ueCloud and internet access are not required for Weidmueller’s new edgeML machine learning solution. Learning algorithms can run directly on the edge. edgeML is manufacturer-independent as a Docker container and can be used on all common industrial controllers, provided they support Docker containers. The solution is particularly suitable for automation engineers. Thanks to Weidmueller’s no-code approach, users can generate ML models without any knowledge of Python or data science.
Capturing, storing and processing data directly on the shop floor eliminates the need to transfer data to the cloud. This means that sensitive data does not have to leave the company. Detecting discrepancies in the production process directly at the machine, for example, speeds up troubleshooting, prevents long downtimes and reduces rejects.
With edgeML, the customer also saves on cloud licenses and fees for data transfer and storage. Even production lines whose machines and systems cannot be connected to the internet for security reasons can be optimized with edgeML using machine learning.
Weidmueller’s ModelBuilder allows automation engineers or other experts to create ML models based on collected data to be analyzed in edgeML. The software supports the standard ONNX format. This means that users can also use existing ONNX models or create their own in Python.
To save time and resources when creating ML solutions, Weidmüller will in future also enable the calibration of created models. This function is already available in ModelRuntime. A standard model for a machine family becomes a template that can be applied to other machines of the same class. The model continues to learn from these in order to adapt to the respective system. This enables the scalable reuse of ML models. edgeML also supports the Weidmueller tool MLOps for managing the life cycle of models. (Image press photo Weidmueller)