Prediction of Complex Permittivity of Polar Liquids Using Regression Based Artificial Neural Network

Abstract

Artificial Neural Network (ANN) based models are very powerful tools to learn the complex relationships in data. This paper proposes a regression-based ANN model to predict the complex permittivity of polar liquids for a range of temperature and frequency using the Debye parameters -static permittivity and high frequency permittivity. The model was built based on the dataset obtained from National Physical Laboratory report MAT 23. The proposed model predicts the complex permittivity without the prior knowledge about the best fit Debye equation and related calculations. The model development was initiated with data preprocessing technique followed by parametric tuning and performance evaluation. This model offers a result of low mean square error and it was validated by comparing the actual data with the predicted data.