Tightening Class Estimation Using Torque, Angle, and Context Data

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods struggle to accurately determine the tightening class of a tightening operation performed by a tightening tool, as it is difficult for human operators or machines to analyze sensor data from the tool, and machine-learning models often require appropriate training to perform accurate estimations, which can be inconsistent across different tools or environments.

Innovation Solution

A method and device using machine-learning models that incorporate context data such as environmental conditions, previous tightening operations, and tool settings to improve the accuracy of tightening class estimation by associating torque and angle values with specific tightening operations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If machine-learning models are used to analyze torque and angle tightening results, then the automation and speed of tightening class determination are improved, but the reliability and accuracy of the estimation deteriorate due to insufficient or inappropriate training data

Engineering Contradiction:
Improveautomation of tightening class determinationVSAvoidaccuracy of tightening class estimation
Core Design Contradiction:
Extent of automationVSReliability

Solution Approach 1:

The system performs preliminary training of the machine-learning model using extensive torque and angle data collected from multiple tightening operations before actual use. This preliminary action ensures the model is properly trained and reliable before it is deployed for automated tightening class determination, resolving the contradiction between automation and reliability.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system implements a feedback mechanism where the machine-learning model continuously learns from new tightening data. The model's predictions are compared with actual outcomes, and this feedback is used to refine and improve the model's accuracy over time, ensuring reliable automated determination of tightening classes.

Inventive Principle:
Principle #23Feedback

2Measurement precision

If extensive training data is collected to improve machine-learning model accuracy, then the measurement precision of tightening class determination is improved, but the loss of time and complexity of the process increases

Engineering Contradiction:
Improveprecision of tightening class determinationVSAvoidtime for data collection and model training
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system collects and processes training data in advance, before actual tightening operations. By performing the time-consuming data collection and model training as a preliminary action, the system achieves high measurement precision without causing time delays during critical tightening operations.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system uses a substantial amount of training data beyond the minimum required to achieve good model performance. This excessive action in data collection during the training phase ensures high precision during operation, while the actual tightening process remains fast since the model is already trained.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4458516B1ML estimation of tightening classes
Publication Date: 2026.01.28 ATLAS COPCO IND TECHNIQUE AB INTELLECTUAL PROPERTY DEPARTMENT
  • EP4458516B1 patent drawingFigure 1
  • EP4458516B1 patent drawingFigure 2a~2b
  • EP4458516B1 patent drawingFigure 2c~2d

AI summary

The present disclosure relates to a method of a device (20, 35) for determining a tightening class of a tightening operation performed by a tightening tool (10). The determination is based on training a machine-learning model with an acquired set of observed torque and angle values, at least one set of context data and one or more associated tightening classes. Subsequently, the trained model is utilized for determining from supplied torque and angle values, as well as context data, one or more tightening classes associated with the supplied values and context data