Machine Learning refers to a class of AI methods that can infer relationships from data. These correlations can be derived from the data as predictive models or as explicitly usable knowledge from the data.

Automatic Machine Learning (AutoML) is a development, in which the learning methods are automatically selected, validated and optimized for the respective data sets and tasks. Our ClearVu Analytics product family provides these functionalities.

Due to adaptive learning methods the models created in this way are continuously checked and automatically retrained in case of necessary adjustments. Machine Learning is a powerful tool for the automatic modeling of data and can be used in many areas, from development to production and logistics processes.

The combination of artificial intelligence and optimization is ideally suited for optimizing production processes in terms of product quality, process stability, resource consumption or other criteria. For this purpose, process data are jointly analyzed and prediction models are derived using machine learning that can predict the target criterion – for example product quality – from the process data.

In combination with a nonlinear optimization method, process parameters can then be derived that enable further improvement of the target criterion. Conversely, it is also possible to identify rather disadvantageous parameter settings and then avoid them. If several target criteria are to be optimized simultaneously, this can be implemented using methods of so-called multi-objective optimization. The combination of machine learning with optimization is the key to successful process optimization.

Machine learning, AI and optimization are based on data that can come from different sources and are often incomplete or insufficiently prepared for analysis. The automated preprocessing of this data can also be supported with machine learning techniques. The combination and integration of data from different sources is another important prerequisite for the application of learning and optimization methods.

We have the appropriate tools for both, which we adapt to your requirements so that AI methods can be used in the best possible way. We make the results of machine learning available in your IT environment through interfaces so that they can be used directly in the process.

With the help of data-driven modeling using AutoML methods it is possible to derive predictive models for predictive maintenance from suitable data. By means of such models, it is possible to calculate the remaining useful life (RUL) of plant components.

This enables significantly improved planning, thus optimization of maintenance intervals and maintenance measures, and the avoidance of unplanned production downtimes.

The quality of produced components often depends on a variety of parameters, which include the process settings, material properties and other influencing variables. Due to digitization, this data is increasingly available today. By means of data-driven modeling with AutoML methods, such data can be used to create predictive models for product quality depending on these influencing parameters. From this we can derive key parameters that influence product quality and, in combination with our optimization algorithms, recommendations for optimal parameter settings that further improve product quality, scrap rates or rework efforts.