Control device of internal combustion engine and control method of same and learning model for controlling internal combustion engine and learning method of same

Patent metadata

Field	Value
Publication number	US-10947909-B2
Application number	US-201916566019-A
Country	US
Kind code	B2
Filing date	Sep 10, 2019
Priority date	Oct 17, 2018
Publication date	Mar 16, 2021
Grant date	Mar 16, 2021

Abstract

A control device of an internal combustion engine is configured to output a predicted value of an output parameter by using a learning model if actually measured values of input parameters are input, control the internal combustion engine based on the predicted value of the output parameter, learn the learning model by using a gradient method and by using a combination of actually measured values of the input parameters and an actually measured value of the output parameter as teacher data, and adjust the learning rate so that the learning is performed by a smaller learning rate when an amount of noise superposed on an actually measured value of at least one parameter among the input parameters and the output parameter is relatively large compared with when the amount of noise superposed on the actually measured value of the parameter is relatively small.

First claim

The invention claimed is:

1. A control device of an internal combustion engine for controlling an operation of an internal combustion engine of a vehicle, the control device of an internal combustion engine being configured to:
output a predicted value of an output parameter by using a learning model if actually measured values of input parameters are input,
control the internal combustion engine based on the predicted value of the output parameter output from the parameter output part,
learn the learning model by using a gradient method and by using a combination of actually measured values of the input parameters of the learning model and an actually measured value of the output parameter of the learning model as teacher data, and
adjust a learning rate so that the learning is performed by a smaller learning rate when an amount of noise superposed on an actually measured value of at least one parameter among the input parameters and the output parameter is relatively large compared with when the amount of noise superposed on the actually measured value of the parameter is relatively small.

2. The control device of an internal combustion engine according to claim 1 , the control device of an internal combustion engine is configured to detect the amounts of noise superposed on the actually measured values of the input parameters and the actually measured value of the output parameter, and the control device of an internal combustion engine is configured to calculate a standard deviation or a reciprocal of an SN ratio of each parameter based on the actually measured values of the input parameters and output parameter detected when the engine operating state is a steady state and using the standard deviation or the reciprocal of the SN ratio of each parameter as the amount of noise of the parameter.

3. The control device of an internal combustion engine according to claim 1 , wherein the control device of an internal combustion engine is configured to adjust the learning rate so that the learning rate becomes smaller the larger the amount of noise superposed on the actually measured value of at least one parameter.

4. The control device of an internal combustion engine according to claim 1 , wherein when the amounts of noise superposed on the actually measured values of the input parameters and the output parameter are equal to or less than threshold values set for the parameters, the control device of an internal combustion engine is configured to adjust the learning rate so as to be maintained at a constant learning rate larger than the learning rate when the amount of noise superposed on the actually measured value of at least one parameter is larger than the threshold value.

5. The control device of an internal combustion engine according to claim 1 , wherein the control device of an internal combustion engine is an electronic control unit provided at the vehicle.

6. The control device of an internal combustion engine according to claim 1 , wherein
the control device of an internal combustion engine comprises an electronic control unit provided at the vehicle and a server installed at an outside of the vehicle and configured to be able to communicate with the electronic control unit,
the electronic control unit is configured to send the teacher data to the server,
the server is configured to adjust a learning rate based on the amount of noise superposed on the actually measured values of the input parameters and the actually measured value of the output parameter included in the teacher data,
the server is configured to learn the learning model by using the teacher data received from the electronic control unit and sends the learning model after learning to the electronic control unit, and
the electronic control unit is configured to use the learning model after learning sent from the server to the electronic control unit.

7. A control method of an internal combustion engine using a learning model to control an operation of the internal combustion engine of a vehicle,
the control method of an internal combustion engine cause a processor to perform steps of:
adjusting a learning rate so as to become smaller when the amount of noise superposed on the actually measured value of at least one parameter among the input parameters of the learning model and the output parameter of the learning model is relatively large compared with when the amount of noise superposed on the actually measured value of the parameter is relatively small,
learning of the learning model by a gradient method by using the adjusted learning rate and by using a combination of the actually measured values of the input parameters and the actually measured value of the output parameter as teacher data,
outputting a predicted value of the output parameter by using the learning model if the actually measured values of the input parameters are input, and
controlling the internal combustion engine based on the output predicted value of the output parameter.

8. A learning model for controlling an internal combustion engine, the leaning model causes a processor to function so as to output a predicted value of an output parameter used for control of an internal combustion engine if actually measured values of input parameters are input,
wherein the learning model is learned by a gradient method using a combination of actually measured values of the input parameters and an actually measured value of the output parameter as teacher data, and,
in the learning, a learning rate adjusted to become a smaller learning rate is used when the amount of noise superposed on the actually measured value of at least one parameter among the input parameters and the output parameter is relatively large compared with when the amount of noise superposed on the actually measured value of that parameter is relatively small.

9. A learning method of a learning model for controlling an internal combustion engine, the learning method causes a processor to function so as to output a predicted value of an output parameter used for control of an internal combustion engine if actually measured values of input parameters are input,
wherein the learning model is learned by a gradient method using a combination of actually measured values of the input parameters and an actually measured value of the output parameter as teacher data, and,
in the learning, a learning rate adjusted to become a smaller learning rate is used when the amount of noise superposed on the actually measured value of at least one parameter among the input parameters and the output parameter is relatively large compared with when the amount of noise superposed on the actually measured value of that parameter is relatively small.

Assignees

• Toyota Motor Co Ltd

Inventors

• Yokoyama Daiki
• Kitagawa Eiki
• Oda Tomihisa

Classifications

• G05B13/027

  using neural networks only · CPC title

• F02D41/1405 Primary

  Neural network control · CPC title

• F02D28/00 Primary

  Program control of engines · CPC title

• F02D2200/1004

  Estimation of the output torque · CPC title

• F02D41/2451

  characterised by what is learned or calibrated · CPC title