?The hot metal temperature can't be stabilized, resulting in production loss and poor quality in the hot metal composition.
?Since there are large variations in the hot metal temperature, we charge excess fuel to avoid the risk of a drop in the hot metal temperature.
?Provides the optimum guidance by utilizing not only a physical model for predicting 8 to 12 hours in the future, but also a machine learning model that predicts the hot metal temperature 2 to 3 hours in advance with high accuracy, an AI model that learns the operations of operators, etc.
?The physical model automatically corrects parameters by using on-line data.
?It is possible to select the manipulated variable (control input) for hot metal temperature operation from the blast moisture, pulverized coal injection (PCI) rate, and the blast temperature.
?In addition to issuing guidance to the operator, the model can also input the output recommended actions directly to the control system, without going through the operator, making it possible to provide a fully-automatic mode that implements furnace heat adjustments automatically.
By applying this model to blast furnaces in 91视频 Steel, variations in the hot metal temperature were successfully reduced by 13 % (this company's data) when the operator followed guidance. (Before application: σ = 15.9℃ after application: 14.0℃)
Search for
