Anomaly Detection – Smart Factory Glossary

Anomaly detection uses artificial intelligence (AI) to identify deviations from a predefined standard. It contributes to better quality, fewer rejects, and greater efficiency in manufacturing companies. 

This is how anomaly detection works

Machine learning, for example, is ideal for detecting anomalies in the production process at an early stage. This AI technology comprises various forms of self-learning, whereby software recognizes regularities and irregularities in data, draws conclusions, and derives actions. The system must first be trained to detect anomalies reliably. To begin with, all data recorded for a product and every decision based on the data must be imported into the system. Now, the first dependencies are stored in the system.

If in a second step new data is added, the software then accesses data records and compares the new data with the one stored in the system. Everything that is unknown is considered a deviation. At this point, the user enters the arena: They classify the new, detected anomalies. And thus, the software is constantly improving as part of this continuous learning process.

Examples

• Computer Vision or Machine Vision combines machine learning with cameras. The system uses images or videos to recognize objects and can classify them based on various characteristics. High-resolution images make it possible to detect even the smallest changes that are invisible to the human eye. If every intermediate or end product is checked in production, large data volumes are generated on a daily basis. The analysis and correlation of this data reveals influencing factors affecting production and makes it possible to take countermeasures at an early stage
• Advanced Analytics is also based on AI methods. A good example is Predictive Quality, where applications predict the quality of a product based on process values recorded during the production process. Predictive Maintenance is used to analyze and optimize maintenance processes.

Manufacturing Execution System HYDRA X by MPDV meets the requirements of an innovative IT tool and uses technologies from the field of artificial intelligence to control production processes more effectively and plan ahead.