Skip to main content

Research Repository

Advanced Search

AKT: Predictive Manufacturing Optimization using Neural Networks (PROMINN)

People Involved

Project Description

We propose to bring together knowledge of advanced AI techniques held by the University, with the internationally renowned expertise in laser development and manufacture of Luxinar Ltd. This combination provides an exciting opportunity to apply AI to improve processes in high-technology laser manufacturing beginning with the problem of Out-of-Box-Failures (OOBF). Despite a robust quality control process, OOBFs represent a recurring real-term cost and a risk to the company's well-deserved reputation. The company produces a large amount of mixed data during the production of the lasers and would like to explore if AI techniques could be used to predict which units may become problematic. The challenge is to work with a combination of time-series data, laser beam profile images and production reports. The size of the data sets is large for each laser, and the company ships thousands of units each year, resulting in a non-human tractable problem. In this project, we will develop an AI system called, Predictive Manufacturing Optimization using Neural Networks (PROMINN). This PROMINN system will enable a thorough examination of big data to explore the causes behind OOBFs. The PROMINN system will use Convolutional Neural Networks (CNNs) for image data and Recurrent Neural Networks (RNNs) or Transformers for sequential data and production reports. We will apply an innovative fusion model to extract features from these data types and integrate them for comprehensive analysis to investigate the causal knowledge for OOBFs. This novel fusion approach combines the causal features from multi-modal datasets to detect the problematic unit in the process of automated investigation of OOBFs.
With the help of the PROMINN system, our goal is to make OOBFs a vanishingly small occurrence, thereby further enhancing the company’s reputation for quality and reliability. This is estimated to generate an additional annual revenue of £120K. The AKT project marks the start of a collaboration between Luxinar and the University of Hull in the field of AI. Once the PROMINN system is developed, we anticipate further projects to integrate it into the production environment in real-time to give feedback during, rather than post-manufacture

Status Project Complete
Funder(s) Innovate UK
Value £29,358.00
Project Dates Mar 25, 2024 - Jun 24, 2024
Partner Organisations Luxinar

You might also like

A ram pressure stripping toy model for jellyfish galaxies Jul 1, 2021 - Aug 27, 2021
An 8 week Undergraduate Research Bursary to fund a summer project to study the causes of jellyfish galaxies. The student intends to provide a model that could give "clues" about the right initial conditions to solve a real-world physics problem.