CONTROL DEVICE

• US 20200134452A1
• Filed: 10/23/2019
• Published: 04/30/2020
• Est. Priority Date: 10/30/2018
• Status: Active Grant

• Current Assignee: Toyota Jidosha Kabushiki Kaisha (Toyota Motor Corporation)
• Original Assignee: Toyota Jidosha Kabushiki Kaisha (Toyota Motor Corporation)
• Inventors: KOBUNA, Shunsuke, KITAGAWA, Eiki, YOKOYAMA, Daiki
• Granted Patent: US 11,436,488 B2
• Time in Patent Office: Days
• CPC Class Codes: B60W 20/11 using model predictive control [MPC] strategies, i.e. control methods based on models predicting performance utilising navigation and traffic information in the control strategy B60W20/12
• CPC Class Codes: B60W 2050/0005 Processor details or data handling, e.g. memory registers or chip architecture
• CPC Class Codes: B60W 2050/0014 Adaptive controllers
• CPC Class Codes: B60W 2050/0028 Mathematical models, e.g. for simulation
• CPC Class Codes: B60W 2050/0039 of the propulsion unit
• CPC Class Codes: B60W 2050/0075 Automatic parameter input, automatic initialising or calibrating means
• CPC Class Codes: B60W 2050/065 by reducing the computational load on the digital processor of the control computer
• CPC Class Codes: B60W 2556/10 Historical data
• CPC Class Codes: B60W 40/12 related to parameters of the vehicle itself , e.g. tyre models
• CPC Class Codes: B60W 50/0097 Predicting future conditions
• CPC Class Codes: F02D 2041/1412 using a predictive controller
• CPC Class Codes: F02D 29/02 peculiar to engines driving vehicles; peculiar to engines driving variable pitch propellers
• CPC Class Codes: F02D 41/042 for stopping the engine
• CPC Class Codes: F02D 41/1405 Neural network control
• CPC Class Codes: F02D 41/1461 of the exhaust gases emitted by the engine
• CPC Class Codes: F02D 41/2441 characterised by the learning conditions
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/084 Backpropagation, e.g. using gradient descent
• CPC Class Codes: G08G 1/0129 for creating historical data or processing based on historical data
• CPC Class Codes: G08G 1/143 inside the vehicles
• CPC Class Codes: G08G 1/148 : Management of a network of parking areas

Abstract

A control device mounted in a vehicle in which at least one controlled part is controlled based on an output parameter obtained by inputting input parameters to a learned model using a neural network, provided with a parked period predicting part predicting future parked periods of the vehicle and a learning plan preparing part preparing a learning plan for performing relearning of the learned model during the future parked periods based on results of prediction of the future parked periods.

Claims

1. A control device mounted in a vehicle in which at least one controlled part is controlled based on an output parameter obtained by inputting input parameters to a learned model using a neural network,

the control device comprising:

a parked period predicting part configured to predict future parked periods of the vehicle; and

a learning plan preparing part configured to prepare a learning plan for performing relearning of the learned model during the future parked periods based on results of prediction of the future parked periods.

2. The control device according to claim 1, wherein

the learning plan preparing part is configured to:

calculate lengths of time of the future parked periods based on the results of prediction of the future parked periods; and

prepare the learning plan so that relearning of the learned model is performed at the future parked periods having lengths of time enabling relearning of the learned model to be completed.

3. The control device according to claim 1, wherein

the process required for relearning of the learned model is divided in advance into a plurality of steps, and

the learning plan preparing part is configured to prepare the learning plan by successively allocating steps able to be completed within the parked period from an upstream step in the plurality of steps to the future parked periods in order from the closest one among the future parked periods based on the lengths of time of the future parked periods and the processing time for each of the plurality of steps.

4. The control device according to claim 3, wherein

the control device is further configured to:

calculate lengths of time of the future parked periods based on the results of prediction of the future parked periods; and

calculate a processing time for each of the plurality of steps based on an amount of data used at the time of relearning of the learned model.

5. The control device according to claim 1, wherein

the control device further comprises a learning part configured so as to perform relearning of the learned model in accordance with the learning plan.

6. The control device according to claim 3, wherein

the control device further comprises a learning part configured so as to perform relearning of the learned model in accordance with the learning plan, and

the learning part is configured to:

judge which parked period of the future parked periods predicted by the parked period predicting part the current parked period is; and

perform the processing of a step if there is a step allocated by the learning plan preparing part to the current parked period.

7. The control device according to claim 3, wherein

one step among the plurality of steps is a relearning step actually performing relearning of the learned model using training data, and

the learning plan preparing part is configured to allocate the relearning step to a parked period with a length of time of the future parked period of a predetermined length or more.