Skip to main content

Introduction

The Dynamic Network Identification Toolbox is developed by a team of researchers and developers under leadership of:

Prof. Paul M.J. Van den Hof
Control Systems Group
Department of Electrical Engineering
Eindhoven University of Technology
Eindhoven, The Netherlands

E-mail: p.m.j.vandenhof@tue.nl
Project website: www.sysdynet.eu

Version:    beta-0.4.0
Date:       30 October 2025

Contributions to the algorithms:
    Xiaodong Cheng, Stefanie Fonken, Karthik Ramaswamy, Shengling Shi, 
    Tom Steentjes, Harm Weerts, Lizan Kivits, Venkat Rajagopal, 
    Desen Liang, Yibo Shi
Contributions to the software implementation:
    John Baek, Mannes Dreef, Wim Liebregts, Ilja van Oort, 
    Gareth Thomas (VersionBay)

This project has received funding from the European Research Council (ERC), Advanced Research Grant SYSDYNET, under the European Union’s Horizon 2020 research and innovation programme (Grant Agreement No. 694504), and the Proof of Concept Grant SYSDYNETTOOL (2024-2025).

Copyright by Paul Van den Hof. Copyright of the software on the Local Polynomial Method for non-parametric frequency domain identification is held by Rik Pintelon.
The software is made available under a General Public License GNU 3.0.

Functionality and limitations​

The beta-version of the app and toolbox contains a graphical user interface for analyzing structural properties of linear dynamic networks, and for data-driven modeling (identification) of network dynamics and network topology. Tools are provided for the allocation of sensors (where to measure) and external signals (where to excite), for arriving at consistent estimates of either a full network, a subnetwork or a single module. Two types of dynamic networks are facilitated:

  • directed (module) networks, and
  • undirect (diffusively coupled) networks.

The prime app-windows concern:

  • Loading, viewing and editing a network topology
  • Basic tests and operations on network topologies; node immersion
  • Analyzing and synthesizing module/network identifiability
  • Constructing predictor models (inputs and outputs) for local module identification
  • Identification algorithms for full network or single module estimation
  • Topology estimation
  • Residual-based diagnostics (under construction)

Additionally, tools are provided for network simulation, and for loading (directed) network structures from Simulink.

The different technical operations are also available as individual m-files for command window operations. The command line functions are available only when the App is open.

The maximum network size that can be handled by the current implementation is around 50 nodes.

The implemented identification algorithms are to be considered as first versions. Implementation aspects are still subject to refinement and optimization.

Required Resources​

  • The Control Systems Toolbox is required to run the App.
  • CVX (http://cvxr.com/cvx/download/) is required to run the SLS algorithm (Identification of directed networks).
  • The System Identification Toolbox is required to run the Direct PEM and the Multistep algorithm for directed network identification.

Future extensions​

In the near future the app/toolbox will be extended with more identification algorithms, model evaluation and model-based diagnostics tools.

Questions and feedback​

For any questions, feedback and bug reports, please contact:
p.m.j.vandenhof@tue.nl.