Deep Generative Models-Based Anomaly Detection for Spacecraft Control Systems

Abstract

A spacecraft attitude control system provides mechanical and electrical control to achieve the required functions under various mission scenarios. Although generally designed to be highly reliable, mission failure can occur if anomalies occur and the attitude control system fails to properly orient and stabilize the spacecraft. Because accessing spacecraft to directly repair such problems is usually infeasible, developing a continuous condition monitoring model is necessary to detect anomalies and respond accordingly. In this study, a method for detecting anomalies and characterizing failures for spacecraft attitude control systems is proposed. Herein, features are extracted from multidimensional time-series data of a simulation of the attitude control system. Then, the artificial neural network learning algorithms based on two types of generation models are applied. A Bayesian optimization algorithm with a Gaussian process is used to optimize the hyperparameters for the neural network to improve the performance. The performance is evaluated based on the reconstruction error through the algorithm using the newly generated data not used for learning as input data. Results show that the detection performance depends on the operating characteristics of each submode in the operation scenarios and type of generation model. The diagnostic results are monitored to detect anomalies in operation modes and scenarios.

Reframe the Field of Aerospace Engineering Via Machine Learning: Application and Comparison

“…The contrast between the OC-SVM and the K-nearest neighbor (k-NN) classification algorithm based on the Local Outlier Factor (LOF) also indicates OC-SVM better performance especially for symbolic data. Another study, Ahn et al, is also used for anomaly detection，in which an approach for automatically identifying and diagnosing anomalies or failures for attitude control systems is presented [9]. In this method, features are taken from a simulated multidimensional time-series dataset of attitude control.…”

Deep-Learning-Based Approach to Anomaly Detection Techniques for Large Acoustic Data in Machine Operation

“…Some studies [ 12 , 17 , 18 ] have classified abnormal states with the concept of reconstruction errors that were distinguished from normal states. The detection method using reconstruction error can be used for AD using accumulated test data or the real-time result, although it is weaker than the method using prediction error to predict the diagnosis result.…”

“…However, determining the inspection index in the process of automating the noise and vibration quality inspection is not easy and requires a lot of time and effort. For this reason, in recent years, deep learning models have been used to minimize the complexity of data preprocessing, feature extraction, and feature selection [ 11 , 12 ].…”

Soft Sensing Model Visualization: Fine-tuning Neural Network from What Model Learned

“…Deep learning approaches have developed rapidly and have become prevalent in big complex data processing [22], [23]. Besides, the emergence of deep learning techniques reduces the need of domain specialists for feature extraction [24]- [26]. Yuan et al [27] applied the deep belief network (DBN) to learn the dynamics and various local correlations of different variable combinations for quality prediction.…”