| In the blast furnace smelting process,sinter with stable cold strength is not only increased the rate of the blast furnace material usage,but also improved the permeability of blast furnace and the condition of powdering sinters at the same time,thus reduced the cost of the blast furnace production.The method of quality control that is based on the periodic sampling and testing results of sinter tumbler strength can not meet the requirements of industrial production.Since the low detection efficiency,poor accuracy and feedback delay of traditional detecting method,C hina will lose competitiveness global in steel industry.Therefore,it is significant to research online prediction of sinter tumbler strength.Hysteresis of Conventional detection method on sintering quality and detection feeds without in time led to sintering parameters are very difficult to adjust accurate.This paper presents an online prediction research method of sinter qua lity based on artificial neural network.And using BP,RBF and Elman neural network to forecast the sinter drum strength.Take the data from the instrument with PGNAA technology as the imports of the prediction model which is built with historical sinter data,forecasting the tumbler strength and abrasion index.And then,return the prediction results to the sintering process in order to achieve the online control of sinter quality.The simulation results present that the BP neural network forecast is partly accurate and its error fluctuates greatly.The inconformity of RBF neural network predicted value and actual value s reflect that the RBF neural network is highly samples dependent.While the Elman neural network prediction values are more accurate than BP and RBF,its prediction results are almost the same as the actual values.The result of Elman neural network predicts not only fit in with the needs of the practical production but also has the time shortened,thereby improved the quality of sinter. |
PDF Full Text Request