Integrating Artificial Neural Network for predicting roll coating dynamics of couple stress viscoplastic fluid with roll slippage

Experts in engineering field broadly recognize the strengths of Artificial intelligence (AI)-driven machine learning and intelligent computing solvers detecting their robustness, resilience, rapid convergence and stability. One of the most rapidly emerging and dynamic domain of AI is Artificial Neural Network (ANN). This work focuses on the Levenberg–Marquardt Backpropagation based Artificial Neural Network (LMB-ANN) approach to model the roll coating phenomena of couple stress Casson fluid with roll slippage. Lubrication approximation theory (LAT) simplifies the governing system after non-dimensionalization. Then, analytical and mumerical solutions attained for flow, temperature, exiting coating thickness and other mechanical quantities. Reference dataset for four cases with fixed rheological parameters is generated and fed to the LMB-ANN algorithm to predict flow and temperature distribution. Three phases (testing, training and validation) of LMB-ANN alongwith correlation confirm the model accuracy and effectiveness when predicting and optimizing complex industrial systems. The convergence illustrations including mean square error (MSE), regression, error histogram and fitness function further verifies the referenced dataset. The best performance is observed with MSE in range of E−07−E−10 and gradient values in range of E−04−E−06 at 187–1000 epochs for all four cases. Overall, a convergent, highly accurate, and robust LMB-ANN technique verified by above findings. Controlling mechanical quantities like separating forces and exiting coating thickness are important in coating industry, helping in overall coating quality, process reliability, and economic efficiency.