State of Health Estimation using Convolutional and Bidirectional LSTM Neural Networks tuned by Bayesian Optimization
2026-08-31 • Machine Learning
Machine Learning
AI summaryⓘ
The authors propose a new way to estimate the health of a battery (State of Health) using a mix of deep learning models, specifically CNN and BiLSTM, combined with Bayesian Optimization to fine-tune the model settings. They tested three deep learning setups and found that adding fully connected layers in the middle gave the best results. They also carefully selected input features like capacity, voltage, and specific voltage analyses to improve accuracy. Their method was validated on three public datasets, including one created by the authors, and evaluated using common error metrics and computational efficiency measures.
State of Health (SOH)Convolutional Neural Network (CNN)Bidirectional Long Short-Term Memory (BiLSTM)Bayesian OptimizationFully Connected (FC) LayersIncremental Capacity Analysis (ICA)Differential Voltage Analysis (DVA)Mean Absolute Error (MAE)Root Mean Squared Error (RMSE)FLOPs
Authors
Panagiotis Eleftheriadis, Foivos Georgios Kyrgios, Sonia Leva
Abstract
In this research, a novel framework is proposed for the SOH estimation, which employs a hybrid deep learning architecture of a concatenation of a Convolution Neural Network (CNN) and a Bidirectional Long Short-Term Memory (BiLSTM) Neural Network (NN) with the integration of Bayesian Optimization-based hyperparameter tuning for the network. Three different deep learning architectures are being evaluated: standalone recurrent models, CNN-RNN architectures and CNN-RNN combinations enhanced with intermediate Fully Connected (FC) layers. Among the three, the model with the intermediate FC layers demonstrated the highest predictive accuracy. A comprehensive feature engineering approach combines capacity (Q), voltage (V), Incremental Capacity Analysis (ICA), and Differential Voltage Analysis (DVA), with systematic evaluation of multiple combinations to identify the optimal input representation. To validate the proposed method, three publicly available datasets were utilized, ensuring reproducibility of the results, two from external sources and one developed by the author of this study using a unique experimental setup. The comparison study was performed using the Mean Absolute Error (MAE), the Root Mean Squared Error (RMSE) and the FLoating-point OPerations (FLOPs) as evaluation metrics.