Physical system simulation
A device includes one or more processors configured to generate simulation data by providing first sets of first values of first input parameters to a simulation of a physical system. The processors are configured to develop at least one surrogate model based on the simulation data, and to select second input parameters from the first input parameters based on the at least one surrogate model. A count of the second input parameters is less than a count of the first input parameters. The processors are configured to generate second simulation data by providing second sets of second values of the second input parameters to the simulation. The processors are configured to develop at least a second surrogate model based on the second simulation data.
1. A device for physical system simulation, the device comprising:
a memory configured to store a simulation of a physical system; and
one or more processors configured to:
generate initial simulation data by providing first sets of first values of first input parameters to the simulation, each set of the first sets indicating a first value of each of the first input parameters;
develop at least one surrogate model based on the initial simulation data;
select second input parameters from the first input parameters based on the at least one surrogate model, wherein a count of the second input parameters is less than a count of the first input parameters;
generate second simulation data by providing second sets of second values of the second input parameters to the simulation, each set of the second sets indicating a second value of each of the second input parameters, wherein second particular input parameters are predetermined, and wherein the second simulation data is further based on second particular values of the second particular input parameters;
develop at least a second surrogate model based on the second simulation data;
receive sensor data during testing of the physical system from one or more sensors configured to monitor the physical system, wherein at least one of the one or more sensors is proximate to the physical system; and
update the simulation based on the sensor data.
2. The device of claim 1 , wherein the second surrogate model includes at least one of a Kriging model, a least squares analysis of variance (ANOVA) model, a heuristic model, or a metaheuristic model.
3. The device of claim 1 , wherein the one or more processors are further configured to perform adjustment of the second surrogate model until an adjustment criterion is met.
4. The device of claim 3 , wherein the one or more processors are configured to determine that the adjustment criterion is met in response to detecting expiration of a time period, detecting a convergence of the second surrogate model, or both.
5. The device of claim 1 , wherein the one or more processors are configured to perform a particular adjustment of the second surrogate model by:
selecting particular input parameters based on the second surrogate model; and
generating particular simulation data by providing particular sets of particular values of the particular input parameters to the simulation, each set of the particular sets indicating a particular value of each of the particular input parameters.
6. A method of physical system simulation, the method comprising:
generating initial simulation data by providing first sets of first values of first input parameters to a simulation of a physical system, each set of the first sets indicating a first value of each of the first input parameters;
developing at least one surrogate model based on the initial simulation data;
selecting second input parameters from the first input parameters based on the at least one surrogate model, wherein a count of the second input parameters is less than a count of the first input parameters;
generating second simulation data by providing second sets of second values of the second input parameters to the simulation, each set of the second sets indicating a second value of each of the second input parameters, wherein second particular input parameters are predetermined, and wherein the second simulation data is further based on second particular values of the second particular input parameters;
developing at least a second surrogate model based on the second simulation data;
receiving sensor data during testing of the physical system from one or more sensors configured to monitor the physical system, wherein at least one of the one or more sensors is proximate to the physical system; and
updating the simulation based on the sensor data.
7. The method of claim 6 , wherein the physical system includes a large-scale complex engineering system (LSCES).
8. The method of claim 6 , further comprising using the second surrogate model to determine one or more test parameters of a test of the physical system.
9. The method of claim 8 , wherein the test is expensive, destructive to the physical system, or both.
10. The method of claim 6 , wherein at least one of the sensors is coupled to the physical system.
11. The method of claim 6 , wherein a first count of the first sets is less than a second count of the second sets.
12. The method of claim 6 , further comprising generating the first sets by using orthogonal array sampling, Latin hypercube sampling, or random sampling.
13. The method of claim 6 , wherein the simulation of the physical system is based on finite element modeling, finite volume modeling, eigenmode decomposition analysis, eigenvalue analysis, eigenstate analysis, iterative first-principle modeling, or a combination thereof.
14. The method of claim 6 , further comprising:
developing an elastic net surrogate model based on the initial simulation data, the at least one surrogate model including the elastic net surrogate model; and
selecting a group of input parameters from the first input parameters based on the elastic net surrogate model.
15. The method of claim 14 , further comprising:
developing one or more linear models based on the group of input parameters, the at least one surrogate model including the one or more linear models; and
selecting at least one of the second input parameters from the group of input parameters based on the one or more linear models.
16. The method of claim 15 , wherein the second input parameters are further based on user input, a configuration setting, default data, or a combination thereof.
17. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to:
generate initial simulation data by providing first sets of first values of first input parameters to a simulation of a physical system, each set of the first sets indicating a first value of each of the first input parameters;
develop at least one surrogate model based on the initial simulation data;
select second input parameters from the first input parameters based on the at least one surrogate model, wherein a count of the second input parameters is less than a count of the first input parameters;
generate second simulation data by providing second sets of second values of the second input parameters to the simulation, each set of the second sets indicating a second value of each of the second input parameters, wherein second particular input parameters are predetermined, and wherein the second simulation data is further based on second particular values of the second particular input parameters;
develop at least a second surrogate model based on the second simulation data;
receive sensor data during testing of the physical system from one or more sensors configured to monitor the physical system, wherein at least one of the one or more sensors is proximate to the physical system; and
update the simulation based on the sensor data.
18. The non-transitory computer-readable medium of claim 17 , wherein the instructions, when executed by the one or more processors, further cause the one or more processors to use the second surrogate model to determine one or more test parameters of a test of the physical system.
19. The non-transitory computer-readable medium of claim 17 , wherein the one or more processors are configured to:
develop an elastic net surrogate model based on the initial simulation data, the at least one surrogate model including the elastic net surrogate model; and
select a group of input parameters from the first input parameters based on the elastic net surrogate model.
20. The non-transitory computer-readable medium of claim 19 , wherein the one or more processors are configured to:
develop one or more linear models based on the group of input parameters, the at least one surrogate model including the one or more linear models; and
select at least one of the second input parameters from the group of input parameters based on the one or more linear models.
21. The non-transitory computer-readable medium of claim 20 , wherein the second input parameters are further based on user input, a configuration setting, default data, or a combination thereof.