Industrial Automation

Standards based Zero Touch Device Replacement

This project presents a vendor-neutral approach to Zero Touch Device Replacement (ZTDR) using standards like OPC UA and DHCP. By automating secure onboarding and lifecycle management, the prototype significantly reduces downtime and manual effort, paving the way for standardized, interoperable solutions in industrial Operational Technology.

Supervisor:
Prof. Jürgen Jasperneite

#IndustrialIoT #OPCUA #Automation #CyberSecurity

Industrial Automation

Product and Supply Chain Information Security through Distributed Ledger Technologies for Digital Product Passports

This master thesis examines how Distributed Ledger Technologies can secure Digital Product Passports in retail supply chains. After evaluating blockchain and Directed Acyclic Graph systems, it identifies IOTA's Tangle as a strong option and designs a coinless architecture. Experiments show scalable, reliable data integrity for large-scale DPP use.

Supervisor:
Prof. Henning Trsek

#Industrial Automation

Web-developement

Lemgo-Link

Developed by students for students, this platform bridges the gap between campus and community. Using advanced web programming and databases, we match students with local job offers and help them find housing via our platform. Your essential hub for local networking and career growth.

Supervisor:
Prof. Jürgen Jasperneite

#Community #Web-Developement #Programming

Industrial Automation

Design, Development and Evaluation of Cybersecurity Testing Solutions for Operational Technology Components

In this work a CRA-compliant cybersecurity testing for firewalls, vulnerabilities, fuzzing, NTP and SysLog is developed. The solution approach improves automation, reproducibility, secure logging, and safe file transfer, enabling reliable test chaining and resulting in vulnerability-free OT systems.

Supervisor:
Prof. Henning Trsek

#Industrial Automation

Data-Driven Engineering

AI-Generated Sensor Data for Smarter Industrial Processes

This project develops diffusion models to generate realistic multivariate sensor time-series for industrial motors and penicillin production. By producing diverse, high-fidelity signals, variational diffusion models successfully augment scarce training data for analytics and monitoring. The study also evaluates the impact of physics-based constraints on data realism and model validation.

Supervisor:
Prof. Markus Lange-Hegermann

#Generative Diffusion Models #Time-Series Augmentation #Industrial Process Simulation

Data-Driven Engineering

Smarter Testing for Industrial Control Software

That fits the project well because the study focuses on using large language models to generate and evaluate unit tests for Structured Text function blocks in industrial automation, with emphasis on oracle correctness, fault detection, and testability-by-design.

Supervisor:
Prof. Markus Lange-Hegermann

#LLM-Based Testing #Industrial Automation #Testability-by-Design

Data-Driven Engineering

TerminalVision

This thesis uses CAD-based rendering and domain randomization to train industrial object detection with synthetic data. By detecting terminal strip components with near-real performance, it eliminates manual annotation. This scalable approach enables practical AI deployment for quality control in manufacturing settings where real-world training data is scarce.

Supervisor:
Prof. Markus Lange-Hegermann

#Synthetic Data Generation #Industrial Object Detection #Scalable Quality Control

Data-Driven Engineering

AI-Powered Offer Generation

This AI-powered pipeline converts customer requirements into professional quotations using RAG, PDF analysis, and Excel-based pricing. It ensures factual consistency and automates enterprise workflows while minimizing hallucinations. The system evaluates various LLMs and embedding models to ensure scalable, reliable performance for complex industrial automation.

Supervisor:
Prof. Markus Lange-Hegermann

#AI-Driven Quotation #Retrieval-Augmented Generation #Enterprise Workflow Automation

Artificial Intelligence

Cycle4AI

The Cycle4AI ergometer makes AI’s invisible energy footprint tangible. As you pedal, the display translates your physical effort into real-time power consumption data. By comparing giants like ChatGPT with lean, specialized models, it highlights the efficiency of decentralized AI—sensitizing you to the sustainable cost of every prompt.

Supervisor:
Prof. Jürgen Jasperneite

#AI #Python #AI Literacy

Industrial Automation

Exploring the Integration of OPC UA over MQTT in Industrial IoT: A Case Study of a Cloud-enabled Table Sorting Machine

This research project explores how OPC UA PubSub over MQTT operates on a real production asset and integrates with Azure. It describes a clone-and-publish setup, two edge-to-cloud paths with different brokers, and evaluates message flow, latency, and reliability to show scalable, lossless performance.

Supervisor:
Prof. Henning Trsek

#Industrial Automation

Smart Farming

SmartGrow

This Python-powered system automates plant growth using a Raspberry Pi and real-time sensors. Different sensors monitor the climate, while actuators like pumps and LEDs respond instantly to data. It’s a precise, space-saving feedback loop, ensuring optimal resource efficiency and a sustainable harvest.

Supervisor:
Prof. Jürgen Jasperneite

#AI #Python #Sustainability

Data-Driven Engineering

Smarter Sampling of Expensive Data

This project demonstrates how to reduce expensive labeling and testing in engineering. By comparing two Gaussian-process strategies across 13 benchmarks, it identifies when targeted sampling accelerates model building by 20–56% and when exploration is safer. It provides a practical rulebook for efficient, budget-limited active learning in data-modeling tasks.

Supervisor:
Prof. Markus Lange-Hegermann

#Active Learning #Gaussian Processes #Engineering Optimization