基于机器学习的飞机起落架着陆载荷预测模型

随着物联网、大数据技术的深入发展，一型装备交付部队的同时，往往需同步提供数字孪生模型以优化视情维护过程。论文基于某型号飞机试飞数据，提出一种将机器学习技术用于飞机起落架着陆载荷预测模型构建的方法。以某型号飞机飞行参数为输入，以传感器实测的左起落架垂向载荷为输出，经数据清洗和特征降维后，分别建立极端梯度提升（extreme gradient boosting, XGBoost）、随机森林（Random Forest）和多层前馈（back propagation, BP）神经网络模型，并对所建模型进行调优。经对比和评估，XGBoost模型具有最高的预测精度，对起落架载荷绝大多数样本的预测误差均保持在6%以内，同时建模时间少，泛化能力强，为起落架载荷预测最优模型。

Prediction Model of Landing Load of Aircraft Landing Gear Based on Machine Learning

With the in-depth development of the Internet of Things and big data technology, digital twin models are always needed to optimize the situational maintenance process when one equipment is delivered to the army. Based on the test flight data of a certain type of aircraft, a method to construct the landing load prediction model of aircraft landing gear using machine learning technology is proposed in this paper. Taking the flight parameters of a certain type of aircraft as input and the left landing gear vertical load measured by sensors as output, after data cleaning and feature dimension reduction, the extreme gradient boosting (XGBoost), Random Forest and back propagation (BP) neural network models were established, respectively. And the model was optimized. After comparison and evaluation, The XGBoost model has the highest prediction accuracy, the prediction error of most of the landing gear load samples is kept within 6%. At the same time, the modeling time is less and the generalization ability is strong, so it is the best model for landing gear load prediction.