Analytics in On Board Diagnosttics (OBD) and Diagnostic Trouble Codes (DTCs)

What are On Board Diagnostics (OBD)?

Onboard sensors are increasingly capable of monitoring and reporting vehicle status, providing OEMs with reliable, real-world usage metrics. Two key motivations for collecting these metrics are to reduce the occurrences of problems leading to diagnostic trouble codes (DTCs) and to schedule predictive maintenance.

Why Use Analytics on OBD Data?

Analytics must be performed on enormous data sets, often in real time, to determine when nominal ranges are exceeded. If not performed in real time, the data are recorded as summary statistics which attenuates the information content and can obfuscate root cause analysis. Identifying high-dimensional variable combinations and histories which lead to DTCs and predict maintenance issues is non-trivial. To overcome these computational barriers and effectively employ OBD data, OEMs must utilize predictive modeling and advanced analytics.

Challenges of Analyzing OBD Data

Performing analytics on data from OBD systems faces a number of formidable challenges such as sensor noise, the quantity of data and the number of potential inputs, limitations of on-board processing, low bandwidth for data transfer, variability in the as-built vehicle, and uncertainty about use and failure modes. These challenges create a seemingly intractable high-dimensional problem with enormous sample sizes.

SmartUQ Solutions for OBD and Embedded Systems

SmartUQ addresses these challenges using a variety of analytics tools:

• Dimensionality reduction, data subsampling, and filtering tools can reduce the amount of data that needs to be processed, stored, or transmitted.
• Sensitivity analysis allow important inputs to be identified and analyzed.
• Statistical calibration and inverse techniques can be used to identify bias signal in noisy data, determine underlying input uncertainties, and tune operating parameters and models.
• Advanced lightweight predictive modeling tools can be used in virtual sensors and real-time prediction for performance optimization.
• Combining it all together, digital twins and probabilistic analysis can be used for root cause analysis of error codes and engine failures.

Summary

Collecting and analyzing on board diagnostic data successfully can be challenging, but with appropriate tools and support, the knowledge of vehicle operations and the opportunities in root cause analysis, predictive maintenance, and advanced control are unprecedented.

To learn more about analytics for OBD applications, email us at [email protected].