基于试飞数据的涡扇发动机模型辨识及故障检测研究

Research on Model Identification and Fault Detection of Turbofan Engine Based on Flight Test Data

航空发动机结构复杂,工作环境恶劣,属于故障多发系统。为了保证发动机安全可靠地工作,可通过建立精确的模型,用于发动机状态监控及故障检测。采用K-means聚类算法对海量试飞数据进行处理,剔除工作状态相近、重复的数据点,改善训练样本的容量;然后基于BP神经网络对涡扇发动机气路参数进行辨识,将训练好的模型在未参与训练的数据中推广应用,模型计算结果与试飞数据基本吻合。对故障检测算法进行研究,以模型输出与实测值的残差是否超过设定的阈值作为故障检测的依据,模型在一段故障数据上的应用表明,相较于发动机故障告警系统,该故障检测算法能提前检测出故障。

Aeroengine has complex structure and harsh working conditions,and are fault prone systems.In order to ensure the safe and reliable operation of the engine,an accurate engine model can be established for state monitoring and fault detection.In this paper,the K-means clustering algorithm is used to process the massive test flight data,and the data points with similar and repeated working status are eliminated to improve the capacity of training samples.Then,based on the BP neural network,the air path parameters of the turbofan engine are identified,and the trained model is promoted and applied in the data that has not participated in the training.The calculation results of the model are basically consistent with the flight test data.The fault detection algorithm is studied in the paper.Whether the residual error between the model output and the measured value exceeds the set threshold is used as the basis for fault detection.The application of the model in a section of fault data shows that the fault detection algorithm can detect the fault in advance compared with the engine fault alarm system.