Link Engineering Company

Applying Data Mining Techniques to a Population of Acquired Automotive and Commercial Vehicle Brake Testing Data

Background:

Link Engineering Company (LINK) provides test equipment and testing services to the global transportation industry. LINK has developed software tools for the control of test equipment and the acquisition of test data. Additionally, LINK has developed a laboratory information management system (LIMS) that is used at all LINK test facilities and numerous customer sites. Creating tools that extract patterns of knowledge from a large population of acquired test data is the next logical step as the control and acquisition software intersects with the LIMS system. To this end, LINK wishes to develop the tools necessary to indefinitely warehouse the acquired test data in a manner that supports reporting and data analysis across the entire population. Presently, there are 10’s of terabytes of acquired data which is growing daily.

Description:

Develop an AI model to identify unknown trends in acquired test data from inertia dynamometers, vehicle tests and other inputs while providing an interface to query the data population directly. The ability to anonymize data sources (customer, make, model, and platform) is required. Using terabytes of acquired test data in a proprietary format, the team will develop the tools necessary to store, extract and process the data in preparation for analysis. Analysis will first seek to identify unknown trends in the data and provide an interface for reporting those trends. Additionally, the tool will provide an interface for a range default queries and allow for customized queries. Output from the tool will be both graphical and tabular with the ability to export to popular formats for further processing or presentation.

Goals:

1. Review the existing LINK data set and data extraction tools to determine the best mechanism for warehousing the data.
2. Convert and migrate the data in preparation for data analysis.
3. Develop analysis tools to find unknown trends in the data and provide an output mechanism (graphical and tabular).
4. Develop a mechanism to create a set of default queries with related graphical and tabular outputs.
5. Develop a mechanism to create customized queries.
6. Document the system to facilitate integration into other LINK tools.
7. Variably anonymize data for presentation of trends without revealing underlying confidential information specific to a customer, make, model, platform, etc.