APC process control when process parameters are inaccurately measured

• US 7,698,004 B2
• Filed: 12/03/2004
• Issued: 04/13/2010
• Est. Priority Date: 08/27/2004
• Status: Expired due to Fees

• Current Assignee: General Electric Technology GmbH (GE Vernova, Inc.)
• Original Assignee: Alstom Technology Ltd. (Alstom SA)
• Inventors: Piche, Stephen, Boyden, Scott A.
• Primary Examiner(s): BARNES-BULLOCK, CRYSTAL JOY
• Application Number:
  US11/002,439
• Publication Number:
  US 20060045803A1
• Time in Patent Office: 1,957 Days
• Field of Search: 700/28, 700/29, 700/32, 700/44, 700/48, 700/51, 700/52, 700/54, 700/56, 700/266, 700/267, 700/274, 702 22- 24, 702/30, 422/83, 422/88, 422/168, 422/177, 423/166, 423/242.1, 423/242.2, 423/243.01, 423/243.02, 423/243.03, 423/243.06, 423/244.01, 423/244.02, 423/244.09, 423/235, 423/237, 423/239.1, 701101-103, 701/106, 701/109, 436/55, 602/74
• US Class Current: 700/52
• CPC Class Codes: B01D 2257/302 Sulfur oxides
• CPC Class Codes: B01D 2257/404 Nitrogen oxides other than dinitrogen oxide
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: G05B 13/048 using a predictor
• CPC Class Codes: Y10T 436/12 Condition responsive control

Abstract

A controller is provided for directing control of a process performed to control an amount of a pollutant emitted into the air. The process has multiple process parameters (MPPs) The controller includes either a neural network process model or a non-neural network process model. Whichever type model is included, it will represent a relationship between one of the MPPs and other of the MPPs. The controller also includes a control processor having the logic to determine the validity of a measured value of the one MPP based on the one model. The control processor directs control of the process in accordance with the measured value of the one MPP only if the measured value of the one MPP is determined to be valid. On the other hand, if the measured value is determined to be invalid, the control processor may direct control of the process in accordance with an estimated value of the one MPP.

Claims

1. A controller for directing control of a process, having multiple

process parameters (MPPs), performed to control an amount of a pollutant emitted into the air, comprising:

one of a neural network process model and a non-neural network process model, the one model representing a relationship between one of the MPPs and other of the MPPs;

a control processor configured with the logic to determine validity of a measured value of the one MPP based on the one model, and to direct control of the process in accordance with the measured value of the one MPP, only if the measured value of the one MPP is determined to be valid; and

an estimator configured with logic to estimate the value of the one MPP based on actual values of the other MPPs and the one model;

wherein the control processor determines the validity of the measured value of the one MPP based on the one model by comparing the estimated value of the one MPP with the measured value of the one MPP and, if the measured value of the one MPP is determined to be invalid, directs control of the process in accordance with an estimated value of the one MPP;

wherein the process is a wet flue gas desulfurization (WFGD) process which applies a reactant to remove SO₂ from SO₂ laden wet flue gas, and exhausts desulfurized flue gas;

wherein the one MPP is a pH level of the applied reactant;

wherein the one model represents a relationship between the pH level of the applied reactant and the other MPPs;

wherein the estimator estimates the value of the pH level of the applied reactant based on the actual values of the other MPPs and the one model; and

wherein the control processor compares the estimated value of the pH level of the applied reactant with the measured value of the pH level of the applied reactant, to determine the validity of the measured value of the pH level of the applied reactant, and (i) if the measured value of the pH level of the applied reactant is determined to be valid, directs control of the WFDG process in accordance with the measured value of the pH level of the applied reactant, and (ii) if the measured value of the pH level of the applied reactant is determined to be invalid, directs control of the WFGD process in accordance with the estimated value of the pH level of the applied reactant.

2. The controller according to claim 1, wherein:

the reactant is a limestone slurry;

the other MPPs include an amount of the SO₂ in the SO₂ laden flue gas and an amount of the SO₂ in the exhausted desulfurized flue gas;

the one model represents a relationship between the pH level of the applied limestone slurry and the SO₂ in the SO₂ laden flue gas and the SO₂ in the exhausted desulfurized flue gas; and

the estimator estimates the value of the pH level of the applied limestone slurry based on the actual amount of SO₂ in the SO₂ laden flue gas and the actual amount of SO₂ in the exhausted desulfurized flue gas, and on the one model.

3. A controller for directing control of a process, having multiple process parameters (MPPs), performed to control an amount of a pollutant emitted into the air, comprising:

one of a neural network process model and a non-neural network process model, the one model representing a relationship between one of the MPPs and other of the MPPs;

a control processor configured with the logic to determine validity of a measured value of the one MPP based on the one model, and to direct control of the process in accordance with the measured value of the one MPP, only if the measured value of the one MPP is determined to be valid; and

an estimator configured with logic to estimate the value of the one MPP based on actual values of the other MPPs and the one model;

wherein the control processor determines the validity of the measured value of the one MPP based on the one model by comparing the estimated value of the one MPP with the measured value of the one MPP and, if the measured value of the one MPP is determined to be invalid, directs control of the process in accordance with an estimated value of the one MPP;

wherein the process is a selective catalytic reduction (SCR) process which applies a reactant to remove NO_x from NO_x laden flue gas, and exhausts reduced NO_x flue gas;

wherein the one MPP is an amount of the applied reactant;

wherein the one model represents a relationship between the amount of applied reactant and the other MPPs;

wherein the estimator estimates the value of the amount of applied reactant based on the actual values of the other MPPs and the one model; and

wherein the control processor compares the estimated value of the amount of applied reactant with the measured value of the amount of applied reactant, to determine the validity of the measured value of the amount of applied reactant, and (i) if the measured value of the amount of applied reactant is determined to be valid, directs control of the SCR process in accordance with the measured value of the amount of applied reactant, and (ii) if the measured value of the amount of applied reactant is determined to be invalid, directs control of the SCR process in accordance with the estimated value of the amount of applied reactant.

4. The controller according to claim 3, wherein:

the reactant is ammonia;

the other MPPs include an amount of the NO_x in the NO_x laden flue gas and an amount of the NO_x in the exhausted reduced NO_x flue gas;

the one model represents a relationship between the amount of the applied ammonia and the NO_x in the NO_x laden flue gas and the NO_x in the exhausted reduced NO_x flue gas; and

the estimator estimates the value of the amount of the applied ammonia based on the actual amount of NO_x in the NO_x laden flue gas and the actual amount of NO_x in the exhausted reduced NO_x flue gas, and on the one model.

5. An article of manufacture for directing control of a process, having multiple process parameters (MPPs), performed to control an amount of a pollutant emitted into the air, comprising:

computer readable storage media; and

logic stored on the storage media, wherein the stored logic is configured to be readable by one or more computers and thereby cause the one or more computers to operate so as to;

determine validity of a measured value of one of the MPPs based on one of a neural network process model and a non-neural network process model, the one model representing a relationship between the one MPP and other of the MPPs;

direct control of the process in accordance with the measured value of the one MPP, only if the measured value of the one MPP is determined to be valid;

estimate a value of the one MPP based on actual values of the other MPPs and the one model;

compare the estimated value of the one MPP with the measured value of the one MPP; and

if the measured value of the one MPP is determined to be invalid, direct control of the process in accordance with the estimated value of the one MPP;

wherein the validity of the measured value of the one MPP is determined based also on the comparison of the estimated value of the one MPP with the measured value of the one MPP;

wherein the process is a wet flue gas desulfurization (WFGD) process which applies a reactant to remove SO₂ from SO₂ laden wet flue gas, and exhausts desulfurized flue gas;

wherein the one MPP is a pH level of the applied reactant;

wherein the one model represents a relationship between the pH level of the applied reactant and the other MPPs;

wherein the value of the pH level of the applied reactant is estimated based on the actual values of the other MPPs and the one model;

wherein the estimated value of the pH level of the applied reactant is compared with the measured value of the pH level of the applied reactant, to determine the validity of the measured value of the pH level of the applied reactant;

wherein if the measured value of the pH level of the applied reactant is determined to be valid, control of the WFDG process is directed in accordance with the measured value of the pH level of the applied reactant; and

wherein if the measured value of the pH level of the applied reactant is determined to be invalid, control of the WFGD process is directed in accordance with the estimated value of the pH level of the applied reactant.

6. The article of manufacture according to claim 5, wherein:

the reactant is a limestone slurry;

the other MPPs include an amount of the SO₂ in the SO₂ laden flue gas and an amount of the SO₂ in the exhausted desulfurized flue gas;

the one model represents a relationship between the pH level of the applied limestone slurry and the SO₂ in the SO₂ laden flue gas and the SO₂ in the exhausted desulfurized flue gas; and

the value of the pH level of the applied limestone slurry is estimated based on the actual amount of SO₂ in the SO₂ laden flue gas and the actual amount of SO₂ in the exhausted desulfurized flue gas, and on the one model.

7. An article of manufacture for directing control of a process, having multiple process parameters (MPPs), performed to control an amount of a pollutant emitted into the air, comprising:

computer readable storage media; and

logic stored on the storage media, wherein the stored logic is configured to be readable by one or more computers and thereby cause the one or more computers to operate so as to;

determine validity of a measured value of one of the MPPs based on one of a neural network process model and a non-neural network process model, the one model representing a relationship between the one MPP and other of the MPPs;

direct control of the process in accordance with the measured value of the one MPP, only if the measured value of the one MPP is determined to be valid;

estimate a value of the one MPP based on actual values of the other MPPs and the one model;

compare the estimated value of the one MPP with the measured value of the one MPP; and

if the measured value of the one MPP is determined to be invalid, direct control of the process in accordance with the estimated value of the one MPP;

wherein the validity of the measured value of the one MPP is determined based also on the comparison of the estimated value of the one MPP with the measured value of the one MPP;

wherein the process is a selective catalytic reduction (SCR) process which applies a reactant to remove NO_x from NO_x laden flue gas, and exhausts reduced NO_x flue gas;

wherein the one MPP is an amount of the applied reactant;

wherein the one model represents a relationship between the amount of applied reactant and the other MPPs;

wherein the value of the amount of applied reactant is estimated based on the actual values of the other MPPs and the one model; and

wherein the estimated value of the amount of applied reactant is compared with the measured value of the amount of applied reactant, to determine the validity of the measured value of the amount of applied reactant;

wherein if the measured value of the amount of applied reactant is determined to be valid, control of the SCR process is directed in accordance with the measured value of the amount of applied reactant; and

wherein if the measured value of the amount of applied reactant is determined to be invalid, control of the SCR process is directed in accordance with the estimated value of the amount of applied reactant.

8. The article of manufacture according to claim 7, wherein:

the reactant is ammonia;

the other MPPs include an amount of the NO_x in the NO_x laden flue gas and an amount of the NO_x in the exhausted reduced NO_x flue gas;

the one model represents a relationship between the amount of the applied ammonia and the NO_x in the NO_x laden flue gas and the NO_x in the exhausted reduced NO_x flue gas; and

the value of the amount of the applied ammonia is estimated based on the actual amount of NO_x in the NO_x laden flue gas and the actual amount of NO_x in the exhausted reduced NO_x flue gas, and on the one model.

9. A wet flue gas desulfurizing system, comprising:

a wet flue gas desulfurizer configured (i) to receive SO₂ laden wet flue gas, (ii) to apply limestone slurry to remove SO₂ from the received SO₂ laden wet flue gas, and (iii) to exhaust desulfurized flue gas;

a pH sensor to measure a pH level of the applied limestone slurry;

one of a neural network process model and a non-neural network process model, the one model representing a relationship between the pH level of the applied limestone slurry and an amount of the SO₂ in the received SO₂ laden wet flue gas and an amount of the SO₂ in the exhausted desulfurized flue gas; and

a control processor configured with the logic to determine validity of the measured pH level of the applied limestone slurry based on the one model, and to direct control of the process in accordance with the measured pH level of the applied limestone slurry only if the measured pH level of the applied limestone slurry is determined to be valid.

10. The system of claim 9, wherein the sensor is a first sensor, and further comprising:

a second sensor to measure an actual amount of the SO₂ in the received SO₂ laden wet flue gas values; and

a third sensor configured to measure an actual amount of the SO₂ in the exhausted desulfurized flue gas;

a virtual analyzer having the logic to estimate the pH level of the applied limestone slurry based on the measured amount of SO₂ in the received SO₂ laden wet flue gas, the measured amount of SO₂ in the exhausted desulfurized flue gas and the one model; and

a feed back loop configured to transmit the measured pH level of the applied limestone slurry, the measured amount of SO₂ in the received SO₂ laden wet flue gas, and the measured amount of SO₂ in the exhausted desulfurized flue gas, to the virtual analyzer in real time;

wherein the virtual analyzer estimates the pH level of the applied limestone slurry based on the transmitted measured amounts of SO₂ in the received SO₂ laden wet flue gas and in the exhausted desulfurized flue gas, in real time;

wherein the control processor is further configured with the logic to compare the estimated pH level of the applied limestone slurry with the measured pH level of the applied limestone slurry, to determine the validity of the measured pH level of the applied limestone slurry based on the comparison and, if the measured pH level of the applied limestone slurry is determined to be invalid, to direct control of the process in accordance with the estimated pH level of the applied limestone slurry.

11. A selective catalytic reduction system, comprising:

selective catalytic reducer configured (i) to receive NO_x laden flue gas, (ii) to apply ammonia to remove NO_x from the received NO_x laden flue gas, and (iii) to exhaust reduced NO_x flue gas;

an ammonia sensor to measure an amount of ammonia in the exhausted reduced NO_x flue gas;

one of a neural network process model and a non-neural network process model, the one model representing a relationship between the amount of the ammonia in the exhausted reduced NO_x flue gas and an amount of NO_x in the received NO_x laden flue gas and an amount of NO_x in the exhausted reduced NO_x flue gas; and

a control processor having the logic to determine validity of the measured amount of ammonia in the exhausted reduced NO_x flue gas based on the one model, and to direct control of the process in accordance with the measured amount of ammonia in the exhausted reduced NO_x flue gas only if the measured amount of ammonia in the exhausted reduced NO_x flue gas is determined to be valid.

12. The system of claim 11, wherein the sensor is a first sensor, and further comprising:

a second sensor to measure an actual amount of the NO_x in the received NO_x laden flue gas; and

a third sensor configured to measure an actual amount of the NO_x in the exhausted reduced NO_x flue gas;

a virtual analyzer having the logic to estimate the amount of ammonia in the exhausted reduced NO_x flue gas based on the measured amount of NO_x in the received NO_x laden flue gas and the measured amount of NO_x in the exhausted reduced NO_x flue gas, and on the one model; and

a feed back loop configured to transmit the measured amount of NO_x in the received NO_x laden flue gas, and the measured amount of NO_x in the exhausted reduced NO_x flue gas, to the virtual analyzer in real time;

wherein the virtual analyzer estimates the amount of ammonia in the exhausted reduced NO_x flue gas based on the transmitted measured amounts of NO_x in the received NO_x laden flue gas and in the exhausted reduced NO_x flue gas, in real time;

wherein the control processor is further configured with the logic to compare the estimated amount of ammonia in the exhausted reduced NO_x flue gas with the measured amount of ammonia in the exhausted reduced NO_x flue gas, to determine the validity of the measured amount of ammonia in the exhausted reduced NO_x flue gas based on the comparison and, if the measured amount of ammonia in the exhausted reduced NO_x flue gas is determined to be invalid, to direct control of the process in accordance with the estimated amount of ammonia in the exhausted reduced NO_x flue gas.