Machine Learning Techniques For Forecasting Demand Of Energy In Supply Chain: A Case Study In Iran

In today's ever-evolving world, the effective management of energy resources stands as a paramount concern. Iran confronts challenges in harmonizing supply and demand, despite being endowed with nearly ۱۰ percent of the world's total natural resources. This disparity is primarily attributable to its substantial energy consumption. In ۲۰۲۲, due to deficiencies in management, instances arose wherein Iranians experienced interruptions in access to electricity during the summer and gas during the winter, extending through at least part of the day. Beyond its impact on individuals, these circumstances wield significant influence over manufacturers and hospitals, entities that possess a direct and indispensable connection to people's lives. Hence, in view of the gravity of this issue, it would be advantageous to employ intelligent techniques capable of furnishing valuable insights to authorities for addressing similar challenges in the forthcoming years. To realize this objective, the present research will harness two machine learning methods, Autoregressive models and ARIMA Model, and one deep learning method, Long Short-Term Memory, to forecast future energy demand and subsequently assess the efficacy of these methodologies. The results demonstrate superior performance when compared to both machine learning methods.

کلیدواژه ها:

Machine Learning ، Energy Demand ، Demand Forecasting ، Autoregressive models ، Energy Supply chain