PredMod
Facts
Academic Partner: The School of Engineering and Jönköping International Business School at Jönköping University
Industrial Partners: BUFAB External link, opens in new window., Fagerhult External link, opens in new window., Husqvarna Group External link, opens in new window., Redfield External link, opens in new window., Siemens Energy External link, opens in new window.
Duration of the project: 2022-2023
Research Team:
Tuwe Löfström (Project Leader)
AI's abilities to predict different processes and needs are of great interest to the manufacturing industry. The predictions make it easier to control the entire logistical flow – from material procurement and production planning to storage and delivery. PredMod is a collaborative and research project investigating how to produce confident predictions and to put them into practice.
Introduction
PredMod (Predictive Modelling) is one of multiple projects within the AFAIR research profile that applies AI in industrial organizations. PredMod is part of the theme ”Data-driven development of products and services” (ARA1). This is one out of four projects called "Ontime", combining data science, logistics and supply chain management. Within the framework of PredMod, we take advantage of the collected knowledge from the DaDriCa and RAdaBuff projects and immerse ourselves in machine learning and prediction.
In the project we study situations where it's important to be entirely confident in the decisions made with the support of AI and also to evaluate and further develop the ”Prediction with confidence” framework. We do this by combining research and knowledge transfer between the participants of the project.
PredMod can thus be divided into two parts: A collaborative part and a research part. The collaboration part aims to help companies get started with prediction using AI and to find pilot studies within their own operations. Participators in the project are industrial companies with activities that can clearly benefit from AI predictions as well as software developer Redfield. The second research-oriented part of the project focuses on the prediction itself. How can one estimate and reduce the uncertain part of AI’s predictions?
Motivation and purpose
The purpose of PredMod and its predecessor projects is to create a platform for further collaboration with business within the framework of AFAIR. Through knowledge sharing, we want to help companies understand the technology and its possibilities, in order to achieve maturity that enables future collaboration projects. Within our own research, we will focus on algorithm and theory development. As models deliver a prediction there is no quantification of how secure the model is. Before deployment, it is therefore important to get an idea of how confident the model is, both in full, and not the least in each individual prediction.
Expected results
PredMod’s goal is to contribute to business development with the help of AI-driven decision support in the entire demand and supply chain. Through our research, we want to contribute to making the uncertainty in the models more visible. We also want to identify what type of knowledge is required by the decision makers in the companies.
For more information
- Senior Lecturer Computer Science
- School of Engineering
- tuwe.lofstrom@ju.se
- +46 36-10 1108
Are you interested in a future collaboration?
- External Relations Manager
- School of Engineering
- linda.bergqvist@ju.se
- +46 36-10 1074