A low-complexity funnel control approach for non-linear systems of higher-order
Dario Dennstädt · 2025
Read A low-complexity funnel control approach for non-linear systems of higher-order by Dario Dennstädt (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-2503-11370.txt.
Introduction
Dario Dennstadt, a solo author, simplifies control design in a 2025 paper. Many machines are run by rules that must track a target output while the inner behavior stays only partly known. Funnel control handles such systems by forcing the tracking error to stay inside a narrowing funnel of allowed error, without needing a full model.
Older funnel methods grow complex for systems of higher order, mixing nonlinear tricks into inner signals. The new design drops those extras and stays lean. Proofs show the scheme works for a broad family of systems with stable inner dynamics. Simulations compare it with earlier methods and show the simpler controller holding its funnel.
Catalog record
| CREATOR | Dario Dennstädt |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-11370
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A low-complexity funnel control approach for non-linear systems of higher-order Dario Dennstädt Abstract We address the problem of output reference tracking for unknown non-linear multi-input, multi-output systems described by functional differential equations. This class of systems includes those with a strict relative degree, and bounded-input bounded-output (BIBO) stable internal dynamics. The objective is to ensure that the tracking error evolves within a prescribed performance funnel. To achieve this, we propose a novel model-free adaptive controller with lower complexity than existing funnel control methods, avoiding the use of non-linearities in intermediate error signals. We establish the feasibility and effectiveness of the proposed design through theoretical analysis and demonstrate its performance with simulations, comparing it to previous approaches. This work was supported by the German Research Foundation - Deutsche Forschungsgesellschaft (DFG; project number 471539468) keywords non-linear systems, adaptive control, output reference tracking, prescribed performance, funnel control, functional differential equation 1 Introduction Funnel control is a model-free, high-gain output feedback technique first introduced in Ilchmann et al. (2002) . It ensures output tracking of reference signals within prescribed performance bounds for a broad class of non-linear multi-input, multi-output systems. The approach solely invokes certain structural assumptions about the system, namely stable internal dynamics and a known relative degree with a sign-definite high-frequency matrix. Under these conditions, the adaptive controller offers robustness against disturbances and guarantees specified transient behaviour without relying on explicit knowledge about system to be controlled. Funnel control therefore proved useful for tracking problems in various applications such as DC-link power flow control, see Senfelds and Paugurs (2014) , control of industrial servo-systems, see Hackl (2017) , and temperature control of chemical reactor models, see Ilchmann and Trenn (2004) . Recent research on funnel control has focused on addressing practical challenges that might arise in real-world applications. These include issues like measurement losses, see Berger and Lanza (2023) , input constraints, see Berger (2024) , reliance on sampled measurement data, see Lanza et al. (2024) , and leveraging existing model knowledge, see Berger et al. (2022) . For a comprehensive overview of the literature, we also recommend the survey paper Berger et al. (2023) . Despite these advancements, a persistent challenge for funnel control and, more general, for high-gain adaptive control has been the handling of systems with relative degree larger than one, for the latter see e.g. Morse (1996) . To address the problem of output tracking within prescribed boundaries for higher-order systems, Berger et al. (2018) proposed a control strategy utilising auxiliary error variables, with the number of variables corresponding to the system’s relative degree. Each auxiliary error signal is recursively defined and includes a reciprocal penalty term that depends non-linearly on the preceding variable. These signals evolve within their respective funnel boundaries, maintaining the desired tracking performance. However, due to their reciprocal nature, error gains can become very large when an auxiliary variable nears its boundary, posing potential numerical stability issues. Although the controller design was further simplified in Berger et al. (2021) , the general approach remains similar, and the controllers lack the low complexity of the original design for systems of order one. Prescribed performance control is a relative to funnel control and was first proposed in Bechlioulis and Rovithakis (2008) . The tracking within prescribed boundaries is achieved by transforming the system via so-called performance functions. While the original controller design incorporates neural networks to approximate unknown non-linearities, Bechlioulis and Rovithakis (2014) proposed an approximation-free control scheme for systems in pure feedback form. Similar to funnel control, recursively defined intermediate control signals are used, matching the number of the system’s relative degree and depending non-linearly on their predecessors. In this paper, we propose a funnel controller of lower complexity for systems of higher-order. While still relying on recursively defined auxiliary error variables to handle the system’s higher-order nature, the proposed approach eliminates the use of time-varying reciprocal penalty terms, replacing them with constant gains. This modification simplifies the controller design and has the potential to mitigate numerical issues and enhance its practicality for real-world applications. 1.0.1 Nomenclature: \mathds{N} and \mathds{R} denote natural and real numbers, respectively. \mathds{N}_{0}:=\mathds{N}\cup\{0\} and \mathds{R}_{\geq 0}:=[0,\infty) . \left\|x\right\|:=\sqrt{\left\langle x,x\right\rangle} denotes the Euclidean norm of x\in\mathds{R}^{n} . \mathcal{C}^{p}(V,\mathds{R}^{n}) is the linear space of p -times continuously differentiable functions f:V\to\mathds{R}^{n} , where V\subset\mathds{R}^{m} , and p\in\mathds{N}_{0}\cup\{\infty\} . \mathcal{C}(V,\mathds{R}^{n}):=\mathcal{C}^{0}(V,\mathds{R}^{n}) . On an interval I\subset\mathds{R} , L^{\infty}(I,\mathds{R}^{n}) denotes the space of measurable and essentially bounded functions f:I\to\mathds{R}^{n} with norm \left\|f\right\|_{\infty}:=\esssup_{t\in I}\left\|f(t)\right\| , L^{\infty}_{\text{loc}}(I,\mathds{R}^{n}) the set of measurable and locally essentially bounded functions. Furthermore, W^{k,\infty}(I,\mathds{R}^{n}) is the Sobolev space of all k -times weakly differentiable functions f:I\to\mathds{R}^{n} such that f,\dots,f^{(k)}\in L^{\infty}(I,\mathds{R}^{n}) . 2 Problem formulation We consider non-linear multi-input multi-output systems with constant degree r\in\mathds{N} of the form \displaystyle y^{(r)}(t) \displaystyle=f\big(d(t),\mathbf{T}(y,\ldots,y^{(r-1)})(t),u(t)\big) (1) \displaystyle y|_{[0,t_{0}]} \displaystyle=y^{0}\in\mathcal{C}^{r-1}([0,t_{0}],\mathds{R}^{m}) with t_{0}\geq 0 , initial trajectory y^{0} , control input u\in L^{\infty}_{\loc}([t_{0},\infty),\mathds{R}^{m}) , and output y(t)\in\mathds{R}^{m} at time t\geq 0 . In the case of t_{0}=0 , we identify \mathcal{C}^{r-1}([0,t_{0}],\mathds{R}^{m}) with the vector space \mathds{R}^{rm} . Then, the initial value condition in ( 1 ) is replaced by (y(t_{0}),\dot{y}(t_{0}),\ldots,y^{(r-1)}(t_{0}))=y^{0}\in\mathds{R}^{rm} . Note that u and y have the same dimension m\in\mathds{N} . The system consists of the unknown non-linear function f\in\mathcal{C}(\mathds{R}^{p}\times\mathds{R}^{q}\times\mathds{R}^{m},\mathds{R}^{m}) , unknown non-linear operator \mathbf{T}:\mathcal{C}(\mathds{R}_{\geq 0},\mathds{R}^{m})\to L^{\infty}_{\loc}([t_{0},\infty),\mathds{R}^{q}) , and unknown bounded disturbances d\in L^{\infty}([t_{0},\infty),\mathds{R}^{p}) . The operator \mathbf{T} is causal, locally Lipschitz and satisfies a bounded-input bounded-output property. It is characterised in detail in the following definition. Definition 2.1 For n,q\in\mathds{N} , and t_{0}\geq 0 , the set \mathcal{T}^{n,q}_{t_{0}} denotes the class of operators \mathbf{T}:\mathcal{C}(\mathds{R}_{\geq 0},\mathds{R}^{n})\to L^{\infty}_{\loc}([t_{0},\infty),\mathds{R}^{q}) for which the following properties hold: • Causality : \ \forall\,y_{1},y_{2}\in\mathcal{C}(\mathds{R}_{\geq 0},\mathds{R}^{n}) \ \forall\,t\geq t_{0} : y_{1}|_{[0,t]}=y_{2}|_{[0,t]}\quad\Longrightarrow\quad\mathbf{T}(y_{1})|_{[t_{0},t]}=\mathbf{T}(y_{2})|_{[t_{0},t]}. • Local Lipschitz : \ \forall\,t\geq t_{0} \ \forall\,y\in\mathcal{C}([0,t],\mathds{R}^{n}) \ \exists\,\Delta,\delta,c>0 \ \forall\,y_{1},y_{2}\in\mathcal{C}(\mathds{R}_{\geq 0},\mathds{R}^{n}) with y_{1}|_{[0,t]}=y_{2}|_{[0,t]}=y and \left\|y_{1}(s)-y(t)\right\|<\delta , \left\|y_{2}(s)-y(t)\right\|<\delta for all s\in[t,t+\Delta] : \hskip-5.69054pt\esssup_{\mathclap{s\in[t,t+\Delta]}}\left\|\mathbf{T}(y_{1})(s)\!-\!\mathbf{T}(y_{2})(s)\right\|\!\leq\!c\ \sup_{\mathclap{s\in[t,t+\Delta]}}\ \left\|y_{1}(s)\!-\!y_{2}(s)\right\|\!. • Bounded-input bounded-output (BIBO) : \ \forall\,c_{0}>0 \ \exists\,c_{1}>0 \ \forall\,y\in\mathcal{C}(\mathds{R}_{\geq 0},\mathds{R}^{n}) : \sup_{t\in\mathds{R}_{\geq 0}}\left\|y(t)\right\|\leq c_{0}\ \Longrightarrow\ \sup_{t\in[t_{0},\infty)}\left\|\textbf{T}(y)(t)\right\|\leq c_{1}. Note that using this operator many physical phenomena such as backlash , and relay hysteresis , and non-linear time delays can be modelled by \mathbf{T} where t_{0} corresponds to the initial delay, cf. ( Berger et al., 2021 , Sec. 1.2) . In addition, systems with infinite-dimensional internal dynamics can be represented by ( 1 ), see Berger et al. (2020) . While unknown, we assume the function f in the system ( 1 ) to have the so-called high-gain property ( Berger et al., 2021 , Definition 1.2) . It is an essential assumption in high-gain adaptive control which, loosely speaking, ensures that, if a large enough control input is applied, then the system reacts sufficiently fast. We summarise our assumptions and define the general system class under consideration which is only parametrised by two natural numbers m and r . Definition 2.2 We say that the system ( 1 ) belongs to the system class \mathcal{N}^{m,r} , written (d,f,\mathbf{T})\in\mathcal{N}^{m,r} , if, for t_{0}\geq 0 and some q\in\mathds{N} , the following holds: d\in L^{\infty}([t_{0},\infty),\mathds{R}^{p}) , {\mathbf{T}\in\mathcal{T}^{rm,q}_{t_{0}}} , and f\in\mathcal{C}(\mathds{R}^{p}\times\mathds{R}^{q}\times\mathds{R}^{m},\mathds{R}^{m}) has the high-gain property , i.e. there exists \nu\in(0,1) such that, for every compact sets K_{p}\subset\mathds{R}^{p} , K_{q}\subset\mathds{R}^{q} , the function \mathfrak{h}:\mathds{R}\to\mathds{R} defined by \mathfrak{h}(s):=\min\left\{\ \langle v,f(\delta,z,-sv)\rangle\!\!\ \left|\ \vphantom{\langle v,f(\delta,z,-sv)\rangle\!\!}\!\!\!\begin{array}[]{ll}\delta\in K_{p},z\in K_{q},\\ v\in\mathds{R}^{m},\nu\leq\|v\|\leq 1\end{array}\!\!\!\ \right.\right\} satisfies \sup_{s\in\mathds{R}}\mathfrak{h}(s)=\infty . Remark 2.1 Under