Production Line Monitoring System For Welding Defects

          Quality control in production and assembly plants in industries of developing countries is mostly done by taking random samples of specimens and manually inspecting them. This causes inconsistencies from one specimen to another due to human error. Also, inconsisten

Project Title

Production Line Monitoring System For Welding Defects

Project Area of Specialization

Artificial Intelligence

Project Summary

          Quality control in production and assembly plants in industries of developing countries is mostly done by taking random samples of specimens and manually inspecting them. This causes inconsistencies from one specimen to another due to human error. Also, inconsistencies arise when a quality inspector is changed. Manual inspection is preferred over automated systems because of the cost of sensors required in automated systems.

          we are working on developing a monitoring system for welding surface defects that will use machine learning and computer vision algorithms to detect defects. the system will be designed in such a way that it is easily applicable to many different situations and work environments but we will focus on demonstrating the final product on a fabricated production line.

Project Objectives

1) To develop and train of Machine Learning algorithm to detect welding defects in a production line.

2) To design and fabricate a conveyer belt system for demonstration and training purposes.

3) To test the model and perform hyper-parameter tuning to increase the accuracy of the algorithm.

Project Implementation Method

since the system uses only a camera there are a variety of possible methods to implement the system like a handheld detection device, using a mobile camera as an input, mounted camera approach. 

we are more focused on implementing it on a production line where the camera will be mounted via a stand with some lights and backlights for proper lighting. the camera will be directly connected to a computer that will run the algorithm.

Benefits of the Project

1) low cost.

2) easy to set up and use.

3) overcomes the limitations of other NDT surface defect detection methods for welding.

4) reliable.

5) useable in a wide variety of situations.

Technical Details of Final Deliverable

the final deliverable consists of two parts

1) fabricated conveyor belt for demonstration purposes, it will be a simple Arduino-based control system to control the speed and direction.

2) ML model for defect detection mainly uses the classifier approach while being backed by CNN and random forest algorithms.

Final Deliverable of the Project

HW/SW integrated system

Core Industry

Manufacturing

Other Industries

Core Technology

Artificial Intelligence(AI)

Other Technologies

Sustainable Development Goals

Industry, Innovation and Infrastructure

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com