Akıllı Kampüsler için Yapay Zekâ Tabanlı Enerji Optimizasyonunun Tekno-Ekonomik Doğrulaması: Bir Dijital İkiz Simülasyonu Yaklaşımı

Abstract

Orta ve büyük ölçekli eğitim tesisleri ve benzer altyapıya sahip komplekslerde artan enerji tüketimi, dinamik işletme koşullarına uyum sağlayabilen akıllı yönetim sistemlerini zorunlu kılmaktadır. Bu çalışma, akıllı kampüsler için önerilen melez yapay zekâ tabanlı bir enerji optimizasyon modelinin teknik performansını ve ekonomik fizibilitesini değerlendirmektedir. Yüksek kullanıcı yoğunluğuna ve 100.000 metrekarelik kapalı alana sahip orta ölçekli bir üniversite kampüsünün (idari, laboratuvar, derslik ve kafeterya binaları dâhil) dijital ikizi, farklı yaz ve kış senaryoları altında gerçek zamanlı operasyonları simüle etmek için geliştirilmiştir. Kontrol mimarisi, ısıtma, havalandırma ve iklimlendirme optimizasyonu için Derin Q-Ağları, hareket tabanlı aydınlatma kontrolü için Evrişimsel Sinir Ağları ve fiş yükü yönetimi için Çok Katmanlı Algılayıcıları entegre etmektedir. Teorik modellerin aksine, bu çalışma performansı titiz bir simülasyon yoluyla doğrulamaktadır. Sonuçlar önerilen sistemin geleneksel kural tabanlı temellere kıyasla ortalama %44,5 enerji tasarrufu sağladığını ve %91'lik bir kullanıcı konfor puanını koruduğunu göstermektedir. Çalışma ayrıca, bu simülasyonla doğrulanmış tasarrufları kullanarak ekonomik fizibiliteyi yeniden hesaplamaktadır. Finansal analiz, yüksek başlangıç sermaye harcamalarına rağmen sistemin 6 yıl içinde başabaş noktasına ulaştığını ve %20'lik bir İç Getiri Oranında istikrar kazandığını ortaya koymaktadır. Bu bulgular, otonom otomasyonun yalnızca teknik olarak sağlam olmakla kalmayıp, aynı zamanda sürdürülebilir kampüs altyapısı için finansal olarak da uygulanabilir bir çözüm sunduğunu doğrulamaktadır.

Keywords

• Akıllı kampüs
• Melez yapay zekâ
• Dijital ikiz
• Ekonomik fizibilite
• Enerji optimizasyonu.

Techno-Economic Validation of AI-Based Energy Optimization for Smart Campuses: A Digital Twin Simulation Approach

Abstract

Context—The rising trajectory of energy consumption in mid-to-large scale educational facilities presents a significant challenge for modern sustainability goals. University campuses, characterized by complex infrastructure and fluctuating occupancy patterns, often suffer from inefficiencies inherent in traditional, rule-based Building Management Systems (BMS). These static systems frequently fail to adapt to dynamic operational conditions, leading to excessive energy waste and compromised occupant comfort. Consequently, there is a critical need to transition from reactive management to intelligent, predictive systems capable of handling the stochastic nature of campus energy demands. Addressing this gap requires robust validation of adaptive technologies that can harmonize energy efficiency with operational cost-effectiveness.

Objective—In the framework of the problems associated with static energy management, the main purpose of this study is to evaluate the technical performance and economic feasibility of a proposed hybrid Artificial Intelligence-based energy optimization model designed specifically for smart campuses. This research aims to bridge the gap between theoretical AI models and practical applications by investigating whether an autonomous system can significantly reduce energy consumption without degrading user comfort, while simultaneously proving its financial viability as a long-term investment for institutional stakeholders.

Method—To achieve rigorous validation, a high-fidelity digital twin of a mid-sized university campus was developed. The study employs a Counterfactual Simulation assumption, running parallel execution threads to benchmark the AI model against a Rule- Based Control (RBC) baseline under identical TMY meteorological conditions. This virtual environment models a 100,000-square-meter indoor area comprising four distinct building typologies: administrative offices, laboratories, classroom buildings, and cafeterias. The study simulates real-time operations under distinct summer and winter meteorological scenarios to test system resilience. The proposed control architecture integrates three specific neural network models: Deep Q-Networks (DQN) are utilized to optimize heating, ventilation, and air conditioning (HVAC) systems; Convolutional Neural Networks (CNN) are deployed for precise, motion-based lighting control, and Multi-Layer Perceptrons (MLP) are applied to manage plug load anomalies.

Results—The most significant findings obtained from the simulation demonstrate that the proposed hybrid AI system outperforms traditional rule-based baselines by a substantial margin. The system achieves an average energy saving of 44.5% across the simulated campus, while successfully maintaining a user comfort score of 91% in compliance with ASHRAE 55, indicating that efficiency did not come at the cost of habitability. Furthermore, the financial analysis, recalculated using these simulation-verified savings, reveals compelling economic indicators. Despite the high initial capital expenditures required for implementation, the system achieves a break-even point within just 6 years and stabilizes at an Internal Rate of Return (IRR) of 20%.

Conclusion—These findings confirm that autonomous automation is not only technically robust but also presents a financially viable solution for sustainable campus infrastructure. The study implies that hybrid AI models offer a scalable pathway for educational institutions to reduce their carbon footprint and operational costs. Future studies may focus on integrating renewable energy sources into this control architecture to further enhance grid independence and sustainability.

Keywords

• Smart campus
• Hybrid artificial intelligence
• Digital twin
• Economic feasibility
• Energy optimization.

References

1. [1] A. R. Amaral, E. Rodrigues, A. R. Gaspar, A. Gomes, “A review of empirical data of sustainability initiatives in university campus operations”, Journal of Cleaner Production, 250, (2020), 119558.
2. [2] D. E. Putri, N. Krisnawati, I. Adhicandra, J. W. Sitopu, “Integration of Internet of Things (IoT) and Artificial Intelligence for Campus Education: A Case Study of Energy Management and Security”, Global Education Journal, 3(1), (2025), 635.
3. [3] S. Agostinelli, F. Cumo, G. Guidi, C. Tomazzoli, “Cyber-Physical Systems Improving Building Energy Management: Digital Twin and Artificial Intelligence”, Energies, 14(8), (2021), 2338.
4. [4] S. S. Gulyamov, R. Ruziev, J. Babaev, M. Musaev, S. Mamanazarov, S. Musaev, A. Rodionov, “Using AI to Create Digital Twins of Energy Systems and Conduct Virtual Stability Tests”, 2024 6th International Conference on Control Systems, Mathematical Modeling, Automation and Energy Efficiency (SUMMA), Lipetsk, Russian Federation, 13-15 November 2024, 945-948.
5. [5] A. Pranoto, H. Hermawan, N. Albart, “A Systematic Literature Review: Business Feasibility Analysis Using Net Present Value (NPV) and Internal Rate of Return (IRR) Methods in the Automotive Industry”, Jurnal Indonesia Sosial Sains, 6(1), (2025), 171-184.
6. [6] W. J. Wang, Y. F. Shu, Z. Y. Wang, C. L. Ji, H. Huang, R. D. Ji, M. X. Zhou, Y. D. Wang, J. Ji, “A Multi-load balancing control strategy for a novel low carbon integrated energy system for buildings”, PLOS One, 20(1), (2025), e0329309.
7. [7] H. Farzaneh, L. Malehmirchegini, A. Bejan, T. Afolabi, A. Mulumba, P. P. Daka, “Artificial Intelligence Evolution in Smart Buildings for Energy Efficiency”, Applied Sciences-Basel, 11(2), (2021), 763.
8. [8] E. Giglio, G. Luzzani, V. Terranova, G. Trivigno, A. Niccolai, F. Grimaccia, “An Efficient Artificial Intelligence Energy Management System for Urban Building Integrating Photovoltaic and Storage”, IEEE Access, 11, (2023), 18673-18688.

9. [9] B. Muniandi, P. K. Maurya, C. Bhavani, S. Kulkarni, R. R. Yellu, N. Chauhan, "AI-Driven Energy Management Systems for Smart Buildings", Power System Technology, 48(1), (2024), 322–337.
10. [10] J. Aguilar, A. Garces-Jimenez, M. D. R-Moreno, R. Garcia, “A systematic literature review on the use of artificial intelligence in energy self-management in smart buildings”, Renewable & Sustainable Energy Reviews, 151, (2021), 111530.
11. [11] B. Palley, J. P. Martins, H. Bernardo, R. Rossetti, “Integrating Machine Learning and Digital Twins for Enhanced Smart Building Operation and Energy Management: A Systematic Review”, Urban Science, 9(6), (2025), 202.
12. [12] I. Rojek, D. Mikołajewski, A. Mroziński, M. Macko, T. Bednarek, K. Tyburek, “Internet of Things Applications for Energy Management in Buildings Using Artificial Intelligence—A Case Study”, Energies, 18(7), (2025), 1706.
13. [13] J. Y. Li, M. S. Herdem, J. Nathwani, J. Z. Wen, “Methods and Applications for Artificial Intelligence, Big Data, Internet-of-Things, and Blockchain in Smart Energy Management”, Energy and AI, 11, (2023), 100208.
14. [14] A. T. Al Ghazo, M. Salah, “IoT and AI-Driven Smart Power Management System for Optimizing Resource Usage in Smart Buildings: The Faculty of Engineering at the Hashemite University Case Study”, 2025 IEEE 22nd International Multi-Conference on Systems, Signals & Devices (SSD), Monastir, Tunisia, 17-20 February 2025, 637-645.
15. [15] C. T. Madabathula, K. Agrawal, V. Mehta, S. Kasarabada, S. S. Kommamuri, G. N. Liu, J. Gao, “Smart Green Energy Management for Campus: An Integrated Machine Learning and Reinforcement Learning Model”, Smart Cities, 8(1), (2025), 30.
16. [16] T. V. Điền, T. N. Phúc, “Overview Study of Smart Energy Management Solutions for Universities Aimed at Saving and Reducing Costs”, International Journal of Advances in Engineering and Management (IJAEM), 7(5), (2025), 332–337.
17. [17] S. Pimenow, O. Pimenowa, P. Prus, “Challenges of Artificial Intelligence Development in the Context of Energy Consumption and Impact on Climate Change”, Energies, 17(23), (2024), 5965.
18. [18] A. Silvestri, D. Coraci, S. Brandi, A. Capozzoli, E. Borkowski, J. Köhler, D. Wu, M. N. Zeilinger, A. Schlueter, “Real building implementation of a deep reinforcement learning controller to enhance energy efficiency and indoor temperature control”, Applied Energy, 368, (2024), 123447.
19. [19] J. O. Abass, M. Shamsudeen, B. N. Obaju, “Energy Optimization in Smart Buildings Using Deep Q-Network-based Reinforcement Learning”, Asian Journal of Advanced Research and Reports, 19(8), (2025), 30-41.
20. [20] B. Zhang, W. H. Hu, A. M. Y. M. Ghias, X. Xu, Z. Chen, “Multi-agent deep reinforcement learning based distributed control architecture for interconnected multi-energy microgrid energy management and optimization”, Energy Conversion and Management, 277, (2023), 116647.
21. [21] Z. Wang, T. Z. Hong, “Reinforcement learning for building controls: The opportunities and challenges”, Applied Energy, 269, (2020), 115036.
22. [22] A. N. Sayed, Y. Himeur, F. Bensaali, “From time-series to 2D images for building occupancy prediction using deep transfer learning”, Engineering Applications of Artificial Intelligence, 119, (2023), 105786.
23. [23] T. Do-Duy, D. V. Huynh, O. A. Dobre, B. Canberk, T. Q. Duong, “Digital Twin-Aided Intelligent Offloading with Edge Selection in Mobile Edge Computing”, IEEE Wireless Communications Letters, 11(4), (2022), 806-810.
24. [24] A. Al-Fuqaha, M. Guizani, M. Mohammadi, M. Aledhari, M. Ayyash, “Internet of things: A survey on enabling technologies, protocols, and applications”, IEEE Communications Surveys and Tutorials, 17(4), (2015), 2347-2376.
25. [25] J. W. Jolles, “Broad-scale applications of the Raspberry Pi: A review and guide for biologists”, Methods in Ecology and Evolution, 12(9), (2021), 1562–1579.
26. [26] Thermal Environmental Conditions for Human Occupancy, ANSI/ASHRAE Standard 55-2020.
27. [27] M. Rady, M. K. I. Muhammad, S. Shahid, “Evolving typical meteorological year (TMY) data for building energy simulation: a comprehensive review of methods, challenges, and future directions”, Advances in Building Energy Research, 19(3), (2025), 269–299.
28. [28] Q. Dong, Z. K. Ma, C. Sun, “Occupancy of rooms in urban residential buildings by users in cold areas of China”, Building Simulation, 16(3), (2023), 483 – 497.
29. [29] P. D. Andersen, A. Iversen, H. Madsen, C. Rode, “Dynamic modeling of presence of occupants using inhomogeneous Markov chains”, Energy and Buildings, 69, (2014), 213-223.
30. [30] X. Q. Yuan, J. P. Li, E. E. Kuruoglu, “Robustness enhancement in neural networks with alpha-stable training noise”, Digital Signal Processing, 156(A), (2025), 104778.
31. [31] Z. Wang, M. B. Luther, P. Horan, J. Matthews, C. L. Liu, “On site solar PV generation and use: Self-consumption and self-sufficiency”, Building Simulation, 16(10), (2023), 1835-1849.
32. [32] L. Blank, A. Tarquin, Engineering Economy, 8th edition, New York, USA, McGraw-Hill Education, 2018.
33. [33] H. Dai, N. Li, Y. Wang, X. Zhao, “The Analysis of Three Main Investment Criteria: NPV IRR and Payback Period”, 2022 7th International Conference on Financial Innovation and Economic Development (ICFIED 2022), Zhuhai, China, 14-16 January 2022, 185–189.
34. [34] M. M. Mathews, “Mathematics of Finance: Internal Rate of Return (IRR)”, International Journal of Science and Research (IJSR), 12(12), (2023), 863–864.