Abstract:Summary
Identifying a nonlinear radial basis function‐based state‐dependent autoregressive (RBF‐AR) time series model is the basis for solving the corresponding prediction and control problems. This paper studies some recursive parameter estimation algorithms for the RBF‐AR model. Considering the difficulty of the nonlinear optimal problem arising in estimating the RBF‐AR model, an overall forgetting gradient algorithm is deduced based on the negative gradient search. A numerical method with a forgetting facto… Show more
“…For example, in Reference 28, Li et al constructed a kernel function‐based nonlinear system, which had a good performance in fitting nonlinear systems. In Reference 29, Zhou et al studied some recursive parameter estimation algorithms for a nonlinear radial basis function‐based state‐dependent autoregressive (RBF‐AR) time series model. In Reference 30, Brenner et al proposed a regularization method for solving the ill‐posed problem.…”
This article proposes a kernel-based regularization least squares algorithm for nonlinear time-delayed systems with unknown structure. Since the structure and the time-delay of the model are unknown, a model pool constituted of several Volterra series is constructed with the aim of approximating the nonlinear model which has different time-delays. Then, a self-organizing maps method and a kernel-based regularization method are interactively used to update the time-delay and parameters. Compared with the traditional algorithms, the proposed algorithm has no limitation on the nonlinear model, and can describe the dynamics of the nonlinear model with a simple structure. Finally, simulation examples are given to show the effectiveness of the algorithm.
“…For example, in Reference 28, Li et al constructed a kernel function‐based nonlinear system, which had a good performance in fitting nonlinear systems. In Reference 29, Zhou et al studied some recursive parameter estimation algorithms for a nonlinear radial basis function‐based state‐dependent autoregressive (RBF‐AR) time series model. In Reference 30, Brenner et al proposed a regularization method for solving the ill‐posed problem.…”
This article proposes a kernel-based regularization least squares algorithm for nonlinear time-delayed systems with unknown structure. Since the structure and the time-delay of the model are unknown, a model pool constituted of several Volterra series is constructed with the aim of approximating the nonlinear model which has different time-delays. Then, a self-organizing maps method and a kernel-based regularization method are interactively used to update the time-delay and parameters. Compared with the traditional algorithms, the proposed algorithm has no limitation on the nonlinear model, and can describe the dynamics of the nonlinear model with a simple structure. Finally, simulation examples are given to show the effectiveness of the algorithm.
“…In terms of the identification of RBF-AR models, several variable projection (VP) algorithms were proposed based on the orthogonal projection [17,18,19]. Because the VP algorithms are off-line methods, in order to realize the on-line identification of this kind of models, a multi-innovation recursive algorithm was developed by employing the innovation modification [20]. However, the parameter estimation performance of the multi-innovation recursive algorithm has a lot to do with the selection of the innovation length, and it often takes many tests to get an appropriate innovation length.…”
“…[20][21][22] In order to continue to update the parameter estimates of the RBF-AR models when new sample data come, some on-line gradient-based methods have also been developed by applying the recursive technique. 23 In engineering, the input and output signals always involve the noise interferences from the process environments. 24,25 There are two forms of external noises, the white noise and the colored noise.…”
Section: Introductionmentioning
confidence: 99%
“…Regarding for the identification of the RBF‐AR models, some off‐line methods, including the structured nonlinear parameter optimization methods and the variable projection methods, have been proposed 20‐22 . In order to continue to update the parameter estimates of the RBF‐AR models when new sample data come, some on‐line gradient‐based methods have also been developed by applying the recursive technique 23 …”
This article studies the parameter estimation problems of radial basis function-based state-dependent autoregressive models with autoregressive noises (RBF-ARAR models). To reduce the effect of the colored noise to parameter estimation, the data filtering technique is applied and a filtering based generalized stochastic gradient algorithm is derived for the RBF-ARAR models. In order to achieve more accurate parameter estimates, a filtering based multiinnovation generalized stochastic gradient (F-MI-GSG) algorithm is proposed by utilizing the current and past innovations. Introducing two forgetting factors, a filtering based multiinnovation generalized forgetting gradient algorithm is developed to improve the transient performance of the F-MI-GSG algorithm. The effectiveness of the proposed algorithms is verified through the simulation examples.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.