This paper describes a model-based prognosis method for the health management of a planetary rover. Using a hybrid model of the rover, including a continuous part and a discrete part, a prognoser is generated that relies on the hybrid particle Petri nets (HPPN) data structure. The prognosis process uses the current diagnosis of the system to predict its future states and to determine its end of life (EOL) or its remaining useful life (RUL). The HPPN-based prognoser is initialized with a stochastic scaling algorithm that selects the diagnosis hypotheses with the highest beliefs. The SSA provides a compromise between performance and available computational resources through the setting of scaling parameters. The prognoser then uses the future commands to determine the hypotheses over the rover future trajectory and the RUL/EOL. The set of future hypotheses associated with their belief degrees forms the current rover prognosis. The prognosis method is tested on different scenarios, with different scaling parameters, considering whether the future commands are known or not. Experimental results show that the approach is robust to real system data and computational performance constraints.

中文翻译：

---

基于混合粒子 Petri 网的行星探测器预测

本文描述了一种用于行星探测器健康管理的基于模型的预测方法。使用漫游车的混合模型，包括连续部分和离散部分，生成依赖于混合粒子 Petri 网 (HPPN) 数据结构的预测器。预测过程使用系统的当前诊断来预测其未来状态并确定其寿命终止 (EOL) 或剩余使用寿命 (RUL)。基于 HPPN 的预测器使用随机缩放算法进行初始化，该算法选择具有最高置信度的诊断假设。SSA 通过设置缩放参数在性能和可用计算资源之间进行折衷。然后，预测器使用未来命令来确定关于漫游车未来轨迹和 RUL/EOL 的假设。与其可信度相关的一组未来假设形成了当前的漫游者预测。考虑未来命令是否已知，预测方法在不同场景下使用不同的缩放参数进行测试。实验结果表明，该方法对真实系统数据和计算性能约束具有鲁棒性。