Combining feature selection and surrogate models for the forecast of material concentration in fluids

Patent metadata

Field	Value
Publication number	US-10839113-B2
Application number	US-201715599692-A
Country	US
Kind code	B2
Filing date	May 19, 2017
Priority date	May 19, 2017
Publication date	Nov 17, 2020
Grant date	Nov 17, 2020

Abstract

Embodiments for intelligent forecasting of material concentrations in a fluid by a processor in a computing environment. A material concentration of a material in a fluid may be predicted according to one or more continuous stirred tank reactor (CSTR) surrogate models on statistical flow trajectories of the fluid defined by a principle component analysis (PCA) operation of a system.

First claim

The invention claimed is:

1. A method for intelligent forecasting of material concentrations in a fluid by a processor within a cloud computing environment, comprising:
receiving and processing input data from an advection-diffusion (AD) model of a fluid;
selecting one or more features and parameters from historical flow data of a material in the fluid as input for a principle component analysis (PCA) operation of a system, wherein the historical flow data is part of the input data received from the AD model;
performing the PCA operation according to the one or more features and parameters selected as input, wherein the PCA operation includes:
utilizing the one or more features and parameters to determine a maximum number of principal components (PCs), a maximum number of continuous stirred tank reactor (CSTR) systems per each one of the PCs, and criteria for convergence, and
performing a transformation operation to convert material flow trajectories of the AD model into each one of the PCs;

generating one or more CSTR surrogate models according to an output of the PCA operation, wherein the PCs of the PCA operation define one or more input parameters for the CSTR surrogate models; and
predicting a material concentration of the material in the fluid according to an analyzation of the one or more continuous CSTR surrogate models on statistical flow trajectories of the fluid defined by the PCA operation of the system.

2. The method of claim 1 , further including parameterizing the one or more CSTR surrogate models using the PCA operation of a physical model of the system for the predicting.

3. The method of claim 1 , wherein the system is one or more CSTR systems connected in series.

4. The method of claim 1 , further including calibrating the one or more CSTR surrogate models by iteratively adjusting one or more of the features and parameters of the PCA operation according to the input data received from the AD model.

5. A system for intelligent forecasting of material concentrations in a fluid within a computing environment, comprising:
one or more computers with executable instructions that when executed cause the system to:
receive and process input data from an advection-diffusion (AD) model of a fluid;
select one or more features and parameters from historical flow data of a material in the fluid as input for a principle component analysis (PCA) operation of a system, wherein the historical flow data is part of the input data received from the AD model;
perform the PCA operation according to the one or more features and parameters selected as input, wherein the PCA operation includes:
utilizing the one or more features and parameters to determine a maximum number of principal components (PCs), a maximum number of continuous stirred tank reactor (CSTR) systems per each one of the PCs, and criteria for convergence, and
performing a transformation operation to convert material flow trajectories of the AD model into each one of the PCs;

generate one or more CSTR surrogate models according to an output of the PCA operation, wherein the PCs of the PCA operation define one or more input parameters for the CSTR surrogate models; and
predict a material concentration of the material in the fluid according to an analyzation of the one or more continuous CSTR surrogate models on statistical flow trajectories of the fluid defined by the PCA operation of the system.

6. The system of claim 5 , wherein the executable instructions further parameterize the one or more CSTR surrogate models using the PCA operation of a physical model of the system for the predicting.

7. The system of claim 5 , wherein the system is one or more CSTR systems connected in series.

8. The system of claim 5 , wherein the executable instructions further
calibrate the one or more CSTR surrogate models by iteratively adjusting one or more of the features and parameters of the PCA operation according to the input data received from the AD model.

9. A computer program product for, by a processor, intelligent forecasting of material concentrations in a fluid, the computer program product comprising a non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising:
an executable portion that receives and processes input data from an advection-diffusion (AD) model of a fluid;
an executable portion that selects one or more features and parameters from historical flow data of a material in the fluid as input for a principle component analysis (PCA) operation of a system, wherein the historical flow data is part of the input data received from the AD model;
an executable portion that performs the PCA operation according to the one or more features and parameters selected as input, wherein the PCA operation includes:
utilizing the one or more features and parameters to determine a maximum number of principal components (PCs), a maximum number of continuous stirred tank reactor (CSTR) systems per each one of the PCs, and criteria for convergence, and
performing a transformation operation to convert material flow trajectories of the AD model into each one of the PCs;

an executable portion that generates one or more CSTR surrogate models according to an output of the PCA operation, wherein the PCs of the PCA operation define one or more input parameters for the CSTR surrogate models; and
an executable portion that predicts a material concentration of the material in the fluid according to an analyzation of the one or more continuous CSTR surrogate models on statistical flow trajectories of the fluid defined by the PCA operation of the system.

10. The computer program product of claim 9 , further including an executable portion that parameterizes the one or more CSTR surrogate models using the PCA operation of a physical model of the system for the predicting.

11. The computer program product of claim 9 , wherein the system is one or more CSTR systems connected in series.

12. The computer program product of claim 9 , further including an executable portion that calibrates the one or more CSTR surrogate models by iteratively adjusting one or more of the features and parameters of the PCA operation according to the input data received from the AD model.

Assignees

• IBM

Inventors

• Arandia Ernesto
• Mckenna Sean A
• O'Donncha Fearghal
• Ragnoli Emanuele

Classifications

• G06F2111/10

  Numerical modelling · CPC title

• G06F30/20 Primary

  Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30) · CPC title