Revolutionizing wood-based panel production

We develop AI processes for industry and bring them into long-term productive operation. To do this, we do not simply use "ready-made solutions from a kit", but have a fundamental understanding of the topic.

Strengthen your competitiveness by using artificial intelligence in industry. AI promises greater efficiency, shorter start-up times, higher quality and much more. But can AI achieve this?

Our experience shows that it can: Yes, if you do it right.

Here are some examples of applications from our portfolio. We have developed specially adapted methods and software solutions for our customers.

Quality prediction

By analyzing historical production data, models can be trained to identify potential quality problems at an early stage. These predictions enable manufacturers to take proactive measures to reduce rejects. Another application is the targeted parameterization of systems to minimize resource consumption for a defined quality target.

Anomaly detection

Durch die Auswertung von Anlagendaten werden Modelle geschaffen, die den Betriebszustand einer Anlage erkennen. Abweichungen von diesem Zustand, die auf potenzielle Defekte oder Probleme hinweisen könnten, werden als Anomalien erkannt. Dies ermöglicht einen frühzeitigen Eingriff, um Ausfälle zu verhindern. Maschinelles Lernen kann dabei komplexe Muster erkennen, die von traditionellen statistischen Methoden übersehen werden könnten.

Root Cause Analysis

Faults and anomalies in industrial systems often propagate and result in a so-called alarm shower, i.e. several simultaneous alarms at different points. This makes it difficult for operators to identify the cause of the fault and address it in a targeted manner. We use techniques that combine causal models and machine learning to localize the causes of faults in the patterns of a complex plant condition.

Parameter optimization and process control

In addition to data analysis and forecasting, AI models can also be used for active optimization interventions. Parameter optimization makes suggestions for process settings based on the currently prevailing conditions, recipe properties, and so on. Human-in-the-loop approaches are often used here, where the operators can still modify the suggestions. Process control goes one step further and intervenes in the process in real time.

Runtime monitoring and customization

Systems in productive use change over time. Causes include fluctuating input materials, wear and maintenance, structural modifications and so on. To ensure the long-term operation of an AI solution, it must be able to adapt to relevant changes in the systems and processes. We are talking about drifts here. Our "AutoDrift" software package enables the automatic detection and adaptation of AI models to such drifts. This ensures long-term operation.

© Dieffenbacher GmbH

Together with the company Dieffenbacher GmbH, we have developed AI-based software solutions that exploit a large savings potential in wood-based panel production. The manufacturing process of wood-based panels consists of a large number of processing steps, from the chipping of the logs and the gluing of the chips to the compaction and curing of the chip mixture in the continuous press. The highly specialized and complex machines are manufactured by Dieffenbacher.

To ensure the quality of the plates, the plant is continuously monitored and controlled. With many hundreds of sensor signals, the complexity of the entire process is overwhelming, especially due to the multiple interactions between the production steps. Variations in production conditions and material properties present an additional challenge and require the expertise of an experienced plant operator to ensure optimal production.

Samples are measured for quality in the laboratory about one to three times a day. The results are only available after several hours, sometimes even days. Unfortunately, this delay can lead to the production of rejects if it is found that the panels produced do not meet the quality standards.

This is where the AI solutions integrated in the Diffenbacher EVORIS platform come in.

1. Anomaly detection

The AI-powered application analyses the complex sensor system and detects deviations from the norm during operation. Whether it's a subtle shift or a significant irregularity, the system's real-time notifications enable operators to quickly take corrective action and prevent unplanned production downtime.

2. Live Quality forecast

The AI application enables a continuous forecast of the quality of the panels just produced. This information helps operators to compensate for quality fluctuations and gives them more certainty when adjusting the process, e.g. with regard to the input materials required for the desired quality. In this way, resources can be saved without compromising quality.

The developed AI applications prioritize production efficiency, optimize production processes and save resources. By integrating AI directly into Dieffenbacher's EVORIS platform, the production of wood-based panels not only becomes smarter, but also more robust in the face of changing input materials or human operating errors.

In our collaboration with Dieffenbacher, we demonstrate the fusion of innovation and tradition, where AI applications empower plant operators to produce high quality wood-based panels with unprecedented efficiency.

Privacy warning

With the click on the play button an external video from is loaded and started. Your data is possible transferred and stored to third party. Do not start the video if you disagree. Find more about the youtube privacy statement under the following link:

Interview with Dr. Julius Pfrommer and Dr. Constanze Hasterok

More information


Department Cognitive Industrial Systems

You want to learn more about our projects in the area of Cognitive Industrial Systems? Then visit the page of our department.