Abstract:The identification of electrical, mechanical, and biological systems using data can benefit greatly from prior knowledge extracted from physical modeling. Parametric continuous-time identification methods can naturally incorporate this knowledge, which leads to interpretable and parsimonious models. However, some applications lead to model structures that lack parsimonious descriptions using unfactored transfer functions, which are commonly used in standard direct approaches for continuous-time system identification. In this paper we characterize this parsimony problem, and develop a block-coordinate descent algorithm that delivers parsimonious models by sequentially estimating an additive decomposition of the transfer function of interest. Numerical simulations show the efficacy of the proposed approach.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively obtain a parsimonious model description in continuous - time system identification. Specifically, when dealing with electrical, mechanical or biological systems, system identification using data can greatly benefit from prior knowledge extracted from physical modeling. However, for some applications, using an undecomposed transfer function to describe the model structure lacks parsimony. The paper points out that traditional direct methods in continuous - time system identification usually use undecomposed transfer functions, which may lead to too many model parameters and lack of parsimony. To overcome this problem, the author proposes a method based on the block - coordinate descent algorithm to generate a parsimonious model by sequentially estimating the additive decomposition of the transfer function of interest. This method is particularly suitable for systems that can be more easily interpreted as the sum of transfer functions with different denominators, which usually correspond to different resonance modes. Numerical simulations show the effectiveness of the proposed method. The main contributions of the paper include: 1. **Deterministic Conditions**: Provide an explicit condition for determining when an additive model should be estimated instead of an undecomposed transfer function, which is very important for deciding the model structure to be considered in continuous - time system identification. 2. **Global Convergence**: Prove the global convergence of the block - coordinate descent algorithm for additive continuous - time systems when the sample size is large enough. 3. **Closed - Form Expressions**: Provide closed - form expressions for SRIVC iterations, and prove that these iterations can reach the critical points of the cost function in each block - coordinate descent step under a finite sample size and mild conditions. Through these contributions, the paper aims to improve the efficiency and accuracy of continuous - time system identification, especially when dealing with high - order or highly resonant systems.

Parsimonious Identification of Continuous-Time Systems: A Block-Coordinate Descent Approach

Identification of Additive Continuous-time Systems in Open and Closed loop

Identification of nonlinearly parameterized nonlinear models: application to mass balance systems.

Identifiability and Identification of Switching Dynamical Networks: A Data-Based Approach

Statistical Analysis of Block Coordinate Descent Algorithms for Linear Continuous-time System Identification

Rapid Identification of Switched Systems: A Data-Driven Method in Variational Framework

A new closed-loop identification technique for step-response model based on subspace approach

An asymptotically optimal indirect approach to continuous-time system identification

Identification of block-oriented nonlinear systems starting from linear approximations: A survey

Combining Off-White and Sparse Black Models in Multi-step Physics-based Systems Identification -- EXTENDED VERSION

Parametric identification of parallel Wiener-Hammerstein systems

Identifying Latent Dynamic Components in Biological Systems

Beyond Occam's Razor in System Identification: Double-Descent when Modeling Dynamics

Model Reduction for Control System Design

Structural Identifiability of Dynamic Systems Biology Models

Modulating Function-Based Subspace Identification for Continuous-Time Fractional Order Systems

Decentralized Identification for Multivariable Integrating Processes with Time Delays from Closed-Loop Step Tests.

State estimation-based parameter identification for a class of nonlinear fractional-order systems

Parsimonious System Identification from Fragmented Quantized Measurements

Incremental Design of Simplex Basis Function Model for Dynamic System Identification

Iterative Identification of Block-Oriented Nonlinear Systems Based on Biconvex Optimization.