Automotive on-board diagnostics systems enable vehicle self-diagnostic and reporting capabilities. They provide the vehicle owner or a repair technician access to the information related to the performance of various vehicle sub-systems. Since the introduction of on-board vehicle computers, the amount of diagnostic information available via such on-board diagnostics systems has increased dramatically, reaching in some cases thousands of possible codes per vehicle. In addition to standardized diagnostic trouble codes, these systems provide various real-time vehicle performance data, which allow the technicians to rapidly identify and remedy malfunctions within the vehicle. Rapid expansion of codes and other information available through on-board diagnostics systems creates a need for fast identification and classification of the associated problems.
Researchers at the University of Michigan have developed a machine learning technology that generates a knowledge base for vehicle diagnostic classification. This knowledge base relies on a training set of vehicle diagnostic text descriptions associated with the appropriate diagnostic code. In addition, they also developed a vehicle diagnostic text classification technology that takes any vehicle problem description and accurately assigns it a diagnostic code.
Applications and Advantages
- Correlation of vehicle diagnostic codes the vehicle malfunction/performance problems
- Easy correlation between diagnostic code and related vehicle problem
- Ability to derive a diagnostic code based on vehicle performance/problem description
- Could enable end-users to independently determine the severity of the problem indicated by the on-board diagnostic system.