Driveshaft Overtorque Prediction for Accurate Vehicle Dimensioning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for dimensioning driveshafts in vehicles are imprecise and time-consuming, often relying on rules of thumb or trial and error, which can lead to driveshaft failure or unnecessary weight and material costs.
Innovation Solution
A computer-implemented method using a trained machine learning model based on reference data from torque sensors to predict overtorque events in driveshafts, allowing for accurate dimensioning without direct torque measurement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the driveshaft is made more robust to prevent failure, then reliability improves, but weight and materials cost increase
Solution Approach 1:
The patent uses machine learning to predict torque classifications and overtorque events, enabling data-driven optimization of driveshaft dimensioning parameters. By analyzing patterns in torque data, the system identifies the minimum necessary driveshaft specifications to prevent failure while minimizing weight and material usage.
2Ease of manufacture
If traditional rules of thumb are used for driveshaft dimensioning, then the process is simple, but precision and accuracy deteriorate
Solution Approach 1:
The patent replaces traditional mechanical engineering rules of thumb with a machine learning-based predictive system. The model processes torque classification data and overtorque event predictions to provide precise, data-driven dimensioning recommendations, substituting empirical rules with algorithmic analysis for improved accuracy.
3Reliability
If trial and error methods are used for driveshaft dimensioning, then comprehensive testing can be performed, but time and cost increase significantly
Solution Approach 1:
The patent performs preliminary analysis by training the machine learning model on historical torque data and predicting overtorque events before finalizes driveshaft dimensioning. This preliminary action identifies potential failure conditions and optimizes dimensions in advance, eliminating the need for time-consuming trial and error testing during product development.
4Measurement precision
If torque sensors are installed for direct measurement, then torque data accuracy improves, but device complexity and cost increase
Solution Approach 1:
The patent introduces a machine learning model as an intermediary that infers torque classifications and predicts overtorque events from indirect vehicle sensor data. This mediator enables accurate torque analysis without requiring direct torque sensor installation, reducing system complexity while maintaining measurement precision through computational inference.
Data Source
Figure 1~2A
Figure 2B
Figure 2B
AI summary
A computer implemented method for predicting overtorque events of a driveshaft of a vehicle without using a direct measurement of a torque of the driveshaft of the vehicle, the method comprising collecting vehicle data related to the vehicle (302), preprocessing the vehicle data (304), applying a trained machine learning model to the pre-processed vehicle data to generate torque classifications (306); and computing overtorque events of the driveshaft of the vehicle based on the generated torque classifications (308), wherein the trained machine learning model has been developed based on reference data collected from a plurality of reference vehicles, the reference data comprising torque information received from torque sensors of the reference vehicles.