Optimal Design Of Large Mode Area Multicore OAM Optical Fiber Based On Neural Network And Particle Swarm Optimization Algorithm

With the rapid development of Internet technology,the demand for capacity of fiber optic communication systems is growing exponentially.However,there is a limit to the information transmission capacity of traditional single-mode fiber optic communication systems.To address this challenge,researchers have explored mode-division multiplexing technology,which improves existing single-mode fiber optic communications to multimode communications and can greatly increase the transmission capacity of fiber optic transmission systems.Orbital Angular Momentum(OAM)optical communication technology based on mode division multiplexing is rapidly becoming a research hotspot in the field of fiber optic communication because of its theoretical ability to greatly broaden the communication capacity.However,like ordinary single-mode fibers,OAM fibers also face nonlinear effects such as self-phase modulation,and increasing the mode field area of the fiber can reduce the optical power density in the fiber,thus effectively reducing the adverse effects caused by nonlinearity.Therefore,it is of great importance to study OAM fibers with large mode field areas.Unlike ordinary optical fibers,OAM beams are susceptible to interference during transmission;therefore,special fiber structures need to be designed to ensure stable transmission of OAM modes over long distances.Traditional optical fiber design methods are mainly based on numerical simulation methods and manual trial-and-error framework,and the computational analysis and optimization process of optical fiber structure and properties are filled with a lot of redundant work,which makes the design process time-consuming and inefficient,and the design results are heavily dependent on the researchers’intuition and experience.To address the current problems in the design of optical fiber structures,this paper proposes an optical fiber design method based on artificial neural network and particle swarm optimization algorithm,which provides a new idea for the optimal design of optical structures with extreme characteristics.The main research work and the innovative results achieved in this paper are as follows.1.An optical property calculation method based on artificial neural network is proposed.First,a dataset between fiber structure parameters and corresponding optical properties was established,and an artificial neural network was built to learn the potential relationship between structure parameters and corresponding optical properties.By tuning the network structure as well as hyperparameters,the neural network can quickly and accurately predict the optical properties(including OAM mode purity,mode field area,dispersion,and nonlinear coefficients)of OAM modes in fiber structures with a prediction error of less than 5%.2.Combining artificial neural network models with multi-objective particle swarm algorithms to achieve automatic iterative optimization of optical structures.By setting different fitness functions,the particle swarm algorithm can achieve structural optimization within the range of customized fiber structure parameters,and the algorithm can find the optical structure parameters that best meet the requirements within the set range.Comparative verification with finite element simulation software shows that the error between the optical properties of the algorithm-designed fiber structure and the finite element simulation results is less than 5%,which indicates that the combination of neural network and particle swarm optimization algorithm can achieve fast and accurate automatic iterative structural optimization design work.3.Based on the above method,a large mode field area multi-core OAM transmission fiber is designed.The simulation results show that the fiber can support the transmission of 10 OAM modes at 1550 nm,and the purity of the supported OAM modes exceeds 99%and the mode field area exceeds 3000μm~2.