Flooding alert system for Kristianstad Municipality using machine learning

Kristianstad municipality has a big problem with flooding every year because Kristianstad city is one of the lowest cities in Sweden. The Kristianstad municipality uses 550 sensors to detect the water in rivers, seas, and channels. Those sensors communicate with the IoT portal, where the data visualization is preprocessed. Using a machine-learning algorithm can help the municipality reduce the time and resources used to detect the risk of flooding. Collecting data and preprocessing is an important prerequisite to using machine learning. We use the LSTM Long Short-Term Memory algorithm for machine learning in this work. LSTM is one of the artificial recurrent neural networks (RNN) that uses Deep Learning (DL). This recurrent algorithm can capture long-range dependencies. Our solution makes predictions of future water levels. It is desired to make rapid forecast with real-time virtualization.