基于多尺度特征Informer模型的受热面积灰预测研究
Research on Prediction of Heating Surface Ash Accumulation Using Informer Model with Multi-scale Features
王鲁君 1孙永华 2刘洪涛 2于秋红 2郝浚杰2
作者信息
- 1. 国能浙江宁海发电有限公司,浙江 宁波 315612
- 2. 山东鲁软数字科技有限公司,山东 济南 250001
- 折叠
摘要
针对电厂变负荷工况频发,污染因子波动较大的问题,提出一种融合多尺度特征的Informer预测模型,首先通过小波变换对传感器数据进行去噪预处理,然后对污染因子、机组负荷以及其他相关参数进行建模,预测锅炉的积灰状态.利用某电厂2022年1月至9月间的锅炉相关数据对模型进行训练和验证,结果表明融合多尺度特征的Informer模型的预测误差显著降低,验证了模型的有效性.
Abstract
In allusion to the frequent load variations and severe fluctuations in pollution factors in power plants,a predictive model utilizing the Informer model with multi-scale features is proposed.Base on sensor data denoised by wavelet transformation,pollu-tion factors,unit loads,and other relevant parameters are modelled to forecast boiler fouling status.The model is trained and vali-dated using boiler data of a power plant from January to September in 2022.The results indicate the prediction errors of the Inform-er model with multi-scale features reduce significantly,validating its effectiveness.
关键词
污染因子/Informer预测模型/多尺度特征融合Key words
pollution factor/Informer prediction model/fusion of multi-scale features引用本文复制引用
出版年
2024
NETL
NSTL
万方数据