A Multiparameter RUL Prediction Method for UAV Lithium-Ion Battery Based on Physical Information

With the rapid development of unmanned aerial vehicles (UAVs), low-cost electrically driven UAVs have been widely used in various fields, and the battery module of UAVs is a critical component that directly affects their operational safety. The actual UAV operation scenario’s working condition mode is more complex. It is difficult to realize the full charge and discharge of the lithium-ion battery, so extracting the battery degradation characteristics is challenging. Moreover, the flight endurance and power parameters of UAVs are mainly determined by the capacity and internal resistance (IR) of Li-ion batteries. For UAVs, flight endurance and load power are equally important parameters. For UAVs in operation, it is crucial to simultaneously predict the capacity and IR, which is also a challenge. This article introduces the incremental capacity analysis (ICA) method to extract the half-peak width integral area and peak voltage as health indicators (HI) as degradation characteristics to characterize battery performance degradation. It establishes the multitask learning (MTL) model for indirect life prediction to address these issues. It builds an MTL model for indirect life prediction to solve these problems. The proposed MTL indirect prediction model is compared with the direct prediction and the single-task learning (STL) models. The results show that the MTL indirect prediction model has excellent prediction performance, with the relative periodic error (RUL_err) less than 1.5%, the root-mean-square error (RMSE), and the mean absolute percentage error (MAPE) held below 1.5% and 1.0%, respectively, and saves 55% of the computational time cost. Therefore, this approach is more suitable for embedded retrofit and applied to UAV-like working environments.