The neural network model of optical fiber direction coupler

The coupling length and coupling ratio is defined as input and output respectively, which are used to train the neural network model. RBF neural network model is built for optical fiber direction coupler, which is in turn simulated and designed through the model. This method has the advantages of speediness, accuracy and reliability, which are identified by the design example. It can be used to design the other kind of optical passive device. This way is a novel design approach, which can save the cost of device design, decrease period of design and has a good prospect of application.