Lorant Kovacs

Power system reliability assessment based on Large Deviation Theory bounds

The ability of the power system to meet the expected demand is one of the main issues of electricity system reliability assessment. In the case of the load is greater than the generation capacity the system is at risk, not to be able to serve the demand. The most commonly used reliability measure in electricity power systems is the Loss of Load Probability (LOLP). Our paper introduces a new approach to the assessment of reliability of a power system, which is viewed as its ability to meet the capacity limit. More precisely, we give an upper limit on the probability of overconsumption. Our method needs the bottom-up modelling of the aggregate load of a consumption area and two states Markovian Models are shown to be an adequate tool for this purpose.

Automatic segmentation of electricity consumption data series with Jensen-Shannon divergence

In Smart Grids the Information and Communication Technologies (ICT) could be used to better manage both consumption and production of electricity. The increasing presence of renewable energy sources in production and the permeation of novel consumption types (e.g. Plug-in Hybrid Electric Vehicles (PHEV)) will obviously cause the increase the fluctuation of electrical energy. One possible solution to these problems is development of novel methods for investigating electrical power consumption data series. As the existing learning algorithms of pattern classification are suitable for discovering internal structures of large datasets, it is important to generate a training/testing/validation learning database from existing measurements (e.g. from smart meters), actually via segmentation and labeling by hand. In this paper we propose a novel method for the automatic segmentation with a predefined confidence level. The algorithm is based on the generalized Jensen-Shannon divergence (JSD), and it estimates the change-points (CPTs) in electrical power consumption data. Both the method and some recent results in segmenting one households power consumption data are presented in this paper.

Approximate minimum bit error rate equalization for fading channels

A novel channel equalizer algorithm is introduced for wireless communication systems to combat channel distortions resulting from multipath propagation. The novel algorithm is based on minimizing the bit error rate (BER) using a fast approximation of its gradient with respect to the equalizer coefficients. This approximation is obtained by estimating the exponential summation in the gradient with only some carefully chosen dominant terms. The paper derives an algorithm to calculate these dominant terms in real-time. Summing only these dominant terms provides a highly accurate approximation of the true gradient. Combined with a fast adaptive channel state estimator, the new equalization algorithm yields better performance than the traditional zero forcing (ZF) or minimum mean square error (MMSE) equalizers. The performance of the new method is tested by simulations performed on standard wireless channels. From the performance analysis one can infer that the new equalizer is capable of efficient channel equalization and maintaining a relatively low bit error probability in the case of channels corrupted by frequency selectivity. Hence, the new algorithm can contribute to ensuring QoS communication over highly distorted channels.

Recurrent neural network based user classification for smart grids

Power consuming users and buildings with different power consumption patterns may be treated with different conditions and can be taken into consideration with different parameters during capacity planning and distribution. Thus the automated, unsupervised categorization of power consumers is a very important task of smart power transmission systems. Knowing the behavioral categories of power consumers better models can be created which can be used for better behavior forecast which is an important task for load balancing. One of the existing best solutions for consumer classification is the consumption forecast based scheme which applies nonlinear forecast techniques to determine the class assignment for new consumers. In this paper, we present new results on the classification of consumers using recurrent neural networks in the forecast based classification framework. The results are compared with existing classification methods using real, measured power consumption data. We demonstrate that consumer classification performed by recurrent neural networks can outperform existing methods as in several cases the correct class assignment rate is near to 100%.

A probabilistic demand side management approach by consumption admission control

New generation electricity network called Smart Grid is a recently conceived vision for a cleaner, more efficient and cheaper electricity system. One of the major challenges of electricity network is that generation and consumption should be balanced at every moment. This paper introduces a new concept for controlling the demand side by the means of automatically enabling/disabling electric appliances to make sure that the demand is in match with the available supplies, based on the statistical characterization of the need. In our new approach instead of using hard limits we estimate the tail probability of the demand distribution and control system by using the principles and the results of statistical resource management.

Deep Learning Based Consumer Classification for Smart Grid

Classification of different power consumers is a very important task in smart power transmission grids as the different type of consumers may be treated with different conditions. Furthermore, the power suppliers can use the category information of consumers to forecast better their behavior which is a relevant task for load balancing.

Jensen-Shannon divergence based algorithm for adaptive segmentation and labelling of household's electricity power consumption data series

The increasing presence of renewable energy sources and the novel consumption types will obviously cause the increase of the fluctuation of electrical power in households. In order to better manage the electrical power consumption and production the integration of information and communication technologies and power grid is necessary, which is obviously a recent research topic. The availability of large amount of measurement data provided by households smart meter(s) offers new possibilities in analyzing the internal structure of power consumption data series. One of them is discovering typical power consumption patterns with their duration-distributions. Our recent achievements in this direction are presented in the paper, namely a novel on-line, Jensen-Shannon divergence-based adaptive and automatic segmentation algorithm, the segment descriptors and the results of clustering using Kohonens self-organizing map.

On the convexity of Chernoff bound in the context of consumption admission control in smart grids

In this paper we prove the convexity of the Chernoff bound in the context of estimating the overconsumption in a bottom-up consumption modelling framework in smart grid. We also introduce three application areas for the Chernoff bound in the context of power systems, more specifically the reliability assessment, consumption admission control and capacity sizing. The proof of convexity shows the method to be a computationally feasible tool for these three application areas.

Minimum Probability of Error-Based Equalization Algorithms for Fading Channels

Novel channel equalizer algorithms are introduced for wireless communication systems to combat channel distortions resulting from multipath propagation. The novel algorithms are based on newly derived bounds on the probability of error (PE) and guarantee better performance than the traditional zero forcing (ZF) or minimum mean square error (MMSE) algorithms. The new equalization methods require channel state information which is obtained by a fast adaptive channel identification algorithm. As a result, the combined convergence time needed for channel identification and PE minimization still remains smaller than the convergence time of traditional adaptive algorithms, yielding real-time equalization. The performance of the new algorithms is tested by extensive simulations on standard mobile channels.