Neural network and statistical modeling techniques for electronic stress prediction
Conference Paper
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
Electronic components are constantly under stress due to factors such as signal density, temperature, humidity, and high current and voltage. There has been relatively little emphasis on stress level prediction under voltage stress. The purpose of this study was to develop an electronic temperature profile model for stress level prediction utilizing neural network and statistical approaches, such as multivariate regression models. Electronic components were removed from boards, subjected to different levels of stress, then replaced. An infrared camera was then used to capture information about component temperature changes over time under normal operating conditions. Neural network and statistical approaches were used to model temperature change profiles for components that had been stressed at different levels, and their predictive ability was compared. Separate data sets were used for model development and model verification. Neural network prediction rates were around 70%, compared to 30% for the statistical approach. Experiments were also conducted to evaluate the noise-tolerance of the neural network model. The neural network accommodated the presence of noise much more easily than statistical approaches. Resilient back propagation learning functions performed better than functions studied. A 3-2-1 topology performed better than 3-3-1 or 3-2-2-1 topologies.
