Estimation of Combustion Temperature Field From the Electrical Admittivity Distribution Obtained by Electrical Tomography

Abstract

Measurement of combustion temperature field is essential for combustion diagnosis. As intrusive devices for temperature measurement interfere the flow field of flame, nonintrusive thermometry techniques are highly desired. Electrical tomography, which is a noncontact technique and can reconstruct the complex admittivity distribution in the flame, is promising for combustion diagnosis. However, the admittivity distribution needs to be mapped to temperature field in practical applications. In this article, an ion current probe and a thermocouple measured the electrical admittivity and temperature in the flame, served as calibration devices for establishing the mapping model from the reconstructed admittivity to temperature. Polynomial, exponential, and eXtreme Gradient Boosting (XGBoost) regression modeling methods were used to fit the mapping model. Accuracies of training and test data sets of the XGBoost mapping model were the highest, and their mean relative error were only 0.69% and 16.69%, respectively. An alcohol lamp flame was monitored by using the dual-modality electrical tomography. Reconstructed images of conductivity and permittivity were mapped to temperature distributions with the established mapping model. The estimated highest temperature was ring-shaped distributed, and it was verified by the longitudinal image of temperature reconstructed from a flame photo of 432-nm radiation. The proposed method was also used to monitor the temperature variations during the ignition and blowout processes of the alcohol lamp flame. The experimental results showed that the proposed method was effective to capture the evolution of temperature distributions in highly dynamic combustion field.