Vehicle Collision Classification via Time-Shifted Sensor Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current airbag control systems struggle to classify collisions accurately due to changes in vehicle structures, leading to suboptimal restraint device control, as existing algorithms rely on synchronized sensor signals that fail to differentiate between triggering and non-triggering collisions effectively.
Innovation Solution
The method involves determining two feature characteristic curves with a time shift, correlating them to create a correlation profile, and comparing this profile to a separation characteristic to classify collisions, allowing for improved discrimination between triggering and non-triggering events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If synchronized sensor signals are used for collision classification, then the algorithm complexity is reduced, but the collision discrimination accuracy deteriorates due to inability to differentiate between triggering and non-triggering collisions
Solution Approach 1:
The patent applies preliminary action by shifting the timing of sensor signal processing in advance. Specifically, the acceleration signal is shifted by a first time period and the structure-borne noise signal is shifted by a second time period before correlation analysis. This time-shifting allows the algorithm to capture characteristic collision processes at different stages, improving discrimination accuracy between triggering and non-triggering collisions while maintaining algorithmic feasibility
Solution Approach 2:
The patent introduces a time-dimension shift to the traditional synchronized signal analysis. By processing sensor signals at different time points and creating time-shifted characteristic curves, the method adds a temporal dimension to the feature space. This dimensional expansion enables better separation of collision types that appear similar in synchronized analysis, thereby improving measurement precision without excessive complexity increase
2Device complexity
If traditional core algorithms based on central sensor are used, then the device structure is simple, but the separation performance deteriorates due to changes in vehicle structures
Solution Approach 1:
The patent introduces structure-borne noise sensors as intermediary elements between the collision event and the control decision. These sensors capture vibration and acoustic information from the vehicle structure, providing additional intermediate features that enhance the separation performance. The structure-borne noise acts as a mediator that reveals collision characteristics not visible to acceleration sensors alone, improving reliability while maintaining relatively simple device structure
Solution Approach 2:
The patent changes the parameters used for collision classification by incorporating structure-borne noise features alongside acceleration signals. Instead of relying solely on acceleration magnitude, the method uses correlation analysis of time-shifted acceleration and structure-borne noise signals. This parameter transformation enables reliable separation of collision types even with simple device structure, as the combined features provide better discrimination capability
3Device complexity
If feature values are related at the same point in time, then the correlation analysis is simple, but the identification of characteristic collision processes deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-shifting the sensor signals in time before correlation analysis. The acceleration signal is shifted by a first time period and the structure-borne noise signal by a second time period, allowing the correlation algorithm to capture relationships between features that occur at different moments during the collision. This preserves characteristic collision process information that would be lost in synchronous analysis
Solution Approach 2:
The patent introduces dynamic time-shifting to the correlation analysis. Instead of using fixed synchronous time points, the method dynamically adjusts the time offsets for different signal pairs. The first and second time periods can be optimized based on the specific collision type and sensor characteristics, allowing the system to adaptively capture the dynamic temporal relationships in different collision scenarios, thereby reducing information loss
Data Source
Figure 1~2
Figure 3A~3B
Figure 4A~4B
AI summary
The method involves determining a characteristic curve that represents time course of a characteristic over a collision course. Another characteristic curve is determined for representing time course of another characteristic over the collision course, where the time course is adjusted around a predetermined time interval. The characteristic curves are provided based on multiple sensor signals, and the time course and a predetermined time course of the characteristics are set in reference to each other. Correlation intervals are compared with separated characteristics to classify collision. Independent claims are also included for the following: (1) a controlling device for executing a method for classifying a collision course of a vehicle (2) a computer program product with a computer program having a set of instructions for executing a method for classifying a collision course of a vehicle.