摘 要

机械运转会产生振动，而振动会产生噪声，噪声具有能量，能被声呐技术检测，极大的影响潜艇的隐蔽性。因此，隔振技术已经成为降低船舶振动和噪声的关键技术。隔振分为被动隔振和主动隔振。被动隔振结构简单，非常稳定；主动隔振更加精确，对复杂的干扰环境具有良好的适应能力，同时主动隔振具有良好的低频隔振能力，正适用于船舶的低频运行环境。主动隔振与浮筏被动隔振相结合能更好的减小潜艇的振动，进而提高潜艇的隐身性能。因此，本文以带电磁作动器的浮筏隔振系统为研究对象，研究自适应前馈控制系统的控制规律，并通过Simulink完成对基于滤波x-LMS算法的SISO以及MIMO仿真，对隔振成果加以验证。

首先，研究模型辨识法的分类、步骤及最小二乘法的原理。针对本文要求，将数据采集范围定在0Hz到80Hz之间，辨识模型与实验模型拟合度均达90%以上，并且极点均分布在单位圆之中，验证结果表明辨识得到的四条通道模型具有高拟合度且均具有稳定性，能够用于下面的控制器设计及仿真实验。

其次，研究前馈控制的基本原理。接下来介绍了最简单的自适应滤波器FIR，它的结构简单，适应性强，注重介绍了FIR滤波器的工作原理。使用瞬时误差平方与梯度下降法能简化滤波系数更新的计算量，由此得到的新的迭代公式就是LMS算法的体现。将LMS算法应用到工程实际中后，对LMS算法形式进行演化修改后得到的算法就是x-LMS算法。x-LMS与LMS本质上原理是相同的，区别在于更新滤波系数的输入信号从x(k)变成了fx(k)。最后进行了基于x-LMS的SISO自适应仿真，通过对比干扰特性不同的两个干扰信号，得出结论x-LMS算法既能保证收敛性、稳定性，也能保证自适应性。

最后，对SISO于MIMO两种情况分别用基于滤波x-LMS算法进行主动控制，并通过Simulink进行仿真，重点介绍了MIMO情况下设计Simulink的基本原理。仿真结果显示，误差信号幅值控制后快速接近于0，且具有稳定性，同时在更换干扰信号后，自适应滤波器控制器仍能将新的误差信号取得较好的控制效果。

关键词：主动浮筏隔振，系统辨识法，自适应滤波前馈控制，x-LMS，Simulink仿真

Abstract

Mechanical operation will generate vibration, and vibration will generate noise. The noise has energy and can be detected by sonar technology, which greatly affects the concealment of the submarine. Therefore, vibration isolation technology has become a key technology for reducing ship vibration and noise. Vibration isolation is divided into passive vibration isolation and active vibration isolation. The passive vibration isolation structure is simple and very stable; the active vibration isolation is more accurate and has good adaptability to complex interference environments. At the same time, the active vibration isolation has good low frequency vibration isolation capability, which is suitable for the low frequency operation environment of ships. The combination of active vibration isolation and floating raft passive vibration isolation can better reduce the submarine's vibration, thereby improving the submarine's stealth performance. Therefore, this paper takes the floating raft vibration isolation system with electromagnetic actuator as the research object, studies the control law of the adaptive feedforward control system, and completes the simulation of SISO and MIMO based on the filtered x-LMS algorithm through Simulink The results are verified.

First, study the classification and steps of model identification and the principle of least square method. According to the requirements of this paper, the data collection range is set between 0Hz and 80Hz, the fitting between the identification model and the experimental model is more than 90%, and the poles are distributed in the unit circle. The verification results show that the four channel models obtained by the identification have High fitting and stable, can be used in the following controller design and simulation experiments.

Second, study the basic principles of feedforward control. Next, the simplest adaptive filter FIR is introduced, which has a simple structure and strong adaptability, and focuses on introducing the working principle of the FIR filter. The use of instantaneous error square and gradient descent method can simplify the calculation of filter coefficient update, and the new iteration formula obtained from this is the embodiment of LMS algorithm. After applying the LMS algorithm to engineering practice, the algorithm obtained after the evolutionary modification of the LMS algorithm form is the x-LMS algorithm. The principle of x-LMS and LMS is essentially the same, the difference is that the input signal for updating the filter coefficient is changed from x (k) to fx (k). Finally, a SISO adaptive simulation based on x-LMS is carried out. By comparing two interfering signals with different interference characteristics, it is concluded that the x-LMS algorithm can not only ensure convergence, stability, but also adaptability.

Finally, for the two cases of SISO and MIMO, they are actively controlled by filtering x-LMS algorithm, and simulated by Simulink. The basic principle of designing Simulink under MIMO is mainly introduced. The simulation results show that the amplitude of the error signal quickly approaches zero after control, and it has stability. At the same time, after replacing the interference signal, the adaptive filter controller can still obtain a better control effect for the new error signal.

Key words: Active floating raft vibration isolation, system identification method, adaptive filter feedforward control, x-LMS, Simulink simulation

目录

摘要 I

Abstract II

目录 IV

第1章绪论 1

1.1论文的研究背景 1

1.2研究的目的、意义 2

1.3国内外发展现状 2

1.3.1主动浮筏隔振的研究现状 2

1.3.2自适应控制的研究现状 4

1.4论文的主要研究内容 5

第2章浮筏隔振系统建模 6

2.1建模方法 6

2.1.2最小二乘法原理 6

2.2浮筏隔振系统介绍 7

2.2.1电磁作动器 8

2.2.2辨识实验装置 9

2.3模型辨识实验 10

2.3.1实验流程 10

2.3.2实验结果 11

2.3.3模型验证 15

2.4本章小结 17

第3章自适应滤波前馈控制原理 18

3.1前馈控制基本原理 18

3.2FIR滤波器（Finite Impulse Response） 20

3.3最小均方（LMS）算法 23

3.4单通道基于滤波x-LMS 23

3.5 SISO基于滤波 x­LMS控制仿真 26

3.6本章小结 27

第4章基于滤波X-LMS算法的自适应前馈控制的simulink仿真 28

4.1单一振源SISO仿真 28

4.1.1FxLMS算法实际控制时的计算过程 28

4.1.2 Simulink仿真框图 29

4.1.3仿真结果 29

4.2单一振源MIMO仿真 32

4.2.1 FxLMS算法实际控制时的计算过程 32

4.2.2 Simulink仿真框图 34

4.2.3仿真结果 34

4.3本章小结 36

第5章总结与展望 37

5.1总结 37

5.2展望 38

参考文献 39

附录 41

A.筛选相干性大于0.95的数据程序 41

B.拟合辨识模型与实验模型的程序 41

C.SISOx-LMS算法程序 42

D.MIMOx-LMS算法程序 43

致谢 46

第1章绪论

1.1论文的研究背景

在现代化海事战争中，各种先进制导武器的使用率以及命中率不断增加，杀伤力也在越来越高。保障自身安全一直是目前各国海事研究的极其关注并且非常重要的问题，潜艇怎样才能有效隐藏其身影，同时如何精准地识别敌人。其中涉及到的知识就是现代潜艇建造中的声学隐身技术。水下声纳检测是水下探测的重要手段之一。自己的潜艇发出的噪音越少，被敌方声呐搜索和检测到的可能性就越小，同时由于自身干扰噪声减少，己方声呐能探索的范围就越大，发现敌方潜艇的踪影的可能性就越高。船舶噪声主要来源于动力装置的振动，动力装置的振动通过传递会引发船体、管道、基座等等位置的振动，这些振动都能产生噪声。这种类型的振动会传播到机舱中，并在机舱中产生空气噪声，或者直接通过船体结构散发到水中。它的噪声特性以线性频谱为特征，并且处于低频范围内，因此分布广泛，成为海下潜艇的标志性目标信号，严重影响潜艇的隐身能力。

在常规船舶上，船上噪声的主要来源是由柴油机和辅助发动机等机械设备产生的机械噪声。除了提高柴油机和其他动力装置的制造精度来减少机械振动外，还有三种主要方法：一种是合理设计各部分的机械结构以减少自身振动，例如设计低噪音柴油发动机和高性能低噪音发动机；第二是设计屏蔽装置与采用吸声材料用于噪声隔离。第三是使用减震器，例如机械弹簧、橡胶弹簧，以隔离从动力设备通过底座传递到机身的振动。隔振降噪技术在船舶上的应用，不仅可以减少振动源对船舶机械结构的破坏，改善船员的生活和工作环境，还可以降低噪音，减少声音强度的变化并防止船舶被检测到，从而提高其作战能力。