Smart Systems

Fraunhofer IWU: Quality predictions in production – reliable and in real time thanks to artificial intelligence

While a machining step is still in progress, it is possible to predict with a high degree of certainty whether the component will meet the quality specifications: Artificial intelligence (AI) makes it possible. The AI solutions developed at Fraunhofer IWU represent an improvement over previous in-line inspection systems, which eliminate the need for time-consuming ejection for inspection purposes. Such quality predictions while machining is still in progress can be integrated into many industrial manufacturing processes, usually even in conjunction with existing, low-cost sensor technology. But this AI can also be used for optimization purposes. With its help, process input parameters can be controlled, for example to avoid rejects from the outset. Or to reduce energy consumption in production without compromising quality.

Share this Post
Fraunhofer IWU trade show demonstrator at Automatica: The AI recognizes, for example, when drilling is being done in wood or plastic, even in composite materials. This can be used to adjust process parameters online in an AI-controlled manner. (Photo: Fraunhofer IWU)

Contact info

Silicon Saxony

Marketing, Kommunikation und Öffentlichkeitsarbeit

Manfred-von-Ardenne-Ring 20 F

Telefon: +49 351 8925 886

Fax: +49 351 8925 889

Contact person:

Quality predictions
Typical future areas of application for the newly developed AI solution are machining steps such as drilling, turning and milling. In drilling, for example, speed, feed rate and measurements by a vibration sensor provide information about the expected qualitative result. This makes it possible, for example, to obtain a quality statement for deep-hole drilling without having to destroy the workpiece for a measurement.

Process optimization
In the area of metalworking, the use of AI has proven itself not least in hot forming. In this process, the workpiece is heated above the austenitizing temperature (approx. 880 °C) in the furnace before pressing. As soon as the desired target temperature for optimum hardening results is reached, the hot sheet is inserted into the press by a handling system and formed. A martensitic structure is formed, and the material is thus hardened. Since product quality is the primary concern in this energy-intensive process, the furnace temperature is often set higher than it should be. By predicting the foreseeable hardness, the AI provides data-based assistance for fine-tuning the hardening temperature.

In injection molding processes, the AI monitors specific parameters such as the temperature of the mold, the rotation speed of the screw conveyor for the granules, the melt temperature, the holding time of the mold and the cooling time. Timely countermeasures in the event of an unfavorable quality forecast thus help to significantly reduce scrap.

100% inspections, low number of training data sets

In all application scenarios, the AI can be used directly in the manufacturing process (inline) to monitor the entire batch (100% inspections). Sole random checks are a thing of the past.

For training different AI models, a two-digit number of data sets is sufficient in many applications, supplemented by expert knowledge of the process. In operation, the computing power of (local) edge computing is often sufficient.

Jointly with Synapticon GmbH: Fraunhofer IWU presents new architecture for human-robot collaboration (MRC)
For robots intended for MRC (cobots), particularly high requirements apply to the safety of the motion sequences: the central safety module must process data from numerous sensors, which requires many cable connections for sensors and actuators in traditional robot architectures and proprietary solutions. The new safety architecture developed by Fraunhofer IWU, NexCOBOT and Synapticon GmbH is designed in a decentralized manner; it enables safe interaction between humans and industrial robots even when work situations change dynamically – with significantly reduced cabling requirements. Only power and communication connections need to be routed to the drives. Another advantage: Since the safety of the motion sequences is monitored directly at the drive axis, significantly less reaction time is required. At Automatica, the new solution can be seen at the Synapticon GmbH booth.

Further links

Photo: Fraunhofer IWU

You may be interested in the following