Zheng Yi, Zuo Kanglin. Prediction Model of Desulfurizer Consumption Based on BP Neural Network and Regression[J]. IRON STEEL VANADIUM TITANIUM, 2017, 38(4): 130-134. doi: 10.7513/j.issn.1004-7638.2017.04.023
Citation:
Zheng Yi, Zuo Kanglin. Prediction Model of Desulfurizer Consumption Based on BP Neural Network and Regression[J]. IRON STEEL VANADIUM TITANIUM, 2017, 38(4): 130-134. doi: 10.7513/j.issn.1004-7638.2017.04.023
Zheng Yi, Zuo Kanglin. Prediction Model of Desulfurizer Consumption Based on BP Neural Network and Regression[J]. IRON STEEL VANADIUM TITANIUM, 2017, 38(4): 130-134. doi: 10.7513/j.issn.1004-7638.2017.04.023
Citation:
Zheng Yi, Zuo Kanglin. Prediction Model of Desulfurizer Consumption Based on BP Neural Network and Regression[J]. IRON STEEL VANADIUM TITANIUM, 2017, 38(4): 130-134. doi: 10.7513/j.issn.1004-7638.2017.04.023
The effect of hot metal temperature,hot metal amount,initial sulfur content,end-point sulfur content,etc,on the consumption of Mg powder was analyzed.It is found out that the consumption of magnesium powder increased with elevation of hot metal temperature,and obviously increased with decrease of end-point sulfur content by amplitude enlarge gradually.In order to reduce cost,the depth of desulfurization should be controlled properly.The desulfurizer consumption prediction model based on BP neural network and regression was established to ascertain appropriate powder consumption. The BP neural network was the main output and the regression model was used to constrain output range. The prediction model was realized on-line control without human intervention,and good application effect was achieved.If deviation between end-point sulfur content and aim sulfur content was within [-0.001 5%,0.001 5%],the coincidence rate of end-point sulfur content reached 90.85%,so the model could effectively achieve the control of the sulfur content after desulfurization.