Data-efficient multi-acquisition strategy for selecting high-cost computational objective functions
A method of optimizing parameters for an industrial process is described, along with media and systems, using a digital twin, physics based model and multiple types of acquisition functions. Output data from the model is analyzed by multiple types of Bayesian acquisition functions, such as an expected improvement acquisition function and a model variance acquisition function. The different acquisition functions tune better parameters, and then then model is re-run in parallel for each to output more data. The data from one acquisition function's run of the model may be co-mingled with data from the other acquisition function's run of the model such that the acquisition functions' exploration and exploitation of the parameter space are intertwined, thus achieving a more globally optimal solution than using just one type of Bayesian acquisition function.
1 . A method of optimizing input parameters for an industrial process, the method comprising:
executing a computation model of an industrial process using seed input parameters to generate a first output, the seed input parameters and first output forming a first data set;
applying a first Bayesian acquisition function to the first data set to generate a second input parameter and then executing the computation model using the second input parameter to generate a second data set;
applying a second Bayesian acquisition function to the first data set to generate a third input parameter and then executing the computation model using the third input parameter to generate a third data set, wherein the first and second Bayesian acquisition functions are different types of Bayesian acquisition functions from one another;
using the first and second Bayesian acquisition functions to analyze the first, second, and third data sets;
running the first Bayesian acquisition function based on data from the third data set generated by the second Bayesian acquisition function to generate a fourth input parameter and then executing the computation model using the fourth input parameter to generate a fourth data set;
running the second Bayesian acquisition function based on data from the second or third data set to generate a fifth input parameter and then executing the computational model using the fifth input parameter to generate a fifth data set; and
selecting a best input parameter from among an output from the first and second types of Bayesian acquisition functions, wherein the output comprises the first, second, third, fourth, or fifth input parameter; and
setting the best input parameter on a physical component that performs the industrial process.
2 . The method of claim 1 , wherein the types of Bayesian acquisition functions are selected from the group consisting of:
an expected improvement acquisition function,
a probability of improvement acquisition function,
a negative lower confidence bounds acquisition function, and
a model variance acquisition function.
3 . The method of claim 2 , wherein the first Bayesian acquisition function is the expected improvement acquisition function, and the second Bayesian acquisition function is the model variance acquisition function.
4 . The method of claim 1 wherein running the second Bayesian acquisition function is based on data from the second data set generated by the first Bayesian acquisition function.
5 . The method of claim 1 further comprising:
using the first and second Bayesian acquisition functions to analyze the first, second, third, fourth, and fifth data sets; and
running the first Bayesian acquisition function based on data from the fourth data set generated by the first Bayesian acquisition function.
6 . The method of claim 1 wherein:
the data sets include target output defined by an objective function as well as field data that is auxiliary to the target output; and
the first and second Bayesian acquisition functions rely upon the field data for generating input parameters.
7 . The method of claim 1 wherein the first and second Bayesian acquisition functions continue to generate input parameters, and the computation model is executed with the input parameters, until a number of iterations is completed, compute power budget is reached, or a target output defined by an objective function reaches a target.
8 . The method of claim 7 wherein the first Bayesian acquisition function is not allowed to analyze the data sets until the computation model has completed execution using input parameters from all other Bayesian acquisition functions.
9 . A non-transitory machine-readable tangible medium embodying information indicative of instructions for causing one or more machines to perform operations for optimizing input parameters for an industrial process, the instructions comprising:
executing a computation model of an industrial process using seed input parameters to generate a first output, the seed input parameters and first output forming a first data set;
applying a first Bayesian acquisition function to the first data set to generate a second input parameter and then executing the computation model using the second input parameter to generate a second data set;
applying a second Bayesian acquisition function to the first data set to generate a third input parameter and then executing the computation model using the third input parameter to generate a third data set, wherein the first and second Bayesian acquisition functions are different types of Bayesian acquisition functions from one another;
using the first and second Bayesian acquisition functions to analyze the first, second, and third data sets;
running the first Bayesian acquisition function based on data from the third data set generated by the second Bayesian acquisition function to generate a fourth input parameter and then executing the computation model using the fourth input parameter to generate a fourth data set;
running the second Bayesian acquisition function based on data from the second or third data set to generate a fifth input parameter and then executing the computational model using the fifth input parameter to generate a fifth data set;
selecting a best input parameter from among an output from the first and second types of Bayesian acquisition functions, wherein the output comprises the first, second, third, fourth, or fifth input parameter; and
setting the best input parameter on a physical component that performs the industrial process.
10 . The medium of claim 9 , wherein the types of Bayesian acquisition functions are selected from the group consisting of:
an expected improvement acquisition function,
a probability of improvement acquisition function,
a negative lower confidence bounds acquisition function, and
a model variance acquisition function.
11 . The medium of claim 10 , wherein the first Bayesian acquisition function is the expected improvement acquisition function, and the second Bayesian acquisition function is the model variance acquisition function.
12 . The medium of claim 9 wherein running the second Bayesian acquisition function is based on data from the second data set generated by the first Bayesian acquisition function.
13 . The medium of claim 9 wherein the instructions further comprise:
using the first and second Bayesian acquisition functions to analyze the first, second, third, fourth, and fifth data sets; and
running the first Bayesian acquisition function based on data from the fourth data set generated by the first Bayesian acquisition function.
14 . The medium of claim 9 wherein:
the data sets include target output defined by an objective function as well as field data that is auxiliary to the target output; and
the first and second Bayesian acquisition functions rely upon the field data for generating input parameters.
15 . A system for optimizing input parameters for an industrial process, the system comprising:
a memory; and
at least one processor operatively coupled with the memory and executing program code from the memory for:
executing a computation model of an industrial process using seed input parameters to generate a first output, the seed input parameters and first output forming a first data set;
applying a first Bayesian acquisition function to the first data set to generate a second input parameter and then executing the computation model using the second input parameter to generate a second data set;
applying a second Bayesian acquisition function to the first data set to generate a third input parameter and then executing the computation model using the third input parameter to generate a third data set, wherein the first and second Bayesian acquisition functions are different types of Bayesian acquisition functions from one another;
using the first and second Bayesian acquisition functions to analyze the first, second, and third data sets;
running the first Bayesian acquisition function based on data from the third data set generated by the second Bayesian acquisition function to generate a fourth input parameter and then executing the computation model using the fourth input parameter to generate a fourth data set;
running the second Bayesian acquisition function based on data from the second or third data set to generate a fifth input parameter and then executing the computational model using the fifth input parameter to generate a fifth data set;
selecting a best input parameter from among an output from the first and second types of Bayesian acquisition functions, wherein the output comprises the first, second, third, fourth, or fifth input parameter; and
setting the best input parameter on a physical component that performs the industrial process.
16 . The system of claim 15 , wherein the types of Bayesian acquisition functions are selected from the group consisting of:
an expected improvement acquisition function,
a probability of improvement acquisition function,
a negative lower confidence bounds acquisition function, and
a model variance acquisition function.
17 . The system of claim 16 , wherein the first Bayesian acquisition function is the expected improvement acquisition function, and the second Bayesian acquisition function is the model variance acquisition function.
18 . The system of claim 15 wherein running the second Bayesian acquisition function is based on data from the second data set generated by the first Bayesian acquisition function.
19 . The system of claim 15 wherein the program code further comprises:
using the first and second Bayesian acquisition functions to analyze the first, second, third, fourth, and fifth data sets; and
running the first Bayesian acquisition function based on data from the fourth data set generated by the first Bayesian acquisition function.
20 . The system of claim 15 wherein:
the data sets include target output defined by an objective function as well as field data that is auxiliary to the target output; and
the first and second Bayesian acquisition functions rely upon the field data for generating input parameters.