First, there is plenty effective ways to analyze information from assets by simple acceleration or velocity RMS level of vibration. Second simple parameter is also temperature.
After this you can have still more sophisticated filtering, envelope filters as example and then also the FFT conversion discussed.
There is indeed new ways of using AI algorithms and for example anomality detection is emerging.
ST’s vision on AI is to embed the right level of intelligence in the different stages of the data path. As example inside ISM330DHCX sensor, included in STWIN board, there is Machine Learning Core inside the sensor that enables extremely power efficient 1st level of processing for system wake-up as example, we are discussing uAmps here.
For additional level such as anomaly classification, ST offers a tool that converts Neural Networks into optimized C code running on Microcontrollers with STM32Cube.AI. More information on this can be found in st.com/stm32cubeai.
ST is also working with different partners in this field, and we will soon have Press Release announcing a Function Pack demonstrating anomaly detection on STM32.
