An IFIC team participates in the development of a traffic alert system based on artificial intelligence to anticipate episodes of high pollution in Valencia

The system makes it possible to anticipate 30 minutes in advance if a stretch of road is going to register a high level of traffic, thus facilitating the adoption of preventive measures to reduce pollution and protect the health of citizens. In addition, it has been trained with data from 1,472 traffic sensors distributed throughout the city and classifies each track segment into three alert levels 

A team from the ITACA Institute of the Polytechnic University of Valencia (UPV) and the Institute of Corpuscular Physics (IFIC), located in the scientific-academic area of the University of Valencia Science Park (PCUV), and joint centre of the Higher Council for Scientific Research (CSIC) and the University of Valencia (UV), has developed an innovative urban traffic prediction and early warning system based on deep learning techniques. 

Applied to the city of Valencia, this system makes it possible to anticipate 30 minutes in advance if a stretch of street is going to register a high level of traffic, thus facilitating the adoption of preventive measures to reduce pollution and protect the health of citizens. 

The ITACA (UPV) and IFIC (CSIC-UV) team’s work is based on one premise: reducing transport emissions not only helps to mitigate climate change, but also directly improves air quality in cities. In the case of Valencia, for example, traffic accounts for around 60% of total greenhouse gas (GHG) emissions. 

The study is based on a premise: reducing transport emissions not only helps to mitigate climate change, but also directly improves air quality in cities

 "Urban traffic is a major source of harmful air pollutants. We must not forget that air pollution is the main environmental cause of premature deaths," says Edgar Lorenzo-Sáez, a researcher at the ITACA Institute and one of the authors of the study.

The ITACA researcher recalls that poor air quality has been linked to diseases such as asthma, lung cancer or cardiovascular problems, "responsible for some 300,000 premature deaths per year in the EU". 

A precise, reliable and scalable system 

The system developed by the UPV and IFIC team has been trained with data from 1,472 traffic sensors distributed throughout the city and supplemented with meteorological variables (wind, rain, atmospheric pressure...). Classifies each track segment into three alert levels and, thanks to the use of neural networks such as LSTM (Long Short-Term Memory), achieves high accuracy in real time, even at peak hours. 

In addition, the model has shown that traffic data can serve as a reliable indicator of levels of NOx (nitrogen oxides), one of the most harmful pollutants to health, especially useful in environments where there is no dense network of air quality sensors. This capacity would strengthen the effectiveness of Low Emission Zones (LHE), with measures more localized and adjusted to the real risk of each street, avoiding generalized restrictions of greater social impact. 

"Our system is correct in 90% of the cases when traffic is fluid and 70% when it anticipates episodes of high traffic. This opens the door to more agile decisions that avoid exceeding legal pollution limits in sensitive areas", adds Edgar Lorenzo-Sáez. 

"Artificial intelligence can be a great ally for cities to breathe better. This system, developed in Valencia, is ready to be exported and help improve air quality in urban environments around the world", Verónica Sanz, IFIC researcher and co-author of the study

For his part, Javier Urchueguía, also a researcher at ITACA, points out: "We have found a direct correlation between traffic flows and recorded NOx levels, which allows us to generate alerts even without a complete network of air quality sensors. It is a key finding for many European cities with limited resources".  

Verónica Sanz, UV professor, IFIC researcher and co-author of the study, explains that system 'intelligence' has been developed using AI models able to learn how the city 'breathes' and anticipate changes in traffic and pollution. "We have worked to make these models robust and adaptable to different scenarios, which opens the door to their application in many other populations," she says. "Artificial intelligence can be a great ally for cities to breathe better. This system, developed in Valencia, is ready to be exported and help improve air quality in urban environments around the world," she says. 

A step towards more sustainable and resilient cities 

This work represents a significant advance in data-driven urban management, integrating artificial intelligence as a tool to address complex environmental challenges. According to the authors, the system can become an essential instrument for designing more dynamic, effective and socially accepted interventions, especially aimed at protecting vulnerable groups such as schoolchildren, elderly people or patients with respiratory diseases. 

Future development lines include the creation of a digital twin in the city of Valencia to simulate measurements before their actual application and the incorporation of additional IoT sensors to improve direct prediction of pollutants. 

The system can become an essential tool for designing more dynamic, effective and socially accepted interventions, especially aimed at protecting vulnerable groups such as schoolchildren, elderly people or patients with respiratory diseases

The study was published in the scientific journal Neural Computing and Applications and has been supported by institutions such as the Generalitat Valenciana and the Ministry of Science, Innovation and Universities. 

This research follows the line of work recently undertaken by IFIC and UPV in the development of a system to accurately measure urban traffic emissions in cities, that has made it possible to know the neighborhoods most affected by traffic pollution in Valencia. 

Source: UPV e IFIC