top of page


Early damage detection using big data analyzes on workshop data

The task

In order to avoid possible damage to vehicles, abnormalities should be identified at an early stage and appropriate measures taken. For this purpose, the memory data of defective vehicles are read out during the repair and compared with other collected findings data. If sufficient data is available, regularities can be identified that will in future be used for the early detection of errors in vehicles (predictive maintenance).


The challenge

In order to be able to make statistically reliable statements, the storage data must be evaluated over a very large number of individual vehicles and surveys over several years. However, the sheer size of the collected measurement data can no longer be managed with classic analysis tools, and the use of new advanced analytics methods is also required.

our solution

The evaluations were transferred to a specially designed Big Data & Advanced Analytics environment. The data loading sections set up on this allow the scalable embedding of existing domain software for quality assurance and the preparation of the data for analysis. The existing analysis approaches were supplemented by pattern recognition for comparison with defined error patterns and by an automated analysis of abnormalities.

The customer benefit

The transfer to regular operation enabled the expansion and execution of the analyzes on new data in self-service and use by the entire development department. As a result, a measurement data set collected over several years can be searched for patterns and abnormalities within a few minutes instead of several days. The statistical evaluations set up on the vehicle fleet support the early detection of errors and provide valuable information for data-driven vehicle development.



Our role

  • Support of the customer by data scientists, data engineers, DevOps

Our activities

  • Setup and configuration of a Hadoop cluster, transfer to and safeguarding of regular operations (including automation of data ingest)

  • Transfer of existing analyzes to Hadoop

  • Training the customer to take over the system

Technologies & methods

  • Applications: HDP, DaSense

  • Data / databases: HDFS, XML, ASCII, MDF

  • Languages / frameworks: Python, Java, Matlab, MapReduce, Yarn, Spark, Oozie, HBase, Knox, Jupyter, Ambari

  • Methods: Dynamic Time Warping, trend analysis, outlier detection

bottom of page