Skip to main content

The role of deep learning and AI in the NAIADES project

Waterand AI

Smart Water and Water Treatment applications and services is a critical aspect of the NAIADES project. Towards that direction the adoption of various AI technologies is required for the development of a novel ICT network. Recent advancements in machine learning and deep learning (two main AI technologies) rendered machines capable of performing tasks in tech industry that typically required human intelligence until now. These technologies enable multi-tasking and easiness of workload, the capabilities of individuals with different skillsets are augmented and the decision-making process is carried out faster and smarter.

Digital rainNAIADES project aims to deliver a variety of applications and services to the local and regional water providers and authorities, such as failure prediction, comfort assessment, optimal resource allocation and water treatment management as well as enhanced decision-making technologies. As part of the DSS, a weather forecasting toolkit is being developed. Its main features are the monitoring of the micro-environment parameters (temperature, wind speed, humidity etc.) and the prediction of various factors by harnessing the power of deep learning. It should be highlighted that a short-term weather prediction (24 hours interval), along with a prediction of the comfort level in both indoor and outdoor environments and the indication of possible infrastructure failures are included in the toolkit underdevelopment.

Deep learning models must be generally effective across a wide range of inputs. Thus, a crucial factor that improves the accuracy of the weather predictions is the usage of data from areas with different physical characteristics. For that reason, deep learning models are being trained with data collected from European locations with heterogeneous water management cycles. In that way, the weather forecast toolkit will be demonstrated in 3 pilots in Switzerland, Spain and Romania and will be evaluated against performance and effectiveness indicators.