Closed Blast Furnace Smelting Process Intelligent Fault Diagnosis System

Imperial Smelting Furnace (ISF) smelting process is a quite complex process, the traditional fault diagnosis methods generally need relative accurate mathematics model, accordingly their applications come up against many insurmountabilities. So adopting artificial intelligence to this fault diagnosis system is an effective technique, by which advanced diagnosis theory and detection technology are combined into an implement for ISF conditions monitoring and fault diagnosing.hi this thesis, GO-STOP expert system and fault diagnosis system based on Fuzzy Neural Network are mainly discussed. Under the common production conditions, GO-STOP process is adopted to predict present furnace conditions and their tendency. Once furnace conditions are abnormal, reasoning based on Fuzzy Neural Network is started. GO-STOP system is relatively reliable and simple for reasoning, which can keep system running in real time. While detailed and useful diagnosis result depends on Fuzzy Neural Network reasoning when a fault happens. Knowledge-base building is also discussed in this thesis. Large quantity of furnace knowledge is described by fuzzy production rules, which are derived into a layered neural network, as well as their membership functions. Here we adopt a Fuzzy-Neural Technology (FNT) of knowledge acquisition.The thesis is organized as follows. In chapter 1 the origin and significance of the subject is presented. The problems existing in the Imperial Smelting Process (ISP) fault diagnosis and its solution are briefly introduced. Chapter 2 introduces fault diagnosis theory and technology, including intelligent fault diagnosis system structure. Chapter 3 describes the design of ISP fault diagnosis system. In chapter 4 the design is implemented step by step. Here knowledge-base building and diagnosis reasoning are discussed in detail. Chapter 5 gives the software design and realization of ISP fault diagnosis system. At the end of the thesis, we summarize the whole research work and propose the work to be continued.