Authors: Dirk Vries, Karel Keesman, Hans Zwart

Abstract: A linear regressive model structure and output predictor, both in algebraic form, are deduced from an LTI state space system with certain properties without the need of direct matrix inversion. On the basis of this, explicit expressions of parametric sensitivities are given. As an example, a diffusion process is approximated by a state space discrete time model with n compartments in the spatial plane and is then reparametrized. The system output can then be explicitly predicted by ŷk = θT φk-n - ेk-n as a function of n, the sensor position, the parameter vector θ, and input-output data. This method is attractive for estimation, prediction and insight in experimental design issues, when physical knowledge is to be preserved.

Keywords: linear estimation,linear prediction,distributed parameter systems,regression analysis,sensitivity analysis

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Jonas Gillberg, Lennart Ljung

Abstract: This paper treats the identification of continuous-time output error (OE) models based on sampled data. The exact method for doing this is well known both for data given in the time and frequency domains. The time domain approach however, becomes somewhat complex, especially for non-uniformly sampled data. In this paper we assume that the system input is a zero order hold signal and that the sampling rate is so high that for high frequencies the system behaves as a set of integrators. The conclusion is that if the system has relative degree l then the output should be interpolated using an l order polynomial spline function.

Keywords: continuous-time systems,parameter estimation,continuous-time,output error,B-splines

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Mats Tallfors, Alf J. Isaksson, Mattias Nordin

Abstract: This paper studies the problem of estimating mechanical parameters in drive trains. The task is to fit a two-mass model with backlash to experimental data. Normally in, for example, rolling mill applications the measured output is just the motor speed. However, for situations with a light load in comparison to the motor, measurement of motor speed only is insufficient to reach good estimates of all parameters in a two-mass model. As a potential solution we advocate use of a shaft torque measurement in addition to the motor speed. The suggested approach is to use a procedure based on three dedicated experiments: 1) One experiment with a sequence of setpoint steps while maintaining the controller in automatic. This enables estimation of the static gain yielding the total inertia damping and static friction. 2) Then an experiment tailored to guaranteeing that no gap openings are encountered, thus enabling estimation of the other mechanical parameters using linear black-box techniques. 3) Finally yet another experiment that with certainty contains gap openings. A non-linear mechanical model is then used, in combination with a numerical search method to also estimate the gap size. The developed procedure has then been tested successfully on real data from a test rig.

Keywords: grey-box identification,black-box identification,mechanical parameters,drive train,backlash,shaft torque

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Yoshinori Takei, Hidehito Nanto, Shunshoku Kanae, Zi-Jiang Yang, Kiyoshi Wada

Abstract: Using a unified approach, recursive algorithms of the error covariance matrices in subspace methods are derived for the MOESP type of subspace methods. The proposed approach is based on the fact that the subspace extraction amounts to computing singular value decomposition of the Schur complement (SC) of the input submatrix in data product moments and the SC can be interpreted as the least squares residuals. The recursion of the error covariance matrix can be applied to derive recursive subspace identification algorithms.

Keywords: interpretation,subspace methods,identification,MIMO,stochastic systems

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Jay D. Schwartz, Daniel E. Rivera

Abstract: The development of control-oriented decision policies for inventory management in supply chains has received considerable interest in recent years, and demand modeling to supply forecasts for these policies is an important component of an effective solution to this problem. Drawing from the problem of control-relevant identification, we present an approach for demand modeling based on data that relies on a control-relevant prefilter to tailor the emphasis of the fit to the intended purpose of the model, which is to provide forecast signals to a tactical inventory management policy based on Model Predictive Control. Integrating the demand modeling and inventory control problems offers the opportunity to obtain reduced-order models that exhibit superior performance, with potentially lower user effort relative to traditional "open-loop" methods. A systematic approach to generating these prefilters is presented and the benefits resulting from their use are demonstrated on a representative production/inventory system case study.

Keywords: control-relevant identification,supply chain management,demand modeling,model reduction

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Erik Wernholt, Svante Gunnarsson

Abstract: In the work presented here, a three-step identification procedure for rigid body dynamics, friction, and flexibilities, introduced in (Wernholt and Gunnarsson, 2005), will be utilized and extended. Using the procedure, the parameters can be identified only using motor measurements. In the first step, rigid body dynamics and friction will be identified using a separable least squares method, where a friction model describing the Striebeck effect is used. In the second step, initial values for flexibilities are obtained using inverse eigenvalue theory. Finally, in the last step, the remaining parameters of a nonlinear physically parameterized model are identified directly in the time domain. The procedure is exemplified using real data from an experimental industrial robot.

Keywords: identification,robotics,flexible arms,friction,manipulators

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Jason Wong, Geoffrey M. Shaw, Christopher E. Hann, Thomas Lotz, Jessica Lin, J. Geoffrey Chase

Abstract: Metabolic modelling can significantly improve control of hyperglycaemia. Clinical control demands physiological accuracy in identifying patient specific parameters. However, typically used non-linear and non-convex identification methods and models can deliver sub-optimal results, affecting control prediction. This research compares a typical non-linear method and a novel linear, convex method for an accepted metabolic control model using retrospective clinical control data. Results show increased errors in fitting for the non-linear fitting method. A significant (140-660X) increase in computational efficiency is also reported. The methods and results presented can be readily applied and generalised to a wider set of pharmacokinetic and pharmacodynamic systems that use similar linear and non-linear models.

Keywords: medical systems,physiological models,system identification,insulin sensitivity,integrals

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Bjorn Sohlberg

Abstract: This paper deals with a case study of grey box modelling where known parts are modelled using a priori information and unknown parts are described as general continuous nonlinear functions, which are approximated by means of Taylor series including higher order terms. In the Taylor series, the partial derivatives are estimated from measured data by minimising the maximum likelihood function. This approach is used to keep the number of estimated parameters low. The modelling procedure follows a structured approach including; basic modelling, data acquisition, model calibration, expanded modelling, stochastic modelling and model appraisal.

Keywords: grey box modelling,identification,nonlinear models,steel industry,Taylor series

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: M. Blanke, M. Knudsen

Abstract: Grey box model identification preserves known physical structures in a model but with limits to the possible excitation, all parameters are rarely identifiable, and different parametrizations give significantly different model quality. Convenient methods to show which parameterizations are the better would be very useful. This paper shows how we can assess the parameter interdependence and model quality. Hessian matrix decomposition is employed to show linear dependencies between variables and to put a quality tag on different parameterizations. The method determines parameter relations that need be constrained to achieve satisfactory convergence. Identification of nonlinear models for a ship illustrate the concept.

Keywords: grey-box identification,marine systems,parameter inter-depence

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Han-Fu Chen, Xiao-Li Hu

Abstract: For the Hammerstein and Wiener systems the paper gives i) the strongly consistent estimates for the coeffcients contained in the linear subsystem; ii) the strongly consistent estimate for f(u) at any u; iii) the optimal adaptive regulation control. No assumption is made on the structure of f(ċ). The estimates and adaptive control are given by the stochastic approximation algorithms with expanding truncations.

Keywords: Hammerstein system,Wiener system,nonparametric nonlinearity,recursive estimate,adaptive regulation,strong consistency,stochastic approximation

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Mark J. W. Musters, Danie J. W. Lindenaar, Aleksandar Lj. Juloski, Natal A. W. van Riel

Abstract: Disentangling the complexity of biochemical networks requires knowledge about the quantitative relationships between the individual components. However, the nonlinear dynamics of biochemical processes are difficult to identify with traditional identification methods. We therefore propose to model the complex nonlinear biochemical processes with several simpler systems (modes), together with a switching rule that determines which mode is active, i.e. with a hybrid system. We consider the example of a nonlinear biochemical oscillator, and propose a simple piecewise affine (PWA) approximation. Qualitative analysis shows that the PWA model is able to capture the dynamics of the nonlinear oscillator. Hybrid identification procedure is subsequently applied to identify the parameters of the PWA model.

Keywords: hybrid models,identification algorithms,least-squares identification,parameter estimation,physiological models,prediction error methods

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Erik Etien, Damien Halber, Gerard Champenois, Regis Ouvrard

Abstract: Jiles Atherton model is described in discrete form. Least square identification is improved using normalization of sensibility functions. Experimental tries validate the proposed method.

Keywords: magnetic hysteresis,Jiles-Atherton model,least square identification,convergence improvement

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Thomas Lotz, J. Geoffrey Chase, Kirsten A. McAuley, Jessica Lin, Jason Wong, Chris E. Hann, Steen Andreassen

Abstract: Modelling can enhance the diagnosis and control of metabolic disorders. Clinical effectiveness demands physiological accuracy, patient specificity and identification with limited data. A two-compartment insulin kinetics model and associated insulin-glucose pharmacodynamics are presented. Similarities with C-peptide kinetics are used to simplify parameter identification. Critical patient specific parameters are identified using a novel convex, integral-based method. The model and methods are validated within physiological ranges using euglycaemic clamp (N=146) and IVGTT data. The mean absolute errors in the resulting glucose and insulin profiles are eG = 5.9% ± 6.6% SD and eI = 6.2% ± 6.4% SD for the clamps and area under glucose and insulin profiles deviated eAG = 1.6% and eAI = 6.7% during IVGTT.

Keywords: medical systems,physiological models,system identification,insulin sensitivity,integrals

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Girish N. Nair, Minyi Huang, Robin J. Evans

Abstract: This paper considers the optimal control of linear systems where measurement data is transmitted from the plant output to the controller over a noiseless communication channel with limited instantaneous data rate. The cost is defined to be the average, over a random initial state, of the usual infinite horizon quadratic regulation criterion, and the number of bits transported by the channel during each sampling interval is bounded. Several fundamental properties of the optimal cost functional are derived for initial state densities that satisfy a mild moment condition. Using these properties, precise expressions for the optimal cost and policy are obtained assuming a uniformly distributed initial state. These expressions agree with the classical optimal LQR results in the high data rate limit and with recent minimum rate results in the low rate regime. Extensions to the case of non-uniform densities and vector-valued states are discussed.

Keywords: optimal control,communication channels,quantization

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Keith R. Godfrey, Ai Hui Tan, H. Anthony Barker, W. Dhammika Widanage

Abstract: Three computer programs, all freely available on the World Wide Web, for designing and generating different classes of perturbation signals for system identification are described. Two of the programs are for designing pseudorandom signals, which have fixed power spectra. The first program is for designing signals based on five known classes of pseudorandom binary or near-binary sequences, and the second program is for designing pseudorandom signals based on maximum-length sequences, both binary and multilevel. The third program is for designing multilevel multiharmonic signals, in which the user can specify the harmonic pattern required, and a computer optimization routine is then used to meet the specification as closely as possible.

Keywords: binary signals,frequency responses,input signals,multilevel codes,pseudorandom sequences,system identification,time-domain responses

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Luca Pivano, Thor Inge Fossen, Tor Arne Johansen

Abstract: This paper proposes a nonlinear dynamics model for a marine propeller able to reproduce the propeller thrust over the full four-quadrant range of propeller shaft speed and vessel speed. A two-state model has been identified from experimental data. The model includes a state equation for the propeller shaft speed and one that describes the dynamics of the axial flow velocity. The model reproduces accurately propeller thrust and torque over a wide range of operation.

Keywords: identification,marine,modeling,nonlinear,propulsion

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Marta Barenthin, Martin Enqvist, Bo Wahlberg, Hakan Hjalmarsson

Abstract: In this paper, we discuss and compare three different approaches for L2- gain estimation of Hammerstein systems. The objective is to find the input signal that maximizes the gain. A fundamental difference between two of the approaches is the class, or structure, of the input signals. The first approach involves describing functions and therefore the class of input signals is sinusoids. In this case we assume that we have a model of the system and we search for the amplitude and frequency that give the largest gain. In the second approach, no structure on the input signal is assumed in advance and the system does not have to be modelled first. The maximizing input is found using an iterative procedure called power iterations. In the last approach, a new iterative procedure tailored for memoryless nonlinearities is used to find the maximizing input for the unmodelled nonlinear part of the Hammerstein system. The approaches are illustrated by numerical examples.

Keywords: Hammerstein system,L2-gain,power iterations,describing function

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Christian P. Robert

Abstract: The toolbox available in Bayesian Statistics has increased considerably in the past decade and it has opened new avenues for Bayesian inference, the first and foremost being Bayesian model choice. The MCMC and particle filter technologies have hugely increased the potential for Bayesian applications, in particular in missing variable models, as illustrated in this short tutorial. We will also mention a new direction in this field, namely the development of adaptive algorithms that avoid a lenghty tuning to fit the problem at hand by automatically modifying the parameters of the algorithm.

Keywords: adaptivity,Bayesian inference,MCMC algorithm,Monte Carlo techniques,missing variables,model choice,particle filter,population Monte Carlo

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Bo Wahlberg, Hakan Hjalmarsson, Marta Barenthin

Abstract: System identification concerns the construction and validation of mathematical models of dynamical systems from experimental data. The objective of this contribution is to discuss new research directions in experiment design, e.g. how to design informative experiments which satisfy specifications on the resulting model quality and practical limitations such as constraints on input and output signals, but also experimental time. In particular, we discuss how input design is instrumental for alleviating the problem of modelling complex systems. Many optimal experiment design problems can be formulated as optimal control problems, but with nonstandard cost functions. A difficulty is that the solution often depends on the true system. To solve optimal control problems, we can exploit recent advances in numerical optimization for control design, including convex optimization and relaxation methods. As a more concrete example, we study how to estimate the H∞ norm of a system.

Keywords: system identification,input design,robust control,model validation,linear quadratic control

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Christopher E. Hann, Steen Andreassen, Bram W. Smith, Geoffrey M. Shaw, J. Geoffrey Chase, Per L. Jensen

Abstract: A minimal cardiac model that accurately captures the essential cardiovascular system dynamics has been developed. Standard parameter identification methods for this model are highly non-linear and non-convex, hindering clinical application, given the limited measurements available in an intensive care unit. This paper presents an integral based identification method that transforms the problem into a linear, convex problem. Five common disease states including four fundamental types of shock, are identified to within 10% without false identification. Clinically, it enables medical staff to rapidly obtain a patient specific model to assist in diagnosis and therapy selection.

Keywords: biomedical systems,physiological models,integrals,parameter identification,diagnosis

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Piet M. T. Broersen

Abstract: The program ARMAsel automatically selects a single time series model for given stationary stochastic data. Three model types are candidates for selection. The type can be AR or autoregressive, MA or moving average and the combined ARMA type. The parameters of that selected model accurately represent the power spectral density and the autocorrelation function of the data. The reduced statistics ARMAsel-rs algorithm uses a long AR model as only input to compute models of the other types and to select the best. ARMAsel-mis is a new program that can handle missing data. It computes AR models with a numerically stable maximum likelihood algorithm and uses ARMAsel-rs to determine MA and ARMA models. The order and the type of the best candidate are automatically selected with dedicated missing data criteria, supposing that data are randomly missing. Missing less than 10 % of the data has not much influence on the accuracy, missing 50 % generally looses a factor two in the accuracy in comparison with the model estimated from the same number of contiguous observations. Low order time series models can be computed as long as the product of the remaining fraction and the remaining number of observations is greater than about 10.

Keywords: ARMA model,autoregressive model,autocovariance estimation,missing observations,order selection,parametric model,spectral estimation

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Sunil L. Kukreja, Johan Lofberg, Martin J. Brenner

Abstract: Identification of parametric nonlinear models involves estimating unknown parameters and detecting its underlying structure. Structure computation is concerned with selecting a subset of parameters to give a parsimonious description of the system which may afford greater insight into the functionality of the system or a simpler controller design. In this study, a least absolute shrinkage and selection operator (LASSO) technique is investigated for computing efficient model descriptions of nonlinear systems. The LASSO minimises the residual sum of squares by the addition of a l1 penalty term on the parameter vector of the traditional l2 minimisation problem. Its use for structure detection is a natural extension of this constrained minimisation approach to pseudolinear regression problems which produces some model parameters that are exactly zero and, therefore, yields a parsimonious system description. The performance of this LASSO structure detection method was evaluated by using it to estimate the structure of a nonlinear polynomial model. Applicability of the method to more complex systems such as those encountered in aerospace applications was shown by identifying a parsimonious system description of the F/A-18 Active Aeroelastic Wing using flight test data.

Keywords: system identification,nonlinear systems,structure detection,aeroelasticity

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Benjamin Golding, Andrew Ross, Thor I. Fossen

Abstract: This paper presents a method for online estimation of nonlinear viscous damping forces in the horizontal plane for a surface vessel at low speed. The method is based on parameter estimation in combination with qualitative knowledge about longitudinally distributed drag coefficients, derived from experimental data. Case studies indicate that the method provides accurate estimates under realistic conditions.

Keywords: modelling,system identification,marine systems

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Rong Deng, Patricia Davies, Anil K. Bajaj

Abstract: A five-parameter fractional derivative model and a hereditary model are being studied to predict polyurethane foam's uniaxial responses under harmonic excitation. A system identification procedure is developed to estimate the model parameters. The prediction results from both models are presented and compared. Both models provide reasonably good prediction of the observed responses from different input levels at a given compression level.

Keywords: system identification,fractional derivative model,hereditary model,polyurethane foam,harmonic response

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006

Authors: Marion Gilson, Guillaume Mercere

Abstract: This paper deals with an optimal instrumental variable method dedicated to subspace-based closed-loop system identification. The presented solution is based on the MOESP technique but requires to modify the original scheme by proposing a new PO MOESP method which uses reconstructed past input and past output data as instrumental variables. The developed approach is then illustrated via a simulation example and a comparison with other subspace-based methods.

Keywords: closed-loop identification,MIMO systems,subspace methods,instrumental variable,optimal estimation

Identifier: None

Conference: 14th IFAC Symposium on System Identification, 2006

Location: , Australia

Start Date: Wed Mar 29 2006 - End Date: Fri Mar 31 2006