基于遥测参数建模的无人飞行器发动机故障诊断

Fault Diagnosis Method of UAV Engine based-on Telemetry Data

摘要:

无人机系统工作处于外回路,从故障发生到判定需要一定的时间做出反馈与控制,若未及时处理,将影响无人机系统运行的稳定性。无人机作为一个大的迟滞延迟复杂系统,只能通过遥测遥控数据掌握飞行器状态。而无人飞行器故障预测与健康管理技术(PHM),是利用先进的传感器的集成,实时下传无人机遥测数据,并借助各种算法和智能模型来预测、监控和管理无人机的状态。本文以遥测数据作为基础,结合无人机的实际工程应用需求,分析无人机发动机典型故障模式,建立无人机发动机典型故障的粒子滤波、K-Means聚类、多层感知器等三种诊断模型。并在最后利用试验数据对诊断结果进行了比较和分析,对三种方法的适用性展开了阐述和说明。实验结果表明,提出的诊断方法能够有效地用于无人机发动机故障诊断中,在工程应用方面具有较高的实用价值。

Abstract:

It takes time to do feedback and control for the outer loop UAV system from failure to determine in the outer loop which will affect the stability of the system if not handled in time. As a large hysteresis delay complex systems it is the only way for UAV to master craft condition telemetry and remote data. UAV’s prognostics and health management (PHM) technology, is to predict, monitor and manage the condition of UAV by using advanced sensor integration, real-time telemetry downlink data, and various algorithms and intelligent models. Based on the telemetry data and combined with actual engineering application demands UAVs, this paper analyzed the typical fault modes, established three UAV typical fault detection models, such as particle filter, K-Means cluster, Multi Layered Perceptron neural network. The experimental results prove the fault diagnosis method can be effectively used in the maintiance of UAV, and has high practical value in engineering application..