Understanding spatiotemporal correlations in power systems is crucial for maintaining grid stability, reliability, and efficiency. By discerning connections between spatial and temporal dimensions, operators can anticipate and address issues such as congestion, voltage instability, and equipment failures. Recent advancements in power system analysis have leveraged spatiotemporal correlations through sophisticated data-driven algorithms. In this survey paper, we conduct a comprehensive examination of deep learning frameworks tailored to tackle the complexities inherent in spatiotemporal data analysis within power systems. We categorize machine learning methodologies into discriminative, generative, and reinforcement learning, providing a structured overview of their mathematical foundations, advantages, and limitations in processing dynamic power system measurements. Through empirical evaluations, we assess the performance of these methodologies across various spatiotemporal applications, including cyber attack detection, fault identification, demand response, and renewable energy forecasting, offering insights into their efficacy and applicability. Additionally, we identify emerging topics within the machine learning domain that hold promise for future endeavors in power systems analysis.

Nowadays, rooftop photovoltaic (PV) panels are getting enormous attention as clean and sustainable sources of energy due to the increasing energy demand, depreciating physical assets, and global environmental challenges. In residential areas, the large-scale integration of these generation resources influences the customer load profile and introduces uncertainty to the distribution system’s net load. Since such resources are typically located behind the meter (BtM), an accurate estimation of BtM load and PV power will be crucial for distribution network operation. This article proposes the spatiotemporal graph sparse coding (SC) capsule network that incorporates SC into deep generative graph modeling and capsule networks for accurate BtM load and PV generation estimation. A set of neighboring residential units are modeled as a dynamic graph in which the edges represent the correlation among their net demands. A generative encoder–decoder model, i.e., spectral graph convolution (SGC) attention peephole long short-term memory (PLSTM), is devised to extract the highly nonlinear spatiotemporal patterns from the formed dynamic graph. Later, to enrich the latent space sparsity, a dictionary is learned in the hidden layer of the proposed encoder–decoder, and the corresponding sparse codes are procured. Such sparse representation is used by a capsule network to estimate the BtM PV generation and the load of the entire residential units. Experimental results on two real-world energy disaggregation (ED) datasets, Pecan Street and Ausgrid, demonstrate more than $9.8\%$ and $6.3\%$ root mean square error (RMSE) improvements in BtM PV and load estimation over the state-of-the-art, respectively.

Due to the increasing demand for electricity and the inherent uncertainty in power generation, finding efficient solutions to the stochastic alternating current optimal power flow (AC-OPF) problem has become crucial. However, the nonlinear and non-convex nature of AC-OPF, coupled with the growing stochasticity resulting from the integration of renewable energy sources, presents significant challenges in achieving fast and reliable solutions. To address these challenges, this study proposes a novel graph-based generative methodology that effectively captures the uncertainties in power system measurements, enabling the learning of probability distribution functions for generation dispatch and voltage setpoints. Our approach involves modeling the power system as a weighted graph and utilizing a deep spectral graph convolution network to extract powerful spatial patterns from the input graph measurements. A unique variational approach is introduced to identify the most relevant latent features that accurately describe the setpoints of the AC-OPF problem. Additionally, a capsule network with a new greedy dynamic routing algorithm is proposed to precisely decode the latent features and estimate the probabilistic AC-OPF problem. Further, a set of carefully designed physics-informed loss functions is incorporated in the training procedure of the model to ensure adherence to the fundamental physics rules governing power systems. Notably, the proposed physics-informed loss functions not only enhance the accuracy of AC-OPF estimation by effectively regularizing the deep learning model but also significantly reduce the time complexity. Extensive experimental evaluations conducted on various benchmarks demonstrate our proposed model's superiority over both probabilistic and deterministic approaches in terms of relevant criteria.

In recent years, the accurate recognition of traffic scenes has played a key role in autonomous vehicle operations. However, most works in this area do not address the domain shift issue where the classification performance is degraded when the distribution of the source and target images are different due to weather changes. Also, lack of sparsity of current studies results in sample complexity and overfitting issues. To mitigate these challenges, this paper proposes a novel Sparse Adversarial Domain Adaptation (SADA) model for traffic scene classification. Our objective is to learn a scene classifier on a source domain with sunny weather and transfer its knowledge to a target domain with different weather (i.e., cloudy, rainy, and snowy) to enhance the classification performance on the target. First, a sparse representation is learned from the source traffic scenes via nonlinear dictionary learning in the latent space of a deep classifier. Then, a conditional generative adversarial network is devised to capture the distribution of the source sparse codes. Finally, a domain invariant sparse feature extractor is developed via a minimax game to align the sparse codes of the target domain images with the source; hence, providing domain adaptation for traffic scene classification using a deep neural network. Experimental results on a real-world dataset collected by the Honda Research Institute (HRI) indicate the superior performance of the proposed SADA compared to current traffic scene classification methods and state-of-the-art domain adaptation frameworks.

Semantic segmentation in diverse real-world traffic scenes is a challenging task for autonomous vehicles to have a reliable understanding of the outside environment. Although deep visual models show a remarkable performance on traffic scene segmentation, they need significant improvements when the source and target scene distributions are different. The weather and illumination differences between source and target datasets result in dramatic distribution shifts that degrade the target segmentation performance. This article develops an unsupervised domain adaptation (UDA) model to address complex domain shift problems where the dataset, weather, and illumination of source and target traffic scenes are different. We devise a sparse adversarial multitarget UDA approach to capture powerful domain-invariant features to segment traffic scenes in different conditions. First, a sparse representation of the source traffic scenes is captured via a new spectral low-rank dictionary learning technique in the latent space of a deep encoder–decoder segmentation architecture. The feature sparsity provides high generalization capacity for our deep feature extractor that enables us to compute complex visual patterns of source images. Then, the distribution of source sparse features is learned using a generative adversarial framework. Finally, the sparse representations of source and target scenes are aligned via a sparse domain-invariant feature extractor trained by min–max optimization. The aligned features serve as domain-invariant scene representations that can best describe both source and target scenes; hence providing deep domain adaptation for traffic scene semantic segmentation. Experiments on a real-world dataset indicate the superiority of the proposed model compared to the state-of-the-art methodologies.

In recent years, deep neural networks have shown significant progress in computer vision due to their large generalization capacity; however, the overfitting problem ubiquitously threatens the learning process of these highly nonlinear architectures. Dropout is a recent solution to mitigate overfitting that has witnessed significant success in various classification applications. Recently, many efforts have been made to improve the Standard dropout using an unsupervised merit-based semantic selection of neurons in the latent space. However, these studies do not consider the task-relevant information quality and quantity and the diversity of the latent kernels. To solve the challenge of dropping less informative neurons in deep learning, we propose an efficient end-to-end dropout algorithm that selects the most informative neurons with the highest correlation with the target output considering the sparsity in its selection procedure. First, to promote activation diversity, we devise an approach to select the most diverse set of neurons by making use of determinantal point process (DPP) sampling. Furthermore, to incorporate task specificity into deep latent features, a mutual information (MI)-based merit function is developed. Leveraging the proposed MI with DPP sampling, we introduce the novel DPPMI dropout that adaptively adjusts the retention rate of neurons based on their contribution to the neural network task. Empirical studies on real-world classification benchmarks including, MNIST, SVHN, CIFAR10, CIFAR100, demonstrate the superiority of our proposed method over recent state-of-the-art dropout algorithms in the literature.