基于卷积神经网络的无人机结构故障诊断

Structural Fault Diagnosis of UAV Based on Convolutional Neural Network

针对无人机结构故障难以发现的问题,提出了基于振动信号与卷积神经网络的无人机结构故障诊断方法。分别在无人机正常状态下和故障状态下进行振动信号采集,搭建了含有三层卷积层和三层池化层的卷积神经网络,对采集的无人机振动数据信号进行故障分类,利用实验室无人机测试平台进行验证,该方法准确率能够达到97.5%,高于传统的机器学习方法,可以用于无人机结构故障诊断。

Due to the difficulty in detecting structural faults of UAV,a structural fault diagnosis method for UAV is proposed based on vibration signal and deep learning.The vibration signals are collected under the normal and fault conditions of the UAV respectively.A convolutional neural network with three convolution layers and three layers of pooling is proposed.The collected UAV vibration data signals are used for fault classification.The laboratory UAV test platform is used to verify with the accuracy rate can reach 97.5%,which is higher than the traditional machine learning method,and can be used to diagnose the structural fault of the UAV.