Gear Surface Profile AI Classification for Transmission Vibration Screening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying gears that cause vibrations in transmissions are inefficient, leading to prolonged manufacturing times and error analysis, as they rely on end-of-line testing and geometric measurements, which are not reliable in predicting faulty gears before installation.
Innovation Solution
A computer-implemented method using a generic AI model trained on reference gear profiles to categorize gears as either causing vibrations above or below a threshold, allowing for precise identification of faulty gears through surface profile analysis, utilizing machine learning approaches like neural networks or random forests.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If end-of-line testing and geometric measurements are used to identify faulty gears, then manufacturing reliability is improved, but manufacturing time and error analysis time increase significantly
Solution Approach 1:
The patent applies preliminary action by performing surface profile measurement and AI-based classification of gears before they are installed in transmissions. The generic AI model is trained on reference gear profiles and used to categorize incoming gears as faulty or fault-free based on their surface profiles, enabling early identification without waiting for end-of-line testing.
Solution Approach 2:
The patent replaces traditional mechanical testing methods (end-of-line vibration testing and geometric measurements) with an AI-based classification system. The generic AI model uses machine learning algorithms to analyze surface profiles and predict gear performance, substituting physical testing with computational analysis that is both faster and equally reliable.
2Manufacturing precision
If traditional geometric measurement methods are used to detect faulty gears, then manufacturing precision is maintained, but measurement precision and reliability of prediction decrease
Solution Approach 1:
The patent replaces traditional geometric measurement methods with surface profile measurement combined with AI-based classification. Instead of relying on conventional dimensional checks, the system captures detailed surface profiles and uses a generic AI model to analyze them, providing more accurate fault prediction while maintaining manufacturing precision requirements.
Solution Approach 2:
The patent changes the measurement parameter from traditional geometric dimensions to surface profile characteristics. By measuring and analyzing the actual surface profile of gear teeth rather than just nominal dimensions, the system gains deeper insight into manufacturing variations that affect performance, enabling more precise fault detection.
3Reliability
If all gears undergo comprehensive testing before installation, then reliability is improved, but productivity decreases due to extended inspection time
Solution Approach 1:
The patent performs preliminary classification of gears using surface profile measurement and AI-based categorization before installation. This early sorting identifies faulty gears quickly without requiring time-consuming comprehensive testing, maintaining reliability by catching defects early while preserving productivity through rapid classification.
Solution Approach 2:
The patent extracts only the essential information needed for fault detection by measuring surface profiles and using AI to identify critical patterns. Instead of performing comprehensive testing on all gears, the system extracts key surface characteristics and uses the generic AI model to make rapid classification decisions, maintaining reliability with minimal inspection time.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer-implemented method for identifying a gear (5) that causes vibrations in a gearbox is presented and described, the method comprising the following steps: training a generic AI model using a training dataset of a group of reference gears (19), wherein for each reference gear (19) of the group at least one reference profile of the surface (21) of the reference gear (19) and a gear category are provided, wherein each reference gear (19) of the group that causes a vibration in the gearbox equal to or above the vibration threshold is assigned a first gear category (23) of the gear categories, and each reference gear (19) that causes a vibration in the gearbox below the vibration threshold is assigned a second gear category (25) of the gear categories, wherein the trained AI model can assign a gear category of the gear categories to each gear (5);Capturing a profile of a surface (11) of a gear (5); and assigning one of the gear categories to the gear (5) based on the captured profile by the trained AI model. The invention further relates to a corresponding device, as well as a corresponding computer program and a corresponding computer-readable medium.