Vehicle Brake Torque Learning Control for Smooth Deceleration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle braking systems face challenges in accurately controlling brake torque, leading to undesired dynamic responses such as oscillations and jerky behavior due to insufficient or inaccurate application of brake torque by service and auxiliary brakes.
Innovation Solution
A learning model trained using machine learning to determine target brake torque data by associating braking conditions with penalties for undesired vehicle dynamic responses, using operating condition data, brake torque data, and vehicle dynamic response data to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If brake torque is increased to ensure sufficient deceleration, then deceleration performance is improved, but vehicle dynamic stability deteriorates due to oscillations and jerky behavior
Solution Approach 1:
The system implements a feedback mechanism where the learning model continuously receives vehicle dynamic response data and adjusts brake torque commands accordingly. The model learns from past braking events to predict and avoid conditions that cause oscillations, while maintaining effective deceleration. This closed-loop control allows the system to balance deceleration performance with dynamic stability.
Solution Approach 2:
The learning model dynamically adjusts brake torque parameters based on learned patterns from training data. By changing the brake torque magnitude and application rate according to vehicle state and environmental conditions, the system achieves smooth deceleration without oscillations. The model optimizes parameters such as brake force distribution and application timing to prevent dynamic instability.
2Measurement precision
If a learning model is trained with penalty-associated braking conditions to improve accuracy, then braking control precision is improved, but training data requirements and system complexity increase
Solution Approach 1:
The system performs preliminary training action by collecting and processing braking condition data before actual operation. During the training phase, the learning model is exposed to various braking scenarios with associated penalties for undesired dynamic responses. This pre-training prepares the model to make accurate brake torque decisions during actual braking events, improving control precision without adding complexity to the operational system.
Solution Approach 2:
The learning model serves itself by learning from its own performance through the penalty mechanism. When the model produces brake torque commands that result in undesired dynamic responses, it receives penalty signals that guide its self-correction and improvement. This self-learning capability allows the system to achieve high precision without requiring complex external training infrastructure.
Data Source
AI summary
A computer system has a learning model and processing circuitry configured to train the learning model for use in determination of target brake torque data, indicative of a target brake torque for at least each one of a service brake and an auxiliary brake of a vehicle. The processing circuitry receives braking condition information including operating condition data of a current or predicted operating condition of the vehicle during the braking condition; brake torque data, indicative of an applied brake torque for at least each one of the service brake and the auxiliary brake during the braking condition, and vehicle dynamic response data indicative of a vehicle dynamic response of the vehicle during the braking condition. The braking condition is associates with a penalty in response to determining that the vehicle dynamic response data is indicative of a vehicle dynamic response of the vehicle during the braking condition being outside an allowable vehicle dynamic response range.