Automated control of circumferential variability of blast furnace

Patent metadata

Field	Value
Publication number	US-10755200-B2
Application number	US-201715712255-A
Country	US
Kind code	B2
Filing date	Sep 22, 2017
Priority date	Sep 22, 2017
Publication date	Aug 25, 2020
Grant date	Aug 25, 2020

Abstract

Controlling circumferential variability in a blast furnace may include generating a predictive model that sets up a relationship between a standard deviation of a selected state variable, state variables and one or more control variables in blast furnace operation for predicting the standard deviation. A number of circumferential sections of the blast furnace is defined, and the predictive model associated with the selected state variable for each of the circumferential sections is trained based on process data of the blast furnace. A plurality trained predictive models is generated associated with different circumferential sections and different selected state variables. One or more future control variable set points that minimize a sum of the plurality of predictive models, is determined. One or more future control variable set points is transmitted to a control system to control the blast furnace operation.

First claim

We claim:

1. A computer program product for controlling circumferential variability in a blast furnace, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions readable by a device to cause the device to:
generate a predictive model that sets up a relationship between a standard deviation of a selected state variable, state variables and at least one control variable in blast furnace operation for predicting a future circumferential variability index associated with the selected state variable;
define a number of circumferential sections of the blast furnace;
receive process data associated with the blast furnace operation;
train the predictive model associated with the selected state variable for each of the circumferential sections based on the process data, wherein a plurality of trained predictive models is generated associated with different circumferential sections and different selected state variables, the different selected state variables comprising temperature and pressure, wherein running the plurality of trained predictive models outputs a plurality of future circumferential variability indices, each of the future circumferential variability indices corresponding to a respective circumferential section and selected state variable;
determine at least one future control variable set point that minimizes a sum of the future circumferential variability indices by solving an optimization problem; and
transmit the at least one future control variable set point to a control system to control the blast furnace operation.

2. The computer program product of claim 1 , wherein the blast furnace operation is a continuous operation.

3. The computer program product of claim 1 , wherein the device caused to determine at least one future control variable set point that minimizes a sum of the future circumferential variability indices, comprises the device caused to determine at least one future control variable set point that minimizes a weighted sum of the future circumferential variability indices.

4. The computer program product of claim 1 , wherein the at least one future control variable set point comprises control variable set points associated with a plurality of future time points.

5. The computer program product of claim 1 , wherein the optimization problem includes a regularization term that controls value fluctuations of future control variable set points between future time points.

6. The computer program product claim 1 , wherein the at least one control variable includes at least one of a dumping rate of input material, a flow rate of blast air, moisture content of blast air, an oxygen enrichment amount of blast air, and a flow rate of pulverized coal.

7. The computer program product of claim 1 , wherein the predictive model's relationship between a standard deviation of a selected state variable, state variables and at least one control variable in blast furnace operation defines a future standard deviation value of the selected state variable as a function of past standard deviations of the selected state variable, past values of the state variables, past values of the at least one control variable and future values of the at least one control variable.

8. The computer program product of claim 1 , wherein the predictive model comprises a deep learning model comprising long short-term memory.

9. A system of controlling circumferential variability in a blast furnace, comprising:
at least one hardware processor operable to generate a predictive model that sets up a relationship between a standard deviation of a selected state variable, state variables and at least one control variable in blast furnace operation for predicting a future circumferential variability index associated with the selected state variable, the at least one hardware processor further operable to define a number of circumferential sections of the blast furnace;
a storage device storing process data associated with the blast furnace operation and coupled to the at least one hardware processor;
the at least one hardware processor further operable to receive the process data associated with the blast furnace operation, train the predictive model associated with the selected state variable for each of the circumferential sections based on the process data, wherein a plurality of trained predictive models is generated associated with different circumferential sections and different selected state variables, the different selected state variables comprising temperature and pressure, wherein running the plurality of trained predictive models outputs a plurality of future circumferential variability indices, each of the future circumferential variability indices corresponding to a respective circumferential section and selected state variable, the at least one hardware processor further operable to determine at least one future control variable set point that minimizes a sum of the future circumferential variability indices by solving an optimization problem;
the at least one hardware processor coupled to a control system, and operable to transmit the at least one future control variable set point to control the blast furnace operation.

10. The system of claim 9 , wherein the at least one hardware processor determines at least one future control variable set point that minimizes a sum of the future circumferential variability indices by determining at least one future control variable set point that minimizes a weighted sum of the future circumferential variability indices, wherein a weight is associated with a section of the circumferential sections.

11. The system of claim 9 , wherein the at least one future control variable set point comprises control variable set points associated with a plurality of future time points.

12. The system of claim 9 , wherein the optimization problem includes a regularization term that controls value fluctuations of future control variable set points between future time points.

13. The system of claim 9 , wherein the at least one control variable includes at least one of a dumping rate of input material, a flow rate of blast air, moisture content of blast air, an oxygen enrichment amount of blast air, and a flow rate of pulverized coal.

14. The system of claim 9 , wherein the predictive model's relationship between a standard deviation of a selected state variable, state variables and at least one control variable in blast furnace operation defines a future standard deviation value of the selected state variable as a function of past standard deviations of the selected state variable, past values of the state variables, past values of the at least one control variable and future values of the at least one control variable.

15. The system of claim 9 , wherein the predictive model comprises a deep learning model comprising long short-term memory.

Assignees

• IBM

Inventors

• Lee Young Min
• Yeo Kyong Min

Classifications

• G06N5/01

  Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/09

  Supervised learning · CPC title