1,721,034 research outputs found
One-shot set-membership identification of Generalized Hammerstein–Wiener systems
No full textIn this paper we consider the identification of Generalized Hammerstein–Wiener systems in the presence of bounded noise affecting both the input and the output sequences. The considered class of systems is characterized by the cascade connection of Hammerstein building blocks, where the nonlinear static input map is a polynomial, which is not assumed to be invertible. The problem of computing the parameter uncertainty intervals is formulated in terms of a set of suitable semialgebraic optimization problems, where both the parameter to be estimated and the inner unmeasurable signals connecting all the subsystems are considered as decision variables. The effectiveness of the proposed algorithm is shown by means of a numerical simulation example. The presented approach is also applied to the problem of identifying the mathematical model of a JFET amplifier from a set of experimentally collected data
Fast sparse optimization via adaptive shrinkage
Full text linkThe need for fast sparse optimization is emerging, e.g., to deal with large-dimensional data-driven problems and to track time-varying systems. In the framework of linear sparse optimization, the iterative shrinkage-thresholding algorithm is a valuable method to solve Lasso, which is particularly appreciated for its ease of implementation. Nevertheless, it converges slowly. In this paper, we develop a proximal method, based on logarithmic regularization, which turns out to be an iterative shrinkage-thresholding algorithm with adaptive shrinkage hyperparameter. This adaptivity substantially enhances the trajectory of the algorithm, in a way that yields faster convergence, while keeping the simplicity of the original method. Our contribution is twofold: on the one hand, we derive and analyze the proposed algorithm; on the other hand, we validate its fast convergence via numerical experiments and we discuss the performance with respect to state-of-the-art algorithms
A kernel-based approach to errors-in-variables identification of stable multivariable linear systems
Full text linkIn this paper, we present a kernel-based non-parametric approach to identifying stable multi-input multi-output linear systems in the presence of bounded noise affecting both the input and the output measurements. Firstly, we formulate the considered problem in terms of robust optimization techniques. Then, we show that the formulated robust optimization problem can be solved using semidefinite optimization. Since the involved optimization problem is computationally demanding, we also provide a result that allows the user to compute a bound on the approximation error introduced by considering reduced complexity models. We present some simulation examples to show the effectiveness of the proposed approach. Finally, we apply the proposed identification method to the dataset experimentally collected on a linear electronic filter
A non-convex adaptive regularization approach to binary optimization
Full text linkBinary optimization is a long-time problem ubiquitous in many engineering applications, e.g., automatic control, cyber-physical systems and machine learning. From a mathematical viewpoint, binary optimization is an NP-hard problem, to solve which one can find some suboptimal strategies in the literature. Among the most popular approaches, semidefinite relaxation has attracted much attention in the last years. In contrast, this work proposes and analyzes a non-convex regularization approach, through which we obtain a relaxed problem whose global minimum corresponds to the true binary solution of the original problem. Moreover, because the problem is non-convex, we propose an adaptive regularization that promotes the descent towards the global minimum. We provide both theoretical results that characterize the proposed model and numerical experiments that prove its effectiveness with respect to state-of-the-art methods
Lasso-based state estimation for cyber-physical systems under sensor attacks
Full text linkThe development of algorithms for secure state estimation in vulnerable cyber-physical systems has been gaining attention in the last years. A consolidated assumption is that an adversary can tamper a relatively small number of sensors. In the literature, block-sparsity methods exploit this prior information to recover the attack locations and the state of the system. In this paper, we propose an alternative, Lasso-based approach and we analyse its effectiveness. In particular, we theoretically derive conditions that guarantee successful attack/state recovery, independently of established time sparsity patterns. Furthermore, we develop a sparse state observer, by starting from the iterative soft thresholding algorithm for Lasso, to perform online estimation. Through several numerical experiments, we compare the proposed methods to the state-of-the-art algorithms
A Pitch Wave Force Prediction Algorithm for the Inertial Sea Wave Energy Converter
No full textIn this paper we introduce a pitch wave force predictor for a wave energy converter. The designed predictor is composed of two subsystems. First, we design a virtual sensor in order to provide a real-time estimate of the pitch wave force acting on the converter, from measurements provided by a set of on-board sensors. Then, we build an autoregressive predictor on such a virtual measurement to predict the future value of the wave force over a finite prediction horizon length. Simulation results, performed by applying the proposed algorithm on a first-principles based Simulink model of the considered sea wave energy converter, demonstrate the effectiveness of the proposed approach
A New Framework for Constrained Optimization via Feedback Control of Lagrange Multipliers
Full text linkThe continuous-time analysis of iterative algorithms for optimization has a long-standing history. This work introduces a novel framework for equality-constrained optimization based on control theory. The central concept is to design a feedback control system in which the Lagrange multipliers serve as the control inputs while the output represents the constraints. This system converges to a stationary point of the constrained optimization problem through suitable regulation. Concerning the Lagrange multipliers, we explore two control laws: proportional-integral control and feedback linearization. These choices lead to a variety of different methods. We rigorously develop the related algorithms, analyze their convergence theoretically, and present several numerical experiments that demonstrate their effectiveness compared to the state-of-the-art approaches
Direct data-driven controller design from bounded errors-in-variables measurements
Full text linkThe work focuses on direct data-driven controller design using input-output measurements corrupted by bounded additive noise. The main goal is to tackle the data-driven model reference control problem. To this end, we employ a set-membership framework and define the feasible controller parameter set, i.e., the set of parameters consistent with the noise bounds and the model-matching condition. We determine the controller parameters as the Chebyshev center of this set. We also provide a data-driven condition that is sufficient for establishing stability. This condition is also robust against the presence of bounded noise. The approach utilizes polynomial optimization and achieves global optimality via semidefinite relaxation techniques. A simulation example demonstrates the effectiveness of the proposed approach
A unified framework for the identification of a general class of multivariable nonlinear block‐structured systems
No full textThis article addresses the parametric identification of block-structured nonlinear systems in a general form, characterized by the feedback interconnection of a multivariable linear system and a static multivariate nonlinear map. We assume that both the input and the output collected data are affected by bounded noise, which casts the problem in the context of set-membership (SM) errors-in-variables identification. We introduce a single-stage SM identification algorithm for the computation of the parameter uncertainty intervals. The proposed solution exploits the formulation of a suitable optimization problem solved through convex relaxation techniques. Numerical simulations and an experimental test show the effectiveness of the proposed approach
- …
