Accurate prediction of wind power generation is of great significance for the efficient operation of wind farms. However, traditional deep learning-based methods predict the wind power without simultaneously considering the temporal features of wind power and spatial features between variables, which leads to low prediction accuracy. This article proposes a novel wind power forecasting approach based on a graph convolution network (GCN) and a multiresolution convolution neural network (CNN), combining spatial features and temporal features. In this approach, GCN merged with maximum information coefficient (MIC) is proposed to extract the spatial correlation features between input variables, which considers the effects of multiple variables on wind power and provides interpretability for deep learning-based forecasting. On the other hand, multiresolution CNN combines multiscale convolution kernels with a new self-attention mechanism to understand local and long-term temporal features, which enables simultaneous prediction of wind power and other variables. Experiments on a real dataset prove that the proposed method is effective and accurate in short-term wind power forecasting. Comparisons with the other three state-of-the-art methods and ablation experiments also reveal the advantages of the proposed approach.

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.

Voltage fault diagnosis is critical for detecting and identifying the lithium (Li)-ion battery failure. This article proposes a voltage fault diagnosis algorithm based on an equivalent circuit model-informed neural network (ECMINN) method for Li-ion batteries, which aims to learn the voltage fault observer by embedding the equivalent circuit model (ECM) into neural network structures. This method directly embeds the deterministic mechanism part in the ECM and designs the uncertain part into neural networks, which takes advantage of the high precision of the physical model and the strong nonlinear processing ability of the neural network to improve the fault diagnosis effect. In this method, the voltage prediction module and residual-based parameter calibration module are developed to describe state space equations and automatically learn the state parameters and model parameters, respectively. Moreover, a fault score specifically designed by combining the identified model parameters is presented, which is proved to be effectively in detecting different fault types. Finally, experiments on two public datasets are conducted, by which the proposed ECMINN method is demonstrated to be effective and accurate in predicting voltage data and detecting faults. Comparison with three other state-of-the-art methods also reveals the advantages of the proposed ECMINN method.

A novel RUL prediction approach for lithium-ion batteries using quantum particle swarm optimization (QPSO)-based particle filter (PF) is proposed. Compared to particle swarm optimization (PSO)-based PF, QPSO-based PF is proved to have a better performance in global searching and has fewer parameters to control, which makes QPSO-PF easier for applications. Moreover, fewer particles are required by QPSO-PF to accurately track the battery's health status, leading to a reduction of computation complexity. RUL prediction results using real data provided by NASA and compared with benchmark approaches demonstrates the superiority of the proposed approach.

Detecting anomalies for multivariate time series is of great importance in modern industrial applications. However, due to the complex temporal dynamics in modern systems, finding a distinguishable judge criterion is hard, which makes accurate anomaly detection still a challenging task. In order to better capture the anomalous features and design a more informative judge criterion, this paper presents an unsupervised generative adversarial network (GAN) for multivariate time series anomaly detection, which highlights a novel Active Distortion Transformer (ADT) block. Different from the vanilla Transformer, the ADT block can make good use of the prior knowledge of time sequences’ overall associations by actively conducting distortion during the reconstruction of input sequences. Benefiting from the ADT block, the network simultaneously utilizes the sequence associations and reconstruction error to recognize anomalies. In the online detection phase, anomalous data points tend to be less correlated with the overall sequence and have greater reconstruction errors than normal ones, so that an irrelevance score and a reconstruction error score can be obtained. We combine the two scores to generate a more powerful anomaly score as the judge criterion. Extensive experiments are conducted on four publicly available sensor datasets, and we also make comparisons with the recent baselines. Results show that our model outperforms the recent state-of-the-art methods, demonstrating the effectiveness of our method.