Study On Intelligent Flame Digital Image Monitoring For Power Plant System

It's very important to protect life and fortune of the country to make boiler of power plant running smoothly. After analyzing the tradition flame detecting method, we find it's has the problem of badly dynamic Auto-adoptions, low perspicacity and single function. So the article design a Image Flame detecting System, first we establish a module of image flame and analyze character of flame and giving a formula about the image flame and combustion condition. In order to enhance image quality, the article disposal a method by using method of color-gray transform, median filter and sharp image enhance. For image segment the article using a two-coordinate maximum entropy segment method. To training the sample the article put forward a monitor type Fuzz-C cluster arithmetic and using it to establish a standard pattern database. Based on human cognize character ,the article put forward a Weightiness Amend Coefficient(WAC) to denote the important extent of different character. And the article treated the sample with maximum Fuzz Similar Extent arithmetic to recognize it.For the purpose of enhancing the intelligent and auto-adoptions to complex combustion condition, the article put forward a new Cluster Center Initiative Optimize and Auto-adoptions Rework arithmetic. Finally we implement the system and relative arithmetic using Visual C++ 6.0 program. Experimentations show that the system and relative arithmetic is doable.