A hybrid approach for lithium-ion battery remaining useful life prediction using signal decomposition and machine learning DOI Creative Commons
Yibiao Fan, Ziyong Lin, Fan Wang

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 10, 2025

Lithium-ion batteries are widely used in many fields, and accurate prediction of their remaining useful life (RUL) was crucial for effective battery management safety assurance. In order to solve the problem reduced RUL accuracy caused by local capacity regeneration phenomenon during degradation, this paper proposed a novel method, which combined complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) technique an innovative hybrid strategy that integrated support vector regression (SVR) long short-term memory (LSTM) networks. First, CEEMDAN decompose data into high-frequency low-frequency components, thereby reducing impact regeneration. Subsequently, SVR model predicted component characterized main degradation trend, contained features using LSTM network optimized sparrow search algorithm (SSA). Finally, final obtained combining predictions two models. Experimental results on NASA public datasets showed method significantly outperformed existing methods: RMSE methods were all less than 0.0086 Ah, MAE 0.0060 R2 values higher 0.96, errors controlled within one cycle. This gave full play complementary advantages provided reliable solution lithium-ion batteries.

Language: Английский

A hybrid approach for lithium-ion battery remaining useful life prediction using signal decomposition and machine learning DOI Creative Commons
Yibiao Fan, Ziyong Lin, Fan Wang

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 10, 2025

Lithium-ion batteries are widely used in many fields, and accurate prediction of their remaining useful life (RUL) was crucial for effective battery management safety assurance. In order to solve the problem reduced RUL accuracy caused by local capacity regeneration phenomenon during degradation, this paper proposed a novel method, which combined complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) technique an innovative hybrid strategy that integrated support vector regression (SVR) long short-term memory (LSTM) networks. First, CEEMDAN decompose data into high-frequency low-frequency components, thereby reducing impact regeneration. Subsequently, SVR model predicted component characterized main degradation trend, contained features using LSTM network optimized sparrow search algorithm (SSA). Finally, final obtained combining predictions two models. Experimental results on NASA public datasets showed method significantly outperformed existing methods: RMSE methods were all less than 0.0086 Ah, MAE 0.0060 R2 values higher 0.96, errors controlled within one cycle. This gave full play complementary advantages provided reliable solution lithium-ion batteries.

Language: Английский

Citations

0