神经网络方法：未知动力学自主水面车辆跟踪控制

77 浏览量更新于2024-09-01 收藏 889KB PDF 举报

"这篇论文提出了一种针对具有完全未知动力学特性的自主水面车辆(ASV)的高效神经网络（NN）跟踪控制方法。该方法适用于受到显著不确定性影响的情况。神经网络采用单层结构，利用车辆回归动态模型，将高度非线性的动力学特性表示为已知和未知动态参数的形式。学习算法简单且计算效率高，源于Lyapunov稳定性理论的推导。" 这篇研究论文探讨了如何利用神经网络技术来解决自主水面车辆在面对未知动力学特性时的跟踪控制问题。传统的控制系统设计通常依赖于对系统模型的精确知识，但在实际应用中，这种知识往往难以获取，尤其是对于环境影响和机械复杂性较高的自主系统。作者提出的方法解决了这个问题，允许ASV在面临重大不确定性的情况下实现有效的跟踪控制。文章的核心是设计了一个单层神经网络，其结构简单但功能强大。这个网络利用车辆回归动态，将复杂的非线性动力学转换为与已知和未知动态参数相关的表达式。这样做可以使得网络能够学习和适应车辆的行为，即使在动态特性未知的情况下也能进行有效控制。神经网络的学习算法是关键。论文指出，这个算法虽然简单，但在计算效率上表现出色。它基于Lyapunov稳定性理论，确保了系统在学习过程中能够保持稳定，避免了由于模型不确定性可能导致的系统震荡或不稳定。Lyapunov稳定性理论是一种广泛应用于控制理论的方法，用于分析系统的长期行为和稳定性。此外，关键词"Autonomous surface vehicles"表明了研究领域是自主水面交通工具，"Unknown dynamics"强调了处理未知动力学的挑战，"Tracking control"指明了目标是实现精确的轨迹跟踪，"Neural networks"和"Lyapunov stability"揭示了所用的技术手段和理论基础。这篇论文为自主水面车辆的控制提供了一种创新方法，通过神经网络学习未知动态，实现了在复杂不确定性环境中的高效跟踪控制，展示了神经网络在解决现实世界控制问题中的潜力和优势。

An efﬁcient neural network approach to tracking control of an autonomous

surface vehicle with unknown dynamics

Chang-Zhong Pan

a,b,c

, Xu-Zhi Lai

a,b,

⇑

, Simon X. Yang

, Min Wu

a,b

School of Information Science and Engineering, Central South University, Changsha, Hunan 410083, China

Hunan Engineering Laboratory for Advanced Control and Intelligent Automation, Changsha, Hunan 410083, China

Advanced Robotics and Intelligent Systems (ARIS) Laboratory, School of Engineering, University of Guelph, Guelph, Ontario, Canada N1G 2W1

article info

Keywords:

Autonomous surface vehicles

Robots

Unknown dynamics

Tracking control

Neural networks

Lyapunov stability

abstract

This paper proposes an efﬁcient neural network (NN) approach to tracking control of an autonomous sur-

face vehicle (ASV) with completely unknown vehicle dynamics and subject to signiﬁcant uncertainties.

The proposed NN has a single-layer structure by utilising the vehicle regressor dynamics that expresses

the highly nonlinear dynamics in terms of the known and unknown dynamic parameters. The learning

algorithm of the NN is simple yet computationally efﬁcient. It is derived from Lyapunov stability analysis,

which guarantees that all the error signals in the control system are uniformly ultimately bounded (UUB).

The proposed NN approach can force the ASV to track the desired trajectory with good control perfor-

mance through the on-line learning of the NN without any off-line learning procedures. In addition,

the proposed controller is capable of compensating bounded unknown disturbances. The effectiveness

and efﬁciency are demonstrated by simulation and comparison studies.

1. Introduction

An autonomous surface vehicle (ASV) is also called an un-

manned surface vessel (USV), which can operate on the surface

of lakes, rivers and oceans. Over the past few years, there has been

renewed interest in the development of ASVs, see Benjamin and

Curcio (2004), Caccia et al. (2007), Desa et al. (2007), Martins,

Almeida, Silva, and Pereira (2006) and Xu, Chudley, and Sutton

(2006) for example. Due to their emerging applications in commer-

cial, civilian and military missions (Størkersen, Kristensen, Indree-

ide, Seim, & Glancy, 1998), the control of ASVs has become an

intensive research area. However, how to control the ASVs efﬁ-

ciently is still very challenging, which might stem from the fact

that (Fossen, 2002): (1) it is hard to use current modelling

techniques to obtain an accurate vehicle dynamic model, which

is generally highly nonlinear, time-varying, and coupled in nature;

and (2) the operation environment of marine vehicles is usually

very complex and unstructured, which brings unpredictable per-

turbations to the control system, such as ocean currents, waves

and wind.

Basic control problems of surface vehicles are classiﬁed into

point stabilisation (Liu, Yu, & Zhu, 2011), way-point manoeuvring

(Fredriksen & Pettersen, 2006), path following (Almeida, Silvestre,

& Pascoal, 2007; Bibuli, Bruzzone, & Caccia, 2009), trajectory track-

ing (Zou, Chen, Feng, & Liu, 2011), and formation control (Peng,

Wang, Chen, Hu, & Lan, 2012). Many control approaches have been

proposed in the literature, such as conventional PID control, non-

linear recursive/adaptive control, sliding-mode control (SMC),

and intelligent control. PID control is very simple and easy to

understand. But due to the complexities of the surface vehicles,

the designed controllers (Escario, Jimenez, & Giron-Sierra, 2012;

Fossen, 2002; Moreira, Fossen, & Guedes Soares, 2007) may cause

poor control performance. In addition, the PID tuning is still a

difﬁcult problem, even though there are only three parameters.

To improve the system control performance, nonlinear control-

lers that usually employ Lyapunov method, recursive backstepping

technique, or adaptive control have been widely proposed. For

example, with the aid of Lyapunov’s direct method, Jiang (2002)

designed two systematic tracking controllers for an underactuated

ship. Based on backstepping technique, Du and Guo (2004) estab-

lished an uncertain nonlinear mathematical model and designed

a nonlinear adaptive controller for the course-tracking control.

Aguiar and Hespanha (2007) presented an adaptive switching

supervisory control combined with a nonlinear Lyapunov-based

tracking control law. The controllers using SMC have low sensitiv-

ity to plant parameter variations and disturbances. They can also

greatly improve the system control performance. For instance,

Cheng, Yi, and Zhao (2007) proposed a multi-variable sliding mode

control law for the trajectory tracking of a ship, where the posi-

tions and yaw angle were simultaneously tracked. Ashraﬁuon

http://dx.doi.org/10.1016/j.eswa.2012.09.008

⇑

Corresponding author at: School of Information Science and Engineering,

Central South University, Changsha, Hunan 410083, China. Tel./fax: +86 731

88836091.

E-mail address: xuzhi@csu.edu.cn (X.-Z. Lai).

Expert Systems with Applications 40 (2013) 1629–1635

Contents lists available at SciVerse ScienceDirect

Expert Systems with Applic ations

journal homepage: www.elsevier.com/locate/eswa

下载后可阅读完整内容，剩余6页未读，立即下载

weixin_38698943

粉丝: 2
资源: 900

神经网络方法：未知动力学自主水面车辆跟踪控制

A Fuzzy Neural Network Approach to Evaluation of Slope Failure Potential

An Extremely Efficient Convolutional Neural Network for Mobile Devices

subsasgn Assign fields of a neural network.

Write pytorch-based Python code to implement a neural network that solves a regression problem with an output layer of a positively weighted sub-network plus a negatively weighted sub-network.

how can i trained a neural network for identify the cat and dog

class NeuralNet(nn.Module): ''' A simple fully-connected deep neural network '''

MATLAB中下载Neural Network Toolbox

matlab中下载Neural Network Toolbox

最新资源