A novel AI-driven framework for trusted decisions and sustainable flood management in smart cities

Abstract

The management of floods is a vital problem that climate change exacerbates, as it produces more frequent and more severe flooding events. This study develops a combined artificial intelligence system which improves flood prediction and decision-making processes for urban environments. The system uses Convolutional Neural Networks (CNNs) to conduct drainage image analysis while using K-Means clustering to identify rainfall and blockage patterns and Random Forest classification to process structured data together with a semantic rule-based system which operates on real-time sensor information and weather API data. The model achieved an F1-score of 0.97 which demonstrates strong predictive performance while it offers explainable recommendations that assist stakeholders with managing flood situations. The research demonstrates how urban planners and disaster management authorities can develop advanced flood-prevention measures through their work. The study’s original contribution is the uniting of multiple real-time data sources with hybrid artificial intelligence methods and Explainable AI (XAI) to achieve flood prediction with transparent and precise results in smart cities.