磨煤机一次风量长短期记忆网络预测模型  

Long and Short Term Memory Network Prediction Model of Coal Mill Primary Air Volume

作　　者：孙传瑜 杨耀权[1] SUN Chuanyu;YANG Yaoquan(Department of Automation,North China Electric Power University,Baoding 071003,China)

机构地区：[1]华北电力大学自动化系,河北保定071003

出　　处：《电力科学与工程》2022年第6期45-53,共9页Electric Power Science and Engineering

基　　金：浙江省电力科学研究院科技项目(1103/9210314014)。

摘　　要：针对燃煤电厂磨煤机一次风量预测精度不高的问题,提出了一种基于麻雀算法和核主成分分析(KPCA)–长短期记忆网络(LSTM)的预测模型。首先,优化数据结构。通过对磨煤机一次风量进行机理分析,预选出与一次风量相关的辅助变量,再利用相似度函数法去除相似度较高的数据样本;利用核主成分分析最大限度地抽取数据之间的特征,降低数据维数。然后,利用LSTM建立一次风量预测模型。为提高预测精度,结合麻雀算法对模型内部存在的多个超参数进行寻优。用某600MW火电机组实际运行数据对算法进行实验检验。实验结果表明,KPCA算法能够有效抽取原数据集的数据特征,提高模型的泛化能力;相似度函数法有效去除了冗余样本,提高了预测精度;与现有神经网络方法相比,所建立的预测模型具有精度更高、波动性更小的优点。Aiming at the problem of low prediction accuracy of primary air volume of coal mill in coal-fired power plants, a prediction model based on sparrow algorithm and KPCA-LSTM was proposed.First, the data structure is optimized.Through the mechanism analysis of primary air volume of coal mill, the auxiliary variables related to primary air volume are pre-selected, and the data samples with high similarity are removed by similarity function method.Kernel principal component analysis is used to extract the features among data to the maximum extent and reduce the data dimension.Then, LSTM is used to establish a prediction model of primary air volume.In order to improve the prediction accuracy,the sparrow algorithm was used to optimize the multiple hyperparameters in the model.The algorithm is tested with the actual operation data of a 600 MW thermal power unit.The experimental results show that KPCA can extract the features of the original data set effectively and improve the generalization ability of the model.The similarity function method effectively removes the redundant samples and improves the prediction accuracy.Compared with the existing neural network methods, the established prediction model has the advantages of higher precision and less fluctuation.

关 键 词：火力发电机组 磨煤机 一次风量 热工自动控制 麻雀算法 长短期记忆网络 核主成分分析 

分 类 号：TK32[动力工程及工程热物理—热能工程] TK224.1