Linear regression

Turbulence

Nusselt number

Engineering

Statistics

Materials science

Reynolds number

Air gap (plumbing)

Airflow

Mechanical engineering

Wind speed

Thermodynamics

Physics

Mathematics

Optics

Thermal

Composite material

Meteorology

Approximation error

Mechanics

Mean squared error

Thermal Science and Engineering Progress

To predict the thermal performance of a solar air heater (SAH) with a C-shape finned absorber panel, Machine Learning (ML) models, namely Random Forest (RF), Linear Regression (LR) and K-Nearest Neighbors (KNN) are used. Experiments are carried out with air velocity ranging from 0.57 to 4.4 m/s and by varying the geometric parameters such as relative pitch-to-gap ratios, relative height ratios, and relative perforated ratios to study the effects on maximum heat transfer enhancement. The experiments produced 64 data sets, out of which 54 samples are used for training and 10 samples for testing the ML models. Eight input parameters (mass flow rate, wind speed, ambient air temperature, intake air temperature, mean exit air temperature, average plate temperature, test zone pressure differential, and solar intensity) are used as independent variables in the LR, RF, and KNN models. Three parameters (Nusselt number, friction factor, and thermal efficiency) are used as dependent variables. The RF model performed better than the KNN and LR models because of its lowest error and highest R2 value. The RF model’s MAE, RMSE, and R2 values are 2.202757, 3.382498, and 0.9783, respectively, with the model efficiency of 98 % being the most significant among the other models. This suggests that the RF model is a good fit for predicting the thermal performance of SAHs.

Thermal performance prediction of a solar air heater with a C-shape finned absorber plate using RF, LR and KNN models of Machine learning