Predicting the behavior of solution alternatives within product improvement processes
Editor: Udo Lindemann, Srinivasan V, Yong Se Kim, Sang Won Lee, John Clarkson, Gaetano Cascini
Author: Abramovici, Michael; Lindner, Andreas; Dienst, Susanne; Fathi, Madjid
Institution: 1: Ruhr University Bochum, Germany; 2: University of Siegen, Germany
Nowadays an increasing number of industrial products are equipped with sensors, allowing a complete monitoring of the product and its working conditions during the use phase. The data generated by such sensors is mainly used for maintenance purposes. The evaluation of that data can offer valuable input for the improvement of existing product generations. The presented approach offers a methodology to identify improvement potentials and to support decisions within product improvement processes. This approach is based on prescriptive decision theory and uses feedback data in addition to product-specific characteristics and properties. A prediction of future product solution alternatives behavior is realized on the basis of object-oriented Bayesian Networks. The validation of the proposed solution has been demonstrated on the basis of decision processes for the improvement of centrifugal pumps.