论文总字数：32580字

摘 要

水面无人艇（Unmanned Surface Vessel，USV），是一种能够自主航行、控制，并完成任务的水面舰艇。USV相对于传统的舰艇，具有体积小，无人驾驶，反应迅速，灵活，隐蔽，而且续航能力较强等特性，因此广泛应用于自然灾害的监测与救助，争议海域的水文监测，海洋资源的探测，以及无人作战平台的构建等。

USV的核心技术在于其自主性，主要体现在两个方面：一方面具有自主的规划能力，能够根据任务和运行环境规划出一条合理的行进路线，并且能够根据危险和障碍动态调整路线；另一方面就是能够实现航迹控制，使USV能够跟踪到自主规划的航迹上。本文主要研究的内容就是第二个方面，即USV的航迹控制。USV是一种时滞、惯性、非线性以及不确定的运动系统，加上其复杂的运行环境，使其控制变得较为复杂，传统的控制方法很难满足其控制要求。

本文首先建立了地理坐标系和载体坐标系，推导出标量和矢量在两个坐标系之间的转换。分析USV的受力，包括推进器推力和水动力，根据动量定理和动量矩定理得到了USV六自由度数学模型，最后对模型进行了解耦简化。然后介绍了神经网络方面的基本概念，详细整理了神经元和RBF神经网络方面的知识。分析了PID控制方法和神经网络各自的优点与局限性，PID控制具有使用简单、鲁棒性等优点，但是也具有局限性：不适应具有不确定性、非线性、时变、多变量的系统的控制。神经网络具有非线性、容错性、自适应性、并行分布\处理等优点，但同样具有一些局限性：收敛速度很慢、神经网络的结构选取尚无定则、网络计算量较大等。本文基于以上分析，提出了一种将传统控制方法和神经网络结合在一起的控制方法，即基于RBF神经网络整定的PID控制方法，该控制方法既具有传统PID控制的使用简单、鲁棒性强等特点，还具有神经网络的自适应能力，能够根据模型的变化动态地调整控制参数。

在Simulink仿真平台上，针对USV模型的控制，分别使用了经典PID控制和基于RBF神经网络整定的PID控制，对比控制结果，验证了神经网络PID控制具有更好的控制效果。改变USV模型的参数，得到一系列的经典PID控制和神经网络PID控制的响应结果，通过纵向和横向的比较，神经网络PID控制性能优于经典PID控制，经典PID控制的结果受模型参数改变的影响较大，神经网络PID控制受模型参数的影响较小，说明基于RBF神经网络整定的PID控制具有一定的自适应性，能够根据模型参数变化调整网络结构和控制参数。整合USV模型控制，电机模型控制得到USV整体航迹控制，通过分析控制结果，得出了神经网络控制符合航迹控制要求的结论，并实现了USV对直线航迹、椭圆航迹、圆形航迹以及弓形航迹的跟踪控制。

关键词：水面无人艇；六自由度数学模型；PID控制；RBF神经网络整定PID控制；航迹控制

TRACK CONTROL DESIGN AND SIMULATION OF USV

Abstract

Unmanned Surface Vessel(USV), is a capable of surface ships, which can autonomous navigation, control, and complete the tast. USV compared to traditional vessels, with a small, unmanned, responsive, flexible, subtle, and strong endurance capacity and other characteristics, so, it is widely used to monitor natural disasters and relief, the hydrological monitoring of disputed waters, marine resources detection, as well as the construction of unmanned combat platforms.

USV core technology lies in its independence, mainly in two aspects: one with independent planning capacity, to plan out a reasonable route of travel according to the task and the operating environment, and can dynamically adjust the route based on the hazard and barriers; the other aspect is the ability to achieve tracking control, so that USV can track the autonomous planning track. The main content of this paper is the second aspect, namely USV track control. USV is a kind of motion system,which is delay, inertia, nonlinearity and uncertainty, with its complex operating environment, so it^’s control becomes more complex, and the traditional control methods are difficult to meet the control requirements.

Firstly, this paper establish the inertial coordinate system and the carrier coordinate system, and dedude the scalar and vector’s conversion relationship between two coordinate system. Analyzing the force of USV, including the propeller thrust and hydrodynamic, according to the theorem of momentum and moment of momentum theorem, obtain the USV six degrees of freedom mathematical model. Finally, the model was simplified decoupling. Then introduces the basic concepts of neural networks. By analyzing the advantages and limitations of the PID control method and neural network, we get that PID control is simple to use, robustness, etc., but also has limitations: not suited uncertain, nonlinear, time-varying, multivariable control systems. Neural networks has the advantages of nonlinear, fault tolerance, adaptability etc, but also has some limitations. Based on the above analysis, this paper put forward a control method combined with traditional control methods and neural networks, which is tuning PID control method based on RBF neural network, the control method has the advantages of traditional PID control, and also has the adaptive ability of neural networks, the control parameters can be adjusted dynamically according to changes in the model.

On the Simulink simulation platform, aiming to the control of USV model, using the classic PID control and RBF neural network tuning PID control respectively, contrast control results to verify the neural network PID control has better control effect. Changing USV model parameters, we get a series of the control results of classic PID control and neural network PID control,and find than neural network PID control performs better than classic PID control through vertical and horizontal comparison. The model parameters changing has a greater impact on the results of the classic PID control, and a less impact on the results of RBF neural network tuning PID control, proving that RBF neural network tuning PID control has a certain adaptive, can adjust the network structure and control parameters according to the model parameters. USV has achieved the path Tracking of straight track, oval track, circular track and bow track.

Key words: Unmanned Surface Vessel(USV); six degrees of freedom mathematical model;PID control; RBF neural networks PID control; track control.

目录

摘要 I

Abstract II

目录 III

第一章 绪论 1

1.1课题研究背景与意义 1

1.2 USV的研究现状 1

1.3常见控制方法简介 3

1.4 USV航迹控制的研究现状 3

1.5本文研究内容 4

第二章 USV建模 5

2.1 USV的结构与运行环境 5

2.2坐标系和运动参数 5

2.2.1坐标系描述 5

2.2.2坐标系之间的转换 5

2.2.3 USV运动参数定义 6

2.3 USV的受力情况分析 7

2.3.1水动力模型 7

2.3.2静力模型 10

2.3.3螺旋桨推力模型 10

2.3.4 USV整体受力与力矩 12

2.4 USV运动方程 14

2.4.1建立平移运动方程 14

2.4.2建立旋转运动方程 15

2.4.3 USV六自由度空间运动方程 15

2.5 USV六自由度完整动力学模型 15

2.6动力学模型解耦简化 17

2.7本章小结 18

第三章 基于RBF神经网络整定的PID控制 19

3.1神经网络的定义与特性 19

3.2神经元 19

3.3 RBF神经网络 20

3.3.1 RBF神经网络基础 20

3.3.2 RBF神经网络结构 20

3.4传统PID控制的特点 21

3.5基于RBF神经网络整定的PID控制 21

3.5.1 PID控制和神经网络控制各自的局限性 21

3.5.2 RBF神经网络整定的PID控制 22

剩余内容已隐藏，请支付后下载全文，论文总字数：32580字