Study On Sintering Burn-through Point Prediction Model And Control Method

Sintering is the pretreatment for the iron and steel production,and its production process contains a large number of physical and chemical reactions,which make the strong hysteresis of sintering process.And there is a strong coupling relationship between the variables.Burn-through point is the corresponding position of bellow at which the sintering materials are burned through.BTP has great influence on the production and quality.But due to the poor conditions and the limitation of the detection means,there is no instrument for BTP detection directly at home and abroad.So the research of BTP detection method,prediction model and control algorithm has an important theoretical significance and application value for improving the quality and the output of sinter.For the difficult of detecting and controlling of sintering BTP,a detection method was carried out and a prediction model to BTP was established,which could predicate the BTP 5 minutes ahead.The variable universe fuzzy controller was designed to fit the characteristics of sintering process.And the following researches were made in this thesis:(1)Fist,the theory of judgment to BTP according to exhaust gas temperature and related calibration method were studied.Meanwhile the physical meaning of the exhaust gas temperature rising point and calculation method were analyzed.(2)The detecting method of exhaust gas temperature and flue oxygen concentration was presented.And the influences of pallet speed,mixing material moisture,ignition temperature,the temperature of the 20 bellow and other variables to the BTP were analyzed.(3)The training algorithm about RBF neural network based on subtractive clustering and particle swarm algorithm was designed.Through the identification of a nonlinear model,the network's generalization ability was studied.And BTP prediction model was established based on this neural network.(4)Finally,the model between sintering pallet speed and the BTP was identified with the least squares method,and the variable universe fuzzy controller was designed based on this model.Through the simulation,the feasibility of the controller was verified.The results showed that the maximum error of prediction model was 0.19,and the accurate was 82%when the error was in the range of ±0.1.Variable universe fuzzy controller could track the change of the given value very well.When there was a disturbance of ±0.3 in the system,the output fluctuation was within ±0.3.