Novel hybrid algorithm approach for modeling pressure drops in fluidized beds using dimensionless parameters

This study proposes an innovative approach for modeling the pressure drop in the fluidized layers of cylindrical and conical beds by employing four different algorithms, Support Vector Regression (SVR) with a default optimization algorithm: SVR coupled with the Dragonfly Algorithm (DA), Gray Wolf Optimizer (GWO), and Particle Swarm Optimization (PSO). In contrast to known methods, our approach utilizes dimensionless parameters such as the Reynolds number in the modeling process. After conducting a comprehensive evaluation, our dimensionless model using PSO-SVR achieved the best results with an average absolute relative deviation (AARD) of 0.5235%, a Root Mean Square Error (RMSE) of 0.047, and a high coefficient of determination (R²) of 0.9972. When using dimensionless data, the model showed a superior performance. The use of dimensionless numbers facilitates comparability across different experimental settings and conditions. Additionally, it allows robust and accurate models to be trained by eliminating the scale differences in raw data. The results of this experiment show that certain dimensionless numbers can effectively predict the pressure drop in fluidized beds and, therefore, can provide a new basis for future work.