Hybrid Neuro-Symbolic Architectures Integrating Physics-Informed Models for Real-Time Risk Prediction in Large-Scale Infrastructure Projects
International Journal on Advanced Science, Engineering and Information Technology · 25.10.2026 · 16 (5) · 1813-1819
DOI: 10.18517/ijaseit.16.5.14642 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal on Advanced Science, Engineering and Information Technology, online erschienen am 25.10.2026. Verfasst von Samer Adel Abd Hussein und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Large-scale infrastructure projects are sensitive to and complex in their response to environmental and structural conditions, and this requires a risk prediction mechanism that is accurate and real-time. While traditional AI algorithms are good at pattern recognition, they do not provide interpretability or do not necessarily follow the domain's physical laws, causing potentially unsafe and inefficient results when applied to critical engineering work. This paper presents a novel hybrid neuro-symbolic AI architecture to enhance prediction performance for infrastructure systems, in response to these challenges. The proposed framework includes three components: neural networks that learn adaptively from the data, symbolic reasoning modules that interpret the data, and physics-informed constraints that enforce physical laws during the learning process. The three layers ensure a high level of prediction accuracy, physical consistency, and human-readable argumentation. The experimental evaluation by simulation and infrastructure risk assessment benchmarking demonstrates that the proposed hybrid approach yields high prediction accuracy (93.5%), low average prediction latency (57ms), and interpretability thanks to the use of a symbolic model part. Physical fidelity is provided by PDEs as constraints; the symbolic layer gives high-level insights of actionable value. When combined, these features allow effective use of the model in the design, construction, and operational phases for dynamic risk assessment. This research lays a solid foundation for future developments in explainable and physics-grounded AI, showcasing its applicability in safety-critical infrastructure domains. Such a hybrid solution may pave the way for new opportunities for responsible and interpretable integration of AI into engineering systems.
Zitieren
APA
Hussein, S. A. A., & Abed, H. A. (2026). Hybrid Neuro-Symbolic Architectures Integrating Physics-Informed Models for Real-Time Risk Prediction in Large-Scale Infrastructure Projects. International Journal on Advanced Science, Engineering and Information Technology, 16(5), 1813-1819. https://doi.org/10.18517/ijaseit.16.5.14642
Vancouver
Hussein SAA, Abed HA. Hybrid Neuro-Symbolic Architectures Integrating Physics-Informed Models for Real-Time Risk Prediction in Large-Scale Infrastructure Projects. International Journal on Advanced Science, Engineering and Information Technology. 2026;16(5):1813-1819. doi:10.18517/ijaseit.16.5.14642
BibTeX
@article{Hussein2026,
author = {Hussein, Samer Adel Abd and Abed, Husam Abbas},
title = {Hybrid Neuro-Symbolic Architectures Integrating Physics-Informed Models for Real-Time Risk Prediction in Large-Scale Infrastructure Projects},
journal = {International Journal on Advanced Science, Engineering and Information Technology},
year = {2026},
volume = {16},
number = {5},
pages = {1813--1819},
doi = {10.18517/ijaseit.16.5.14642}
}
EUVIMED