摘要
电站锅炉高效低NOx燃烧优化技术越来越受到人们的重视,而锅炉燃烧效率和NOx排放模型是高效低NOx燃烧优化的基础。从提高网络的泛化能力着手,对最小资源分配网络算法的隐节点删减策略进行改进,加入惩罚策略和合并策略,并把改进的MRAN算法应用到对电站锅炉NOx排放与效率的实时建模上。仿真结果表明,改进的MRAN算法除了具有一般MRAN算法的优点外,还具有比MRAN网络更加紧凑的结构。提出的网络算法具有多输出结构,可同时预测NOx排放与效率,适于用在电站锅炉的NOx排放与效率的燃烧实时整体优化中。
There have been considerable attractions to high efficiency and low NO, emissions combustion optimizing technology for power station boilers these years. High efficiency and low NOx emissions combustion optimization is based on boilers combustion efficiency and NO, emissions model. From the point of improving the generation capabilities, this paper improves the hidden nodes reducing strategies of MRAN by adding punishing and incorporating strategies, applies the improved algorithm to model NOx emissions and efficiency on line for a power station boiler. Simulation results show the improved model has more compact structure than MRAN except the general excellence. The improved model is multi-output and can predict NOx emissions and efficiency at the same time, and suitable for NOx emissions and efficiency globe optimizing on line for power plant boilers.
出处
《锅炉技术》
北大核心
2009年第3期5-9,41,共6页
Boiler Technology
基金
国家教育部博士点基金(2050286041)
河海大学科技创新基金资助(2006406086)
关键词
最小资源分配网络
氮氧化物
锅炉效率
Minimal Resource Allocating Networks （MRAN）
NOx
boiler efficiency
