This technology converts analog signals, such as voltage and current, from individual battery cells or packs within a battery system into digital data. It then uses neural network models—including MNN, LSTM, and GRU—to precisely estimate the State of Charge (SOC) for each unit, physically switching out degraded cells or packs with spare batteries or energy storage systems.
Voltage deviations and energy imbalances between cells in a battery pack or between packs in a battery system have historically led to over-discharge and internal short circuits. These issues often resulted in battery failure, system downtime, and even safety hazards like fires.
This technology utilizes a relay circuit to monitor the status of each cell and pack individually, transmitting digitized signals via a converter to a neural network-based SOC estimation unit. Based on the estimated SOC, the control unit identifies degraded packs. If the number of packs requiring replacement is below a certain threshold, the system switches to a spare battery pack; if it exceeds the threshold, it switches to an energy storage system to maintain power output. Applicable to data center UPS systems requiring uninterruptible power, large-capacity battery systems for electric buses and ships, and backup power for communication base stations, this system allows operations to continue without shutting down the entire unit due to a single faulty pack.
N/A
