Welding Quality Determination Using Learned Monitoring Features

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional welding quality determination systems require manual setting of threshold values and rule-based assessments, which are inefficient and need reconfiguration when processing conditions change, such as different materials are used.

Innovation Solution

A welding system that uses a learning device to create a welding result estimation model based on feature information extracted from monitoring data, automatically determining welding quality without manual setting of check items, by calculating the contribution of feature elements in the monitoring information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual threshold setting and rule-based assessment are used for welding quality determination, then the system can operate with simple structure, but the productivity and adaptability deteriorate when processing conditions change

Engineering Contradiction:
Improvewelding quality determination efficiencyVSAvoidsystem structure complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The welding quality determination device automatically extracts feature information from monitoring data and performs quality assessment without requiring manual threshold setting or rule configuration. The system serves itself by autonomously learning from teaching data and making quality determinations based on extracted features, eliminating the need for operator intervention in parameter setup

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system extracts multiple types of feature information including time-series waveform features, statistical features, and frequency-domain features from monitoring data. By dynamically calculating and utilizing these varied parameter representations, the system adapts to different welding conditions and materials without requiring manual reconfiguration of assessment rules

Inventive Principle:
Principle #35Parameter changes

2Adaptability or versatility

If manual threshold setting is required for quality determination, then the device complexity is low, but the adaptability to different materials and processing conditions deteriorates

Engineering Contradiction:
Improveadaptability to different materials and conditionsVSAvoidprocessing complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system automatically adapts to different materials and welding conditions by extracting feature information from monitoring data and using it to determine welding quality. The device performs self-adjustment by learning from teaching data associated with specific materials and conditions, eliminating the need for manual threshold reconfiguration when processing parameters change

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The welding quality determination device is designed to handle multiple welding processes (laser welding, resistance welding, arc welding) and different materials through a universal feature extraction and assessment framework. The system can process various types of monitoring data and extract relevant features regardless of the specific welding method or material being processed

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Productivity

If feature information is extracted automatically from monitoring data, then the productivity improves, but the device complexity increases

Engineering Contradiction:
Improvequality determination speedVSAvoidprocessing complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The feature extraction process is divided into distinct segments: time-series waveform feature extraction, statistical feature extraction, and frequency-domain feature extraction. Each segment processes specific aspects of the monitoring data independently, allowing the system to handle complex analysis through modular, manageable steps that improve processing efficiency

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system introduces feature information as an intermediary between the raw monitoring data and the final quality determination. By calculating and utilizing feature information that represents essential characteristics of the welding process, the system bridges the gap between complex raw data and simple quality assessment, enabling automated decision-making

Inventive Principle:
Principle #24Intermediary (Mediator)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Enables accurate and automated welding quality determination during the process, reducing the need for manual adjustments and improving consistency across varying processing conditions.

Implementation Method 1

a welding monitor device configured to use a sensor to monitor the progress of welding by the welding device and output obtained monitoring information

Methodology Applied
Scientific EffectOptical detection: Photoelectric Effect

Data Source

PatentEP4420819B1Welding system and welding result determination device
Publication Date: 2026.03.11 AMADA CO LTD
  • EP4420819B1 patent drawingFigure 1
  • EP4420819B1 patent drawingFigure 2
  • EP4420819B1 patent drawingFigure 3

AI summary

A welding system according to one aspect includes: a welding device; a welding monitor device configured output monitoring information about welding by the welding device; a learning device configured to accept the input of welding result information associated with welding results obtained after welding and feature information extracted from the monitoring information at the time of welding when the welding results were obtained as teaching data, and create and output a welding result estimation model on the basis of the teaching data; and a welding result determination device configured to extract feature information from monitoring information output at the time of welding for which a welding result is to be determined, input the feature information as data for estimation into the welding result estimation model, and output a quality determination result for the welding, wherein the learning device is configured to detect the values of a preset plurality of feature elements from the monitoring information which includes time series data, calculate the contribution to the welding result, and determine the feature information on the basis of the calculated contribution.