Multi-Source Data-Driven Switching Strategy Profiling for User-Side Energy Storage

With the growing complexity of user-side energy storage applications, multi-source operational data, including active power trajectories, operating state records, valid operating day information, and time-of-use tariff windows, increasingly contain standby periods and atypical conditions, making the true switching strategies of devices difficult to identify. Accordingly, this study aims to extract stable typical operating patterns from multi-source real-world operational data and to establish a hierarchical switching-strategy profile for practical business use. To achieve this objective, this paper proposes a multi-source data-driven switching strategy profiling method based on typical curves and hierarchical clustering using real-world operational data, in which a standby-interference-resistant typical-curve extraction algorithm first reconstructs the core output pattern of each device by adaptively screening effective operating conditions. On this basis, a hierarchical shape–feature profiling framework is constructed: OPTICS is used to determine baseline physical shapes, and temporal anchoring together with execution consistency features is used for fine-grained K-Means subdivision. A case study using 151 user-side energy storage devices from January 2023 to December 2025 shows that the proposed method identifies three macro operating shapes and six micro execution types, revealing differences in duration, timing deviation, and execution quality. The results provide interpretable support for refined operation, performance evaluation, and asset management of distributed user-side energy storage.