Inductance Modelling of Planar Meander Structure Using RBM and kNN

Meander Line is a type of inductor that consists of a compact structure while maintaining a certain length of the conducting lines in the combination of vertical and horizontal lines because of this nature it reduces the size of the device. In this paper, the prediction of the meander structured line inductance using the Restricted Boltzmann Machine (RBM) and k-Nearest Neighbor (kNN) has been demonstrated. The modelling for inductance has been done with physical parameters of the meander structure as input parameters with 10, 000 and 1, 20, 120 data points created using analytical formulation and simulation. Analytical data is created using the analytical formula available for the meander structure. The simulation data is created by performing the simulations in the 3D software Computer Simulation Tool (CST) Microwave Studio. The Mean Squared Error (MSE) in the RBM and kNN model is 0.869, 4.34, respectively on analytical data and 8.869 and 1.129, respectively on simulation data. Further, the desired generalizability of the developed model has been validated by creating the new dataset in-between the range selected and also out of the range selected apart from the train-test and the performance of the model showing good agreement.