SFREOM: A Smart Flood-Resilient Energy Optimization Model for Sustainable Disaster-Response Solar PV-Battery Systems
In this paper, propose the SFREOM (Smart Flood Resilient Energy Optimization Model), a hybrid physics-informed and artificial intelligence-based framework that aims to provide solar PV-battery systems with enhanced reliability, such as during disaster-response. SFREOM incorporates three components: a Hybrid Metaheuristic Optimization (HMO) process for optimal configuration of solar-battery systems which uses immediately-discovered Particle Swarm Optimizations (PSO) and Genetic Algorithms (GA); a Flood Impact Prediction Module (FIPM) which builds on Physics-Informed Neural Networks (PINNs) to predict demand surge due to flooding; and a Dynamic Load Balancing Controller (DLBC) based on fuzzy logic with adaptive prioritization of loads for critical infrastructures, whilst shedding loads for non-critical infrastructures. The overarching impact of the mentioned modules enhances the responsiveness of renewable micro-grids to disaster events to optimize sustainable energy planning for disaster-resilient environments. Our experimental results showed significant improvements on prediction, optimization, and resilience outcomes: (1) FIPM enhanced the accuracy of predicting flooding-induced demand surge by 15-20% compared to traditional neural nets, (2) HMO improved lifecycle costs by 35.5% compared to traditional linear optimization, (3) the DLBC allowed for set-point supply continuity of 100% whilst shedding loads for all non-critical infrastructures therefore improving resilience by 30-33% versus a baseline controller. Finally, stress-test simulation runs of SFREOM during an extreme flooding scenario demonstrated a reduction of 25% of lost supply contract performance (blackout events), and (4) our computational analysis of SFREOM showed that the convergence of HMO was 30-40% faster than traditional optimization; therefore, enabling decision-making in real-time. In conclusion, the experimental outcomes provide its confirmation to be an effective framework for developing disaster-resilient smart energy networks.