Vehicle Machine Learning Device Power Management for Disaster Response
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In vehicles with large battery capacities, such as plug-in hybrid vehicles, the electric power required for training machine learning models can deplete the available power supply during disasters, making it difficult to secure enough power for external distribution.
Innovation Solution
A machine learning device that reduces power consumption during training by acquiring disaster information and adjusting processing based on vehicle position and anticipated power supply needs, including stopping training processes or reducing their frequency, and obtaining driver permission for power conservation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If processing relating to training of a machine learning model is performed in a vehicle, then the machine learning model can be trained to improve vehicle functions, but the amount of electric power consumed in the vehicle increases and the electric power required at the time of a disaster is liable to be unable to be secured
Solution Approach 1:
The system dynamically adjusts the training process based on real-time power availability and disaster risk assessment. The processor modifies training parameters, data processing frequency, and computational intensity according to the vehicle's power state and environmental conditions, transitioning between high-performance training modes and power-conservation modes as needed
Solution Approach 2:
The system changes key training parameters such as batch size, learning rate, data sampling frequency, and computational precision based on power availability. When power constraints are detected, the system reduces these parameters to maintain training effectiveness while minimizing energy consumption
2Use of energy by moving object
If the processor stops the processing relating to training to reduce power consumption, then the electric power available for external supply during disasters is increased, but the machine learning model training progress is halted
Solution Approach 1:
The system performs preliminary actions by pre-processing training data, pre-calculating model parameters, and preparing training batches in advance when power availability is sufficient. This allows the actual training execution to be minimized or interrupted during power-constrained periods without significant loss of training progress
Solution Approach 2:
The system implements periodic training cycles where training operations are executed intermittently rather than continuously. The processor alternates between training phases and power-conservation phases, scheduling training tasks during periods of adequate power availability while maintaining training momentum through periodic progress
3Reliability
If the processor acquires disaster information and position information to determine power supply needs, then the ability to respond to disaster situations is improved, but the electric power consumption for data acquisition and processing increases
Solution Approach 1:
The system uses the vehicle's existing sensors, communication devices, and processor to acquire and process disaster information without requiring additional dedicated hardware or significant extra power. The same computational resources used for normal vehicle operations and training are leveraged for disaster detection and response
Data Source
AI summary
A machine learning device is provided in a vehicle able to supply electric power to an outside, and includes a processor configured to perform processing relating to training a machine learning model used in the vehicle. The processor is configured to lower an electric power consumption amount in the processing relating to training when acquiring disaster information compared with when not acquiring the disaster information.