Test Event Weighting Coefficients for a Machine Structure from Customer-sampled D-Optimally Placed Wireless Fatigue Nodes

Abstract

Wireless fatigue nodes (WFN) were used to create a mapping between proving ground test severity and observed 90th percentile customer application severity. This mapping takes the form of coefficients that are associated with each event in the test schedule. The locations of the wireless nodes were influenced by a D-Optimal row selection algorithm that sought to minimize the variance in the calculated test event coefficients. These D-Optimal sensor location selections were based upon simulated machine loads associated with the various events being passed through finite element models, resulting in calculated strain response and fatigue damage rates. After the WFN locations were chosen, the events were all executed experimentally in order to verify that the physically observed damage rate basis matrix would be well conditioned for the purposes of test event coefficient calculation. Finally, random customer-sampled damage rates from WFN were used to estimate the 90th percentile customer damage rate for each node location followed by a least-squares estimate of the test event coefficients. Future designs for the machine can be validated by conducting and/or simulating the selected test events and weighting their damage rates by the calculated coefficients in order to obtain a final 90th percentile customer severity composite fatigue life.