Search

An effective model to represent real-world wind power production scenarios is essential for an accurate assessment of the impact of wind power generation on power systems. Such a model should capture the spatial and temporal correlations between different wind turbines that are part of a wind farm. Accurate modeling of these correlations will ensure that the wind power uncertainty can be properly quantified. This is especially critical when analyzing systems with high penetration of wind power productions. This paper addresses this critical need by developing the Hourly Mixed Copula Model, a new probabilistic model, based on Gaussian and Archimedean copulas, to model the uncertainty and variability of wind power generation. This model enables the simulation of realistic scenarios of a wind farm power production. Metrics such as the Euclidean distance and Kullback-Leibler divergence are used to demonstrate the effectiveness of the developed model in incorporating the spatio-temporal dependence among the turbine power productions. The impact of accurate characterization of uncertainty in wind power modeling on power system steady-state operational planning is also demonstrated by case studies. These studies indicate that, in order to correctly evaluate the effect of wind power variation on steady-state operational planning, the dependence among wind power resources should not be neglected for power flow studies.

Deep Learning (DL) is vulnerable to out-of-distribution and adversarial examples resulting in incorrect outputs. To make DL more robust, several posthoc (or runtime) anomaly detection techniques to detect (and discard) these anomalous samples have been proposed in the recent past. This survey tries to provide a structured and comprehensive overview of the research on anomaly detection for DL based applications. We provide a taxonomy for existing techniques based on their underlying assumptions and adopted approaches. We discuss various techniques in each of the categories and provide the relative strengths and weaknesses of the approaches. Our goal in this survey is to provide an easier yet better understanding of the techniques belonging to different categories in which research has been done on this topic. Finally, we highlight the unsolved research challenges while applying anomaly detection techniques in DL systems and present some high-impact future research directions.

This paper presents novel sparse Bayesian learning (SBL)-based target imaging and parameter estimation techniques in monostatic multiple-input multiple-output (MIMO) radar systems for practical scenarios with insufficient observation samples and unknown target parameters. First, the SBL framework is developed for a single measurement vector setting with an underlying sparse target reflectivity parameter vector. This is subsequently extended to scenarios with multiple observation snapshots considering uncorrelated as well as correlated target reflectivity parameters. Variants are also proposed for challenging scenarios considering the presence of ground clutter. Cramér-Rao bounds are derived for the reflectivity, Doppler, and range estimates to comprehensively characterize the performance of the proposed estimation schemes. A joint parameter estimation and imaging scheme is developed based on a Taylor series expansion of the MIMO radar dictionary matrix. Simulation results demonstrate enhanced imaging and estimation accuracy of the proposed SBL schemes in comparison with the existing techniques for MIMO radar systems.