Showing papers on "System identification published in 1995"

A least-squares approach to blind channel identification

Abstract: Conventional blind channel identification algorithms are based on channel outputs and knowledge of the probabilistic model of channel input. In some practical applications, however, the input statistical model may not be known, or there may not be sufficient data to obtain accurate enough estimates of certain statistics. In this paper, we consider the system input to be an unknown deterministic signal and study the problem of blind identification of multichannel FIR systems without requiring the knowledge of the input statistical model. A new blind identification algorithm based solely on the system outputs is proposed. Necessary and sufficient identifiability conditions in terms of the multichannel systems and the deterministic input signal are also presented.

Structural-System Identification. I: Theory

Abstract: The investigation reported in this paper looks into the application of a number of system-identification techniques to problems of earthquake engineering. A number of techniques for structural-system identification have been developed over the past few years. Many of these techniques have been successful at identifying properties of linearized and time-invariant equivalent structural systems. Most of these techniques were verified using mathematical models simulated on the computer. In this paper, a number of structural-identification algorithms are reviewed and applied to the identification of structural systems subjected to earthquake excitations. The algorithms are applied to experimental data obtained in controlled laboratory conditions. The data pertain to the acceleration records from two building models subjected to various loading conditions. The performance of the various identification algorithms is critically assessed, and guidelines are obtained regarding their suitability to various engineeri...

Blind channel identification based on second-order statistics: a frequency-domain approach

Abstract: In this communication, necessary and sufficient conditions are presented for the unique blind identification of possibly nonminimum phase channels driven by cyclostationary processes. Using a frequency domain formulation, it is first shown that a channel can be identified by the second-order statistics of the observation if and only if the channel transfer function does not have special uniformly spaced zeros. This condition leads to several necessary and sufficient conditions on the observation spectra and the channel impulse response. Based on the frequency-domain formulation, a new identification algorithm is proposed. >

Optimal fuzzy inference for short-term load forecasting

Abstract: This paper proposes an optimal fuzzy inference method for short-term load forecasting. The proposed method constructs an optimal structure of the simplified fuzzy inference that minimizes model errors and the number of the membership functions to grasp nonlinear behavior of power system short-term loads. The model is identified by simulated annealing and the steepest descent method. The proposed method is demonstrated in examples.

Identification of ARMAX model for short term load forecasting: an evolutionary programming approach

Abstract: This paper proposes a new evolutionary programming (EP) approach to identify the autoregressive moving average with exogenous variable (ARMAX) model for one day to one week ahead hourly load demand forecasts. Typically, the surface of forecasting error function possesses multiple local minimum points. Solutions of the traditional gradient search based identification technique therefore may stall at the local optimal points which lead to an inadequate model. By simulating natural evolutionary process, the EP algorithm offers the capability of converging towards the global extremum of a complex error surface. The developed EP based load forecasting algorithm is verified by using different types of data for practical Taiwan power (Taipower) system and substation load as well as temperature values. Numerical results indicate the proposed EP approach provides a method to simultaneously estimate the appropriate order and parameter values of the ARMAX model for diverse types of load data. Comparisons of forecasting errors are made to the traditional identification techniques.

System identification for adaptive and reconfigurable control

Abstract: In this work, an efficient static system identification method is developed that incorporates prior information. The prior information from flight mechanics enhances the performance of the identification algorithm, yielding more accurate parameter estimates. The proposed static system identification approach is superior to dynamic system identification for on-line identification of rapidly varying plant parameters, making it suitable for adaptive and reconfigurable control. The effectiveness of the static system identification scheme is illustrated in an example problem. A derivative F-16 aircraft control surface (elevator) failure is simulated and the abruptly changing pitch channel control system's stability and control derivatives are successfully identified on-line in the presence of unmodeled dynamics, process noise (clear air turbulence), and realistic measurement noise.

Structural System Identification. II: Experimental Verification

Abstract: The investigation reported in this paper looks into the application of a number of system-identification techniques to problems of earthquake engineering. A number of techniques for structural-system identification have been developed over the past few years. Many of these techniques have been successful at identifying properties of linearized and time-invariant equivalent structural systems. Most of these techniques were verified using mathematical models simulated on the computer. In this paper, a number of structural-identification algorithms are reviewed and applied to the identification of structural systems subjected to earthquake excitations. The algorithms are applied to experimental data obtained in controlled laboratory conditions. The data pertain to the acceleration records from two building models subjected to various loading conditions. The performance of the various identification algorithms is critically assessed, and guidelines are obtained regarding their suitability to various engineering applications.

Non-Linear System Identification with Neural Networks

Abstract: This thesis addresses the non-linear system identification problem, and in particular, investigates the use of neural networks in system identification. An overview of different possible mode! structures is given in a common framework. A nonlinear structure is described as the concatenation of a map from the observed data to the regressor, and a map from the regressor to the output space. This divides the model structure selection problem into two problems with lower complexity: that of choosing the regressor and that of choosing the non-linear map.The possible choices for the regressors consists of past inputs and outputs, and filtered versions of them. The dynamics of the mode! depends on the choice of regressor, and families of different mode! structures are suggested based on analogies to linear black-box models. State-space models are also described within this common framework by a special choice of regressor. It is shown that state-space models which have no parameters in the state update function can be viewed as an input-output mode! preceded by a pre-filter. A parameterized state update function, on the other hand, can be seen as a data driven regressor selector. The second step of the nonlinear identification is the mapping from the regressor to the output space. It is often advantageous to try some intermediate mappings between the linear and the general non-linear mapping. Such non-linear black-box mappings are discussed and motivated by considering different noise assumptions.The validation of a linear mode! should contain a test for non-linearities and it is shown that, in general, it is easy to detect non-linearities. This implies that it is not worth spending too much energy searching for optimal non-linear validation methods for a specific problem. lnstead the validation method should be chosen so that it is easy to apply. Two such methods, based on polynomials and neural nets, are suggested. Further, two validation methods, the correlation-test and the parametric F-test, are investigated. It is shown that under certain conditions these methodscoincide.Parameter estimates are usually based on criterion minimization. In connection with neural nets it has been noted that it is not always optimal to try to find the absolute minimum point of the criterion. Instead a better estimate can be obtained if the numerical search for the minimum is prematurely stopped. A forma! connection between this stopped search and regularization is given. It is shown that the numerical minimization of the criterion can be view as a regularization term which is gradually turned to zero. This closely connects to, and explains, what is called overtraining in the neural net literature.

Identifying MIMO Wiener systems using subspace model identification methods

Abstract: In this paper we show that the multivariable output-error state-space model (MOESP) class of sub-space model identification (SMI) schemes can be extended to identify Wiener systems, a series connection of a linear dynamic system followed by a static nonlinearity. In this paper, we restrict to present these extensions for the case the Taylor series expansion of the static nonlinearity contains odd terms. It is shown that the extension allows to identity the linear part of the Wiener systems as if the static nonlinearity is not present. In this way, it is related to cross-correlation analysis techniques.

Improved structure selection for nonlinear models based on term clustering

Abstract: In this paper the concepts of term clusters and cluster coefficients are defined and used in the context of system identification. It is argued that if a certain type of term in a nonlinear model is spurious, the respective cluster coefficient is small compared with the coefficients of the other clusters represented in the model. Once the spurious clusters have been detected, the corresponding terms can be deleted from the set of candidate terms. The consequences of doing this are (i) a drastic reduction in the size of the set of candidate terms and, consequently, a substantial gain in computation time is achieved; (ii) the final estimated model is more likely to reproduce the dynamics of the original system; and (iii) the final model is more robust to overparametrization. Numerical examples are included to illustrate the new procedure.

Low-order black-box models for control system design in large power systems

Abstract: The paper studies two multi-input multi-output (MIMO) procedures for the identification of low-order state space models of power systems, by probing the network in open loop with low-energy pulses or random signals. Although such data may result from actual measurements, the development assumes simulated responses from a transient stability program, hence benefiting from the existing large base of stability models. While pulse data is processed using the eigensystem realization algorithm, the analysis of random responses is done by means of subspace identification methods. On a prototype Hydro-Quebec power system, including SVCs, DC power lines, series compensation, and more than 1100 buses, it is verified that the two approaches are equivalent only when strict requirements are imposed on the pulse length and magnitude. The 10th-order equivalent models derived by random-signal probing allow for effective tuning of decentralized power system stabilizers (PSSs) able to damp both local and very slow inter-area modes.

Worst case system identification in H/sub /spl infin//: validation of a priori information, essentially optimal algorithms, and error bounds

Abstract: Resolves several important open issues pertaining to a worst-case control-oriented system identification problem known as identification in H/sub /spl infin//. First, a method is presented for developing confidence that certain a priori information available for identification is not invalid. This method requires the solution of a certain nondifferentiable convex program. Second, an essentially optimal identification algorithm is constructed. This algorithm is (worst-case strongly) optimal to within a factor of two. Finally, new upper and lower bounds on the optimal identification error are derived and used to estimate the identification error associated with the given algorithm. Interestingly, the development of each of these results draws heavily upon the classical Nevanlinna-Pick interpolation theory. As such, the authors' results establish a clear link between the areas of system identification and optimal interpolation theory. Both the formulation and techniques in this paper are applicable to problems where the frequency data available for identification may essentially be arbitrarily distributed. >

Modeling and identification of parallel nonlinear systems: structural classification and parameter estimation methods

Abstract: Structural classification and parameter estimation (SCPE) methods are used for studying single-input single-output (SISO) parallel linear-nonlinear-linear (LNL), linear-nonlinear (LN), and nonlinear-linear (NL) system models from input-output (I-O) measurements. The uniqueness of the I-O mappings of some model structures is discussed. The uniqueness of I-O mappings of different models tells us in what conditions given model structures can be differentiated from one another. Parameter uniqueness of the I-O mapping of a given structural model is also discussed, which tells us in what conditions a given model's parameters can be uniquely estimated from I-O measurements. These methods are then generalized so that they can be used to study single-input multi-output (SIMO), multi-input single-output (MISO), as well as multi-input multi-output (MIMO) nonlinear system models. Parameter estimation of the two-input single-output nonlinear system model, which was left unsolved previously, can now be obtained using the newly derived algorithms. Applications of SCPE methods for modeling visual cortical neurons, system fault detection, modeling and identification of communication networks, biological systems, and natural and artificial neural networks are also discussed. The feasibility of these methods is demonstrated using simulated examples. SCPE methods presented in this paper can be further developed to study more complicated block-structured models, and will therefore have future potential for modeling and identifying highly complex multi-input multi-output nonlinear systems. >

A numerical approach to genetic programming for system identification

Abstract: This paper introduces a new approach to genetic programming (GP), based on a numerical technique, which integrates a GP-based adaptive search of tree structures, and a local parameter tuning mechanism employing statistical search (a system identification technique). In traditional GP, recombination can cause frequent disruption of building blocks or mutation can cause abrupt changes in the semantics. To overcome these difficulties, we supplement traditional GP with a local hill-climbing search, using a parameter tuning procedure. More precisely, we integrate the structural search of traditional GP with a multiple regression analysis method and establish our adaptive program, called STROGANOFF (STructured Representation On Genetic Algorithms for NOn-linear Function Fitting). The fitness evaluation is based on a minimum description length (MDL) criterion, which effectively controls the tree growth in GP. We demonstrate its effectiveness by solving several system identification (numerical) problems and compare the performance of STROGANOFF with traditional GP and another standard technique (radial basis functions). We then extend STROGANOFF to symbolic (nonnumerical) reasoning by introducing multiple types of nodes, using a modified MDL-based selection criterion and a pruning of the resultant trees. The effectiveness of this numerical approach to GP is demonstrated by successful application to symbolic regression problems.

Model validation tests for multivariable nonlinear models including neural networks

Abstract: A fast and concise MTMO nonlinear model validity test procedure is derived, based on higher order correlation functions, to form a global-to-local hierarchical validation diagnosis of identified MEMO linear and nonlinear models. The new procedure is applied to four MIMO nonlinear system models including a neural network training example, to demonstrate the effectiveness of the tests.

Issues in nonlinear stochastic grey box identification

Abstract: Grey box identification refers to the practice of identifying dynamical systems in model structures exploiting partial prior information. This contribution reviews a method for stochastic grey box identification and surveys experiences and lessons of applying it to a number of industrial processes. Issues to be addressed include advantages and costs of introducing stochastics into the model, the question of what contribution must be expected from the model designer as opposed to what can be formalized in computer algorithms, and an outlook on future plans to resolve present shortcomings.

A system identification approach to the detection of changes in both linear and non‐linear structural parameters

Abstract: This paper explores the potential of a new time domain identification procedure to detect changes in structural dynamic characteristics on the basis of measurements. This procedure is verified using mathematical models simulated on the computer. The experiments involve two eight-storey steel structures with and without energy devices, and a 47-storey building at San Francisco during the Loma Prieta earthquake. The recursive instrumental variable method and extended Kalman filter algorithm are used as identification algorithms. An exploratory investigation is made of the usefulness of various indices, such as mode shape and storey drift, that can be extracted accurately from identification to quantify changes in the characteristics of the physical system. It is concluded that the change of storey drift is the key information to the detection of changes in structural parameters, from which the proposed system identification algorithm can be applied with an appropriate inelastic model to simulate the dynamic behaviour of real structures undergoing strong ground motion excitations.