Application Story

Challenge: At Robert Bosch GmbH Diesel Systems, Measurement data has become large scale; the number of files and data volume has skyrocketed. Today, engineers and experts in the domain are losing time and resources by manually developing and executing duplicate analysis programs, which they run only once, locally, on their hard-drives. In 80% of the cases, the tasks are repetitive as they inherit Standard data Validation and evaluation.

Solution: The approach to a solution was identified by integrating an analysis framework into the existing system architecture of the Product Lifecycle Management Database (PLCD), which includes an ASAM ODS Database for Measurement data management. The server-side framework defines and provides a workflow, which integrates automated evaluation calculations based on, for example, DIAdem scripts, MATLAB scripts or pure Java. The concept of the analysis server enables the system to register ASAM ODS server event notifications and combine them with measurements of a defined MIME-type and a corresponding, predefined third-party tool evaluation.

Key Benefits: The key benefit of the analysis server is that it takes away most of the repetitive analysis tasks from the engineers, thereby allowing them to spend their time on specific evaluations. In addition, analysis results, such as reports, are generated in a standardized manner and remain comparable for further evaluation within ASAM ODS. Next to being able to process large-scale Measurement data in a fully automated manner, the solution enables a convenient approach for evaluating characteristics over multiple tests.

The PLCD project at Bosch started with the Automotive Diesel Systems Department and is to be rolled out throughout the corporation. It currently covers multiple test stands for injector testing but is designed to cover other domains in the future.

The current implementation of the PLCD project runs in a productive ASAM ODS environment and is very advanced in data acquisition, data preparation and Validation; it also integrates a web-based front-end data management and work-flow GUI. Although these steps are already up and running, the focus is on improving the process of evaluating managed data.

As of now, evaluations are mostly done by downloading files from the ASAM ODS server to a common shared drive to then process it further. As a requirement for the year 2015, Bosch wants to standardize this part of the work-flow in order to provide an overall streamlined solution to their internal costumers.

The idea of integrating an analysis server with PLCD was primarily influenced by the proof-of-concept at Daimler AG, where such a system for road-load data analysis has been running well in a productive environment since 2010; that system was partly developed by HighQSoft GmbH. The objective for Bosch is a re-launch of the concept with a generic implementation that incorporates lessons learned, new requirements and features.

Saving Measurement data on a shared drive led to two major problems: first, the download of files from the ASAM ODS server onto the shared drive created redundancies and second, diversified analysis scripting led to varying implementation and non-comparable results.

These two problems created inefficiency in the following ways: 

• Repetitive tasks occupied the employees’ time and prevented them from focusing on their “actual work.” In addition, the number of analyses to be executed was rising faster than employees could track. Finally, the resulting data was not properly fed back into the ASAM ODS system; therefore, comparison of analysis results was not feasible.
• These challenges led to the decision to implement a server-side framework to integrate automated execution of evaluations. The objective is to process new and incoming ASAM ODS measurement data based on their MIME-type and pre-defined analysis programs in order to retrieve standardized and comparable result sets.

We solved the aforementioned problems by implementing an analysis server application based on the ASAM ODS Standard. The analysis server manages workflow by executing an evaluation. Its tree-like structure incorporates calls, analysis programs and parameters.The analysis program itself is a third-party script to be run by the analysis server with the third party’s tools, such as DIAdem, MATLAB or pure JAVA. 

The result of an evaluation might be multifaceted and depend on the engineers’ case-by-case requirements. An evaluation may produce simulated Measurement data rewritten into ASAM ODS, evaluation reports to be attached to instance elements or a notification sent to a user group. A web interface also allows engineers to administer analyses programs and their parameter sets for automated scheduling. In addition, they may retrieve results and are also able to trigger specific evaluations. 

An additional challenge related to the manual measurements downloads and individual analyses is that the data volume on the shared drives grew exponentially. Due to the large number of files, we have developed a tool to support the identification of redundant data on ASAM ODS and on the shared folder. The tool allows an administrator to set up searches to identify, copy, move and/or delete the files based on various comparison modes. The administrator can dispense with the data according to the company’s data guidelines. This tool is available on a web interface.

The integration of the analysis server and the setup of an automated data workflow could significantly reduce the efforts of engineers to work through redundant tasks. The solution increases the quality of result sets through standardization, thereby enabling comparability.

As evaluation structure may be defined very flexibly. Moreover, an evaluation may contain multiple subsequent analysis programs and integrate multiple measurements into, for example, fleet analysis. Therefore, the solution meets the requirement of Bosch to be reused in different domains. 

Another important fact is that the solution is designed to provide the workflow to automate data analysis; the know-how of the analysis is to be provided by the individual engineers and therefore stays within the enterprise. 

In terms of IT infrastructure, the solution permits the analysis of largescale data volumes right where they are produced. The scalability of the solution is given vertically (machine performance, CPUs and memory) as well as horizontally over multiple machines. Data traffic over networks and data redundancy on shared drives is reduced.