A Noise Reduction and Bubble Identification Algorithm for Conductivity Probe Signal Processing in Steam-Water Two-Phase Flow Based on VMD-WTD, SNIP, and SEVA

The conductivity probe serves as an effective instrument for measuring two-phase flow parameters; however, its application in high-temperature and high-pressure environments faces constraints due to structure and signal processing algorithm. Under such conditions, the probe signal is vulnerable to noise, baseline drift and waveform distortion. Conventional algorithms, predominantly designed for standard temperature and pressure conditions, demonstrate limited efficacy in addressing these complex challenges. Therefore, a comprehensive signal processing framework was proposed. First, variational mode decomposition (VMD) was integrated with wavelet threshold denoising (WTD) to effectively suppress noise. Subsequently, the sensitive nonlinear iterative peak (SNIP) algorithm was employed for accurate baseline estimation. Finally, the slope and extreme value analysis (SEVA) algorithm was utilized for precise identification of bubble signals. Experimental validation was conducted under pressures ranging from 0.8 MPa to 4.1 MPa. The results demonstrate that the proposed algorithm effectively processes conductivity probe signals under such conditions. The conductivity probe achieved an expanded uncertainty of 3.88% (coverage factor k = 2), corresponding to a 95% confidence level. Experimental void fraction results were compared with predictions from the classical drift-flux models, demonstrating relative deviations within 30%. Additionally, a comparison with image processing measurements revealed a smaller margin of relative deviation, limited to 12%.