【摘要】 不稳定燃烧存在多种模式，不同模式会导致后继燃烧状态的不同演化，对燃烧室燃烧模式的实时预测能够为抑制不稳定燃烧提供依据并提高燃烧室设计效率。首先对模型旋流燃烧室不同工况下不稳定燃烧特性进行实验研究，获得动态压力脉动数据；随后建立不稳定燃烧模式-压力脉动数据库；最后在该数据库的基础上，采用隐马尔可夫模型（HMM）作为学习和预测工具，建立模式与压力脉动信号之间的概率模型并对不稳定燃烧模式进行在线识别与预测。研究结果表明，基于实验数据的HMM方法能够在毫秒级准确预测不稳定燃烧是否发生，并且能够较准确识别不稳定燃烧模式。更多还原

【Abstract】 There are different modes of unstable combustion,which will lead to various evolution of combustion states.The real-time prediction of combustion modes in combustor can provide a basis for suppressing unstable combustion and improve the efficiency of combustor design. First of all,the experiment was conducted to study the unstable combustion characteristics of the model swirl combustor under different working conditions,and the dynamic pressure pulsation data was obtained. Secondly,an unstable combustion model-pressure pulsation database was established. Finally,on the basis of the database,Hidden Markov Model(HMM)was used as a learning and prediction tool to establish the probability model between the model and the pressure pulsation signal,and to identify and predict the unstable combustion mode online. The results of the study show that the HMM method based on experimental data can accurately predict the occurrence of unstable combustion at the millisecond level,and can accurately identify the unstable combustion mode.更多还原