光纤方向耦合器的神经网络仿真与设计

为了能简单、高效地设计光纤方向耦合器,将人工神经网络方法引入方向耦合器设计中.以两光纤耦合长度作为输入,耦合比作为输出,建立光纤方向耦合器的径向基函数神经网络模型.用此模型进行了光纤方向耦合器仿真和设计,结果证明,该方法具有速度快、精确高、可靠性好、节约器件的设计成本和缩短设计周期等优点,是光纤方向耦合器的一种全新设计方法.该方法可推广到其他光无源器件的设计.

Simulation and Design of Optical Fiber Direction Coupler Using Neural Network

Artificial neural network has been introduced to design optical fiber direction couplers.The coupling length and coupling ratio are defined as input and output respectively,which are used to train the neural network model.The radial basis function neural network model is built for optical fiber direction couplers,which are in turn simulated and designed through the model.The results show that this method has the advantages of speediness,accuracy and reliability,which are identified by the design example.In addition,this way, as a novel design approach, can save the cost of device design and decrease period of design.It can be used to design other kind of optical passive device.