Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{175651,
        author = {P Ramya and Dr. K Venkataramana and V Sridhar},
        title = {.FFM: REVOLUTIONIZING FLOOD FORECASTING USING DECENTRALIZED FFNN AND CNN2D ALGORITHMS},
        journal = {International Journal of Innovative Research in Technology},
        year = {2025},
        volume = {11},
        number = {11},
        pages = {3495-3503},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=175651},
        abstract = {Floods are one of the most common natural disasters that often occur and cause serious damage to property, agriculture, economy and life. Flood forecasting presents a major challenge for researchers who have been battling against forecasting floods for a long time. The flood prediction model was proposed using federal learning techniques which ensures data protection, guarantees data availability, promises data security, and predicts flooding by banning data transferred over the network for model training. Flood Forecasting Model (FFM) is the most advanced machine learning technology (ML) that conducts ding tests. Federal Learning technology seeks training local data models in the field instead of sending huge data records to central servers for local models aggregation and training, it focuses on transferring these local models within the network server. This proposed model integrates a local training models data segregated from eighteen clients investigation at which station flooding is about to happen and generates flood alarms at a 5-days lead time. Local models of Feed Forward Neural Networks (FFNN) are trained at client stations where tides were expected. The flood forecasting module of the local FFNN model predicts the expected water level by taking several regional parameters as inputs. Data records for five different rivers and barrels were collected between 2015 and 2021 and took into account which includes four aspects such as rainfall-runoff, snow melting, hydrodynamics and flow routing. The proposed flood forecasting model predicted that previous floods in selected zones occurred with an accuracy of 84% from 2010 to 2015.},
        keywords = {Feed Forward Neural Networks(FFNN), Federal learning, Flood Forecasting Model, Hydrodynamics, Machine Learning.},
        month = {April},
        }