EN摘要
燃煤电厂排放的氮氧化物是环境污染的主要来源之一，燃烧优化可以有效降低锅炉NOx排放量，NOx排放量预测模型作为燃烧优化的基础，受到了人们广泛的关注。针对火电厂W火焰锅炉，提出了一种基于稳态特征提取的模型样本集构造方法，在此基础上，提出一种组合加权最小二乘支持向量机(CWLS-SVM)建立NOx排放量预测模型。首先通过机理分析确定模型输入变量，基于滑动窗口对海量历史运行数据进行稳态特征搜索，以组合相似度判断法进一步筛选特征，构造模型样本集；然后，针对实际生产中LS-SVM对异常值和噪声干扰敏感、不同输入变量对结果的差异性影响等问题，采用基于局部异常因子的经验风险项加权和基于最大信息系数的特征变量加权的方法对LS-SVM进行了改进；最后进行了多种仿真对比实验。结果表明，CWLS-SVM相比于LS-SVM与其他神经网络模型，具有更强的鲁棒性和泛化能力，对实现锅炉燃烧优化具有重要意义。
Abstract
NOx emission from coal-fired power plants is one of the main sources of environmental pollution. Combustion optimization is an effective method to reduce NOx emission from boilers. As the basis of combustion optimization
NOx emission prediction model has attracted wide attention. For W flame boilers in thermal power plants
we propose a model sample set construction method based on steady-state feature extraction. Furthermore
we propose a combined weighted least squares support vector machine(CWLS-SVM) to establish a NOx emission prediction model. Firstly
we determine the input variables of the model through mechanism analysis. Based on sliding windows
we conduct the steady-state feature search on the massive historical operation data
with the features further refined using a combined similarity judgment method to construct the model sample set. Then
aiming at the the problems such as the sensitivity of LS-SVM to outlier and noise interference in actual production
as well as the differential impact of different input variables on the results
we enhance the LS-SVM by employing the empirical risk weighted method based on local anomaly factors and the characteristic variable weighted method based on the maximum information coefficient. Finally
we conduct various simulation and comparison experiments. The results show that CWLS-SVM exhibits superior robustness and generalization ability compared to LS-SVM and other neural network models
marking its significance for boiler combustion optimization.
关键词
Keywords
references
王锐.基于改进LOF的高维数据异常检测方法[J].电信工程技术与标准化,2023(03).
张晖,刘成成,于鹏峰,马双忱.超临界W火焰锅炉燃用劣质燃料燃烧优化研究[J].热能动力工程,2022(10).
张强,武明路,韩文学,包伟伟,张楠,李明轩,翟永杰.高噪声环境下电厂设备声音融合特征生成方法研究与实现[J].热力发电,2022(12).
唐振浩,朱得宇,李扬.基于数据驱动的燃煤锅炉NO_x排放浓度动态修正预测模型[J].中国电机工程学报,2022(14).
张东东,艾小川,刘畅.基于相似样本特征提取的装备性能退化研究[J].系统工程与电子技术,2022(07).
李劲松,董泽,马宁.一种滑动窗口检验的稳态检测方法[J].计算机仿真,2021(10).
邹洋,吴和成,姜允志,赵应丁.改进填补法和多权重相似度相结合的推荐算法[J].计算机应用研究,2020(12).
朱钰森,金晓明,张泉灵.基于聚类与加权连接的锅炉NOx排放量多模型建模[J].控制工程,2019(04).
丁知平,刘超,牛培峰.IGSA-LSSVM软测量模型预测燃煤锅炉NO_x排放量[J].计量学报,2018(03).
顾燕萍,赵文杰,吴占松.最小二乘支持向量机鲁棒回归算法研究[J].清华大学学报(自然科学版),2015(04).
Dehghani Mohammad,Trojovský Pavel.Osprey optimization algorithm: A new bio-inspired metaheuristic algorithm for solving engineering optimization problems[J].Frontiers in Mechanical Engineering,2023.
Wu Qiqiang,Zhang Xianmin,Zhao Bo.A novel adaptive kernel-guided multi-condition abnormal data detection method[J].Measurement,2023.
Xi Yan,Kai Wang,Chao Zhao,Kun Lu.Efficiency improvement technology of front and rear wall hedging boilers based on combustion optimization and balancing heating surface wall temperature[J].Energy Reports,2022.
Yu Chuanjin,Li Yongle,Chen Qian,Lai Xiaopan,Zhao Liyang.Matrix-based wavelet transformation embedded in recurrent neural networks for wind speed prediction[J].Applied Energy,2022.
Abdalla Hassan I.,Amer Ali A..On the integration of similarity measures with machine learning models to enhance text classification performance[J].Information Sciences,2022.
Cao Dan,Chen Yuan,Chen Jin,Zhang Hongyan,Yuan Zheming.An improved algorithm for the maximal information coefficient and its application[J].Royal Society Open Science,2021.
Xiong Wei,Mo Yimin,Yan Cong.Online State-of-Health Estimation for Second-Use Lithium-Ion Batteries Based on Weighted Least Squares Support Vector Machine[J].IEEE ACCESS,2021.
Chen Yiying,Li Zhaowen,Zhang Gangqiang.Attribute Reduction in an Incomplete Interval-Valued Decision Information System[J].IEEE ACCESS,2021.
Sid Ahmed Bessedik,Rabah Djekidel,Aissa Ameur.Performance of<?show [AQ ID=Q1]?> different kernel functions for LS-SVM-GWO to estimate flashover voltage of polluted insulators[J].IET Science, Measurement & Technology,2018.
陈俊.基于YOLOv3算法的目标检测研究与实现[D].电子科技大学,2020(07).
0
浏览量
190
下载量
0
CSCD
- Meta-XML下载
- BibTeX下载
- Endnote下载
- RefMan 下载
- NoteExpress下载
- NoteFirst下载
关联资源
相关文章
相关作者
相关机构
AI问答- 技术支持由北京北大方正电子有限公司提供 京ICP备09064830号-19
京公网安备11010802024621 - 本系统建议在Chrome、 IE9+ 以上版本浏览器阅读本站内容，360浏览器请切换至极速模式
- Cookies帮助我们提供服务并提供个性化体验。使用本网站，即表示您同意我们使用Cookies