Unsupervised Identification of Discharging Curve Stage and State Evaluation for lead-Acid Battery

Lead-acid batteries are important backup power supply for equipment in substations, playing a crucial role in storing electrical energy and responding to emergency situations. Evaluating the battery status is of great significance for ensuring the stable operation of substations. During the check and discharge of lead-acid batteries, the DC discharge data of the battery contains the information about the battery's aging status, which is crucial for battery status evaluation. Based on subsequence quality evaluation and density based spatial clustering of applications with noise (DBSCAN) algorithm, an unsupervised identification method for the boundary between the initial discharge stage and the platform is firstly proposed in this paper. To meet the digital evaluation need of lead-acid battery pack, the voltage variation characteristics of the initial discharge stage and the discharge platform are extracted separately. The initial discharge voltage, the voltage difference of the initial discharge stage, and the voltage drop rate of the discharge platform are used as evaluation indicators to construct a multi-index evaluation model for battery health status based on local outlier detection (LOF) algorithm. The experiment shows that the method has a high level of intelligence in unsupervised recognition of discharge stage and can quantitatively evaluate battery status, which is beneficial to lead-acid battery status waring and battery maintenance.