Vehicle Shift Control Using Deep Learning Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle shift control technologies require numerous shift patterns and are inefficient in preventing busy shift and acceleration delay phenomena due to their reliance on traditional shift patterns rather than predictive models.
Innovation Solution
A deep learning model-based system that predicts vehicle speed and accelerator position sensor values for each future time point, allowing for the prediction and control of gear stages to prevent busy shift and acceleration delay by maintaining or adjusting gear stages based on predicted conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional shift patterns are used to control vehicle shift, then the shift control logic becomes complex with hundreds of patterns required, but the system can prevent busy shift and acceleration delay phenomena
Solution Approach 1:
The patent transforms the complex shift control problem by changing the control parameters from discrete shift patterns to continuous prediction variables (vehicle speed, APS value, gear stage) with associated uncertainty levels. This allows the system to determine shift timing based on predicted values and uncertainty comparisons rather than consulting hundreds of predefined patterns, thereby reducing logical complexity while maintaining control reliability
Solution Approach 2:
The system performs preliminary prediction of future vehicle speed, APS value, and gear stage before actual shift execution. By predicting these parameters in advance and comparing uncertainty levels, the system proactively determines optimal shift timing and prevents busy shift phenomena before they occur, rather than reacting to current conditions with complex pattern matching
2Device complexity
If deep learning model prediction is used to control shift, then the shift control logic complexity is reduced, but the ability to accurately predict future conditions and prevent busy shift and acceleration delay phenomena may be affected
Solution Approach 1:
The system incorporates feedback mechanisms by comparing predicted values with actual sensor readings and calculating uncertainty levels. The uncertainty level serves as a feedback parameter that indicates prediction reliability, allowing the system to adjust control decisions based on the quality of predictions. This feedback loop ensures that the simplified prediction-based control logic maintains reliability by adapting to actual system behavior
Solution Approach 2:
The patent replaces the mechanical system of predefined shift patterns with an intelligent prediction system using deep learning models. Instead of mechanically selecting from hundreds of predefined patterns based on current conditions, the system uses neural network predictions of future states to determine shift timing, substituting complex mechanical logic with adaptive intelligent control that reduces logical complexity while maintaining or improving reliability
Data Source
AI summary
An apparatus configured for controlling shift of a vehicle and a method therefore are provided. The apparatus include a storage storing a deep learning model, learning of which is completed; and a controller that predicts a vehicle speed and an accelerator position sensor (APS) value for each future time point according to the deep learning model, predicts a gear stage for each future time point using the predicted vehicle speed and the predicted APS value, and controls the shift of the vehicle based on the gear stage for each future time point, thus preventing a busy shift phenomenon and preventing an acceleration delay phenomenon.