AI on the edge | Topics
Keeping a close eye on their machine
Holistic monitoring via AI in field-level devices
As part of the digitalisation and optimisation of industrial processes, information on the current status of machines and plant, for example, is becoming increasingly important. This is because, in addition to data from regular operations, details on maintenance requirements and intervals, or any deviations, are now crucial for the planning and execution of such production processes. But how can machine and plant operators be informed of changes as and when they occur?
You can find our English-language video here.
The solution presented by SYS TEC electronic and PowerBrain. Shop uses structure-borne sound analysis for machinery and plant to demonstrate how data can be generated and processed accurately, effectively and in real time. Using a practical example in the video of the same name, vibration patterns from data recorded during rotation can be used to illustrate, amongst other things, downtime, normal operation, as well as deviations or other anomalies.
The conversion of the recorded data from the time domain to the frequency domain takes place in the sysWORXX SSM-100 analysis module using Fast Fourier Transformation (FFT). The frequency spectrum, together with the continuous root mean square (RMS) value, is transmitted to the controller, the sysWORXX CTR-700, and the data is linked to the PowerBrain. Shop AI via Node-RED. The data from the FFT analysis itself is also sent via Node-RED to the AI PowerBrains, which are integrated using so-called Docker containers. The controller thus serves as an execution platform, linking sensors and actuators with local artificial intelligence. This allows self-sufficient artificial intelligence to be conveniently ported to field devices using AI PowerBrain technology.
The captured and labelled data is communicated to the PowerBrains via an MQTT instance. Within the AI itself, the information undergoes verification and is subsequently classified accordingly. A Node-RED interface designed by SYS TEC electronic enables straightforward integration and operation of the system, for example, initiating the training of the PowerBrains and monitoring progress. This is displayed visually in the monitor window.
Conclusion: FFT data is complex sensor data that can be effectively processed using the SYS TEC electronic/PowerBrain. Shop AI solution. This approach enables direct on-site monitoring and control with low latency, as well as an optional network and cloud connection for remote monitoring and control. Operators of machinery and plant benefit from AI as well as from the monitoring and processing of machine data. Depending on the scenario, various conclusions can be drawn regarding maintenance requirements, wear and tear, or loosening of mechanical connections. Consequently, anomalies or damage can be detected at an early stage and countermeasures initiated.
The advantages of edge AI technology over cloud applications are clear: the system operates autonomously and is secure against attacks; it can be run even without internet connectivity, thereby offering the highest level of cyber security. As data sovereignty is a given, the data remains on the operator’s or company’s machine; this can lead to cost savings, for example by eliminating the costs of aggregating data from the field level to the cloud, as well as transaction costs within the cloud.
Our solution is implemented using our analysis module, our controller and the low-code Node-RED interface implemented therein, as well as a PowerBrainShop AI.