Abstract:Predicting the remaining useful life (RUL) of lithium-ion batteries is crucial for enhancing their reliability and safety. Addressing the issue of inaccurate RUL predictions caused by the nonlinear decay resulting from capacity regeneration, this paper proposes an indirect lithium-ion battery RUL prediction method based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and convolutional neural network (CNN)–bidirectional gated recurrent unit (BiGRU). The method extracts Health Indicators (HI) from the battery-charging stage and employs CEEMDAN to decompose HI into several components. These components are then input into a component prediction model for forecasting. Finally, the predicted component results are fused and input into a capacity prediction model to achieve indirect RUL prediction. Validation is conducted using the lithium-ion battery dataset provided by NASA. The results indicate that, under prediction starting points (STs) of 80 and 100, the maximum average absolute errors do not exceed 0.0096 and 0.0081, and the maximum root mean square errors do not exceed 0.0196 and 0.0115, demonstrating high precision and reliability.

What problem does this paper attempt to address?

### Problems Addressed by the Paper The paper aims to address the issue of inaccuracy in predicting the remaining useful life (RUL) of lithium-ion batteries, particularly the impact of nonlinear degradation caused by capacity regeneration. The paper proposes an indirect lithium-ion battery RUL prediction method based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and Convolutional Neural Network-Bidirectional Gated Recurrent Unit (CNN-BiGRU). ### Main Contributions 1. **Health Indicator (HI) Extraction**: By analyzing the current and voltage variation characteristics during the battery charging phase and using correlation analysis, the paper extracts indirect health indicators (HI) that highly reflect battery capacity degradation, optimizing the input data for battery RUL prediction. 2. **Multi-Scale Decomposition**: The CEEMDAN method is used to perform multi-scale decomposition on the HI, separating the overall trend part and the fluctuation part related to capacity regeneration into different components, suppressing the regeneration phenomenon and improving prediction accuracy. 3. **CNN-BiGRU Prediction Model**: A prediction model based on CNN-BiGRU is proposed, which effectively extracts global and local features of time series data using CNN and Patch-Aware design. The BiGRU network is introduced to capture long-term dependencies in time series data, and a dual prediction head mechanism is adopted to further enhance the model's prediction capability. 4. **Experimental Validation**: Experimental validation is conducted using the lithium-ion battery dataset provided by NASA. The results show that when the starting point (ST) for prediction is 80 and 100, the maximum mean absolute error does not exceed 0.0096 and 0.0081, and the maximum root mean square error does not exceed 0.0196 and 0.0115, demonstrating the high accuracy and reliability of the proposed method. ### Method Overview 1. **Indirect Health Indicator Extraction**: By analyzing the current and voltage variation characteristics during the battery charging phase, three indirect health indicators (HI) are extracted, including constant current charging time, current variation within 1000 seconds during constant voltage charging, and voltage variation within 500 seconds before charging. 2. **CEEMDAN Decomposition**: The CEEMDAN algorithm is used to decompose the HI, separating the overall trend part and the fluctuation part related to capacity regeneration, improving the accuracy of subsequent predictions. 3. **CNN-BiGRU Prediction Model**: A prediction model based on CNN-BiGRU is constructed, utilizing CNN to extract global and local features of time series data, BiGRU to capture long-term dependencies in time series, and a dual prediction head mechanism to further enhance the model's prediction performance. 4. **Capacity Prediction Model**: The prediction results of the decomposed components are integrated into the capacity prediction model to achieve RUL prediction, thereby mitigating the impact of capacity regeneration on RUL prediction accuracy. Through the above methods, the paper effectively addresses the key issues in lithium-ion battery RUL prediction and demonstrates its high accuracy and reliability in practical applications.

Indirect Prediction of Lithium-Ion Battery RUL Based on CEEMDAN and CNN-BiGRU

A Bayesian Mixture Neural Network for Remaining Useful Life Prediction of Lithium-Ion Batteries

Lithium-Ion Battery Remaining Useful Life Prediction Based on Hybrid Model

Remaining Useful Life Estimation of Lithium-Ion Batteries Based on Small Sample Models

Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Neural Network and Adaptive Unscented Kalman Filter

Early prediction of remaining useful life for lithium-ion batteries based on CEEMDAN-transformer-DNN hybrid model

Predicting the Future Capacity and Remaining Useful Life of Lithium-Ion Batteries Based on Deep Transfer Learning

Remaining Useful Life Prediction for Lithium-Ion Battery Based on CEEMDAN and SVR.

Remaining useful life indirect prediction of lithium-ion batteries using CNN-BiGRU fusion model and TPE optimization

Remaining Useful Life Prediction of Lithium‐Ion Batteries Based on a Combination of Ensemble Empirical Mode Decomposition and Deep Belief Network–Long Short‐Term Memory

Prediction of the Remaining Useful Life of Lithium-Ion Batteries Based on the 1D CNN-BLSTM Neural Network

Remaining useful life prediction for lithium-ion batteries with a hybrid model based on TCN-GRU-DNN and dual attention mechanism

A Remaining Useful Life Indirect Prediction Method for Lithium-Ion Batteries Based on SA-DBN

Integrated Method of Future Capacity and RUL Prediction for Lithium‐Ion Batteries Based on CEEMD‐Transformer‐LSTM Model

A Hybrid Data-Driven Method Based on Data Preprocessing to Predict the Remaining Useful Life of Lithium-Ion Batteries

Remaining Useful Life Prediction for Lithium-Ion Batteries Based on a Hybrid Deep Learning Model

Remaining Life Prediction of Li-Ion Batteries Considering Sufficiency of Historical Data

Hybrid Data-Driven Approach for Predicting the Remaining Useful Life of Lithium-Ion Batteries

Prediction of Remaining Useful Life for Lithium‐Ion Batteries Using Complete Ensemble Empirical Mode Decomposition with Adaptive Noise for Feature Analysis, and Bidirectional Long Short‐Term Memory Coupled with a Gaussian Process Regression Model

A complete ensemble empirical mode decomposition with adaptive noise deep autoregressive recurrent neural network method for the whole life remaining useful life prediction of lithium-ion batteries

Residual life prediction of lithium-ion batteries based on data preprocessing and a priori knowledge-assisted CNN-LSTM