Prediction of optical specifications through ANN model to design a monochromatic optical filter for all three optical windows

Abstract Plane wave expansion method is extensively implemented in the literature [1–5] to design monochromatic photonic filters based on silicon grating structure like Silicon (Si) and Silicon Monoxide (SiO) where the thickness of these materials is the significant factor for the selection of the particular admissible wavelength of the optical signal. The thicknesses of these materials are selected by speculation manner while designing which leads to time intense affairs as PWE simulation itself takes time to accomplish. To evade this time-consuming process, a machine learning technique using ANN algorithm is proposed in this work to establish a model for the selection of thickness of Si and SiO to realize the filters. The ANN model is developed on Google Colab using TensorFlow framework which is provided by Google generously to implement the machine learning algorithm. Here, three ANN models are proposed for the three optical windows (800–900 nm, 1300–1400 nm, 1500–1600 nm) from where the optical specifications (thickness of materials) can be found out for any of the allowable wavelengths of the optical signal within the range for designing the filter. The ANN model is established by collecting the data arbitrarily by PWE simulation method earlier to train the model and finally the model data are validated again by PWE simulation method to check the accuracy of the model.