基于机器学习LSTM网络的SO_(2)排放浓度预测

Prediction of SO_(2) Emission Concentration Based on Machine Learning LSTM Network

CFB机组炉内炉外联合脱硫过程因其非线性、强耦合、大滞后等特点,SO_(2)排放浓度难于在线实时、准确测量,控制效果并不精准,因此建立有效的SO_(2)排放浓度预测模型很有意义。该文采集并分析研究某电厂350MWCFB锅炉现场运行数据,在脱硫系统大样本、数据分布分散、样本函数模糊的情况下,选取影响SO_(2)排放浓度的主要变量,建立基于机器学习的LSTM网络的SO_(2)排放浓度预测模型。结果显示此模型预测精度较高、误差较小,可为控制系统提供可靠参考,有助于提高脱硫系统控制精度,具有一定的工程实用价值。

Due to the characteristics of non-linearity,strong coupling,large lag,and other characteristics of the combined desulfurization process inside and outside the furnace of the CFB unit,the SO_(2) emission concentration is difficult to measure online,real-time and accurately,and the control effect is not accurate.Therefore it is meaningful to establish an effective SO_(2) emission concentration prediction model.This paper collects and analyzes the on-site operating data of a 350MWCFB boiler in a power plant.In the case of a large sample of desulfurization system,scattered data distribution,and fuzzy sample function,the main variables that affect the SO_(2) emission concentration are selected,and the SO_(2) emission concentration of the LSTM network based on machine learning is established.Forecast model.The results show that this model has high prediction accuracy and small error,which can provide a reliable reference for the control system,help improve the control accuracy of the desulfurization system,and has certain engineering practical value.