基于改进HHO的飞行数据异常诊断  

Flight data anomaly diagnosis method based on improved HHO

作　　者：张迪 马文彬 柴源通 曾佩佩 ZHANG Di;MA Wen-bin;CHAI Yuan-tong;ZENG Pei-pei(Engineering Technology Training Center,Civil Aviation University of China,Tianjin 300300,China;College of Electronic Information and Automation,Civil Aviation University of China,Tianjin 300300,China)

机构地区：[1]中国民航大学工程技术训练中心,天津300300 [2]中国民航大学电子信息与自动化学院,天津300300

出　　处：《计算机工程与设计》2025年第3期934-940,F0003,共8页Computer Engineering and Design

基　　金：民航安全能力建设基金项目([2023]50)。

摘　　要：为解决民机飞行数据异常诊断方法准确率低、鲁棒性不足等问题,提出一种改进HHO-MCN-BiLSTM飞行数据异常诊断方法。采用改进的时空网络,通过融合多尺度输入卷积、多尺度残差和双向长短时记忆网络对飞行数据进行特征提取,获取更丰富的特征信息;在网络输出端添加多头注意力机制对特征信息进行加权处理;利用改进的哈里斯鹰优化算法对网络结构以及模型超参数进行寻优。实验结果表明,改进模型的检测精度可达94.96%,性能优于对比算法,可有效改善飞行数据异常诊断的准确率。To solve the problems of low accuracy and insufficient robustness of flight data anomaly diagnosis methods for civil aircraft,an improved HHO-MCN-BiLSTM flight data anomaly diagnosis method was proposed.The improved spatio-temporal networks were utilized to extract features from flight data by multi-scale input convolution,multi-scale heterogeneous residuals,and bi-directional long short-term memory networks,which aimed to capture richer feature information.The multi-headed attention mechanism was added at the output of the network to weight the feature information.The network structure and the model hyper-parameters were optimized using the improved Harris Hawk optimization algorithm.Experimental results show that the detection accuracy of the improved model can reach 94.96%,which is superior to the comparative algorithms,and effectively enhances the accuracy of flight data anomaly diagnosis.

关 键 词：飞行数据 异常诊断 多尺度残差卷积 双向长短时网络 多头注意力机制 改进哈里斯鹰搜索算法 focal loss函数 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]