"This site requires JavaScript to work correctly"


Simulation of Metal Anomaly Research Detection

I hope there is no metal in my food!

28.10.2021 | Faculty NuW

On 16 August, the Bavarian Ministry of Economic Affairs, Regional Development and Energy approved the SMARD project - "Simulation of Metal Anomaly Research Detection". It will be funded with a good 0.85 million euros. Of this, 447,000 euros will go to the Deggendorf Institute of Technology (DIT). Together with the project partner Mesutronic Gerätebau GmbH, DIT will research an innovative further development in the field of metal detection and integrate it into existing series solutions.

Existing detection concepts for anomalies caused by metal contamination in the industrial manufacturing sector must be evaluated and their concrete implementations tested in detail. There are solutions for evaluating changes in the electromagnetic field caused by metals. However, these solutions are only partially suitable in test environments for anomaly tests with hard real-time requirements, as is the case in manufacturing systems. Existing approaches with threshold-based algorithms are only conditionally suitable for use in the validation of future detection systems for product anomalies due to rigid decision criteria. Other anomaly testing approaches, such as image processing or artificial intelligence algorithms, are not yet established in the field of metal detection. However, the integration of such advanced methods in anomaly environments provides significant added value for safeguarding future systems. Lack of research activity in this context testifies to the relevance of the project.

The aim of the project is to research and develop an AI-based evaluation method for metal detectors in the industrial manufacturing sector. This detection system, developed on the basis of an electromagnetic FEM process and an AI-controlled software analysis, enables improved detection of metallic contaminants in a manufacturing product with simultaneous reliable non-detection of non-contaminated products. The AI-based learning is performed using the electromagnetic simulation results of the FEM software Ansys (EM) Maxwell. In addition, the thermodynamic influences on the overall system consisting of the detection unit and the test product are investigated, analysed and included in the AI algorithm.

Bild (DIT): Front right Prof. Andreas Grzemba, front centre DIT project manager Prof. Frank Denk, front left project manager Mesutronic authorised signatory Manfred Artinger, back left development engineer Mesutronic Stefan Wittenzellner, back right research assistant DIT Tobias Hofbauer.