Kömür Yakıcılarında Yanma Verimi Tahmini için Gömülü Platformda Çalışabilen Evrişimsel Sinir Ağının Parametre Analizi

Accurately and effectively calculating combustion efficiency in coal burners is crucial for industrial boiler manufacturers. Two main approaches can be used to calculate efficiency: 1) Analyzing the gas emitted from flue; 2) Visualizing chamber boiler. Flue analyzers, which are not user-friendly, come with high costs. Additionally, physical distance between flue causes measurement delayed. Methods based on visualizing do have these disadvantages. This study proposes a system has two contributions literature: first time, modern Convolutional Neural Networks (CNN) approach estimate efficiency; CNN architecture optimal parameters work an embedded platform. When classical classification techniques CPU-supported processor card used, calculated one flame image 1.7 seconds, while this number increases approximately 20 frames per second (34 times faster) when proposed GPU-supported used. The results obtained demonstrate superiority of hardware over estimating efficiency.