摘要
针对生物质锅炉燃烧过程的动态特性,提出一种改进的长短期记忆-自注意力机制全卷积神经网络(LSTM-SAFCN)模型用于预测NO_(x)排放浓度。首先利用完全自适应噪声集合经验模态分解法(CEEMDAN)对数据进行预处理,消除数据噪声对NO_(x)排放浓度预测的影响;其次融合自注意力机制与长短时记忆-全卷积神经网络(LSTM-FCN)进行特征提取与预测建模,该拓展方法能够同时兼顾时间序列数据的局部细节与长期趋势特征;最后,利用生物质热电联产系统的实际运行数据验证了所提算法的有效性。
In view of the dynamic characteristics of the biomass boiler combustion process,this paper proposes a long short-term memory-self attention fully convolutional network(LSTM-SAFCN)to predict NO_(x) emission.Firstly,a complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)is applied to preprocess the noise existing in input data.Secondly,the long short-term memory fully convolutional network(LSTM-FCN)is combined with self-attention method for feature extraction and prediction modeling,which takes both the local details of series data and long-term prediction tendency into account.Finally,the effectiveness of the proposed algorithm is verified on a biomass cogeneration system.
作者
何德峰
刘明裕
孙芷菲
王秀丽
李廉明
HE Defeng;LIU Mingyu;SUN Zhifei;WANG Xiuli;LI Lianming(College of Information Engineering,Zhejiang University of Technology,Hangzhou 310023;Jiaxing New Jies Heat&Power Co.Ltd.,Jiaxing 314016)
出处
《高技术通讯》
CAS
北大核心
2024年第1期92-100,共9页
Chinese High Technology Letters
基金
浙江省重点研发计划(2021C03164)资助项目。
