神经网络与模拟退火算法结合的锅炉低NOx燃烧优化

对某600MW燃煤电站锅炉进行了多工况热态NO_x排放特性测量,在利用多层前向神经网络对该锅炉的NO_x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NO_x燃烧的优化,计算得到可获得低NO_x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=50K,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NO_x燃烧控制的在线优化技术打下了基础.

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

With the developing restrict environmental protection demand, more attention was paid on the low NOx combustion optimizing technology for its cheap and easy property. In this work, field experiments on the NOx emissions characteristics of a 600 MW 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 NOx 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 T0 = 50 K, alpha = 0.6 can lead to a better optimizing process. This work can give the foundation of the boiler low NOx combustion on-line control technology.