Generative models have demonstrated significant success in anomaly detection and segmentation over the past decade. Recently, diffusion models have emerged as a strong alternative, outperforming previous approaches such as GANs and VAEs. In typical diffusion-based anomaly detection, a model is trained on normal data, and during inference, anomalous images are perturbed to a predefined intermediate step in the forward diffusion process. The corresponding normal image is then reconstructed through iterative reverse sampling.However, reconstruction-based approaches present three major challenges: (1) the reconstruction process is computationally expensive due to multiple sampling steps, making real-time applications impractical; (2) for complex or subtle patterns, the reconstructed image may correspond to a different normal pattern rather than the original input; and (3) choosing an appropriate intermediate noise level is challenging because it is application-dependent and often assumes prior knowledge of anomalies, an assumption that does not hold in unsupervised settings.We introduce Reconstruction-free Anomaly Detection with Attention-based diffusion models in Real-time (RADAR), which overcomes the limitations of reconstruction-based anomaly detection. Unlike current state-of-the-art (SOTA) methods that reconstruct the input image, RADAR directly produces anomaly maps from the diffusion model, improving both detection accuracy and computational efficiency. We evaluate RADAR on real-world 3D-printed material and the MVTec-AD dataset. Our approach surpasses state-of-the-art diffusion-based and statistical machine learning models across all key metrics, including accuracy, precision, recall, and F1 score. Specifically, RADAR improves F1 score by 7% on MVTec-AD and 13% on the 3D-printed material dataset compared to the next best model.Code: https://github.com/mehrdadmoradi124/RADAR

Supervisory control and data acquisition (SCADA) systems are widely used for industrial control of critical infrastructures, such as power plants and manufacturing systems. There is abundant evidence of SCADA systems being subject to cyberattacks. With increasing interest in industrial digitization, the cybersecurity of SCADA systems is poised to be even more important. Equipment faults and cyberattacks can manifest themselves in a similar fashion, i.e., they can exhibit similar signatures. This article focuses on methods that are capable of distinguishing equipment faults from bona fide cyberattacks. Especially, we consider a relatively sophisticated form of cyberattack known as the “replay attack” (RA). We derive mathematical formalisms that distinguish the RA from several classes of equipment faults and verify our methodology through an extensive numerical study.Note to Practitioners—This article is motivated by the problem of detecting replay cyberattacks in industrial control systems and differentiating it from equipment faults. Existing approaches mainly focus on the detection aspect but usually ignore the importance of differentiation. We an ensembled statistical process monitoring approach based on five statistical metrics. The statistical metrics are derived based on a theoretical analysis that shows the data characteristics under each system anomaly, including replay attack (RA), controller fault, and plant fault. We mathematically prove that the signatures generated by the derived metrics can be used to differentiate an RA from the equipment faults. We conduct a sensitivity analysis of the detection delay of our method regarding the magnitude of the cyberattack. Physical experiments on a rotating machinery setup show that the proposed approach applies to some simple real-world settings. In future research, we will address the scalability issue of our method as well as more generalized nonlinear system settings.

Cybersecurity of Industrial Control Systems (ICS) is drawing significant concerns as data communication increasingly leverages wireless networks. A lot of data-driven methods were developed for detecting cyberattacks, but few are focused on distinguishing them from equipment faults. In this paper, we develop a data-driven framework that can be used to detect, diagnose, and localize a type of cyberattack called covert attacks on smart grids. The framework has a hybrid design that combines an autoencoder, a recurrent neural network (RNN) with a Long-Short-Term-Memory (LSTM) layer, and a Deep Neural Network (DNN). This data-driven framework considers the temporal behavior of a generic physical system that extracts features from the time series of the sensor measurements that can be used for detecting covert attacks, distinguishing them from equipment faults, as well as localize the attack/fault. We evaluate the performance of the proposed method through a realistic simulation study on the IEEE 14-bus model as a typical example of ICS. We compare the performance of the proposed method with the traditional model-based method to show its applicability and efficacy.

We present an algorithm for multi-robot coverage of an initially unknown spatial scalar field characterized by a density function, whereby a team of robots simultaneously estimates and optimizes its coverage of the density function over the domain. The proposed algorithm borrows powerful concepts from Bayesian Optimization with Gaussian Processes that, when combined with control laws to achieve centroidal Voronoi tessellation, give rise to an adaptive sequential sampling method to explore and cover the domain. The crux of the approach is to apply a control law using a surrogate function of the true density function, which is then successively refined as robots gather more samples for estimation. The performance of the algorithm is justified theoretically under slightly idealized assumptions, by demonstrating asymptotic no-regret with respect to the coverage obtained with a known density function. The performance is also evaluated in simulation and on the Robotarium with small teams of robots, confirming the good performance suggested by the theoretical analysis.

Learning the relationship between a response variable (e.g., a quality characteristic) and a set of predictors (e.g., process variables) is of special importance in process modeling, prediction, and optimization. In many applications, not only is the number of these variables large but these variables are also high-dimensional (HD) (e.g., they are represented by waveform signals). This high dimensionality requires a systematic approach to both modeling the relationship between the variables and removing the noninformative input variables. This article proposes a functional regression method in which an HD response is estimated and predicted through a set of informative and noninformative HD covariates. For this purpose, the functional regression coefficients are expanded through a set of low-dimensional smooth basis functions. In order to estimate the low-dimensional set of parameters, a penalized loss function with both smoothing and group lasso penalties is defined. The block coordinate decent (BCD) method is employed to develop a computationally tractable algorithm for minimizing the loss function. Through simulations and case studies, the performance of the proposed method is evaluated and compared with benchmarks. The results illustrate the advantage of the proposed method over the benchmarks.

This article proposes two methods based on deep learning for estimating time-to-failure (TTF) of an industrial system using its degradation image. This provides an effective tool for predictive maintenance practitioners toward digitization of maintenance processes in Industry 4.0 transformation. Both methods utilize the long short-term memory (LSTM) networks for capturing temporal information. First methodology consists of two convolutional layers preceding a single LSTM layer to extract compact information from the individual images and rescue LSTM network from curse of dimensionality. Then, an LSTM layer estimates the TTF value from the extracted features. In the second approach, the dimension of the individual images are decreased by a fully connected neural network, which is trained as an autoencoder. A separate LSTM network is trained and run over this lower dimensional space. The strength of suggested architectures is shown using simulation data and a dataset of infrared image streams collected from rotating machinery. The performance comparison of proposed methods and other methods is also provided.

Image and video sensors are increasingly being deployed in complex systems due to the rich process information that these sensors can capture. As a result, image data play an important role in process monitoring and control in different application domains such as manufacturing processes, food industries, medical decision-making, and structural health monitoring. Existing process monitoring techniques fail to fully utilize the information of color images due to their complex data characteristics including the high-dimensionality and correlation structure (i.e., temporal, spatial and spectral correlation). This paper proposes a new image-based process monitoring approach that is capable of handling both grayscale and color images. The proposed approach models the high-dimensional structure of the image data with tensors and employs low-rank tensor decomposition techniques to extract important monitoring features monitored using multivariate control charts. In addition, this paper shows the analytical relationships between different low-rank tensor decomposition methods. The performance of the proposed method in quick detection of process changes is evaluated and compared with existing methods through extensive simulations and a case study in a steel tube manufacturing process.