Software Realization Of Image Recognition System For Transmission Equipment Based On Convolution Neural Network

In recent years,the application of neural network technology in deep learning has achieved good results in the field of image recognition.The monitoring of the operation status of power equipment is an important part of the operation and maintenance of transmission lines.Power patrol personnel usually need to manually observe a large number of images to identify the status of power equipment.Power patrol personnel can use the deep learning technology to intelligently analyze the inspection image to improve work efficiency.In this paper,based on the structure and texture characteristics of important power transmission equipment such as insulators,spacers and shock-proof hammers,an image recognition algorithm for power transmission equipment based on convolutional neural network Faster R-CNN is proposed.The algorithm implements an algorithm for the identification of multiple types of power equipment,and achieves a high recognition accuracy under a variety of complex backgrounds.Compared with many traditional image recognition methods,it has obvious advancement.The discussion in this paper is divided into four steps.In the first step,the data collection,image screening,image size adjustment,image annotation,image classification and data set matching are completed in order,and the principles of image screening and labeling are also formulated.In the second step,three kinds of convolutional neural networks,ResNet101+Faster R-CNN,VGG16+Faster R-CNN and YOLOv2,were used to train the model and train the model.Comparing the results of the measurement and identification,it is concluded that the Faster R-CNN convolutional neural network based on ResNet101 has a high recognition rate for all kinds of transmission equipment.In the third step,for the problem that the recognition rate of some types of power equipment is low in the previous test results,the data set is expanded by adopting image cropping,rotation,noise addition,etc.,further improving the recognition accuracy and model of the model.In the generalization ability.In the fourth step,combined with the practical application scenarios of power operation and maintenance work,an image recognition software system for power transmission equipment suitable for operation and maintenance personnel is designed.In particular,the scientific management of images through hierarchical design not only facilitates the management of operation and maintenance personnel but also improves the image storage efficiency.Finally,the paper also discusses the shortcomings of this identification system and the direction that can be further studied in the next step.In this paper,a large number of pictures of the proposed image recognition method based on convolutional neural network are tested experimentally,which proves that it has a good recognition effect.The transmission equipment identification software system designed in this paper can realize the practical application of the algorithm described in the paper and provide convenience for the use of power patrol personnel,which has certain engineering application value.