Method for blast furnace top temperature measurement based on multi-source data fusion

Abstract

A method for blast furnace top temperature measurement based on multi-source data fusion. First, a blast furnace is provided. The blast furnace includes an acoustic sensing device, a non-acoustic sensing device, and a top thermometer. The acoustic sensing device and the top thermometer are respectively installed on the top of the blast furnace. After acquiring the sound wave propagation velocity parameter in the gas medium at the top of the blast furnace from the acoustic sensing device and the blast furnace condition parameters from the non-acoustic sensing device, the multi-source data fused from the acoustic wave propagation velocity parameter and the blast furnace condition parameters are used as input features for training and learning. Using the set of measurements from the top thermometer as the target, machine learning algorithms are used to optimize the input features and construct a blast furnace top temperature field transformation model.

Claims

1. A method for measuring the temperature at the top of a blast furnace based on multi-source data fusion, comprising: providing a blast furnace, the blast furnace including an acoustic sensing device, a non-acoustic sensing device, and a furnace top thermometer, wherein the acoustic sensing device and the furnace top thermometer are respectively disposed at the top of the blast furnace; acquiring acoustic propagation velocity parameters in the gas medium at the top of the blast furnace from the acoustic sensing device; acquiring blast furnace condition parameters from the non-acoustic sensing device; and using the multi-source data fused from the acoustic propagation velocity parameters and the blast furnace condition parameters as input features for training and learning, and using the set of measurements from the furnace top thermometer as the target, optimizing the input features using machine learning algorithms to construct a furnace top temperature field conversion model.

2. The method as described in claim 1, wherein the sound wave sensing device acquires a complex array of sound wave path velocity data.

3. The method as described in claim 1, wherein the blast furnace condition parameters are selected from a group consisting of a tuyere flow rate, a blast pressure, a hot blast temperature, a furnace top pressure, a furnace wall temperature, and charging-related information.

4. The method as described in claim 1, wherein the furnace top thermometer acquires a complex array of planar distributed measurements to form the set of measurements.

5. The method as described in claim 4, wherein the set of measurements excludes the planar distribution measurements located in the central region of the blast furnace.

6. The method as described in claim 1, wherein a coefficient of determination between the input feature and the set of measurements is greater than 0.

7.

7. The method as described in claim 6, wherein the coefficient of determination is greater than 0.

9.

8. The method as described in claim 1, wherein the furnace top temperature field conversion model infers a virtual furnace top temperature.

9. The method as described in claim 8, wherein the furnace top temperature field conversion model integrates the set of measurements and the virtual furnace top temperature to output a furnace top temperature field distribution map.

10. The method as described in claim 5, wherein the furnace top temperature field conversion model integrates the set of measurements with a virtual furnace top temperature and outputs a furnace top temperature field distribution map.

Images