基于TCN-Transformer模型的低温省煤器入口烟气参数预测OA
Prediction of Inlet Flue Gas Parameters of Low Temperature Economizer Based on TCN-Transformer Model
以某火电厂 1 055 MW 机组的低温省煤器为例,提出了一种融合时序卷积网络(TCN)与 Trans-former架构的低温省煤器入口烟气参数预测模型.充分考虑机组运行特性对多变量特征时间序列进行建模,实现了对低温省煤器入口烟气温度和烟气流量的预测.通过与传统的反向传播(BP)神经网络、长短期记忆网络(LSTM)模型和最小二乘支持向量机(LSSVM)模型的对比分析表明:所建TCN-Transformer模型在烟气参数预测方面具有更高的准确性和泛化能力,对低温省煤器入口烟气温度和烟气量的预测相对误差分别为0.22%和 1.67%.
Taking the low-temperature economizer of a 1 055 MW unit in a thermal power plant as an example,a low-temperature economizer inlet flue gas parameter prediction model that integrates the Temporal Convolutional Network(TCN)and the Transformer architec-ture is proposed.The multivariate characteristic time series is modeled by fully considering the operating characteristics of the unit,and the prediction of the flue gas temperature and flow rate at the inlet of the low-temperature economizer is realized.Through comparative analysis with the traditional Back Propagation(BP)neural network,Long Short-Term Mem-ory(LSTM)model and Least Squares Support Vector Machine(LSSVM)model,the results show that the constructed TCN-Transformer model has higher accuracy and generalization ability in flue gas parameter prediction,and the relative errors of the prediction of the flue gas temperature and flow rate at the inlet of the low-temperature economizer are 0.22%and 1.67%,respectively.
姜正雄;韦栋梁;周昊
上海电气电站环保工程有限公司,上海 201612浙江大学 能源高效清洁利用全国重点实验室,浙江 杭州 310027浙江大学 能源高效清洁利用全国重点实验室,浙江 杭州 310027
能源科技
低温省煤器烟气温度TCN-Transformer时序预测
low-temperature economizerflue gas temperatureTCN-Transformertime series prediction
《锅炉技术》 2026 (2)
11-17,7
中央高校基础研究基金(2022ZFJH04)
评论