Periodic Reporting for period 1 - CTDM (CLIMB THE DATA MOUNTAIN)
Reporting period: 2020-11-01 to 2021-10-31
The factory of the future will have only two employees – a man and a dog. This quote from the MIT researcher Warren Bennis indicates. While much is being shown in the literature and in marketing in the area of digitization and intelligent data use, very little of this is being used in SMEs in Europe. The software provider DatenBerg offers software for this purpose to support manufacturing companies in their digital transformation. With the software, subjectivity can be removed from the production process and people can be supported in their decision-making. This enables manufacturing companies to maintain their sites in Europe in a tough global market environment with large margin pressures and generate wealth through work for society. The goal of the CTDM project is to solve complex use cases in production with the help of data analysis. For this purpose, two prototypes are to be created, which solve specific problems for a manufacturing company. In addition to the implementation of concrete applications, further possibilities for the application of data-based approaches are to be evaluated and suitable technical implementations selected. Project partner DatenBerg - itself a small company with limited resources - benefits from the project with support from an Innovation Associate with deep knowledge of data analytics. The funding enables the SME to implement the application fields and offer added value to its customers, which otherwise would not have been possible in the short time due to lack of capacity. This creates an effect of scale and other SMEs in the European market benefit indirectly from the funding. The project brought the SME DatenBerg a big step further in technical development and benefited from the Associate in various levels. In addition to the technical support with a deep understanding of methods, the internationality could also enrich the work in the company.
The first project supported by the innovation associate focused on remaining useful live of electrodes in welding. The aim is to estimate wear and predict failures before they happen. We were able to complete a proof-of-concept using customer data from screw manufacturing. In the future we plan to integrate these results in a traffic light system for easy customer application.
The innovation associate also enabled us to improve these and other prediction models using explainable AI approaches. Using these approaches, we are now able to improve prediction performance and validate recommendations.
By working with the innovation associate, we achieved significant improvements in our DeepDive technology. Starting as a simple statistics toolbox, it now uses machine learning to automatically identify cause and effect relationships in a process and aid in the finding of root causes for quality issues. These newly developed algorithms have been integrated into the software and are being tested by pilot customers in the rubber compounding industry.
A combination of the explainable AI approaches and modelling was used for project autoHINT, in which we developed a proof of concept for automated process adjustment recommendations in manufacturing. The associate was able to demonstrate significantly improved stability and quality in a simulated production process.
The DeepDive technology and autoHINT will be further developed towards a higher TRL in the near future and a commercial exploitation is planned for the end of 2022. Next to the technical work the associate also took part in different trainings.
The innovation associate also enabled us to improve these and other prediction models using explainable AI approaches. Using these approaches, we are now able to improve prediction performance and validate recommendations.
By working with the innovation associate, we achieved significant improvements in our DeepDive technology. Starting as a simple statistics toolbox, it now uses machine learning to automatically identify cause and effect relationships in a process and aid in the finding of root causes for quality issues. These newly developed algorithms have been integrated into the software and are being tested by pilot customers in the rubber compounding industry.
A combination of the explainable AI approaches and modelling was used for project autoHINT, in which we developed a proof of concept for automated process adjustment recommendations in manufacturing. The associate was able to demonstrate significantly improved stability and quality in a simulated production process.
The DeepDive technology and autoHINT will be further developed towards a higher TRL in the near future and a commercial exploitation is planned for the end of 2022. Next to the technical work the associate also took part in different trainings.
Process understanding and control are some of the biggest issues manufacturers face today. Especially small companies struggle in the competition for innovation and personnel. Using the technologies developed with support from the associate, we were able to demonstrate significant advances in this area.
Machine learning assistance in finding root causes for failures helps users to navigate the increasing amount of data in their process. By making these tools available for all companies and people without a background in data science, they enable wider participation in industry 4.0 technology.
Automated process recommendations for both quality and maintenance application can be used to compensate for a loss of experience and knowledge as qualified personnel retires and businesses are looking to improve their manufacturing technology. This enables small companies to stay competitive with a limited budget for research and innovation.
Several of these developments have reached the pilot stage and are being actively tested by customers across a wide range of industries.
Machine learning assistance in finding root causes for failures helps users to navigate the increasing amount of data in their process. By making these tools available for all companies and people without a background in data science, they enable wider participation in industry 4.0 technology.
Automated process recommendations for both quality and maintenance application can be used to compensate for a loss of experience and knowledge as qualified personnel retires and businesses are looking to improve their manufacturing technology. This enables small companies to stay competitive with a limited budget for research and innovation.
Several of these developments have reached the pilot stage and are being actively tested by customers across a wide range of industries.