China
Africa
Explore chapters and articles related to this topic
Noise Reduction Techniques
Published in David C. Swanson, ®, 2011
Adaptive Noise Cancellation significantly suppresses, if not completely removes, the coherent components of an available noise reference signal from the signal of interest through the use of optimal Wiener filtering. Figure 15.9 shows a block diagram for a straightforward Wiener filter used to remove the interference noise correlated with an available “reference noise,” thus leaving the residual error containing mostly the desired signal [6–8]. If coherence is high between the reference noise and the waveform with our desired signal plus the noise interference, the cancellation can be nearly total. This of course requires that we have a very accurate model of the transfer function between the reference noise and our waveform. The Wiener filtering in Section 11.2 is mainly concerned with the identification of this transfer function system, or “system identification.” A valuable by-product of precise system identification is the capability of separating the reference noise components from the output waveform to reveal the “desired signal” with the noise interference removed.
Discrete-Time Signals and Systems
Published in Anastasia Veloni, Nikolaos I. Miridakis, Erysso Boukouvala, Digital and Statistical Signal Processing, 2018
Anastasia Veloni, Nikolaos I. Miridakis, Erysso Boukouvala
System analysis includes the study of system inputs and outputs over time and frequency. Through this study, an attempt is made to understand the complexity of the system and classify it into a category. System modeling aims to “limit the description” of the system to the basic mathematical relations that govern its behavior. These mathematical relations are expressed in terms of parameters that need to be defined to respond to the operation of the particular system at some point in time. These parameters may be constant or time-varying. System identification is intended to determine the parameters of the model so that its behavior is “as close as possible” to that of the system. Finally, the quality of the model must be checked. Different criteria are used for this purpose.
Preliminaries
Published in Wong Gabriyel, Wang Jianliang, Real-Time Rendering: Computer Graphics with Control Engineering, 2017
The goal of system identification is to derive a mathematical model of a dynamic system based on observed input and output data. Usually a priori information pertaining to a system will be useful for postulating the preliminary model structure. The system may then be modelled according to empirical data (black-box modelling) or conceivable mathematical functions such as physical laws (white-box modelling). Often, real world systems are non-linear and operate with reliance on state memory. The systems are dynamic and thus their outputs may depend on a combination of previous inputs, outputs, and states. The combination provides the basis for time series and regression mathematical expressions (models) for different reproducible systems.
Neural network-based parametric system identification: a review
Published in International Journal of Systems Science, 2023
Aoxiang Dong, Andrew Starr, Yifan Zhao
System identification can generally be divided into two types: parametric approaches and nonparametric approaches. The nonparametric system identification investigates the system’s specific properties by analysing the observed data directly without a model. In contrast, the parametric system identification unveils the system’s inherent dynamics by building a universal approximation model based on observed data. The analysis or prediction of the actual system is based on the model instead of the measurement data. One prominent advantage of such an approach is that it is still able to represent the system well when the noise is nonlinearly and causally correlated to the system inputs and outputs. A mathematic model of noise can be built and accommodated in the general model creating an unbiased estimation of the mean of the system output distribution (Billings, 2013). There are two significant challenges in parametric system identification.
Adaptive Control Based on LMS Algorithm for Grid-Connected Inverters
Published in Electric Power Components and Systems, 2023
Goran S. Nikolić, Tatjana R. Nikolić, Goran Lj. Djordjević
In the considered power system, the inverter block, which includes the PWM block and the power inverter, presents a system to identify. The nonlinear system comprises the inverter block and the block for phase and amplitude adjustment and synchronization of the grid voltage, see Figure 4. The block for adjustment of the grid voltage can be represented as a nonlinear static block, while the inverter block can be represented as a linear dynamic block. System identification as a method for modeling dynamic systems is used to approximate the unknown system with a linear model based on the available input and output data. For identifying the model of an unknown system, two system identification approaches can be used: parametric and nonparametric estimation techniques [36]. Nonparametric methods use spectral and correlation analysis to directly compute the frequency response or impulse response of the system and, in this way, estimate the behavior of the system. In parametric techniques, the overall structure and the order of the system model (number of poles, zeros) are necessary to define before estimating the unknown system. The parameters of the model are identified using information extracted from the system. The selection of the candidate model depends on the application and its complexity is often influenced by the approximations that can be made.
Optimal sensors placement for structural health monitoring based on system identification and interpolation methods
Published in Journal of the Chinese Institute of Engineers, 2021
In terms of time domain analyses, several methods have been widely used. First, the Hilbert-Huang Transform (HHT) method, which could be characterized as a version of real-time method, has been particularly influential in contributing insights to the modal analysis field. Second, those methods which are derived from non-real-time methods, such as the autoregressive model method (Bhansali 1978), and the subspace identification method have seen their use mounting steadily during recent years (Qin 2006). System identification is one kind of numerical analysis that uses measured data to build mathematical models of dynamic systems. In general, system identification methods have been commonly branched into either the time domain or the frequency domain (Hoa et al. 2010). The SSI is one of the methods extended from subspace identification. Since SSI does not need external forces data and is more convenient and accurate than other subspace identification methods, the SSI is adopted as the system identification method for this study.