基于BP神经网络整定的蒸汽发生器水位PID控制及仿真研究

将神经网络与PID控制结合起来,提出了基于BP神经网络整定的蒸汽发生器水位PID控制方法,采用BP学习算法调整控制器神经网络的连接权值,通过对系统性能的学习实现了控制器参数的在线整定,仿真结果表明,所设计的控制器具有良好的控制性能。     

深水Spar平台在不规则波中的运动响应

南海水深一般在500m到2000m,属于深水区,Spar平台的适用水深为500m到3000m,是深水开发的经济型平台.对系泊于2000m水深的Classic spar平台在不规则波作用下的运动响应和缆索张力进行了数值模拟.首先通过频域方法得到Spar平台的波浪作用力传递函数、附加质量和辐射阻尼,然后通过快速傅立叶变换将频域结果转换到时域;基于全量的拉格朗日表述和两节点的等参索单元,应用几何非线性有限元方法和Newmark方法计算系泊缆索张力;应用四阶龙哥库塔法计算时域运动方程,得到Spar平台的运动响应和缆索张力.采用JONSWAP波能谱生成不规则波面,考虑了谱峰升高因子不同取值的影响,分析了二阶漂移力、不同缆索初张力和缆索刚度对于系泊系统的影响.     

Experimental study on the nonlinear pressure acting on a high-speed vessel in irregular waves

The present study focuses on the nonlinear behavior of pressure on the hull surface of a high-speed vessel in irregular waves, particularly the pressure responses of alternately wet and dry areas near the waterline and on the bow zone. The vessel has high deadrise angles that may be subject to slight impact and water pile-up effects. A series of experiments in regular and irregular head waves were conducted, and the validity of applying Volterra modeling was investigated. In a previous article using experimental data in regular waves, it was confirmed that the approximate third-order Volterra model adequately simulated the variation of pressure responses in regular waves of different steepness up to a wave amplitude with a wavelength ratio of 0.01, even for the highly nonlinear pressures acting on the abovementioned areas of the hull surface. In this article, further validation for the second part of the study was obtained using experimental data in irregular waves. The frequency response functions obtained from the previous study’s experimental data in regular waves were applied to the third-order Volterra model by combining the input of irregular waves to simulate the responses in irregular waves of sea state five. Then, the spectra and statistics were analyzed. For the motions, accelerations, and pressure responses in irregular waves (as well as for the simulated time histories) variance spectra and statistics such as cumulative distributions of peak values and probability density functions were compared with the experimental results. It was confirmed that even for highly nonlinear and non-Gaussian pressures on the abovementioned areas of the hull surface, the approximate third-order Volterra model simulates the pressure responses in irregular head waves up to a sea state of five with adequate accuracy on deterministic and statistical bases.     

URANS analysis of a broaching event in irregular quartering seas

Ship motions in a high sea state can have adverse effects on controllability, cause loss of stability, and ultimately compromise the survivability of the ship. In a broaching event, the ship losses control, naturally turning broadside to the waves, causing a dangerous situation and possibly capsizing. Classical approaches to study broaching rely on costly experimental programs and/or time-domain potential or system-based simulation codes. In this paper the ability of Reynolds averaged Navier–Stokes (RANS) to simulate a broaching event in irregular waves is demonstrated, and the extensive information available is used to analyze the broaching process. The demonstration nature of this paper is stressed, as opposed to a validated study. Unsteady RANS (URANS) provides a model based on first principles to capture phenomena such as coupling between sway, yaw, and roll, roll damping, effects of complex waves on righting arm, rudders partially out of the water, etc. The computational fluid dynamics (CFD) method uses a single-phase level-set approach to model the free surface, and dynamic overset grids to resolve large-amplitude motions. Before evaluating irregular seas two regular wave cases are demonstrated, one causing broaching and one causing stable surf riding. A sea state 8 is imposed following an irregular Bretschneider spectrum, and an autopilot was implemented to control heading and speed with two different gains for the heading controller. It is concluded that the autopilot causes the ship to be in an adverse dynamic condition at the beginning of the broaching process, and thus is partially responsible for the occurrence of the broaching event.     

不规则波在斜坡地形破碎区波高分布的数值试验研究

在近岸浅水波浪分布的研究中,格鲁霍夫斯基给出适用于深水波至破碎波的整个浅水域(相对水深η>2)的波高经验分布公式,但该式在波浪发生破碎后的适用性研究欠缺。基于FLOW-3D软件对深水不规则波传播到斜坡地形上波浪发生破碎进行模拟,验证波高沿程分布与试验值的一致性,并模拟在130斜坡地形条件下波浪从有限水深传播到近岸破碎区的沿程波高分布变化。结果表明,在该坡度相对水深η<2.75情况下,格鲁霍夫斯基经验分布公式出现较大误差,不再适用;破碎区各累积率波高与平均波高的比值随相对水深变小呈递减趋势,经验公式值与之相比,总体上呈现出大波偏大、小波偏小的情况。     

数值水池短峰不规则波模拟研究

基于粘性数值波浪水池技术,对短峰不规则波进行数值模拟.文中首先改进了长峰不规则波的数值模拟方法,并使用改进后的方法进行了不同海况下长峰不规则波数值模拟,模拟效果明显改善,数值模拟结果与目标值储吻合相当好.之后进行了短峰不规则波的数值模拟,模拟结果与特征值/目标谱也相当接近.     

采用Green-Naghdi方程和有限元法计算浅水船波

采用波动方程／有限元法求解Green-Naghdi(G-N)方程计算船舶在有限水深区域的兴波和波浪阻力。把行驶船舶对水面的扰动作为移动压力直接加在Green-Naghdi方程里，以描述运动船体和水面的相互作用，并经此来计算不面波动、船底水动压力和波浪阻力。G－N方程比浅水方程增加一个非线性的频散项，以补充有限水深对浅水船波的影响。采用随船运动网格的有限方法，以Series 60 CB=0.6船作为算例给出浅水船波的计算结果，并与浅水方程的结果进行了比较。计算结果表明，当船速小于临界速度时，由于频散的影响，G－N方程级出的船后尾波波高比浅水方程的结果大，同时波浪阻力也比浅水方程的结果有所提高。当船速大于临界速度时，G－N方程的计算结果与浅水方程基本相同，频率散射无明显影响。     

某型水下航行器控制方案的半实物仿真

在以ＶＡＸｌａｂ计算机为主机的小型通用动力学半实物仿真系统上对某型水下航行器的运动进行了半实物仿真。给出了半实物仿真系统的基本组成和工作原理，讨论了水下航行器运动的仿真模型和实时积分算法。结果表明，该仿真系统结构合理，精度高，可靠性好，所采用的控制方案可使其满足使用要求，达到了预期的目的。     

移动式水下观测网络试验节点研制

研制了一种低成本的试验节点作为移动式水下观测网络(MUSN)的有效技术验证平台。以一种现有的可在水面航行的小型运载器为基础,引入深度控制系统,得到了一种可以在浅水域完成可控运动的移动试验节点。针对MUSN节点间距较小的特点,提出了可抵偿偏差的移动定位算法,利用水声MODEM的测距信息实现节点在水下移动时较精确的定位。露天水域试验验证了该节点路径跟踪的性能,精度满足应用需要。该节点具备自主航行和节点间通信能力,可以用于MUSN协调控制技术的验证研究。     

Bottom effects on the tower base shear forces and bending moments of a shallow water offshore wind turbine

In this paper the tower base shear forces and bending moments of a shallow water offshore wind turbine have been rigorously calculated by using a nonlinear simulation method taking into account the bottom effects. It has been found that nonlinearly simulated realistic waves with bottom effects should be used as inputs in the stochastic time domain simulation in order to design an un-conservative support structure for the offshore wind turbine. In order to further improve the simulation efficiency, a transformed linear simulation method has been utilized in this paper for generating equivalent waves as those obtained from the nonlinear simulation method. The accuracy and efficiency of the transformed linear simulation method have been convincingly substantiated through the subsequent calculation examples in this article.     

On the dispersion relation of hydroelastic waves in a plate

This paper deals with the dispersion relation of hydroelastic waves in pontoon-type very large floating structures (VLFS) using a simple beam modeling, where the term hydroelastic waves means propagation of deflection vibrations in VLFS. The purpose of this paper is to show the properties of the hydroelastic waves. The dispersion relation of hydroelastic waves propagating in an infinite plate floating on the water is derived based on the linear water wave theory. The effects of the water depth and of the bending rigidity of the floating plate on the wavelength, phase velocity, and group velocity of the hydroelastic waves are shown theoretically or numerically. Then, the dispersion relation of hydroelastic waves in a finite plate floating on shallow water is investigated. It is shown that the wavelength or the phase velocity of the hydroelastic waves varies with the location in the plate. Received for publication on April 7, 1999; accepted on Aug. 20, 1999     

不规则波浪在陡坡上非线性传播变形的试验研究

通过试验研究了不规则波浪在一陡坡(1:10)上的传播过程中的非线性特征。为达到研究目的,在试验水槽中以JONSWAP谱为靶谱生成了两组随机波浪。试验结果显示,在坡前常水深区域和坡顶,两种波况下波高分布均符合瑞利分布;但是在变浅区域两种波况的波高分布却不尽相同。应用基于小波变换的二阶相位谱来分析波浪在传播过程中的非线性相位耦合特征,结果表明：随着水深的变浅,波浪的非线性逐渐增强并且参与非线性相互耦合的频率成份也越来越多。通过二阶相位谱发现：随着水深的减小,参与最强的非线性相位耦合频率向高频处移动。另外通过傅立叶频谱、二阶谱以及二阶频谱共同分析陡坡上两组波况下低频波浪的演化情况。     

浅水条件下船舶通过船闸时的水动力性能数值研究

文章通过应用CFD方法数值模拟在浅水条件下通过船闸的船舶粘性绕流,对船舶通过船闸时的水动力性能进行了数值预报研究。通过UDF编程定义船舶的运动,使用动网格方法和滑移交界面技术进行船舶运动过程中的网格更新,计算作用在船体上的水动力,并由计算得到的水动力求得船体下沉和纵倾。为了验证所采用的数值方法,以一艘通过比利时泽布吕赫Pierre Vandamme船闸的船舶为例,在模型尺度下进行了计算,并将计算结果和佛兰德水利研究所的模型试验基准数据进行了比较。通过分析不同船速、偏心距和水深条件下的数值结果,给出了这些因素对船舶通过船闸时的水动力性能的影响。该文研究结果可为浅水条件下船舶通过船闸时的安全操纵和控制提供一定的指导。     

无人水下航行器控制系统CAN总线通信设计与仿真研究

由于水下环境的复杂性及无人水下航行器自身特点,对其实施实时高效控制需处理大量数据。文章采用CAN总线对UUV控制系统通信网络进行设计和分析,讨论了CAN总线网络数据帧在系统中的具体应用设计和传输参数设置,并使用CANoe仿真软件对所设计的UUV控制系统CAN网络进行了仿真,通过仿真验证了UUV控制系统CAN网络通信设计的正确性。该方法应用于UUV控制系统通信网络设计,将有效提高UUV系统控制性能和特征模拟性能。     

基于卷积神经网络的AUV水下识别算法设计与实现

自主式水下机器人(AUV)是应用于复杂海洋环境中的高智能化无人装备,其需要具备良好的环境感知能力进行自主导航,包括水下目标识别能力。随着人工智能的高速发展,卷积神经网络作为图像处理领域的深度学习架构,在图像特征提取和图像识别上有着强大的性能和卓越的优势。本文利用卷积神经网络,实现了自主式水下机器人水下目标的自主识别。同时,通过采用三段式全连接方式和增加卷积层深度的方式对卷积神经网络进行进一步改进,提高了卷积神经网络的训练速度、准确率和泛化能力。     

某型地空导弹作战模拟训练器的系统设计

某型地空导弹作战模拟训练器采用多微机分布式体系结构,由中央主控单元、搜索指挥单元、发射制导单元构成,通过网络交换设备,以UDP协议实现网络互连。该模拟训练器通过模拟实装面板上的按键和开关,采集学员的操作动作,实现训练评估,故障模拟,目标仿真,雷达仿真,并向学员提供智能操作指导模式,向教员提供航迹与航队编辑模式和故障设置模式,从而全方位模拟实装设备的操作。此外,文章还提出了较为新颖的三维视景的产生与处理方法,航迹与航队编辑方法,以及训练评估算法,对模拟训练器的性能有很大的提升。     

The estimation of wave elevation and wave disturbance caused by the wave orbital motion of a fully submerged hydrofoil craft

The current control system of a fully submerged hydrofoil craft has manual input of fore-foil depth and control mode selection to improve the performance of the control system. However, the manual input needs skillful human operation and observation of waves the encountered to work well over a wide range of waves. In order to use information about the waves encountered in the control system, we considered the estimation of wave elevation and wave disturbance which was caused by the orbital motion of the waves in irregular waves. First, we investigated the wave disturbance by a fully submerged hydrofoil craft, in a state-space model of wave disturbance, and in hydrofoil craft motion, etc. We than considered estimations of the wave elevation and wave disturbance using a shaping filter, a Kalman filter, an autoregressive (AR) model, etc. Finally, we confirmed through simulations that the estimation results and estimation error of wave elevation and wave disturbance were valid.     

Shallow water effects on high order statistics and probability distributions of wave run-ups along FPSO broadside

Ship hydrodynamics in shallow water becomes especially complicated since the nonlinearities in both the incident waves and the wave–hull interactions will be affected by the water depth. For a ship-shaped Floating Production, Storage and Offloading unit (FPSO) operating in shallow water, the broadside often suffers from the wave run-up and green water incidents in non-collinear harsh ocean environments. By applying the methods of ordinary moments and L-moments and the empirical Weibull distribution on the data measured in a series of model experiments, the high order statistics and the exceedance probability distribution of the run-ups along the FPSO broadside are evaluated and the effects of the shallow water depth and the incident environments are analyzed in this paper. It is seen that both the incident waves and the wave run-ups are non-Gaussian in shallow water and that the wave run-up characteristics are significantly influenced by the water depth and the incident environments, while the contribution due to the vessel vertical motions is negligible for the FPSO used in this study. The exceedance probabilities of the wave run-ups show that the broadside will be more likely to suffer from serious wave run-up and green water incidents in shallower water, in a higher incident wave and a non-collinear environment, especially so at locations around the FPSO midship within a range of 3/8L_pp ∼ 5/8L_pp. The dependency of the shape and scale parameters of the wave run-up probability distributions on the locations and the environment is quantified by model tests. The present study leads to the conclusion that the wave run-up characteristics and the shallow water effects should be considered carefully in determining the wave loads and the freeboard of a large FPSO in non-collinear environment conditions.     

基于神经网络的围网渔船横摇运动研究

为了快速预报一艘65 m的围网渔船在任意载况规则波中横摇运动频率响应函数及不规则波中各海况下的横摇角有义值,应用神经网络的原理和算法,以该船7种实际航运载况的耐波性计算数据为训练样本,建立满足精度要求的围网渔船横摇运动神经网络预估模型。并分析吃水、重心高度、航速和浪向对横摇运动的影响。该模型可应用于风浪中船舶航行安全性评估。     

基于神经网络的柴油机遥控系统故障智能诊断研究

为了克服传统模拟电路故障诊断方法的不足,通过对船舶柴油机遥控系统工作原理的分析,提出采用BP神经网络诊断船舶主机遥控系统的智能诊断方法。介绍BP神经网络结构确定方法及其数值优化技术,并以具体电路模块为例探讨神经网络在船舶柴油机遥控系统故障诊断中的应用。通过Matlab仿真可以发现基于BP神经网络的电路故障诊断方法具有自适应性好、训练时间短、准确性高等特点。