Development and comparison of synthetic data generation models for on-board diagnostic data

Context

Data plays a vital role in Machine Learning (ML) by providing the foundation for training and improving ML models. In real-world systems like On-Board Diagnostics (OBD), the implementation of ML can be set back by the poor quality and limited amount of data, which can result in underfitting of the ML models. OBD systems have the potential to enhance the durability and reliability of vehicles by utilizing AI-based analysis, predictive diagnostics, predictive maintenance, etc. However, acquiring a high volume of OBD data, such as SAE J1699 Compliance Test Log files, poses challenges due to the risk of exposing sensitive information. To address this issue, synthetic data generation offers a promising alternative. By creating synthetic data, it becomes possible to generate large amounts of OBD data that can be used to train high-quality ML models. Synthetic data generation enables the free use of vast amounts of OBD data, ensuring the security and privacy of customers' sensitive information.

Targets

• State of the Art in Synthetic Data Generation Models
• Investigation of the principal attributes of On-Board Diagnostics data, with focus on SAE J1699 Data Log files
• Successful implementation of the suitable synthetic data generation model
• Research and development the evaluation concept for Synthetic generated data

Requirements

• Interest in the development of AI/ML systems
• Basic knowledge and experience in data processing
• Programming skills (Python/C++)

We at RA Consulting are ready to continue the cooperation with you after your Bachelor thesis through future Internship, and/or Master thesis!