This paper addresses the critical challenge of accurately predicting both the load demand and state-of-health (SOH) for user-side energy storage systems under time-specific operation strategies. By leveraging advanced machine learning models and real-time operational data, the proposed methodology aims to enhance system efficiency, prolong battery lifespan, and optimize energy dispatch decisions. The study incorporates a comprehensive analysis of temporal patterns in energy consumption and degradation mechanisms, enabling precise forecasts that account for dynamic usage scenarios and environmental factors. Experimental results demonstrate that the proposed approach achieves significant improvements in prediction accuracy compared to conventional techniques, with reductions in error rates and enhanced adaptability to varying operational conditions. Furthermore, the integration of SOH prediction facilitates proactive maintenance scheduling, thereby minimizing downtime and operational costs. Overall, this work provides a robust framework for optimizing energy storage management, which is crucial for supporting sustainable energy infrastructures.