Prediction of heat transfer value in the automotive industry with an approach based on internet of things and machine learning

Abstract

Purpose: The aim of this study is to predict the heat that will be sent to the buildings from the heating center in enterprises or facilities with many buildings and a single heating center. Theory and Methods: Machine learning-based regression models were developed with an expanded data set by generating new variables from existing temperature data to predict the heat required for the selected pilot plant of a factory in the automotive industry. Results: Among the nine different machine learning algorithms evaluated, the Linear Regression algorithm with the highest prediction accuracy was selected. Conclusion: Temperature regulation was made with the developed model. Costs have been reduced thanks to the effects of many negative factors such as heat losses in the heating process of the facility, changes in outdoor conditions, overheating or cooling of the environment, loss of effect of the sent heat after a while, and the prevention of heat losses.