Temperature-field reconstruction algorithm based on reflected sigmoidal radial basis function and QR decomposition

Abstract

Information on temperature-field distribution can reflect the combustion state within utility boiler furnaces, which is conducive for the formulation of control strategies to ensure safe boiler operation; therefore, it is significant for the realization of automatic boiler control. Ultrasonic thermometry is a noncontact acoustic pyrometry technique that utilizes the time-of-flight of ultrasonic waves on limited wave propagation paths to continuously reconstruct distributed temperature fields, and thus overcomes the shortcomings of traditional physical temperature measurement methods. To address the issue of low reconstruction accuracy of the temperature-field edges caused by traditional acoustic pyrometry algorithms and realize high-accuracy reconstruction of two-dimensional temperature fields, this study proposes a temperature-field reconstruction algorithm based on a combination of the reflected sigmoidal radial basis function and QR decomposition (referred to as the RSQR algorithm). The results of simulation experiments indicate that the accuracy of the reconstruction results obtained using the proposed algorithm increase by 2.03% on an average compared to the commonly used temperature-field reconstruction algorithms, with an increase of 1.88% in the reconstruction accuracy of the temperature-field edges.