Repository der Technischen Hochschule Ingolstadt
Not a member yet
4170 research outputs found
Sort by
Relevance-based Feature Masking: Improving Neural Network based Whale Classification through Explainable Artificial Intelligence
No full textSmartManPy – Open Source Synthetic Manufacturing Data
Get PDFData Science projects in manufacturing often suffer from difficult access to the necessary high-quality data, such as sensor readings or product quality records. In addition, strict rules due to data protection and intellectual property concerns hinder broad use of the provided data for innovative research, especially in the academic context. To enable the research and evaluation of approaches for a variety of Industry 4.0 use cases, we developed SmartManPy, an extension to the ManPy discrete event simulation environment. We extended the scope of the project to meet our needs and are now able to simulate production lines and the resulting data in just a few lines of code. SmartManPy (Smart Manufacturing Python) includes complex synthetic feature generation, quality control and failures based on recorded features, and data drift simulation. We additionally added a basic environment to train AI agents using reinforcement learning. Our code is released to the public on GitHub
Practical Comparison of UiPath and Power Automate by Creating an Automation Use Case from Logistics
No full text
Methodology and Performance Assessment of Three-Dimensional Vehicular Ad-Hoc Network Simulation
No full textPacket-based simulation is a key tool for the research and development of Vehicular Ad-hoc Networks (VANETs). Over the last decade, many models throughout the communication stack have been presented, which have increased the degree of realism that can be achieved with popular simulation frameworks. Nevertheless, the three-dimensional aspects of many real-world traffic situations barely find consideration. In this paper, we present a holistic approach to simulate large-scale three-dimensional VANET scenarios. We briefly summarize our previously presented models covering different aspects of communication in 3D scenarios, including an environmental diffraction model, an n-ray ground interference model, and the consideration of multi-floor communication. We then describe the principle of a model selector, which applies the appropriate models depending on the environment of the currently transmitted packet. Subsequently, we use the outlined methodology implemented in our Veins 3D framework to simulate a large urban reference scenario. The results differ significantly from comparable 2D simulations, demonstrating the necessity of three-dimensional considerations. However, they also show strongly increased execution times. Therefore, we further suggest different approaches to improve the simulation performance. Based on these optimizations, simulation durations in the same order of magnitude as a comparable 2D simulation can be achieved
On the Impact of priority-based MAC Layer Scheduling in 5G V2N multi-application Scenarios
No full textAI-driven point cloud framework for predicting solder joint reliability using 3D FEA data
No full textCrack propagation in solder joints remains a major challenge impacting the thermo-mechanical reliability of electronic devices, underscoring the importance of optimizing package and solder pad designs. Traditional Finite Element Analysis (FEA) techniques for predicting solder joint lifespan often rely on manual post-processing to identify high-risk regions for plastic strain accumulation. However, this manual process can fail to detect complex and subtle failure mechanisms and purely based on averaging the creep strain and correlating it to lifetime values collected from experiments using Coffin Manson equation. To address these limitations, this study presents an Artificial Intelligence (AI) framework designed for automated 3D FEA post-processing of surface-mounted devices (SMDs) assembled to Printed Circuit Board (PCB). This framework integrates 3D Convolutional Neural Networks (CNNs) and PointNet architectures to automatically extract complex spatial features from 3D FEA data. These learned features are then linked to experimentally measured solder joint lifetimes through fully connected neural network layers, allowing the model to capture complex and nonlinear failure behaviours. The research specifically targets crack development in solder joints of ceramic-based high-power LED packages used in automotive lighting systems. This dataset included variations in two-pad and three-pad configurations, as well as thin and thick film metallized ceramic substrates. Results from the study demonstrate that the PointNet model outperforms the 3D CNN, achieving a high correlation with experimental data (R2 = 99.91%). This AI-driven, automated feature extraction approach significantly improves the accuracy and provide the more reliable models for solder joint lifetime predictions, offering a substantial improvement over traditional method