Electric power image classification method based on deep learning

Abstract

The invention discloses an electric power image classification method based on deep learning. The electric power image classification method comprises the following steps: realizing gradual emission of input electric power images by adopting a disordered emission mechanism: adopting a first caching region and a second caching region in the disordered emission mechanism, wherein the first caching region is used for storing a plurality of images to be processed; selecting the plurality of images from the first caching region and storing into the second caching region, wherein the images in the second caching region wait for being emitted in a queue; carrying out region extraction on the electric power images; carrying out enhanced processing on the images subjected to the region extraction; carrying out size adjustment on the images subjected to the enhanced processing; formatting the images so as to meet the input of a classifier; inputting the data into the classifier and selecting different training set sizes and training speeds to carry out a plurality of times of experiments, and comparing and analyzing classification accuracy and efficiency; and selecting an optimal training parameter to obtain an optimal training effect. Compared with the prior art, the electric power image classification method has the obvious advantages, and the convergence speed of a training is more rapid and the accuracy is higher.

Description

technical field [0001] The invention relates to a power image classification method based on deep learning. Background technique [0002] As the lifeblood of the national economy and security, the safe operation of the power grid has always been the core and essential requirement of the power grid. In order to ensure the safety of the high-voltage power transmission and transformation grid, a large number of sensors have been deployed in the power transmission and transformation system. However, the function of a sensor with a single function cannot fully meet the needs of power grid security. In recent years, an important auxiliary solution is to use monitoring equipment to collect video images, manually evaluate the status of power transmission and transformation equipment, and take corresponding measures to solve the fault. However, manually classifying huge streaming media files will consume a lot of manpower and time, and the efficiency is also very low, which makes th...

AI Technical Summary

Problems solved by technology

[0005] In some existing deep learning-based processing platforms, such as the invention patent with application number 201310739026.3 and the invention patent with application number 201410138343.4, there are obvious limitations in processing large-scale image data similar to electric equipment: that is, when a When the same type of training data is continuously input for a period of time, the training model will fail

Images