基于回声状态网络的船舶摇荡连续预报方法研究

回声状态网络（ ESNS）是一种新型递归神经网络,可通过对有限的已知样本进行训练,建立非线性模型来预报未知样本。该算法在解决非线性问题时具有一定优势。无需知道海浪的先验信息和船舶航行姿态的状态方程,仅利用实测的船舶横摇、纵摇历史数据,寻求规律即可进行实测摇荡数据的极短期预报。仿真结果表明,该算法在预报15 s以内可达到较高的预报精度,通过预报窗口的平移,可以进行连续在线预报。     

Forecasting wind with neural networks

Wind forecasts over a varying period of time are needed for a variety of applications in the coastal and ocean region, like planning of construction and operation-related works as well as prediction of power output from wind turbines located in coastal areas. Such forecasting is currently done by adopting complex atmospheric models or by using statistical time-series analysis. Because occurrence of wind in nature is extremely uncertain no single technique can be entirely satisfactory. This leaves scope for alternative approaches. The present work employs the technique of neural networks in order to forecast daily, weekly as well as monthly wind speeds at two coastal locations in India. Both feed forward as well as recurrent networks are used. They are trained based on past data in an auto-regressive manner using back-propagation and cascade correlation algorithms. A generally satisfactory forecasting as reflected in its higher correlation and lower deviations with actual observations is noted. The neural network forecasting is also found to be more accurate than traditional statistical time-series analysis.     

神经网络模型在港口吞吐量预测中的应用与误差分析

应用神经网络BP算法对杭州港的吞吐量预测实例进行了详细分析。通过对网络各种参数的调试与组合得出,当隐含层节点数为15,训练控制误差为0.035,分级迭代级数为4级,平滑因子参数为0.2,学习速率参数为1.5时,网络性能最佳。将网络预测结果与时间序列和回归分析2种方法进行了比较,得出神经网络方法在短期预测中要优于传统方法。通过对模型预测误差产生原因的简要分析,得出神经网络方法并不适用于吞吐量长期预测。最后对其应用过程中可能存在的一些问题提出了建议。     

Forecasting the Suez Canal traffic: a neural network analysis

Although the Suez Canal is the most important man-made waterway in the world, rivaled perhaps only by the Panama Canal, little research has been done into forecasting its traffic flows. This paper uses both univariate ARIMA (Autoregressive Integrated Moving Average) and Neural network models to forecast the maritime traffic flows in the Suez Canal which are expressed in tons. One of the important strengths of the ARIMA modelling approach is the ability to go beyond the basic univariate model by considering interventions, calendar variations, outliers, or other real aspects of typically observed time series. On the other hand, neural nets have received a great deal of attention over the past few years. They are being used in the areas of prediction and classification, areas where regression models and other related statistical techniques have traditionally been used. The models obtained in this paper provide useful insight into the behaviour of maritime traffic flows since the reopening of the Canal in 1975—following an 8-year closure during the Arab–Israeli wars (1967–1973)—till 1998. The paper also compares the performance of ARIMA models with that of neural networks on an example of a large monthly dataset.     

Forecasting the Suez Canal traffic: a neural network analysis

Although the Suez Canal is the most important man-made waterway in the world, rivaled perhaps only by the Panama Canal, little research has been done into forecasting its traffic flows. This paper uses both univariate ARIMA (Autoregressive Integrated Moving Average) and Neural network models to forecast the maritime traffic flows in the Suez Canal which are expressed in tons. One of the important strengths of the ARIMA modelling approach is the ability to go beyond the basic univariate model by considering interventions, calendar variations, outliers, or other real aspects of typically observed time series. On the other hand, neural nets have received a great deal of attention over the past few years. They are being used in the areas of prediction and classification, areas where regression models and other related statistical techniques have traditionally been used. The models obtained in this paper provide useful insight into the behaviour of maritime traffic flows since the reopening of the Canal in 1975—following an 8-year closure during the Arab-Israeli wars (1967-1973)—till 1998. The paper also compares the performance of ARIMA models with that of neural networks on an example of a large monthly dataset.     

基于聚类的港口吞吐量预测方法及其适用性分析

在统计分析历史数据的基础上,选取港口吞吐量、GDP值等指标,采用SPSS统计分析软件中的层次聚类分析法,将我国具有代表性的港口按照吞吐量增长规律分成平稳增长型、加速增长型和波动增长型3类。然后选择时间序列法、回归分析法、灰色模型理论和神经网络模型法,对不同类型的港口吞吐量预测的适用性进行了理论分析。最后以上海港和镇江港为实例进行计算,并对不同预测方法的适用性进行了验证。     

基于多源信息聚融的港域航道智能监测系统

为通过对航道工程、数据挖掘、时空推理与航海新技术的深入研究,提出了基于时间序列的水文信息分析模型和基于神经网络的气象信息预测模型,实现了航道水文、气象信息预警功能,为管理者决策提供了数据支持。     

船舶柴油机NOx排放特性神经网络预测中的试验设计

研究船舶柴油机NOx排放特性神经网络预测中的学习样本选取试验设计方法。根据用于主机的船舶柴油机可能持续运行范围的工况变化特点,提出采用功率因素变边界的均匀设计法进行试验设计选取样本,并验证了其可行性。研究结果表明,变边界均匀设计法选取的样本用于神经网络训练,预测精度明显高于随机样本选取法。4位级变边界均匀设计法选取的样本训练得到的神经网络模型,NOx排放浓度预测误差小于3.8%,NOx比排放预测误差小4.5%。     

Forecasting container throughput based on wavelet transforms within a decomposition-ensemble methodology: a case study of China

To improve predictive accuracy, new hybrid models are proposed for container throughput forecasting based on wavelet transforms and data characteristic analysis (DCA) within a decomposition-ensemble methodology. Because of the complexity and nonlinearity of the time series of container throughputs at ports, the methodology decomposes the original time series into several components, which are rather simpler sub-sequences. Consequently, difficult forecasting tasks are simplified into a number of relatively easier subtasks. In this way, the proposed hybrid models can improve the accuracy of forecasting significantly. In the methodology, four main steps are involved: data decomposition, component reconstruction based on the DCA, individual prediction for each reconstructed component, and ensemble prediction as the final output. An empirical analysis was conducted for illustration and verification purposes by using time series of container throughputs at three main ports in Bohai Rim, China. The results suggest that the proposed hybrid models are able to forecast better than do other benchmark models. Forecasting may facilitate effective real-time decision making for strategic management and policy drafting. Predictions of container throughput can help port managers make tactical and operational decisions, such as operations planning in ports, the scheduling of port equipment, and route optimization.     

On-line wave prediction

Operational prediction of wave heights is generally made with the help of complex numerical models. This paper presents alternative schemes based on stochastic and neural network approaches. First order auto regressive moving average and auto regressive integrated moving average type of models along with a three-layered feed forward network are considered. The networks are trained using three different algorithms to make sure of the correct training. Predictions over intervals of 3, 6, 12 and 24 h are made at an offshore location in India where 3-hourly wave height data were being observed. Comparison of model predictions with the actual observations showed generally satisfactory performance of the chosen tools. Neural networks made more accurate predictions of wave heights than the time series schemes when shorter intervals of predictions were involved. For long range predictions both the stochastic and neural approaches showed similar performance. Small interval predictions were made more accurately than the large interval ones.     

A short-term adaptive forecasting model for BIFFEX speculation: a Box—Jenkins approach

There are numerous possible advantages to be gained from the accurate prediction of future movements in the Baltic Freight Index (BFI). Because of the difficulties inherent in long-range forecasting, however, the potential for such predictions to provide insight into the future state of the physical dry bulk market is perhaps limited. The greater accuracy of short-term forecasts, on the other hand, facilitates the development of a forecasting model form is justified by the inevitably continuous nature of futures market speculation. Such a model is developed through the application of the Box—Jenkins approach to time series analysis and forecasting. The methodology is presented and the resulting model is evaluated on the basis of objective measures of predictive power and by comparison with alternative forecasting models. Finally, the applicability of the model to the practice of BIFFEX speculation is assessed by judging its performance within a simulated BIFFEX trading environment.     

The question of safety

There are numerous possible advantages to be gained from the accurate prediction of future movements in the Baltic Freight Index (BFI). Because of the difficulties inherent in long-range forecasting, however, the potential for such predictions to provide insight into the future state of the physical dry bulk market is perhaps limited. The greater accuracy of short-term forecasts, on the other hand, facilitates the development of a forecasting model form is justified by the inevitably continuous nature of futures market speculation. Such a model is developed through the application of the Box—Jenkins approach to time series analysis and forecasting. The methodology is presented and the resulting model is evaluated on the basis of objective measures of predictive power and by comparison with alternative forecasting models. Finally, the applicability of the model to the practice of BIFFEX speculation is assessed by judging its performance within a simulated BIFFEX trading environment.     

Error correction of a predictive ocean wave model using local model approximation

Constructing models from time series with nontrivial dynamics is a difficult problem. The classical approach is to build a model from first principles and use it to forecast on the basis of the initial conditions. Unfortunately, this is not always possible. For example, in fluid dynamics, a perfect model in the form of the Navier–Stokes equations exists, but initial conditions and accurate forcing terms are difficult to obtain. In other cases, a good model may not exist. In either case, alternative approaches should be examined. This paper describes an alternative approach of combining observations and numerical model results in order to produce an accurate forecast. The approach is based on application of a method inspired by chaos theory for building nonlinear models from data called Local Models. Embedding theorem based on the time lagged embedded vectors is the basis for the local model. This technique is used for analysis and updating of numerical model output variables to forecast and correct the errors created by numerical model. The local model approximation is a powerful tool in the forecasting of chaotic time series and has been employed for wave prediction in a forecasting horizon from a few hours to 24 h. The efficacy of the local model as an error correction tool (by combining the model predictions with the observations) compared with the predictions of linear auto regressive models has been brought up. In the present study, the parameters driving the local model are optimized using evolutionary algorithms.     

基于组合预测模型的世界海运周转量预测分析

世界海运周转量是衡量未来航运市场运力需求的直接体现,在确定航运市场和船舶市场的发展趋势方面具有关键作用。针对世界海运周转量受到众多复杂因素影响的现实,基于传统的单个预测方法,分别采用时间序列、灰色系统、神经网络方法对世界海运周转量进行预测,然后再对单个预测方法进行加权组合,建立组合预测模型进行海运周转量的预测,预测结果表明:组合预测模型能够得到更加可靠的结果。     

基于BP神经网络的船用柴油机振动状态监测

通过监测柴油机表面振动信号,用时间序列分析方法提取柴油机故障的振动特征参数,以此建立相应的神经网络,用于船用柴油机的状态监测,提高诊断的准确性。试验研究在中速四冲程增压柴油机上进行。文中以柴油机气阀间隙异常的诊断和柴油机负荷状态的识别为例阐述了该方法的实现过程,并给出了振动信号的特征参数与柴油机工作状态之间的关系。研究表明,利用神经网络监测柴油机运行状态的变化是可行的和有效的。     

组合预测模型在上海内河港口吞吐量预测中的应用

建立时间序列和二元线性回归的组合预测模型,对上海内河港口2010年、2015年和2020年的货物吞吐量水平进行了预测。研究发现,组合预测模型相比单个预测方法具有较高的精度,能够较准确地预测上海内河港口货物吞吐量。     

一种新型的船舶横摇运动实时预报方法

本文在阐述强化学习的基本原理，方法的基础上，提出了一种强化学习的TD算法与BP算法相结合的BPTD方法，并将基用于对角回归神经网络的在线训练，最后以在船舶横摇运动时预报技术上的应用为例，说明这种算法有很强的实时多步预报能力。     

江都散货泊位船舶操纵仿真模拟研究

拟建江都散货泊位工程水域附近流态复杂。为保证进出港船舶和靠、离泊作业船舶的安全,需要对拟建工程进行模拟仿真研究。根据实地拍摄视景图片和定位,制作出码头附近的三维视景库,利用当地实测地形资料和相关资料,制作出包括码头设计尺寸、航道的三维地形图和视景图及电子江图,运用水流模拟软件进行模拟和验证,建立了6自由度的设计船舶模型,根据不同风、流组合条件,进行了模拟航行和操纵试验,得到试验数据和模拟研究结论。本次模拟研究,首次使用专业水流模拟软件,结合大型船舶操纵模拟器进行试验,使结果更加具有参考价值和科学依据。     

Forecasting front displacements with a satellite based ocean forecasting (SOFT) system

Relatively long term time series of satellite data are nowadays available. These spatio–temporal time series of satellite observations can be employed to build empirical models, called satellite based ocean forecasting (SOFT) systems, to forecast certain aspects of future ocean states. The forecast skill of SOFT systems predicting the sea surface temperature (SST) at sub-basin spatial scale (from hundreds to thousand kilometres), has been extensively explored in previous works. Thus, these works were mostly focussed on predicting large scale patterns spatially stationary. At spatial scales smaller than sub-basin (from tens to hundred kilometres), spatio–temporal variability is more complex and propagating structures are frequently present. In this case, traditional SOFT systems based on Empirical Orthogonal Function (EOF) decompositions could not be optimal prediction systems. Instead, SOFT systems based on Complex Empirical Orthogonal Functions (CEOFs) are, a priori, better candidates to resolve these cases.In this work we study and compare the performance of an EOF and CEOF based SOFT systems forecasting the SST at weekly time scales of a propagating mesoscale structure. The SOFT system was implemented in an area of the Northern Balearic Sea (Western Mediterranean Sea) where a moving frontal structure is recurrently observed. Predictions from both SOFT systems are compared with observations and with the predictions obtained from persistence models. Results indicate that the implemented SOFT systems are superior in terms of predictability to persistence. No substantial differences have been found between the EOF and CEOF-SOFT systems.     

Wave tranquility studies using neural networks

Information on heights of waves and their distribution around harbor entrances is traditionally obtained from the knowledge of incident wave, seabed and harbor characteristics by using experimental as well as numerical models. This paper presents an alternative to these techniques based on the computational tool of neural networks. Modular networks were developed in order to estimate wave heights in and around a dredged approach channel leading to harbor entrance. The data involved pertained to two harbor locations in India. The training of networks was done using a numerical model, which solved the mild slope equation. Test of the network with several alternative error criteria confirmed capability of the neural network approach to perform the wave tranquility studies. A variety of learning schemes and search routines were employed so as to select the best possible training to the network. Mutual comparison between these showed that the scaled conjugate method was the fastest among all whereas the one step secant scheme was the most memory efficient. The Brent's search and the golden section search routines forming part of the conjugate gradient Fletcher–Reeves update approach of training took the least amount of time to train the network per epoch. Calibration of the neural network with both mean square as well as the sum squared error as performance functions yielded satisfactory results.