基于卷积神经网络的机场助航灯故障诊断  

Fault Diagnosis of Airfield Lighting Based on Convolutional Neural Network

作　　者：姜巍 JIANG Wei(Civil Aviation Engineering Consulting Company of China,Beijing 100621,China)

机构地区：[1]中国民航工程咨询公司,北京100621

出　　处：《软件导刊》2018年第11期113-115,119,共4页Software Guide

摘　　要：针对人工目视检测机场助航灯故障诊断效率较低且存在主观因素的弊端,提出一种基于卷积神经网络深度学习的助航灯故障自动诊断方法。该方法通过卷积网络模型自动提取图像特征,并对故障进行分类,最终实现助航灯故障自动诊断。用采集到的助航灯等光强图对网络模型进行训练,再用测试集对模型测试,测试集诊断结果准确度高达94.84%。通过理论和实验数据分析说明,训练后的卷积神经网络模型能对助航灯故障进行高效、准确的自动诊断。In the fault diagnosis of airfield lighting,the efficiency of manual visual inspection is low and subjective factors are affected.For this reason,an intelligent diagnosis method of airfield lighting fault based on deep learning of convolutional neural network is proposed.The method automatically extracts image features through a convolutional network model,and classifies the faults,finally diagnoses the faults of the airfield lighting.The network model is trained by using the airfield lighting intensity maps,and the model is tested with the testing set.The accuracy of the diagnosis fault is as high as 94.84%.Theoretical analysis and experimental data show that the trained convolutional neural network model can effectively and accurately diagnose the faults of airfield lighting.

关 键 词：卷积神经网络 等光强图 故障诊断 深度学习 助航灯光 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]