Artificial Neural Network Model for Optical Fiber Direction Coupler Design

摘要

A new approach to the design of the optical fiber direction coupler by using neural network is proposed. To train the artificial neural network,the coupling length is defined as the input sample, and the coupling ratio is defined as the output sample. Compared with the numerical value calculation of the theoretical formula, the error of the neural network model output is 1% less.Then, through the model, to design a broadband or a single wavelength optical fiber direction coupler becomes easy. The method is proved to be reliable, accurate and time saving. So it is promising in the field of both investigation and application.