Image-Based Prognostics Using Penalized Tensor Regression

TL;DR

This paper introduces a novel tensor regression approach for predicting system lifetime from degradation image sequences, combining tensor algebra with regression techniques to handle high-dimensional data effectively.

Contribution

It develops a penalized tensor regression framework with tensor decompositions and optimization algorithms for accurate prognosis using image streams.

Findings

Validated on simulated data showing high accuracy

Successfully applied to infrared machinery degradation images

Demonstrated effective handling of high-dimensional tensor data

Abstract

This paper proposes a new methodology to predict and update the residual useful lifetime of a system using a sequence of degradation images. The methodology integrates tensor linear algebra with traditional location-scale regression widely used in reliability and prognosis. To address the high dimensionality challenge, the degradation image streams are first projected to a low-dimensional tensor subspace that is able to preserve their information. Next, the projected image tensors are regressed against time-to-failure via penalized location-scale tensor regression. The coefficient tensor is then decomposed using CANDECOMP/PARAFAC (CP) and Tucker decompositions, which enables parameter estimation in a high-dimensional setting. Two optimization algorithms with a global convergence property are developed for model estimation. The effectiveness of our models is validated using a simulated dataset and infrared degradation image streams from a rotating machinery.