Research on dynamic prediction of NO <i><sub>x</sub></i> emission of thermal power plants based on PSO-XGBoost ensemble algorithm

SUN Zunqiang; TIAN Yichun; MA Xiuyuan; ZHENG Chenghang; SU Nan; YANG Hongmin

doi:10.13205/j.hjgc.202512020

Volume 43 Issue 12

Dec. 2025

Turn off MathJax

Article Contents

Article Navigation > ENVIRONMENTAL ENGINEERING > 2025 > 43(12): 178-185

SUN Zunqiang, TIAN Yichun, MA Xiuyuan, ZHENG Chenghang, SU Nan, YANG Hongmin. Research on dynamic prediction of NO x emission of thermal power plants based on PSO-XGBoost ensemble algorithm[J]. ENVIRONMENTAL ENGINEERING , 2025, 43(12): 178-185. doi: 10.13205/j.hjgc.202512020

Citation:

SUN Zunqiang, TIAN Yichun, MA Xiuyuan, ZHENG Chenghang, SU Nan, YANG Hongmin. Research on dynamic prediction of NO _x emission of thermal power plants based on PSO-XGBoost ensemble algorithm[J]. ENVIRONMENTAL ENGINEERING , 2025, 43(12): 178-185. doi: 10.13205/j.hjgc.202512020

Citation:

PDF( 1648 KB)

Research on dynamic prediction of NO _x emission of thermal power plants based on PSO-XGBoost ensemble algorithm

doi: 10.13205/j.hjgc.202512020

1. State Power Environmental Protection Research Institute, Nanjing 210031, China;
2. National Institute of Clean-and-Low-Carbon Energy, Beijing 102200, China;
3. College of Energy Engineering, Zhejiang University, Hangzhou 310027, China;
4. School of Energy and Mechanical Engineering, Nanjing Normal University, Nanjing 210042, China

Received Date: 2024-01-14
Accepted Date: 2024-03-15
Rev Recd Date: 2024-02-28

Available Online: 2026-01-09

Abstract

Abstract

The single model for NO _x generation in thermal power plants can only be trained based on partial data features， failing to fully explore and utilize the potential value of the data. To effectively predict NO _x emissions from coal-fired power plants， this paper proposed a feature selection method based on Pearson correlation analysis and an extreme gradient boosting （XGBoost） algorithm optimized by particle swarm optimization （PSO）. Firstly， Pearson correlation analysis was used to calculate the correlation coefficients of various features and conduct feature selection. Secondly， the particle swarm optimization algorithm was employed to fine-tune the hyperparameters of the XGBoost model， ensuring its robustness and generalization ability under different operating conditions. Finally， the prediction results were compared with those of other machine learning algorithms for verification. Predictions were conducted based on the actual data of a boiler unit. The mean absolute percentage error （MAPE） between the predicted and actual values was 0.93%， the root mean square error （RMSE） was 5.959， the mean absolute error （MAE） was 3.564， and the coefficient of determination （R²） was 0.97. The results indicate that the improved model outperforms other machine learning algorithms in predicting NO _x emissions， significantly reducing prediction errors and greatly enhancing the accuracy and practicality of the model.
- machine learning,
- XGBoost,
- particle swarm optimization,
- NO _x,
- emissions prediction

FullText(HTML)

References(20)

References

[1]	YOU C Y. Research on monitoring and optimization strategies for carbon dioxide emissions from coal-fired power plants［J］. Resources Economization& Environmental Protection，2024（8）：17-21. 尤晨昱. 燃煤电厂二氧化碳排放监测与减排优化策略研究［J］. 资源节约与环保，2024（8）：17-21.
[2]	QIU Z L，ZHANG X H，ZHAO D，et al. Analysis of nitrogen oxide generation mechanism and emission control technology in thermal power plants［J］. Journal of Shenyang Institute of Engineering（Natural Science），2023，19（1）：23-29. 邱正霖，张小辉，赵頔，等. 火电厂氮氧化物生成机理及排放控制技术分析［J］. 沈阳工程学院学报（自然科学版），2023，19（1）：23-29.
[3]	ZHUO J K，JIAO W H，SONG S P，et al. A review on nitrogen oxides prediction model in combustion optimization of boilers［J］. Journal of Combustion Science and Technology，2016，22（6）：531-540. 卓建坤，焦伟红，宋少鹏，等. 锅炉燃烧优化中 NOₓ预测模型研究进展［J］. 燃烧科学与技术，2016，22（6）：531-540.
[4]	LI Z B，CHEN S Q. Influencing factors of low-temperature SCR denitrification efficiency for coal-fired boiler flue gas［J］. Chemistry& Bioengineering，2024，41（9）：57-59. 李子彬，陈绍清. 燃煤锅炉烟气低温 SCR 脱硝效率影响因素研究［J］. 化学与生物工程，2024，41（9）：57-59.
[5]	WANG L，HOU Y T，XUE J Z，et al. Principal component analysis-based NOₓ formation prediction for SCR flue gas denitrification system［J］. Thermal Power Generation，2015，44（8）：84-89. 王林，侯玉婷，薛建中，等. 基于主元分析的 SCR 烟气脱硝系统入口 NOₓ生成量预测模型［J］. 热力发电，2015，44（8）：84-89.
[6]	ZHAO Q. Simulation optimization of SCR flue gas denitrification system and optimal control of ammonia injection amount［D］. Chongqing：Chongqing University，2012. 赵乾. SCR 烟气脱硝系统模拟优化及喷氨量最优控制［D］. 重庆：重庆大学，2012.
[7]	LI Y，ZHUO J K，WU Y F，et al. Interpretable prediction model for NOₓ mass concentration at SCR reactor inlet in coal-fired power plants under flexible operating conditions［J］. Thermal Power Generation，2024，53（7）：119-128. 李影，卓建坤，吴逸凡，等. 可解释的变负荷下燃煤机组 SCR 反应器入口 NOₓ质量浓度预测模型［J］. 热力发电，2024，53（7）：119-128.
[8]	LIU Y，YU J，JIN X Z. NOₓ mass concentration prediction based on feature optimization and improved LSTM network［J］. Thermal Power Generation，2021，50（7）：162-169. 刘岳，于静，金秀章. 基于特征优化和改进长短期记忆神经网络的NOₓ质量浓度预测［J］. 热力发电，2021，50（7）：162-169.
[9]	LU Y，LIU J Z，YANG T T，et al. NOₓ emission characteristic modeling based on feature extraction using PLS and LS-SVM［J］. Chinese Journal of Scientific Instrument，2013，34（11）：2418-2424. 吕游，刘吉臻，杨婷婷，等. 基于 PLS 特征提取和 LS-SVM 结合的 NOₓ排放特性建模［J］. 仪器仪表学报，2013，34（11）：2418-2424.
[10]	YANG G T，WANG Y N，LI X L，et al. Dynamic prediction of boiler NO _x emission based on mutual information variable selection and LSTM［J］. Journal of North China Electric Power University（Natural Science Edition），2020，47（3）：66-74. 杨国田，王英男，李新利，等. 基于互信息变量选择与 LSTM 的电站锅炉 NO _x 排放动态预测［J］. 华北电力大学学报（自然科学版），2020，47（3）：66-74.
[11]	LAN M W，LI Y，ZHAO G Q，et al. Study on boiler combustion modeling based on MAPSO optimizing LSSVM model parameters［J］. Journal of Central South University（Science and Technology），2022，53（4）：1506-1515. 蓝茂蔚，李杨，赵国钦，等. 基于 MAPSO 优化 LSSVM 的锅炉燃烧建模研究［J］. 中南大学学报（自然科学版），2022，53（4）：1506-1515.
[12]	MA L Y，CHENG S Z，WANG Y J. Application of sine algorithm optimized regularized ELM in boiler NOₓ emission modeling［J］. Journal of North China Electric Power University（Natural Science Edition），2022，49（3）：112-118. 马良玉，程善珍，王永军. 正弦算法优化正则化 ELM 在 NO _x 排放量建模中的应用［J］. 华北电力大学学报（自然科学版），2022，49（3）：112-118.
[13]	WANG B，ZHAO L，ZHAO C C，et al. Prediction of NO _x concentration at outlet of SCR denitrification system in thermal power plant based on EMD-SVR［J］. Computer Measurement& Control，2020，28（5）：71-75. 王博，赵亮，赵长春，等. 基于 EMD-SVR 的火电厂选择性催化还原脱硝系统出口 NO _x 浓度预测研究［J］. 计算机测量与控制，2020，28（5）：71-75.
[14]	JIN X Z，ZHANG J，CHEN J Z. NO _x emission prediction of thermal power plant based on sparrow search algorithm optimized TPA-LSTM［J］. Control Engineering of China，2024，31（10）：1937-1944. 金秀章，张瑾，陈佳政. 基于麻雀搜索算法优化 TPA-LSTM 的火电厂 NO _x 排放预测［J］. 控制工程，2024，31（10）：1937-1944.
[15]	WANG X，LIU W，WANG Y，et al. A hybrid NO _x emission prediction model based on CEEMDAN and AM-LSTM［J］. Fuel，2022，310：121895.
[16]	ZHOU L，ZHU X J. Univariate short-term electrical load based on MA-CNN-LSTM-Self attention［J］. Science Technology and Engineering，2024，24（22）：9408-9416. 周磊，竺筱晶. 基于 MA-CNN-LSTM 和自注意力机制的单变量短期电力负荷预测［J］. 科学技术与工程，2024，24（22）：9408-9416.
[17]	SUN H B，YANG J G，JIN H W，et al. NO _x prediction model for coal-fired boiler based on bayesian optimization and random forest regression［J］. Journal of Chinese Society of Power Engineering，2023，43（7）：910-916. 孙胡彬，杨建国，金宏伟，等. 基于贝叶斯优化— 随机森林回归的燃煤锅炉 NO _x 预测模型［J］. 动力工程学报，2023，43（7）：910-916.
[18]	WANG Y D，LI J H，YUE X，et al. Prediction of main steam pressure of power plant boiler based on CEEMDAN combined with WTD-XGBoost-LSTM［J］. Chinese Journal of Electron Devices，2024，47（3）：780-787. 王宇冬，李家翰，岳显，等. 基于 CEEMDAN 混合 WTD-XGBoost-LSTM 的电厂锅炉主蒸汽压力预测［J］. 电子器件，2024，47（3）：780-787.
[19]	WEN X Q，LI K C，WANG J G，et al. NO _x emission predicting for coal-fired boilers based on ensemble learning methods and optimized base learners［J］. Energy，2023，264：126171.
[20]	DUAN H，MENG X，TANG J，et al. NO _x emissions prediction for MSWI process based on dynamic modular neural network［J］. Expert Systems with Applications，2024，238：122015.