CS Paper Radar

Sample-Covariance Regularization for Data-Driven Predictive Control from Frequency Response Functions under Colored Noise

M. Wind, T. J. Meijer, V. S. Dolk, W. P. M. H. Heemels

eess.SY ·submitted 2026-11-06 21:14 UTC · v1

Abstract

In this work, we investigate how sample covariance obtained from frequency-response functions (FRFs) can be exploited to improve performance of data-driven frequency-domain predictive control (FreePC) in the presence of colored measurement noise. In case of periodic multisine excitation, this covariance is obtained directly from repeated periods and provides a frequency-dependent measure of uncertainty without requiring an additional parametric noise model. To incorporate this information into both indirect and direct predictive control, we develop a frequency-domain subspace predictive controller (FD-SPC) and establish its equivalence to the direct FreePC formulation in the noiseless setting. Regularization terms for both formulations are then motivated by expected prediction-error criteria and constructed from the sample covariance of the FRF estimates. Across all tested noise models and data sizes, regularization consistently reduces the median realized control cost relative to the corresponding controller without covariance regularization, with reductions of up to 35%.

Closest papers

Nearest neighbours by meaning across the whole index.

Stochastic Data-driven Predictive Control of Linear Systems with Sub-Gaussian Disturbances using Causal Predictors

eess.SY ·2026-09-22 ·Johannes Teutsch, Marion Leibold
published similarity 0.63
Abstract · This manuscript is a revised version of the article published in IFAC Journal of Systems and Control (2026), vol. 25, pp. 100399. It contain…

We present a stochastic data-driven predictive control (DPC) framework for discrete-time linear time-invariant systems subject to sub-Gaussian additive disturbances based solely on input--output data. In contrast to related methods that rely on exact disturbance data or at least sample generation for closed-loop guarantees, the proposed approach leverages a disturbance data estimate. By enforcing consistency of the disturbance data estimate with the available input--output data and system class, we first identify data-driven and provably causal subspace predictors for use in DPC. Then, we analyze statistical properties of the corresponding prediction error, yielding tightened constraints for the nominal predictions that guarantee satisfaction of chance constraints. The proposed DPC scheme comes with guarantees on recursive feasibility and conditional chance constraint satisfaction in closed-loop under standard assumptions. A numerical evaluation study demonstrates the performance of the proposed controller.

0.51

Fundamental Limits of Data-Driven Predictive Control

eess.SY ·2026-11-11 ·Jiabao He, Feiran Zhao, Yushan Li, Yue Ju, Florian Dörfler +1 more
benchmarktheoryhot direction similarity 0.62
Abstract

Data-driven predictive control (DDPC) has attracted substantial interest, and many variants have been proposed; however, comparatively less is known about the statistical closed-loop performance of these methods and the fundamental limits they are subject to. In this contribution, we apply the statistical decision framework developed in our earlier work to study fundamental limits of DDPC. The performance of a data-driven controller is measured by its risk, the expected deterioration in control performance relative to an oracle model-based controller at a given system, and by its total risk that measures the overall performance across all systems of interest. Lower bounds on these risks are derived using the bias-variance decomposition and the Cramer-Rao inequality. In particular, a lower bound on the total risk is obtained by a calculus-of-variations method. This bound applies to any DDPC design, thereby quantifying unavoidable performance limitations and providing a benchmark for assessing existing DDPC methods. In an illustrative example, we compare the risks of certainty-equivalence model predictive control, subspace predictive control, and data-enabled predictive control with the derived lower bounds, providing a quantitative performance assessment of these methods and some guidance on tuning regularization parameters in DDPC.

0.26

Multirate Data-Driven Predictive Control: Rank Obstructions, Identifiability, and Noise Conditioning

eess.SY math.OC ·2026-11-01 ·Robert Vrabel
similarity 0.60
Abstract

We develop rank, identifiability, and noise-conditioning results for compact input--output predictors in multirate data-driven predictive control. The inverse of a square zero-input response matrix used in an existing compact-lifting construction is shown to be undefined under its observability and minimal-polynomial assumptions. We derive exact replacements using rank-complete data, characterize all predictor extensions, and establish the minimum-norm and minimax Euclidean noise properties of the canonical Moore--Penrose extension. A joint predictor is identified directly from mixed experiments with arbitrary initial states. Deterministic perturbation bounds quantify reconstruction errors through informative singular-value margins, while a positive-definite reduced weighting preserves the nominal quadratic cost. Sampling-rate observability and event-triggering analyses identify additional conditions needed for reconstruction and stability guarantees.

0.30

Forgetting While Remembering, an Invariant Online Data-Driven Predictive Control Formulation

eess.SY ·2026-09-16 ·Alessandro Chiuso, Florian Dörfler, Keith Moffat
similarity 0.59
Abstract · 8 pages, 3 figures, conference paper (CDC)

Low signal-to-noise ratio (SNR) data is a core challenge of online Data-Driven Predictive Control (DPC) for linear, time-varying systems. This paper proposes a Bayesian, online DPC framework based on autoregressive models with exogenous inputs (ARX) that uses an externally-provided prior, which encodes inductive bias such as smooth system dynamics and stability, to safeguard performance when SNR is low. The posterior estimate of the ARX parameter is propagated forward in time using a Kalman filter with a state equation defined by an adaptation-rate hyperparameter, which is adjusted online to track the rate at which the underlying system dynamics evolve. The Kalman filter's posterior mean and covariance determine the DPC's Final Control Error cost function, which is the posterior expectation of the quadratic cost function. Critically, the Kalman filter's process equation is chosen so that, regardless of the hyperparameter adaptation, the prior distribution is invariant over time. Thus, while old data is forgotten, the prior is not. DPC tracking experiments on a time-varying second-order system demonstrate the efficacy of the proposed method.

0.56

Time-series Foundation Models for Predictive Control: The Role of Excitation

cs.LG eess.SY ·2026-10-05 ·Mazen Amria, Jasper Hoffmann, Philipp Bordne, Anna Rothenhäusler, Lilli Frison +3 more
published similarity 0.59
Abstract · Accepted at the TS-LIMITS Workshop at NeurIPS 2026

Deploying model predictive control (MPC) requires constructing or identifying a predictive model for each target system. Time-series foundation models (TSFMs) offer an attractive option thanks to strong zero-shot forecasting capabilities across systems. However, low forecast error does not guarantee that a TSFM captures the system's response to the alternative actions considered by the controller. We study this gap using residential heat-pump control as a test bed, measuring the agreement between predicted and ground-truth effects of control interventions. Importantly, we find that TSFMs can recover the system's input-response relationship when the context contains sufficient independent control excitation. Common fine-tuning pipelines and feature smoothing reduce, but do not eliminate, the need for in-context excitation. Our results indicate that current TSFMs used for predictive control require sufficiently informative control variation in the inference context. Initial closed-loop results show promise for shorter context windows.

0.36

DeoPC: Data-Enabled Output-Feedback Predictive Control for State Regulation

eess.SY math.DS ·2026-11-19 ·Jingyuan Li, Dawei Shi, Ling Shi
similarity 0.58
Abstract · 14 pages, 4 figures

Constrained regulation of the internal state of an unknown linear system is challenging when only input-output measurements are available online. In this work, we develop a data-enabled output-feedback predictive control (DeoPC) framework for constrained state regulation using input-output data. Specifically, a data-driven state estimate is constructed from past input-output data via a subspace-based singular value decomposition. Based on this state representation, we formulate a predictive controller that directly penalizes the state and control effort while enforcing the system constraints. Under persistent excitation and initialization conditions, we establish recursive feasibility and closed-loop stability of DeoPC in both noise-free and bounded-noise settings. In addition, we show that DeoPC can be calibrated to the physical state coordinates using a few offline state measurements, or treated as a model predictive control tuning problem when no state measurements are available. The effectiveness of DeoPC is demonstrated through simulations on an unstable batch reactor system.

0.26

Fast-DeePC: Fast Data-Enabled Predictive Control using Soft-constrained Scheme

eess.SY ·2026-10-09 ·Terry Zhang, Jan Drgona, Malika Sader, Di Liu, Liang Wu
similarity 0.58
Abstract · 8 pages, submission to ACC 2027

Data-enabled predictive control (DeePC) replaces the explicit system model of model predictive control with a data-driven predictor built directly from recorded input-output trajectories, but the original hard-equality formulation keeps its coordinate vector as an online decision variable whose size grows with the offline data length, so the online problem grows more expensive as more data is used to enrich the behavioral representation. This article proposes Fast-DeePC, a framework which relaxes this consistency relation into the control objective with a quadratic ridge on the coordinate vector in place of the customary sparsity-promoting penalty, admitting a closed-form, offline elimination of the coordinate vector and reducing the online problem to a fixed-size box-constrained quadratic program solved by a structure-exploiting interior-point method. On identical soft quadratically regularized problems, the result matches a general-purpose solver's solution and is up to two orders of magnitude faster.

0.24

Robust Data-Driven Controller Synthesis from Sampled and Uncertain Frequency Response Estimates using Integral Quadratic Constraints

eess.SY ·2026-10-14 ·Luuk Marceau Spin, Maurice W. P. M. H. Heemels, Nathan van de Wouw, Roland Toth
similarity 0.58
Abstract · 15 pages, 6 figures, none other publications

In this paper, we develop a novel MIMO controller synthesis method that relies solely on frequency-domain data of discrete-time systems. The approach is based on an iterative design procedure using \emph{integral quadratic constraints} (IQCs). It generalizes existing data-driven methods that approximate semi-infinite IQC conditions by enforcing them on a finite set of measured frequency points. Such approximations do not provide guarantees on the continuous frequency axis and may fail to capture inter-frequency behavior. We address this limitation by formulating a convex optimization problem that preserves robustness guarantees over the entire frequency range. In particular, the proposed method explicitly accounts for inter-frequency behavior of both plant and controller, leading to certified robust stability and performance. The effectiveness and computational efficiency of the approach are demonstrated using experimentally obtained frequency response data from industrial-grade semiconductor equipment. The synthesized controller is implemented on the physical setup, and its performance is validated in real-time experiments.

0.35