Input Signal Decorrelation Applied to Adaptive Second-Order Volterra Filters in the Time Domain

Author(s):  
Fabian Kuech ◽  
Marcus Zeller ◽  
Walter Kellermann
2013 ◽  
Vol 655-657 ◽  
pp. 2202-2206
Author(s):  
Yuan Sheng Wang ◽  
Gui Ying Lu ◽  
Juan Yu ◽  
Bo Li

Influence of the damping ratio on the response fast performance to under-damped second-order system in the time domain has been discussed. The relationship between peak time and the input signal, the adjust time, and the system type has been analyzed. The response’s fast performance indicators are relative, and it is related to the input signal, the response of the system, and the type of system and its initial states. In conclusion, the peak time and the adjust time cannot reach a minimum at the same time. The fast response issue must be discussed in relation to specific cases, and it cannot be generalized.


1997 ◽  
Vol 05 (04) ◽  
pp. 355-370 ◽  
Author(s):  
E. K. Skarsoulis

A scheme for approximate normal-mode calculation of broadband acoustic signals in the time domain is proposed based on a second-order Taylor expansion of eigenvalues and eigenfunctions with respect to frequency. For the case of a Gaussian impulse source a closed-form expression is derived for the pressure in the time domain. Using perturbation theory, analytical expressions are obtained for the involved first and second frequency-derivatives of eigenvalues and eigenfunctions. The proposed approximation significantly accelerates arrival-pattern calculations, since the eigenvalues, the eigenfunctions and their frequency-derivatives need to be calculated at a single frequency, the central frequency of the source. Furthermore, it offers a satisfactory degree of accuracy for the lower and intermediate order modes. This is due to the fact that essential wave-theoretic mechanisms such as dispersion and frequency dependence of mode amplitudes are contained in the representation up to a sufficient order. Numerical results demonstrate the efficiency of the method.


2017 ◽  
Vol 2017 ◽  
pp. 1-11 ◽  
Author(s):  
Bing Bai ◽  
Lixiang Zhang

In order to identify the quadratic Volterra system simplified from the hydroturbine shaft system, a blind identification method based on the third-order cumulants and a reversely recursive method are proposed. The input sequence of the system under consideration is an unobservable independent identically distributed (i.i.d.), zero-mean and non-Gaussian stationary signal, and the observed signals are the superposition of the system output signal and Gaussian noise. To calculate the third-order moment of the output signal, a computer loop judgment method is put forward to determine the coefficient. When using optimization method to identify the time domain kernels, we combined the traditional optimization algorithm (direct search method) with genetic algorithm (GA) and constituted the hybrid genetic algorithm (HGA). Finally, according to the prototype observation signal and the time domain kernel parameters obtained from identification, the input signal of the system can be gained recursively. To test the proposed method, three numerical experiments and engineering application have been carried out. The results show that the method is applicable to the blind identification of the hydroturbine shaft system and has strong universality; the input signal obtained by the reversely recursive method can be approximately taken as the random excitation acted on the runner of the hydroturbine shaft system.


2019 ◽  
Vol 100 (1) ◽  
Author(s):  
Hong-Guo Li ◽  
Rui-Xue Zhang ◽  
Zhan-Dong Liu ◽  
Zong-Guo Li

2014 ◽  
Vol 14 (5) ◽  
pp. 270-278 ◽  
Author(s):  
P. Sovilj ◽  
M. Milovanović ◽  
D. Pejić ◽  
M. Urekar ◽  
Z. Mitrović

Abstract Measurement methods, based on the approach named Digital Stochastic Measurement, have been introduced, and several prototype and small-series commercial instruments have been developed based on these methods. These methods have been mostly investigated for various types of stationary signals, but also for non-stationary signals. This paper presents, analyzes and discusses digital stochastic measurement of electroencephalography (EEG) signal in the time domain, emphasizing the problem of influence of the Wilbraham-Gibbs phenomenon. The increase of measurement error, related to the Wilbraham-Gibbs phenomenon, is found. If the EEG signal is measured and measurement interval is 20 ms wide, the average maximal error relative to the range of input signal is 16.84 %. If the measurement interval is extended to 2s, the average maximal error relative to the range of input signal is significantly lowered - down to 1.37 %. Absolute errors are compared with the error limit recommended by Organisation Internationale de Métrologie Légale (OIML) and with the quantization steps of the advanced EEG instruments with 24-bit A/D conversion


Author(s):  
Yan-Lin Shao ◽  
Jens Bloch Helmers

A time-domain Higher-Order Boundary Element Method (HOBEM) based on cubic shape functions for second-order wave-current-body interaction developed by Shao & Faltinsen [1] is further refined by investigating the feasibility of adopting the unstructured meshes on the free surface and body surfaces from an open source mesh generator [2]. When the steady local flow effect is considered in the time-domain boundary-value-problem formulation, the advection terms in the free surface are part of the sources of numerical instability. In this paper, the advection terms are taken care of in an implicit way in a 4th order Runge-Kutta scheme with much better stability. Some numerical examples extensively studied in the literature are studied in order to validate the present numerical model.


1981 ◽  
Vol 21 (01) ◽  
pp. 129-140 ◽  
Author(s):  
J.N. Sharma ◽  
R.G. Dean

Abstract Most methods for wave force computation incorporate either the nonlinearities of the ocean surface for a single fundamental component or the random and/or directional characteristics using superposition of linear wave components. One exception is the intuitive "hybrid" method, which combines elements of linear and nonlinear waves. This paper describes and applies a method correct to the second order in wave height for calculating waves and wave forces caused by a directional wave spectrum on an offshore structure.Starting with a prescribed linear spectrum of directional waves, a set of random phases is generated and the second-order spectrum computed with phases defined by all contributing pairs of first-order components. Thus, with one realization of the spectrum complete up to the second order, the wave profile and water particle kinematics can be profile and water particle kinematics can be simulated in the time domain. The wave forces also are computed in the time domain, taking full account of their nonlinear and directional properties. The resulting wave forces at any level vary in direction and magnitude. The total wave forces summed over all piling of a structure are less than those for a unidirectional train of waves with the same one-dimensional spectrum.Several examples are presented to illustrate reductions in maximum wave forces caused by the directional distribution of waves. We found that for a single piling the maximum force decreases by a factor ranging from 1.0 to 0.61 as the directional spread increases from unidirectional to uniformity over a half plane. For a four-pile group on a square array of 300-ft (91.4-m) spacing, the corresponding decrease in the factor is 1.0 to 0.51 for a Bretschneider spectrum with a peak period of approximately 12 seconds. The results of this complete model are compared with the more intuitive and approximate hybrid method and are found to agree quite well. Force spectra are presented and discussed for the inline and transverse directions. Introduction The nonlinearity, randomness, and directionality of a real sea preclude a simple but realistic determination of wave loading on a single- or multiple-pile group. Presently, there are two essentially different but complementary methods for computing wave loadings. One method represents nonlinearities of a single wave composed of a characteristic fundamental period and its higher harmonics. A number period and its higher harmonics. A number of such theories have been-developed. Dalrymple extended the stream function approach of Dean, to waves on a shear current. Some of these theories adequately account for the nonlinearities; however, they avoid the random and directional characteristics of the sea surface. The second method uses the principle of linear superposition of an infinite principle of linear superposition of an infinite number of waves with given frequencies, amplitudes, and directions of propagation but independent phases; the total energy is distributed over a phases; the total energy is distributed over a continuum of frequencies and directions. In this manner, a three-dimensional Gaussian sea can be represented fully. However, ignoring the nonlinearities makes the random Gaussian model unrealistic - especially for large waves. SPEJ P. 129


Sign in / Sign up

Export Citation Format

Share Document