How StellarAi's Advanced 'Event-based Search' helped an OEM locate a crucial problem from their big data.

How StellarAi's Advanced 'Event-based Search' helped an OEM locate a crucial problem from their big data.

The Problem 
While working closely with one of our OEM customers, DriveTech was tasked diagnosing a critical issue they were facing with their proto vehicle fleet. Essentially the requirement was to collect network data from affected vehicles, analyze it and come up with the correct diagnosis in a week.  

 

The Solution 

DriveTech installed data logger devices and recorded the CAN network data from the vehicles and their multiple networks. This data acquisition was done for a week in hope to catch a replication of the failure event on the CAN log. During this time the vehicles were routinely driven for 100 to 150 kms every day, generating large amounts of data. The devices captured multiple failure events, and now the data had to be analyzed.  This case was very critical and demanded urgent resolution as the failure was adversely affecting the test program. DriveTech was left with ~ 100+ GBs of data (500+ files) which had to be analyzed. Each file would take around 45 minutes for analysis and analyzing 500+ files would mean spending a month or more just isolating the events. 


The Turning Point 
     To specifically address this, we decided to leverage our in-house Data Analytics platform, StellarAi. This platform excels at handling vast amounts of data with minimal processing time. The standout feature that proved invaluable in this case was the Advanced ‘Event-based’ Search functionality. We were looking for specific events within this huge amount of CAN data, and the advanced event-based search needed only a 1-line query to locate the files containing failure events in less than a minute.      

We were able to quickly visualize the data from these files, which helped us pinpoint the precise instance when failure occurred. We were able to visualize multiple such files simultaneously which helped us narrow down the potential causes.    
Collective analysis of files from all the affected vehicles enabled us to precisely identify the cause of failure and come up with a diagnosis. We were also able to recreate the failure on a bench set up, validating our diagnosis.


Conclusion  

After some more on-road testing & inspection, we were able to definitively conclude the reason for failure and notified the Automaker with potential insights. DriveTech was able to use StellarAi’s value added features and converge the analysis quickly. Without features such as event based Advanced Search it would have taken anyone a very long time to resolve the issue; while losing valuable time in getting new models into the market.