Reconstruction of Polarity Distribution in Laminar Premixed Flames Based on Electrostatic Tomography

Detecting the polarity distribution of flames is significant for understanding flame electrical properties and optimizing the combustion control method of applying an electric field. Compared with invasive methods like mass spectrometers and probes, this paper adopts non-invasive electrostatic tomography (EST) to reconstruct the polarity distribution of laminar premixed flames. The induced charge of the electrostatic sensor comes from mutual offsetting of positive and negative charged particles in flames, which makes it difficult to reconstruct the polarity distribution. To solve this problem, the one-dimensional residual network (1-D ResNet) is used to establish a nonlinear mapping between static boundary measurement values of EST and the polarity distribution. It is difficult to obtain a large amount of polarity distribution in flames and corresponding EST boundary measurement values through experiments. Thus, the training dataset for 1-D ResNet is constructed by combining combustion and electrostatic field simulations. The dynamic induced signal collected by the sensor reflects the change of induced charge over time. A method is proposed to characterize the boundary measurement value considering polarity. It converts the dynamic induced signal into the static boundary measurement value. The method not only characterizes the charge value but also the polarity of the induced charge. A belt-type electrostatic experiment is designed to validate the effectiveness of the method. The characterizing values are used as input to the 1-D ResNet trained solely on simulation data. Simulation and experimental results show that the 1-D ResNet successfully reconstructs the polarity distribution of laminar premixed flames at different heights.