基于发参数据的发动机泵调系统的健康监测

Health Monitoring of Engine Pump Regulating System Based on Engine Data

中文关键词:  飞机发动机  泵调系统  健康监测  马氏距离  K均值聚类

英文关键词:aero-engine  pump regulating system  health monitoring  Mahalanobis distance  K-means clustering

中文摘要:

      为探究飞机发动机泵调系统健康监测自动化的可行性，开展了基于发参数据的飞机发动机泵调系统的健康监测方法的研究。从发参数据中提取出泵调系统在起动阶段的相关特征参数，构成样本数据；采用马氏距离技术筛选出健康样本，建立泵调系统特征参数基线模型；计算实际飞行样本与基线模型之间的马氏距离作为健康指数，确定预警阈值。采用3种K均值聚类算法对实际发参数据进行离群点检测，结果表明根据样本分布范围随机生成质心的K均值聚类算法能够有效地对泵调系统故障进行诊断，误报率为0，为实现发动机泵调系统健康监测自动化提供了有效的技术途径。

英文摘要:

      In order to explore the feasibility of the health monitoring automation of the engine pump regulating system,the health monitoring method of aircraft engine pump regulating system based on the engine data is studied.The relevant characteristic parameters of the pumping system at the starting stage are extracted from the engine data to form sample data.The Mahalanobis distance technology is used to screen out healthy samples and a baseline model of characteristic parameters of the pump regulating system is estabsished.The Mahalanobis distance between the actual flight sample and the baseline model is then calculated as a health index to determine the warning threshold.Three k-means clustering algorithms are used to detect outliers of real engine data.The results show that the K-means clustering algorithm that randomly generates the center of mass according to the sample distribution range can effectively diagnose the fault of the pump regulating system,and the false alarm rate is 0.It provides an effective technical approach to realize the health monitoring automation of the engine pump regulating system.