Active fault-tolerant systems are crucial for ensuring the reliability and safety of spacecraft attitude control, especially when dealing with actuator faults that could jeopardize mission success. This article introduces a comprehensive framework for fault detection and identification (FDI) and fault-tolerant control (FTC) tailored to meet stringent real-time processing requirements and limited onboard resources. We propose a neural network (NN)-based fault identification approach, termed FaultNet, which leverages long short-term memory (LSTM) architectures to achieve millisecond-level inference ( $\approx3$ ms). Compared to traditional observer-based FDI methods with multisecond convergence times, FaultNet offers a lightweight solution (model size $\approx 550$ kB) suitable for onboard deployment, delivering rapid and precise fault estimation. It enables accurate identification of both constant and time-varying faults, including effectiveness loss and additive bias. For fault-tolerant attitude control in spacecraft, we develop a reinforcement learning (RL)-based controller, incorporating carefully designed state representations and reward functions to ensure robust performance. The proposed framework is rigorously evaluated across four customized fault scenarios—including both constant and time-varying faults—and achieves rapid convergence (average attitude alignment within 20 s) and sustained stability (average convergence duration exceeding 280 s). The controller effectively compensates for actuator faults, maintaining high control precision with minimal attitude oscillation and consistent recovery performance. Simulation outcomes validate the NN-based FDI’s efficiency and accuracy, as well as the RL-based FTC’s (RL-FTC’s) reliability, showcasing minimal fault estimation errors and consistent fault recovery performance. The source code and associated data are available at https://github.com/LUWENL/fault-tolerant-control

The tracking of moving targets using satellite constellations presents significant challenges due to the need for time-sensitive and effective scheduling solutions. While static target observation has been extensively studied, stereo tracking of moving targets remains a critical research gap, particularly when considering the complexities introduced by real-time movement and operational constraints. To address these challenges, this paper proposes a pioneering approach that integrates stereo tracking with a Parallel Island Differential Evolutionary Algorithm (PIDEA) for multi-satellite mission scheduling. The PIDEA harnesses the computational power of Graphics Processing Units (GPUs) and employs island evolution strategies to balance computational efficiency with solution diversity, ensuring timely and effective scheduling in dynamic scenarios. Additionally, a reinforcement learning-based attitude control system is introduced to enable agile satellites to maintain accurate and stable tracking of moving targets, even under challenging conditions. To further enhance operational adaptability, we incorporate event-driven mechanisms to dynamically trigger rescheduling when significant changes occur, such as satellite availability or energy constraints. Extensive experiments conducted in multiple moving target tracking scenarios demonstrate the effectiveness and efficacy of the proposed method. The results validate its ability to generate near-optimal scheduling solutions that meet the dual demands of time sensitivity and tracking effectiveness, marking a significant step forward in autonomous satellite mission planning for dynamic environments. The source code of PIDEA is available at https://github.com/LUWENL/PIDEA.