Condition Monitoring and Predictive Maintenance systems include a several smart sensor nodes in the equipment, which are directly connected to the cloud directly or connected via intermediate gateways.
Computation is performed inside the smart sensor or on the local microprocessor, either in gateways or in the cloud, depending on the expected latency and on how far the raw data and processed data are sent over connectivity.
Edge processing occurs when the computation of data is carried out directly in the smart sensor node or at the gateway, in order to save power consumption and ensure data is kept confidential, allowing companies to analyze critical information at the node level and to reduce anomaly detection time.
Combining edge and cloud computing technologies could help tpo develop predictive maintenance techniques and enhance their efficiency and effectiveness. Indeed, working on data at the edge, close to the sensor, allows companies to detect machine deterioration at the node level and to take immediate and informed corrective actions, thus preventing further damage and machine failures. Long-term analysis and actions to determine trends and optimize local analysis models can be managed on the cloud, enabling more complex analytics on large amounts of pre-processed data coming from multiple nodes.
Edge processing offers the following benefits:
- Confidentiality: data is not sent to the cloud and is locally stored on the device or the equipment
- Cost reduction: latency and throughput of high-volume time-series asset data is significantly optimized. Reducing the amount of useless machine data sent and stored in the cloud leads to significant benefits, as it enables real-time distributed applications and eliminates the need for complex systems
- Low latency: minimal delay in the repair of equipment is essential for assets which are mission-critical.
To easily implement Predictive Maintenance algorithms on MCU and MPU edge devices, STM32 Tools were enriched with the STM32Cube. AI ecosystem, extending STM32CubeMX software capabilities with the automatic conversion of pre-trained Neural Networks. STM32Cube. AI supportsing several deep learning frameworks and includes optimized libraries which can be used to implement AI algorithms in embedded applications level.
STMicroelectronics also provides a complete offer of microprocessors and optimized Power Management solutions for industrial gateways as key enablers of aggregated processing in the field. Software development kits and cloud applications for Edge and Cloud processing (starter kits) as well as AI studio tools, like STM32Cube. AI software and code examples, are included our function packs and solutions.