Abstract
The uncertainty associated with the increasingly wind power penetration in power systems must be considered when performing the traditional day-ahead scheduling of conventional thermal units. This uncertainty can be represented through a set of representative wind power scenarios that take into account the time-dependency between forecasting errors. To create robust Unit Commitment ( UC) schedules, it is widely seen that all possible wind power scenarios must be used. However, using all realizations of wind power might be a poor approach and important savings in computational effort can be achieved if only the most representative subset is used. In this paper, the new hybrid metaheuristic DEEPSO and clustering techniques are used in the traditional stochastic formulation of the UC problem to investigate the robustness of the UC schedules with increasing number of wind power scenarios. For this purpose, expected values for operational costs, wind spill, and load curtailment for the UC solutions are compared for a didactic 10 generator test system. The obtained results shown that it is possible to reduce the computation burden of the stochastic UC by using a small set of representative wind power scenarios previously selected from a high number of scenarios covering the entire probability distribution function of the forecasting uncertainty.

Abstract
The optimal solution provided by metaheuristics can be viewed as a random variable, whose behavior depends on the value of the algorithm's strategic parameters and on the type of penalty function used to enforce the problem's soft constraints. This paper reports the use of parametric and non-parametric statistics to compare three different penalty functions implemented to solve the Security Constrained Optimal Power Flow (SCOPF) problem using the new enhanced metaheuristic Differential Evolutionary Particle Swarm Optimization (DEEPSO). To obtain the best performance for the three types of penalty functions, the strategic parameters of DEEPSO are optimized by using an iterative algorithm based on the two-way analysis of variance (ANOVA). The results show that the modeling of soft constraints significantly influences the best achievable performance of the optimization algorithm.

Abstract
The deployment of Smart Grid technologies opens new opportunities to develop new forecasting and optimization techniques. The growth of solar power penetration in distribution grids imposes the use of solar power forecasts as inputs in advanced grid management functions. This paper proposes a new forecasting algorithm for 6 h ahead based on the vector autoregression framework, which combines distributed time series information collected by the Smart Grid infrastructure. Probabilistic forecasts are generated for the residential solar photovoltaic (PV) and secondary substation levels. The test case consists of 44 micro-generation units and 10 secondary substations from the Smart Grid pilot in Evora, Portugal. The benchmark model is the well-known autoregressive forecasting method (univariate approach). The average improvement in terms of root mean square error (point forecast evaluation) and continuous ranking probability score (probabilistic forecast evaluation) for the first 3 lead-times was between 8% and 12%, and between 1.4% and 5.9%, respectively. (C) 2015 Published by Elsevier Ltd.

Abstract
The solar power penetration in distribution grids is growing fast during the last years, particularly at the low-voltage (LV) level, which introduces new challenges when operating distribution grids. Across the world, distribution system operators (DSO) are developing the smart grid concept, and one key tool for this new paradigm is solar power forecasting. This paper presents a new spatial-temporal forecasting method based on the vector autoregression framework, which combines observations of solar generation collected by smart meters and distribution transformer controllers. The scope is 6-h-ahead forecasts at the residential solar photovoltaic and medium-voltage (MV)/LV substation levels. This framework has been tested in the smart grid pilot of vora, Portugal, and using data from 44 microgeneration units and 10 MV/LV substations. A benchmark comparison was made with the autoregressive forecasting model (AR-univariate model) leading to an improvement on average between 8% and 10%.

Abstract
This paper offers an efficient and robust concept for a decentralized bad data processing, able to supply in real-time a power system state estimator with a repaired measurement set. Corrupted measurement vectors are funneled through a denoising auto-associative neural network in order to project the biased vector back to the data manifold learned during an offline training process. In order to improve accuracy, a maximum similarity with the solution manifold, measured with Correntropy, is searched for by a meta-heuristic. The extreme robustness and scalability of the process is demonstrated in multiple characteristic case studies.

Abstract
Optimization problems in electric power systems under high levels of uncertainty have been solved using stochastic programming methods for years. This is especially the case for medium-term problems and systems with a large share of hydro storages. The increased uncertainty in power system operation coming from volatile renewables has made the stochastic techniques interesting in shorter time frames as well. In the stochastic models the uncertainty is typically included by a discretized set of scenarios. This increases the computational burden significantly so a common approach is to preprocess and reduce the number of scenarios. Scenario reduction methods have been shown to function relatively well in expected value stochastic optimization. However, such setting of stochastic optimization is often criticized as being risk-prone so other risk-averse models exist. The evolutionary computation algorithms' flexibility permits the implementation of these risk models with relative simplicity. In this paper, a population-based evolutionary computation algorithm is applied to unit commitment problem under uncertainty and the paper illustrates several approaches to including risk policies in an evolutionary algorithm fitness function and illustrates its flexibility along with the link between scenario reduction similarity metric and risk policy.