Predictive maintenance on a compressor air production unit of a subway train

Project

Produce a demonstrator integrating our own predictive maintenance model combining different AI algorithms and a performance evaluation engine enabling the best result to be selected according to context.

#IA #LSTM #CNN #XIA #RUL

Challenges & objectives

Needs analysis

State of the art

Setting up and preparing datasets

Design and development of AI models (Remaining Lifetime Estimation – RUL)

    • Long Short Memory Recurrent Neural Networks (LSTM)
    • A more complex hybrid sequence-to-sequence architecture
    • Setting up a data splitting strategy

Optimization: taking into account limited available resources

Explainability of results

Requirements

Data quality and volume

Complexity of appropriate AI models (interoperability, adaptability)

Industrial environment constraints

Integration and deployment

Our achievements

Each project reflects our ability to transform a critical requirement into a concrete solution. From the embedded software layer to functional safety, Médiane Système mobilizes all its expertise to deliver innovative, robust solutions ready for industrialization.

Predictive maintenance on a compressor air production unit of a subway train

Discover the realization
See all achievements