Battery Energy Storage systems play a significant role in renewable energy grids, where fault detection is critical to ensuring reliability, safety, and optimal performance. Existing methods for fault detection, which are model-based Kalman filters and data-driven signal processing techniques, could be effective but face limitations in detecting slight issues or early signs of deterioration. This paper presents a novel hybrid approach to real-time fault detection which integrates signal processing techniques with Kalman filter-based state estimation. The proposed model is designed to detect faults and predict degradation trends, thereby enhancing the overall health monitoring of battery systems. This detailed methodology covers signal acquisition, preprocessing, feature extraction, Kalman filter design, and residual analysis. Kalman filter predicts system state and performs residual analysis to identify faults. Post-Kalman filter signal processing methods like wavelet transforms are used to analyze the residuals and detect fault-specific patterns, allowing for precise fault classification, including overcharging and short circuits. This hybrid approach combines the strengths of real-time state estimation and signal processing to advance real-time battery health monitoring, which results in a robust, scalable solution for fault detection in battery Storage Systems.