Hybrid ANFIS-BiLSTM-ANFIS model for remaining useful life prediction of Li-ion batteries in electric trains

• Innovative hybrid method: Presenting a hybrid ANFIS-BiLSTM-ANFIS method for more accurate battery life prediction of electric trains. • Increased safety and economic efficiency: Improving battery health monitoring and optimizing the efficiency of electric trains. • Superior prediction accuracy: The proposed model outperforms other machine learning algorithms in predicting the remaining useful life. • Practical application in electric rail systems: Using real simulation data with a constant and repeatable train speed profile. In recent years, the use of batteries as sources of power in electric trains has expanded dramatically due to developments in battery technology and the desire for sustainable and clean transportation. The importance of battery state-of-health (SOH) monitoring is growing due to the widespread use of batteries, and it serves a crucial function in enhancing safety. Accurate and reliable estimation of battery life to evaluate their economic efficiency is one of the fundamental challenges in the field of using batteries. This paper introduces a novel hybrid method that combines the characteristics of the Adaptive Neuro-Fuzzy Inference System (ANFIS) with a Bidirectional Long Short-Term Memory (BiLSTM) to increase the accuracy of the remaining useful life prediction of lithium-ion batteries in electric trains. This simulation assumes a steady and repeatable train speed profile. The battery life is estimated using machine learning (ML) models by applying the train's current usage to the battery over a specific time period. The proposed model is assessed in two stages of training and testing with other ML algorithms based on the root-mean-square error (RMSE), mean absolute error (MAE), and R-squared error ( R 2 ) metrics. The evaluation results in the test phase demonstrate that the proposed model achieves a RMSE, MAE, and R² that are at least 31.7%, 41.9%, and 2% more accurate than other models, respectively. These findings highlight the superior performance and accuracy of the proposed model in predicting the remaining useful life (RUL) of batteries in electric trains.