Premium
Battery capacity degradation prediction using similarity recognition based on modified dynamic time warping
Author(s) -
Tao Laifa,
Lu Chen,
Yang Chao
Publication year - 2018
Publication title -
structural control and health monitoring
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.587
H-Index - 62
eISSN - 1545-2263
pISSN - 1545-2255
DOI - 10.1002/stc.2024
Subject(s) - dynamic time warping , similarity (geometry) , degradation (telecommunications) , battery (electricity) , computer science , test data , path (computing) , dynamic similarity , artificial intelligence , data mining , pattern recognition (psychology) , image (mathematics) , power (physics) , physics , quantum mechanics , programming language , reynolds number , turbulence , thermodynamics , telecommunications
Summary Battery degradation prediction is a significant recent challenge given the complex physical and chemical processes that occur within batteries, various working conditions, and limited performance degradation data and/or ground test data. In this study, we describe an approach called dynamic spatial time warping, which is used to determine the similarities of two arbitrary curves. Unlike classical dynamic time warping methods, this approach can maintain the invariance of curve similarity to the rotations and translations of curves, which is vital in curve similarity search and can recognize the intrinsic relationship between two curves. Moreover, it can be applied for battery degradation prediction even when rare data are available and do not require special assumptions, which fulfill the requirements of degradation prediction for batteries subject to extreme limited available data. The accuracy of this approach is verified by using both simulation data and NASA battery datasets. Results suggest that the proposed approach provides a highly accurate path of predicting battery degradation even with very limited data.