Search results

In many applications, very fast methods are required for estimating of parameters of harmonic signals distorted by noise. Most of the known digital algorithms are not fully parallel, so that the speed of processing is quite limited. In this paper new parallel algorithms are proposed, which can be implemented by analogue adaptive circuits employing some neural networks principles. Algorithms based on the least‐squares (LS) and the total least‐squares (TLS) criteria are developed and compared. The problems are formulated as optimization problems and solved by using the steepest descent continuous‐time optimization algorithm. The corresponding architectures of analogue neuron‐like adaptive processors are also shown. The developed networks are more robust against noise in the measured signal than other known neural network algorithms. The network based on the TLS criterion optimizes the estimation under the assumption that the signal model can also be perturbated (frequency or sampling interval fluctuation and so forth). The TLS estimates are better and more reliable than the corresponding LS estimates, when applying a higher sampling frequency and a wider sampling window. The TLS algorithm is a generalization of the well known LMS rule and could be in some applications superior to the family of LMS algorithms. Extensive computer simulations confirm the validity and performance of the proposed algorithms.

We present an approach for blind separation of acoustic sources produced from multiple speakers mixed in realistic room environments. We first transform recorded signals into the time‐frequency domain to make mixing become instantaneous. We then separate the sources in each frequency bin based on an independent component analysis (ICA) algorithm. For the present paper, we choose the complex version of fixedpoint iteration (CFPI), i.e. the complex version of FastICA, as the algorithm. From the separated signals in the time‐frequency domain, we reconstruct output‐separated signals in the time domain. To solve the so‐called permutation problem due to the indeterminacy of permutation in the standard ICA, we propose a method that applies a special property of the CFPI cost function. Generally, the cost function has several optimal points that correspond to the different permutations of the outputs. These optimal points are isolated by some non‐optimal regions of the cost function. In different but neighboring bins, optimal points with the same permutation are at almost the same position in the space of separation parameters. Based on this property, if an initial separation matrix for a learning process in a frequency bin is chosen equal to the final separation matrix of the learning process in the neighboring frequency bin, the learning process automatically leads us to separated signals with the same permutation as that of the neighbor frequency bin. In each bin, but except the starting one, by chosen the initial separation matrix in such a way, the permutation problem in the time domain reconstruction can be avoided. We present the results of some simulations and experiments on both artificially synthesized speech data and real‐world speech data, which show the effectiveness of our approach.

The purpose of this paper is to develop the new method of estimation of the short‐term largest Lyapunov exponent of electroencephalogram (EEG) waveforms for the detection and prediction of the epileptic seizure.

The paper proposed the modifications concerned with the way of selection of the segments of EEG waveforms taking part in estimation of Lyapunov exponent, as well as determination of the distances between two time series. The proposed method is based on Kolmogorov‐Smirnov test of similarity of two vectors. Through the application of this test more accurate and less parameterized approach to the estimation of the short‐term largest Lyapunov exponent of EEG waveforms has been obtained.

The results of performed experiments have shown that in most cases our modified method has outperformed the classical procedure, leading to more stable results, closer to the neurologist indications. The analysis of the data has proved that the change of the largest Lyapunov exponent provides a lot of information regarding the epileptic seizure. The minimum value of Lyapunov exponent indicates fairly well the seizure moment. The T_index applied for few different electrode sites can provide good advanced prediction of the incoming epileptic seizure.

After additional experiments this method may find practical application for supporting the medical diagnosis of the epilepsy.

The proposed modification of the estimation of the short‐term largest Lyapunov exponent of the EEG waveforms eliminates some arbitrarily chosen parameters tuned by the user and leads to more accurate estimate. Such estimation results are better suited for the characterization of the epileptic activity.

Proposes a non‐negative matrix factorization method.

Presents an algorithm for finding a suboptimal basis matrix. This is controlled by data cluster centers which can guarantee that the coefficient is very sparse. This leads to the proposition of an application of non‐matrix factorization for blind sparse source separation with less sensors than sources.

Two simulation examples reveal the validity and performance of the algorithm in this paper.

Using the approach in this paper, the sparse sources can be recovered even if the sources are overlapped to some degree.

The aim of this paper is to develop the accurate computer method of predicting the average PM10 pollution for the next day on the basis of some measured atmospheric parameters, like temperature, humidity, wind, etc. This method should be universal and applicable for any place under consideration.

The paper presents the new approach to the accurate forecasting of the daily average concentration of PM10. It is based on the application of the ensemble of neural networks and wavelet transformation of the time series, representing PM10 pollution.

On the basis of numerical experiments, the paper finds that application of many neural predictors cooperating with each other can significantly improve the quality of results. The paper shows that the developed forecasting system checked on the data of PM10 pollution in Warsaw generated good overall accuracy of prediction in terms of root mean squared error, mean absolute error and mean absolute percentage error.

The main novelty of the proposed approach is the application of the wavelet transformation and many neural networks organized in the form of ensemble. The individual neural predictors are integrated into one forecasting system using different forms of integrations, including the blind source separation method and neural‐based integration.

This paper aims to present a prototype of the diagnostic system for the examination of the distribution of the force applied by foot to substrate during usual human moving. Presented system is competitive to other currently available devices, thanks to sensors reliability, user-friendly operation manner and design based on cheap parts. The results of examinations are transmitted by radiomodem. Its recording and visualization are possible on either personal or mobile computers.

During selection of the sensors substrate, many polymeric electrets were examined. Polyvinylidene fluoride films were selected, because they have good charge uniformity across the surface, wide range of acceptable temperatures, linear relation between mechanical stress and output signal and high resistance for squeezing. The system measures the charge generated in film.

The pressures are recorded in relation to maximum value; therefore, measuring system does not require calibration. The simultaneous recording of data from all eight sensors allows tracking the signal without distortion.

An array of sensors is installed in the shoe insole. The measuring device is fixed to the outer surface of the shoe. Its weight is 75 g. The range of transmission is suitable for examination in the natural environment, outside traditional consulting room. Software is dedicated for analysis of the pressure distribution in every moment of the foot movement. The system is suitable for examination of flat feet, diabetic foot and recovery progress after injures.

This paper aims to present the details of isotactic polypropylene (it-PP) films with a cellular structure (air-cavities) dedicated to pressure sensors. The polymer composites (thin films enriched with 5 and 10 wt% of mineral fillers as Sillikolloid P 87 and glass beads) should exhibit suitable structural elasticity within specific stress ranges. After the deformation force is removed, the sensor material must completely restore its original shape and size.

Estimating the stiffness tensor element (C₃₃) for polymer films (nonpolar space-charge electrets) by broadband resonance ultrasound spectroscopy is a relatively simple method of determining the safe stress range generated in thin pressure sensors. Therefore, ultrasonic and piezoelectric studies were carried out on four composite it-PP films. First, the longitudinal velocity (v_L) of ultrasonic waves passing through the it-PP film in the z-direction (thickness) was evaluated from the ω-position of mechanical resonance of the so-called insertion loss function. In turn, the d₃₃ coefficient was calculated from accumulated piezoelectric charge density response to mechanical stress.

Research is at an early stage; however, it can be seen that the mechanical orientation of the it-PP film improves its piezoelectric properties. Moreover, the three-year electric charge stability of the it-PP film seems promising.

Ultrasonic spectroscopy can be successfully handled as a validation method in the small-lot production of polymer films with the air-cavities structure intended for pressure sensors. The structural repeatability of polymer films is strongly related to a homogeneous distribution of the electric charge on the electret surface.

Aims to evaluate different thick‐film materials for use in strain sensors and temperature sensors on low‐temperature co‐fired ceramic (LTCC) substrates.

LTCC materials are sintered at the low temperatures typically used for thick‐film processing, i.e. around 850°C, The thick‐film resistor materials for use as strain and temperature sensors on LTCC tapes are studied. Thick‐film piezo‐resistors in the form of strain‐gauges are realised with 10 kΩ/sq. 2041 (Du Pont)and 3414‐B (ESL), resistor materials; thick‐film temperature‐dependent resistors were made from PTC 5093 (Du Pont), and NTC‐4993 (EMCA Remex) resistor materials.

The X‐ray spectra of the 2041 and 3414‐Bb low TCR resistors after drying at 150°C and after firing display more or less the same peaks. The electrical characteristics of 2041 resistors fired on alumina and LTCC substrates are similar indicating that the resistors are compatible with the LTCC material. After firing on LTCC substrates the sheet resistivities and TCRs of the 3414‐B resistors increased. Also, there is a significant increase in the GFs from 13 to over 25.

Investigates the compatibility of thick‐film materials and the characteristics of the force and temperature sensors.