Multi-Scale Signal Processing for High-Voltage Lithium-Ion Battery Performance Optimization: Noise Suppression and Feature Enhancement via Deep Learning and Time-Frequency Analysis

High-voltage lithium-ion batteries (HV-LIBs) are essential for energy-intensive applications such as electric vehicles and renewable energy storage, but their efficient operation requires advanced battery management systems (BMS). This paper presents a comprehensive framework integrating signal processing and deep learning to optimize battery performance through accurate state-of-charge (SOC) and state-of-health (SOH) estimation. Traditional approaches like Kalman filtering and particle filtering, while effective for noise suppression, fail to capture the multi-scale and non-linear nature of HV-LIB signals under dynamic conditions. The proposed methodology leverages wavelet transform for multi-resolution signal analysis and a deep neural network (DNN) for predictive modeling, enabling robust feature extraction and accurate battery health monitoring. Experimental results demonstrate significant noise reduction, with a 30 dB increase in signal-to-noise ratio (SNR), and SOC and SOH prediction accuracies of 95% and 92%, respectively. Scalability tests across varying battery capacities and configurations further validate the framework’s adaptability. This approach bridges the gap between traditional signal processing and modern machine learning, offering a scalable, real-time solution for advanced BMS applications. The proposed framework shows great potential for enhancing the safety, efficiency, and lifespan of HV-LIBs in diverse operational environments.