Research On Slagging Characteristics Of Coal-fired Boiler Using Fuzzy Neural Network

In operation of coal-fired boilers in powerplants, slagging is a multiple problem, bothering the operators and plaguing the security,economy in operation and boiler operating availability.So,it's important to take efficient measures to forecast the slagging property of boilers stanchly.Firstly, influence factors of slagging are introduced in detail, an open comment is contributed which was on the advances both at home and abroad in predicting slagging propensity of coal-fired boiler, while the characteristic,merits and faults of the methods from single and muti-judging indicator is discussed and the reasons is analyzed.Secondly,this article presents fuzzy theory and artifical neural network.Based on an overall consideration of characteristic of coal itself , boiler structer design and operation conditions,seven indexes are determined to be inputs of network model,which are softening temperature,silica alumina ratio, base acid ratio,silica ratio,comprehensive target,nondimensional furnace maximum temperature, nondi-mensional acutal diameter of tangent round. Fuzzy inputs of given samples by using six diffierent membership functions. This article introduces thirty-four samples,first twenty-eight as training samples,last six as testing samples. According to maximum membership principle,slagging propensity of testing samples were determined.Using this method, without consideration of boiler stucture design and operation conditions,results of six testing samples are obtained again.Contrasting to seven indexes model,there was a large deviation between the results,which shows characteristic of coal-fired boilers not only relates with characterisitc of coal itself,but the structure and operation conditions.This method takes accout of different membership functions to divide slagging level for discrimination indexes, providing higher accuracy in forecasting slagging characteristic of boiler samples.