Multicriteria decision-making framework for robust energy management AI solutions
Bibliographic Data
| ID | 7397341 |
|---|---|
| Authors | Salem Garfan (0000-0003-1310-6931), A H Alamoodi (0000-0003-4393-5570), Suliana Sulaiman (0000-0002-2440-8831), O S Albahri (0000-0002-7844-3990), A S Albahri (0000-0003-3335-457X), Iman Mohamad Sharaf (0000-0002-0878-5696) |
| Year | 2025 |
| Volume | 10 |
| Pages | 100822-100822 |
| Publication date | 2025-06-17 |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Sustainable Futures (JOURNAL) |
| Journal identifiers | ISSN: 2666-1888 |
| Publisher | Elsevier BV (PUBLISHER) |
| DOI | 10.1016/j.sftr.2025.100822 |
| OpenAlex | W4411370993 |
| Language | EN |
| References cited | 26 |
In recent years, global attention has been shifted toward energy issues, prompting significant support from major countries toward nearly zero-energy structures. However, this transition faces challenges, particularly regarding the financial implications of implementation despite diverse methodologies. The emergence of artificial intelligence (AI) has catalyzed advancements in energy conservation and management, leading to the development of numerous smart energy management systems leveraging the internet of things and AI methodologies. Various machine learning (ML) models have been utilized for energy-saving and consumption prediction solutions, posing challenges in selecting the most effective model. Multicriteria decision-making (MCDM) models offer a solution to this challenge and have been applied across domains, including energy management. This study aims to utilize MCDM approaches, specifically the fuzzy-weighted zero-inconsistency (FWZIC) and combinative distance-based assessment (CODAS) methods, to select the best energy management ML model. The study used data for eight ML alternatives based on the assessments by three field experts with respect to five criteria. The results of the criteria evaluation weights indicate that robustness (C1) received the highest criterion weight with a value of 0.298. The results of the alternative evaluation indicated the hybrid artificial neural network (A1) as the best model for performance. Additionally, a comparison analysis was performed between FWZIC and various criteria weighting methods, as well as between CODAS and different alternative ranking methods. This framework enables decisionmakers to consider an AI solution that optimizes for accuracy, costs, and resilience in a move towards zero-energy infrastructure
Management science · Operations research · Computer Science · Energy Efficiency and Management · Energy Load and Power Forecasting · Engineering · Smart Grid Energy Management · Artificial Intelligence
| Highly cited | No |
|---|