Yazzoom in SmartFactory
Yazzoom
Yazzoom provides software and services for AI services in industry, both in the form of custom software and with the software product Yanomaly - an analytics environment for creating, training, applying and analysing AI models on machine data. Yanomaly is a plug-in software on top of existing data collection platforms and extends their functionality with:
- Signal validation and detection of sensor drift
- Predictive analytics
- Univariate and multivariate anomaly detection
- Models for monitoring specific assets such as VSDs, motors, pumps, valves and control loops
- Centre lining on the basis of golden runs
- Diagnostic analytics
Yazzoom's role in SmartFactory
With Yanomaly, Yazzoom provides monitoring software for various aspects of the SmartFactory demonstrator's operation. In the three modules, different components are monitored with AI-based models and problems are detected and immediately sent to Flagstone's HMI. The user can do a further drilldown in the Yanomaly browser-based interface.
The appropriate Yanomaly detector is used in each case to monitor the operation of various components:
- Multivariate anomaly detection to monitor the relationship between position, velocity and force of the Z-spindle in the storage module
- Monitoring of binary control signals for the operation of the conveyor belt in the insertion module
- Monitoring of the torque of the screw unit as a function of the part being inserted into the screw module
Advantages of Yazzoom in SmartFactory
Yanomaly is applied to all kinds of signals, both analogue and digital, from production machines, drawing attention to abnormal behaviour. Particularly in complex machines whose operation depends on many factors, Yanomaly helps to detect problems relating to productivity, quality and speed more quickly, as well as determining the root cause of the problem. This saves on manpower and increases the machine's OEE. Furthermore, using data-driven AI models means that there is no need for extensive configuration and maintenance of various expert rules - the models learn what normal operation means on their own, based on example data.