I'm an engineer in electronics and information technologies. My primary working/research focuses on machine learning and deep learning driven approaches for image classification, object detection as well as predictive maintenance applications in industrial context.
Email: wagner-tobias@outlook.de
The doctoral thesis proposes a framework for the current signal based bearing fault detection on rotating machinery by using a pipeline of chained data-manipulation steps where the individual steps to be used in the pipeline as well as their hyperparameterization is carried out by autoML mechanisms to reduce the human induced bias on the solution search.
This study addresses the problem of bearing damage detection on Permanent Magnet Synchronous Motors (PMSMs) based on the internal phase current data sources. It focusses on tackling the accuracy degradations caused by variations of the motor parameters like rotational speed, load torque and radial forces. Therefore, we propose a feature based deep unsupervised domain adaptation method to improve the classification accuracies of two operating points by use of only one label set.
The objective of this study is the automatic bearing fault detection of permanent magnet synchronous motors (PMSM) using phase current data. Our research proposes a method using sensor fusion to improve the information quantity as well as the gained quality from all available sensor sources using a multi stage workflow. As an initial feature extraction stage a deep neural network architecture based on a 1D-CNN-LSTM is applied on the raw current data to create baseline probability distributions. Then, probability merging is applied to combine the results of all available baseline classifiers to a new feature matrix which is considered as the feature-set for the final classification stage which is build up on an multi learner ensemble of k-Nearest-Neighbor classifiers.
The objective of this study was the automatic classification of boxing punches to help athletes improving their punch performance. Our research proposes a system consisting of a smartwatch as the measuring device and serversided machine learning classification models.
The research within the PhD aims to develop concepts for the use of machine learning models for intelligent fault detection in electronic drives. The goal is to develop a strategy for transferable analysis models that enable the advantages of model-based data analysis without being tied to the drive systems in general. By using those transferable models, the results of the research are to be applied and evaluated in a prototypical manner. The focus of the component faults to be recognized are bearing damages in permanent magnet synchronous motors (PMSM).
As part of the smartPunch project, movement data is analyzed using machine learning. In cooperation with sports clubs, a dataset is created from annotated three-dimensional, direction-oriented acceleration data. This dataset forms the training basis of the models for the automatic classification of the box punches.