基于混沌理论和改进极限学习机的船舶升沉预报

船舶升沉运动预报是主动升沉补偿系统中的重要组成部分.为了满足船舶升沉运动预测的实时性和准确性要求,本文提出了一种混沌理论与增强搜索极限学习机相结合的混合方法(CES-ELM).在混沌动力系统相空间重构的基础上,采用基于误差最小化的方法生成ELM隐藏节点并不断更新权值;利用优化后的模型参数建立船舶运动预测模型.不同海况下的仿真结果表明,该方法的预测平均绝对百分误差小于10％,与传统的ELM和LSSVM模型相比,该模型能有效提高预测精度和鲁棒性.     

改进二阶灰色极限学习机在船舶运动预报中的应用

为提高船舶运动预报的精度,基于海上船舶运动姿态具有灰色特性和周期性振荡特性的特点,提出一种以误差平方和最小为准则的改进二阶灰色极限学习机组合预测模型,对船舶运动姿态进行预报。该方法利用五点三次平滑算法对船舶运动姿态序列进行平滑降噪,采用余弦函数变换构建GM(2,1)预测模型;利用自适应粒子群算法(Adaptive Particle Swarm Optimization,APSO)优化极限学习机权值和阈值参数,对不同模型预测结果进行加权求和,构建改进二阶灰色极限学习机组合预测模型。对2组船模水池试验纵摇时历进行预报,并将其与其他传统的预测方法相比较,结果表明,建立的组合预测模型具有更好的预测精度和泛化能力。     

基于VDM与APSO优化极限学习机的船舶运动姿态预报

为提高船舶在海上运动的耐波性与适航性,并为解决具有非线性、随机性和非平稳性特点的船舶运动姿态难以准确预测的问题,提出运用一种基于变分模态分解和自适应粒子群算法优化极限学习机的组合预测模型。该算法首先利用变分模态分解将船舶运动姿态序列分解为一系列限带内本征模态函数,并且变分模态分解可以避免经验模态分解技术所产生的模态混叠和端点效应,可以降低序列的非平稳性对预测精度的影响;然后对各模态分量分别建立极限学习机预测模型,并用改进的粒子群算法对极限学习机的初始权值和阈值进行优化;最后将各模态分量预测结果进行叠加,得到最终的船舶运动姿态预测值。通过模拟试验测试并与其他传统的预测方法进行比较,结果表明所建立的组合预测模型具有更高的预测精度。     

基于IPESN的船舶升沉运动预报方法

船舶在复杂海况下的升沉运动具有很强的随机性与非线性特征,为提高船舶升沉运动的预报精度,提出基于内在可塑性回声状态网络(IPESN)的船舶升沉运动预报方法.将具有内在可塑性的神经元引入回声状态网络(ESN)的储备池结构,以提高网络对动态系统的映射能力;采用岭回归的方法对IPESN输出连接权值进行学习,以提高网络的泛化能力.将IPESN应用于3级、4级和5级海况下的船舶升沉运动极短期预报,并将结果与传统ESN和径向基函数网络(RBFN)进行对比.结果 表明,IPESN的平均绝对误差分别为0.0028、0.0039和0.0095;均方根误差分别为0.0035、0.0049和0.0117,优于经典ESN与RBFN的预报精度,验证了改进方法的有效性.     

基于混沌理论与RBF神经网络的船舶运动极短期预报研究

文章基于混沌动力系统相空间重构理论,利用关联维数法和最大Lyapunov指数法,对船舶运动时间序列的混沌特性进行了判定。并利用RBF神经网络较强的非线性映射功能,结合相空间重构理论建立了船舶运动极短期直接多步预报模型。实例预报结果表明,所建立的预报模型应用于船舶运动极短期预报取得了令人满意的预报精度,预报时间可达10 s。     

加勒比海近海风能评估和预报（英文）

The exploitation of wind energy is rapidly evolving and is manifested in the ever-expanding global network of offshore wind energy farms.For the Small Island Developing States of the Caribbean Sea(CS),harnessing this mature technology is an important first step in the transition away from fossil fuels.This paper uses buoy and satellite observations of surface wind speed in the CS to estimate wind energy resources over the 2009-2019 11-year period and initiates hour-ahead forecasting using the long short-term memory(LSTM) network.Observations of wind power density(WPD) at the 100-m height showed a mean of approximately 1000 W/m~2 in the Colombia Basin,though this value decreases radially to 600-800 W/m~2 in the central CS to a minimum of approximately 250 W/m~2 at its borders in the Venezuela Basin.The Caribbean LowLevel Jet(CLLJ) is also responsible for the waxing and waning of surface wind speed and as such,resource stability,though stable as estimated through monthly and seasonal coefficients of variation,is naturally governed by CLLJ activity.Using a commercially available offshore wind turbine,wind energy generation at four locations in the CS is estimated.Electricity production is greatest and most stable in the central CS than at either its eastern or western borders.Wind speed forecasts are also found to be more accurate at this location,and though technology currently restricts offshore wind turbines to shallow water,outward migration to and colonization of deeper water is an attractive option for energy exploitation.     

船舶结构设计准则和衡准的研究进展（英文）

围绕船舶与海上设施设计准则和衡准,聚焦IACS、IMO、ISO/TC8五年来的最新研究进展,考虑了IACS各成员国的新规范、指南,以及地区海事结构安全规则的研发,讨论了数值孪生、营运数据、可靠性和风险评估、设计阶段中人为因素等对结构设计准则发展的影响等热点问题,给出了未来海事规则制定的总体方向的考虑。     

基于AR-EMD方法的扩展非平稳船舶运动极短期预报AR模型（英文）

准确的极短期预报技术能够提高对船舶摇荡运动敏感的海洋特种作业安全性和效率。自回归(auto-regressive,AR)预报模型由于其自适应性强、计算效率高而被广泛应用于船舶运动的极短期预报研究。但该模型基于平稳随机假设,因而在非平稳船舶运动的极短期预报中存在困难。针对非平稳船舶运动极短期预报,文章提出一种基于AR-EMD方法的扩展AR模型,称为EMD-AR预报模型。其中,AR-EMD方法是指在经验模态分解(empirical mode decomposition,EMD)的过程中,采用AR预报的方法处理端点效应问题。EMD-AR预报模型将非平稳信号分解成若干平稳的固有模态函数分量及余项,然后对各个分量分别用AR模型预报,得到最终的预报结果,以此克服非平稳性对AR预报模型的影响。研究基于船舶试验数据将EMD-AR模型与线性AR模型、非线性支持向量机回归(support vector regression,SVR)预报模型进行对比分析,结果表明,AR-EMD方法能够有效处理船舶运动非平稳性对AR预报模型的影响,提高该模型的预报精度,且EMD-AR模型预报性能较线性AR模型和非线性SVR模型更优。     

基于神经网络的船舶操纵运动建模研究（英文）

In this paper, Neural Networks(NNs) are used in the modeling of ship maneuvering motion. A nonlinear response model and a linear hydrodynamic model of ship maneuvering motion are also investigated. The maneuverability indices and linear non-dimensional hydrodynamic derivatives in the models are identified by using two-layer feed forward NNs. The stability of parametric estimation is confirmed. Then, the ship maneuvering motion is predicted based on the obtained models. A comparison between the predicted results and the model test results demonstrates the validity of the proposed modeling method.     

基于固定网格小波神经网络的不规则波中船舶横摇运动在线预报（英文）

船舶横摇运动预报对于船舶安全与作业非常重要。本文应用固定网格小波神经网络在线预报不规则波中的船舶横摇运动。该固定网格小波神经网络由离散的小波激活函数组成,其结构和参数可以基于滑动数据窗在线调整;在每一个滑动数据窗,误差下降比判据被用来从小波函数库中选择重要的小波函数项来构建小波神经网络模型,直到该模型可以较好地表达所研究的非线性系统,获得的模型一般比较简洁。预报结果表明,仅仅几个小波函数项就可以很好地捕捉到不规则波中船舶横摇运动的非线性动力学内在特性,这不仅展示了小波函数很强的非线性表达能力,也证实了所采用的建模方法对于预报船舶在不规则波中的横摇运动的有效性。     

基于BP和LSTM组合优化的船舶升沉运动预测

预测船舶升沉运动有助于增强波浪补偿系统的补偿效果,解决补偿系统滞后问题。为提高预测模型的预测精度,提出一种基于误差反向传播（BP）神经网络和长短时记忆（LSTM）神经网络组合优化的船舶升沉运动预测方法。以采用计算流体动力学（CFD）方法获取的船舶在规则波浪作用下的升沉运动和在突发性干扰下的升沉运动为对象,基于PYTORCH框架和LINGO软件,建立以加权方式组合优化BP神经网络和LSTM神经网络的预测模型。研究结果表明,无论是船舶在规则波浪作用下的升沉运动,还是船舶在突发性干扰下的升沉运动,BP-LSTM组合模型的预测精度均高于BP神经网络和LSTM神经网络,有助于提高补偿精度。     

水下船舶清洗技术综述（英文）

This paper presents a comprehensive review and analysis of ship hull cleaning technologies. Various cleaning methods and devices applied to dry-dock cleaning and underwater cleaning are introduced in detail, including rotary brushes, high-pressure and cavitation water jet technology, ultrasonic technology, and laser cleaning technology. The application of underwater robot technology in ship cleaning not only frees divers from engaging in heavy work but also creates safe and efficient industrial products. Damage to the underlying coating of the ship caused by the underwater cleaning operation can be minimized by optimizing the working process of the underwater cleaning robot. With regard to the adhesion technology mainly used in underwater robots, an overview of recent developments in permanent magnet and electromagnetic adhesion, negative pressure force adhesion, thrust force adhesion, and biologically inspired adhesion is provided. Through the analysis and comparison of current underwater robot products, this paper predicts that major changes in the application of artificial intelligence and multirobot cooperation, as well as optimization and combination of various technologies in underwater cleaning robots, could be expected to further lead to breakthroughs in developing next-generation robots for underwater cleaning.     

船舶和离岸结构波浪载荷与运动（英文）

正The special issue on wave loads and motions of ships and offshore structures is the outcome of a workshop on the same topic that was organised in Harbin Engineering University in November 2017 with the objective of bringing together recent work done on the subject area and providing a forum for discussing these results.     

基于动态执行机构的船舶循迹反步积分控制（英文）

文章针对连续曲率路径,用一种简单的几何方法生成连续曲率的路径。基于该几何方法生成的连续路径,文中利用line-of-sight(LOS)引导律解决了循迹控制中横向偏差最小的问题。为了减弱控制输出的振荡和获得平滑的控制输出,一种基于动态执行机构的改进反步积分控制器在过驱动船舶循迹控制中得到了应用。值得注意的是,文中用积分操作来抵抗风浪流环境力。数值分析结果展示了该控制器的有效性。     

用于船舶伴流场预报的一种非几何相似的模型（英文）

The scale effect leads to large discrepancies between the wake fields of model-scale and actual ships, and causes differences in cavitation performance and exciting forces tests in predicting the performance of actual ships. Therefore, when test data from ship models are directly applied to predict the performance of actual ships, test results must be subjected to empirical corrections. This study proposes a method for the reverse design of the hull model. Compared to a geometrically similar hull model, the wake field generated by the modified model is closer to that of an actual ship. A non- geometrically similar model of a Korean Research Institute of Ship and Ocean Engineering(KRISO)'s container ship(KCS) was designed. Numerical simulations were performed using this model, and its results were compared with full-scale calculation results. The deformation method of getting the wake field of full-scale ships by the non-geometrically similar model is applied to the KCS successfully.     

基于改进神经网络的船舶姿态准确预报技术

为了有效解决当前船舶姿态预测准确性问题,结合当前船舶姿态数据特征,改进传统神经网络并以此为基础建立新型船舶姿态预报技术。重构神经网络格式特征区,添加脉冲输出和神经网络数据放大和衰减参数量,构建耦合神经网络作为主要计算网络,结合达尔文进化算法和传统遗传算法特征,构建交叉概率算法,顶替传统经验算法获取放大衰减真实值,通过PC端数据传输和样本导入,实现船舶姿态准确预测。仿真实验数据表明,改进后的神经网络船舶姿态预报技术对于船舶横纵斜度的预测均提高30%以上,达到了提高船舶姿态预测准确度的目标。     

基于改进BP神经网络的船舶操纵性能预报

以某单桨大型船舶在海上的回转性能为例，探讨了应用改进的BP神经网络（Back-pmpagation Neural Network）建立船舶操纵性预报数学模型的方法，并利用matlab语言对其进行了仿真。研究结果表明，改进的BP算法有更快的收敛速度和更好的计算精度。     

基于惯性测量和卡尔曼多步观测器的船舶升沉运动测量方法

为了得到精确的船舶升沉运动信息,解决升沉加速度二次积分后的漂移和相位超前现象,基于卡尔曼滤波算法建立船舶升沉运动的多步观测模型,提出一种基于惯性测量和卡尔曼多步观测器的船舶升沉运动测量方法。首先,升沉平台模拟船舶升沉运动,使用惯性测量元件采集升沉加速度,经过滤波二次积分得到升沉位移。然后,对升沉加速度进行误差分析,运用升沉位移与升沉加速度的关系建立船舶升沉运动的状态空间模型并利用状态转移矩阵进行多步观测,消除升沉位移的漂移和延时。实验结果表明,本文所提出的测量方法可以解决不同幅值船舶运动中的偏移和相位超前问题,测量误差在0.01 m之内,体现了测量的精确性。     

船舶进水应急风险评估DCEFM模型（英文）

This paper proposes a risk assessment model considering danger zone, capsizing time, and evaluation time factors(DCEFM)to quantify the emergency risk of ship inflow and calculate the degree of different factors to the emergency risk of water inflow. The DCEFM model divides the water inflow risk factors into danger zone, capsizing time, and evacuation time factors. The danger zone, capsizing time, and evacuation factors are calculated on the basis of damage stability probability,the numerical sim...     

基于B样条应用的船舶波浪增阻计算（英文）

Making an exact computation of added resistance in sea waves is of high interest due to the economic effects relating to ship design and operation. In this paper, a B-spline based method is developed for computation of added resistance. Based on the potential flow assumption, the velocity potential is computed using Green's formula. The Kochin function is applied to compute added resistance using Maruo's far-field method, the body surface is described by a B-spline curve and potentials and normal derivation of potentials are also described by B-spline basis functions and B-spline derivations. A collocation approach is applied for numerical computation, and integral equations are then evaluated by applying Gauss–Legendre quadrature. Computations are performed for a spheroid and different hull forms; results are validated by a comparison with experimental results. All results obtained with the present method show good agreement with experimental results.