Name: Neda Shafiei
Title: Estimating The Reliability of a New Consumer Product Using User Survey Data and Reliability Test Data
Date: 7/27/2022 at 11am via Zoom only
Zoom Link: https://umd.zoom.us/j/9968970907
Committee Members:
Mohammad Modarres, Chair
Jefferey Herrmann
Abhijit Dasgupta
Katrina Groth
Vasiliy Krivtsov
Mohamad Al-Sheikhly, Dean’s Representative
Abstract: Collecting data for estimating the reliability of a new consumer product is challenging. Using test data for estimating reliability is usually insufficient because, in reliability tests, the number of samples, test stress levels, and test times are restricted. Collecting and analyzing reliability data of the same class of product obtained from user surveys offer a cost-effective and quick way to obtain the field prior reliability characteristics of the new device. User survey data, however, is biased. This dissertation provides a guideline for designing a reliability-informed survey. An approach is proposed that corrects any bias in the survey responses through the Kullback-Leibler (KL) divergence method and uses bias-corrected survey data and test data about the similar product and test data about the new product to estimate the reliability model. The approach considers a product with many or several cycles to failure. The application of the approach is illustrated using the simulated survey and test datasets for a consumer electronic product with the failure mode of cracking caused by accidental drops.