ZETA: a library for Zonotope-based EsTimation and fAult diagnosis of discrete-time systems
Brenner S. Rego, Joseph K. Scott, Davide M. Raimondo, Marco H. Terra, Guilherme V. Raffo · 2025
Read ZETA: a library for Zonotope-based EsTimation and fAult diagnosis of discrete-time systems by Brenner S. Rego, Joseph K. Scott, Davide M. Raimondo, Marco H. Terra, Guilherme V. Raffo (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-06467.txt.
Introduction
Brenner Rego, Joseph Scott, Davide Raimondo, Marco Terra, and Guilherme Raffo release ZETA, a software library for tracking systems and catching their faults, in a 2025 paper. Engineers who estimate a machine's hidden state must handle uncertainty honestly. Zonotopes — tidy geometric sets that wrap the possible states — do this rigorously, but tooling was scattered.
The MATLAB library implements zonotopes in three flavors plus interval arithmetic, with the operations each needs. It propagates sets through nonlinear behavior, shrinks set size without losing safety, and supports reachability checks, state estimation, and active fault diagnosis for linear, nonlinear, and descriptor systems. Worked examples show the routines in action. A shared toolkit spreads set-based methods beyond their home labs.
Catalog record
| CREATOR | Brenner S. Rego, Joseph K. Scott, Davide M. Raimondo, Marco H. Terra, Guilherme V. Raffo |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 39 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-06467
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
ZETA: a library for Zonotope-based EsTimation and fAult diagnosis of discrete-time systems Brenner S. Rego, Joseph K. Scott, Davide M. Raimondo, Marco H. Terra, Guilherme V. Raffo Abstract This paper introduces ZETA, a new MATLAB library for Zonotope-based EsTimation and fAult diagnosis of discrete-time systems. It features user-friendly implementations of set representations based on zonotopes, namely zonotopes, constrained zonotopes, and line zonotopes, in addition to a basic implementation of interval arithmetic. This library has capabilities starting from the basic set operations with these sets, including propagations through nonlinear functions using various approximation methods. The features of ZETA allow for reachability analysis and state estimation of discrete-time linear, nonlinear, and descriptor systems, in addition to active fault diagnosis of linear systems. Efficient order reduction methods are also implemented for the respective set representations. Some examples are presented in order to illustrate the functionalities of the new library. I Introduction In the last decades, many areas of scientific investigation have started to use set-based algorithms for different goals. From reachability analysis of dynamic systems [ 1 , 2 , 3 , 4 ] , to state [ 5 , 6 ] and parameter estimation [ 7 ] , fault diagnosis [ 8 , 9 ] , invariant [ 10 ] and controllable sets [ 11 ] , and model predictive control [ 12 , 13 ] , set-based operations have gained attention because they allow to generate guaranteed enclosures for dynamic systems subject to bounded uncertainties. Thanks to their advantages for important set operations such as the Minkoswki sum and linear image, zonotopes have been used to obtain tight enclosures for the trajectories of discrete-time linear systems [ 14 , 15 ] . Such enclosures have been successfully used in accurate state estimation and active fault diagnosis [ 16 ] . Constrained zonotopes (CZs), an extension of zonotopes proposed in [ 17 ] , allowed further improvements in these topics [ 8 ] , mainly due to their capability of representing arbitrary convex polytopes. CZs retain key computational advantages of zonotopes, including efficient complexity reduction algorithms. Constrained zonotopes were instrumental in achieving efficient state estimation and active fault diagnosis of discrete-time linear descriptor systems [ 18 ] , whose trajectories have both a dynamic and static nature. To overcome the limitations of CZs in representing unbounded sets, [ 19 ] introduced line zonotopes (LZs), an extension of the original framework that retains support for effective order reduction methods. Zonotopic sets have been especially important for the development of computationally affordable reachability analysis and state estimation algorithms for discrete-time nonlinear systems [ 20 ] . Zonotope methods have been proposed since the 90’s starting with a Mean Value Theorem approach [ 21 ] . This has been extended in [ 5 ] , and a Taylor’s Theorem approach has been proposed in [ 22 ] . These two algorithms have been improved by using constrained zonotopes in [ 23 ] , and further extended for nonlinear measurement equations and invariants in [ 24 ] . Difference of convex (DC) programming principles have been used in [ 25 ] and [ 26 ] for state estimation using zonotopes and CZs, respectively. Finally, polyhedral relaxation techniques have been used in [ 27 ] and [ 28 ] to obtain improved CZ enclosures for discrete-time nonlinear systems. The main objective of this paper is to introduce ZETA, a new MATLAB library for Zonotope-based EsTimation and fAult diagnosis of discrete-time systems 1 1 1 See https://github.com/Guiraffo/ZETA-releases . . Thanks to their computational advantages, zonotopic sets are very important for the development of set-based algorithms. However, the underlying concepts for such algorithms can be difficult to tackle by the general audience, while robust and reliable implementations are not straightforward. A few libraries are available in different languages [ 29 , 30 , 31 ] , providing algorithms for basic operations and reachability analysis using zonotopic sets. However, robust implementations of several of the mentioned methods are still not available as off-the-shelf algorithms, especially the polyhedral relaxation techniques proposed in [ 27 , 28 ] , active fault diagnosis methodologies [ 16 ] , and line zonotopes [ 19 ] . The aim of ZETA is to fulfill this relevant gap, including (besides various zonotopic set representations and their basic operations): (i) efficient implementations of CZ complexity reduction algorithms, (ii) state estimation and fault diagnosis of discrete-time linear systems, (iii) several approximation methods for enclosing the trajectories of discrete-time nonlinear systems, (iv) nonlinear state estimation, and (v) active fault diagnosis methods. Notation Lowercase italic letters denote scalars, lowercase bold letters denote vectors, uppercase bold letters denote matrices, and uppercase italic letters denote general sets. Moreover, \bm{0}_{n\times m} and \bm{1}_{n\times m} denote n\times m matrices of zeros and ones, respectively. The n\times n identity matrix is denoted by {\mathbf{I}}_{n} . The set of real numbers is denoted by \mathbb{R} . Additionally, let \bm{\alpha}:\mathbb{R}^{n_{s}}\to\mathbb{R}^{n_{\alpha}} be a nonlinear function with argument {\mathbf{s}} . Functions with set-valued arguments S satisfying {\mathbf{s}}\in S denote the exact image of the set under the function, i.e., \bm{\alpha}(S)\triangleq\{\bm{\alpha}({\mathbf{s}}):{\mathbf{s}}\in S\} . II Mathematical background Consider a real matrix {\mathbf{R}}\in\mathbb{R}^{m\times n} and sets Z,W\subset\mathbb{R}^{n} , Y\subset\mathbb{R}^{m} . The linear mapping, Minkowski sum, generalized intersection, and the Cartesian product are defined as \displaystyle{\mathbf{R}}Z \displaystyle\triangleq\{{\mathbf{R}}{\mathbf{z}}:{\mathbf{z}}\in Z\}, (1) \displaystyle Z\oplus W \displaystyle\triangleq\{{\mathbf{z}}+{\mathbf{w}}:{\mathbf{z}}\in Z,\,{\mathbf{w}}\in W\}, (2) \displaystyle Z\cap_{{\mathbf{R}}}Y \displaystyle\triangleq\{{\mathbf{z}}\in Z:{\mathbf{R}}{\mathbf{z}}\in Y\}, (3) \displaystyle Z\times Y \displaystyle\triangleq\{({\mathbf{z}},{\mathbf{y}}):{\mathbf{z}}\in Z,\,{\mathbf{y}}\in Y\}, (4) respectively. Moreover, \text{conv}(Z,W) denotes the convex hull of Z and W . To accomplish its various capabilities, ZETA implements a few different set representations: intervals, strips, zonotopes, constrained zonotopes, line zonotopes, and convex polytopes in halfspace representation. These are defined below. Definition 1 Let \mathbb{I}\mathbb{R}^{n} denote the set of all non-empty compact intervals in \mathbb{R}^{n} . If {{\mathbf{x}}}^{\text{L}},{{\mathbf{x}}}^{\text{U}}\in\mathbb{R}^{n} with {{\mathbf{x}}}^{\text{L}}\leq{{\mathbf{x}}}^{\text{U}} , an interval X\in\mathbb{I}\mathbb{R}^{n} is defined as X\triangleq\{{\mathbf{x}}\in\mathbb{R}^{n}:{{\mathbf{x}}}^{\text{L}}\leq{\mathbf{x}}\leq{{\mathbf{x}}}^{\text{U}}\}\triangleq[{{\mathbf{x}}}^{\text{L}},{{\mathbf{x}}}^{\text{U}}] . Definition 2 A set S\subset\mathbb{R}^{n_{s}} is a strip if there exists a tuple ({\mathbf{p}}_{s},d_{s},\sigma_{s})\in\mathbb{R}^{n_{s}}\times\mathbb{R}\times\mathbb{R} such that S=\{{\mathbf{s}}\in\mathbb{R}^{n_{s}}:|{\mathbf{p}}_{s}^{T}{\mathbf{s}}-d_{s}|\leq\sigma_{s}\} . We use the shorthand notation ({\mathbf{p}}_{s},d_{s},\sigma_{s})_{\text{S}} for strips. Definition 3 A set Z\subset\mathbb{R}^{n_{z}} is a zonotope with n_{g} generators if there exists a tuple ({\mathbf{G}}_{z},{\mathbf{c}}_{z})\in\mathbb{R}^{n_{z}\times n_{g}}\times\mathbb{R}^{n} such that Z=\left\{{\mathbf{c}}_{z}+{\mathbf{G}}_{z}\bm{\xi}:\|\bm{\xi}\|_{\infty}\leq 1\right\} . We use the shorthand notation ({\mathbf{G}}_{z},{\mathbf{c}}_{z})_{\text{Z}} for zonotopes. Definition 4 [ 17 ] A set Z\subset\mathbb{R}^{n_{z}} is a constrained zonotope with n_{g} generators and n_{c} constraints if there exists a tuple ({\mathbf{G}}_{z},{\mathbf{c}}_{z},{\mathbf{A}}_{z},{\mathbf{b}}_{z})\in\mathbb{R}^{n\times n_{g}}\times\mathbb{R}^{n}\times\mathbb{R}^{n_{c}\times n_{g}}\times\mathbb{R}^{n_{c}} such that Z=\left\{{\mathbf{c}}_{z}+{\mathbf{G}}_{z}\bm{\xi}:\|\bm{\xi}\|_{\infty}\leq 1,{\mathbf{A}}_{z}\bm{\xi}={\mathbf{b}}_{z}\right\} . We use the shorthand notation ({\mathbf{G}}_{z},{\mathbf{c}}_{z},{\mathbf{A}}_{z},{\mathbf{b}}_{z})_{\text{CZ}} for constrained zonotopes. Definition 5 A set Z\subseteq\mathbb{R}^{n_{z}} is a line zonotope with n_{\ell} lines, n_{g} generators, and n_{c} constraints, if there exists a tuple ({\mathbf{M}}_{z},{\mathbf{G}}_{z},{\mathbf{c}}_{z},{\mathbf{S}}_{z},{\mathbf{A}}_{z},{\mathbf{b}}_{z})\in\mathbb{R}^{n_{z}\times n_{\ell}}\times\mathbb{R}^{n_{z}\times n_{g}}\times\mathbb{R}^{n}\times\mathbb{R}^{n_{c}\times n_{\ell}}\times\mathbb{R}^{n_{c}\times n_{g}}\times\mathbb{R}^{n_{c}} such that Z=\{{\mathbf{M}}_{z}\bm{\delta}+{\mathbf{G}}_{z}\bm{\xi}+{\mathbf{c}}_{z}:\bm{\delta}\in\mathbb{R}^{n_{\ell}},\|\bm{\xi}\|_{\infty}\leq 1,{\mathbf{S}}_{z}\bm{\delta}+{\mathbf{A}}_{z}\bm{\xi}={\mathbf{b}}_{z}\}. We use the shorthand notations ({\mathbf{M}}_{z},{\mathbf{G}}_{z},{\mathbf{c}}_{z},{\mathbf{S}}_{z},{\mathbf{A}}_{z},{\mathbf{b}}_{z})_{\text{LZ}} and ({\mathbf{M}}_{z},{\mathbf{G}}_{z},{\mathbf{c}}_{z})_{\text{LZ}} for line zonotopes with and without constraints, respectively. Definition 6 A set P\subset\mathbb{R}^{n} is a convex polytope in halfspace representation if there exists ({\mathbf{H}}_{p},{\mathbf{k}}_{p},{\mathbf{A}}_{p},{\mathbf{b}}_{p})\in\mathbb{R}^{n_{h}\times n}\times\mathbb{R}^{n_{h}}\times\mathbb{R}^{n_{c_{p}}\times n}\times\mathbb{R}^{n_{c_{p}}} such that P=\{{\mathbf{x}}\in\mathbb{R}^{n}:{\mathbf{H}}_{p}{\mathbf{x}}\leq{\mathbf{k}}_{p},\ {\mathbf{A}}_{p}{\mathbf{x}}={\mathbf{b}}_{p}\} . We use the shorthand notation ({\mathbf{H}}_{p},{\mathbf{k}}_{p},{\mathbf{A}}_{p},{\mathbf{b}}_{p})_{\text{P}} for convex polytopes. III Core features of ZETA The ZETA toolbox makes extensive use of Object Oriented Programming (OOP) and operator overloading to implement set computations. Auxiliary methods necessary for set-based state estimation and active fault diagnosis are also present. This section describes the core features of the library. III-A Structure, dependencies, and external libraries ZETA files are organized according to the