DH electronics GmbH
scroll Scroll Down down

3D printing: Anomaly detection with STM32MP15-based IoT Gateway and AWS Greengrass

Experts from DH electronics, Klika Tech, STMicroelectronics and AWS  developed a solution that detects equipment-level anomalies by finding actionable insights in collected data. The solution is based on the AWS Greengras Qualified Industrial IoT Gateway by DH electronics which uses a STM32MP15-based System on Module to pre-process machine-level data at the edge. The IIoT Anomaly Development solution accelerator runs tinyML and Amazon SageMaker NEO at the edge to find abnormal behavior and ensure ongoing ML model optimization. These advanced technologies allow customers to eliminate system inefficiencies, increase asset uptime and utilization, and extend equipment lifecycles. The IIoT Gateway is designed for industrial environment: the robust housing can be mounted on a DIN rail, operating temperature is 0 to +50 °C, Dual Ethernet enables highest security, long-term availability of 10+ years is guaranteed and Software will be maintained.

The 3D printer is equipped with two acceleration sensors whose motion data is routed to two ST Nucleo boards via Wireless Bluetooth and Wireless LoRa. The NUCLEO-WB55RG is connected wirelessly to the STM32MP15-based IIoT Gateway by DH electronics with certified Greengrass 2.0 build. The machine learning algorithm for anomaly detection of NUCLEO-WB55RG is performed on the IIoT Gateway. Complementary, the machine learning algorithm to detect anomalies in the motion data from the LoRa sensor is performed on NUCLEO-WL55JC1 to reduce communication bandwidth. NUCLEO-WL55JC1 is connected to the LoRa base station and the IoT Core for LoRaWAN via the IIoT Gateway. A graphical display to show if an anomaly is detected is done on AWS cloud. The machine learning algorithm was built with the same setup in data collect mode using AWS Sagemaker, Sagemaker neo and Cube.AI.