A signal analysis and identification scheme for an online multiphase micron-sized particle analyzer system

Abstract

Online microparticle detection is of utmost importance for industrial production. This paper proposes a signal processing and feature identification strategy to achieve particle size statistics for online measurement in a three-phase stirred tank reactor based on the electrical sensing zone (ESZ) method. Signal denoising and de-interference are achieved using the wavelet soft threshold method combined with mathematical morphological filtering. Pulse selection is implemented using pulse width limiting conditions. The key features that distinguish the pulse waveforms are defined based on the differences in the motion characteristics of the different types of particles through the aperture. Finally, the unsupervised classification algorithm balanced iterative reducing and clustering using hierarchies clustering is employed to distinguish the pulsed features between hard particles and bubbles. The results show that the particle size distribution identified by this strategy agrees with offline measurements indicating the effectiveness of the scheme. The effects of electromagnetic noise and the interference of small bubbles that approximate the particle size in solution in the online, in-situ measurement task are solved. This study scheme has a guiding and facilitating role in applying the ESZ principle to the industrial online measurement environment.