1,720,982 research outputs found
Model Predictive Control for Electrical and Mechatronic Systems
No full textL'attività di ricerca presentata in questa tesi riguarda il progetto e lo sviluppo di sistemi di controllo basati sulla predizione del comportamento del processo da controllare, in modo da ottenere prestazioni di controllo ottimizzate. I processi presi in considerazione sono convertitori elettrici di potenza e sistemi meccatronici. L'uso di algoritmi di controllo basati sul modello per predire l'evoluzione del processo è stata limitato a lungo a sistemi caratterizzati da dinamiche lente. Ciò è dovuto al tempo necessario a calcolare la legge di controllo, che prevede la risoluzione di un problema di minimizzazione vincolata da efettuarsi nell'intervallo di campionamento indicato nelle specifiche di progetto (in tempi minori in caso di unità di controllo che gestiscono più processi). Negli ultimi anni, la potenza di calcolo delle schede di controllo embedded è significativamente aumentata, permettendo di applicare le tecniche di controllo predittivo a sistemi caratterizzati da dinamiche veloci.
Considerando le potenzialità del controllo predittivo, in questa tesi differenti leggi di controllo sono state sviluppate per far fronte a problematiche di controllo relative a diferenti casi di studio. Approcci MPC basati su modelli lineari tempo invariati (LTI) sono state considerati per il controllo di sistemi di micro-posizionamento piezoelettrici e di convertitori di potenza. In questo ambito sono state presentate le basi teoriche che permettono di applicare con successo tali tecniche su schede di controllo caratterizzate da ridotta potenza computazionale. Controllori per sistemi avionici e marini sono stati inoltre considerati. In questo ambito, sono stati sviluppati modelli lineari a parametri tempo varianti (LPV) rispetto ai quali sono stati sviluppati i sistemi di controllo. In questo ambito è stata sviluppata inoltre un'innovativa tecnica che permette di ridurre il carico computazionale richiesto dagli algoritmi MPC basati su modelli LPV, di modo da permetterne l'implementazione su schede embedded. Tutti i sistemi di controllo sviluppati sono stati testati su processi reali laddove possibile, ed in alternativa su simulatori altamente realistici.The research activity presented in this thesis concerns the design and the development of control systems based on the prediction of the behavior of the process to be controlled, in order to obtain optimized control performance. The processes taken into account are electrical power converters and mechatronic systems. The use of model-based algorithms to predict the process evolution has been limited for a long time to systems featured by slow dynamics. This issue is related to the time needed to compute the control law, which requires the real time resolution of a constrained optimization problem, that must be always solved in a prescribed amount of time, usually corresponding to the sampling period or even shorter in multi-purpose control units. In recent years the computational the increasing power of control units has allowed the application of predictive control algorithms to complex and faster dynamic processes.
Considering the potential of the predictive control to solve multivariable control problems, thanks to the optimized performance and the efective handling of systems constraints, in this thesis diferent control algorithms have been developed to solve problems linked to several case studies. Linear Time Invariant (LTI) model-based MPC approaches have been considered for micro-positioning piezoelectric systems and power converters. In this context theoretical conditions to efectively apply MPC for pre-compensed systems on low computational hardware have been given. Control approaches for avionic and marine systems have been also considered. In this context Linear Parameter Varying (LPV) models have been developed and an innovative solution has been studied to reduce the computational efort of LPV-based MPC on em- bedded control platform. All the above controllers have been tested on real processes, wherever possible, and through highly realistic simulators
Model predictive control for pre-compensated power converters: Application to current mode control
No full textCurrent Mode Control (CMC) is the standard approach to regulate DC-DC power converters in high performance applications, allowing to obtain a faster time-response and better closed-loop stability if compared to Voltage Mode Control (VMC). In the last decade, several algorithms have been proposed to improve standard CMC, most of them requiring to replace the original controller. However, it is common to have either analog or embedded CMC controllers which cannot be replaced easily in commercial power converters. Inspired by very recent results in the topic, this paper proposes a Model Predictive Control (MPC) external loop aimed at optimally modifying the set-point of a CMC loop to improve converter performance. The proposed configuration is directly applicable to any pre-compensated converter as it does not require changes on the already-in-place controller. Moreover, by leveraging a multi-rate implementation, the benefits of MPC are introduced in power conversion without affecting much the computational cost of the over-all control system, contrary to what would happen for a direct MPC implementation. Simulation and experimental results on a synchronous DC-DC buck converter, controlled by a standard CMC algorithm, confirm the benefits of the approach. © 2019 The Franklin Institut
Fault tolerant model predictive control for an over-actuated vessel
No full textThis paper presents a Fault Tolerant-Model Predictive Control (FT-MPC) for a vessel featured by the presence of a greater number of actuators with respect to the number of controlled outputs, classified in the category of over-actuated systems. Over-actuated plants are usually controlled by a main controller in conjunction with a Thrust Allocation (TA) algorithm in order to guarantee the required control performances. In this work an unconstrained Quadratic Programming (QP) TA policy is considered in conjunction with a MPC to drive the vessel. The proposed solution has been tested on an over-actuated vessel called Cybership II. The main contribution of this paper is the introduction in the MPC of a fault tolerant action, in order to improve control performance in actuators’ fault scenarios. A Linear Parameter-Varying (LPV) model has been used to described the Cybership II dynamics and to develop the proposed controller. Considering this model, a MPC has been developed to drive the vessel, verifying controller performance in standard control scenarios. The proposed FT-MPC has been compared with respect to a standard MPC in actuators’ fault scenarios, considering several wave noise disturbances. Reported simulation results show the effectiveness of the proposed approach. © 2018 Elsevier Lt
A fast model predictive control algorithm for linear parameter varying systems with right invertible input matrix
No full textDevelopment and experimental validation of a LQG control for a pre-compensated multi-axis piezosystem
No full textAdaptive Reference Governor for DC–DC Converters Based on Model Predictive Control
Full text linkIn this article, we propose a time-varying model predictive control (MPC)-based scheme to enhance the dynamic performance of dc–dc converters. The proposed approach employs MPC as a reference governor (RG), addressing industrial certification constraints that may limit modifications to the low-level controller. To accommodate the computational limitations of conventional control boards, we introduce a highly efficient real-time optimization algorithm for solving equality-constrained quadratic programming (QP) problems. The algorithm is based on a tailored QR factorization that outperforms well-known linear algebra libraries, and it is shown to be superior to condensing with state elimination. Furthermore, we implement an efficient recursive least-squares (RLS) method to provide a linear-time varying model for the adaptive MPC-based RG. No information regarding the topology of the converter nor the structure of the low-level controller is required for such adaptation, making the proposed method self-tuning and eliminating the need for prior identification steps. The proposed control scheme has been tested on various simulated and real dc–dc converters, demonstrating its computational and memory efficiency, as well as its versatility across different converter topologies
Fixed-size LS-SVM LPV System Identification for Large Datasets
Full text linkIn this paper, we propose an efficient method for handling large datasets in linear parameter-varying (LPV) model identification. The method is based on least-squares support vector machine (LS-SVM) identification in the primal space. To make the identification computationally feasible, even for very large datasets, we propose estimating a finite-dimensional feature map. To achieve this, we propose a two-step method to reduce the computational effort. First, we define the training set as a fixed-size subsample of the entire dataset, considering collision entropy for subset selection. The second step involves approximating the feature map through the eigenvalue decomposition of the kernel matrices. This paper considers both autoregressive with exogenous input (ARX) and state-space (SS) model forms. By comparing the problem formulation in the primal and dual spaces in terms of accuracy and computational complexity, the main advantage of the proposed technique is the reduction in space and time complexity during the training stage, making it preferable for handling very large datasets. To validate our proposed primal approach, we apply it to estimate LPV models using provided inputs, outputs, and scheduling signals for two nonlinear benchmarks: the parallel Wiener-Hammerstein system and the Silverbox system. The performances of our proposed approach are compared with the dual LS-SVM approach and the kernel principal component regression
A Kernel-based Learning Approach for Nonlinear MIMO Systems in an Iterative Learning Control Framework
No full textThis paper introduces a kernel-based method for learning feedforward controllers within an Iterative Learning Control (ILC) framework tailored for nonlinear processes. Unlike traditional ILC algorithm that relies on the knowledge of first principle-based models, this approach leverages a data-driven methodology to develop an iterative control update rule using kernel-based training. We compared this method against a traditional ILC scheme and a baseline neural network-based approach. The effectiveness of the proposed method is demonstrated through a unicycle path-following control problem, evaluated across various simulated test scenarios. Performance metrics include vehicle tracking error and ILC convergence speed, confirming the effectiveness of the proposed data-driven approach
Computationally efficient model predictive control for a class of linear parameter-varying systems
No full textThe use of linear parameter varying (LPV) prediction models has been proven to be an effective solution to develop model predictive control (MPC) algorithms for linear and non-linear systems. However, the computational effort is a crucial issue for LPV-MPC, which has severely limited its application especially in embedded control. Indeed, for dynamical systems of dimension commonly found in embedded applications, the time needed to form the quadratic programming (QP) problem at each time step, can be substantially higher than the average time to solve it, making the approach infeasible in many control boards. This study presents an algorithm that drastically reduces this computational complexity for a particular class of LPV systems. They show that when the input matrix is right-invertible, the rebuild phase of the QP problem can be accelerated by means of a coordinate transformation which approximates the original formulation. Then they introduce a variant of the algorithm, able to further reduce this time, at the cost of a slightly increased sub-optimality. The presented results on vehicle dynamics and electrical motor control confirm the effectiveness of the two novel methods, especially in those applications where computational load is a key indicator for success. © 2018 The Institution of Engineering and Technology. All rights reserved
Model predictive control for the reference regulation of current mode controlled DC-DC converters
No full text- …
