Background: The purpose of the paper is to propose different arrangements of neural networks for short-time 24-h load forecasting in Power Systems. Methods: The study discusses and compares different techniques of data processing, applying the feedforward and recurrent neural structures. They include such networks as multilayer perceptron, radial basis function, support vector machine, self-organizing Kohonen networks, deep autoencoder, and recurrent deep LSTM structures. The important point in getting high-quality results is the composition of many solutions in the common ensemble and their fusion to create the final forecast of time series. The paper considers and compares different methods of fusing the individual results into the final forecast, including the averaging, application of independent component analysis, dynamic integration, and wavelet transformation. Results: The numerical experiments have shown a high advantage of using many individual predictors integrated into the ensemble which are responsible for the final forecast. Especially efficient is the application of non-standard wavelet application in the formation of an ensemble, as well as the use of LSTM as the basic prediction unit. The novelty of the paper is the critical comparative analysis of the time series prediction methods applied for load forecasting in the power system. The presented approach may be useful for the users involved in power system operation management.
1 December 2016 in Bulletin Of The Polish Academy Of Sciences-Technical Sciences
This paper presents an originally-developed system for design and optimization of AC-DC converters dedicated in particular to operation in distributed generation systems. The proposed procedure is based on a multi-objective discrete optimization and expert knowledge of electrical engineering, especially power electronics. The required accuracy of calculations is obtained by using the database with real components, while the parameters applied in calculations are based on parameters provided by the manufacturer. The paper presents the foundations and basic system properties, the design and optimization process, and selected optimization results, obtained with a fully functional prototype of the design and optimization system (DaOS).
17 May 2022 in Bulletin Of The Polish Academy Of Sciences-Technical Sciences
The paper considers the problem of increasing the generalization ability of classification systems by creating an ensemble of classifiers based on the CNN architecture. Different structures of the ensemble will be considered and compared. Deep learning fulfills an important role in the developed system. The numerical descriptors created in the last locally connected convolution layer of CNN flattened to the form of a vector, are subjected to a few different selection mechanisms. Each of them chooses the independent set of features, selected according to the applied assessment techniques. Their results are combined with three classifiers: softmax, support vector machine, and random forest of the decision tree. All of them do simultaneously the same classification task. Their results are integrated into the final verdict of the ensemble. Different forms of arrangement of the ensemble are considered and tested on the recognition of facial images. Two different databases are used in experiments. One was composed of 68 classes of greyscale images and the second of 276 classes of color images. The results of experiments have shown high improvement of class recognition resulting from the application of the properly designed ensemble.
This paper discusses some numerical aspects of the simulation of electroconvulsive therapy (ECT). A realistic finite-element model of the human head is used to discuss two approaches to modeling the electrodes applied to human head skin. The first approach models the electrode by a mixed-boundary condition, while the second one uses additional subdomain imitating electrode-to-skin contact for that purpose [three-dimensional (3-D) model]. An algorithm of grid modification used to add an external subdomain modeling the electrode contact resistance is presented. The authors examine the influence of the electrode model on the convergence speed of the iterative solver. The authors state that the 3-D model is better conditioned, and, thus, it converges faster
The recent trends in software development transform classical desktop-based programs into distributed systems offered over the Internet. The authors adapt this approach to the radiotherapy planning software. A general concept and necessary components of such a system are presented.