Robust Adaptive Boundary Control of a Thermal Process with Thermoelectric Actuators: Theory and Experimental Validation

Basic Information

• Paper ID: 2510.09169
• Title: Robust Adaptive Boundary Control of a Thermal Process with Thermoelectric Actuators: Theory and Experimental Validation
• Authors: Paul Mayr (Graz University of Technology), Alessandro Pisano (University of Cagliari), Stefan Koch (Graz University of Technology), Markus Reichhartinger (Graz University of Technology)
• Classification: eess.SY (Systems and Control), cs.SY (Computer Science - Systems and Control)
• Publication Date: October 10, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2510.09169

Abstract

This paper proposes an adaptive sliding mode-based boundary control law for a class of uncertain thermal reaction-diffusion processes subject to matched disturbances. Since the disturbances are assumed to be bounded but with unknown bounds, an adaptive control strategy is required. The proposed boundary control law comprises a proportional term and a discontinuous term, where the amplitude of the discontinuous relay term is adjusted through a gradient-based adaptive algorithm. Depending on the parameterization of the adaptive algorithm, the adaptive gain can be a non-decreasing function of time (one-way adaptation) or can increase and decrease (two-way adaptation). The convergence and stability of both solutions are studied through Lyapunov analysis, yielding two distinct stability results: asymptotic stability for one-way adaptation and global uniformly ultimately bounded solutions for two-way adaptation. The proposed algorithm is subsequently applied to the temperature distribution stabilization control problem of a metal beam equipped with thermoelectric boundary actuators and experimentally validated.

Research Background and Motivation

Problem Background

1. Control Challenges in PDE Systems: Many industrial processes such as tubular reactors, lithium-ion batteries, and distillation processes are described by diffusion-type PDEs, and control of these systems is often limited to boundary actuation in the spatial domain
2. Uncertainty and Disturbances: Real systems frequently face severe uncertainties, such as spatial variations in material properties, external disturbances, and actuator defects
3. Limitations of Existing Methods: Traditional sliding mode control requires known disturbance bounds, but in practical applications, disturbance bounds are often unknown

Research Motivation

• Limited application of existing adaptive sliding mode control in PDE systems, particularly for diffusion processes containing reaction terms
• Need to address unknown parameters under Robin boundary conditions
• Practical applications require comparison of performance differences between one-way and two-way adaptive algorithms

Core Contributions

1. Theoretical Extension: Extension of adaptive sliding mode boundary control to thermal diffusion processes with reaction terms, addressing unknown parameters under Robin boundary conditions
2. Dual Stability Analysis: Complete Lyapunov stability proofs for both one-way adaptation (asymptotic stability) and two-way adaptation (uniformly ultimately bounded)
3. Experimental Validation: Verification of theoretical results on an aluminum beam experimental platform equipped with thermoelectric modules (TEM)
4. Algorithm Comparison: Systematic comparison of two adaptive strategies in practical applications, particularly regarding chattering suppression

Methodology Details

Problem Formulation

Consider a scalar field z(x,t) varying in space and time, where x ∈ 0,1 is the spatial variable and t ≥ 0 is the time variable. The system is described by the following perturbed initial-boundary value problem:

PDE Equation:

z_t(x,t) = [θ(x)z_x(x,t)]_x + λ(x)z(x,t)

Robin Boundary Conditions:

-b_0 z_x(0,t) + a_0 z(0,t) = u_0(t) + ψ_0(t)
b_1 z_x(1,t) + a_1 z(1,t) = u_1(t) + ψ_1(t)

Where:

• θ(x): spatially-varying diffusion coefficient
• λ(x): spatially-varying reaction coefficient
• u_i(t): controllable boundary control inputs
• ψ_i(t): unknown bounded matched disturbances

Controller Architecture

Adaptive Sliding Mode Boundary Control Law:

u_i(t) = -k_i z(x_i,t) - M_i(t) sign(z(x_i,t))

Adaptive Law:

Ṁ_i(t) = -α_i M_i(t) + γ_i |z(x_i,t)|

Parameter Descriptions:

• k_i > 0: proportional gain
• M_i(t): adaptive switching gain
• α_i ≥ 0: adaptation parameter (α_i = 0 for one-way, α_i > 0 for two-way)
• γ_i > 0: adaptation rate

Technical Innovations

1. Extended System Class: Compared to reference 13, this paper addresses more general systems containing reaction term λ(x)z(x,t)
2. Robin Boundary Conditions: Handling more complex boundary conditions including unknown parameters a_i, b_i
3. Dual Stability Theory:
   • One-way adaptation: global asymptotic stability
   • Two-way adaptation: global uniformly ultimately bounded
4. Application-Oriented: Considers dead-zone correction and chattering issues in practical implementation

Experimental Setup

Experimental Platform

• Main Component: Aluminum alloy beam (EN AW-6060)
• Dimensions: Length 315 mm, thickness 3 mm, width 25 mm
• Actuators: Thermoelectric modules (TEM/Peltier elements) at both ends
• Sensors: Thermal imaging camera measuring temperature distribution across the beam
• Disturbance Source: Additional TEM for applying matched boundary disturbances

System Parameters

Control Parameter Settings

• Desired boundary temperatures: T_{e,0} = 30°C, T_{e,L} = 34°C
• Ambient temperature: T_a ≈ 24.6°C (one-way) / 24.1°C (two-way)
• Proportional gain: k_i = 15A
• Sampling time: T_s = 0.1s
• Dead-zone parameter: ε_0 = ε_1 = 0.5°C (one-way adaptation only)

Experimental Results

Main Results

Experimental Phase Design:

1. t = 0-2min: Initialization phase
2. t = 2-6min: Proportional control only (M_i(t) = 0)
3. t = 6-10min: Sinusoidal disturbance applied, no adaptation
4. t = 10-16min: Adaptation enabled, large-amplitude disturbance
5. t = 16-32min: Disturbance amplitude reduced

Key Findings:

1. Disturbance Rejection: Temperature oscillations significantly reduced after enabling adaptation
2. Steady-State Performance: Two-way adaptation can reduce gains after disturbance reduction, minimizing chattering
3. Chattering Comparison:
   • One-way adaptation: E = 16K²
   • Two-way adaptation: E = 10.6K² (where E is the energy measure of temperature deviation)

Adaptive Gain Evolution

• One-way adaptation: Gain monotonically increases to a level sufficient to suppress disturbances and remains constant
• Two-way adaptation: Gain dynamically adjusts based on disturbance magnitude, decreasing when disturbances reduce

Temperature Control Accuracy

The measured temperature distribution at the end of the experiment closely matches the computed reference equilibrium state, validating the effectiveness of the control algorithm.

Related Work

Sliding Mode Control of PDE Systems

• Pisano and Orlov 6: Sliding mode control of parabolic PDEs with ISS properties
• Gu and Wang 7: n-coupled reaction-diffusion PDEs with boundary input disturbances
• Koch et al. 8: Sliding mode boundary control of coupled PDE/ODE systems

Adaptive Sliding Mode Control

• Plestan et al. 12: Survey of adaptive sliding mode control for finite-dimensional systems
• Roy et al. 14: Adaptive sliding mode control without prior uncertainty bounds
• Mayr et al. 13: Adaptive sliding mode boundary control of perturbed diffusion processes

Advantages of This Work

Compared to existing work, this paper is the first to systematically address thermal diffusion processes with reaction terms, providing comprehensive theoretical analysis and experimental validation.

Conclusions and Discussion

Main Conclusions

1. Theoretical Contribution: Established rigorous stability theory for both adaptive strategies
2. Practical Value: Two-way adaptation demonstrates superior performance in practical applications, particularly in chattering suppression
3. Experimental Validation: Successfully verified theoretical predictions on thermoelectric actuation systems

Limitations

1. Implementation Constraints: One-way adaptation requires dead-zone correction to prevent unbounded gain drift
2. Model Assumptions: Requires matched and bounded disturbances
3. Parameter Requirements: Certain system parameters (such as λ_M, θ_m, b_{M,i}) must be known for parameter tuning

Future Directions

1. Unmatched Disturbances: Addressing unmatched disturbances in the domain
2. Input Saturation: Handling actuator saturation constraints
3. Multi-variable Extension: Extension to multi-input multi-output systems

In-Depth Evaluation

Strengths

1. Theoretical Rigor: Provides complete Lyapunov stability analysis with rigorous mathematical derivations
2. Strong Practicality: Addresses practical implementation issues such as dead-zones and chattering
3. Comprehensive Experiments: Detailed comparative experiments on real physical systems
4. Clear Innovation: Distinct theoretical and application contributions in PDE adaptive control

Weaknesses

1. Limited Scope: Restricted to one-dimensional diffusion processes and specific boundary conditions
2. Parameter Tuning: Requires certain prior knowledge for control parameter settings
3. Computational Complexity: Real-time implementation may face computational burden

Impact

1. Academic Value: Provides new theoretical framework for adaptive control of PDE systems
2. Application Prospects: Broad application potential in thermal management, chemical reactors, and related fields
3. Reproducibility: Detailed experimental setup and parameters facilitate reproduction

Applicable Scenarios

• Thermal management systems (electronic device cooling, building temperature control)
• Chemical reactor control
• Battery thermal management
• Temperature control in material processing

References

The paper cites 22 relevant references covering multiple domains including PDE control, sliding mode control, and adaptive control, providing a solid theoretical foundation for this research.

Overall Assessment: This is a high-quality paper emphasizing both theory and practice, making important contributions to the field of adaptive control of PDE systems. The theoretical analysis is rigorous, experimental validation is comprehensive, and the work provides significant guidance for practical applications.