Research on calibration method of multispectral radiation temperature measurement system based on fully connected network

Explosive fireballs generate high temperature, high pressure, and strong shock waves during the combustion process, and accurate measurement of their temperature is crucial in engineering applications. Multi spectral radiation temperature measurement technology, as an efficient non-contact detection method, has significant advantages. This study conducted temperature calibration experiments on a multispectral temperature measurement system using a 25 channel multispectral camera and a standard medium temperature blackbody furnace, and processed the calibrated grayscale images using a fully connected neural network. Generate a high-precision temperature distribution map by inputting grayscale images at 25 wavelengths into the network. This method not only improves the accuracy of temperature measurement, but also lays the experimental foundation for subsequent measurement of explosion temperature field. The research results provide key technical support for temperature measurement in engineering applications and are expected to play an important role in related fields.