Abstract:In order to identify one system (module) in an interconnected dynamic network, one typically has to solve a Multi-Input-Single-Output (MISO) identification problem that requires identification of all modules in the MISO setup. For application of a parametric identification method this would require estimating a large number of parameters, as well as an appropriate model order selection step for a possibly large scale MISO problem, thereby increasing the computational complexity of the identification algorithm to levels that are beyond feasibility. An alternative identification approach is presented employing regularized kernel-based methods. Keeping a parametric model for the module of interest, we model the impulse response of the remaining modules in the MISO structure as zero mean Gaussian processes (GP) with a covariance matrix (kernel) given by the first-order stable spline kernel, accounting for the noise model affecting the output of the target module and also for possible instability of systems in the MISO setup. Using an Empirical Bayes (EB) approach the target module parameters are estimated through an Expectation-Maximization (EM) algorithm with a substantially reduced computational complexity, while avoiding extensive model structure selection. Numerical simulations illustrate the potentials of the introduced method in comparison with the state-of-the-art techniques for local module identification.

What problem does this paper attempt to address?

This paper aims to solve the problem of local module identification in dynamic networks. Specifically, the paper focuses on how to effectively identify specific modules (systems) given the known network topology without performing full - network identification on the entire network. Traditional direct methods, when dealing with multi - input single - output (MISO) structures, need to estimate a large number of parameters and select an appropriate model order, which increases the computational complexity of the algorithm and makes it infeasible. Moreover, if some modules in the MISO structure are unstable, the existing prediction error identification methods cannot be used either, because these methods assume that the predictor is stable. To overcome these problems, the paper proposes a new method based on the regularized kernel method. The main contributions of this method include: 1. **Identifying local modules through direct methods**: Utilize the information of external excitation signals and perturbation signals to enhance the usefulness of data. 2. **Avoiding model order selection for large - scale problems**: Reduce the number of parameters, especially those that are not related to the main interests of the experimenter. 3. **Not depending on module stability**: This method can be used even if some modules in the MISO structure are unstable, without prior knowledge of which modules are unstable. The paper achieves the above goals by modeling the impulse responses of the remaining modules as zero - mean Gaussian processes (GP) and using a first - order stable spline kernel to encode the stability and smoothness of the process. In addition, the paper also introduces the empirical Bayes method to perform parameter estimation by maximizing the marginal likelihood of data and uses the expectation - maximization (EM) algorithm to iteratively solve the nonlinear optimization problem. In conclusion, this paper proposes a new method that can effectively identify local modules in dynamic networks, while reducing computational complexity and being able to handle the situation of unstable modules.

Learning linear modules in a dynamic network using regularized kernel-based methods

A frequency domain approach for local module identification in dynamic networks

Closed-Loop Subspace Identification Based on Kpls

Selective Recursive Kernel Learning for Online Identification of Nonlinear Systems with NARX Form

A Spline Kernel-Based Approach for Nonlinear System Identification with Dimensionality Reduction

On the estimation of initial conditions in kernel-based system identification

Adaptive regularised kernel-based identification method for large-scale systems with unknown order

Applications of Landweber Interation Regularization Method in Dynamic Load Identification

Adaptive Kernel Leaning Networks with Application to Nonlinear System Identification

System identification under non-negativity constraints

System Model Identification Based on Kernel Function

On Kernel Design for Regularized Non-Causal System Identification

Kernel-Based Learning of Stable Nonlinear Systems

Parameter Identification for Nonlinear State-Space Models of A Biological Network Via Linearization and Robust State Estimation

Robust On-Line Nonlinear Systems Identification Using Multilayer Dynamic Neural Networks with Two-Time Scales

Identification methods of nonlinear systems based on the kernel functions

Nonlinear system identification in coherence with nonlinearity measure for dynamic physical systems—case studies

Novel Nonlinear System Identification Method Based on Kernel Function

A scalable multi-step least squares method for network identification with unknown disturbance topology

SID-Net: machine learning based system identification framework for rigid and flexible multibody dynamics

Robust Identification for Unknown Nonlinear Multivariable Systems Based on Dynamic Neural Networks