一种基于PCA-BP的直升机起落架着舰载荷评估方法  

A method of helicopter landing gear landing load evaluation based on PCA-BP

作　　者：郑甲宏[1] 赵敬超[1] ZHENG Jiahong;ZHAO Jingchao(Chinese Flight Test Establishment,Xi’an 710089,China)

机构地区：[1]中国飞行试验研究院,陕西西安710089

出　　处：《中国测试》2021年第5期156-161,共6页China Measurement & Test

摘　　要：为了提高直升机起落架着舰载荷的评估精度,提出采用主成分分析(PCA)与后退式神经网络(BP)相结合的方法建立评估模型。首先通过对影响起落架着舰载荷的飞行状态参数进行主成分分析,降低参数的维度并使参数正交化,然后将获得的主成分作为BP神经网络模型的输入变量获得PCA-BP评估模型,通过起落架着舰试飞数据进行训练、测试和验证,并进一步与全要素的BP神经网络模型进行均方根误差和拟合优度的对比分析。结果表明:PCA-BP评估模型可减少输入变量的个数,消除参数之间的相关性,提高着舰载荷的评估精度。该方法可为直升机起落架在飞行包线边界及包线扩展状态下的着舰载荷评估提供技术支持。In order to improve the landing load evaluation accuracy of helicopter landing gear,the evaluation model was established by combining principal component analysis(PCA)with backward neural network(BP).First,the flight state parameters that affect loading load were analyzed by principal component analysis to reduce the dimension of the parameters and orthogonalize the parameters.Then,the obtained principal components were used as input variables of the BP neural network model to obtain the PCA-BP evaluation model.The PCA-BP evaluation model were trained,tested and validated by flight test data and were further compared with the whole element BP neural network model by the root mean square error and fitting coefficient.It reveals that the PCA-BP evaluation model can reduce the number of input variables,eliminate the correlation between parameters and improve the accuracy of load evaluation.This method can provide technical support for the load assessment of helicopter landing gear under the condition of flight envelope boundary and envelope expansion.

关 键 词：主成分分析 神经网络 起落架 着舰载荷 直升机 

分 类 号：V217[航空宇航科学与技术—航空宇航推进理论与工程]