Innova-tsn wins Renfe’s challenge to develop a Predictive System for Graffiti Prevention using AI 

• The solution, selected in Renfe’s first Artificial Intelligence Project Competition, will make it possible to anticipate and dissuade acts of vandalism on trains and facilities, optimising surveillance and reducing operating costs.

Renfe has selected Innova-tsn, a leading consulting company specialising in the integral data lifecycle and artificial intelligence, as the winner of the challenge “Predictive System for Graffiti Prevention using AI” within its first Artificial Intelligence Project Competition, which seeks to promote the use of advanced analytics to solve key operational challenges in the railway environment.

Innova-tsn’s proposal stands out for its proactive, data-based focus to anticipate and reduce acts of vandalism, especially graffiti attacks on rolling stock and certain critical facilities.

An operating, economic and reputational challenge

According to Renfe, vandalism with unauthorised graffiti has a significant impact on railway operations: direct cleaning and repair costs, temporary removal of trains from service and deterioration of passenger experience and corporate image. In fact, in 2024, direct costs linked to graffiti exceeded 11 million euros for Renfe, highlighting the urgent need to move from reactive post-damage cleaning models to predictive and preventive strategies.

In addition to affecting trains, the phenomenon extends to depots, tunnels, stations and other assets, and has incorporated more organised tactics, such as the use of the emergency brake or “palancazo” to force stops, which require new capacities for anticipation and coordinated response between security, operations and maintenance.

The challenge posed by Renfe:

Renfe’s AI Laboratory launched the challenge with the aim of identifying solutions to:

• Analyse the history of graffiti incidents (location, type of attack, day, time, context) to discover recurrent patterns.
• Generate predictive alerts at least two hours in advance that pinpoint high-risk areas and time slots.
• Optimise surveillance resource allocation and prioritise patrol rounds depending on dynamic risk.
• Activate machine vision capabilities on existing cameras to detect suspicious movements or anomalous activities in depots and vulnerable areas at later stages.
• Ensure model explainability, modular scalability and alignment with Renfe’s safety and regulation compliance standards.

The solution presented by Innova-tsn:

Innova-tsn proposed a predictive system for graffiti prevention based on explainable AI and advanced analytics, designed to be gradually deployed and generate value from the first weeks, focused on three pillars:

1. Historical analytics and context enrichment: Integration of graffiti incident databases with external variables such as the weather, public events, work calendar and other environmental factors that influence the probability of attacks.
2. Time-space modelling of risk: supervised models which generate risk heatmaps by location and time slot for the next 24 hours, updated periodically with new operational information.
3. Early alerts and surveillance replanning: automatic issuance of prioritised alerts (high/medium/low) ≥2 hours in advance when risk thresholds are exceeded, accompanied by suggestions for dynamic adjustment of patrol rounds.

The coming weeks will see the construction of a secure cloud infrastructure, the integration of historical and context data, initial model training and the deployment of an operational dashboard for a set of high-incidence pilot depots.

The pilot will establish a measurement framework based on operational and economic KPIs that aim, among other variables, to reduce vandalism by 20%, issue 50% of alerts at least 2 hours in advance to enable preventive action, achieve estimated annual savings of €150,000 in cleaning, logistics and operational availability, and obtain a satisfaction level of around 80% among security personnel using the system.

After the first phase, the extension to new geographical areas and the progressive incorporation of machine vision capabilities on existing cameras will be evaluated.

Why Innova-tsn is the appropriate partner for Renfe

The proposal is backed by Innova-tsn’s experience as a leading consulting company in the integral data lifecycle, with a proven track record in advanced analytics, applied artificial intelligence, and the deployment of end-to-end solutions in critical operating environments. The company has demonstrated its technological and collaborative capabilities in leading innovation ecosystems, where it provides leadership in data, AI and sectoral expertise to address complex challenges in digitalisation, advanced detection (fraud, anomalies) and distributed infrastructure management.

According to Begoña Vega, Head of AI Models & Applications at Innova-tsn:

“We are very thankful to Renfe for placing its trust in us to address such a relevant challenge. At Innova-tsn we believe in applying AI where it really makes a difference: reducing the impact of vandalism, optimising resources and improving public services. We will bring our expertise in advanced analytics to bear to ensure that the pilot generates tangible results from the outset.”