Digital image flame detector using high-speed computer and digital image processing technology is the new technology of furnace safeguard supervisory and protection developing in recent years. In this dissertation, furnace flame image detecting system has been studied and designed on DSP platform and a method of judging furnace outfire was presented. For the characteristics of furnace flame image, effective flame image character region was determined and three feature vectors of furnance flame image were extracted. According to furnace operation conditions, burner outfire was judged when powder burner off. And combustion states were classified with the Support Vector Machine (SVM) & Fisher linear discriminant algorithm when powder burner on. Result shows that the method is simple, convenient and practical and has high correct classification rate. |