The increase in quality standards in the automotive industry requires specifications to be propagated across the supply chain, a challenge exacerbated in domains where the quality is subjective. In the daily operations of ThyssenKrupp Presta AG, requirements imposed on the vibroacoustic quality of steering gear need to be passed down to their subcomponents. We quantify the influence of ball nut assemblies on the steering gear by finding optimal encodings of their respective vibroacoustic signals, iteratively maximizing the correlation of their order spectra under orthogonality constraints. We compare the performance of linear and non-linear variants of this approach known as Canonical Correlation Analysis and establish the superiority of the neural network based variant in terms of attainable correlation. The practical relevance of our findings is guaranteed, since the visualization of the weights enables the identification of core influence areas.
Technical report, Alpen-Adria Universität Klagenfurt, Mathematics, Optimization Group, TR-AAUK-M-O-28-06-18, 2018.