Two teams of students from the Faculty of Computer and Information Sciences at Misr University for Information Technology developed two applications targeting practical challenges in the manufacturing sector: improving production and machinery efficiency, and detecting the risks of delays in raw-material deliveries. The two projects were developed in cooperation with experts from Germany, Spain, and the Netherlands, and underwent an evaluation session via Zoom attended by supervisors and experts.
According to the university, the two applications were designed in response to the needs of institutions associated with smart industries and the Fifth Industrial Revolution, benefiting from the Smart, Flexible, and Resilient Manufacturing Networks project funded by the Horizon Europe program. The external evaluation committee was chaired by German Professor Bernd Kramer and included Professor Mike Papazoglou, Dr. Maria Ramirez Gutierrez, and Mr. Fernando Gigante from the Spanish organization AIDIMME.
From Sensor Data to Maintenance Decisions
The first application is called “Knowledge-Model-Based Production and Machinery Efficiency Optimizer.” It aims to address a common problem in modern manufacturing platforms: systems remaining in a reactive mode, dealing with machinery failures or production bottlenecks only after they occur.
The application relies on a hybrid decision-support platform that combines machine-learning models, including Isolation Forest and XGBoost, with ontology-based reasoning using the OWL language. The system links sensor-data streams to a hierarchy of factory assets, then uses SWRL rules to trace causal relationships between component degradation, system bottlenecks, and production-efficiency losses.
The platform includes an explanation module based on large language models to convert complex semantic situations into maintenance recommendations in simpler language. The project also includes a knowledge-base expansion layer that manages six competing artificial-intelligence agents tasked with monitoring new behavioral patterns and proposing SWRL rules for engineers to review before approval. The university states that the design is intended to operate in a factory environment through a containerized, low-latency execution path.
A Digital Twin for Raw-Material Risks
The second application, “Semantic Digital Twin for Detecting and Addressing Raw-Material Supply Risks,” focuses on the lack of real-time visibility in manufacturing logistics and the difficulty of handling unstructured service-level agreement conditions. The system aims to detect delivery delays in advance and activate alternative suppliers when necessary.
The application combines machine learning to predict the likelihood of delays, large language models to convert contract texts into structured “ontology triples,” and an ontology reasoner to support decision execution. Through a knowledge graph linking data and dependencies, the framework seeks to shift supply management from responding after delays occur to proactive intervention, while providing a unified source of information related to inventory availability.
Why Does This Development Matter?
The importance of the two projects lies in combining artificial-intelligence technologies with knowledge-representation methods within two specific industrial scenarios, rather than merely presenting a general overview of AI capabilities. In practical terms, the first project focuses on making sensor data interpretable and usable for decision-making, while the second focuses on transforming scattered contract and supply-chain data into a model that can be used for prediction and intervention.
Nevertheless, the material does not present quantitative operational results, such as a reduction in failures or an improvement in supply times, nor does it clarify whether the two applications have entered commercial use or field trials. Therefore, the announced industrial impact remains linked to the development and evaluation stage, rather than to verified performance results in operating factories.
The first application team consisted of Abdel Karim Ahmed, Seif Wafik, Mohamed Hussein, Renad Sameh, and Malak Montaser, under the supervision of Dr. Sherine El-Fayoumi. The second application team consisted of Matthew Wael, Youssef Fares, Youssef Osama, and Mark Tadros, under the same supervision.