Research On Optical Fiber Multi Parameter Detection And Pulse Evolution Based On Artificial Intelligencer

The propagation and evolution of optical pulses in nonlinear media is always an important research content in the field of optical fiber and laser.After passing through the nonlinear system,ultrashort pulse often evolves into chaotic complex signals,which carry a lot of information of the transmission system.Through the information extraction and data analysis of these chaotic signals,it is not only helpful to the research of nonlinear system,but also of great significance to obtain the internal characteristics of transmission medium and the parameter sensing of medium environment.However,the traditional methods currently used are extremely inefficient in this regard.Especially in the research of the inverse problem of parameter extraction,the traditional methods often encounter great challenges in practical application because of these complex and non-linear problems without clear rules.In this paper,with the help of the powerful data analysis and information extraction ability of machine learning algorithm,the optical fiber nonlinear system is studied.A variety of machine learning algorithms are proposed to analyze the complex nonlinear propagation law in optical fiber,realize the effective extraction of the information of nonlinear system,and put forward a new practical scheme to complete optical fiber multi parameter detection only using incomplete information(power spectrum information).The main work of this paper is as follows:(1)A new machine learning algorithm is proposed to extract the transmission medium information from the chaotic and irregular signal evolved under the combined action of nonlinearity and dispersion after ultrashort pulse is transmitted through optical fiber and other media for multi parameter detection and sensing.K-nearest neighbor algorithm,decision tree algorithm,random forest algorithm,fully connected network,convolutional neural network and other machine learning algorithms are used to detect fiber dispersion and nonlinearity.The results of these algorithms are compared and optimized.(2)Based on the above scheme,a new measurement method is proposed by using the power spectrum information without phase information under multiple different conditions instead of the complete signal information with phase information.The multi parameter detection based on incomplete signal information is realized,which avoids the difficulty of phase measurement and makes the measurement method more practical.(3)For the optical fiber with small nonlinear coefficient,the discrimination of power spectrum becomes worse under the action of near simple dispersion effect,which affects the measurement effect of using the power spectrum scheme without phase information;An improved method is proposed.By artificially introducing high nonlinear optical fiber with known parameters into the measurement system,the power spectrum discrimination is improved,and the measurement range and sensitivity of the original scheme are improved.It can be used for the measurement of Low Nonlinearity and even pure dispersion devices.(4)In order to solve the difficulty that optimizing pulse propagation requires a large number of numerical simulation with high computational requirements,machine learning is used to simulate and predict the complex nonlinear propagation process in optical fiber.The evolution process of pulse in nonlinear optical fiber system and the prediction of output signal are realized.The automatic modeling ability of neural network for optical fiber nonlinear evolution system is verified.