Abstract
This paper presents a new method to identify classes of events, by processing phasor measurement units (PMU) frequency data through deep neural networks. Deep tapered Multi-layer Perceptrons of the half-autoencoder type, Deep Belief Networks and Convolutional Neural Networks (CNN) are compared, using real data from Brazil. A sound success is obtained by a transformation of time-domain signals, from dynamic events recorded, into 2D images; these images wee processed with a CNN, taking advantage of the strong dependency existing among neighboring pixels in images. The training, computing and processing was achieved with a GPU (Graphics Processing Unit), allowing speeding-up of the process up to 30 times and rendering the process suitable to increase the online situational awareness of network operators.

Abstract
This paper presents a data-driven spatial net-load forecasting model that is applied to the distribution network expansion problem. The model uses population census data with Information Theory-based Feature Selection to predict spatial adoption patterns of residential electric vehicle chargers and photovoltaic modules. Results are high-resolution maps (0.02 km(2)) that allow distribution network planners to forecast asymmetric changes in load patterns and assess resulting impacts on installed HV/MV substation transformers in distribution systems. A risk analysis routine identifies the investment that minimizes the maximum regret function for a 15-year planning horizon. One of the outcomes from this study shows that traditional approaches to allocate distributed energy resources in distribution networks underestimate the impact of adopting EV and PV on the grid. The comparison of different allocation methods with the presented diffusion model suggests that using conventional approaches might result in strong underinvestment in capacity expansion during early uptake and overinvestment in later diffusion stages.

Abstract
This paper proposes a new probabilistic model for active low-voltage prosumers suitable for distribution expansion planning studies. The load uncertainty of these consumers is considered through a range of load profiles by segmenting the energy consumption according to the different energy uses. Then, consumption adjustments are simulated using a nonhomogenous Poisson process based on the energy usage preferences and the financial gains according to the tariff scheme. A case study based on the modified IEEE 33-Bus test system with real data collected from a Brazilian distribution company is performed in order to analyze the impact of the load profiles in scenarios with high penetration of renewable distributed generation (DG). The experiments carried out reveal that considerable monetary savings in the expansion of the distribution grid can be achieved for this case study (up to 37%) as compared with the alternative with no active demand (AD) by exploiting the flexibility associated with the active behavior of prosumers as a response to price signals and/or by permitting adequate levels for the integration of DG into the distribution grid.

Abstract
This paper provides an answer to the problem of State Estimation (SE) with multiple simultaneous gross errors, based on Generalized Error Correntropy instead of Least Squares and on an interior point method algorithm instead of the conventional Gauss-Newton algorithm. The paper describes the mathematical model behind the new SE cost function and the construction of a suitable solver and presents illustrative numerical cases. The performance of SE with the data set contaminated with up to five simultaneous gross errors is assessed with confusion matrices, identifying false and missed detections. The superiority of the new method over the classical Largest Normalized Residual Test is confirmed at a 99% confidence level in a battery of tests. Its ability to address cases where gross errors fall on critical measurements, critical sets or leverage points is also confirmed at the same level of confidence.

Abstract
Classical Weighted Least Squares (WLS) State estimation (SE) in power systems is known for not performing well in the presence of Gross Errors (GE). The alternative using Correntropy proved to be appealing in dealing with outliers. Now, a novel SE method, generalized correntropy interior point method (GCIP) is being proposed, taking advantage of the properties of the Generalized Correntropy and of the Interior Point Method (IPM) as solver. This paper discusses how the choice of different central path neighborhoods, an essential concept in IPM, is critical in the identification of gross errors. The simulation results indicate that a one-sided infinity norm neighborhood successfully identifies outliers in the SE problem, making GCIP a competitive method. © 2019 IEEE.

Abstract
The adoption of energy transition technologies for residential use is accelerated through incentive designs. The structure of such incentives affects technology adoption patterns, that is, the locations where new technologies are installed and used. These spatial adoption patterns influence network expansion costs and provide indication on potential cross-subsidization between population groups. While until today, most programs have been involuntarily favoring households with high-income and above-average educated population groups, incentive designs are currently under review. This paper presents a spatiotemporal technology adoption model that can predict adoption behavior of residential electric vehicle (EV) chargers and photovoltaic (PV) modules up to a predefined time horizon. A set of EV and PV adoption patterns for nine incentive design combinations are compared in order to assess potential synergies that may arise under orchestrated EV and PV adoption. Effects on adoption asymmetries are evaluated using an Information-Theoretic inequality metric. Results for Continental Portugal show that global network expansion costs can be reduced while minimizing technology adoption asymmetries, if specific incentive designs are combined.