Study on real-time prediction and warning technology for negative acceleration flight test of civil airplanes

民用客机负加速度试飞实时预测及告警技术研究

¹ School of Automation, Northwestern Polytechnical University, Xi'an 710072, China
² COMAC Flight Test Center, Shanghai 201324, China
³ Shanghai Civil Aircraft Flight Test Engineering Technology Research Center, Shanghai 201324, China

Received: 8 April 2023

Abstract

The flight test of modern civil aircraft verifies the limits of aircraft design performance through a series of rigorous flight tests. Flight test mission is characterized by high risk and complex technology. The negative acceleration flight test is to verify that the aircraft power unit, auxiliary power unit, or any component or system related to it shall not have dangerous faults during negative acceleration. The risk level of negative acceleration flight test is high-risk. This paper presents a real-time prediction and alarm technology for negative acceleration flight test of civil airliners. Firstly, a fusion simulation system for the flight test scene of negative acceleration was developed. The accuracy verification results indicate that the system can meet the requirements of engineering applications. Secondly, the main factors that affect the negative acceleration flight test are given through theoretical analysis, which provides a guidance for simulation. Finally, the negative acceleration prediction model based on compensation factor is established by using BP neural network algorithm and XGBoost algorithm. And the real-time prediction and alarm program of negative acceleration flight test is developed, which is used in negative acceleration flight test of a certain civil aircraft. The prediction results indicate that the accuracy of real-time prediction and alarm program can meet the requirements of flight test monitoring.

摘要

现代民用飞机通过一系列严苛的飞行试验来验证飞机设计性能的极限, 因此飞行试验任务具有高风险、技术复杂的特点。其中负加速度试飞旨在验证飞机动力装置、辅助动力装置以及与之有关的任何部件或系统在负加速度条件下不会发生危险故障。负加速度试飞的风险等级是高风险。针对民用客机负加速度试飞提出一种实时预测及告警技术: 开发针对负加速度试飞场景的融合仿真系统, 精度验证结果表明系统可以满足工程应用的需要; 通过理论分析给出影响负加速度试飞的主要因素, 为仿真计算提供方向; 利用BP神经网络算法和极限梯度提升(XGBoost)算法, 建立基于补偿因子的负加速度试飞预测模型, 并开发负加速度试飞实时预测及告警程序。所提技术应用于某型民机的负加速度试飞中, 预测结果表明负加速度试飞实时预测及告警程序的精度可以满足试飞监控的要求。