Reasearch On Optical Fiber Vibration Signal Recognition Algorithm Based On Stochastic Neural Network

Nowadays,pipeline transportation of oil and gas resources has become an important part of resource allocation system in our country.Compared with traditional transportation methods,pipeline transportation has some irreplaceable advantages,such as more economical and convenient.Nevertheless,due to the flammability and volatility of oil and gas resources,with the economy develops,the threats to pipelines caused by human behavior such as construction are increasing,which makes people suffer huge losses in production and life.Therefore,it is necessary to design a new type of early-warning system with high precision,high sensitivity,high timeliness and high stability.The optical fiber pre-warning system(OFPS)has been designed to solve this problems.This system has the advantages of wide range,high precision,anti-electromagnetic interference,small floor space and low economy,which has received extensive attention.Since the working environment of the optical fiber pre-warning system is relatively complicated,it is very important to improve the anti-interference ability of the system.The main contents of this article are as follows:This paper first introduces the research background,then summarizes the application situation and research status of the optical fiber pre-warning system at home and abroad.After that,the problem of low input signal-to-noise ratio in the system is expounded,and the significance of improving the robustness of the optical fiber data identification method is highlighted.Feature extraction of the optical fiber intrusion signal is one of the most basic problems in the OFPS.This paper proposes a spectrum analysis method in the optical fiber pre-warning system.This method solves the limitation of Fourier transform for fiber optic signals and improve the performance of the recognition algorithm.Finally,this paper introduces the development history of stochastic neural networks and the superiority of stochastic neural networks compared with traditional identification methods,and analyzes the limitations of existing stochastic neural networks in terms of robustness.According to the limitations of the generalization ability of random neural networks,an improved model based on truncated singular value decomposition is proposed to improve the robustness of the model.This method eliminates redundant parts of the data and selects the hidden layer nodes dynamically.With this method the generalization ability of the network has improved,so that it can meet the requirements of the optical fiber early warning system.The experimental results show that the proposed algorithm has higher recognition accuracy and robustness to fiber-optic signals,so the algorithm has higher practical value.