Utilizing Artificial Neural Networks to Correlate Energy Consumption and Intensity in Metal Industries for the Midwest States

Energy optimization and analysis are significant in mapping energy consumption and savings for various industries. This study sheds light on utilizing two main energy evaluation parameters: Energy Intensity Utilization (EIU) and Specific energy consumption (SEC). This research paper is founded on an extensive 10-year study, from 2013 to 2023, on more than 196 on-site energy assessments performed on various manufacturing companies in Wisconsin and Northern Illinois. This research is considered a comparative study, which aims to explore the correlation between those two parameters using linear regression analysis using SPSS (Statistical Package for Social Sciences) and validate the outcomes using Artificial Neural Networks (ANN). Besides, it is also employed to assess the most updated methods, procedures, and technological applications that are meant to increase energy efficiency. This study demonstrates the relationship between energy consumption, energy savings, and the effects of energy assessment across several industrial sectors. The results have shown a linear correlation between the EIU and SEC as two of the most significant parameters in measuring energy footprint in the energy-auditing sector. Also, the results have shown that the Annual Energy Savings Percentage (AES%) could have interdependence with both the SEC and EIU energy evaluation parameters, which could provide an energy footprint baseline for any stakeholder interested in penetrating the fabricated metal products industry. The ANN results have shown that underfitting occurs when the data model cannot accurately capture the relationship between the input and output variables. The reasons for the underfitting are that the data points are small (95 data points) for a neural network, and the complexity of the model is simplistic for a machine learning application.