Master Thesis: Development of AI-agent for Single-Cylinder Engine Testing

Background

Heavy-duty road transport is undergoing a significant transition toward CO2 neutrality. One promising carbon-free energy carrier is hydrogen, and an internal combustion engine is a key technology for converting hydrogen’s chemical energy into mechanical work. Development of an AI-agent (“Test CoPilot”) could potentially accelerate the development cycle by streamlining the testing process, potentially leading to faster breakthroughs in efficient and clean propulsion technologies.

About us

At Volvo Powertrain Strategic Development, we shape technology strategies and inform business decisions for new powertrain solutions. We actively explore emerging powertrain technologies, regulatory shifts, and future societal needs to prepare a powertrain roadmap. Furthermore, we develop technology solutions that support the transition from concept to production, enabling the next generation of sustainable transport solutions.

Thesis background

The use of single-cylinder engine (SCE) tests, coupled with design, CFD, and 0D/1D simulations, enables an efficient evaluation of future engine concepts before investing resources into full-scale prototypes. To maximize this potential, it would be of value to investigate how the development of an AI-agent, Test CoPilot, could contribute to the intent. This AI-agent could theoretically enhance the iterative cycle of SCE testing, including test preparation, test execution, data evaluation, and reporting. The AI-driven assistant could bring significant value by improving the efficiency and precision of these activities as future transport solutions demand increased knowledge productivity in this domain.

Content

Initially, a literature study should be conducted to gain an understanding of the target domain, including hydrogen pre-mixed combustion, SCE testing, and previous development of AI-agents within the department. The next step is to consider what key adaptations should be made to the agent for domain-specific adjustments. Several potential sub-topics could be prioritized, and depending on progress, this can be adapted towards a final demonstration based on a dataset aligned with the time plan.

A proposed thesis plan (to be aligned with examinator and students):

• Analyze current SCE testing processes with focus on data evaluation and reporting as key activities to “close the test cycle” identifying AI improvement opportunities.
• Develop a conceptual framework for the AI-agent (Test CoPilot).
• Implement a prototype focusing on data evaluation and reporting.
• Conduct case studies to evaluate the AI-agent’s performance against traditional methods.
• Refine the AI-agent based on results and explore expansion to other testing phases.
• The final step would be to demonstrate the Test CoPilot and verify its effectiveness in enhancing the overall test cycle, with emphasis on data evaluation and reporting improvements. The thesis should conclude with an analysis of the AI-agent’s impact on SCE-testing efficiency and decision-making support.

Research Depth

Research into how an AI-agent can enhance the cyclic test iteration by optimizing part of or the complete test cycle phase. The research will explore how the AI can adapt to the specific challenges of hydrogen pre-mixed combustion in single-cylinder engine testing, streamline data processing, and provide actionable insights. The objective is to enable more efficient decision-making and deeper integration of data-driven methodologies into the engineering workflow, ultimately contributing to the advancement of future powertrain solutions.

Suitable Background

We are looking for candidates in the final year of their Master’s studies, preferably from faculties of Computer Science, Mechanical Engineering, or Applied Physics. We highly value diverse perspectives and encourage applications from all backgrounds. An interest in automotive engineering, internal combustion engines, and AI-driven data evaluation is highly valued. Experience or interest in AI applications, data processing, or simulation tools will also be beneficial.