We attempt to detect a precursor of thermoacoustic combustion oscillations in a staged multisector combustor for an aeroengine by the combined methodology of recurrence-based analysis and machine learning. Combustion experiments are conducted at the combustor inlet temperature and pressure of 740 K and 1340 kPa, respectively. The joint dimensional vicinity $J_d$ and the determinism of recurrence plots $D_e$, effectively capture significant changes in dynamical state during a transition from stable combustion to well-developed combustion oscillations. The feature space of the $J_d-D_e$ plane, which is obtained using a support vector machine, is found to be a promising tool for detecting a precursor of combustion oscillations with increasing main fuel flow rate while maintaining the pilot fuel flow rate.

我们尝试通过基于递归的分析和机器学习相结合的方法，在航空发动机的分级多扇区燃烧器中检测热声燃烧振荡的前兆。燃烧实验分别在燃烧器入口温度和压力分别为740 K和1340 kPa的条件下进行。关节尺寸附近${Ĵ}_d$ 和递归图的确定性 $d_{Ë}$在从稳定燃烧到发展良好的燃烧振荡的过渡过程中，可有效捕获动态状态的重大变化。的特征空间${Ĵ}_d-d_{Ë}$ 发现使用支持向量机获得的飞机是一种有前途的工具，用于在增加主燃料流量的同时检测引燃燃料流量的同时检测燃烧振荡的前兆。