The Welding Process Monitoring solution aims an online welding defect detector which allows to estimate the presence of defects during the welding. This early detection of defects can translate into great savings in time and material.

Project: ZDZW

Publisher: AIMEN, GRI, UPV, DIRA

Technology: Multi-modal data acquisition software

Industry: Welding business

INTRODUCTION

The Welding Process Monitoring solution is deployed in two stages. First, a multi-modal data acquisition software (SW) and hardware (HW) is defined and deployed on the welding facilities. The monitoring system is used to create a comprehensive and extensive dataset integrating real data from the welding. The HW will be interoperable with the existing machines by integrating communication capabilities (industrial fieldbuses, digital/analogue input-outputs...) to read data from machine controllers and/or existing sensors. Ad-hoc advanced sensors -e.g., high-current probes, high speed oscilloscope- will be used to monitor process parameters (current, voltage, wire speed).
In the second stage, the dataset is thoroughly analyzed to select the most relevant sensors and it is correlated with Non-Destructive Test (NDT) information to try to elaborate models for online defect prediction. Finally, the monitoring system is updated by eliminating the not-used sensors and by deploying the defect prediction models on its processing unit. These updates enable the online process data analysis and the early warnings when a defect is detected/predicted during the welding.

FEATURES/BENEFITS

  • Features:
    • Online defect prediction  
    • Scalable 
    • Adaptable
    • Process monitoring
  • Benefits
    • Can be adapted to different welding processes.  
    • Easy to add new sensors and communication protocols.  
    • Offers a powerful tool to analyze the real welding process. 
    • Enables an early actuation on defect correction, reducing waste on reworking time and scrap.

TECHNICAL INFORMATION

  • In-house SW development using open source or permissive licensed dependencies.
  • AI analysis & online/early defect prediction

RESOURCES

USE CASE

SERVICES LINKED

PUBLISHERS