Data-driven process modelsIntelligent process monitoring and optimization

Rule-based algorithms are used in the classical statistical evaluation of process data to determine key figures, temporal trends and for regression analysis. When processing large amounts of data, however, these algorithms lead to high or no longer acceptable evaluation times. Due to the number of parameters or influencing variables to be considered, no useful model for the prediction or optimization of process parameters can be found.

Where classical statistical evaluations reach their limits, IconPro implements machine learning algorithms for the automatic evaluation of process capabilities as well as for trend analysis and derivation of correlations between process and quality parameters. Through data-driven modeling, customers receive solutions in a comparatively short time that are more robust and capable in comparison to the state of the art or that would not be possible at all according to the state of the art. In addition, the combination of process mining and data mining can also be used to derive and analyze process flows and optimize them in terms of productivity and quality.

Select your field of application

Your advantages

Fewer preconfigurations
Higher evaluation accuracy
Shorter evaluation times

Machine Learning for determining process capability

As quality or production manager, take advantage of the benefits of artificial intelligence when evaluating your process and machine capabilities. Machine learning methods are particularly well-suited for the fully automated and fast determination of cp and cpk values. This requires fewer manual preconfigurations and specifications than with conventional evaluation software.

The assignment of a distribution model to existing process data, e.g. normal distribution or Weibull distribution, which is important for capability determination, is performed automatically without the need for specifying additional presumptions. In addition, our solution offers higher evaluation accuracy. This allows you to benefit from process capability values that reflect your processes more realistically. The required computing time is also significantly reduced, which gives you advantages especially with large data sets.

Best practices

Quality assurance in production

Inspection of form and position tolerances of manufactured workpieces

  • Process capabilities
  • Live visualization

Food production

Monitored compliance with process parameter specifications in food production

  • Process capabilities
  • Sample analysis

Plant operation

Monitored compliance with operating parameter specifications of process plants

  • Process capabilities
  • Trend detection

Your advantages

More transparent processes
Reduced scrap
Shortened throughput times

Production process mining

With the help of Process Mining, real process sequences can be derived from event data, ongoing processes in the process landscape can be visually monitored and deviations from specifications can be detected. In addition, bottlenecks can be detected and eliminated in the next process planning. By combining process and data mining methods, processes and characteristic values such as scrap rates or lead times can be predicted.

Recommendations for process paths to optimize quality and productivity can be derived from this. The event data records required for this analysis are typically extracted from MES (Manufacturing Execution Systems) or ERP (Enterprise Resource Planning) systems. The data records, also known as event logs, must contain at least an activity name, a time stamp and a process number.

Best practices

Semiconductor technology equipment

Reduced throughput time of functional tests with large variety of variants

  • Process modelling
  • Bottleneck analysis

Tool construction

Minimization of scrap in production lines with a large variety of variants

  • Process modelling
  • Quality prediction

Manufacturing of turbochargers

In-line minimization of scrap during series production of rotary parts

  • Process modelling
  • Bottleneck analysis

Our services

Checking data quality

After you have described your application and objectives to us, we check the quality of the data provided with regard to feasibility.

Application implementation

A positive proof-of-concept is followed by the implementation of the application, whose operation and interfaces are tailored to your technical constraints.

Preprocessing of data

The type of data format or database is irrelevant to us. We process all types of data structures and are completely oriented towards you.


We provide the final software as an executable file for the environment you specify and test it extensively beforehand.

Machine Learning Evaluation

After the rough selection of the procedure depending on the application, we evaluate your data using the most modern machine learning algorithms.

Privacy Policy

The confidentiality of your data is a natural priority for us. The protection of sensitive information is an integral part of our service.

Have we sparked your interest?

Get in touch!

Arrange a video call with us or let us talk personally about your use case.