神经网络与模拟退火算法结合的锅炉低NO_x燃烧优化  

The Utility Boiler Low NO_x Combustion Optimization Based on ANN and Simulated Annealing Algorithm

作　　者：周昊[1] 钱欣平[2] 郑立刚[1] 翁安心[1] 岑可法[1] 

机构地区：[1]浙江大学热能工程研究所能源清洁利用和环境工程教育部重点实验室,杭州310027 [2]浙江大学出版社,杭州310027

出　　处：《环境科学》2003年第6期63-67,共5页Environmental Science

基　　金：国家自然科学基金资助项目 (5 0 2 0 6 0 18) ;国家重点基础研究发展规划项目 (G2 0 0 1CB40 96 0 0 ;G19990 2 2 2 0 4)

摘　　要：对某 6 0 0MW燃煤电站锅炉进行了多工况热态NOx 排放特性测量 ,在利用多层前向神经网络对该锅炉的NOx 排放特性进行建模的基础上 ,将神经网络模型与模拟退火全局优化算法相结合 ,实现了锅炉的低NOx 燃烧的优化 ,计算得到可获得低NOx 排放浓度的具体燃烧配风方案 .文中对 2种不同退火参数的模拟退火算法进行了比较 ,结果说明采用T0 =5 0K ,α=0 .6的参数可以获得较好的寻优效果 .本文研究结果为实现大型电站锅炉低NOx 燃烧控制的在线优化技术打下了基础 .With the developing restrict environmental protection demand, more attention was paid on the low NO x combustion optimizing technology for its cheap and easy property. In this work, field experiments on the NO x emissions characteristics of a 600MW coal-fired boiler were carried out, on the base of the artificial neural network (ANN) modeling, the simulated annealing (SA) algorithm was employed to optimize the boiler combustion to achieve a low NO x emissions concentration, and the combustion scheme was obtained. Two sets of SA parameters were adopted to find a better SA scheme, the result show that the parameters of T 0=50K,α=0.6 can lead to a better optimizing process. This work can give the foundation of the boiler low NO x combustion on-line control technology.

关 键 词：锅炉 氮氧化物 模拟退火算法 燃煤电站 排放特性 多层前向神经网络 

分 类 号：X701[环境科学与工程—环境工程]