Steering Torque Compensation Sign Check for Over-Learning Protection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Steering systems in vehicles can over-learn torque steer mitigation functions, leading to undesirable pulling effects due to increased gain values, which are not corrected on new key cycles.
Innovation Solution
A method and system that detect differences in mathematical signs between compensation and handwheel torque, apply filters to prevent false detection, and adjust learned gains or invert compensation torque values to mitigate over-learning, using ramp rates to stabilize torque.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If torque steer mitigation functions use calibrations set to quickly learn, then torque offset compensation speed is improved, but over-learning occurs causing pulling effects
Solution Approach 1:
The system uses feedback by comparing the sign of compensation torque with the sign of handwheel torque to detect over-learning conditions. When the signs differ, the system identifies that over-learning has occurred and takes corrective action by adjusting learned gains or inverting the compensation torque value, thus eliminating the harmful pulling effect while maintaining fast learning capability.
Solution Approach 2:
The system changes parameters dynamically by adjusting learned gains or inverting compensation torque values based on the detected sign mismatch. This parameter adjustment prevents over-learning from causing harmful effects while preserving the fast learning capability of the torque steer mitigation function.
2Measurement precision
If learned gain is stored as long term gain, then torque compensation accuracy is improved, but over-learning accumulates on consecutive key cycles
Solution Approach 1:
The feedback mechanism compares compensation torque sign with handwheel torque sign across key cycles to detect when over-learning has accumulated. This allows the system to maintain long-term gain storage for accuracy while periodically correcting accumulated over-learning errors that occur on consecutive key cycles.
Solution Approach 2:
The system performs periodic checks on consecutive key cycles to detect sign mismatches indicating over-learning accumulation. This periodic detection allows the system to maintain long-term gain storage while periodically correcting accumulated errors, balancing accuracy with prevention of over-learning.
3Speed
If compensation torque is applied without sign verification, then steering response speed is improved, but instability occurs due to inverted compensation
Solution Approach 1:
The system performs preliminary verification by checking the sign of compensation torque against the handwheel torque sign before applying compensation. This preliminary action prevents inverted compensation from causing instability while maintaining fast steering response, as the sign check occurs before the compensation is applied.
Solution Approach 2:
The sign verification acts as a preliminary anti-action by preventing the harmful effect of inverted compensation torque before it can occur. By checking signs in advance, the system blocks the potential instability-causing action while preserving the beneficial fast response characteristic.
Data Source
AI summary
A method for providing over learn protection for torque steer mitigation includes receiving a compensation torque value corresponding to a torque offset associated with a transmission torque and receiving a handwheel torque value associated with a handwheel of a steering system. The method also includes detecting a mathematical sign of the compensation torque value, detecting a mathematical sign of the handwheel torque value, and determining whether the mathematical sign of the compensation torque value is the same as the mathematical sign of the handwheel torque value. The method also includes, in response to a determination that the mathematical sign of the compensation torque value is not the same as the mathematical sign of the handwheel torque value, adjusting one or more learned gains, and generating an inverted compensation torque value by inverting the mathematical sign of the compensation torque value.