A change-point based reliability prediction model using field return data

• A practical yet accurate reliability prediction model using short-term field data.
• A change point detection method using forward and backward data analysis.
• Constructed on a Weibull-exponential hazard rate scheme.
• Applicable to high-volume electronic products.

In this study, we propose an accurate reliability prediction model for high-volume complex electronic products throughout their warranty periods by using field return data. Our model has a specific application to electronics boards with given case studies using 36-month warranty data. Our model is constructed on a Weibull-exponential hazard rate scheme by using the proposed change point detection method based on backward and forward data analysis. We consider field return data as short-term and long-term corresponding to early failure and useful life phases of the products, respectively. The proposed model is evaluated by applying it to four different board data sets. Each data set has between 1500 and 4000 board failures. Our prediction model can make a 36-month (full warranty) reliability prediction of a board with using its field data as short as 3 months. The predicted results from our model and the direct results using full warranty data match well. This demonstrates the accuracy of our model. We also evaluate our change point method by applying it to our board data sets as well as to a well-known heart transplant data set.