Power Plant Furnace Flame Stability Detection Based on ResNet18-SimAM

Assessing the stability of flame combustion within the furnace is crucial for ensuring the safe operation and efficient combustion of power plants, as well as for preventing accidents and avoiding energy wastage. This paper presents an algorithm for effectively and accurately detecting the stability of flame combustion inside a furnace. The algorithm is based on Residual network 18 (ResNet18), with the incorporation of SimAM attention mechanism at the model's bottom layers, allowing the model to focus more on crucial areas and reduce attention to redundant information. Experimental results demonstrate that this model, in comparison to ResNet18, not only accelerates convergence but also enhances the model's accuracy, achieving accuracy rates of 97.54% and 94.01 % on the training and test sets, respectively.