Big Data in the field of automotive testing is mainly about the analysis of huge amounts of data that result from a wide range of test benches and driving tests. The aim is to identify new patterns and correlations in these test data in order to describe a complex system of different components and to predict the future behavior of this system. In this way, e.g. it is possible to gain important insights for the development of new simulation models and to achieve considerable savings in the execution of cost-intensive tests. Among other things, the basis for this are tools for Data Mining, Predictive Analytics or Machine Learning, which are nowadays available in a large number on the market – in some cases even free of charge.
However, the use of these tools requires that well-documented test data is available. In practice, this is often the point at which the problem begins: On the one hand, many test engineers do not consider the maintenance of meta information as “sexy”. On the other hand, extensive knowledge of the test context is crucial for interpreting measurements correctly in the long term and independently of individual persons. If the systematic documentation of the test context is missing, the measured data is only a column of figures that lose value very quickly.
Up to now, many test departments neglect metadata management – if one can speak of it at all. In particular, its embedding in the test processes is at best deficient. Guidelines for dealing with metadata, which prescribe uniform processes and responsibilities for data entry and maintenance (= Test Data Governance), are often looked for in vain. In addition, many of the systems used for measurement data management (MDM) do not offer sufficient functionalities to support and enforce such Test Data Governance. Unfortunately, the companies still give not enough attention to this fact during the planning and selection of MDM platforms.
This results in numerous, often department-specific and isolated solutions with individual and mostly rudimentary documented test data stocks. These are only little suitable as a basis for the implementation of Big Data analyzes deployed across many domains.
Conclusion: Test departments that want to tap the potential of Big Data in the near future are doing well to deal with the topic of Test Data Governance today. Standards such as ASAM ODS provide the appropriate basis for this. Software platforms, such as openMDM or Peak Resource Planner, help to anchor the topic firmly in the company-specific test processes.