# US 2024/0202540 A1 · Setting device, setting method, and recording medium

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2024/0202540 A1
- Document ID: US20240202540A1
- Status: Published application
- Kind: A1
- Application number: 18/554,336
- Filed: 2021-04-20
- Published: 2024-06-20
- Earliest claimed: 2021-04-20
- Assignees: NEC Corporation
- Applicants: NEC Corporation
- Inventors: Shumpei KUBOSAWA; Takashi ONISHI
- Web page: https://patentgrep.com/patent/US20240202540A1

## CPC

- G06N 3/092: Reinforcement learning
- G05B 13/02: electric
- G05B 13/04: involving the use of models or simulators
- G06F 30/20: Design optimisation, verification or simulation (optimisation, verification or simulation of circuit designs G06F30/30)
- G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
- G06N 20/00: Machine learning
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G05B: CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS (systems for controlling or regulating non-electric variables G05D; systems for regulating electric or magnetic variables G05F; control devices or systems insofar as characterised by mechanical features only G05G)
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A setting device sets a parameter value of a simulator that simulates a simulation subject. The setting device calculates weights of respective output items based on an output value of the simulator and a reference output value. The reference output value is subject to simulation according to the output value of the simulator. The setting device controls, by using an evaluation function, learning of a setting of a parameter value of the simulator. The evaluation function is based on the output value of the simulator, the reference output value, and the weights.

## Claims

### Claim 1

A setting device comprising:
a memory configured to store instructions; and
a processor configured to execute the instructions to:
set a parameter value of a simulator that simulates a simulation subject;
calculate weights of respective output items based on an output value of the simulator and a reference output value, the reference output value being subject to simulation according to the output value of the simulator; and
control, by using an evaluation function, learning of a setting of a parameter value of the simulator, the evaluation function being based on the output value of the simulator, the reference output value, and the weights.

### Claim 2 (depends on 1)

The setting device according to claim 1, wherein the processor is configured to execute the instructions to calculate the weights of the respective output items such that the output item having a larger normalized error obtained by normalizing errors of the respective output items has a smaller weight.

### Claim 3 (depends on 2)

The setting device according to claim 2, wherein the processor is configured to execute the instructions to calculate, for each unit of the learning, an error index value indicating magnitude of a weighted error between the output value of the simulator and the reference output value for the entire output items, and when the error index value after executing one unit of learning indicates that an error is larger than the error index value before executing the one unit of learning, set the weight back to the weight before executing the one unit of learning.

### Claim 4 (depends on 1)

The setting device according to claim 1,
wherein the processor is configured to execute the instructions to calculate, for each output item of the simulator, a gradient vector indicating the gradient of the error between the simulator output value and the reference output value in the parameter value space, and
wherein the processor is configured to execute the instructions to calculate the weights for the respective output items such that the weight increases as a difference between the gradient vector and an average gradient vector decreases, the average gradient vector being obtained by averaging the gradient vectors for all the output items.

### Claim 5 (depends on 4)

The setting device according to claim 4, wherein the processor is configured to execute the instructions to calculate the gradient vector based on an amount of change in the error when the parameter value is randomly changed.

### Claim 6 (depends on 1)

The setting device according to claim 1, wherein the processor is configured to execute the instructions to use the evaluation function based on the output value of the simulator at a time in simulation of the simulation subject performed by the simulator, the time including a time in future after time at which the parameter value is set.

### Claim 7 (depends on 1)

The setting device according to, claim 1,
wherein the reference output value is an output value of a second simulator that simulates the simulation target, and
the processor is configured to execute the instructions to control learning of the setting of the parameter value of the simulator, by using an evaluation function based on an output value of a first simulator obtained when an input value that is same as an input value to the second simulator is entered to the first simulator, the reference output value, and the weights, the first simulator being the simulator for which the parameter value setting means sets the parameter value.

### Claim 8 (depends on 7)

The setting device according to claim 7,
wherein the processor is configured to execute the instructions to change a parameter value of the second simulator, and
wherein the reference output value includes an output value of the second simulator when the parameter value of the second simulator is changed.

### Claim 9 (depends on 8)

The setting device according to claim 8,
wherein the processor is configured to execute the instructions to control, by using a second evaluation function, learning of a setting of the parameter value of the second simulator so as to increase a difference between the output value of the first simulator and the reference output value, the second evaluation function being based on an output value of the first simulator obtained when an input value that is same as an input value to the second simulator is entered to the first simulator, and the reference output value, and
wherein the processor is configured to execute the instructions to control the learning of the setting of the parameter value of the first simulator so as to reduce a difference between the output value of the first simulator and the reference output value.

### Claim 10 (depends on 9)

The setting device according to claim 9,
wherein the processor is configured to execute the instructions to perform scaling to increase a value of the second evaluation function when the learning progresses beyond a predetermined condition.

### Claim 11 (depends on 8)

The setting device according to claim 8,
wherein the processor is configured to execute the instructions to set, to the second simulator, a sensor fault in a simulation model.

### Claim 12 (depends on 11)

The setting device according to claim 11, wherein the processor is configured to execute the instructions to set the weight for an output value of the sensor for which a fault is set, to a value indicating exclusion from a calculation subject of the evaluation function.

### Claim 13

A setting device comprising:
a memory configured to store instructions; and
a processor configured to execute the instructions to:
set a parameter value of a first simulator that simulates a simulation subject; and
control, by using an evaluation function, learning of a setting of the parameter value of the first simulator, the evaluation function being based on an output value of a second simulator that simulates the simulation subject, and an output value of the first simulator obtained when an input value that is same as an input value to the second simulator is entered to the first simulator.

### Claim 14 (depends on 13)

The setting device according to claim 13,
wherein the processor is configured to execute the instructions to change a parameter value of the second simulator, and
wherein the output value of the second simulator includes an output value when the parameter value of the second simulator is changed.

### Claim 15 (depends on 14)

The setting device according to claim 14,
wherein the processor is configured to execute the instructions to control, by using a second evaluation function, learning of a setting of the parameter value of the second simulator so as to increase a difference between the output value of the first simulator and the output value of the second simulator, the second evaluation function being based on an output value of the first simulator obtained when an input value that is same as an input value to the second simulator is entered to the first simulator, and the output value of the second simulator,
wherein the processor is configured to execute the instructions to control the learning of the setting of the parameter value of the first simulator so as to reduce a difference between the output value of the first simulator and the output value of the second simulator.

### Claim 16 (depends on 15)

The setting device according to claim 15,
wherein the processor is configured to execute the instructions to perform scaling to increase a value of the second evaluation function when the learning progresses beyond a predetermined condition.

### Claim 17 (depends on 14)

The setting device according to
claim 14,
wherein the processor is configured to execute the instructions to set, to the second simulator, a sensor fault in a simulation model.

### Claim 18 (depends on 17)

The setting device according to claim 17, wherein the processor is configured to execute the instructions to use the evaluation function in which an output value of the sensor for which fault setting is performed is excluded from a reference subject in calculation of the evaluation function.

### Claim 19

A setting method executed by a computer, comprising:
setting a parameter value of a simulator that simulates a simulation subject;
calculating weights of respective output items based on an output value of the simulator and a reference output value, the reference output value being subject to simulation according to the output value of the simulator; and
controlling, by using an evaluation function, learning of a setting of a parameter value of the simulator, the evaluation function being based on the output value of the simulator, the reference output value, and the weights.

### Claim at position 20

-22. (canceled)
