Energy Performance Indicator (EnPI) Prediction of a Compressed Air System in a Can Manufacturing Factory

Greenhouse gas emissions and energy usage are increasing rapidly, posing a significant threat to the environment and the sustainability of the Earth. One of the ways to mitigate this issue is through the implementation of Energy Management Systems (EnMS) in industries, where overall efficiency improvement and energy expenditures cutback are targeted. This work aims to address the energy consumption in a can manufacturing factory. Here, machine learning approach is used to optimize the energy performance of the compressed air system (CAS) which is the major contributor to the factory's energy usage at 38%. In this work, the Energy Performance Indicators (EnPI) of the CAS is predicted using KNN by employing the variables from CAS itself. The collected variables are first analyzed using Dataiku software and correlation matrix and heatmap technique, giving two sets of key features. The predicted EnPI showed high accuracy when compared to the actual values.