Friction Coefficient Estimation Using Regression Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the friction coefficient between a vehicle and a roadway are limited in accuracy and reliability, especially in dynamic conditions, and often require dedicated sensor systems, which are costly and not universally applicable.
Innovation Solution
A method and device using a time sequence-based statistical approach with sensor data processing via regression models, including linear regression algorithms, to estimate friction coefficients as probability distributions, leveraging swarm knowledge and reducing sensor errors, and enabling cloud-based estimation and prediction for vehicle control functions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If dedicated friction coefficient sensor systems are used, then measurement precision is improved, but device complexity and cost increase
Solution Approach 1:
The patent replaces dedicated mechanical friction coefficient sensors with a computational approach using existing vehicle sensors (acceleration, steering angle, brake pressure sensors) combined with regression models and neural networks to estimate friction coefficients, thereby eliminating complex dedicated sensing hardware while achieving comparable measurement precision
Solution Approach 2:
The patent introduces regression models and neural networks as intermediary computational layers that process data from existing vehicle sensors to derive friction coefficients, acting as a mediator between raw sensor data and the desired friction measurement without requiring direct physical contact sensors
2Measurement precision
If dedicated friction coefficient sensor systems are used, then measurement precision is improved, but setup costs increase
Solution Approach 1:
The patent makes existing vehicle sensors serve multiple functions - they are used for both standard vehicle control functions and for estimating friction coefficients, eliminating the need for dedicated friction sensors and reducing setup costs while maintaining measurement precision through multi-purpose data utilization
Solution Approach 2:
The patent enables the vehicle's existing sensor system to serve itself by processing its own operational data through regression models and neural networks to generate friction coefficient estimates, eliminating the need for external dedicated sensing systems and reducing overall system cost
3Reliability
If statistical approach with swarm knowledge is used, then reliability is improved, but loss of information increases due to data aggregation
Solution Approach 1:
The patent implements feedback mechanisms where the neural network continuously learns from aggregated data while providing individualized friction estimates, and where the system refines its models based on ongoing data collection, thereby maintaining reliability through iterative improvement while managing information loss through intelligent data processing
Data Source
AI summary
A method for determining a friction coefficient for a contact between a tire of a vehicle and a roadway. The method includes processing sensor signals in order to generate processed sensor signals. The sensor signals represent state data that are read in at least by at least one detection device and that are correlatable with the friction coefficient. The processed sensor signals represent at least one preliminary friction coefficient. The method also includes ascertaining the friction coefficient using the processed sensor signals and a regression model.