FP-AI-MONITOR1

Description

The FP-AI-MONITOR1 function pack helps to jump-start the edge AI implementation and development for sensor-monitoring-based applications designed with X-CUBE-AI or with the NanoEdge™ AI Studio . It covers the entire design of the Machine Learning cycle from the data set acquisition to the integration on a physical node. The examples provided allow the user to create, in a matter of minutes, a proof of concept for a predictive maintenance solution with anomaly detection and classification based on both vibration and ultrasound, but also on activity recognition. These examples can be fine-tuned to fit the user's dedicated use cases by retraining the models with the user's data set.

X-CUBE-AI is an STM32Cube Expansion Package part of the STM32Cube.AI ecosystem. It extends the STM32CubeMX capabilities with the automatic conversion of pretrained Neural Network or Machine Learning models and the integration of the generated optimized library into the user's project. X-CUBE-AI offers also several means to validate AI models both on desktop PC and STM32, as well as to measure performance on STM32 devices without user handmade specific C code. The support vector classifier used for human activity recognition (HAR) example is generated by X-CUBE-AI. Other applications can be created using optimized ML and DNN code generated by X-CUBE-AI.

NanoEdge™ AI Studio (NanoEdgeAIStudio) simplifies the creation of autonomous Machine Learning libraries with the possibility of running training on target and inference on the edge. For instance, condition-based monitoring applications using vibration and motion data can be created easily by recompiling the function pack with NanoEdge™ AI anomaly detection models.

FP-AI-MONITOR1 runs the learning session and the inference in real time on an STM32L4R9ZI ultra-low-power microcontroller (Arm^® Cortex^®‑M4 at 120 MHz with 2 Mbytes of flash memory and 640 Kbytes of SRAM), taking physical sensor data as input. The SensorTile wireless industrial node (STEVAL-STWINKT1B) embeds industrial-grade sensors, including very high frequency audio and ultrasound spectra detection, 6-axis IMU, 3-axis accelerometer, and vibrometer to record any inertial and vibrational data with high accuracy at high frequencies.

FP-AI-MONITOR1 implements a wired interactive CLI to configure the node, and manage the learn and detect phases. For simple operation in the field, a standalone battery-operated mode allows basic controls through the user button, without using the console.