(1990). ENERGY FUNCTION ANALYSIS FOR POWER SYSTEM STABILITY. Electric Machines & Power Systems: Vol. 18, No. 2, pp. 209-210.

Energy function analysis for power system stability

http://ira.lib.polyu.edu.hk/bitstream/10397/5259/1/EAAI7.pdf

Artificial neural network simulation of hourly groundwater levels in a coastal aquifer system of the Venice lagoon

Control in Power Electronics : selected problems

Soft computing techniques are becoming even more popular and particularly amenable to model the complex behaviors of most geotechnical engineering systems since they have demonstrated superior predictive capacity, compared to the traditional methods. This paper presents an overview of some soft computing techniques as well as their applications in underground excavations. A case study is adopted to compare the predictive performances of soft computing techniques including eXtreme Gradient Boosting (XGBoost), Multivariate Adaptive Regression Splines (MARS), Artificial Neural Networks (ANN), and Support Vector Machine (SVM) in estimating the maximum lateral wall deflection induced by braced excavation. This study also discusses the merits and the limitations of some soft computing techniques, compared with the conventional approaches available.

https://dr.ntu.edu.sg/bitstream/10356/146887/2/1-s2.0-S1674987119302361-main.pdf

State-of-the-art review of soft computing applications in underground excavations

The contextual-based convolutional neural network (CNN) with deep architecture and pixel-based multilayer perceptron (MLP) with shallow structure are well-recognized neural network algorithms, representing the state-of-the-art deep learning method and the classical non-parametric machine learning approach, respectively. The two algorithms, which have very different behaviours, were integrated in a concise and effective way using a rule-based decision fusion approach for the classification of very fine spatial resolution (VFSR) remotely sensed imagery. The decision fusion rules, designed primarily based on the classification confidence of the CNN, reflect the generally complementary patterns of the individual classifiers. In consequence, the proposed ensemble classifier MLP-CNN harvests the complementary results acquired from the CNN based on deep spatial feature representation and from the MLP based on spectral discrimination. Meanwhile, limitations of the CNN due to the adoption of convolutional filters such as the uncertainty in object boundary partition and loss of useful fine spatial resolution detail were compensated. The effectiveness of the ensemble MLP-CNN classifier was tested in both urban and rural areas using aerial photography together with an additional satellite sensor dataset. The MLP-CNN classifier achieved promising performance, consistently outperforming the pixel-based MLP, spectral and textural-based MLP, and the contextual-based CNN in terms of classification accuracy. This research paves the way to effectively address the complicated problem of VFSR image classification.

A hybrid MLP-CNN classifier for very fine resolution remotely sensed image classification

A new simple and effective algorithm of arcing fault detection in distribution networks with the application of a wavelet transform technique is presented in this paper. The protection algorithm developed observes the phase displacement between wavelet coefficients calculated for zero-sequence voltage and current signals at a chosen high-level frequency. The final decision in regards to feeder switching off (or alarm issuing) is met either with a deterministic logic scheme or with the use of a neural net trained especially for that purpose. The developed wavelet-based high-impedance fault (HIF) detector has been tested with Electromagnetic Transients Program-Alternative Transients Program (ATP)-generated signals, exhibiting better performance than traditionally used algorithms and methods. The protection method proposed may be used for HIF detection independent of the network neutral-point grounding mode. The scheme proved to be robust against transients generated during normal events such as feeder energizing and de-energizing as well as capacitor bank switching

High-impedance fault detection in distribution networks with use of wavelet-based algorithm

The signal processing methods and algorithms described in preceding chapters were expressed in form of explicit equations, transfer functions and/or logic rules, either in crisp or in fuzzy versions. There are, however, specific tasks and power system operation conditions when, especially for the problems that are complex and difficult to express in terms of traditional means, other solutions should be applied. In such situations, both for signal processing and decision making Artificial Neural Networks may constitute a good solution.

Application of Artificial Neural Networks

Application of two new ANN-based algorithms for arcing high impedance fault (HIF) detection in multigrounded medium-voltage (MV) distribution networks is presented in this paper. The paper provides an evaluation of two new structures of artificial neural networks (ANNs) that may be used for reliable HIF detection in multigrounded as well as isolated, compensated, and grounded via small resistance distribution grids. The results obtained by use of both neural nets are presented. The performance was tested using data obtained from staged HIFs in real MV network as well as from electromagnetic transients program-alternative transients program simulations. A small number of necessary neurons in developed ANNs, short measuring sliding data window, and easy interpretation of obtained output signals are the main advantages of the proposed approach. Satisfactory results of ANN performance were observed for all examined HIF cases in which the ground fault current was greater than 16 A. The selected ANNs of best performance show high reliability and immunity to transients resulting from switching operations in protected feeders and from capacitor bank switching.

New ANN-Based Algorithms for Detecting HIFs in Multigrounded MV Networks

A new algorithm of power system frequency estimation is presented. The algorithm applies orthogonal signal components obtained with use of two orthogonal FIR filters. The essential property of the algorithm presented is its outstanding immunity to both signal orthogonal component magnitudes and FIR filter gains variations which ensures 1.5 mHz accuracy of estimation over typical (±2 Hz) range of measured frequency deviation. The results of the algorithm simulation tests using the EMTP software package are also included. It was confirmed that the algorithm dynamics are good enough to track power system frequency even under transient conditions.

Power system frequency estimation

If a current transformer (CT) saturates while transforming the heavy primary current, in each period, there are sections when the magnetic core is unsaturated and transformation is correct, and when it is saturated, it causes enormous errors. The key to correct the errors is to detect instants of saturation and moments when the saturation ends. This paper presents simple and efficient methods of doing that. In addition, it shows the way how to predict up to three samples after saturation with the acceptable degree of accuracy, thus artificially expanding the unsaturated section. It may substantially facilitate operation of protective devices. The ultimate solution of the problem is to estimate the values of the fundamental frequency component amplitude and the decaying DC component in each period. This paper presents the method of doing that on the ground of four consecutive samples taken during the unsaturated section. If the primary current consists of the fundamental and DC exponential components only, the accuracy of the method is excellent. Neglecting contamination of the current signal with harmonics and noise, it is possible to reconstruct the CT primary current based on the measured values of fundamental and DC components.

Waldemar Rebizant

Papers

High-impedance fault detection in distribution networks with use of wavelet-based algorithm

Application of Artificial Neural Networks

New ANN-Based Algorithms for Detecting HIFs in Multigrounded MV Networks

Power system frequency estimation

Correction of Current Transformer Transient Performance