摘要
针对焦炉现场废气含氧量不能实时检测和样本数据点过少的问题,该文采用支持向量机组合预测方法建立烟道吸力预测模型。首先,利用启发式搜索算法得到基于RBF核函数和多项式核函数的支持向量机子模型。然后,基于模糊组合原理,将子模型进行加权集成,得到组合预测模型。实例分析表明,烟道吸力模型能有效地预测空气系数,指导烟道吸力的调整,从而使加热燃烧过程优化,最终达到节能降耗的目的。
The fact that flue gas oxygen content can not be detected real-time on-line and less actual sample data points can be achieved in coking plant,in this paper a prediction model is established by the method of combinat-ing support vector machines.Firstly,a heuristic searching algorithm is presented,which is valuable for getting the support vector machines submodels based on RBF and polynomial kernel functions.Then the combination model is integrated by the submodels according to the principle of fuzzy combination.The results of simulation express that the flue suction model can effectively predict the air coefficient and provide the best suction settings,so as to optimize the combustion process for coke oven and ultimately achieve the purpose of energy saving.
出处
《自动化与仪表》
北大核心
2010年第8期5-9,共5页
Automation & Instrumentation
基金
国家863计划重点项目(2008AA042902)
关键词
焦炉
烟道吸力
核函数
支持向量机
组合预测
coke oven
suction in flue
kernel function
support vector machines（SVM）
combination prediction
