Learning device, learning method, and program therefor

1. A control system comprises:

a learning device;

a controller which outputs a command value obtained by compensating a target value based on a compensation amount; and

a control object controlled to perform a predetermined process on an object to be processed, wherein the command value output by the controller is input to the control object, and the control object outputs a control variable as a response to the command value,

wherein the learning device provides, to an adjuster including a learned model learned to output the compensation amount to the controller based on a specific parameter of the object to be processed, the learned model, and

wherein the learning device operates to:

obtain operation data including the target value, the command value, and the control variable to evaluate the quality of the control variable by determining whether the control variable is within a predetermined allowable range;

generate a candidate compensation amount based on the operation data, performs learning with the generated candidate compensation amount and the specific parameter of the object to be processed as teacher data, and generates a learned model; and

provide the learned model to the adjuster when the control variable evaluated is within the predetermined allowable range and when the command value obtained by compensating the target value based on the compensation amount output by the generated learned model is imparted to the control object.

2. The learning device according to claim 1 , wherein when the specific parameter of the object to be processed provided to the control object is equal to a parameter, whose evaluation has not been performed yet, of the generated learned model, wherein the learning device outputs a compensation amount output by the generated learned model to the controller and to evaluate the quality.

3. The learning device according to claim 2 , wherein the learning device performs learning again when the quality evaluated based on the compensation amount output by the generated learned model is out of the predetermined allowable range, and regenerates the learned model.

4. The learning device according to claim 1 , wherein the learning device generates the candidate compensation amount by data-driven control.

5. The learning device according to claim 4 , wherein the data-driven control is any one of virtual reference feedback tuning (VRFT), fictitious reference iterative tuning (FRIT), and estimated response iterative tuning (ERIT).

6. A learning method executed to operate a learning device in a control system, the learning method comprises:

outputting, by a controller, a command value obtained by compensating a target value based on a compensation amount; and

controlling a control object to perform a predetermined process on an object to be processed, wherein the command value output by the controller is input to the control object, and the control object outputs a control variable as a response to the command value,

wherein the learning device provides, to an adjuster including a learned model learned to output the compensation amount to the controller based on a specific parameter of the object to be processed, the learned model, and

wherein the learning method further comprises operating the learning device to perform:

obtaining operation data including the target value by determining whether the control variable is within a predetermined allowable range, the command value, and the control variable to evaluate the quality of the control variable;

generating a candidate compensation amount based on the operation data, performing learning with the generated candidate compensation amount and the specific parameter of the object to be processed as teacher data, and generating a learned model; and

providing the learned model to the adjuster when the control variable evaluated is within the predetermined allowable range and when the command value obtained by compensating the target value based on the compensation amount output by the generated learned model is imparted to the control object.

7. A non-transitory computer-readable recording medium storing a program which, when executed by a processor causes the processor to perform a learning method to operate a learning device in a control system, the learning method comprises:

outputting, by a controller, a command value obtained by compensating a target value based on a compensation amount; and

controlling a control object to perform a predetermined process on an object to be processed, wherein the command value output by the controller is input to the control object, and the control object outputs a control variable as a response to the command value,

wherein the learning device provides, to an adjuster including a learned model learned to output the compensation amount to the controller based on a specific parameter of the object to be processed, the learned model, and

wherein the learning method further comprises operating the learning device to perform:

obtaining operation data including the target value, the command value, and the control variable to evaluate the quality of the control variable by determining whether the control variable is within a predetermined allowable range;

generating a candidate compensation amount based on the operation data, performing learning with the generated candidate compensation amount and the specific parameter of the object to be processed as teacher data, and generating a learned model; and

providing the learned model to the adjuster when the control variable evaluated is within the predetermined allowable range and when the command value obtained by compensating the target value based on the compensation amount output by the generated learned model is imparted to the control object.