Data Scientist Intern

Description de l'offre

Alstom

The Railway industry today is characterized by both a strong and sustained growth across the world. The trends that drive this are well known: environmental challenges, population growth, urbanization and increasing demands for mobility.
With 6B€ of Sales and around 30,000 employees based in over 60 countries, Alstom develop & market the most complete range of systems, equipment and services offered today in the railway sector. Today we offer our customers solutions that feature a seamless blend of diverse technologies, ensuring optimal interfaces, along with flexible implementation and real synergy in innovation.

Context

Prognostics and health management (PHM) consists of assessing the current and predicting the future health state of a system to optimize its maintenance. PHM is performed from the embedded extraction of relevant features named health indicators (HIs), from which symptoms are generated. Then, some processing algorithms elaborate diagnosis and prognosis that are used to maximize the system’s availability and reliability by optimizing the maintenance schedule.

Alstom Digital Mobility is launching various other innovative programs which will require data science skills , such as ‘intelligent traffic management”,  driver assistance, energy management, advanced field reliability analysis, etc.

Job description

The health assessment of railway infrastructure is challenging and expensive, mainly because of the extensive length of the railway network and also due to the wide variety of defects that could affect the tracks. Early warning systems that minimize the disruption of the network can have a beneficial impact on the reliability and on the availability of the infrastructure.

In the context of its predictive maintenance program, ALSTOM aims at offering Prognostics and Health Management tools to diagnose different types of failure modes on the track infrastructure and prevent them.

Rolling stock has been instrumented with advanced data acquisition technologies to collect data about rail conditions and to localize rail defects. Damaged rail infrastructure is associated with localized deformations of the track which lead to deviation in the measured acceleration of the vehicle and ground power supply issues that are monitored using a supervisor system. Appropriate data analysis techniques need to be designed to:

·  Clean up the data (on board data from rolling stock & off board data from the supervisor system)
·  Define some relevant features to diagnose the presence of irregularities arising from the infrastructure damage (respectively to each source of data)
·  Identify the links between on board data coming from rolling stock and off board data coming from infrastructure supervisor system
·  Elaborate a fusion approach to gather these both source of data

Candidate profile

Student in final year of engineering school or master 2 in engineering of industrial systems with option in dependability, monitoring systems, data science

Candidate skills

·  dependability, statistics, monitoring systems;
·  Computation software: Matlab, R, MySQL,NoSQL Minitab...;
·  Technical English;
·  Autonomy and initiative. 

The candidate  will spend 6 months in the Predictive Maintenance Center of  Excellence