出口温度进行预测，并与工业测量数据进行对比分析。结果表明，由激冷系统理论计算模型可以得到气化炉出口温度，但因测量参数敏感度低，导致温度预测精度和稳定性较差。采用GABP神经网络模型总体上可以提高预测温度的精度和稳定性，但反应系统中由于煤量波动和煤质数据缺乏等原因，导致部分区间预测误差较大；采用激冷系统参数可大幅提高绝大部分区间内温度的预测精度，预测误差保持在15 K以下，可满足不同工况下的气化炉温度实时在线监测需要。Abstract: Gasification temperature is the most important operating parameter of entrained flow gasifier. However, reliable methods to long-term measure the gasification temperature of coal gasifier units remain elusive. In order to monitor the operation state of entrained flow gasifier in real time and ensure the safe and stable operation of gasification system, measurable data including gasifier cooling system and reaction system are collected. The outlet temperature of gasifier was predicted by a theoretical calculation model and BP neural based on the genetic algorithm model (GABP). The results were compared to those obtained from industrial measurement. The outlet temperature of gasifier can be obtained by the theoretical calculation of quench system, while the accuracy and stability of the prediction results are unsatisfactory due to the low sensitivity of measurement parameters. GABP neural network model greatly improves the prediction performances. Based on the gasification chamber parameters, the prediction error is large due to the fluctuation of coal water slurry flow rate and a lack of coal property data. Using quench system parameters as the input of GABP neural network greatly improves the prediction accuracy, and the absolute value of the prediction error is less than 15 K. Both the train set and verification set have excellent prediction results, and the average absolute error of GABP model with quench system parameters as input is about 5 K. The GABP model has good performances in the face of complex working conditions. When predictions are carried out under different conditions, the results under steady and variable coal loads are produced with good prediction precision and stability, which meet the requirements of online monitoring of gasifier temperature.
-
Key words:
- gasification /
- quench system /
- neural network /
- genetic algorithm /
- prediction
-
图 3 理论计算相关参数的敏感性分析
Figure 3. Sensitivity analysis of relevant factors calculated theoretically
1—Syngas pressure of quench chamber outlet；2—Black water temperature；3 —Syngas temperature of quench chamber outlet；4—Syngas temperature of water scrubber outlet
图 6 理论计算的出口温度(a)及误差分布(b)
Figure 6. Outlet temperature (a) and error distribution (b) of theoretical calculation
图 7 激冷系统参数对出口温度的预测结果（a~b）及GABP验证集的误差分布（c）
Figure 7. Predict results of outlet temperature (a~b) and error distribution of GABP-test (c) with quench system
图 8 反应系统参数对出口温度的预测结果（a~b）及GABP验证集的误差分布（c）
Figure 8. Predict result of outlet temperature (a~b) and error distribution of GABP-test (c) with reaction system
图 9 稳定负荷(a)与变化负荷(b)件下温度预测情况对比
Figure 9. Comparison of temperature prediction under steady load (a) and variable load (b)
图 10 不同预测方法的气化炉出口温度误差分布频率
Figure 10. Frequency of temperature error distribution in gasifier outlet by different prediction methods
表 1 气化炉激冷系统相关参数
Table 1. Relevant parameters of quench system of gasifier
Parameter Parameter name Q1 Syngas temperature of water scrubber outlet Q2 Syngas volume of water scrubber outlet Q3 Syngas pressure of water scrubber outlet Q4 Quenching water flow Q5 Quenching water temperature Q6 Black water temperature Q7 Syngas temperature of quench chamber outlet Q8 Syngas pressure of quench chamber outlet Q9 Liquid level of quench chamber 
下载: 导出CSV
表 2 气化炉反应系统相关参数
Table 2. Relevant parameters of reaction system of gasifier
Parameter Parameter name R1 Differential pressure of slag tap R2 Gasification pressure R3 Coal slurry flow rate R4 Oxygen flow rate
下载: 导出CSV
表 3 气化炉参数与出口温度变化关系
Table 3. Variation of gasifier parameters and outlet temperature
Parameter $\Delta T/{\rm{K}}$ R4 142.4 Q7 71.7 R3 42.8 Q5 19.1 Q6 18.5
下载: 导出CSV
-
[1] HIGMAN C, VAN B M. Gasification[M]. Amsterdam: Gulf Professional Publications, 2008. [2] 刘霞, 田原宇, 乔英云. 国内外气流床煤气化技术发展概述[J]. 化工进展, 2010, 29(S2): 120-124. [3] 郭庆华, 卫俊涛, 龚岩, 等. 多喷嘴对置式气流床气化炉内热态行为的研究进展[J]. 煤炭学报, 2020, 45(1): 403-413. [4] GONG X, LU W, GUO X, et al. Pilot-scale comparison investigation of different entrained-flow gasification technologies and prediction on industrial-scale gasification performance[J]. Fuel, 2014, 219(8): 37-44. [5] 管蕾. 激冷式气流床粉煤气化炉模拟研究[D]. 上海: 华东理工大学, 2016. [6] ZHANG B, JIN J, LIU H. Modeling study of the slag behaviors and SiC refractory wall corrosion on the top cone of a membrane wall entrained-flow gasifier[J]. Energy Fuels, 2020, 34(10): 12440-12448. doi: 10.1021/acs.energyfuels.0c02477 [7] WANG H, LUIS A, RICARDEZ S. Dynamic optimization of a pilot-scale entrained-flow gasifier using artificial recurrent neural networks[J]. Fuel, 2020, 272: 117731. doi: 10.1016/j.fuel.2020.117731 [8] NAVID K, ZHOU A, NAZAM M, et al. Modelling of municipal solid waste gasification using an optimised ensemble soft computing model[J]. Fuel, 2021, 289: 119903. doi: 10.1016/j.fuel.2020.119903 [9] ALI Y, OZGUN Y. An artificial intelligence based approach to predicting syngas composition for downdraft biomass gasification[J]. Energy, 2018, 165(A): 895-901. [10] MAHSHAD V, KATHRYN M, LUIS A, et al. State estimation and sensor location for entrained-flow gasification systems using Kalman filter[J]. Control Engineering Practice, 2021, 108: 104702. doi: 10.1016/j.conengprac.2020.104702 [11] WANG H, CHAFFART D, RICARDEZ-SANDOVAL L A. Modelling and optimization of a pilot-scale entrained-flow gasifier using artificial neural networks[J]. Energy, 2019, 188: 116076. doi: 10.1016/j.energy.2019.116076 [12] LI G, LIU Z, LI J, et al. Application of general regression neural network to model a novel integrated fluidized bed gasifier[J]. International Journal of Hydrogen Energy, 2018, 43(11): 5512-5521. doi: 10.1016/j.ijhydene.2018.01.130 [13] Ö GREN Y, TÓTH P, GARAMI A, et al. Development of a vision-based soft sensor for estimating equivalence ratio and major species concentration in entrained flow biomass gasification reactors[J]. Applied Energy, 2018, 226: 450-460. doi: 10.1016/j.apenergy.2018.06.007 [14] WANG K, ZHANG J, SHANG C, et al. Operation optimization of shell coal gasification process based on convolutional neural network models[J]. Applied Energy, 2021, 292: 116847. doi: 10.1016/j.apenergy.2021.116847 [15] 孙钟华, 代正华, 周志杰, 等. 灰含量及助熔剂对气流床粉煤气化炉性能的影响[J]. 中国电机工程学报, 2011, 31(20): 7-12. [16] PAN L, NOVAK L, LEHKY D, et al. Neural network ensemble-based sensitivity analysis in structural engineering: Comparison of selected methods and the influence of statistical correction[J]. Computers and Structures, 2021, 242: 106376. doi: 10.1016/j.compstruc.2020.106376 [17] 左锋, 王玺. 化工测量及仪表[M]. 第4版. 北京: 化学工业出版社, 2020. [18] 冷仓田, 王德祯, 周邵萍. 有源噪声控制中基于神经网络的次级通道辨识优化[J]. 华东理工大学学报(自然科学版), 2021, 47(6): 761-768. [19] CYBENKO G. Approximation by superpositions of a sigmoidal function[J]. Mathmatics Control Signals System 2, 1989, 2(4): 303-314. doi: 10.1007/BF02551274 [20] Wang Y, SHENG T, HE L, et al. Calibration method of magnetometer based on BP neural network[J]. Journal of Computer and Communications, 2020, 8(6): 31-41. doi: 10.4236/jcc.2020.86004 [21] WANG X, MA L, WANG Y, et al. Research and application of BP neural network in specific power harmonic detection[J]. Advanced Materials Research, 2012, 588: 379-383. [22] WILL S. Fintech model: The random neural network with genetic algorithm[J]. Procedia Computer Science, 2018, 126: 537-546. doi: 10.1016/j.procs.2018.07.288 [23] MAHYA M, ALIREZA A, MASOUD R. Transmission and generation expansion planning of energy hub by an improved genetic algorithm[J]. Energy Sources: Part A. Recovery, Utilization, and Environmental Effects, 2019, 41(24): 3112-3126. doi: 10.1080/15567036.2019.1568640 [24] 都月, 孟晓辰, 祝连庆. 遗传算法和神经网络的重叠光谱解析[J]. 光谱学与光谱分析, 2020, 40(7): 2066-2072. -
点击查看大图
计量
- 文章访问数: 225
- HTML全文浏览量: 138
- PDF下载量: 83
- 被引次数: 0
