A Hybrid Interval Type-2 Fuzzy Inference System-orangutan Optimization Algorithm (IT2FIS-OOA) for Multi-year Forecasting of Hospital Electrical Energy Consumption during Special Holidays
Year: 2026
Type: article
Abstract: In large-scale hospitals, accurate electrical energy forecasting is essential to ensure a reliable power supply, especially during special holidays characterized by severe, non-linear load fluctuations.This study proposes a hybrid Interval Type-2 Fuzzy Inference System integrated with the Orangutan Optimization Algorithm (IT2FIS-OOA) to forecast multi-year hospital electrical energy consumption during special holidays.The method is implemented for the Energy Management System (EMS) of Ulin General Hospital using data from 14 special holidays over a five-year period (2020-2024).The OOA optimizes the Footprint of Uncertainty (FoU) in the IT2FIS to achieve a balanced exploration-exploitation trade-off for determining optimal membership function parameters.The proposed method achieves the highest accuracy, with the lowest pooled Mean Absolute Percentage Error (MAPE) of 6.583%, outperforming standard machine learning regressors and baselines such as Support Vector Regression (7.024%), XGBoost (8.350%), and Historical Mean (7.094%).It also demonstrates superiority over Fuzzy Type-1 (6.666%), standard IT2FIS (6.612%), IT2FIS-Cuckoo Search Algorithm (6.613%), IT2FIS-Firefly Algorithm (6.606%), and IT2FIS-ZOA (6.605%).Additional metrics, including the Coefficient of Determination (R 2 ), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE), validate its predictive precision.Furthermore, a stability data analysis utilizing the Population Stability Index (PSI) yielded a highly stable score of 0.0591, demonstrating the framework's sustained reliability and robustness against significant distributional data shifts.
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7213628129