Automated generation of optimization model for system-wide plant optimization

Patent metadata

Field	Value
Publication number	US-2022027685-A1
Application number	US-202016938884-A
Country	US
Kind code	A1
Filing date	Jul 24, 2020
Priority date	Jul 24, 2020
Publication date	Jan 27, 2022
Grant date	—

Abstract

A computer implemented method for automatically generating an optimization model for site-wide plant optimization includes mapping a process flow diagram of a plant process to a graph comprising nodes and edges, wherein the nodes represent processes and the edges represent flows between processes. A behavior is learned for each node of the graph based at least on historic data of the plant process. One or more regression functions are modeled for each node to predict an output of each of the processes, wherein the one or more regression functions are modeled based on the learned behavior for each node.

First claim

What is claimed is:

1 . A computer implemented method, comprising:
mapping a process flow diagram, of a plant process to a graph comprising nodes and edges, wherein the nodes represent processes and the edges represent flows between processes; learning a behavior for each node of the graph based at least on historic data of the plant process; and modeling one or more regression functions for each node to predict an output of each of the processes, wherein the one or more regression functions are modeled based on the learned behavior for each node.

2 . The computer implemented method of claim 1 , wherein the graph is a two-level fully connected feed-forward network of a plant.

3 . The computer implemented method of claim 1 , wherein inputs for each node include controlled variables and uncontrolled variables.

4 . The computer implemented method of claim 1 , further comprising encoding network topology by generating an adjacency matrix for the graph, the adjacency matrix identifying interconnected nodes.

5 . The computer implemented method of claim 4 , further comprising decoding the network topology using one or more matrices and positions of nodes and edges from the graph to plot a directed acyclic graph for input verification.

6 . The computer implemented method of claim 4 , further comprising generating an optimization model from the regression functions for each node and the adjacency matrix, the optimization model providing setpoints for each node for an optimization of one or more parameters.

7 . The computer implemented method of claim 6 , wherein the model generator outputs a continuous optimization model with a function value estimator and a gradient estimator.

8 . The computer implemented method of claim 6 , wherein the model generator outputs a mixed-integer linear program for modelling the one or more regression functions.

9 . The computer implemented method of claim 6 , further comprising providing different key performance indicators as input and producing options for setpoints to achieve the inputted key performance indicators.

10 . The computer implemented method of claim 1 , further comprising automatically determining whether an input of the regression function is correct.

11 . The computer implemented method of claim 1 , wherein the graph is a two-level fully connected feed-forward network.

12 . The computer implemented method of claim 1 , wherein the regression functions include piece-wise linear and non-linear types of regression models.

13 . A computer implemented method for automatically generating an optimization model for a site-wide optimization of a plant, comprising:
defining a graphical representation for a process flow diagram of the plant; encoding a network topology of the graphical representation to generate an adjacency matrix for the graphical representation; automatically generating a set of equations defining the network topology; modeling one or more regression functions using a machine learning platform to predict an output of each process of the plant based on inputs received at each process; and generating an optimization model from the one or more regression functions for each node and the adjacency matrix, the optimization model providing setpoints for each process of the plant.

14 . The computer implemented method of claim 13 , wherein the graphical representation is a two-level fully connected feed-forward network with no skip layer assumption.

15 . The computer implemented method of claim 13 , wherein the encoding of the network topology is performed with fewer inputs than the inputs of the plant processes.

16 . The computer implemented method of claim 13 , further comprising decoding the network topology using matrices and positions of nodes and edges from the graphical representation to plot a directed acyclic graph for input verification.

17 . The computer implemented method of claim 13 , wherein the encoding of the network topology includes generating an adjacency matrix for the graph, the adjacency matrix identifying interconnected nodes.

18 . The computer implemented method of claim 17 , further comprising generating an optimization model from the regression functions for each node and the adjacency matrix.

19 . The computer implemented method of claim 18 , wherein the model generator outputs a continuous optimization model with a function value estimator and a gradient estimator.

20 . The computer implemented method of claim 18 , wherein the model generator outputs a mixed-integer linear program for modelling the one or more regression functions.

21 . A non-transitory computer readable storage medium tangibly embodying a computer readable program code having computer readable instructions that, when executed, causes a computer device to carry out a method of improving computing efficiency of a computing device for automatically generating an optimization model for site-wide plant optimization, the method comprising:
defining a graphical representation comprising nodes and edges for a process flow diagram of the plant processes, wherein the nodes of the graphical representation represent processes and the edges of the graphical representation represent flows between the plant processes; encoding network topology of the graphical representation by generating an adjacency matrix for the graph, the adjacency matrix identifying interconnected nodes; automatically generating a set of equations defining the network topology; modeling one or more regression functions using a machine learning platform to predict an output of each process of the plant processes based on inputs received at each process; and generating an optimization model from the one or more regression functions for each node and the adjacency matrix, the optimization model providing setpoints for each process of the plant.

22 . The non-transitory computer readable storage medium of claim 21 , wherein the execution of the code by the processor further configures the computing device to perform an act comprising decoding the network topology using matrices and positions of nodes and edges from the graph to plot a directed acyclic graph for input verification.

23 . The computer implemented method of claim 21 , wherein the model generator outputs a continuous optimization model with a function value estimator and a gradient estimator.

24 . The computer implemented method of claim 22 , wherein the model generator outputs a mixed-integer linear program for modelling the one or more regression functions.

25 . The non-transitory computer readable storage medium of claim 22 , wherein the model generator outputs a mixed-integer linear program for modelling the one or more regression functions.

Assignees

• IBM

Inventors

• Phan Dzung Tien
• Nguyen Lam
• Murali Pavankumar
• Zhou Nianjun

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F18/29 Primary

  Graphical models, e.g. Bayesian networks · CPC title

• G06N5/01

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

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title

• G06N3/0499

  Feedforward networks · CPC title