电动垂直起降飞机（electric Vertical Take-Off and Landing aircraft, eVTOL）是绿色低空飞行器发展的重要趋势之一。为满足eVTOL安全可靠运行需求，需要电池管理系统结合飞行工况特性准确高效估算机载锂离子电池系统的健康状态。目前，大部分面向eVTOL应用的电池健康状态估计算法仅考虑了电池容量损失，缺乏根据飞行工况数据同时对电池能量衰退及功率衰减进行全面精准感知。针对上述问题，提出使用eVTOL起飞阶段电池电流和电压数据集作为估计算法输入参数，并选取电池容量、欧姆内阻和极化内阻作为健康指标综合表征电池性能衰退。为了提高多指标估算问题中模型的训练效率和估算精度，提出了一种基于多输出最小二乘支持向量回归的电池多健康指标联合估算方法，通过考虑输出变量间的耦合关系实现电池容量和内阻参数的快速准确估算。最后，基于eVTOL电池数据集进行了实验验证。对比分析结果表明，基于该方法估算所得电池容量和内阻的平均绝对百分比误差均低于2.5%，且模型训练时间控制在0.05 s以内，成功实现了面向eVTOL应用锂离子电池健康指标的全面、高效和准确估算。

Electric Vertical Take-Off and Landing aircraft (eVTOL) is one of the significant trends in the development of the renew-able low-altitude aircraft. To meet the safe and reliable operation requirements of eVTOL, it is necessary for the battery management system to accurately and efficiently estimate the state-of-health (SoH) of the on-board lithium-ion battery system based on the characteristics of the flight profile. At present, most SoH estimation algorithms for eVTOL applica-tions only consider the battery capacity loss, and lack the ability to comprehensively and precisely estimate both the battery energy and power degradation based on the data recorded during the specific flight phase. To address the aforementioned issue, the battery current and voltage measured during the take-off phase are selected as the inputs for the estimation algorithm, and the battery capacity, the Ohmic resistance, and the polarization resistance are utilized as the health indicators to comprehensively characterize the battery degradation. In order to improve the training effi-ciency and the estimation accuracy of the model for the multi-indicator estimation, a multi-output least squares support vector regression-based method for the joint estimation of multiple health indicators is proposed, which can achieve the rapid and accurate estimation of the battery capacity and resistances by considering the coupling relationship among the output parameters. Finally, the experimental verification is conducted based on the eVTOL Battery Dataset. The results of the comparative study demonstrate that based on the proposed method, the mean absolute percentage error of the estimated battery capacity and internal resistances are overall below 2.5%, and the model training time is within 0.05 s, successfully realizing a comprehensive, efficient, and accurate estimation of lithium-ion battery health indicators for eVTOL applications.