Novel and robust machine learning approach for estimating the fouling factor in heat exchangers

• DOI: 10.1016/j.egyr.2022.06.123
• Publication Date: 2022-07-07T01:35:48Z
• Authors: Saleh Hosseini; Amith Khandakar; Muhammad E.H. Cho...; Mohamed Arselene ...; Tawsifur Rahman; Moajjem Hossain C...; Behzad Vaferi
• Keywords: Factor (programming language)

The fouling factor (Rf) is an operating index for measuring undesirable effect of solids' deposition on the heat transfer ability exchangers. Accurate prediction helps appropriate scheduling cleaning cycles. Since diverse factors affect this feature, it sometimes hard to estimate accurately using simple empirical or traditional intelligent methods. Therefore, study employs four up-to-date machine-learning algorithms (Gaussian Process Regression, Decision Trees, Bagged Support Vector Regression) and a model (Linear as function constructing variables. 5-fold cross-validation 9268 data samples determines structure considered estimators, 2358 external datasets have been utilized models' testing. relevancy analysis confirms that most accurate predictions are achieved when square root (√Rf) simulated. Gaussian Regression (GPR) shows highest level agreement with experimental in both construction testing stages. trained GPR scored R2 value 0.98770 0.99857 internal datasets, respectively. predicts overall 11626 (Davoudi Vaferi, 2018) MAPE = 13.89%, MSE 7.02 × 10−4, R2=0.98999. proposed outperforms previously suggested artificial neural network estimating