| At present,China's underground cable tunnel is still in the stage of large-scale construction.Cable lines can effectively reduce the surface area and alleviate the contradiction between power transmission,urban traffic and residents' life while satisfying the need of large-scale transmission of electricity.However,due to the complexity of underground environment and the difficulty of manual inspection,there have been some cable tunnel safety accidents in China.Therefore,it is of great importance to obtain timely and accurate information of cable tunnels.Real-time image detection algorithm and equipment for cable tunnels based on patrol video have been emerged as the times goes by.In this paper,Convolutional Neural Networks(CNN)is used as the automatic inspection data analysis model of robots.From data storage,data analysis to training convolutional neural networks,the problem of cable tunnel abnormal state identification is systematically studied.This paper uses video enrichment technology by extracting key frames,and condenses a large number of tunnel inspection data without changing the correlation between data.Most of the processed data is used as the training sample of the training convolutional neural network(a small part is used for verification).Cable tunnel equipment identification model,tunnel equipment corrosion model and tunnel water accumulation identification mode are established by means of cable equipment state recognition and other technologies.The artificial experience is expressed as a mathematical model,and the data collected by the manual inspection and the tunnel inspection robot are processed in depth to improve the analysis value and utilization efficiency.Finally,the robot patrols apply the data to the model proposed in the previous paragraph.The experimental results show that the convolutional neural network has good image analysis and search ability.It is an intelligent algorithm worthy of promotion in the power field. |
PDF Full Text Request