Abstract The paper presents a novel data-driven method for fault detection and isolation of control moment gyroscopes onboard satellites. The proposed method uses the Chebyshev Neural Network and genetic algorithm in conjunction with satellite attitude rate data only. A data-driven model is first developed that fuses the symmetric property of the data and the system orientation property of actuators that reduces the need for historical data by 93.75%. Next, the data is trained using Chebyshev Neural Network. An adaptive threshold-based fault detection algorithm is applied to detect the faults in the spin and gimbal motors of the control moment gyroscopes. A fault isolation scheme is developed wherein an objective function is optimized using a genetic algorithm for different cases of system parameters. The proposed scheme has a success rate of 93.5% in isolating faults of 8 motors (4 gimbal and 4 spin) that can fail in 254 different ways. Overall, the proposed methodology can be regarded as a promising fault diagnostic tool for satellites using limited historical data and measurements.

中文翻译：

---

一种新的数据驱动卫星控制力矩陀螺故障检测与隔离方法

摘要 本文提出了一种新的数据驱动方法，用于卫星控制力矩陀螺仪的故障检测和隔离。所提出的方法仅结合卫星姿态率数据使用切比雪夫神经网络和遗传算法。首先开发了一个数据驱动模型，该模型融合了数据的对称特性和执行器的系统方向特性，将历史数据的需求减少了 93.75%。接下来，使用 Chebyshev 神经网络训练数据。应用基于自适应阈值的故障检测算法来检测控制力矩陀螺仪的自旋和万向节电机的故障。开发了一种故障隔离方案，其中针对系统参数的不同情况使用遗传算法优化目标函数。建议方案的成功率为 93。5% 的 8 个电机（4 个万向节和 4 个自旋）可以以 254 种不同的方式发生故障 总体而言，所提出的方法可以被视为使用有限历史数据和测量的卫星故障诊断工具。