Vehicle Control Device Neural Network Relearning Scheduling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The limited processing resources of a control device in a vehicle make it difficult to perform neural network relearning during operating periods, and even during parked periods, if the parked time is short, the relearning may not be completed suitably.
Innovation Solution
A control device with a parked period predicting part and a learning plan preparing part that predicts future parked periods and allocates the relearning process into multiple steps, allowing for efficient utilization of processing resources by scheduling relearning during predicted parked periods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If relearning of the neural network is performed during the operating period of the vehicle, then the precision of estimation can be improved, but the limited processing resources make it difficult to perform relearning in parallel with other control
Solution Approach 1:
The patent applies dynamics by making the relearning execution flexible and adaptive based on vehicle conditions. The control device dynamically determines whether to perform relearning during operating or parked periods based on available time and processing resource status, allowing the system to adapt its behavior to current conditions rather than following a fixed schedule
Solution Approach 2:
The patent applies preliminary action by performing relearning during parked periods before the next operating period begins. This allows the system to complete relearning tasks in advance when processing resources are more available, so that improved estimation precision is ready for use when the vehicle operates again
2Productivity
If relearning of the neural network is performed during parked periods, then processing resources can be utilized more effectively, but if the parked period is short, it is not possible to complete the relearning
Solution Approach 1:
The control device dynamically adjusts relearning execution based on the actual duration of parked periods. When a parked period is sufficient, relearning is performed; when it is too short, relearning is deferred or adjusted, ensuring completion quality while maintaining resource utilization efficiency
Solution Approach 2:
The patent changes the parameter of relearning execution timing based on parked period duration. Instead of fixed scheduling, the system varies when and how relearning is performed according to the available time window, optimizing between completion success and resource utilization
3Measurement precision
If relearning of the neural network is performed during parked periods, then the precision of estimation can be improved, but a large amount of time becomes required for relearning
Solution Approach 1:
The patent performs relearning during parked periods before the next operating period, utilizing otherwise idle time. This preliminary action ensures that relearning completes before precision improvements are needed for operation, transforming wasted parked time into productive relearning time without extending total vehicle downtime
Solution Approach 2:
The system dynamically schedules relearning to occur during naturally occurring parked periods rather than requiring dedicated time allocation. This allows the system to perform time-intensive relearning tasks using existing idle periods in the vehicle's operation cycle
Data Source
AI summary
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.