Experimental and Machine Learning Research on a Multi-Functional Trombe Wall System

Abstract

Performance prediction tools can assist architects and engineers in designing and sizing TWs without the extensive effort, time, and costs associated with experimental evaluations. This study aims to develop an artificial neural network (ANN) model for predicting the performance of a multi-functional TW by using 57 experimental datasets and the Levenberg-Marquardt algorithm as the training algorithm. The developed model was found to be capable of TW performance prediction with error rates < 0.23%. The performance parameters for the ANN model, namely the mean squared error (MSE) and the coefficient of determination (R), were calculated to be 0.034 and 0.99917, respectively.