Abstract: In this study, a new method for evaluating the combustion performance of material monomers based on flame image recognition technology was proposed. In view of the limitations of the existing combustion test methods, such as insufficient measurement of process parameters and the influence of test scale, we used image recognition technology to measure the variation law of flame in the single burning item test (SBI), and established the mapping relationship between the combustion heating power of building materials and the area of flame luminous pixels. In the experiment, an intelligent camera system based on the Raspberry Pi and IMX477R camera module was used to process the flame image through OpenCV and Python algorithms, and the number of flame pixels was quantified. The FDS simulation software was used to simulate the flames with different combustion powers, and compared with the experimental data, the feasibility of the proposed method was verified. The results show that there is a nearly linear relationship between the flame burning power and the number of white pixels at the same viewing angle. In addition, the combustion performance level judgment based on visual analysis uses 60 kW of combustion energy as the safety threshold, which can quickly distinguish between refractory and combustible materials.

Key words: combustibility, heat release rate, flame image recognition, single burning item (SBI), pixel area