Online Fault Diagnosis, Prognosis, and Health Monitoring of Small Satellites

该研究的题目是“空间系统的在线故障诊断、预测和健康监测”。“应当指出，该方案的重点将是小型卫星星座健康监测，有两个主要目标，（1）为星座中的小型卫星开发高保真可配置动态模型，（2）使用目标1中开发的模型开发卫星星座健康监测算法。针对第一个目标，建立了星座中小卫星的高保真可配置动力学模型，利用PI方法建立了具有反作用飞轮（RW）和控制力矩陀螺（CMG）的整体卫星的高保真模型，并采用非线性滑模控制器（SMC）实现了三轴稳定。这些模型只包括一个卫星姿态控制系统（ACS），而在星座和编队飞行轨道控制也需要纳入站保持。因此，该模型需要扩展到包括姿态和轨道控制系统（AOCS）。单个卫星的现有模型将扩展到多个卫星。该任务包括两个部分：（a）为编队中的卫星开发动力学模型（B）为编队组件开发控制器。星座组装可以在三种主要拓扑中实现：（a）集中式，（B）分层式，（c）分布式/分散式，并且最终模型需要能够在这三种拓扑之间切换。 对于第二个目标，即空间系统的故障诊断和预测，应研究新的方法。应探索将基于物理的方法和数据驱动的方法相结合的混合方法。在基于物理的方法中，应使用不可测量系统参数的状态/参数估计来最小化检测延迟。在数据驱动方法中，除了目标1中开发的模型的模拟数据外，模拟数据还应与机器学习算法一起沿着使用，以对系统、其性能和可能的故障模式进行建模。此外，应开发故障预测算法，以估计机组的剩余使用寿命，并建议潜在的补救措施。这项工作将为加拿大航天工业提供用于小型卫星星座组件的高保真可扩展和可配置模型，这将有助于将加拿大在小型卫星发射领域的市场份额从目前的2.2%提高到2.2%。这为空间发现提供了新的机会，而单一卫星飞行任务的初始成本和经常性成本较高。此外，该计划还将提供智能诊断和预后技术以及经过培训的人员（HQP），使该行业能够提高系统的可用性并降低数百万美元的维护成本。该计划的成功为出口提供了有利的贸易平衡，使加拿大成为这一领域的领导者。

The topic of the research being proposed is "Online Fault Diagnosis, Prognosis, and Health Monitoring of Space Systems." It should be noted that the focus of the program will be on small satellite constellation health monitoring with two major objectives, (1) development of the high-fidelity configurable dynamics model for small satellites in constellation and (2) developing health monitoring algorithms for the satellite constellation using the developed models in objective 1. For the first objective, development of the high-fidelity configurable dynamics model for small satellites in constellation, high-fidelity models for a monolithic satellite with reaction wheels (RW) and control moment gyros (CMG) with 3-axis stabilized via nonlinear sliding mode controller (SMC) are developed by the PI. These models only incorporate a satellite attitude control system (ACS) while in constellation and formation flying orbital control also needs to be incorporated for station keeping. Hence, the model needs to be expanded to include attitude and orbital control system (AOCS). Available models from individual satellites will be extended to multiple satellites. This task comprises of 2 parts: (a) developing dynamics models for the satellites in the formation (b) developing controllers for the formation assembly. The constellation assembly can be achieved in three main topologies (a) Centralized, (b) Hierarchical, (c) Distributed/Decentralized and the final model needs to be able to switch between these three topologies. For the second objective, fault diagnosis and prognosis of space systems, new approaches shall be investigated. Hybrid approaches, combining physics-based and data-driven methods, shall be explored. In the physics-based approach, state/parameter estimation of the non-measurable system parameters shall be used to minimize delay in detection. In the data-driven approach, simulated data, in addition to simulated data from models developed in objective 1, shall be used along with machine learning algorithms to model the system, its performance, and possible failure modes. Furthermore, fault prognosis algorithms shall be developed to estimate remaining useful life of the unit(s) and suggest potential remedial actions. This work will provide Canadian space industries with high-fidelity scalable and configurable models for small satellite constellation assemblies that would help increase the Canadian market share in the small satellite launch sector from 2.2% as it stands today. This yields new opportunities for space discovery compared to monolithic satellite missions to due initial and recurring costs. Furthermore, this program will provide intelligent diagnosis and prognosis technologies and trained personnel (HQP) that enable the industry to increase the system's availability and reduce maintenance costs by millions of dollars. The success of this program offers favorable trade balance to exports, presenting Canada as a leader in this area.