Skip to main content
To KTH's start page

PICTORIAL - Predictive Intelligent Control for Resistance Welding

The primary contribution of this project is to develop machine learning (ML) models for the quality control of resistance spot welding (RSW) in the automotive industry, giving real-time cognitive support to personnel involved in production.

A distinct approach to training the ML models is to apply real production data from the automotive factories, which is rare in literature. Previously, simulated or experimental data have been used. Resistance welding is the predominate method in automotive industry today. The developed models will make it easier to reduce material extraction through increased use of circular, thinner and more advanced materials that are demanding to join.

New materials and new possibilities regarding sensors and computing systems will, together with the new predictive intelligent digital control system developed in this project, make a huge step in manufacturing capability.

Expected results and effects

The new predictive intelligent digital control system developed in the project, will make a leap forward in the quality assessment of RSW, the most used joining method within automotive. The project addresses the FFI sub-programme’s mission 2b), “Social sustainability throughout the value chain, e.g., Retain employees and attract new skills through techniques and solutions for cognitive and physical support ...”. At the same time, it will increase the digitalization knowledge in automated welding production.

Planned approach and implementation

Participants in the project consist of people with deep knowledge in production and spot welding, and in machine learning (ML). In the start-up of the project, time will be used to find a common language. Then relevant production data must be retrieved, after which the data must be curated and significant data translated into data language. Several different ML algorithms will be evaluated and combined, then trained against the acquired and curated data sets. When the final model is chosen, it will be tested in a demonstrator in industry.

Page responsible:vickid@md.kth.se
Belongs to: Engineering Design
Last changed: Nov 13, 2023
EXHILO: Real-time exoskeleton control for human-in-the-loop optimization
HARU: On the compliance, reliability and motion control of a tabletop robot
IRIS: Novel Mechatronic Systems and Soft Robotics enabled by 4D Printing and Machine Learning
SocketSense
Excellence in production research (XPRES)
WILDCAT
FIRE - the Future of mItigating Rural firEs
PICTORIAL - Predictive Intelligent Control for Resistance Welding
SALIENCE4CAV
CAVeaT: Connected Automated Vehicles trialling and Trustworthiness
TADDO2: Trustworthy Automated Driving DevOps, continued
AD-EYE
TADDO: Trustworthy Automated Driving DevOps
Entice - Enablers for trustworthy, infrastructure supported, autonomous vehicles
StorAIge: Embedded storage elements on next MCU generation ready for AI on the edge
BOAUT - Boundaries Of AUTonomy
Susan's Ride on Campus 2030
WASP: AD-EYE – an Open Modular Testbed
HUBCAP
TRUSST-E
ADinSOS
CAMPUS 2030 – Enabling systemic solutions for smart roads
ATRIUM: ArchiTectural RefInement using Uncertainty Management
FED4SAE
HI2OT
InSecTT – Intelligent Secure Trustable Things
PRYSTINE
Rigorous Systems Engineering
SMP Tool