lithium-ion battery life cycle monitoring system based on IoT edge computing and cloud big data
The reliable lifecycle monitoring of lithium-ion batteries is critical for microgrid energy storage systems, where battery degradation, dynamic load variation, and abnormal operating conditions may directly affect system safety and energy efficiency. However, existing battery monitoring systems still have limitations in long-term degradation analysis, adaptive threshold adjustment, and edge–cloud collaborative decision-making. Conventional battery management systems mainly focus on local protection and short-term state estimation, IoT-based monitoring systems often emphasize data acquisition rather than lifecycle analysis, and cloud-only approaches may suffer from delayed responses to safety-critical events. To address these challenges, this study proposes a lithium-ion battery lifecycle monitoring system based on IoT edge computing and cloud big data for a BYD LiFePO4 battery pack in a 5 kW microgrid. The principal contribution of the proposed system is architectural and system-oriented rather than the development of a new standalone estimation or prediction algorithm. It organizes established battery monitoring, edge computing, and cloud analytics techniques into a lifecycle-oriented edge–cloud closed-loop architecture, in which latency-sensitive monitoring and warning tasks are executed at the edge, while long-term degradation analysis and model updating are performed in the cloud. The edge layer uses a BYD battery management unit and a Raspberry Pi node to collect and process voltage, current, temperature, and SOC data, while the cloud layer performs long-term data storage and advanced analysis through synchronized communication using Modbus TCP/IP and REST API. The system was validated using laboratory test data and continuous microgrid operation data collected from 2020 to 2022. The experimental results show that the proposed system improves SOC estimation accuracy under different operating conditions. Under dynamic load conditions, the proposed system reduced the SOC mean absolute error to 2.17%, compared with 3.85% for the Coulomb-counting/BMU-based SOC baseline without edge-side correction. The SOH prediction RMSE was also reduced to 2.13%, compared with 4.87% for the conventional capacity-ratio SOH baseline. These results demonstrate that the proposed system provides an effective solution for improving real-time response, prediction accuracy, and lifecycle management capability of lithium-ion batteries in microgrid applications.