Metro de Madrid Adopts Machine Learning for Predictive Maintenance in Tunnels

“We have created a degradation model of the catenary that allows us to anticipate and optimize the maintenance actions.”

Key Outcomes

  • Saved time in the data validation and analysis phase
  • Integrated data from different sources
  • Shared algorithms with non-MATLAB users

Every day, Metro de Madrid stores more than 10 GB of new data acquired from different sources. Many available tools can only analyze data from a single sensor, and such approaches lack domain expertise. In order to use all the data they acquire for predictive maintenance, Metro de Madrid needed to integrate the data from a wide variety of sensors and customize their signal analysis algorithms.

Metro de Madrid used MATLAB® and Statistics and Machine Learning Toolbox™ to automate the data merging, signal analysis, and algorithm sharing, which enables people without MATLAB experience to perform advanced signal analysis.