航空科学技术2025，Vol.36Issue(1)：22-32,11.DOI:10.19452/j.issn1007-5453.2025.01.003
基于多尺度时序卷积网络的晴空湍流颠簸预测研究
Research on Clear Air Turbulence Prediction Based on Multi-Scale Temporal Convolutional Network
摘要
Abstract
Atmospheric turbulence can easily induce airplane turbulence,and even leads to serious symptoms and accidents.Accurately predicting atmospheric turbulence and its impact on flight is extremely important for ensuring flight safety.This paper addresses the problems of sparse prediction datasets,low spatiotemporal resolution,and high false alarm rates in current turbulence intensity estimation methods based on flight data.Firstly,a maximum likelihood estimation(EDR)algorithm based on the Kolmogorov model is established,and experiments are designed using the von Kármán turbulence theory model to verify the effectiveness of the EDR algorithm.Secondly,a multi-scale temporal convolutional network(MT-CNN)is constructed to predict the EDR index of flight routes based on flight parameter time series.Experimental analysis shows that the turbulence prediction accuracy based on MT-CNN reaches 92.77%.The method proposed in this article can provide effective prediction of clear sky turbulence intensity in flight routes,helping pilots and controllers make decisions on route turbulence risks and ensuring flight safety.关键词
晴空湍流/EDR指数/von Kármán模型/卷积神经网络/深度学习Key words
clear air turbulence/EDR index/von Kármán model/CNN/deep learning分类
航空航天引用本文复制引用
张其霖,高振兴,齐凯..基于多尺度时序卷积网络的晴空湍流颠簸预测研究[J].航空科学技术,2025,36(1):22-32,11.基金项目
航空科学基金(2022Z066052002) Aeronautical Science Foundation of China(2022Z066052002) （2022Z066052002）
