The satellite must continue working properly under different working environments and working loads. The power system is an essential component. Due to different working tasks, loads, and attitudes, a power system has many diverse working states. Therefore, it is necessary to accurately recognize the working state online for fault diagnostics and health management. However, under different working loads, measurement errors, environmental noise, environmental interference, and other uncertain factors, the output voltage value of a satellite power system has different levels of uncertainties. If these uncertainties and various working states are not considered, the recognition results can be of low quality. To address this problem and the uncertainty factors, we present an online dynamic working-state recognition system for satellite power systems based on uncertain data classification. In the system, we first explore the uncertain-data clustering center to model the working state. Then, with a slide-window processing strategy, we compute the distances between the uncertain cluster centers and the uncertain voltage data for the satellite power system online. Thus, we can obtain more accurate dynamic working-state recognition results. The evaluation results of real data demonstrate that the presented system is valid for working-state recognition and can be applied to a satellite power system.

卫星必须在不同的工作环境和工作负载下继续正常工作。电力系统是必不可少的组件。由于工作任务，负载和态度不同，电源系统具有许多不同的工作状态。因此，有必要在线准确地识别工作状态，以进行故障诊断和健康管理。但是，在不同的工作负载，测量误差，环境噪声，环境干扰和其他不确定因素的影响下，卫星电源系统的输出电压值具有不同程度的不确定性。如果不考虑这些不确定性和各种工作状态，则识别结果的质量可能很低。为了解决这个问题和不确定因素，我们提出了一种基于不确定数据分类的卫星动力系统在线动态工作状态识别系统。在系统中，我们首先探索不确定数据聚类中心以对工作状态进行建模。然后，通过滑动窗口处理策略，我们为卫星电力系统在线计算了不确定的群集中心和不确定的电压数据之间的距离。因此，我们可以获得更准确的动态工作状态识别结果。实际数据的评估结果表明，所提出的系统对于工作状态识别是有效的，可以应用于卫星电力系统。我们在线计算了不确定的星团中心和不确定的电压数据之间的距离。因此，我们可以获得更准确的动态工作状态识别结果。实际数据的评估结果表明，所提出的系统对于工作状态识别是有效的，可以应用于卫星电力系统。我们在线计算了不确定的星团中心和不确定的电压数据之间的距离。因此，我们可以获得更准确的动态工作状态识别结果。实际数据的评估结果表明，所提出的系统对于工作状态识别是有效的，可以应用于卫星电力系统。