基于深度学习网络的电气设备图像分类  

Electrical Equipment Image Classification Based on Deep Learning Network

作　　者：王雨滢 赵庆生[1] 梁定康 WANG Yu-ying;ZHAO Qing-sheng;LIANG Ding-kang(Shanxi Key Laboratory of Power System Operation and Control,Taiyuan University of Technology,Taiyuan 030024,China)

机构地区：[1]太原理工大学电力系统运行与控制山西省重点实验室,太原030024

出　　处：《科学技术与工程》2020年第23期9491-9496,共6页Science Technology and Engineering

基　　金：山西省自然科学基金(201801D221362)。

摘　　要：为了对变电站中智能巡检系统采集到的海量图片进行快速分析和识别,提出一种深度学习和支持向量机(support vector machine, SVM)相结合的图像分类模型。首先,运用旋转、翻折等方法对采集到的原始数据进行扩充。然后,合并扩展图像集,并在相同类型的条件下将其随机划分为训练集和测试集。基于实际图像改进卷积神经网络(convolutional neural network, CNN),并提取训练集的图像特征。最后,通过使用训练集图片的深度特征来训练SVM分类器,并且在测试集图片上实现分类测试。利用巡检机器人采集到的8 000张图片对模型精度进行实验验证,结果表明,该模型具有较强的分类性能。In order to analyze and identify massive images collected by the intelligent inspection system in the substation quickly, an image classification model combining deep learning and support vector machine(SVM) was proposed. Firstly, the original collected data was expanded by methods such as rotation and folding. The expanded images were then randomly divided into training sets and test sets according to types, and the parameters in the convolutional neural network(CNN) were improved by actual data, and the training set image features were extracted on the trained CNN models. Finally, the SVM classifier was trained by using the depth features of the training set image, and the test set image was classified and verified. The experiment used 8 000 pictures collected by the inspection robot in a substation to verify the accuracy of the model. The results show that the model has strong classification performance.

关 键 词：电气设备 图像分类 深度学习 卷积神经网络(CNN) 支持向量机(SVM) 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]