Prediction method for propagating crack length of carbon-fiber-based composite double cantilever beam using its electromechanical behavior and particle filter

Abstract

Prognostics and health management (PHM) algorithms to predict the electromechanical behavior of double cantilever beam (DCB) composites were proposed using the particle filter method, which is a physics-based prognostics tool. The composites consisted of carbon fibers and carbon-glass hybrid fibers, whose warp and weft were carbon and glass fibers, respectively. The electrical resistance was changed by mechanical deformation as the cracks propagated. Finite element analysis (FEA) supported the empirical data in that mechanical stress along the through-thickness direction of each section and a detour of the electrical path via the crack tip caused resistance changes. Previous prognostics research which used data-driven or physics-based prediction were limited to predict future material behavior in real time due to long calculating time or high dimensional physical model. To develop the real time prognostic tool, the particle filter required sectioning and guiding, even though it only analyzed numerical values with a given physical equation. The sectioning and guiding processes were based on the empirical analysis and the FEA results. The handling was optimized by controlling the number of sections and coefficients of the physical equation. Therefore, this study successfully extracted the remaining useful lifetimes of composites within 4.54% and 5.71% of prediction error for carbon fiber and carbon-glass hybrid fiber based DCB composites, respectively.