Abstract
This letter proposes a new concept applied to state estimation based on replacing traditional regression models by a criterion of maximizing error correntropy introducing a novel way to identify and correct large errors.

Abstract
This paper presents the proof of concept for a new solution to the problem of recomposing missing information at the SCADA of energy/distribution management systems (EMS/DMS), through the use of offline trained autoencoders. These are neural networks with a special architecture, which allows them to store knowledge about a system in a nonlinear manifold characterized by their weights. Suitable algorithms may then recompose missing inputs (measurements). The paper shows that, trained with adequate information, autoencoders perform well in recomposing missing voltage and power values, and focuses on the particularly important application of inferring the topology of the network when information about switch status is absent. Examples with the IEEE RTS 24-bus network are presented to illustrate the concept and technique.

Abstract
The main purpose of this work is the development of an optimization model for the design of the grounding grid in electrical substations. The problem is formulated as a mixedinteger linear programming problem, in terms of the constructive characteristics and the peculiar requirements to construct and to install the grounding grid. The model incorporates the variables that define the grid characteristics of all possible configurations including the grid geometry and the depth and conductor size. The optimization problem is subject to safety constraints related with the maximum allowed touching and step voltages. It also includes the equivalent impedance of the transmission line connected to the substation where it will be located the grounding grid to be designed. The methodology allows selecting the optimum grid of the possible configurations, so that is a very useful tool for the engineering design. The formulation and specifications used is based in IEEE Std. 80-2000.

Abstract
A fuzzy model for power system operation is presented. Uncertainties in loads and generations are modeled as fuzzy numbers. System behavior under known (while uncertain) injections is dealt with by a DC fuzzy power flow model. System optimal (while uncertain) operation is calculated with linear programming procedures in which the problem nature and structure allow some efficient techniques such as Dantzig-Wolfe decomposition and dual simplex to be used. Among the results, one obtains a fuzzy cost value for system operation and possibility distributions for branch power flows and power generations. Some risk analysis is possible, as system robustness and exposure indices can be derived and hedging policies can be investigated.

Abstract
Neural networks have broad applicability to real power system problems. One of the areas in power system with huge interest in appliance of neural networks is load forecasting. In this paper the neural networks were trained and tested using 15-minute load data collected in Portugal by the electric power company EDP during a 44 day period. The artificial neural networks showed as a good nonlinear approximator, giving promising results. The main objective of the presented work is to interest power companies in the Region for possible practical implementations.

Abstract
This paper is written with the didactic purpose of exploring and indicating possibilities to power companies in the Balkan region for the application of adaptive neuro-fuzzy inference system (ANFIS) models in load prediction with real load data set. ANFIS models were trained and tested using 15-minute load data collected in Portugal by the electric power company EDP during a 42 day period. Simulation results gave promising results especially considering small size of used data set. Although the objective of the paper is to demonstrate possibilities for practical implementation, further research and improvement including the contributions of similar approaches in the world must he done.