Fault detection and prognostics in train traction auxiliary motors

1. Context
Maintaining a train in optimal operating conditions requires a smart monitoring of its critical components (pantograph, bogie, doors, motors, etc.). Condition monitoring aims at collecting data, which are then processed and analyzed to track the health state of the subsystems and, consequently, that one of the train. This work is will be done in the framework of PHM (Prognostics and Health Management (PHM). PHM concerns the development of methods and algorithms for condition monitoring of systems, fault detection, diagnostics, prognostics and decision support. The implementation of PHM allows the user to increase the part of smart maintenances (condition-based and predictive) while reducing the part of traditional maintenances (corrective and time-planned). This leads to an improvement of the availability, reliability and security of the train while decreasing the exploitation costs.

2. Objective of the study
In this context, ALSTOM desires to develop tools and methods that allow assessing and predicting the health state of its trains’ sub-systems. The purpose of this work is to propose a method that tracks the health state of the traction auxiliary motors in a train. In the first time, this work will focus on the ventilation systems used to cool the components of the traction chain. The other components on which ALSTOM wants to apply the developed method are the motors of the pumps and also the motors of the braking rheostats ventilators. Moreover, the proposed method may be applied on the traction motors. The objective is to track the health state of the auxiliary motors, particularly that one of their bearings, in order to plan the maintenance interventions.
In this study, we will focus, in priority, on fault detection of bearings by working on the analysis of the electrical signals provided by the on-board inverter. This analysis will allow online fault detection, during the operation of the train. It will be completed by the analysis of the vibration signals for offline fault detection, during the maintenance interventions.

3. Workflow
The study is carried out within the framework of this project will last for 18 months. The project is structured in four actions: a first action dedicated to the understanding of the system, its environment and the completion of the test bench at the PRIMES platform in Tarbes. A second action related to fault detection on bearings based on electrical analysis. A third action concerning the fault detection based on vibration analysis. Finally, the fourth action will deal with fault prognostics.

- Action 1: it concerns the understanding of the system (the chosen auxiliary motor, its components, its operation, its environment, the monitoring module, etc.). This work will be carried out in parallel to the implementation of the test bench for fault detection and fault prognostics of the auxiliary motor’s bearings (installed in ALSTOM’s trains).
The test bench will be composed of two auxiliary ventilation motors provided by ALSTOM (one for fault detection and one for fault prognostics). Each one of these motors will be equipped with a load.
This action, particularly the experimental tests, will be carried out during the whole period of the project. Intermediate review meetings will be organized to discuss the work progress. In this project, the experimental scenarios and the construction of health indicators (action 2) will be achieved offline, on a dedicated computer for measurements and signal processing.

- Action 2: it deals with fault detection, based on electrical signal processing, and with the construction of health indicators that track the evolution of bearing degradation.

- Action 3: it will be carried out in parallel with the action 2 and concerns fault detection of bearings by using vibration analysis. This consists in testing different fault detection methods to highlight their limits and propose an approach that allow a better fault detection of the bearings. Also, it will be interesting to identify a hierarchy in the cost of each proposed solution (e. g. possibility of using the sensors already present on the train)

- Action 4: it concerns the development of a fault prognostic method (prediction of remaining useful life of bearings), by using the health indicators constructed in action 2.

4. Contacts
Bertrand CHAUCHAT (bertrand.chauchat@alstom.com)
Diego Alejandro TOBON-MEJIA (diego.tobon@alstom.com)
Kamal MEDJAHER (Kamal.medjaher@enit.fr)
Pascal MAUSSION (Pascal.Maussion@laplace.univ-tlse.fr)
Antoine PICOT (Antoine.Picot@laplace.univ-tlse.fr)

5. Required skills
We are looking for a PhD who has defended his/her thesis in the domain of fault diagnostics and/or fault prognostics. In addition to these latter skills, the candidate should have strong knowledge in electrical engineering (electrical machines) and signal processing. He/she has also to be good in setting experimental test benches (hardware and software) and possesses skills in programming (Matlab and Labview). Finally, the candidate should speak English fluently, intermediate French speaking would be appreciated.


follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed