华北电力大学学报(自然科学版) Journal of North China Electric Power University(Natural Science Edition) ,河北 保定 071003)
摘要:针对燃煤电厂 SCR 脱硝系统入口 NOX 浓度难以测量的问题,提出了基于改进鲸鱼算法(Improved Whale Optimization Algorithm,IWOA)优化双向长短时记忆神经网络(Bi-directional Long Short-Term Memory Neural Network,Bi-LSTM)的 SCR 入口 NOX 浓度预测模型。利用 LightGBM 进行特征选择,运用最大时间周期的方法计算迟延时间;采用加入 Relu 层的 Bi-LSTM 神 经网络提取时序特征,建立预测模型,并利用 IWOA 确定 Bi-LSTM 的最优超参数,最后与传统 Bi-LSTM、LSTM、LightGBM 预测模型 进行对比验证。仿真结果表明,IWOA-Bi-LSTM 模型的均方根误差、平均绝对百分比误差、平均绝对误差最小,能够实现对 NOX 浓 度的准确预测。 关键词:NOX 建模;鲸鱼优化;特征处理;双向长短时记忆;SG 滤波 中图分类号:TK39 文献标志码:A NOX Modeling of Denitrification System by Bi-LSTM Optimized With Improved Whale Algorithm YAO Ning,JIN Xiuzhang,LI Yangfeng (School of Control and Computer Engineering, North China Electric Power University, Baoding hebei 071003, China) Abstract: Aiming at the problem that it is difficult to measure the NOX c