Accurate battery capacity estimation is essential for the effective and reliable operation of lithium-ion battery management systems. Battery impedance is a key parameter that encapsulates electrochemical information, closely correlating with the internal states of batteries. This study proposes a novel capacity estimation framework that effectively balances accuracy, efficiency, and practicality. Firstly, a novel feature extraction method is introduced to extract health features from the imaginary impedance at a single frequency. The extracted feature demonstrates a strong and stable correlation with battery degradation under various operation conditions, while significantly reducing data requirements. To address the impact of diverse degradation patterns on estimation accuracy, an initial adjustment method is applied to precisely retrace the relative degradation paths of batteries. The results show that the mean absolute percentage error of battery capacity estimation decreases from 15.65% to 2.87%. Additionally, a transformer-based capacity estimation model is developed, which integrates a feature fusion unit to explicitly eliminate the influence of temperature on model performance. As a result, the model's accuracy improves by over 28% under various thermal conditions.

@en

The State of Health (SOH) is a crucial component of battery management systems (BMSs), offering important health information and protection against unsafe usage. In this paper, an accurate model for SOH estimation of Li-ion batteries was developed, which is uniquely characterized by using only the imaginary part of impedance at a specific frequency for precise SOH estimation. Through the identification of the relationship between impedance at a specific frequency and capacity degradation using correlation coefficients, the feature data most closely related to battery aging was selected. Next, the battery aging modeling and SOH estimation were validated on nine batteries across three different temperatures using a Feed-forward Neural Network (FNN). The validation results indicated that the proposed method has a high estimation accuracy, achieving a Mean Absolute Percentage Error (MAPE) of merely 2.05% throughout the entire lifecycle of the battery 45C02 during tests at a temperature of 45°C.

@en

Traditional methods such as Steinmetz's equation (SE) and its improved variant (iGSE) have demonstrated limited precision in estimating power loss for magnetic materials. The introduction of Neural Network technology for assessing magnetic component power loss has significantly enhanced accuracy. Yet, an efficient method to incorporate detailed flux density information—which critically impacts accuracy—remains elusive. Our study introduces an innovative approach that merges Fast Fourier Transform (FFT) with a Feedforward Neural Network (FNN), aiming to overcome this challenge. To optimize the model further and strike a refined balance between complexity and accuracy, Multi-Objective Optimization (MOO) is employed to identify the ideal combination of hyperparameters, such as layer count, neuron number, activation functions, optimizers, and batch size. This optimized Neural Network outperforms traditionally intuitive models in both accuracy and size. Leveraging the optimized base model for known materials, transfer learning is applied to new materials with limited data, effectively addressing data scarcity. The proposed approach substantially enhances model training efficiency, achieves remarkable accuracy, and sets an example for Artificial Intelligence applications in loss and electrical characteristic predictions with challenges of model size, accuracy goals, and limited data.@en