检索结果-内蒙古大学图书馆

Adaptive parameterized model predictive control based on reinforcement learning: A synthesis framework

ENGINEERING APPLICATIONS OF ARTIFICIAL INTELLIGENCE 2024年第PartB期136卷

作者： Sun, Dingshan Jamshidnejad, Anahita De Schutter, Bart Delft Univ Technol Delft Ctr Syst & Control Mekelweg 2 NL-2628 CD Delft Netherlands Delft Univ Technol Dept Control & Operat Kluyverweg 1 NL-2629 HS Delft Netherlands

Parameterized model predictive control (PMPC) is one of the many approaches that have been developed to alleviate the high computational requirement of model predictive control (MPC), and it has been shown to significantly reduce the computational complexity while providing comparable control performance with conventional MPC. However, PMPC methods still require a sufficiently accurate model to guarantee the control performance. To deal with model mismatches caused by the changing environment and by disturbances, this paper first proposes a novel framework that uses reinforcement learning (RL) to adapt all components of the PMPC scheme in an online way. More specifically, the novel framework integrates various strategies to adjust different components of PMPC (e.g., objective function, state-feedback control function, optimization settings, and system model), which results in a synthesis framework for RL-based adaptive PMPC. We show that existing adaptive (P)MPC approaches can also be embedded in this synthesis framework. The resulting combined RLPMPC framework provides a solution for an efficient MPC approach that can deal with model mismatches. A case study is performed in which the framework is applied to freeway traffic control. Simulation results show that for the given case study the RL-based adaptive PMPC approach reduces computational complexity by 98% on average compared to conventional MPC while achieving better control performance than the other controllers, in the presence of model mismatches and disturbances.

关键词： Parameterized control Model predictive control learning-based control Deep reinforcement learning Freeway traffic management Synthesis control framework

来源：评论

学校读者我要写书评

暂无评论

Safe and Fast learning-based Model Predictive control of Nonlinear Systems with Applications to Cold Atmospheric Plasmas

Safe and Fast Learning-Based Model Predictive Control of Non...

引用

作者： Bonzanini, Angelo Domenico University of California Berkeley

学位级别：Ph.D., Doctor of Philosophy

Model predictive control (MPC) is a widely-used optimization-based control strategy for the control of multivariable constrained systems. However, the effectiveness of MPC is hindered by computational and modeling challenges, especially when dealing with systems that are hard-to-model, safety-critical, and/or exhibit fast sampling times. This has led to the advent of learning-based MPC (LB-MPC), which combines machine learning (ML) with MPC to address these challenges. Even so, enabling fast LB-MPC with safety guarantees remains an open area of research. Thus, the overarching objective of this dissertation is to develop novel LB-MPC frameworks to enable safe and fast predictive control of hard-to-model dynamical systems subject to uncertainty. This is primarily explored through the lens of real-time control of cold atmospheric plasmas (CAPs) towards automated processing of (bio)materials as well as automated medical applications. Specifically, the objective of this work is broken down into the following sub-aims: (i) control- oriented model learning with uncertainty quantification; (ii) safe control; and (iii) fast control. To learn a dynamical model for the purposes of control, we start from an approximate base (linear) state-space model and capture the plant-model mismatch using Gaussian Process (GP) regression. This is closely related to safe control, as it enables decoupling safety from performance. On the one hand, the GP term corrects the baseline model, while, on the other hand, it yields state-dependent uncertainty bounds in the form of confidence intervals, which can be used to provide safety guarantees; for example, using tube-based or scenario-based approaches. However, solving the MPC problem online, particularly when the GP regression is embedded in the model predictions, constitutes a significant bottleneck in terms of computational cost. In addition, robust approaches such as scenario-trees are notorious for their poor scalability with respect to both

关键词： Chance constraints Cold atmospheric plasmas control systems learning-based control Model predictive control Robust control

来源：评论

学校读者我要写书评

暂无评论

Adaptive Uncertainty Quantification for Scenario-based control Using Meta-learning of Bayesian Neural Networks ⁎

引用

IFAC-PapersOnLine 2024年第28期58卷 486-491页

作者： Yajie Bao Javad Mohammadpour Velni Intelligent Fusion Technology Inc. Germantown MD 20874 USA Clemson University Clemson SC 29634 USA

Scenario-based optimization and control has proven to be an efficient approach to account for system uncertainty. In particular, the performance of scenario-based model predictive control (MPC) schemes depends on the accuracy of uncertainty quantification. However, current learning- and scenario-based MPC (sMPC) approaches employ a single time-invariant probabilistic model (learned offline), which may not accurately describe time-varying uncertainties. Instead, this paper presents a model-agnostic meta-learning (MAML) of Bayesian neural networks (BNN) for adaptive uncertainty quantification that would be subsequently used for adaptive-scenario-tree model predictive control design of nonlinear systems with unknown dynamics to enhance control performance. In particular, the proposed approach learns both a global BNN model and an updating law to refine the BNN model. At each time step, the updating law transforms the global BNN model into more precise local BNN models in real time. The adapted local model is then used to generate scenarios for sMPC design at each time step. A probabilistic safety certificate is incorporated in the scenario generation to ensure that the trajectories of the generated scenarios contain the real trajectory of the system and that all the scenarios adhere to the constraints with a high probability. Experiments using closed-loop simulations of a numerical example demonstrate that the proposed approach can improve the performance of scenario-based MPC compared to using only one BNN model learned offline for all time steps.

关键词： Uncertainty quantification learning-based control scenario-based control model predictive control meta-learning Bayesian neural networks

来源：评论

学校读者我要写书评

暂无评论

learning-based nonlinear model predictive control with accurate uncertainty compensation

引用

NONLINEAR DYNAMICS 2021年第4期104卷 3827-3843页

作者： Xie, Jingjie Zhao, Xiaowei Dong, Hongyang Univ Warwick Sch Engn Coventry CV4 7AL W Midlands England

A learning-based nonlinear model predictive control (LBNMPC) method is proposed in this paper for general nonlinear systems under system uncertainties and subject to state and input constraints. The proposed LBNMPC strategy decouples the robustness and performance requirements by employing an additional learned model and introducing it into the MPC framework along with the nominal model. The nominal model helps to ensure the closed-loop system's safety and stability, and the learned model aims to improve the tracking behaviors. As a core of the learned model construction, an online parameter estimator is designed to deal with system uncertainties. This estimation process effectively evaluates both the current and historical effects of uncertainties, leading to superior estimating performance compared with conventional methods. By constructing an invariant terminal constraint set, we prove that the LBNMPC is recursively feasible and robustly asymptotically stable. Numerical verifications for a two-link manipulator are conducted to validate the effectiveness and robustness of the proposed control scheme.

关键词： Nonlinear model predictive control learning-based control Adaptive control Parameter estimation

来源：评论

学校读者我要写书评

暂无评论

Regret bounds for online-learning-based linear quadratic control under database attacks

引用

AUTOMATICA 2023年第1期151卷

作者： Chekan, Jafar Abbaszadeh Langbort, Cedric Univ Illinois Coordinated Sci Lab Urbana IL 61801 USA Univ Illinois Dept Aerosp Engn Urbana IL USA

This paper is concerned with understanding and countering the effects of database attacks on a learning-based linear quadratic adaptive controller. This attack targets neither sensors nor actuators, but just poisons the learning algorithm and parameter estimator that is part of the regulation scheme. We focus on the adaptive optimal control algorithm introduced by Abbasi-Yadkori and Szepesvari and provide regret analysis in the presence of attacks as well as modifications that mitigate their effects. A core step of this algorithm is the self-regularized on-line least squares estimation, which determines a tight confidence set around the true parameters of the system with high probability. In the absence of malicious data injection, this set provides an appropriate estimate of parameters for the aim of control design. However, in the presence of attack, this confidence set is not reliable anymore. Hence, we first tackle the question of how to adjust the confidence set so that it can compensate for the effect of the poisonous data. Then, we quantify the deleterious effect of this type of attack on the optimality of control policy by bounding regret of the closed-loop system under attack.(c) 2023 Elsevier Ltd. All rights reserved.

关键词： learning-based control Database attacks Regret bound

来源：评论

学校读者我要写书评

暂无评论

Predictive control with learning-based Terminal Costs Using Approximate Value Iteration

引用

IFAC-PapersOnLine 2023年第2期56卷 3874-3879页

作者： Francisco Moreno-Mora Lukas Beckenbach Stefan Streif Technische Universität Chemnitz Chemnitz 09126 Germany Automatic Control and System Dynamics Lab

Stability under model predictive control (MPC) schemes is frequently ensured by terminal ingredients. Employing a (control) Lyapunov function as the terminal cost constitutes a common choice. learning-based methods may be used to construct the terminal cost by relating it to, for instance, an infinite-horizon optimal control problem in which the optimal cost is a Lyapunov function. Value iteration, an approximate dynamic programming (ADP) approach, refers to one particular cost approximation technique. In this work, we merge the results of terminally unconstrained predictive control and approximate value iteration to draw benefits from both fields. A prediction horizon is derived in dependence on different factors, such as approximation-related errors, to render the closed-loop asymptotically stable further allowing a suboptimality estimate in comparison to an infinite-horizon optimal cost. The result extends recent studies on predictive control with ADP-based terminal costs, not requiring a local initial stabilizing controller. We compare this controller in simulation with other terminal cost options to show that the proposed approach leads to a shorter minimal horizon in comparison to previous results.

关键词： model predictive control reinforcement learning learning-based control

来源：评论

学校读者我要写书评

暂无评论

Kernel-based learning for Safe control of Discrete-Time Unknown Systems under Incomplete Observations

Kernel-based Learning for Safe Control of Discrete-Time Unkn...

引用

第43届中国控制会议

作者： Zewen Yang Xiaobing Dai Weijie Yang Bahar Ilgen Aleksandar An?el Georges Hattab Robert Koch Institute Technical University of Munich Taiyuan University of Science and Technology Freie Universit?t Berlin

Safe control for dynamical systems is critical,yet the presence of unknown dynamics poses significant *** this paper,we present a learning-based control approach for tracking control of a class of high-order systems,operating under the constraint of partially observable *** uncertainties inherent within the systems are modeled by kernel ridge regression,leveraging the proposed strategic data acquisition approach with limited state *** achieve accurate trajectory tracking,a state observer that seamlessly integrates with the control law is *** analysis of the guaranteed control performance is conducted using Lyapunov theory due to the deterministic prediction error bound of kernel ridge regression,ensuring the adaptability of the approach in safety-critical *** demonstrate the effectiveness of our proposed approach,numerical simulations are performed,underscoring its contributions to the advancement of control strategies.

关键词： learning-based control kernel ridge regression discrete-time systems high-order systems state observer

来源：评论

学校读者我要写书评

暂无评论

An active energy management distributed formation control for tethered space net robot via cooperative game theory

引用

ACTA ASTRONAUTICA 2025年 227卷 57-66页

作者： Ma, Yifeng Zhang, Yizhai Liu, Ya Huang, Panfeng Zhang, Fan Northwestern Polytech Univ Res Ctr Intelligent Robot Sch Astronaut Xian 710072 Shaanxi Peoples R China Shaanxi Prov Innovat Team Intelligent Robot Techno Xian 710072 Shaanxi Peoples R China Shanghai Jiao Tong Univ Sch Elect Informat & Elect Engn Shanghai 200240 Peoples R China

The current studies for Tethered Space Net Robot (TSNR) typically treat the tension force induced by the net as a disturbance and employ passive suppression for compensation. However, these approaches not only result in excess fuel consumption but also overlook the intrinsic nature of the net dynamics. When one Maneuverable Unit (MU) maneuvers, it generates a tension force on the net that is transmitted to other MUs. This force not only affects the control accuracy of other MUs but also has a positive effect. In this paper, an Active Energy Management Distributed Formation control (AEMC) strategy is proposed to reveal this kind of interaction and maximize its advantage. Firstly, an energy recovery framework is established, allowing each MU can effectively utilize the tension force due to the net. Specifically, a neural network estimator is designed to capture the hysteresis relationship in which MUs influence each other by transmitting forces through the net. Furthermore, to achieve the cooperative completion of tasks, a game based control scheme is proposed to optimize the control input and tension force collectively. Through prediction and optimization, MUs actively manage their impacts on each other, thereby controlling the influence of tension force on the tracking errors of others. Finally, numerical simulations are conducted to showcase the effectiveness of the proposed approach.

关键词： Tethered space net robot Formation control Differential cooperative games learning-based control

来源：评论

学校读者我要写书评

暂无评论

How far back shall we peer? Optimal air handling unit control leveraging extensive past observations

引用

BUILDING AND ENVIRONMENT 2025年 269卷

作者： Li, Rui Zou, Zhengbo Univ British Columbia Civil Engn 6250 Appl Sci Lane Vancouver BC V6T 1Z4 Canada Columbia Univ Civil Engn & Engn Mech 610 SW Mudd New York NY 10027 USA

Heating, Ventilation, and Air Conditioning (HVAC) systems playa critical role in ensuring occupant comfort in buildings. Traditional Rule-based Feedback control (RBFC) systems, while widely deployed for their simplicity, suffer from low adaptability. Recent improvements, such as Model Predictive control (MPC) methods demand complex mathematical modeling and substantial expert knowledge, creating a high barrier for system design and optimization. Reinforcement learning (RL) emerges as a promising solution with its adaptability and model-free nature, albeit challenged by sample inefficiency and suboptimal convergence. Given the intrinsic delayed effects from previous actions and prolonged thermal inertia in the HVAC systems, this study introduces an innovative deep RL framework that can leverage historical observations to refine RL agent performance. By incorporating a state-of-the-art (SOTA) Transformer model, we capture the temporal patterns in HVAC data and build amore precise RL training environment. Implemented on high-resolution, real-world HVAC datasets, our framework showed superior performance in both HVAC system modeling and RL control performance. Specifically, compared to the two baseline models-Bidirectional Long Short-Term Memory (BiLSTM) and vanilla Transformer, our proposed RL environment model achieved an average of 30.5% and 35.8% prediction accuracy improvement, respectively. Moreover, our optimal past observable RL agent swiftly delivers 35.3% of electricity saving and 54.4% of thermal comfort improvement over traditional RBFC. These results show the effectiveness of the pioneering integration of extensive historical observations for HVAC operation optimization.

关键词： learning-based control HVAC optimization Thermal comfort Time-series modeling

来源：评论

学校读者我要写书评

暂无评论

Adaptive Sliding Mode control for Quadrotor UAVs Under Disturbances Using Multi-Layer Perceptron

引用

IEEE ACCESS 2025年 13卷 45518-45526页

作者： Fatemi, Mir Mikael Akbarimajd, Adel Shahed Univ Elect & Elect Engn Dept Tehran *** Iran Univ Mohaghegh Ardabili Dept Elect & Comp Engn Ardebil *** Iran

This paper introduces a novel adaptive sliding mode control (SMC) framework for quadrotor unmanned aerial vehicles (UAVs) operating in the presence of external disturbances and parametric uncertainties. The proposed approach leverages a multi-layer perceptron (MLP) neural network to dynamically tune the SMC parameters in real-time. This synergistic integration of MLP with SMC enables adaptive, efficient, robust, and energy-efficient control, significantly enhancing system performance. By continuously adjusting the SMC controller parameters based on real-time system feedback, the MLP effectively mitigates the impact of external disturbances and parameter uncertainties, achieving optimal hyperparameter values for enhanced trajectory tracking accuracy. The neural network empowers the controller to adapt seamlessly to dynamic changes in system behavior and environmental conditions. Another key contribution of this research lies in the substantial reduction of the chattering phenomenon, a common limitation of traditional SMC systems. Through rigorous simulations, the proposed controller demonstrates superior stability and robustness under diverse external disturbances and dynamic operating conditions. The rapid adaptability of the controller to changing environments surpasses the performance of conventional methods. This work underscores the significant potential of integrating artificial intelligence techniques with classical control methodologies to enhance the resilience and adaptability of UAV control systems in complex and unpredictable scenarios.

关键词： Quadrotors Real-time systems Sliding mode control Uncertainty control systems Artificial neural networks Tuning Mathematical models Accuracy Vehicle dynamics learning-based control sliding mode control adaptive controller optimization robust stability quadrotor UAV MLP artificial neural networks artificial intelligence

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：