AI Welding Position Recognition for Secondary Battery Assembly

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing welding control systems face issues with misrecognition of welding positions due to variations in photographing environments and image quality, leading to potential misrecognition and defects in secondary battery manufacturing.

Innovation Solution

An artificial intelligence-based system that includes a machine vision system, relay system, and welding system, utilizing pre-trained AI models to recognize and verify welding positions, with alarms and defect inspections to improve accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If machine vision technology is used to recognize welding positions, then automation is improved, but recognition accuracy deteriorates due to misrecognition of scratches and environmental variations

Engineering Contradiction:
Improveautomation of welding position recognitionVSAvoidwelding position recognition accuracy
Core Design Contradiction:
Extent of automationVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by pre-training an AI model with diverse image data including various lighting conditions, camera lens contaminations, and electrode configurations before actual welding recognition. This pre-training prepares the system to handle environmental variations and prevents misrecognition issues before they occur in production

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements feedback by using an inspection system to verify welding quality and feed this information back to update and retrain the AI model. This continuous feedback loop improves recognition accuracy over time by learning from actual welding outcomes and correcting misrecognition patterns

Inventive Principle:
Principle #23Feedback

2Productivity

If machine vision technology is used to identify welding positions, then productivity is improved, but reliability deteriorates due to misrecognition under varying photographing conditions

Engineering Contradiction:
Improvewelding process efficiencyVSAvoidwelding position identification reliability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent applies parameter changes by training the AI model to recognize welding positions across a wide range of parameter variations including different lighting conditions, camera angles, lens contaminations, and electrode configurations. This enables reliable recognition despite varying photographing conditions while maintaining high productivity

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent uses a composite approach by combining machine vision technology with AI-based image recognition algorithms. This composite system integrates the speed of machine vision with the pattern recognition capabilities of AI, achieving both high productivity and reliability in welding position identification

Inventive Principle:
Principle #40Composite materials

3Measurement precision

If AI model recognition is used to improve welding position accuracy, then measurement precision is improved, but device complexity increases due to additional systems and model training requirements

Engineering Contradiction:
Improvewelding position recognition accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies segmentation by dividing the welding control system into distinct functional modules: machine vision system for image capture, AI model for position recognition, relay system for coordination, and inspection system for verification. This modular segmentation manages complexity while maintaining high measurement precision

Inventive Principle:
Principle #1Segmentation

Data Source

PatentUS20250245808A1System, Apparatus and Method for Controlling Welding of Secondary Battery Based on Artificial Intelligence
Publication Date: 2025.07.31 SK ON CO LTD
  • US20250245808A1 patent drawing
  • US20250245808A1 patent drawing
  • US20250245808A1 patent drawing

AI summary

The present disclosure provides a system, apparatus and method for controlling welding based on artificial intelligence. The artificial intelligence-based welding control system according to an embodiment of the present disclosure may include a machine vision system configured to photograph at least a portion of a welding object through a camera, and transmit the photographed image (hereinafter, “photography image”) to a relay system; the relay system configured to receive the photography image from the machine vision system, recognize a welding position of the welding object from the received photography image based on a pre-trained first artificial intelligence model, determine whether the recognized welding position is included in a designated position range, and when the recognized welding position is included in the designated position range, transmit the recognized welding position to a welding system; and the welding system configured to receive the welding position from the relay system, and perform welding at the received welding position.