Federated AI Platform for Industrial Technologies
Project Status: running
Start Date: March 2022
End Date: February 2025
Budget (total): 2232.44K€
Effort: 49.06 PY
Project-ID: C2021/1-10
Inosens, Türkiye
KocSistem, Türkiye
TAV Technologies, Türkiye
SAMM Teknology, Türkiye
TORUN, Türkiye
DLIT, South Korea
HUFS, South Korea
SmartCore, South Korea
ISEP/IPP, Portugal
Sistrade Software Consulting S.A., Portugal
SIDONIOS MALHAS S.A., Portugal
Abstract
The newly developed manufacturing or transportation industries are deploying more intelligent and smart technologies and the productivity of their system has found to be increased by 17-20% with improved machine utilization and optimization of energy usage by smart manufacturing or operation management systems. Artificial Intelligence (AI) technologies (along with other concepts such as cyber-physical systems, digital manufacturing, additive manufacturing, big data processing, etc.) contribute to the overarching concept of Industry 4.0. The smart manufacturing or transportation system connects the product design, analytics, manufacturing process, stocks and supply chain system, product customization, real-time machining units, product delivery system, and the end customers through the use of cloud computing which made on-demand manufacturing, product customization and maintain the demand and supply ecosystem more efficient
Current AI-based industrial applications have a linear sequential approach for data collection, processing and model deployment cycles where each part of the cycle has a clear task. However, collecting the data required for learning the desired models in one place may not always be possible and centralized data collection may cause data quality issues The recent advances and trends in federated learning address some of these issues in other domains (such as mobile applications). In this project, we aim to build a federated learning platform for industrial automation that offers solutions, leveraging systems engineering for AI, by building AI models on decentralized data using a balanced approach to the learning process between centralized and distributed processing and using blockchain approach to disseminate data allowing accuracy and privacy as well as potential to pay for the data.
This proposal aims to provide great benefit to the manufacturing and transportation industries by efficiently incorporating artificial intelligence into the production or operation line to resolve and eliminate some invisible and internalized problems that cost a lot. Also, a smart contract system will be implemented on the blockchain, storing all the relevant events and transactions of stakeholders in the Industry 4.0 ecosystem, which will allow industries stakeholders and end consumers to verify the data in an intuitive way.