Increasing Resolution of Wire-Mesh Sensor Data using Statistical Reconstruction Approach
F. A. Dias, D. R. Pipa, M. J. da Silva
1 Department of Electrical Engineering (CPGEI), Universidade Tecnológica Federal do Paraná, Brazil
*Corresponding author: email@example.com
Wire-Mesh sensor (WMS) is a powerful device to measure multiphase flows allowing the imaging and parameter extraction of the flow with high temporal and spatial resolution. Its architecture is based on intrusive electrodes, which may be undesirable in many processes. This paper presents a technique based on image reconstruction with a statistical view of regularization to improve void fraction measurement and image quality from a sensor with less than optimal electrode number. Thus, reducing the intrusive effects on the process. The following methodology was used to achieve such goal: first, WMS models with 8x8, 6x6 and 4x4 spatial resolution were simulated by Finite Element Method (FEM) and a sensitivity function was modeled for each one. Next, a linear model was used to solve the direct problem. Then, the maximum a posteriori technique was applied to recover each one of the models to 16x16 spatial resolution. Statistical parameters as covariance and mean were obtained from practical experiments with real data of 16x16 WMS. After, void fraction error (VFE) and Mean Squared Error (MSE) were used to evaluate all experiments. Simulation results show that this approach is promising and can improve significantly WMS measurements with less than optimal electrode number.
Keywords Wire-Mesh Sensor, Multiphase Flow, Inverse Problem, Statistical Image Reconstruction, Maximum a posteriori (MAP).
Copyright © International Society for Industrial Process Tomography, 2016. All rights reserved.