Violation Identification of Substation Staff Electrical Safety Apparatus Use Based on NASNet-Mobile-SVM

JIAN-BAO ZHU, YU-WEI SUN, JING FU, XIN-CHUN YU, LI-XIA SUN, JING-TAO BAI

Abstract


In order to reduce the occurrence of illegal accidents on the site of electric power production, multi-dimensional safety supervision is implemented for electrical safety apparatus. This paper puts forward a comprehensive supervision scheme of three stages before, during and after the use of electrical safety apparatus to realize active early warning so as to avoid personal injury. In this paper, based on the image data of the monitoring system, the transfer learning model and support vector machine classifier are combined to establish the NASNet-Mobile-SVM model, The model realizes the identification of violation of rules in the use of electrical safety apparatus by substation personnel. Data enhancement method was used to expand the initial data. The expanded data was used to fine-tune the NASNet-Mobile network applicable to the Mobile terminal to obtain the feature vectors. Finally, support vector machine (SVM) classifier was used to classify and identify the extracted features. Compared with resnet-18, resnet-50 and MobileNetv2 convolutional network, the proposed method can be easily applied to mobile terminal and maintain better recognition performance

Keywords


Svm, Substation, Transfer learning, Image recognition, Nasnet-Mobile, Safety Apparatus.Text


DOI
10.12783/dteees/peems2019/34029

Full Text:

PDF

Refbacks

  • There are currently no refbacks.