基于LSTM算法的火电厂智能辅助脱硝系统开发与工程应用

Development and engineering application of intelligent assisted denitrification system for thermal power plants based on LSTM algorithm

随着新能源电站上网电量的快速增长,常规火电机组承担电网调峰任务逐渐增多,锅炉烟气脱硝系统面临更加频繁多变的运行工况。为提高设备运行的可靠性,降低运维人员的监盘工作量,利用大数据分析与人工智能算法赋能电厂传统脱硝设备,为某350 MW超临界锅炉开发了智能辅助监盘系统。经训练优化,脱硝入口参数预测模型、电加热器性能监测模型、热一次风流量异常监测模型预测准确率均达到实用性要求。系统部署应用后,现场未再发生电加热器与热一次风流量不足故障,调峰工况下未再出现操作员过调与欠调问题,有效提升了SCR系统运行可靠性,减轻了人员工作强度,取得了良好的经济效益和环境效益。

With the rapid growth of grid connected electricity in new energy power plants,conventional thermal power units are gradually undertaking more and more grid peak shaving tasks,and boiler flue gas denitrification systems are facing more frequent and variable operating conditions.In order to improve the reliability of equipment operation and reduce the workload of operation and maintenance personnel,big data analysis and artificial intelligence algorithms were used to empower traditional denitrification equipment in power plants,and an intelligent auxiliary monitoring system was developed for a 350 MW supercritical boiler.After training and optimization,the prediction accuracy of the denitrification inlet parameter prediction model,electric heater performance monitoring model,and thermal primary air flow anomaly monitoring model all meet the practical requirements.After the deployment and application of the system,there were no further incidents of insufficient electric heaters and hot air flow on site,and there were no operator over regulation or under regulation issues under peak shaving conditions.This effectively improved the reliability of the SCR system operation,reduced personnel workload,and achieved good economic and environmental benefits.