A Comparative Evaluation of GRNN and OSELM Models for Dynamic Pressure Drop Prediction in Deep Bed Filtration Systems

Abstract

This study is a detailed examination of the filtration process with a primary focus on the correlation of raw flow rates and measurements of pressure at different locations in a filtration system. The data show a significant change in pressure with increasing flow rates, demonstrating the effect of filtration efficiency on operational measures. The analysis illustrates the complex dynamics involved in ensuring optimal filtration by examining the system's pressure at various points. In addition, a comparative study of the Generalized Regression Neural Network (GRNN) and the Online Sequential Extreme Learning Machine (OSELM) is conducted with regard to their predictive accuracy of given data sets. Coefficient of determination (R2), Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE), were used to compare the models' efficiency. The findings show that GRNN is outperforms OSELM, showing a better capability to fit the training data. The results indicate that GRNN is a very strong and reliable predictive model whereas OSELM requires further optimization to increase its precision.