基于神经网络的飞行课目载荷预测研究

Flight Mission Load Prediction Study Based on Artificial Neural Network

精确地跟踪、预测飞行载荷对单机寿命监控至关重要,相比传统的飞参解析法,神经网络在预测复杂飞行状态下的飞行载荷具有明显优势,然而现在国内外大部分研究的预测模型主要用于预测单个部件的飞行载荷,缺乏以飞行课目为输入的全局性飞行载荷预测。本文通过飞行课目归并,挖掘各飞行课目典型飞行架次,建立课目与飞行特性矢量间的映射关系,获得输入数据库,以反向传播神经网络方法为基础,结合主成分分析(PCA)法和遗传算法(GA),构建飞行课目载荷预测方法,训练得到各飞行课目下不同预测对象的载荷预测模型,实现全局性预测,形成新型载荷监控模式。根据校验结果可知,基于神经网络的飞行课目载荷预测方法方便快捷、准确高效。

The prediction of high accuracy flight load plays a key role in aircraft life monitoring.Compared with the traditional flying parameter analysis method,neural network has obvious advantages in flight load prediction in complex flight condition.However,most of the prediction models are mainly used to predict a single component flight load.There is a lack of global flight load prediction by flight subject.This paper merges the flight missions,searching for the typical representative measurements of all flight missions to establish vector mapping between flight mission and flight feature vector.In this paper,combining with Principal Component Analysis(PCA)and Genetic Algorithm(GA),the flight load prediction models of different flight missions are trained by the artificial neural network,making global prediction come true to form a new load monitoring model.The checksum result shows that neural network has high prediction accuracy on flight mission load.