Datenbasierte und durch Maschinelles Lernen erweiterte Entwurfsansätze für elektrische Maschinen (el. knyga) (skaityta knyga) | knygos.lt

Aprašymas

This dissertation, entitled "Data-Driven and Machine Learning-Enhanced Design Approaches for Electrical Machines", presents novel data-driven and machine learning-enhanced approaches for the design and optimization of electrical machines. The growing complexity of electrical machine development and the increasing availability of design data motivate the integration of modern machine learning techniques into established engineering workflows. The work introduces methods for the geometric representation, generation, and evaluation of machine designs, with a particular focus on scalable data-driven design processes. Large-scale design datasets are generated and utilized to train surrogate models and generative machine learning models, including variational autoencoders. These models enable efficient exploration of design spaces, rapid performance estimation, and the synthesis of novel machine geometries. The proposed approaches are applied to practical electrical machine design problems and demonstrate how machine learning can accelerate optimization workflows while maintaining engineering relevance and physical interpretability. By combining geometric modeling, data-driven methods, and optimization techniques, this dissertation contributes to the development of next-generation design methodologies for electrical machines.