Combustion Status Recognition of MSWI process Based on Flame Image by Using YOLOv5

The combustion status is one of the key controlled variables in the municipal solid waste incineration (MSWI) process. At present, the actual industrial process mainly relies on domain experts to observe the flame image. By using such manual mode, it is easy to cause the instability of the MSWI process and lead to the increase of pollution emission concentration. The complexity of the MSW composition can lead to the characteristics of the flame in the furnace, such as large brightness difference and motion blur. To solve these problems, this article proposes a method for combustion status recognition of the MSWI process based on flame image by using YOLOv5. The method uses backbone for feature extraction and uses the head layer for status recognition. The experimental result shows that the proposed method has higher accuracy than Alexnet, Vggnet, Googlenet and other algorithms.