Practical implementation and validation of instantaneous angular speed estimation on resource-constrained embedded systems

In this study, an instantaneous angular speed (IAS) estimation algorithm suitable for low-power embedded systems is demonstrated and evaluated using a gas foil bearing as a challenging test case. While many algorithms for IAS estimation exist, none have been validated as suitable for use in low-power embedded systems to date. Instead, their evaluation is typically done with high-performance computers and laboratory-grade sensors. In this work, the ViBES algorithm is demonstrated running online in a real embedded system comprising a Cortex-M33-based microcontroller and a MEMS accelerometer. The system is part of a prototype fluid bearing that determines operating parameters during operation, thereby providing a dynamic and challenging test case. This system is validated on a test rig with reference measurements, continuously estimating the IAS in real-time during the operation of the mechanical system. The estimation achieves a median accuracy of 0.18 Hz, validating that the algorithm can function within the limitations of low-power hardware. The implementation requires only 65 kB of RAM and 3.0 ms processing time for a window of 4096 samples, confirming the hardware requirements determined by the authors in earlier work. These hardware requirements also indicate that most modern low-power microcontrollers and MEMS accelerometers should be sufficient for the presented algorithm. • Practical demonstration of angular speed estimation on low-power embedded system • Real-time estimation is achieved with 0.18 Hz median accuracy • Computational complexity, memory and sensor requirements are analyzed and benchmarked • Method is compatible with most low-power microcontrollers and sensors