Authors: P. Pietrzak, B. Pekoslawski, M. Makowski, A. Napieralski
Affilation: Technical University of Lodz, Poland
Pages: 182 - 185
Keywords: MEMS accelerometer, adaptive subband filtering, NLMS algorithm, vibration measurement, hardware filter
The main factor that determines the cost of vibration-based condition monitoring system of large rotating machines is the unitary price of the piezoelectric sensors, commonly used in such devices. Silicon microaccelerometers can be considered as an alternative to these sensors. Unfortunately, relatively high noise floor of commercially available MEMS sensors limits the possibility of their usage in condition monitoring systems. The solution of this problem is the method of signal filtering described in the paper. It is based on adaptive subband filtering employing Adaptive Line Enhancer. For filter weights adaptation two novel algorithms have been developed. They are based on the NLMS algorithm. Both of them significantly simplify its software and hardware implementation and accelerate the adaptation process. The paper describes the software (Matlab) and hardware (FPGA) implementation of the proposed noise filter. The results of the performed tests are also reported. They confirm high efficiency of the solution. In summary, the developed method of the noise level reduction makes it possible to apply MEMS accelerometers in condition monitoring systems that utilize harmonic analysis of the vibration signal.