Search

Photonic Bandgap crystals are avidly studied because of their application of optical waveguides, optical diodes, defect mode photonic lasers, to name a few. This paper presents a comparative study of 14 bandgap crystals [13 defective and 1 without defect]. The design, simulation, and classification-based analysis of these defective crystals obtained by modifying dielectric arrangement in photonic bandgap lattice (PBG) lattice is carried out. In addition, the study of the dielectric structure of rods in the air is carried out concerning photonic bandgap for a 9 × 10μm wafer size. The discrete pulse components in real data, imaginary data, and intensity data were obtained using Fast Fourier Transforms and were converted to a clean dataset. With context to defective Photonic Crystal analysis and design explored the predictive and generative models for data-driven approaches. Within the predictive modeling framework, Microsoft AzureML is used to give classification performance using five different algorithms. The relative effectiveness of these algorithms has been studied. The created data have been classified using Multiclass Artificial Neural Network, Multiclass Decision Jungle, Multiclass Logistic Regression, Multiclass Random Forest, and 2-Clsaas Support Vector Machine classifiers. The Multiclass Decision Jungle and Multiclass Random Forest exhibit the maximum accuracy of 91.01% and 91.20%, respectively, while the other three algorithms give the classification accuracy of 87.5%. [ABSTRACT FROM AUTHOR]