The Construction of an Artificial Intelligence Model for the Intelligent Design of Ring Erbium-Doped Fiber Lasers

Abstract

Erbium-doped fiber laser serves as a crucial cornerstone in the development of optical communications while its performance design primarily relies on inefficient manual adjustments. This paper proposes an AI model for the intelligent design of ring erbium-doped fiber lasers, in which the variables are preset based on the complex nonlinear relationships among 5 key design parameters of the ring erbium-doped fiber laser. The model integrates the theoretical mechanisms of the ring erbium-doped fiber laser with 3-layer fully connected neural networks. By employing a new dual-channel neural network architecture with physical constraints, the model enables the prediction of output characteristics under various parameter combinations and facilitates the inversion of target performance indicators to structural parameters. The application of design parameters resulted in a deviation of only 3.91% in the output power of the erbium-doped fiber laser, while reducing the design process to approximately 2 hours. This method facilitates the intelligent design of erbium-doped fiber lasers and offers a novel approach for the intelligent design of ring fiber lasers.