Abstract: In this paper, various algorithms for designing quadrature mirror filter are reviewed and a new algorithm is presented for the design of near perfect reconstruction quadrature mirror filter bank. In the proposed algorithm, objective function is formulated using the perfect reconstruction condition or magnitude response condition of prototype filter at frequency (ω = 0.5π) in ideal condition. The cutoff frequency is iteratively changed to adjust the filters coefficients using optimization algorithm. The performances of the proposed algorithm are evaluated in term of computation time, reconstruction error and number of iterations. The design examples illustrate that the proposed algorithm is superior in term of peak reconstruction error, computation time, and number of iterations. The proposed algorithm is simple, easy to implement, and linear in nature.
Abstract: Robust face recognition under various illumination
environments is very difficult and needs to be accomplished for
successful commercialization. In this paper, we propose an improved
illumination normalization method for face recognition. Illumination
normalization algorithm based on anisotropic smoothing is well known
to be effective among illumination normalization methods but
deteriorates the intensity contrast of the original image, and incurs less
sharp edges. The proposed method in this paper improves the previous
anisotropic smoothing-based illumination normalization method so
that it increases the intensity contrast and enhances the edges while
diminishing the effect of illumination variations. Due to the result of
these improvements, face images preprocessed by the proposed
illumination normalization method becomes to have more distinctive
feature vectors (Gabor feature vectors) for face recognition. Through
experiments of face recognition based on Gabor feature vector
similarity, the effectiveness of the proposed illumination
normalization method is verified.
Abstract: The paper is concerned with developing stochastic delay mechanisms for efficient multicast protocols and for smooth mobile handover processes which are capable of preserving a given Quality of Service (QoS). In both applications the participating entities (receiver nodes or subscribers) sample a stochastic timer and generate load after a random delay. In this way, the load on the networking resources is evenly distributed which helps to maintain QoS communication. The optimal timer distributions have been sought in different p.d.f. families (e.g. exponential, power law and radial basis function) and the optimal parameter have been found in a recursive manner. Detailed simulations have demonstrated the improvement in performance both in the case of multicast and mobile handover applications.
Abstract: Load forecasting has become in recent years one of the major areas of research in electrical engineering. Most traditional forecasting models and artificial intelligence neural network techniques have been tried out in this task. Artificial neural networks (ANN) have lately received much attention, and a great number of papers have reported successful experiments and practical tests. This article presents the development of an ANN-based short-term load forecasting model with improved generalization technique for the Regional Power Control Center of Saudi Electricity Company, Western Operation Area (SEC-WOA). The proposed ANN is trained with weather-related data and historical electric load-related data using the data from the calendar years 2001, 2002, 2003, and 2004 for training. The model tested for one week at five different seasons, typically, winter, spring, summer, Ramadan and fall seasons, and the mean absolute average error for one hour-ahead load forecasting found 1.12%.
Abstract: A complex valued neural network is a neural network
which consists of complex valued input and/or weights and/or thresholds
and/or activation functions. Complex-valued neural networks
have been widening the scope of applications not only in electronics
and informatics, but also in social systems. One of the most important
applications of the complex valued neural network is in signal
processing. In Neural networks, generalized mean neuron model
(GMN) is often discussed and studied. The GMN includes a new
aggregation function based on the concept of generalized mean of all
the inputs to the neuron. This paper aims to present exhaustive results
of using Generalized Mean Neuron model in a complex-valued neural
network model that uses the back-propagation algorithm (called
-Complex-BP-) for learning. Our experiments results demonstrate the
effectiveness of a Generalized Mean Neuron Model in a complex
plane for signal processing over a real valued neural network. We
have studied and stated various observations like effect of learning
rates, ranges of the initial weights randomly selected, error functions
used and number of iterations for the convergence of error required on
a Generalized Mean neural network model. Some inherent properties
of this complex back propagation algorithm are also studied and
discussed.
Abstract: This paper presents a method for the optimal
allocation of Distributed generation in distribution systems. In this
paper, our aim would be optimal distributed generation allocation for
voltage profile improvement and loss reduction in distribution
network. Genetic Algorithm (GA) was used as the solving tool,
which referring two determined aim; the problem is defined and
objective function is introduced. Considering to fitness values
sensitivity in genetic algorithm process, there is needed to apply load
flow for decision-making. Load flow algorithm is combined
appropriately with GA, till access to acceptable results of this
operation. We used MATPOWER package for load flow algorithm
and composed it with our Genetic Algorithm. The suggested method
is programmed under MATLAB software and applied ETAP
software for evaluating of results correctness. It was implemented on
part of Tehran electricity distributing grid. The resulting operation of
this method on some testing system is illuminated improvement of
voltage profile and loss reduction indexes.
Abstract: Local Linear Neuro-Fuzzy Models (LLNFM) like other neuro- fuzzy systems are adaptive networks and provide robust learning capabilities and are widely utilized in various applications such as pattern recognition, system identification, image processing and prediction. Local linear model tree (LOLIMOT) is a type of Takagi-Sugeno-Kang neuro fuzzy algorithm which has proven its efficiency compared with other neuro fuzzy networks in learning the nonlinear systems and pattern recognition. In this paper, a dedicated reconfigurable and parallel processing hardware for LOLIMOT algorithm and its applications are presented. This hardware realizes on-chip learning which gives it the capability to work as a standalone device in a system. The synthesis results on FPGA platforms show its potential to improve the speed at least 250 of times faster than software implemented algorithms.
Abstract: Wind turbines with double output induction
generators can operate at variable speed permitting conversion
efficiency maximization over a wide range of wind velocities. This
paper presents the performance analysis of a wind driven double
output induction generator (DOIG) operating at varying shafts speed.
A periodic transient state analysis of DOIG equipped with two
converters is carried out using a hybrid induction machine model.
This paper simulates the harmonic content of waveforms in various
points of drive at different speeds, based on the hybrid model
(dqabc). Then the sinusoidal and trapezoidal pulse-width–modulation
control techniques are used in order to improve the power factor of
the machine and to weaken the injected low order harmonics to the
supply. Based on the frequency spectrum, total harmonics distortion,
distortion factor and power factor. Finally advantages of sinusoidal
and trapezoidal pulse width modulation techniques are compared.
Abstract: This work consists of three parts. First, the alias-free
condition for the conventional two-channel quadrature mirror filter
bank is analyzed using complex arithmetic. Second, the approach
developed in the first part is applied to the complex quadrature mirror
filter bank. Accordingly, the structure is simplified and the theory is
easier to follow. Finally, a new class of complex quadrature mirror
filter banks is proposed. Interesting properties of this new structure
are also discussed.
Abstract: The aim of this paper is to present a new method
which can be used for progressive transmission of electrocardiogram
(ECG). The idea consists in transforming any ECG signal to an
image, containing one beat in each row. In the first step, the beats are
synchronized in order to reduce the high frequencies due to inter-beat
transitions. The obtained image is then transformed using a discrete
version of Radon Transform (DRT). Hence, transmitting the ECG,
leads to transmit the most significant energy of the transformed
image in Radon domain. For decoding purpose, the receptor needs to
use the inverse Radon Transform as well as the two synchronization
frames.
The presented protocol can be adapted for lossy to lossless
compression systems. In lossy mode we show that the compression
ratio can be multiplied by an average factor of 2 for an acceptable
quality of reconstructed signal. These results have been obtained on
real signals from MIT database.
Abstract: Artificial Neural Network (ANN) has been
extensively used for classification of heart sounds for its
discriminative training ability and easy implementation. However, it
suffers from overparameterization if the number of nodes is not
chosen properly. In such cases, when the dataset has redundancy
within it, ANN is trained along with this redundant information that
results in poor validation. Also a larger network means more
computational expense resulting more hardware and time related
cost. Therefore, an optimum design of neural network is needed
towards real-time detection of pathological patterns, if any from heart
sound signal. The aims of this work are to (i) select a set of input
features that are effective for identification of heart sound signals and
(ii) make certain optimum selection of nodes in the hidden layer for a
more effective ANN structure. Here, we present an optimization
technique that involves Singular Value Decomposition (SVD) and
QR factorization with column pivoting (QRcp) methodology to
optimize empirically chosen over-parameterized ANN structure.
Input nodes present in ANN structure is optimized by SVD followed
by QRcp while only SVD is required to prune undesirable hidden
nodes. The result is presented for classifying 12 common
pathological cases and normal heart sound.