Tightening Class Estimation Using Normalized Torque-Angle Phases
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining the correctness of a tightening operation performed by a tightening tool is burdensome and time-consuming for human operators and machines, as existing methods struggle to accurately analyze torque and angle sensor data, requiring appropriate training of machine-learning models for accurate estimations.
Innovation Solution
A method involving the acquisition and normalization of torque and angle values from a tightening tool, separating the rundown and end-tightening phases, and training machine-learning models to identify the type of tightening operation, with normalization improving data resolution and enabling accurate classification of tightening classes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine-learning models are trained with raw torque and angle values covering the entire tightening range, then the model can process complete tightening data, but the normalization resolution becomes poor especially for the end-tightening phase
Solution Approach 1:
The tightening process is divided into two distinct phases: rundown phase and end-tightening phase. Each phase is normalized separately with its own min/max values, allowing high resolution for the end-tightening phase (30°-40° range) while still processing the complete tightening data. This segmentation resolves the contradiction by enabling precise normalization without requiring complex alternative approaches.
2Reliability
If human operators or machines analyze sensor data to determine tightening correctness, then tightening results can be evaluated, but the process becomes burdensome and time-consuming
Solution Approach 1:
The system performs preliminary normalization of torque and angle data during the tightening process itself, preparing the data in advance for rapid ML model evaluation. The normalization constants (min/max values) are determined beforehand for each phase, enabling quick classification without time-consuming post-processing analysis by operators or machines.
Solution Approach 2:
The manual or machine-based analysis process is replaced with an automated machine-learning model that classifies tightening results based on normalized torque and angle values. This substitution eliminates the burdensome and time-consuming human or traditional machine analysis while maintaining reliable determination of tightening correctness.
3Manufacturing precision
If the entire torque and angle range is used for normalization, then all tightening data can be processed, but the resolution for the end-tightening phase becomes insufficient
Solution Approach 1:
The torque and angle data are segmented into rundown phase and end-tightening phase, with separate normalization applied to each. This allows the end-tightening phase (critical for tightening class determination) to be normalized with high resolution over its small range (30°-40°), preventing information loss while still processing the complete tightening data range.
Data Source
AI summary
The present disclosure relates to a method of a device of enabling determination of a tightening class of a tightening operation performed by a tightening tool. The determination is based on normalizing torque values of an end-tightening phase with a determined torque value range of the end-tightening phase and angle values of the end-tightening phase with a determined angle value range of the end-tightening phase, and training a machine-learning model with the normalized torque and angle values of the end-tightening phase and at least one tightening class associated with the normalized torque and angle values of the end-tightening phase, the tightening class identifying a type of tightening operation having been applied to the fastener.