Design of wireless battery management system monitoring and automated alarm system based on improved long short-term memory neural network.

The battery management system (BMS) can intelligently manage and maintain each battery unit while monitoring its status, thereby preventing any possible overcharge or over-discharge of the battery. In BMS research, battery state parameter collection and analysis are essential. However, traditional data collection methods require personnel to be present at the scene, leading to offline data acquisition. Therefore, this study aimed to develop a wireless BMS monitoring and alarm system based on socket connection that would enable researchers to observe the operating parameters and problem details of the battery pack from a distance. A device like this effectively raises the battery's level of cognitive control. In the study, the researchers first designed the overall scheme of the BMS remote monitoring system, followed by building a wireless BMS monitoring and alarm system. Performance evaluations of the system were then conducted to confirm its effectiveness. A Long Short-Term Memory (LSTM) network enhanced by the Batch Normalization (BN) technique was applied to the time series data of battery parameters to solve the large accuracy inaccuracy in battery state of charge estimate. Furthermore, the Denoise Auto Encoder (DAE) algorithm was utilized to denoise the data and reduce the model's parameter dependence. The accuracy and robustness of the estimation are improved, and the model error is gradually stabilized within 5%.