Abstract
This paper presents a new approach to incipient fault diagnosis in power transformers, based on the results of dissolved gas analysis. A set of autoassociative neural networks or autoencoders is trained, so that each becomes tuned with a particular fault mode or no fault condition. The scarce data available forms clusters that are densified using an Information Theoretic Mean Shift algorithm, allowing all real data to be used in the validation process. Then, a parallel model is built where the autoencoders compete with one another when a new input vector is entered and the closest recognition is taken as the diagnosis sought. A remarkable accuracy of 100% is achieved with this architecture, in a validation data set using all real information available.

Abstract
Acoustic emission monitoring is often used in the diagnosis of electrical and mechanical incipient faults in high-voltage apparatus. Partial discharges are a source of failure in power transformers, and the differentiation from other sources of acoustic emissions is of the utmost importance. This paper reports the development of a new sensor concept-mandrel-based fiber-optic sensor-for the detection of incipient faults in oil-filled power transformers, taking direct measurements inside a transformer. These sensors can be placed in the inner surface of the transformer tank wall, not affecting the insulation integrity of the structure, and improving fault detection and location. The applicability of these acoustic sensors in air, water, and oil is investigated and the paper presents the promising results obtained, which will enable the industrial development of practical solutions.

Abstract
The paper describes how mapping a neural network into a rule-based fuzzy inference system leads to knowledge extraction. This mapping makes explicit the knowledge implicitly captured by the neural network during the learning stage, by transforming it into a set of rules. By applying the method to transformer fault diagnosis using dissolved gas-in-oil analysis, one could not only develop intelligent diagnosis systems, providing better results than the application of the IIEC 60599 Table, but also generate a new rule table whose application also leads to better diagnosis results.

Abstract
This paper presents a new methodology for auto- mated route selection for the construction of new power lines, based on geographic information systems (GIS). It uses a dynamic programming model for route optimization. Environmental restrictions are taken into account together with all of the operating, maintenance, and equipment installation costs, including a new approach to the costs associated with the slope of the terrain crossed by the power lines. The computing and visual representation capacities of GIS are exploited for the selection of economic corridors, keeping the total costs under a threshold imposed by the user. Intensive simulation examples illustrate the power and flexibility of the proposed methodology.

Abstract
In this paper, Artificial Neural Networks (ANN) are used to model an extraction process that uses a supercritical fluid as solvent which its pilot installation is located at the Institute of Experimental and Technological Biology - IBET in Oeiras - Lisbon - Portugal. A strategy is used to complement the experimental data collected in laboratory during extraction procedures of useful compositions for the pharmaceutical industry using Black Agglomerate Residues (BAR) originating from of the cork production as raw material. The strategy involves fitting of data obtained during an operation of extraction. Two neural models are presented: the neural model trained using a Mean Square Error (MSE) minimization algorithm and the neural model which the learning was based on the error entropy minimization. A comparison of the performance of the two models is presented. © 2008 IEEE.

Abstract
This paper analyzes the application of the Information Theoretic (IT) Mean Shift algorithm for modes finding in order to provide the classification of Electricity Customer Load Patterns. The impact of the algorithm parameters is discussed and then clustering indices are used in order to make a comparison with the classical methods available. Results show a good capability of the modes found in capturing the data structure, aggregating similar load patterns and identifying the uncommon patterns (outliers). © 2011 IEEE.