The telemetry data obtained from an on-orbit spacecraft contain important information to indicate anomaly of the spacecraft. However, the large number of monitoring variables and the large amount of data points, as well as the lack of prior knowledge about anomaly due to complicated structure of spacecraft and its working conditions, pose great challenge to the anomaly detection. This article proposes an anomaly detection algorithm based on a spatial-temporal generative adversarial network (GAN) for the anomaly detection in telemetry data. This algorithm establishes a GAN-based model combining convolutional neural network (CNN) and long short-term memory (LSTM) to extract spatial and temporal features of the telemetry data, which facilitates the automatic and simultaneous representation of nonnegligible time-related characteristics of a monitoring variable and complex correlation between variables. Using these features, many kinds of anomalies including multivariate anomalies and contextual anomalies can be detected. Moreover, an anomaly score specifically designed to fit the GAN-based algorithm is also proposed to evaluate the possibility of anomaly by weighted fusion of the generator metric and the discriminator metric, which is proved to be significantly helpful to the accuracy of anomaly detection. Finally, experiments on one real telemetry data set and two public telemetry data sets are conducted, by which the proposed anomaly algorithm is demonstrated to be effective and accurate in detecting outliers in telemetry data. Comparison with three other state-of-the-art methods also reveals the advantages of our proposed algorithm.

Noncontact heart rate (HR) measurement is a very important trend in clinical medicine. Recently, a variety of deep networks have been applied to estimate HRs from facial videos. However, due to limited data resources and poor parameter optimization, few existing models have achieved incredible performance in complicated scenarios, such as those with illumination changes, different skin tones, and facial motion. To address these challenges, this paper proposes a novel multi-stage deep network (MSDN) that can decentralize the learnable parameters into different stages to reduce the difficulty of learning through multiple training steps. Specifically, the proposed network consists of three stages in an end-to-end way. In the first stage, an HR-aware feature extractor utilizes the next convolutional neural network (ConvNeXt) embedded with a newly designed bandpass filter as its backbone to extract spatial-temporal features for determining heart rate changes. Moreover, pseudolabels are generated to make the features compatible with illumination, motion, and color variance. In the second stage, various modules, including singular value decomposition (SVD) pooling and enhanced difference convolution modules, are then designed and combined with a Transformer encoder to construct a feature-compressed remote photoplethysmography (rPPG) generator. In the third stage, an HR estimator with an interbeat interval analyzer and a 1D filter is newly designed for HR estimation. Extensive experiments are performed on three publicly available databases (i.e., VIPL-HR, COHFACE, and PURE), and the results demonstrate the effectiveness of the proposed method through ablation studies and comparison experiments with state-of-the-art (SOTA) methods.