第二届高性能船舶技术创新国际研讨会

—智能技术和高性能船舶

2022年11月18日

武汉理工大学

武汉理工大学高性能船舶技术学科创新引智基地将于2022年11月18日在武汉举办“第二届高性能船舶技术创新国际研讨会—智能技术和高性能船舶” (线上+线下相结合方式)。本次会议邀请国内外领域专家、学者进行主题发言研讨，大会报告专家来自荷兰丹尔伏特理工大学、英国南安普顿大学、英国布鲁内尔大学，英国思克莱德大学、中国船舶科学研究中心及武汉理工大学等，欢迎大家参会！

会议时间：2022年11月18日16:10 （北京时间）

会议地点：武汉理工大学余家头校区图书馆一楼报告厅

在线会议（Microsoft Teams）: https://teams.live.com/meet/9511013823956 

会议主席：吴有生院士

名誉主席：W. G. Price院士

会议执行主席：刘祖源 教授

会议组织与技术委员会委员：范世东 朱凌 蔡薇 程细得 陈辉 李廷秋

会议主持人：朱凌 教授

[Symposium Notice] 

2^nd International Symposium of Innovating Advanced Ship Technology

--Intelligent Technology and High-Performance Ships

（2022）

Time: 18^th November 2022

Hosted by Wuhan University of Technology

The Advanced Ship Innovation and Talent Introduction Base of Wuhan University of Technology will hold the 2^nd International Symposium of Innovating Advanced Ship Technology – Intelligent Technology and High Performance Ships (2022) in Wuhan on 18^th November 2022. The Symposium will be held online and offline. A number of international and domestic specialists and researchers are invited to give keynote speeches in the Symposium, coming from Delft University of Technology (NETHERLANDS), University of Southampton (UNITED KINGDOM), Brunel University (UNITED KINGDOM), University of Strathclyde (UNITED KINGDOM), Chinese Ship Science Research Center (CHINA) and Wuhan University of Technology (CHINA). On behalf of the symposium organization committee, we warmly invite you to participate in the symposium.

Time: 16:10 (UTC+8, Beijing Time) on 18^th November 2022

Venue of the offline meeting: Library Conference Room, Yujiatou Campus, Wuhan University of Technology

Online via Microsoft Teams – link (https://teams.live.com/meet/9511013823956 )  

Symposium Chairmen: Prof. Yousheng WU

Symposium Honorary chairman: Prof. W.G.PRICE

Symposium Executive Chairman: Prof. Zuyuan LIU

Members of the symposium organization and technology committee: Prof. Ling ZHU, Prof. Shidong FAN, Prof. Wei CAI, Prof. Xide CHENG, Prof. Hui CHEN and Prof. Tingqiu LI

Host: Prof. Ling ZHU

研讨会会议议程

Symposium Program

Speakers：

1. Pandeli Temarel (University of Southampton, UNITED KINGDOM)

Title of the Presentation: Wave-induced Loads: Numerical Modelling vs Measurements   

Abstract: The first part of the presentation provides a summary of hydroelasticity methods for the prediction of global wave-induced loads, 2D and 3D, including whipping effects using selective examples and comparing numerical predictions and measurements – both experimental and full-scale. The second part of the presentation focuses on slamming impact pressure with selective examples on comparison between numerical pressure predictions and experimental and full-scale drop test measurements.

2. Min Gu (China Ship Scientific Research Center, CHINA)

Title of the Presentation:  Research progress on intelligent ship testing and verification

Abstract: The test and verification of intelligent technology and equipment is an important link to support the ship intelligent technology into reality. Building an intelligent ship testing and verification system is an important support for the demonstration application and future commercialization of ship intelligent technology. Aiming at eight kinds of intelligent ship technologies, an intelligent ship testing and verification system that covers model test, digital simulation and full scale test is proposed. Combined with the intelligent technology test ship, the research progress on intelligent ship testing and verification in China Ship Scientific Research Center is introduced.

3. Hui Chen (Wuhan University of Technology, CHINA)  

Title of the Presentation: Fault diagnosis of marine machinery based on the data-driven models

Abstract: In order to realize the intelligent condition monitoring of marine machinery, the machine learning algorithms are introduced to establish data-driven models based on condition-monitoring data for the fault isolation and diagnosis of marine machinery. To solve problems of the class imbalance and limited labeled data, the graph learning and self-surpervised learning are adopted in these data-driven models. Graph learning is used to learn and extract the correlation between adjacent samples in addition to samples’ own features, which provide more topological information for this imbalanced fault diagnosis task. Furthermore, the self-supervised learning model is designed to obtain valuable and robust feature representations from abundant unlabeled samples.  After the representative features are learned by multi-layer convolutional operation, the limited annotation data is used to fine-tune the model to get appropriate layers to determine suitable feature representations for the fault diagnosis.

4. Jeroen Pruyn (Delft University of Technology, NETHERLANDS)

Title of the Presentation: No future for our ships?  

Abstract: Ships and shipping are to a large extent responsible for our wealth and prosperity. They play a crucial role in the globalization and outsourcing of production. As a result, ships are responsible for about 80-90% of all ton*miles (weight of the cargo x distance transported). However, ships are also responsible for around 3-5% of all our GHG emissions and score even worse for Sulphur Oxides (SOx), Nitrous Oxides (NOx), particulate matter (PM) and many other pollutants and carcinogens. The pressure is increasing for shipping to become greener. Technology alone is not enough; new fuels will be needed to achieve the goals set by IMO. There are over a dozen potential fuels, but none currently with sufficient production capacity and for most of them, we are competing with other sectors. This is leading to unprecedented levels of uncertainty for ship owners, as any ship ordered today, will need to change in the future and thus its future value is not known. To address this, it is necessary to address this uncertainty and include it explicitly in the design. In this presentation, we will show how various researchers have addressed this as a first step toward robust or future-ready design.

5. Zhiming Yuan (University of Strathclyde, UNITED KINGDOM)

Title of the Presentation: Hydrodynamics of a fleet of ships in a single file formation  

Abstract: When a body is moving on the free water surface, its resistance components are fundamentally different from those moving in a single media. A distinct Kelvin wave pattern will be observed on the interface between the air and water in the wake of the moving body, accompanied by a wave making resistance to the body, which is one of the major components of the total resistance. Therefore, it is essential to investigate the wave interactions among bodies travelling in highly organized groups. It was found by Yuan et al. (2021) that when a duckling swims at the “sweet spot” behind its mother, a destructive wave interference phenomenon occurs and the wave resistance of the duckling turns positive, pushing the duckling forward. More interestingly, this wave-riding benefit could be sustained by the rest of the ducklings in a single-file line formation. Starting from the 3rd one in a queue, the wave resistance of individuals gradually tended towards zero, and a delicate dynamic equilibrium is achieved. Each individual under that equilibrium acted as a wave passer, passing the waves energy to its trailing one without any energy losses. It inspires the present study to investigate the ships in a single file configuration and to see if we can learn this wave-riding and wave-passing skills from the waterfowl to improve our design of a fleet and to eventually design a water train of minimized wave-making resistance. The objective of this presentation is to quantify how much wave-resistance is increased or reduced by ships in single file formation, comparing to that of a single ship.

6. Tingqiu Li (Wuhan University of Technology, CHINA)  

Title of the Presentation: Rapid Prediction on Hydroelasticity of Marine Structures by Machine Learning

Abstract: The high-precision hydroelasticity modeling technology for the strong nonlinear response of marine structure faces the bottleneck challenges of long time-consuming and low efficiency. Within the framework of the machine learning technique for regression prediction, an adaptive-sequential parallel sampling method (KAPS) and a pointwise ensemble metamodeling method (Ensemble) are developed with the introduction of k-fold cross validation method and combinatorial learning concept. It has been used to predict the two degree-freedom motion of damaged ships, and the input-output function mapping relationship for nonlinear motion of ships has been reconstructed by combined agent model. The research results provide a new idea for the stability evaluation of hydroelasticity response. It has the advantages of minimizing the modeling consumption of the hydroelastic-analysis process, with the superiorities of simple concept and convenient implementation.

7. Xiuhan Chen (Delft University of Technology, NETHERLANDS)

Title of the Presentation: Dredging and sustainable energy: from application to equipment

Abstract: The 7th goal of the UN Sustainable Development Goals calls for affordable and clean energy for the current and future human society. This affects the dredging community from two sides. First, the dredging community should involve itself for developing sustainable energy. Second, future dredging projects should be carried out with sustainable energy. Currently two projects are going on to fulfill these two goals, reservoir dredging from the application side and methanol powered dredging vessels from the equipment side.

Studies conducted by WODA Reservoir Dredging Working Group have shown that reservoir dredging is one of the key solutions to resume the capacity of hydro power generation and water supply for many sick reservoirs around the world. This paper will introduce the joint effect on delivering an international guideline on reservoir dredging. Besides, efforts made by the Dutch academy and industries to promote methanol as the next generation fuel for dredging activities will be introduced in this paper.  

8. Bin Wang (Brunel University, UNITED KINGDOM)

Title of the Presentation: Acoustic emission source location in ship damages using finite element generated delta-T mapping 

Abstract: One of the most significant benefits of Acoustic Emission (AE) testing over other Non-Destructive Evaluation (NDE) techniques lies in its damage location capability over a wide area in complex structures such as ships and other marine flatforms. The delta-T mapping technique developed so far has been shown to enable AE source location to a high level of accuracy in complex structures. However, the time-consuming and laborious data training process of the delta-T mapping technique has prevented this technique from large-scale application on large complex structures. In order to solve this problem, a Finite Element (FE) method was applied to model training data for localization of experimental AE events on a complex plate. Firstly, the FE model was validated through demonstrating consistency between simulated data and the experimental data in the study of Hsu-Nielsen (H-N) sources on simple plate. Then, the FE model with the same parameters was applied to a planar location problem on a complex plate. It has been demonstrated that FE generated delta-T mapping data can achieve a reasonable degree of source location accuracy with an average error of 3.76 mm whilst decreasing the effort required for manually collecting and processing the training data.

9. Andrea Coraddu (Delft University of Technology, NETHERLANDS)

Title of the Presentation: Data science and advanced analytics for shipping energy systems

Abstract: In the last years, Data Science and Advanced Analytics are experiencing a fast process of commodification. This characterization is in the interest of big IT companies, but it correctly reflects the current industrialization of Data Science and Advanced Analytics also in the field of shipping energy systems. Designing technologies from a human-centered perspective means incorporating human-relevant requirements such as safety, fairness, privacy, and interpretability, but also considering broad societal issues such as ethics and legislation. These are essential aspects to foster the acceptance of Data Science and Advanced Analytics technologies in a human-oriented environment and the shipping one, as well as to comply with evolving legislation concerning the impact of digital technologies on ethically and privacy-sensitive matters.

Unsurprisingly, their societal impact is coming to the fore of the public discussion. For this purpose, these technologies are now required to satisfy some additional requirements such as Privacy, Fairness, Safety, Security, Reliability, Interpretability, and Explainability.

The problem of learning from data while preserving the privacy of individual observations has a long history and spans multiple disciplines. One way to preserve privacy is to corrupt the learning procedure with noise without destroying the information we want to extract. Another way is to exploit the data in a federated way, leaving the data in the hand of the data generator, centralizing only aggregated information.

Safety, Security, and Reliability are properties of Data Science and Advanced Analytics necessary to provide robust answers and suggestions. For example, recent Deep Learning algorithms have even surpassed human performances on some well-defined benchmark datasets. It has thus been extremely surprising to discover that such algorithms can be easily fooled by adversarial examples, that are, imperceptible, adversarial perturbations that mislead these systems into perceiving things that are not there. This undermined the safety and security properties of such algorithms, and many stakeholders have shown interest in understanding the risks associated with their misuses to develop proper mitigation strategies and incorporate them into their products.

One of the legal bottlenecks hampering the application of Data Science and Advanced Analytics to real problems in the social domain is the “right to explanation” granted to citizens. Such requirement directly collides with the limitations of many Data Science and Advanced Analytics technologies in terms of interpretability and explainability. These issues have lately come to the forefront of researchers, primarily due to the widespread development and application of Deep Learning methods in systems with societal impact. As they amplify shallow neural networks, it is no surprise that Deep Learning may become an extreme case of black box models, further reducing their interpretability and explainability.

Finally, for shipping energy systems, while Descriptive, Diagnostic, and Predictive analytics are quite exploited in research and practice, fewer examples of Prescriptive Analytics can be found. Prescriptive Analytics is the effort to fully automatize the process of taking decisions and actions starting from the data about the problem with no human intervention making specific processes (e.g., maintenance or fuel optimization) autonomous. On the one side, this process is limited by the specific domain of the shipping energy system, which requires (because of the legislation or because of the contracts) that the final decision should be undertaken by a human operator who takes responsibility for that choice (and therefore Visual Analytics is so important). However, on the other side, there is a technological limitation: Prescriptive Analytics requires the knowledge of multiple aspects of Artificial Intelligence and the presence of multiple data sources, which are not always available. For example, to model constraints and preferences of the operators, it is necessary to exploit data in the form of ontology describing the context (and in the shipping energy systems there is still a large gap in this sense), and it is necessary to exploit data and information which is not structured to achieve practical results. This process requires a significant effort in research to adapt and improve the current tools to the ship energy systems context and a significant investment from the companies in developing the skills required to adopt these tools internally. While for more straightforward analytics, this process started already many years ago, for more advanced analytics, this process is still in its early stages, and more effort is required to fill the current gaps.

10. Adam Sobey (University of Southampton, UNITED KINGDOM)

Title of the Presentation:  Applications of Data-Centric Engineering to reduce emissions in shipping

Abstract: Reducing emissions is of increasing global importance. Within shipping, the International Maritime Organisation are putting pressure on companies to quickly change the manner in which they operate. In the long term there are a number of exciting fuels that will help the industry get to net-zero. However, these alternatives come at a cost and as an industry we can’t wait to start lowering the impact we make on the planet. The use of data and AI provides methods to immediately allow a reduction in fuel use and can help to cut the cost of new green fuels in the longer time.

To illustrate the immediate benefits of a Data-Centric approach, 2 projects will be outlined. One on power prediction for ships using Machine Learning. In this research the difficulty in fitting to the expected mean is illustrated in such a stochastic environment. The resulting research has been commercialised as JAWS and provides an 8% reduction in emissions. A second solution is the optimisation of a ship's route. The most recent advances from the Evolutionary Computation field have not been benchmarked on this problem, especially the co-evolutionary algorithms that provide the widest diversity of search. The results show that the top performing algorithms have the potential to reduce in fuel usage, 7.6% on average over the state of the art, and voyage times, 8.4% on average over the state of the art; this has been commercialized at T-VOS.

11. S. A. Miedema (Delft University ofTechnology, NETHERLANDS)

Title of the Presentation: Improving Dredging Efficiency, a dream or reality. 

Abstract: In dredging a number of power intensive processes are present. The main processes are cutting soil, jetting with water and slurry transport by means of centrifugal pumps. The question is, can these processes be carried out more efficient significantly.

Cutting in sand requires a lot of energy and in dragheads is replaced by waterjets. Without waterjets the cutting power has to be generated by the propulsion system with low efficiency. With waterjets the required power is reduced significantly. In cutterheads the use of waterjets could improve the efficiency in sand, but most probably not in clay and rock. Horizontal transport by means of slurry transport is already the most efficient and can hardly be improved.

To make dredging ready for the future, meaning reducing pollution and CO2, there is not too much to gain by improving the efficiency, although everything helps. So the solution is to change from fossile fuels to alternative fuels or full electric. CSD’s could operate full electric with electricity from nuclear powerplants. Clammshells and backhoes also. For TSHD’s alternative fuels should be the solution.

12. Qiyu Liang (Wuhan University of Technology, CHINA)

Title of the Presentation: Springback prediction through deep learning in ship plate forming  

Abstract: One important procedure in shipbuilding is to manufacture strips and plates of desired shapes before welding into hulls and segments. Wang et al. (2010) developed a multi-square punch forming (MSPF) machine (SKWB-400) that can adjust the height and the rotation of the punches to set different mold shapes, reducing high cost of rigid tools due to the uniqueness of hull plates. A threshold factor for the quality of MSPF is the springback of the work piece, and how to predict the springback accurately and efficiently and design the mold are major issues in hull plate forming. In MSPF process, the upper punches provide normally follower loads, which is quite different from traditional tools and introduces superior difficulty in springback prediction. There are generally theoretical interpretations and numerical simulations to solve the problem, and rapid development of deep learning algorithms encourage us to have a preliminary attempt on this topic. This presentation is to reveal the forming mechanism of the strip in MSPF, discuss dominant parameters that influence the springback, forecast the springback through deep learning algorithms and explore extra capability of deep learning. In comparison with conventional approaches to the springback prediction, the deep learning algorithm is more accurate and less time-consuming, thus improving the efficiency of the hull plate forming process. Besides, deep learning makes the mold design possible according to the target shape, so one-step or few-steps forming in MSPF can be achieved, facilitating the intelligent development in shipbuilding.