Thesis Project: 30 hp Artificial neural networks for auxiliary brake diagnostics

Scania is now undergoing a transformation from being a supplier of trucks, buses and engines to a supplier of complete and sustainable transport solutions. The next generation powertrains with increased efficiency and electrification are important steps in this transformation as well as meeting the science based target of carbon dioxide reduction.

In the department for Engine Control Systems we are responsible for development of control systems for combustion engines for all powertrains in the group. An intensive technical development is ongoing to bring our already world leading engine platform to the next level of sustainable transport solutions.

A crucial part of this is the technology shift we are doing within gas exchange system. By controlling gas flow with the engine internal valves replacing external flaps a more efficient engine with further improved customer values can be achieved.

Background
Compression release brake (CRB) is an integral component of engine braking technology. CRB is an engine braking mechanism installed on some diesel engines. When activated, it opens exhaust valves to the cylinders. This is done immediately before the compression stroke ends, which releases the compressed gas trapped in the cylinders and slows the vehicle. This system enhance braking efficiency in diverse engine types. However, challenges such as oil leakage, rocker arm fractures, and valve lash can compromise the functionality of the CRB system.

A notable challenge arises when CRB faults occur in cylinders that are expelling air into a intake manifold. Despite surpassing hardware limitations, the inability to identify affected cylinders or pinpoint the specific crank angle of the fault poses a considerable risk. This highlights the necessity for diagnostic capabilities across a broader range of cylinders to ensure comprehensive fault detection and resolution within engine systems.

Thesis description
This thesis proposal is to utilize artificial neural networks (ANN) for detection by looking at different related inputs and then ultimately predict when CRB is active. Preliminary truck data will be available for analysis and the writers should then conduct a theoretical analysis and test different ANN architectures such as FCN, CNN, LTSM to see which will yield most satisfactory results to detect a malfunctioning CRB. The writers are free to use any programming language of choice, e.g. MATLAB, python, R, etc. for analysis.

Implementation and testing
When the network is ready it should be implemented directly to the engine management system (EMS) in form of embedded C code. The proposed diagnostic method will with the aid of the supervisor be applied in a standalone engine and/or truck and be tested. The resulting data should be used for an final analysis of the product.

The writers
Applied physics, mechatronics, electrical engineering or computer engineering M.Sc. students are well suited for this assignment. Additionally it is beneficial to have experience with machine learning and embedded systems. This thesis could be conducted by one student alone or by two working in pair.

Number of students: 1-2 Student
 
Start date for the thesis work: from 7 jan 2025
 
Estimated time required: 30 hp
 
Contact persons and supervisors:
Emil Sorcini ENEFV (supervisor), Fredrik Löwgren ENEFV (acting manager)

Application
Your application must include a CV, personal letter and transcript of grades.

A background check might be conducted for this position. We are conducting interviews continuously and may close the recruitment earlier than the date specified.