Clutch Torque-Stroke Learning Convergence Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional clutch torque-stroke learning methods fail to ensure convergence of torque-stroke curves across different torque regions, leading to issues like slip and shift shock due to the lack of reliable data for low-torque regions when learning from high-torque regions.
Innovation Solution
A method that divides the torque region into multiple areas, determines learning direction based on torque variation, assesses reliability through temperature and time differences, and updates learning points to ensure convergence by using reliable data from adjacent regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If the T-S curve is learned starting from a high-torque region up to a low-torque region, then the learning process is simplified, but the prestored value for the low-torque region is not utilized causing the learned curve to diverge from the actual T-S curve
Solution Approach 1:
The patent applies preliminary action by determining the learning direction (high-to-low or low-to-high torque regions) based on prestored T-S curve data from adjacent regions. This preliminary determination ensures that reliable prestored values are utilized during the learning process, preventing curve divergence while maintaining a simplified learning procedure. The control unit selects the optimal learning direction before execution, incorporating advance preparation of reliability assessment.
Solution Approach 2:
The patent applies inversion by allowing the learning direction to be reversed based on reliability conditions. Instead of always learning from high-to-low torque regions, the system can invert the direction to low-to-high when prestored data indicates higher reliability in that direction. This inversion principle enables the system to adapt the learning sequence to maximize accuracy while maintaining process simplicity.
2Adaptability or versatility
If the T-S curve is continuously learned for multiple torque regions, then comprehensive coverage is achieved, but the curves from different regions do not converge to each other causing slip and shift shock
Solution Approach 1:
The patent applies feedback by continuously monitoring the reliability of prestored T-S curve data from adjacent torque regions during the learning process. The control unit uses this feedback to determine whether to learn from high-to-low or low-to-high torque regions, ensuring that the learned curve converges with previously learned curves. This feedback mechanism prevents divergence and maintains stability across multiple torque regions while achieving comprehensive coverage.
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting the learning direction parameter based on reliability assessments of prestored data. When learning across multiple torque regions, the system changes the learning direction parameter (high-to-low or low-to-high) to optimize curve convergence. This parameter adjustment ensures that T-S curves from different regions converge properly, preventing slip and shift shock while maintaining comprehensive torque region coverage.
3Ease of manufacture
If the learning direction is fixed, then the process is simple to implement, but it cannot adapt to different reliability conditions in various torque regions
Solution Approach 1:
The patent applies dynamics by making the learning direction adaptive rather than fixed. The control unit dynamically determines the learning direction (high-to-low or low-to-high torque regions) based on real-time reliability assessments of prestored T-S curve data. This dynamic adjustment maintains implementation simplicity through automated decision-making while providing adaptability to different reliability conditions in various torque regions, eliminating the need for complex manual configuration.
Data Source
AI summary
Provided are a method of learning a torque-stroke relationship of a clutch, and more particularly, a clutch torque-stroke learning method in which, during a process of dividing a torque region on a torque-stroke curve (T-S curve) of a clutch into two or more regions and learning the T-S curve passing through two or more torque regions with different torque section values, by learning the curve for a first torque region (e.g., a high-torque region or a low-torque region) when the curve is learned for a second torque region with guaranteed reliability, it is possible to prevent a problem of the T-S curve not converging to a previously learned curve value when the T-S curve is continuously learned for the two or more different torque regions.