Fakultät Maschinenbau
Refine
Document Type
Language
- English (3)
Is part of the Bibliography
- no (3)
Keywords
- Machine learning (3) (remove)
Institute
Begutachtungsstatus
- peer-reviewed (3)
New structural sheet metal parts are developed in an iterative, time-consuming manner. To improve the reproducibility and speed up the iterative drawability assessment, we propose a novel low-dimensional multi-fidelity inspired machine learning architecture. The approach utilizes the results of low-fidelity and high-fidelity finite element deep drawing simulation schemes. It hereby relies not only on parameters, but also on additional features to improve the generalization ability and applicability of the drawability assessment compared to classical approaches. Using the machine learning approach on a generated data set for a wide range of different cross-die drawing configurations, a classifier is trained to distinguish between drawable and non-drawable setups. Furthermore, two regression models, one for drawable and one for non-drawable designs are developed that rank designs by drawability. At instantaneous evaluation time, classification scores of high accuracy as well as regression scores of high quality for both regressors are achieved. The presented models can substitute low-fidelity finite element models due to their low evaluation times while at the same time, their predictive quality is close to high-fidelity models. This approach may enable fast and efficient assessments of designs in early development phases at the accuracy of a later design phase in the future.
Tightening quality requirements of industrial products involving manual assembly lead to the development of assisting workbenches with integrated functions to support workers performing these manual tasks. This contribution discusses a new approach to learning transitions of a finite state automaton representing the sequence of work tasks based on the video stream of a 3D depth camera. Preprocessed video data is fed into a three-stage classification scheme based on support vector machines. The results of the classification are then related to the state automation to trigger state transitions indicating the completion of a specific work task and the start of the next one. The proposed approach has been evaluated at an industrial assembly process of moderate complexity and shows very robust results with respect to disturbances caused by inaccurate object classification.
The increasing global competition demands continuous optimization of products and processes from companies in the process industry. Where conventional methods of Lean Management and Six Sigma reach their limits, new opportunities and challenges arise through increasing connectivity in the Industrial Internet of Things and machine learning. The majority of industrial projects do not reach the deployment or are isolated solutions, as the structures for data integration, training, deployment and maintenance of models are not established. This paper presents the conception of a reference architecture for machine learning in the process industry to support companies in implementing their own specific structures. The focus is on the development process and an exemplary implementation in the brewing industry.