Method and system for performance optimization of flue gas desulphurization (FGD) unit

Patent metadata

Field	Value
Publication number	US-12462901-B2
Application number	US-202017597133-A
Country	US
Kind code	B2
Filing date	Jun 27, 2020
Priority date	Jun 27, 2019
Publication date	Nov 4, 2025
Grant date	Nov 4, 2025

Abstract

State of the art techniques used for Flue Gas Desulpharization (FGD) process monitoring fail to comprehend the relationship between various process parameters, which is crucial in determining the performance of a FGD process being monitored. The disclosure herein generally relates to industrial process monitoring, and, more particularly, to a method and system for performance optimization of Flue Gas Desulphurization (FGD) Unit. The system identifies Key Performance Indicators (KPIs) associated with the process being monitored, and identifies parameters associated with each KPI. This information is used to generate several predictive models, from which a predictive model having the highest value of composite model score amongst the predictive models is selected as the predictive model for processing the input data, which is then used to perform optimization of the FGD process.

First claim

The invention claimed is:

1 . A processor implemented method for optimization of a Flue Gas Desulphurization (FGD) process, comprising:
collecting a plurality of plant data using a plurality of sensors, from a FGD process being monitored, as input data, via one or more hardware processors, wherein the plurality of plant data comprises real-time and non-real-time data; pre-processing the input data, comprising removing one or more unwanted components from the input data via the one or more hardware processors; performing a dimensionality reduction on pre-processed input data, via the one or more hardware processors, comprising:
performing a feature selection, wherein the feature selection comprising identifying a plurality of features affecting each of a plurality of Key Performance Indicators (KPIs) of the FGD process, from the pre-processed input data, wherein the plurality of KPIs are a plurality of Process Variables (PVs) representing a running state of the FGD plant and in turn the FGD process being monitored, and wherein the plurality of PVs comprise chimney inlet Sulphur dioxide (SO 2 ) concentration, absorption tower level, absorption tower pH, booster upper fan power consumption, limestone slurry concentration inside tower, and gypsum conversion; and
extracting the plurality of features by performing a feature extraction;
generating a plurality of predictive models based on the plurality of features extracted, for each KPI, via the one or more hardware processors; selecting one of the plurality of predictive models as a predictive model for processing the input data, via the one or more hardware processors; compensating for one or more unmeasured parameters of the selected predictive model using information generated using one or more soft-sensors, via the one or more hardware processors, wherein the one or more unmeasured parameters comprise one or more PVs among the plurality of PVs; performing the optimization of the FGD process, via the one or more hardware processors, comprising:
simulating the operation of the FGD process using the selected predictive model, comprising predicting a plurality of FGD process parameters by the selected predictive model;
estimating a plurality of optimal set points of operation, in real time, from the predicted plurality of FGD process parameters;
determining at least one performance lapse in the FGD process, based on the estimated plurality of optimal set points of operation; and
generating at least one recommendation to optimize the FGD process, in response to the determined at least one performance lapse, wherein the at least one recommendation comprise an optimal set points of a plurality of Manipulated Variables (MVs) of the FGD plant, wherein the plurality of MVs comprise limestone slurry flow rate, limestone slurry concentration, air flow for oxidation, gypsum purging, number of recirculation pumps, and elevation of spray for the pumps, and wherein the plurality of MVs are adjusted by a control system of the FGD plant to bring desirable effects in the plurality of PVs;
continuously pushing the optimal set points of the plurality of MVs into the FGD plant at predefined time intervals to optimize operation of the FGD process; and detecting deviation in working of one or more equipment in the FGD plant by comparing the selected predictive model with one or more working models of the one or more equipment, wherein upon detecting deviation an alarm is generated to alert a user about a potential equipment failure.

2 . The processor implemented method as claimed in claim 1 , wherein selecting the predictive model from the plurality of predictive models, comprising:
calculating values of different performance metrics of the plurality of predictive models; estimating a composite model score of each of the plurality of predictive models, for a given test dataset; comparing the estimated composite model score of each of the plurality of predictive models with estimated composite model score of each of the other predictive models; and selecting a predictive model having the highest value of the composite model score amongst the plurality of predictive models, as the predictive model for processing the input data.

3 . The processor implemented method as claimed in claim 1 , wherein the selected predictive model generates a plurality of FGD process parameters while simulating the operation of the FGD process.

4 . The processor implemented method as claimed in claim 1 , wherein performance of the selected predictive model is validated, comprising:
estimating the error in prediction of the plurality of FGD process parameters for the selected predictive model; calculating a health score of the selected predictive model, based on the estimated error in prediction of the plurality of FGD process parameters; comparing the calculated health score with a threshold of health score; and correcting the selected predictive model if the calculated health score exceeds the threshold health score.

5 . The method as claimed in claim 1 , wherein an equipment monitoring is performed using one or more of the predictive models, comprising:
monitoring at least one equipment in the FGD process for a faulty operation, based on data from at least one predictive model of the at least one equipment, via the one or more hardware processors; generating an alarm if at least one instance of the faulty operation is detected, via the one or more hardware processors; performing a root-cause analysis to determine at least one cause of the detected faulty operation of the equipment, via the one or more hardware processors; and estimating a Remaining Useful Life (RUL) of the equipment, via the one or more hardware processors.

6 . A system for optimization of a Flue Gas Desulphurization (FGD) process, comprising:
one or more hardware processors; a communication interface; and a memory storing a plurality of instructions, the plurality of instructions when executed, cause the one or more hardware processors to:
collect a plurality of plant data using a plurality of sensors, from a FGD process being monitored, as input data, wherein the plurality of plant data comprises real-time and non-real-time data;
pre-process the input data by removing one or more unwanted components from the input data;
perform a dimensionality reduction on pre-processed input data, comprising:
performing a feature selection, wherein the feature selection comprising identifying a plurality of features affecting each of a plurality of Key Performance Indicators (KPIs) of the FGD process, from the pre-processed input data, wherein the plurality of KPIs are a plurality of Process Variables PVs) representing a running state of the FGD plant and in turn the FGD process being monitored, and wherein the plurality of PVs comprise chimney inlet Sulphur dioxide (SO 2 ) concentration, absorption tower level, absorption tower pH, booster upper fan power consumption, limestone slurry concentration inside tower, and gypsum conversion; and
extracting the plurality of features by performing a feature extraction;

generate a plurality of predictive models based on the plurality of features extracted, for each KPI;
select one of the plurality of predictive models as a predictive model for processing the input data;
compensate for one or more unmeasured parameters of the selected predictive model using information generated using one or more soft-sensors, wherein the one or more unmeasured parameters comprise one or more PVs among the plurality of PVs;
perform the optimization of the FGD process, comprising:
simulating the operation of the FGD process using the selected predictive model, comprising predicting a plurality of FGD process parameters by the selected predictive model;
esti

Assignees

• Tata Consultancy Services Ltd

Inventors

• Kumar Rajan
• Venugopal Minimol Pallavi
• Sakhinana Sagar Srinivas
• Baikadi Abhishek
• Doan Duc
• Masampally Vishnu Swaroopji
• Runkana Venkataramana

Classifications

• G05B13/048 Primary

  using a predictor · CPC title

• G16C20/70

  Machine learning, data mining or chemometrics · CPC title

• G05B23/0294

  Optimizing process, e.g. process efficiency, product quality · CPC title

• G05B23/0221

  Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods · CPC title

• G06N20/00

  Machine learning · CPC title