Research Of The Dynamical Mathematical Model And The Fault Diagnosis Of Condenser

As an important part of the condensing turbo-generator unit, the operation performance of the condenser is important to the whole plant's safe and economic operation .On the base of a great deal of foreign and domestic documents, learning the structure of condenser and steam flow and heat transfer behavior of condenser, the author analyses the effect on condenser vacuum by some factors, such as the cooling water temperature, the cooling water flow rate, the exhaust steam flew rate, cooling surface, clearing coefficient and tightness of vacuum system. The principle, method and process developing dynamic mathematics model of condenser are discussed in detail. By using the model the author discusses the dynamical characteristics of condenser under several typical cases, especially analyses the handling measures of improving vacuum. At the same time the practical ability of the model is proved through establishing condenser fault information library.The judgment of condenser's fault is not easy for it has all kinds of reasons and signs are fuzzy to some extent. In this paper, neural network, which is fairly advanced, is applied in condenser's fault diagnosis. After introducing the neural network basis and discussing the advantage and disadvantage of five algorithm, the neural network is brought forward. A neural network, which has powerful self-study ability and processing data ability, is an advanced fault diagnosis method. According to dynamic mathematics model and operating rules of the condenser, the three-layer BP network with 17 input nodes, 13 middle nodes and 12 output nodes is constructed. By the training data we can get the knowledge base structure of the neural network. Then visual inquiry of network is completed by Simulink toolbox. A condenser's practical fault is diagnosed with the neural network, and the result is satisfied.