Skip to main content

AIMEN’s services deployed in the NAIADES platform!


AIMEN’s development of AI services to provide and/or support water management solutions within the NAIADES platform has concluded. The NAIADES project aims to create a holistic platform for water management in urban areas. An interoperable, modular, scalable and secure cloud platform has been designed within the project, and it is currently on development and testing under the called ‘NAIADES Development platform (Dev)’. The NAIADES platform aims to gather as many as possible smart water management solutions (services and/or combination of them) and AIMEN has recently deployed in Dev two of these services: the Water Quality Forecast and the Dynamical Treatments Suggestions.

The Water Quality Forecast (WQF) is an AI based service whose objective is the provision of up to 48 hours forecast of water quality parameters such as pH, free Chlorine, turbidity and Chlorates (Figure 1). The core of the service is a trained Recurrent Neural Network model that uses past values of water quality and weather observations; and also uses the weather forecasts provided by other NAIADES service (the Weather Forecast) to generate the future predictions. In the NAIADES project context, the service will be applied to monitor the water quality in the public fountains of Carouge (Switzerland). The city of Carouge manage these fountains, which are used mainly during summer by citizens for bathing related activities. The current issues related to the management of these fountains imply dedicated time from city staff to monitor and control the water quality and maintenance activities; being common service stops and fountain closure to prevent health issues due to water quality drops. The WQF aims to inform beforehand the city staff about possible water quality drops so they can proactively plan actions to guarantee water quality and prevent service stops. The WQF service could be applied to any other similar infrastructures such as water reservoirs and water courses, where the weather is constantly affecting water quality.

figure 1

Figure 1. Example of WQF predictions vs the real observations.

The Dynamical Treatments Suggestions (DTS) is an AI based service whose objective is the suggestion of the most efficient treatments dosages to be applied in a drinking Water Treatment Plant (dWTP) for any water quality at its inlet. The service has been developed for dWTPs with specific treatment processes: precipitation (coagulation and sedimentation), filtration and chlorination. Using this very common dWTP configuration as reference, an approach based on Reinforcement Learning has been used to train a model (a transition matrix) that suggest the most efficient values of coagulant and chlorine dosages which always ensure the water quality at the outlet is within the drinking water thresholds defined by the World Health Organization (turbidity < 5NTU and 0.5mg/l < residual chlorine < 1 mg/l).

Both services were designed to be part of the internal NAIADES services (only backend), thus their user interface is provided by the NAIADES Human Machine Interface (HMI), Figure 2. The HMI is currently on its final development stage. Once it is finished and the development platform functionalities are properly tested, the NAIADES development platform will be ready for migration to a Production environment and the NAIADES validation will start, including the validation and evaluation of the aforementioned services.

figure 2

Figure 2. Current HMI Mock-up of the DTS service's User Interface.

AIMEN will continue providing support for both services until the end of the NAIADES project, adapting the services to any requirement requested during the NAIADES validation. After the project, the methodologies applied, and results obtained will be continued in future AIMEN’s research for optimization and application to other areas.