Serap Cekli

Classification of chaotic circuit output patterns with probabilistic neural networks

This study focused on the classification of chaotic circuit behaviors with probabilistic neural network (PNN). Although, chaotic circuit outputs track similar traces for the defined parameters, still the circuit outputs preserve their own random characteristics at each trial. PNN is an effective tool for classification of pattern recognition problems. Inherited features of PNN are very compatible with the chaotic circuit output classification problem and it provides satisfying performance. The selection of the proper features in the feature extraction step defines the performance of the classification significantly. In order to, compare classification performance of the PNN, different feature vectors are employed in the training process. Moreover, the spread parameter is a considerably vital factor for the performance of the network. The simulation results and the corresponding illustrations for the performance analysis are also given.

Position detection with spherical interpolation least squares based on time difference of arrivals using separated acoustic signals by independent component analysis

In this study, locations of two sources have been detected using source signals separated with independent component analysis (ICA) method. The spherical interpolation least squares method which is based on the time difference of arrivals (TDOA) between the microphones is used for the solution of position detection. Although, there are different approaches and methods based on the TDOA for position detection, these methods presents a solution for a single source in general. In addition, the efficient signal separation is another problem when there are more than one target object in the sensor field. Therefore, the emitted signals which are received by the randomly distributed sensors (microphones) in the field are separated by ICA method. The source locations are found using the spherical interpolation least squares method regarding the signal groups which have the relative time differences between each other. The mean square error graphics and results are presented in the study for different signal to noise ratio values.

An efficient SPIHT algorithm and system architecture for image compression

In this study, an efficient modified SPIHT algorithm and related architecture design is proposed for image compression. The proposed architecture is designed as a three stage pipeline structure. Different sized images are compressed using the designed architecture for different bpp values. The computed PSNR values of the compressed images using the proposed algorithm and also the original algorithm is compared in order to test the performance of the proposed algorithm and architecture.

Kinect brain waves with controlled smart automation system for als patients

Nowadays, smart automation systems which directly affect peoples life styles are being studied and designed frequently. These studies are carried out in a home environment, hospital or wheelchair to satisfy users needs effectively. But these studies mainly focused on the healthy people. In this study a smart home automation system is proposed for ALS patients, orthopedic disabilities. In this work, Emotive device, Kinect One and Arduino Yún microcontroller were used to establish automation system which will render especially disabled peoples daily life more safely and practically while they are at home.

Analysis of Transformer Resonant Overvoltages Due to Switching Capacitance

Power transformer transients, which are originated from non-linear operating characteristics, may cause failure of a power system. In this study, a special resonance phenomenon where overvoltages observed due to switching capacitances of the line is investigated. This situation is detected while high voltage side of the transformer is open circuit (no-load) and the low voltage side is connected to a transmission line through nonlinear switch (or breaker). This special resonance is caused by the interaction of leakage inductance of the transformer windings and the switching capacitance. A test setup containing 220V/34.5kV transformer is employed for realization of simulation results obtained by SPICE analysis. In order to simulate switching capacitance, a capacitance of 5.54μF is connected to low voltage side. The recorded and simulated test results show good correspondence. Observed overvoltages are analysed by continuous wavelet transform (CWT) and fast Fourier transform (FFT) techniques. Also the energy distribution, which is generated by overvoltages, is investigated by instantaneous energy levels. Keywords—switching capacity, power transformer, overvoltage, continuous wavelet transform, resonance

Modelling of chaotic surface tracking on the polymeric insulators with Hidden Markov Models

In this study, chaotic surface tracking patterns observed on polymeric high voltage (HV) outdoor insulation materials were investigated and simulated. The polymeric samples are tested according to the IEC 587 Inclined Plane Tracking Test Standard. Since the chaotic surface tracking patterns manifests smutty and disordered images, they are preprocessed and purified by image processing tools. Internal and external effects may severely decrease insulation performance. In order to examine external effects, samples are subjected to moisture and vibration effect. Polymer samples are investigated by their fractal dimension which is a prominent tool for analyzing chaotic images. To simulate these chaotic surface tracking patterns Hidden Markov Models (HMM) are used.

Amplitude and frequency estimation of power system signals using independent component analysis

In this paper independent component analysis (ICA) method for amplitude and frequency estimation of distorted power system signals is proposed. In order to protect system and keep it in safe operation mode the amplitude and frequency estimation should be conducted accurately. Transient disturbances of the power system may reduce estimation performance due to the distortion strength. In this study white noise and pulse noise which are very common for power systems, are used to contaminate power system signal. Thus, the proposed method is employed to decompose noise from system signal and hence to improve the efficiency of the estimation. Computer simulations have been carried out for the performance analysis of the ICA method and the comparison of the results of the proposed method with the conventional filters are displayed by using mean square error (MSE) values.

Three dimentional acoustic source location estimation with maximum likelihood estimation using Kalman filter

In this study, a novel algorithm based on the Maximum Likelihood Estimator (MLE) is proposed to estimate three dimensional position of an acoustic source which is located in a field with randomly deployed sensors. The received sensor signals have been filtered by using Kalman filter in order to eliminate the background noise. Filtered signals have been subjected to the estimation procedure to obtain the three dimensional source location. Consequently, the usage of Kalman filter has provided better estimation performance in accordance with the case without Kalman filter. Moreover, the performance decreases due to the larger parameter space in the procedure of estimating three dimensional location of the source. In the three dimensional estimation scenario Kalman filter drastically improves the performance of the estimator. The performances of the estimation procedures with and without Kalman filter has been investigated by plotting the estimator variances. The performance of the algorithm has been analyzed by comparing the simulation results and the Cramer Rao Bound expressions.