Rozas PhD project has been a collaboration with Volvo Trucks, aiming to optimize vehicles based on customer usage. The focus has been on detecting and estimating the number of events related to steering components, such as curves and manoeuvres, to find out how much damage the steering components are exposed to and to be able to improve the design elements.
The tests have been made with both simulated and measured real data. In general, customer vehicles have no access to measurements dedicated to durability, and using specially equipped vehicles in service is difficult and expensive. Therefore, the events have been detected by using the information available for all vehicles known as Controller Area Network (CAN) bus data. By counting the number of driving events, one can estimate the fatigue damage caused by the same kind of events.
The results have been good and the job has been extended to estimate the number of events online as a new observation arrives without needing to store the previous observations. The proposed statistical method to estimate the number of events in the project was hidden Markov models, and in the end it could be concluded that the method works well both online and offline.
Roza came to Gothenburg in 2009 with a master in mathematical statistics from Allameh Tabatabaei University in Iran to get a second master in financial mathematics at the University of Gothenburg. She learnt how to use mathematical statistics in finance and found the teachers and supervisors to be very good. During her second master, she got the position as PhD at this project.
– This has been a very nice environment and I have also had good colleagues, as well as a good steering group at Volvo who really improved my knowledge about automative applications, I will miss them! But we are going to apply for money and hope that the project will continue to create knowledge that lead to more cost efficient transport vehicle specifications with high energy efficiency.
Roza has also been teaching and is right now the main teacher of the course Mathematical statistics and discrete mathematics, for Computing Science and Engineering. She says that she loves teaching, and concludes that she is really happy that she got this position where she has got good experience for the future.
Roza Maghsood will defend her PhD thesis “Hidden Markov models for detecting steering events and valuating fatigue damage” on September 23 at 13.15 in the room Pascal, Hörsalsvägen 1. Supervisor is Igor Rychlik, co-supervisors are Pär Johannesson and Jonas Wallin.
Text and photo: Setta Aspström