Paris : over 2,200,000 inhabitants which become strong>10.700.000 if we count the ’Unité urbaine de Paris and over 12.500.000 inhabitants if we consider the entire metropolitan area . It is a wide area served by a public transport system consisting of 16 subway lines to which we add the RER line of regional trains, the tram and bus network. It is a widespread, well-organized service for the citizen, essential for daily life around the Eiffel Tower area.

How to contribute to the effective operation of Parisian public transport? NetValue has responded to this question with its NetValue artificial intelligence NetValue project - specifically AI Machine learning – that provides predictive maintenance of the public transport‘s visual signaling panels around the metropolitan area, integrated with the city public lighting network.

The starting point was a precious and illustrated heritage via large dataset that allowed the possibility of a detailed analysis through AI Machine learning. From this analysis - especially if conducted on a large amount of data - it is possible to identify correlations and information still difficult to recognize through traditional technologies. An example of the application of these new potentialities is represented by predictive maintenance with the aim of predicting future systems failures or machinery malfunctions due to defective components, design defects or problematic environmental factors.

The feasibility test of this NetValue's project has provided positive results. The application analyzes the data relating to the operation of the system components (voltage, charger, temperature, video card, communication card, etc.). The information network coming from the single panel is sampled minute per minute (and stored in the data lake, a large archive). It is important to emphasize that this operation was carried out according to these times on more than 600 visual panels that were subjected to tests. Thanks to the AI Machine learning algorithms, it is possible to deal with anomalous situations before the system fails. In short, prevention is better than cure: once the anomaly has been identified, operational plans, including automated ones, are prepared, and aimed at recovering the critical situation.

The advantages of predictive maintenance compared to traditional reactive maintenance are evident. The result is a possible reduction in operating costs and increasing efficiency . This dvantages the rhythms of the city and the satisfaction of citizens.

Summary diagram of the project: