Steam consumption prediction in a tire factory using machine learning approaches

In the tire industry, a considerable amount of energy is used in the form of steam. Thus, developing an adequate steam consumption management system, that supports continual improvement, has a significant impact on energy and water efficiency. In energy management systems, energy baselines are used as a reference point for measuring and assessing the energy performance of the organization or a specific energy-using system. In this method, energy performance indicators are normalized against different factors affecting energy consumption but are not directly related to energy performance. ISO 50006:2017 proposes using linear regression for this purpose. However, linear models cannot capture the non-linearity of complex systems. In this regard, the current study presents a steam consumption model for a tire factory based on more sophisticated machine learning (ML) approaches. This model can be applied as a tool in EnMS for assessing energy performance, establishing energy targets, and measuring the improvement achieved due to energy management efforts.