Research And Application Of Fiber Optic Vibration Signal Recognition Method For Perimeter Security System

The distributed acoustic vibration fiber optic sensing equipment uses vibration fiber as a sensor to detect the surrounding environment of the security area in real time.The application of vibration fiber in the field of perimeter security is an important development in the field of security.In the perimeter security system,the event signal needs to be accurately and efficiently identified,which is the most important task in the perimeter security system.In long-distance monitoring,the surrounding environment is complex and changeable,which not only increases the difficulty of signal identification by the security system,but also increases the difficulty of real-time and efficient signal monitoring.Therefore,it is of great value and significance to study various event signals obtained by vibration fiber induction in a complex environment.In order to solve the problem of perimeter security system identification,the following work was done.First,the process design of the perimeter security system,the data communication protocol technology between the upper computer and the lower computer,the event signal collection and data preprocessing technology of the optical fiber vibration sensing system in different environments are studied;secondly,introducing a One-Dimensional Convolutional Neural Network(1D-CNN),and improved the structure of one-dimensional convolutional neural network,and tested in two different environments,respectively with Support Vector Machine(SVM)and e Xtreme Gradient Boosting(XGBoost)algorithm for comparative analysis.Then,the influence parameters of the recognition algorithm such as sample sequence length,overlapping window,noise reduction and multi-node in the recognition algorithm are studied.Through comparative analysis of the experiments in two different test environments,the better influence parameters are found;finally,the perimeter is studied Security system design plan,and completed the application of perimeter security system in engineering practice.After comparing the experiments of 1D-CNN and traditional algorithms such as SVM and XGBoost in two environments,the 1D-CNN algorithm performs better.In the net-connected environment and the buried environment,the recognition accuracy rates are 99.36% and 93.29%,respectively.Time is higher than traditional methods.In order to improve the recognition accuracy of the algorithm,the study of the impact factor shows that in the network environment,it is suitable to use 5s sequence length,4s overlapping window,no noise reduction,and three-node parameters.In the buried environment,it is suitable to use 4s sequence length,4.5s overlapping window,noise reduction,and five-node parameters.Integrating the 1D-CNN algorithm into the system for field application has a good effect,and the recognition accuracy rate is 98.26%.