Cost-effective and a perfect fit: how SMEs can introduce data-driven services independently
Unplanned machine downtimes, high maintenance costs and untapped efficiency potential – challenges that many companies are familiar with. The machines themselves often provide the solution: operating and sensor data can be used to predict wear, optimize maintenance cycles and avoid breakdowns. The problem is that the analysis and implementation of such data-driven services is usually complex, expensive and requires specialized knowledge – a hurdle that often deters SMEs. This is where the it’s OWL project ‘Industry 4.0 ecosystem for the automated use of data-driven services, I4.0AutoServ for short’ comes in. It creates a solution that automatically generates data-based value-added services (VBS) and makes them usable for companies – even without in-depth IT expertise.
Operating and sensor data generated in production processes often offers untapped potential. Data-based services can be used to detect faults at an early stage, plan maintenance cycles better and make production processes more efficient. Particularly interesting for SMEs: such solutions can extend the service life of machines and minimize downtime – thus reducing costs and increasing competitiveness.
Companies do not need IT expertise
The it’s OWL project ‘I4.0AutoServ’ offers a platform that makes data-based services available quickly and easily. The quality of data-based services is largely dependent on data processing, which often requires complex analyses. The platform supports companies by automatically analyzing historical operating data, selecting suitable features by means of feature engineering and determining and training a suitable algorithm for the respective use case. The individual building blocks are automatically assembled into a ready-to-use package and made available.
Feature engineering refers to the process of extracting meaningful characteristics (so-called “features”) from raw data that are relevant for analysis and machine learning (ML).
For example, the maximum value of an acceleration sensor on a machine could provide information about the wear condition of a machine and therefore be a helpful feature. Increasing signs of wear often lead to stronger vibrations – and therefore to higher maximum values. This feature therefore provides valuable information for condition monitoring.
The success of ML models depends heavily on the quality of the features. Good features make it easier for algorithms to recognize correlations and make predictions. In-depth analyses are usually required to identify suitable features. Thanks to the ‘I4.0AutoServ’ offering, this step is simplified through automation to make it easier for companies to get started with data-based analyses.
How SMEs can unlock the potential of their data
The project’s technology enables small and medium-sized companies to tap into the potential of their machine and production data. Instead of having to rely on external service providers, companies can introduce data-driven services themselves – faster, more cost-effectively and precisely tailored. The use of VBS offers companies:
- Fewer breakdowns: Early fault detection minimizes unplanned downtime.
- Error identification: Recognize known errors.
- Forecast: Estimating the remaining useful life of technical systems
Flexible and application-oriented: automated data processing
The ‘I4.0AutoServ’ project is based on the results of the it’s OWL project ‘ML4Pro²’, in which building blocks for machine learning have already been modularized. These technologies form the basis for the automated generation of VBS. Companies benefit from a solution that is both flexible and user-friendly. This means that companies can flexibly put together modular building blocks for data analysis. Companies can either create their own data processing chains or use tried and tested templates. The system then automatically starts training the models. The result is a ready-to-use data-based service (DBS) that not only covers all the necessary software requirements, but also provides the appropriate source code. This allows companies to use the technology directly in various application areas – from predictive maintenance to production optimization.
