基于主成分分析、聚类和BP神经网络的湍流MILD燃烧初始着火过程的分析  

Ignition Process in a Turbulent MILD Flame Based on Principal Component Analysis,Clustering and Back-Propagation Neural Network

作　　者：谢凡 鲁昊 张翰林 王忠禹 Xie Fan;Lu Hao;Zhang Hanlin;Wang Zhongyu(School of Energy and Power Engineering,Huazhong University of Science and Technology,Wuhan 430074,China)

机构地区：[1]华中科技大学能源与动力工程学院,武汉430074

出　　处：《燃烧科学与技术》2023年第6期685-692,共8页Journal of Combustion Science and Technology

基　　金：国家自然科学基金资助项目(51776082).

摘　　要：在MILD燃烧中,湍流和燃烧的相互作用十分强烈,尤其是在初始着火过程,识别该过程的火焰推进对于理解燃烧形成过程、稳定燃烧条件有很大帮助.采用MILD燃烧HM1工况的大涡模拟数据,探索了一种基于主成分分析(PCA)、聚类和反向传播神经网络(BPNN)的动态燃烧场识别方法,对稳态燃烧场进行PCA和聚类后,基于BPNN对初始着火过程进行动态识别和分析.结果表明,该方法用于着火过程动态燃烧场识别是可行的,与传统方法相比具有客观、高效的特点,是一种实用的工业火焰测量方法.In MILD combustion,the interaction between turbulence and combustion is very strong,especially in the initial ignition process,and identifying the flame advance of this process is very helpful in understanding the combustion formation process and stabilizing the combustion conditions.With the data from the large eddy simula-tion of the HM1 operating condition of MILD combustion,this paper explored a dynamic combustion field identification method based on principal component analysis(PCA),clustering and back propagation neural network(BPNN).After the PCA and clustering of the steady-state combustion field,the initial ignition process is identified and analyzed dynamically based on BPNN.The results show that this method is feasible for dynamic combustion field identification during ignition,and being more objective and efficient than traditional methods,it can be applied to practical industrial flame measurement.

关 键 词：MILD燃烧 初始着火过程 动态燃烧场 聚类 BPNN 

分 类 号：TK11[动力工程及工程热物理—热能工程]