Duration: Completed | 01/2021 – 12/2021
Funded by: EIT Manufacturing
If you have any questions about this completed project please contact our institute management:
Oberingenieure@PTW.TU-Darmstadt.de
Initial situation
In current production systems, large amounts of data are generated and collected on a daily basis. Much of the historical data can contain useful information that can be discovered through ML. How to use this data for production, especially for shop floor management, is the question of this research project. Currently, both data science and artificial intelligence (AI) offer ways to unlock the potential value of data. However, many companies lack the appropriate knowledge to use this data to solve practical problems. In addition, the quality of the data itself presents some difficulties in implementing ML.
Objective
Based on the current situation, the RaisQ project aims to use data from production, testing and rework processes for analysis and prediction. This can help provide decision support to people in the production process and thus reduce rework and defect costs for manufacturing companies. First, SIA must be able to automate the process of data cleansing and data preparation for different types of data in different formats. Then, SIA must use AI algorithms to make recommendations for the rework process based on the available data. Considering that data quality has a great impact on results, Natural Language Processing (NLP) also plays an important role in SIA to process text data. With better data quality, more accurate recommendations can be made to better support the actual production.
Approach
At the beginning of the project, the use cases are defined together with the partners considering the current data and the actual needs. Based on the designed prototype of SIA, the specific framework for each use case will be standardized for practical implementation. In the implementation phase of this project, SIA will be tested as an important part of SFM Systems' Digital Teamboard in the real production line of MAN Truck & Bus. Based on the results of the tests, the performance of SIA will be improved and more scenarios will be developed according to the demand. Different NLP methods will also be developed for text data cleansing. After evaluating the performance of SIA, SFM Systems will develop SIA from a prototype to a product and eventually offer it as a cloud software-as-a-service (SaaS).
Acknowledgements
This project is funded by EIT Manufacturing. We thank for the opportunity to work on this project.
Funding source