Safe Trajectory Tracking of the Stefan Problem with Second-Order Moving Boundary Dynamics
Shumon Koga, Miroslav Krstic · 2026
Read Safe Trajectory Tracking of the Stefan Problem with Second-Order Moving Boundary Dynamics by Shumon Koga, Miroslav Krstic (2026) 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-2605-25084.txt.
Introduction
Shumon Koga and Miroslav Krstic address safe tracking for the Stefan problem with a moving edge. The model pairs a parabolic partial differential equation on a moving domain with an edge governed by second-order rules under a Neumann rule. A series-expansion feedforward control solves the inverse Stefan problem for a reference trajectory, and the series is proven to converge. An energy-shaping controller keeps the constraint safe in closed loop. The system is globally stable with fast decay in tracking error, and a sine-wave simulation shows tracking.
Catalog record
| CREATOR | Shumon Koga, Miroslav Krstic |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 27983 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.25084
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Safe Trajectory Tracking of the Stefan Problem with Second-Order Moving Boundary Dynamics Shumon Koga, Miroslav Krstic Abstract This paper considers a safe trajectory tracking of the Stefan problem with a second-order moving boundary dynamics. The model is given by a parabolic Partial Differential Equation (PDE) defined on a time-varying domain of moving boundary governed by a second-order Ordinary Differential Equation (ODE) associated with the Neumann boundary condition. A feedforward control is designed by a series expansion approach to solve the inverse Stefan problem under given reference trajectory of the moving boundary, and the convergence of infinite series is proven. A trajectory tracking controller is derived based on an energy-shaping, which ensures the safety of the model constraint in the closed-loop system. The closed-loop system is also shown to be globally exponentially stable with respect to the tracking error by performing PDE backstepping transformation and Lyapunov analysis. Numerical simulation illustrates an effective tracking performance of the proposed method under a sinusoidal reference trajectory. Code is released at https://github.com/shumon0423/StefanTracking_ACC2026.git . I INTRODUCTION The Stefan problem is a classical mathematical model representing physical phenomena of thermal phase change, such as melting or solidification, which is widely studied in many applications, including sea ice melting [ 35 ] , additive manufacturing [ 39 ] , continuous casting [ 3 ] , electrosurgery [ 33 ] , and cell therapy [ 37 ] . The Stefan problem comprises a partial differential equation (PDE) for the temperature profile, defined over a time-varying spatial domain whose moving boundary is governed by an ordinary differential equation (ODE). This modeling approach, where the spatial domain grows or shrinks, is also known as a free boundary problem and has been incorporated even beyond thermal dynamics for applications such as modeling chemical reactions in batteries [ 32 ] , biological cell growth [ 13 ] , and option pricing [ 28 ] . Although the Stefan problem has been studied for more than a century, the geometric nonlinearity arising from the coupling between the PDE and the moving boundary presents an inevitable challenge. Consequently, ongoing work has focused on studying the solution of the Stefan problem using state-of-the-art methods such as Bayesian optimization [ 42 ] and physics-informed machine learning [ 41 ] . Control design for the Stefan problem has been developed over the last two decades using several approaches, such as optimal control [ 16 ] , model predictive control [ 12 ] , geometric control [ 26 ] , and enthalpy-based control [ 30 ] . The PDE backstepping method, a systematic approach for designing boundary controllers and observers for many classes of PDEs and delay systems as summarized in a recent survey paper [ 38 ] , has been successfully applied to the Stefan problem, ensuring global exponential stability [ 19 ] and validated in physical experiments [ 25 ] . Several extensions of the backstepping design for the Stefan problem are provided in [ 21 ] . Other classes of PDEs with moving boundaries have also been handled by the backstepping method, including traffic shockwaves [ 44 ] , piston dynamics [ 5 ] , and neuron growth [ 8 ] . Trajectory tracking for the Stefan problem, which aims to follow a desired behavior of the moving boundary, has also been studied in a few works. A pioneering work on motion planning by series expansion [ 10 ] provides a solution for the reference temperature of a nonlinear Stefan problem for an a priori given moving boundary as a function of time. Trajectory tracking control designs to stabilize the Stefan system at a reference solution have been proposed using enthalpy-based feedback [ 31 ] and a flatness-based approach [ 11 ] . Moreover, [ 14 ] proposed an extremum-seeking control for the Stefan problem to drive the system to an unknown optimum by tracking a reference trajectory that includes sinusoidal perturbations on a cost map. However, in all those works, stability analysis has been performed under assumptions on either closed-loop performance or a static reference. The reasons for these assumptions are twofold: first, the requirement that a physical condition being satisfied under closed-loop control, and second, the spatial domain discrepancy between the system and reference profiles. As shown in [ 11 ] , once the reference error PDE state is defined to maintain consistent conditions at the moving boundary, the error dynamics includes a time derivative of the reference state at the boundary. This term cannot be bounded in standard L_{2} or H_{1} norms for the classical Stefan problem with first-order moving boundary dynamics, as the velocity gap is equivalent in norm to the Neumann boundary value. Ensuring that constraints are satisfied in control systems has recently been studied under the notion of safety and Control Barrier Function (CBF) [ 2 ] . For trajectory tracking, combining a CBF-based safety filter with a preset planning and low-level control architecture was addressed in [ 1 ] . Based on an energy-shaping approach and passivity [ 29 ] , an energy-based safety constraint and its associated CBF to preserve passivity were proposed in [ 36 ] for handling kinematic constraints in robotic systems. Several extensions combining passivity with CBF-based safety have since been developed [ 7 , 6 ] . Safe or constrained control of PDE systems has also been studied in several works. Early work by [ 9 ] proposed a model predictive control approach that incorporates state and input constraints through a modal decomposition of a parabolic PDE. Input constraints under saturation for unstable parabolic PDEs have been handled using linear matrix inequalities [ 27 , 43 ] . A Lyapunov-based constraint guarantee for PDEs has been addressed for fluid dynamics [ 18 ] and batteries [ 34 ] . The safe control design by a high-order CBF for PDEs was initially proposed for the Stefan problem with actuator dynamics in [ 23 ] as well as an event-triggered mechanism [ 22 ] , and later for hyperbolic PDEs in [ 40 ] . A learning-based approach for the safe control of PDEs was proposed in [ 17 ] using a neural operator [ 4 ] . This paper is the first to provide a rigorous proof of safety and tracking stability of the Stefan problem with a trajectory tracking control. We consider the Stefan problem with second-order moving boundary dynamics, proposed in [ 24 ] , to represent thermal inertia in the phase change dynamics. The feedforward control is designed following the series expansion method, and convergence of the infinite series under the second-order moving boundary dynamics is also shown. The trajectory tracking control is then designed via energy-shaping with respect to the energy function of the reference solution, and the safety condition is proven to hold by maximum principle. Based on the reference PDE state proposed in [ 11 ] , the stability analysis in the tracking error is employed using a Lyapunov approach and a PDE backstepping transformation. Numerical simulation is performed with a sinusoidal reference trajectory. II PROBLEM STATEMENT Fig. 1 : Schematic of Stefan problem and a reference trajectory. II-A Stefan Problem with Second-Order Moving Boundary We consider the one-phase Stefan problem modeling the dynamics of the temperature profile T(x,t)