On application of Parzen-Rosenblatt estimation for burning torcH studies in infrared band

Abstract

Nowadays, the fuel combustion is widely used in industry. In this context combustion processes investigation and combustion control automated systems development are the actual problems. Recently, the non-contact diagnostic methods have become widespread including the optical control in the infrared band. Analysis of the known methods of the flame scanning data processing was performed, and the result of analysis showed that known methods are useful exclusively in scientific purposes, but not under the industrial conditions. To develop the information processing methodology some experiments were carried out consisting in shooting a burning torch with a thermal imaging camera in the waves band of 1.5-5.1 µm. During data processing, approximation of the pixels distribution over the ranges of relative temperatures was performed by the Parzen-Rosenblatt method. It was shown that calculated dependency has the same pattern on the all processed ranges. To find probability density and accumulated distribution function of abscissas of the distribution extremums, the Parzen-Rosenblatt method was used. It was proved that abscissas of the distribution extremums are stable and quasi continuous parameters, which do not depend on the time factors. Taking into account that the found parametrs can be determined in the automatic mode, it makes it possible to construct a measuring system using the infrared camera that can be placed inside the power or thermomechanical equipment and by this way construct the control and operation system of the torch combustion. © 2020 American Institute of Physics Inc.. All rights reserved.