Occupant Injury Risk Prediction Using Two-Stage Collision Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for predicting collision severity oversimplify the occupant collision process, leading to low accuracy and inadequate decision-making in inevitable traffic accidents, as they rely solely on equivalent energy speed without considering various influencing factors.
Innovation Solution
A method involving two learning models is established to predict collision severity by inputting vehicle and collision scene data into a first model to generate a collision acceleration curve, and then using occupant and restraint system data in a second model to produce kinematics and dynamics curves of human body parts, thereby evaluating collision severity comprehensively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If traditional EES-based method is used to predict collision severity, then calculation is simple, but prediction accuracy is low due to oversimplification of occupant collision process
Solution Approach 1:
The prediction system is segmented into multiple independent modules: vehicle collision module (predicting vehicle acceleration), occupant motion module (predicting occupant kinematics), and injury risk module (calculating injury metrics). Each module handles a specific aspect of the collision process, allowing complex predictions to be broken down into manageable components while maintaining overall accuracy.
Solution Approach 2:
The patent introduces intermediate parameters as mediators between the simple EES calculation and the complex injury prediction. These include vehicle collision acceleration, occupant collision acceleration, and various kinematic parameters that bridge the gap between simple input data and complex injury outcomes, enabling accurate predictions without requiring direct complex calculations.
2Device complexity
If only EES is considered for collision severity prediction, then prediction process is simple, but important factors like occupant posture, restraint system, and vehicle configuration are ignored
Solution Approach 1:
The prediction system is designed with multi-functionality to handle various types of input data (vehicle parameters, occupant characteristics, restraint system configurations, road conditions) and generate comprehensive injury risk assessments. The same core prediction framework can accommodate different vehicle types, occupant demographics, and collision scenarios, making the system universally applicable while considering multiple factors simultaneously.
Solution Approach 2:
The patent combines multiple types of data (vehicle configuration data, occupant feature data, restraint system data, collision scene data) into a composite prediction model. This composite approach integrates diverse information sources to create a more reliable and comprehensive collision severity assessment than any single data type could provide alone.
3Measurement precision
If comprehensive factors are considered in collision severity prediction, then prediction accuracy improves, but calculation complexity and time increase
Solution Approach 1:
The system performs preliminary actions by pre-defining prediction models and parameter relationships before actual collision assessment. The framework establishes predetermined calculation procedures, parameter mappings, and prediction algorithms that can be quickly executed during real-time or post-collision analysis, reducing calculation time while maintaining comprehensive factor consideration.
Solution Approach 2:
The patent utilizes parameter changes to simplify calculations while maintaining accuracy. By transforming complex physical relationships into parameter-based predictions (e.g., using acceleration curves instead of full dynamic simulations, using injury metrics derived from simplified kinematic parameters), the system achieves accurate predictions with reduced computational burden and faster execution time.
Data Source
AI summary
Disclosed is a method for predicting collision severity, including: establishing a first learning model, and inputting vehicle data and collision accident scene feature data into the first learning model; obtaining a predicted collision acceleration curve outputted by the first learning model, the predicted collision acceleration curve being established based on a plane rectangular coordinate system; establishing a second learning model, and inputting the predicted collision acceleration curve, the occupant feature data and the restraint system feature data into the second learning model; obtaining a plurality of predicted collision kinematics and dynamics curves of human body parts outputted by the second learning model; generating a collision severity parameter according to the plurality of predicted collision kinematics and dynamics curves of the human body parts, the collision severity parameter being configured to evaluate the collision severity.