摘要
烟气含氧量是送引风调节系统中的重要参数,直接反映锅炉燃烧过程的风煤比,对锅炉燃烧优化有着至关重要的意义。针对火电厂烟气含氧量的软测量问题,提出了一种新型集成软测量方法。基于相似准则在线选择当前工况的相似性样本集,分别建立基于概率密度函数准则的在线滚动即时学习软测量模型和神经模糊软测量模型。在此基础上,采用加权集成的方法集成上述单一模型,并通过粒子群优化算法寻找最优权值。最后,将集成软测量模型应用于火电厂烟气含氧量的测量中。仿真结果表明,提出的集成软测量模型具有较高的预测精度。
Flue gas oxygen content is an important parameter in the induced draft fan and forced draft fan system, a direct reflection of the air coal ratio of boiler combustion process, and it has a crucial importance for boiler combustion optimization. Aiming at the soft measurement problem of flue gas oxygen content for thermal power plant, a new type of integrated soft measurement method was proposed. Based on similar criteria to select the similarity sample set for the current operating conditions, the just-in-time learning soft measurement model and neural fuzzy soft measurement model were respectively established based on the rule of probability density function. On this basis, the weighted integrated approach integrated the single models, and through the particle swarm optimization algorithm to find the optimal weight. Finally, the integrated soft measurement model was applied to the soft measurement of the flue gas oxygen content. The simulation results show that the proposed integrated soft measurement model has higher prediction accuracy.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2014年第7期1497-1502,共6页
Journal of System Simulation
基金
国家自然科学基金项目(61004019)
上海市科委国际合作项目(12510709400)
上海市教委创新重点项目(14ZZ088)
2013年度上海市人才发展基金
关键词
烟气含氧量
软测量
神经模糊模型
即时学习
概率密度函数
flue gas oxygen content
soft measurement method
neural fuzzy model
just-in-time learning model
probability density function