-
-
-
-
[show abstract]
[hide abstract]
ABSTRACT: This technical note studies identification problems for dual-rate sampled-data linear systems with noises. A hierarchical least squares (HLS) identification algorithm is presented to estimate the parameters of the dual-rate ARMAX models. The basic idea is to decompose the identification model of a dual-rate system into several sub-identification models with smaller dimensions and fewer parameters. The proposed algorithm is more computationally efficient than the recursive least squares (RLS) algorithm since the RLS algorithm requires computing the covariance matrix of large sizes, while the HLS algorithm deals with the covariance matrix of small sizes. Compared with our previous work, a detailed study of the HLS algorithm is conducted in this technical note. The performance analysis and simulation results confirm that the estimation accuracy of the proposed algorithm are close to that of the RLS algorithm, but the proposed algorithm retains much less computational burden.
IEEE Transactions on Automatic Control 12/2011; · 2.11 Impact Factor
-
IEEE Trans. Automat. Contr. 01/2011; 56:2677-2683.
-
Automatica. 01/2011; 47:1646-1655.
-
[show abstract]
[hide abstract]
ABSTRACT: This technical note addresses identification problems of non-uniformly sampled systems. For the input-output representation of non-uniform discrete-time systems, a partially coupled stochastic gradient (C-SG) algorithm is proposed to estimate the model parameters with high computational efficiency compared with the standard stochastic gradient (SG) algorithm. The analysis indicates that the partially C-SG algorithm can give more accurate parameter estimates than the SG algorithm. The parameter estimates obtained using the partially C-SG algorithm converge to their true values as the data length approaches infinity.
IEEE Transactions on Automatic Control 09/2010; · 2.11 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: A multiinnovation least-squares (MILS) identification algorithm is presented for linear regression models with unknown parameter vectors by expanding the innovation length in the traditional recursive least-squares (RLS) algorithm from the viewpoint of innovation modification. Because the proposed MILS algorithm uses p innovations (not only the current innovation but also past innovations) at each iteration (with the integer p > 1 being an innovation length), the accuracy of parameter estimation is improved, compared with that of the RLS algorithm. Performance analysis and simulation results show that the proposed MILS algorithm is consistently convergent. Moreover, a new interval-varying MILS algorithm is proposed, for which the key is to dynamically change the interval in order to deal with cases where some measurement data are missing. Furthermore, an auxiliary-model-based MILS algorithm is derived for pseudolinear models corresponding to output error moving average systems with colored noises. Finally, the proposed algorithms are applied to model an experimental water level control system.
IEEE Transactions on Systems Man and Cybernetics Part B (Cybernetics) 07/2010; · 3.08 Impact Factor
-
IEEE Transactions on Systems, Man, and Cybernetics, Part B. 01/2010; 40:767-778.
-
[show abstract]
[hide abstract]
ABSTRACT: A multiinnovation least-squares (MILS) identification algorithm is presented for linear regression models with unknown parameter vectors by expanding the innovation length in the traditional recursive least-squares (RLS) algorithm from the viewpoint of innovation modification. Because the proposed MILS algorithm uses p innovations (not only the current innovation but also past innovations) at each iteration (with the integer p > 1 being an innovation length), the accuracy of parameter estimation is improved, compared with that of the RLS algorithm. Performance analysis and simulation results show that the proposed MILS algorithm is consistently convergent. Moreover, a new interval-varying MILS algorithm is proposed, for which the key is to dynamically change the interval in order to deal with cases where some measurement data are missing. Furthermore, an auxiliary-model-based MILS algorithm is derived for pseudolinear models corresponding to output error moving average systems with colored noises. Finally, the proposed algorithms are applied to model an experimental water level control system.
IEEE transactions on systems, man, and cybernetics. Part B, Cybernetics: a publication of the IEEE Systems, Man, and Cybernetics Society 10/2009; 40(3):767-78. · 3.01 Impact Factor
-
[show abstract]
[hide abstract]
ABSTRACT: For pseudo-linear regression identification models corresponding output error systems with colored measurement noises, a difficulty of identification is that there exist unknown inner variables and unmeasurable noise terms in the information vector. This paper presents an auxiliary model based multi-innovation stochastic gradient algorithm by using the auxiliary model technique and by expanding the scalar innovation to an innovation vector. Compared with single-innovation stochastic gradient algorithm, the proposed approach can generate highly accurate parameter estimates. The simulation results confirm theoretical findings.
Robotics and Automation, 2009. ICRA '09. IEEE International Conference on; 06/2009
-
2009 IEEE International Conference on Robotics and Automation, ICRA 2009, Kobe, Japan, May 12-17, 2009; 01/2009
-
[show abstract]
[hide abstract]
ABSTRACT: For pseudo-linear regression identification models corresponding output error systems with colored measurement noises, a difficulty of identification is that there exist unknown inner variables and unmeasurable noise terms in the information vector. This paper presents an auxiliary model based multi-innovation extended stochastic gradient algorithm by using the auxiliary model method and by expanding the scalar innovation to an innovation vector. Compared with single innovation extended stochastic gradient algorithm, the proposed approach can generate highly accurate parameter estimates. The simulation results confirm this conclusion.
Signal Processing.
-
[show abstract]
[hide abstract]
ABSTRACT: Gradient based and least-squares based iterative identification algorithms are developed for output error (OE) and output error moving average (OEMA) systems. Compared with recursive approaches, the proposed iterative algorithms use all the measured input–output data at each iterative computation (at each iteration), and thus can produce highly accurate parameter estimation. The basic idea of the iterative methods is to adopt the interactive estimation theory: the parameter estimates relying on unknown variables are computed by using the estimates of these unknown variables which are obtained from the preceding parameter estimates. The simulation results confirm theoretical findings.
Digital Signal Processing.
-
[show abstract]
[hide abstract]
ABSTRACT: This paper considers the identification problems of the Hammerstein nonlinear systems. A projection and a stochastic gradient (SG) identification algorithms are presented for the Hammerstein nonlinear systems by using the gradient search method. Since the projection algorithm is sensitive to noise and the SG algorithm has a slow convergence rate, a Newton recursive and a Newton iterative identification algorithms are derived by using the Newton method (Newton–Raphson method), in order to reduce the sensitivity of the projection algorithm to noise, and to improve convergence rates of the SG algorithm. Furthermore, the performances of these approaches are analyzed and compared using a numerical example, including the parameter estimation errors, the stationarity and convergence rates of parameter estimates and the computational efficiency.
Digital Signal Processing.
