Vehicle-Grid Integration Behavior Prediction with Reliability Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing prediction models for vehicle behavior in vehicle-grid integration (VGI) are inaccurate, especially for vehicles without fixed behavior patterns, as they rely solely on past service record data, failing to account for variations and sudden changes in vehicle behavior.
Innovation Solution
A behavior prediction device and method that integrates both past behavior history and user-provided scheduled data, determining the final predicted value based on the higher reliability between model-predicted and user-provided values, with priority given to user-provided one-off schedules for sudden changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a prediction model based only on past service record data is used, then the device complexity is reduced, but the prediction accuracy deteriorates for vehicles without fixed behavior patterns
Solution Approach 1:
The patent merges two prediction approaches: (1) statistical prediction based on past behavior history, and (2) user-provided scheduled values. The system calculates both predicted values and selects the more reliable one based on calculated reliability metrics, thereby improving overall prediction accuracy while maintaining manageable system complexity through a structured comparison mechanism.
2Adaptability or versatility
If only past behavior history is used for prediction, then data collection is simpler, but the system cannot adapt to sudden changes in vehicle behavior
Solution Approach 1:
The system performs preliminary action by collecting and storing user-provided scheduled values in advance before they become actual behavior data. This allows the system to anticipate and adapt to planned behavior changes (such as scheduled vehicle usage for specific purposes) before they occur, improving adaptability while preventing information loss about intended vehicle usage patterns.
3Measurement precision
If user-provided scheduled values are always prioritized, then prediction accuracy for scheduled behaviors improves, but reliability calculation becomes more complex
Solution Approach 1:
The system applies parameter changes by dynamically adjusting the selection criteria between predicted values based on behavioral parameters (past history) and scheduled values (user input). Reliability is calculated as a quantitative parameter that determines which prediction source to trust, allowing the system to flexibly switch between data sources based on calculated reliability thresholds rather than using a fixed prioritization rule.
Data Source
AI summary
A device predicts the future behavior of a vehicle participating in vehicle-grid integration based on past behavior history of the vehicle, and acquires a user input value representing scheduled future behavior of the vehicle. A first integrated error is obtained by integrating an error between the predicted value and an actual value of the vehicle behavior during a predetermined period, and a second integrated error is obtained by integrating an error between the user input value and the actual value during the same period. The device determines that the user input value is a final predicted value of the future behavior of the vehicle when the first integrated error is smaller than the second integrated error, and determines that the predicted value is the final predicted value when the first integrated error is equal to or larger than the second integrated error.