Study On Control System For Moving Vision Detection Of Breakages And Intrusions Of High Speed Rail Fences

The high-speed rail fence is a safety barrier that isolates the high-speed rail driving area from the outside world,it plays an important role in ensuring the safety of high-speed railway traffic.The tasks of the high-speed rail fence detection system mainly include fence intrusion detection and fence damage detection.The existing fence detection system uses vibration fiber optics,electronic fences,radio detection,intelligent monitoring and other system technologies,However,there are still problems such as singularity of detection method,high false alarm rate,high cost,monitoring blind spots,and the need for inspectors to detect the damage status of the fence in real time around the clock.Therefore,how to solve the problems existing in the fence detection system,reduce the cost,reduce the labor intensity of the staff,and improve the detection accuracy has become a new problem in the security of high-speed railway enclosures.To this end,the subject starts with mobile monitoring,uses excellent computer vision technology,and uses the orbit inspection robot as a carrier to design and develop a high-speed rail fence detection system based on mobile vision,this paper studies the detection algorithm of fence damage and the detection and control system of fence intrusion based on deep learning..The main work and results are as follows:(1)In view of the shortcomings of traditional target detection methods,introduce the relevant knowledge of deep learning.Firstly,analyzing the structure of convolutional neural network,including convolutional layer,pooling layer,fully connected layer,batch normalization layer and activation function.Then,the SVM classifier and Softmax classifier are introduced briefly,and finally analyzing the target detection model based on deep learning.The argument shows that the target detection of Faster RCNN model is the most efficient.(2)This section proposes a deep learning-based target detection algorithm.This algorithm is to improve the existing advanced Faster RCNN model.First,it need to build the fence damage data set,and increase the number of pictures in the data set by means of data enhancement,which effectively avoids the occurrence of overfitting during the training process.Then,aiming at the characteristic that the size of the fence damage is not the same,the target recognition framework based on Faster RCNN is used,and the original VGG-16 network is replaced by the deep residual network RseNet-50 as the feature extraction network of the fence damage model.Improve theratio and make the size of the anchor closer to the actual size of the damaged fence.Finally,it need to train the improved model to realize the classification and positioning of the broken fence part of the fence.The accuracy rate of the trained model on the validation set reached 95.45%,and the recall rate reached 96.61%.Compared with the results of the identification of the damaged fence before and after the improvement,the improved model is better than the original model in the recognition of the broken fence.(3)Aiming at the shortcomings of existing intrusion detection methods for high-speed rail fences,an intrusion detection control system with moving vision is designed.this section design a mobile vision intrusion detection control system,elaborated the hardware structure and software algorithm design scheme.The experimental test platform has tested and analyzed functions such as autonomous inspection of intrusion.it was found that the positioning error of the inspection robot to the intrusion area can be controlled within 0.3m after 200 intrusion tests,which can fully satisfied the requirements of this design.The experimental results show that the fence intrusion detection and control system has the characteristics of high reliability,accurate control and clear video shooting.The work completed above has certain practical significance for the high-speed rail fence intrusion detection.(4)Aiming at the functional requirements of the detection system of high-speed rail fence,a set of man-machine interactive monitoring system is developed,including the design and development of the login interface and operation interface of the monitoring system.Figure [46] table [4] reference [61]...