1

Solution overview

Predictive maintenance applications collect and process data from a wide variety of sensors in order to identify

potential failures in machinery before they happen. A principal requirement of such applications is that the

condition monitoring equipment is placed very close to relevant machine componentry for the data to be reliable,

which is why the STWIN.box node is designed to be small but robust, self-powered and capable of wireless

communication. Another application issue is the high volumes of preferably real-time data processing involved,

which can overwhelm centralized monitoring and control systems, and corresponding communication networks.

Distributed (or decentralized) computing architectures represent a valid solution to this problem by performing

data preprocessing and analytical operations directly on the node. The STWIN kit supports and can demonstrate

this concept through sample applications in the firmware package running on the STM32U5 ultra-low-power

microcontroller embedded on the core system board. An additional possibility is provided by the embedded

processing capabilities of the IIS2ICLX and the ISM330DHCX sensors, thanks to their programmable finite state

machine and machine learning core.

Figure 1. STEVAL-STWINBX1 functional block diagram

STEVAL-STWINBX1

Solution overview

DB4598 - Rev 2

page 3/6