Akademik Veri Yönetim Sistemi
COMMUNICATIONS IN COMPUTER AND INFORMATION SCIENCE, cilt.2669, sa.2, ss.3-17, 2025 (Scopus)
Energy demand and supply in the industrial sector are subject to dynamic and seasonal variations. Hence, energy consumption forecasting models are crucial in the energy management and control system. This study proposes an artificial intelligence-based predictive energy consumption model for an intelligent small-sized steel factory named DAE-WOO in South Korea. This factory's energy data is retrieving using cloud-based systems. The dataset utilized in this study comprises the enterprise’s historical and projected values for reactive power and power factor, along with carbon dioxide emissions and classifications of load types. In addition, the number of seconds from midnight for each day, the state of the week, and the day of the week variables were added to the dataset using the date and time variables. Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) were used to create the training model. LSTM is capable of learning long-term temporal relationships in sequential data, demonstrating superior performance in modelling complex energy consumption patterns. Mean Absolute Error, Mean Square Error, and Root Mean Square Error metrics have been used to calculate the energy consumption estimation performance of the models. Empirical findings prove that the LSTM model is able to predict energy usage with high accuracy and achieve better performance than other algorithms. The findings highlight LSTM's potential as a robust solution for real-time forecasting in industrial applications where temporal dynamics play a significant role. The LSTM model, with its low error rates, supports energy optimization and smart city planning through efficient energy usage.