A Hybrid Model of Artificial Intelligence Integrated into GIS for Predicting Accidents in Water Supply Networks
Image: Accident prediction map generated using ANFIS-GIS hybrid model (Kyiv)Abstract
This article proposes a hybrid neuro-fuzzy model integrated into a geographic information system (GIS) to forecast accident locations in urban water supply networks. Combining ANFIS with genetic algorithms and ant colony optimization (ACO), the system allows spatiotemporal prediction with improved accuracy. The study was conducted using real data from Kyiv’s water infrastructure. Model performance was evaluated using MAE, RMSE, and R², achieving a forecast improvement of up to 11%. The hybrid approach demonstrates the potential for practical deployment in critical infrastructure monitoring.
Type
Publication
System Research and Information Technologies, 2(2024)
Due to the sensitive nature of the infrastructure involved, part of the source code and the dataset used in this study are not publicly available. These materials are classified under wartime restrictions and relate to critical infrastructure, thus cannot be shared or distributed.
This publication is based on research conducted at the Educational and Research Institute for Applied System Analysis (ERIAS) of Igor Sikorsky Kyiv Polytechnic Institute.
The proposed hybrid system combines:
- ANFIS: for fuzzy logic and learning from data;
- GA + ACO: to optimize fuzzy rule sets and clustering;
- GIS: for spatial context, visualization, and decision support.
The integration supports water utilities in making informed, predictive decisions regarding accident response and preventive maintenance. Use of metaheuristics improved predictive power while retaining interpretability.