Vehicle Control Model Updating for Actuator Failure Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In automatic driving systems, the gap between the real vehicle and its model during failures is difficult to reflect in real-time on low-priced electronic control units, making system identification complex and unstable.
Innovation Solution
The vehicle control device updates the vehicle model with a fixed value corresponding to detected failures and fixes the actuator command value, aligning the real vehicle's behavior with the model, allowing for simplified identification and continued operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If system identification is performed to reflect the real vehicle's failure state on the vehicle model, then control stability is improved, but device complexity and computational burden increase significantly
Solution Approach 1:
The patent pre-defines multiple failure patterns and their corresponding vehicle model configurations before actual failure occurs. When a failure is detected, the system simply selects and applies the pre-prepared failure pattern rather than performing complex real-time system identification. This preliminary preparation resolves the contradiction by trading offline computational effort for online simplicity, maintaining control stability without real-time identification complexity.
Solution Approach 2:
The patent changes the parameters of the vehicle model to match the detected failure pattern by selecting from pre-defined failure patterns. Instead of performing complex system identification to determine model parameters, the system directly adjusts parameters by switching to appropriate pre-configured failure patterns, thereby maintaining control stability while avoiding computational complexity.
2Measurement precision
If real-time system identification is performed on low-priced ECUs, then accurate vehicle model updating is achieved, but processing speed and reliability deteriorate
Solution Approach 1:
The patent pre-calculates and stores multiple failure patterns with their corresponding vehicle model parameters before runtime. During actual operation, the system only needs to select the appropriate pre-computed pattern based on detected failure conditions, avoiding complex real-time calculations on low-priced ECUs. This approach maintains measurement precision while ensuring fast real-time processing speed.
3Reliability
If complex system identification algorithms are implemented, then accurate failure state reflection is achieved, but ease of operation and implementation difficulty worsen
Solution Approach 1:
The patent implements complex system identification algorithms offline to generate pre-defined failure patterns, rather than executing them in real-time during operation. The online system only performs simple pattern selection based on detected failure conditions, dramatically improving ease of operation and implementation while maintaining accurate failure state reflection through the pre-computed patterns.
Data Source
AI summary
Provided is an automatic driving system based on a model predictive control, the automatic driving system where, in an event of a failure of an actuator, identification between a real vehicle and a vehicle model is simplified. Based on information regarding the failure of the actuator, the automatic driving system updates a spot in the vehicle model, the spot corresponding to the spot of failure detected, to a fixed value, and causes an actuator control device for the actuator, where the failure is detected, to fix a command value that is overwritten in accordance with a state of the actuator. With this configuration, the identification between the real vehicle and the vehicle model is simplified.