ANN-Based Identification and Prediction Application in a Heating System

In this study, ANN (Artificial Neural Network) and ARX (Auto Regressive eXogenous) prediction structures were compared, and tested on a heating system. Different ANN structures were also compared within the study, and the network RMSE (Root Mean Square Error) and ME (Mean Error) values were highlighted. The used ANN structures vary in a single hidden layer with node counts of $5,7,9$, and 11, while the input values are examined in two groups as two-step back (second order) or one step back (first order), and the training and test results of these structures are investigated.