Search

Structural analysis of signals is an important part of smart energy. It includes adaptive control of the aperiodic, low-frequency, and signal-harmonics levels for relay protection and automation tasks, fault location, and monitoring and diagnostics of electrical systems. For real-time signal processing, adaptive filtering is widely used and elements of artificial intelligence are increasingly being employed. This paper considers feed-forward artificial neural networks (multilayer perceptron) for signal processing. The simplest artificial neural networks (ANNs) can replace neural networks with a more complex structure (convolutional, recurrent), but within the framework of a sequential recurrent algorithm. This makes it possible to control the quality of parameter identification and signal processing at each stage of calculations. The proposed algorithm is tested using the example of sliding window processing of a signal with nonlinear distortions and at CT saturation. It is shown that the amplitude, frequency, and phase of a utility frequency signal with a high level of harmonics and aperiodic component can be identified with an accuracy of a few percent in a time not exceeding a few milliseconds. To improve the accuracy at each calculation step, in addition to ANNs, traditional methods can be used, such as averaging and median smoothing. The presented results demonstrate the possibility of real-time structural analysis of signals. [ABSTRACT FROM AUTHOR]

A method of estimating nonlinear distortions, which differs from the well-known ones in that it enables one to determine the nonlinear distortions of random signals, which carry useful information, is described. A new criterion is proposed for estimating these distortions, the basis of which is a calculation of the mean value of the relative change in the gain of the circuit which introduces the distortions. A block diagram of the measuring circuit is presented. [ABSTRACT FROM AUTHOR]

The paper proposes a nonlinear acoustic echo cancellation setup based on a decoupled scheme of the second-order adaptive Volterra filter. The acoustic distortions that take place in the Loudspeaker-Enclosure-Microphone setup are modeled using measured linear and quadratic Volterra kernels. The adaptation of the filter weights is achieved by exploiting the Normalized Least-Mean-Fourth algorithm. The proposed method is compared with the usual direct adaptation procedure of the second-order Volterra filter using the same algorithm and the same step size values. Comparisons between these techniques regard the evolution of the Echo Return Loss Enhancement in terms of convergence rate and steady-state error. Experimental results for Laplacian distributed excitation show that the new approach increases the speed of convergence in comparison to the one given by the coupled version without enhancing the steady-state error. [ABSTRACT FROM AUTHOR]

The design of pseudorandom maximum length (PRML) signals for linear identification in the presence of nonlinear distortions is considered. For this application, it is advantageous for the signal to have harmonic multiples of two and three suppressed, in order to minimize the effect of non-linearity, thus resulting in a better estimate of the underlying linear dynamics. Such signals may be designed through exhaustive search for the sequence-to-signal conversions. However, for signals generated from Galois fields GF(q) with q large, this method is computationally inefficient. An alternative technique is proposed where a primitive version of the signal, the period of which is considerably shorter than that of the required PRML signal, is first generated as a computer-optimized signal. The primitive signal is then used to define the conversions for the generation of the required PRML signal, which is a member of a new class of hybrid signal. [ABSTRACT FROM AUTHOR]

A frequency domain analysis of the influence of non- linear distortions on linear frequency response function measurements performed using multisine test signals is presented in this paper. The aim is to provide a closer insight into the mechanisms at work when multisine signals are subjected to static power-series nonlinearities. Emphasis is given to the behavior of a class of nonlinear distortions, termed hype II contributions, when different types of multisine signals are applied to a cubic, fifth-, seventh-, and ninth-order static nonlinear element The original results presented in this paper are a development of work conducted by the current authors as well as other authors in this area. [ABSTRACT FROM AUTHOR]

In this letter, we show how different signal processing techniques can be combined to optimize the performance of a typical wideband nonlinear satellite link with spectrally efficient high-level modulation techniques. In particular, we follow two concurrent approaches: on one side, we pursue signal optimization in the form of special amplitude and phase shift keying (APSK) constellations to reduce the effect of nonlinear distortions, while on the other, we analyze the feasibility of adaptive data predistortion (DP) at the transmitter and adaptive nonlinear equalization (NLE) at the receiver. We demonstrate that the use of such optimized constellations relieves the complexity of the nonlinearity compensation techniques, and we also show that by clever adoption of these techniques the sensitivity to nonlinear distortion of both uncoded and turbo coded quadrature amplitude modulation and APSK constellations is greatly reduced. [ABSTRACT FROM PUBLISHER]

Using specially designed broadband periodic random excitation signals, the open loop, the common mode, and the power supply gains of operational amplifiers are measured and modeled. The proposed modeling technique 1) takes into account the measurement uncertainty and the nonlinear distortions, 2) gives information about possible unmodeled dynamics, 3) detects, quantifies, and classifies the nonlinear distortions, and 4) provides opamp parameters (time constants, gain-bandwidth product, etc.) with confidence bounds. The approach is suitable for the experimental characterization of operational amplifiers (see [23]) as well as the fast evaluation of new operational amplifiers designs using network simulators (see Part I). Part I describes the modeling approach and illustrates the theory on simulations. [ABSTRACT FROM AUTHOR]

This paper examines the output properties of static power-series nonlinearities driven by periodic multiharmonic signals with emphasis given to their effect on linear frequency response function (FRF) measurements. The analysis is based on the classification of nonlinear distortions into harmonic and interharmonic contributions. The properties of harmonic contributions are examined in detail and explicit formulae are derived, by which the number of harmonic contributions generated at the test frequencies can be calculated for odd-order nonlinearities up to, and including, the ninth order. Although an analytic solution for any odd-order nonlinearity is still under investigation, a heuristic methodology is developed that solves this problem. It is shown that the derived formulae provide a useful tool in the examination of the behavior of FRF measurements in the presence of nonlinear distortions. Based on these formulae, different approaches in classifying nonlinear distortions are then compared with respect to their suitability in assessing the influence of system nonlinearities on linear FRF measurements. [ABSTRACT FROM AUTHOR]