Artificial intelligent technologies have become an important approach to improving the safety of shipping and reducing the operating costs of shipping companies. In order to further improve the level of ship intelligence and break down the data barriers between different shipping companies, an efficient privacy-preserving federated learning method (EPFL) is proposed in this paper.

Federated learning is adopted to organize multiple ship participants to collaboratively train a global fault diagnosis model, and cryptography technologies are used to protect their local data information. Considering Internet of Ships (IoS) scenarios, this paper introduces sparsification technology to compress the model parameters uploaded by shipping participants and reduce their number.

Theoretical analysis and the experimental results show that the proposed EPFL method can effectively reduce the resource consumption of cryptographic computation and data communication while protecting the local data information of ship participants.

This paper aims to explore new methods for enhancing the abnormal data processing of real ships in inland rivers, improving data comprehension and assisting in ship behavior recognition research.

By constructing a navigation logic level, the time series data is divided to obtain the semantic label of the ship behavior. A navigation logic visualization analysis system is designed on the basis of semantic labels, and the navigation status of the ship is combined with data visualization to assist in analyzing data problems and studying ship characteristics. Relying on a digital waterway, the data of working ships with complex behavior in an inland waterway is selected for example-based testing, and the system is used to analyze abnormal data and conduct research on ship behavior.

Through the interactive visualization of navigation logic, the causes and characteristics of abnormal data with position jumping can be effectively determined, thereby enhancing abnormal data processing. In addition, the qualitative analysis of features and quantitative analysis of thresholds effectively divides the berthing and direct sailing status data, further enriching the semantic labels of ship behavior.

Aiming at the replacement of propellers behind surface ships with pumpjet propulsion systems, this paper introduces a novel method for predicting full-scale power performance based on statistical learning.

Pump performance maps originating from the neural network learning of existing pumpjet thrust coefficient maps and matched to a ship's drag line from model tests are used to determine the pumpjet's full-scale power performance behind a large surface ship. To validate its precision and availability, traditional complete model tests including the ship model drag test, pump model open water test and ship-pumpjet self-propulsion test are completed to determine the full-scale benchmark power performance under different ship speeds.

The prediction errors of the pumpjet's rotation speed, thrust and power under different self-propulsion ship speeds from 18 knots to the design point of 30 knots are smaller than 5.4%, with no more than 2% from the design condition. As for the ship-propulsor interaction amplitude, the surface ship-pumpjet subsystem lies between ship-propeller interaction and ship-waterjet pump interaction with a thrust deduction coefficient approaching zero. From this point of view, the pumpjet propulsion system behind a surface ship can be recognized as a transitional stage from the propeller-shaft configuration to the waterjet propulsion system.

In order to improve the intelligent level of ship engine room operation and maintenance (O&M), a technical route is developed according to the characteristics of the engine room equipment, such as variety, quantity and complex coupling relationship, and propose a ship-shore integrated intelligent O&M system for ship engine room.

By analyzing the technical development trend of the intelligent O&M system of an engine room at home and abroad, an intelligent O&M system that conforms to the characteristics of an engine room is designed by integrating the technical advantages of health management, big data mining, digital twinning, lightweight data transmission and so on.

A data-brain enabled functional framework, platform design, operation system, operation process and key technology application of an engine room intelligent O&M system are designed.

To address the problem of the fault diagnosis of ship propulsion shafting bearings, this paper proposes a visual diagnosis method based on a holographic symmetrical dot pattern (SDP) and similarity recognition.

First, the bearing vibration signals are collected in three directions to comprehensively monitor the non-stationary changes in the time-domain and frequency-domain caused by the regular impact of the bearing faults. Second, based on SDP, multiple one-dimensional time-domain signals and spectrums are merged into a two-dimensional image to amplify the difference between different state signals. Finally, a simple bearing diagnosis is performed based on the similarity recognition method.

The results of engineering experiments show that this method can achieve the effective graphic fusion of multiple signals, fully display the characteristics of equipment signals and accurately diagnose faults.

A ship's system is a complex mechanism composed of multiple pieces of equipment. Due to the dynamic and non-linear characteristics of the parameters of each component, fault diagnosis is complicated. This paper proposes a dynamic feature fusion method for performing efficient fault diagnosis on the system.

Fractal theory, dynamic theory and the kernel principal component analysis (KPCA) method are used to reconstruct, map and filter the system state data, and obtain the principal component characteristic data matrix, square prediction error (SPE) and corresponding control limits. An offline monitoring model based on the health data of a marine diesel engine intake and exhaust system is then constructed and used to diagnose and analyze ship system faults. In order to verify the validity of the model, the fault data of the intake and exhaust system of a marine diesel engine is selected for verification and analysis.

The results show that this method can effectively realize the accurate analysis of a system's dynamic nonlinear state data and efficient analysis and diagnosis of faults, with better fault diagnosis performance than the KPCA and support vector machine (SVM) method.

In order to overcome the disadvantages of the traditional ensemble empirical mode decomposition (EEMD) method in selecting parameters (integration time and white noise amplitude coefficient) based on experience, and reduce the cost of calculation time, a fast ensemble empirical mode decomposition (FEEMD) method is used to extract the characteristic frequency.

By changing the distribution density of the added white noise, different signal envelopes can be obtained. Furthermore, we can identify the optimal envelope by finding the optimal search window width of the moving mean filter, thereby avoiding the defect of EEMD selecting parameters by experience. At the same time, after the abnormal component in the signal is decomposed, the residual component can be decomposed by EMD to further save the calculation time cost. Finally, the method is combined with Hilbert envelope demodulation technology and applied to the fault characteristic frequency diagnosis of the bearing inner ring of an asynchronous motor.

As the results show, compared with the traditional EEMD method, FEEMD can extract the fault frequency more efficiently.

With reference to the definitions and requirements of intelligent engine rooms in the China Classification Society Rules for Intelligent Ships, this paper studies methods for predicting the remaining useful life (RUL) of bearings in order to explore prognostic and health management technologies.

Addressing the poor prediction accuracy of conventional data-driven methods, this study uses the Stacking fusion strategy in integrated learning to construct an R-A-X (Ridge-ANN-XGBoost, with XGBoost and ANN as the base learner, and ridge regression as the meta learner) fusion model. It then designs a prediction performance comparison experiment using the life cycle data in the IEEE PHM 2012 Prognostic Challenge under the same working conditions, with MAE and R² used as performance evaluation indicators to compare the R-A-X fusion model with the single algorithm and average.

The results show that the prediction performance of the R-A-X fusion model are better than those of the other methods involved in this article, with an improvement effect reaching up to 20%.

In view of the difficulty of collecting the full life cycle performance degradation data of a marine diesel turbocharger, a life prediction model based on the Weiner process is proposed to predict the residual life of the turbocharger.

First, the K-Means model is used to cluster the actual operating conditions of the turbocharger and extract the typical operating condition data, then the Bayesian change-point detection model is used to identify the defect points of the turbocharger, and finally, the residual life of a certain type of turbocharger is predicted by the Weiner process degradation model.

The results prove that the proposed method can predict the residual lifetime of the turbocharger without the historical degradation data of similar equipment.

This paper puts forward a method for monitoring and evaluating the lubrication performance of a marine stern tube bearing which combines a lubrication performance decay model and support vector machine (SVM) algorithm.

Aiming at difficulties in the monitoring and recognition of the lubrication regimes of stern bearings, a bearing lubrication decay numerical model is established and validated with experimental data. The effects of load, roughness and radius clearance on the lubrication decay mechanism are then investigated. Based on the SVM algorithm, a lubrication regime classifier is constructed; the hyperparameters are optimized through a grid search algorithm; the datasets of different lubrication regimes are used for training; and lubrication regimes for stern bearings are evaluated.

The results show that with the increase of external load, roughness and radius clearance, the critical speed of the deterioration of the bearing lubrication regime increases, the working range of hydrodynamic lubrication (HL) decreases and the working range of mixed lubrication（ML） increases. The lubrication regime recognition model is then verified by the simulation dataset, and the proposed lubrication regime recognition method has an accuracy rate of 96.88%.

The propulsion shafting system is an important part of a ship, and the bearing load directly affects its operating state and service life. In this paper, bearing load under hull deformation is studied using grey system theory.

First, according to the empirical formula of the relative displacement of each bearing caused by the deformation of the hull of a 57 000 DWT oil tanker, the relative displacement of each bearing is calculated and input into a finite element model, and the load value of each bearing is output. On this basis, grey relationship analysis of grey system theory is introduced to study the influence degree of stern bearing displacement on the load of each bearing, and the relative change of the load of each bearing caused by the displacement of the stern bearings is analyzed. A GM (1,1) prediction model is then established for the bearing load considering the bearing displacement conditions, and the hull deformation-fitting and prediction of each bearing load are made.

The results show that grey relationship analysis can effectively reflect the influence of hull deformation on bearing load. The GM (1,1) prediction model has high accuracy and prediction errors less than 6.0%, and the model test indexes can represent the accuracy of the prediction.

In order to realize the precise control of a certain type of electro-hydraulic position servo system, a feedback-feedforward variable gain iterative learning method is proposed.

First, the simplified model of the electro-hydraulic position servo system is established, then the iterative learning control algorithm is improved and a variable gain learning law with a forgetting factor is adopted. Finally, Matlab simulation is carried out for verification.

The simulation results show that compared with traditional iterative learning and PID control, the improved iterative learning algorithm has better convergence and smaller tracking error, and can make the electro-hydraulic position servo system track the position curve quickly and accurately with improved dynamic characteristics.

In order to improve the energy efficiency of ships, this paper studies the energy efficiency optimization method of fuel cell/lithium battery hybrid ships under the influence of multiple factors.

Based on MATLAB/Simulink simulation modeling software, simulation models of the fuel cell, lithium battery, DC-DC converter, ship hydrostatic resistance, ship navigation environmental resistance, ship motion and ship propeller are built respectively. On this basis, an energy efficiency simulation model of the target ship is established, and the influence of navigational elements such as wind speed, wind direction, water depth and wave height on the energy efficiency of the ship is studied. Aiming at the target ship's hybrid system structure, a power allocation strategy based on fuzzy logic is used to optimize the system's energy flow. Taking the minimum total energy consumption of the system as the optimization goal and considering multiple internal and external constraints, a non-linear model of ship speed optimization in light of multiple factors is established. The intelligent optimization algorithm of the whale population is used to carry out the dynamic optimization of the non-linear ship speed model, and a comparative analysis of energy efficiency optimization under different navigation methods is performed.

The results show that under the condition that the total voyage time remains unchanged, the proposed multi-factor optimization method of ship energy efficiency can reduce total energy consumption by 5.04% and total hydrogen fuel cell consumption by 13.16%.

In order to improve the fault diagnosis level of marine power systems, this paper studies the real-time fault diagnosis of a marine supercharged boiler based on a convolutional neural network (CNN).

First, the simulation program of the marine supercharged boiler is developed based on the GSE platform, and the simulation fault data is obtained. The fault diagnosis model of the boiler is then established using the CNN method. Next, through the change trends of temperature, flow and other parameters, combined with a priori knowledge and the machine learning method, fault identification is carried out. Lastly, the performance of the method is evaluated against criteria such as confusion matrix and accuracy.

According to the comparison results between the feature extracted dataset and the original dataset, the stability of the model output results and the generalization ability of the model are optimized and improved, with an overall fault classification accuracy reaching 99.53%.

To simplify the ship modeling process and improve modeling efficiency, this paper proposes a method for rapidly implementing hull model creation based on hull line drawing.

Using the CATIA platform, this method adopts the component application architecture (CAA) development tool for secondary development. First, by reading the geometry elements and label information of the hull lines in a drawing, the transformation of the offset points from 2D to 3D is realized. On this basis, the creation of the hull lines, stern and bow is completed, and a 3D wireframe model obtained. Finally, the hull 3D modeling is completed in conjunction with the CATIA native surface creation command. The stability and reliability verification of the developed type value extraction and bow generation program is then carried out via application analysis.

The results show that the compiled program can realize the automatic creation of offset points, transverse lines, waterlines, profile lines and other boundary lines except the top line of the wall, and the centerline and tangent lines created by the bow generator are easily modified and simple for users to operate.

Warship damage flooding is an interactive process between progressive flooding and counter-flooding measures under battle-damaged conditions. Traditional damage stability theory cannot simulate the influence of counter-flooding measures on the damaged warship flooding process, and there is no method for predicting this process in the full-time domain.

By analyzing a coupling relationship between the damage flooding process and counter-flooding measures, a flooding process time domain model is established and a numerical solving algorithm obtained. Case studies are analyzed on the basis of the model and algorithm, and an interactive visual simulation system is developed.

The simulation systemcan comprehensively assess the flooding process and various counter-flooding measures, providing an important supplement for warship damage stability theory.

In order to obtain the plane magnetic field data at different heights above a ship and solve the problem of calculating the plane magnetic field from the vaulted magnetic field above the ship, this paper proposes a method for calculating the magnetic field above the ship using the equivalent surface magnetic charge.

Based on the theory of equivalent source, the magnetic field data of the vault is used for inversion modeling. As the ship's magnetic field is equivalent to that generated by the surface magnetic charge, it is calculated accordingly to simulate the magnetic field above the ship. The magnetic field calculated by the surface magnetic charge method is then compared with the ship's actual magnetic field to verify the feasibility of the surface magnetic charge method.

Through numerical simulation, the relative root mean square error of the vertical component of the ship's magnetic field calculation is 0.44%. Using the magnetic field ship model, the relative root mean square error of the vertical component calculation is 7.64%.

In this work, the influence of the initial heeling angle on the roll motion response of a ship in random beam wave is studied.

Specifically, the path integration method is applied to solve the Fokker-Planck equation, which governs the probability properties of the stochastic differential equation for the roll motion. The probability distributions of the roll motion response can then be obtained.

The results show that the initial heeling angle has limited influence on the roll motion response spectrum, but the probability density function of the roll motion response and distribution of extreme roll motion response can be significantly influenced by the heeling angle. As a result, the safety and stability of the ship will deteriorate dramatically.

A string-type mooring system is proposed for the permanent berthing demands of special offshore platforms. So it is necessary to study the hydrodynamic performance of the berthing platform using this mooring system to verify its adaptability to the permanent berthing demands.

Based on the three-dimensional potential flow theory, a numerical model of the platform berthed alongside a terminal wharf is established using the boundary element method, and the frequency domain numerical simulation of the moored platform at different tide levels is carried out. Two mooring systems, i.e., string-type and traditional dock mooring, are used to simulate the hydrodynamic performance of the platform in the time domain under extreme wave conditions, and parameterized analysis is conducted to analyze the influence of different tide levels on the motion response of the platform.

By providing omni-directional restoring force, the string-type mooring system strengthens the rotational motion limitations of the platform, greatly reduce the most violent rolling motion, transfers the energy of environmental load to translational motion, and further strengthens the restriction of the overall motion of the platform; the motion range of the platform is balanced, stable and controllable; and the effects of the system on the motion limit of the platform are less affected by the change in tide level.

In order to solve the problems of safety and smoothness in the path planning of an unmanned surface vehicle (USV), a path planning method with a controllable distance from obstacles is proposed.

First, the raster environment information is generated in combination with the radar image, and the Voronoi field algorithm (VFA) is used to add the danger potential field to each grid and establish the navigation boundary; second, the risk function associated with the navigation boundary is established to improve the evaluation function of the A-star algorithm, and the improved A-star algorithm is used for path planning; finally, for the problem of the large course altering of the navigation path, the gradient descent method (GDM) is used to plot a continuous smooth navigation path that satisfies the actual navigation requirements of the USV.

The simulation results show that the proposed path planning method can control the distance between the path and obstacles by setting different navigation boundaries, and the smoothness meets the navigation requirements.

In order to investigate the rolling-forward law of a spherical underwater vehicle and the influence of mass distribution on its motion, this study carries out the innovative design and analysis of its mechanical mechanism.

First, a dynamic model of the rolling-forward motion is established using the Newton-Euler method. The influence of mass distribution on its motion is then analyzed through the ground test and underwater hydrodynamics theory. Finally, by building a simulation environment and virtual prototype, the rolling dynamics of the vehicle underwater and on land are compared and analyzed.

The results show that when the built-in driving unit rotates constantly, the speed of the vehicle fluctuates and the swing angle of the built-in driving unit also alters periodically. When the driving weight is increased, the period and amplitude of swing angle become smaller, and the rolling-forward motion becomes more stable.

In order to improve the attitude control performance of a remote operated vehicle (ROV) in complex underwater environments, the attitude control of ROV with multi-motor coordinated propulsion is studied.

First, for the structure and algorithm of the multi-motor system, a deviation coupling structure based on a PID speed compensator and a new non-singular terminal sliding mode control (SMC) algorithm are respectively proposed, and a novel ROV attitude control method based on multi-motor cooperative propulsion is designed. The kinematic and dynamic model of an ROV is then established, the thrust analysis of the thruster group is carried out, and the decoupling simplified ROV dynamic model is developed. Finally, the ROV sliding mode attitude controller is designed.

The simulation results show that the structure and algorithm proposed in this paper improve the anti-interference, synchronization and rapid response capabilities of the multi-motor system, thereby enhancing the stability and robustness of the ROV attitude control system.

A spherical bulkhead with openings has many structural parameters, and it is time-consuming and laborious to directly use the finite element method to carry out parameter research. Therefore, a combined surrogate model method for predicting characteristic stresses is proposed.

The geometric characteristic parameters of an opening spherical bulkhead are extracted and used as design variables to construct a surrogate model for the characteristic stresses of a spherical bulkhead with a single opening. After analyzing the correlation between the characteristic stresses and structural parameters using the surrogate model, a combined surrogate model method with lower dimensions of design variables is proposed. The combined surrogate model consists of a surrogate model of characteristic stresses based on the parameters of a cylindrical-toroidal-spherical shell and a surrogate model of characteristic stress correction coefficients based on the opening fence parameters and thickness of the spherical shell.

The results show that the accuracy of the combined surrogate model is significantly improved due to the lower dimensions of design variables. Compared with the results of the directly constructed surrogate model, the maximum error and average error of the characteristic stresses are reduced by 4.96%–22.95% and 0.59%–5.43% respectively. The single-opening combination surrogate model is used to predict the characteristic stresses of a typical spherical bulkhead with multiple openings, and the prediction errors of the characteristic stresses are 0.12%–11.42%.

In this paper, the numerical simulation method is used to study the anti-penetration performance and energy absorption mode of a stiffened plate, as well as the influence of different stiffened bars on the flight attitude of the projectile body.

Finite element software LS-DYNA is used to simulate the process of a truncated oval-nosed projectile penetrating a stiffened plate, and the results of the numerical simulation are compared with an experiment to verify the reliability of the numerical simulation method. The momentum method and mass equivalence method are used to predict the residual velocity of the projectile, and the applicability of different theoretical methods within different velocity ranges is compared. The deformation energy of different regions of the stiffened plate is then extracted to analyze the influence of the initial velocity of the projectile body on the energy absorption mode of the target plate. Finally, the structure of the stiffeners is changed and the influence of the relative position of the stiffeners on the penetration attitude of the projectile body is analyzed.

The results show that the mass equivalence method is more accurate than the momentum method in predicting the residual velocity of the stiffened plate when the initial velocity of the projectile body is in the range of 300–900 m/s. The ratio of the deformation energy of the stiffened plate to the energy loss of the projectile body decreases with the increase of the initial velocity of the projectile body. The effect of a T-stiffened plate on trajectory is greater than that of a rectangular-stiffened plate.

This paper proposes a method for calculating the acoustic and vibration response of underwater cylindrical shell structures based on land-based vibration test results.

An axisymmetric boundary element method (BEM) is introduced to describe the radiation acoustic field of the shell. The relationship of acoustic pressure at the nodal point of the generatrix with velocity is obtained by solving the numerical solution of the boundary element integral equation, then the acoustic radiation impedance matrix of the outer surface of the shell and acoustic transfer vector (ATV) are constructed. Based on the assumption that the low-order vibration mode of an underwater cylindrical shell is the same as that of an onshore cylindrical shell, combined with the modes and acoustic radiation impedance matrix of an onshore cylindrical shell, the modal added mass and damping are calculated.

The natural frequency calculation formula of the underwater cylindrical shell is established on the basis of the onshore mode, and the calculation method of underwater vibration response and acoustic radiation characteristics with the vibration response in air as input is obtained based on mode superposition method.

This study aims to investigate the dynamic coupling effect of "hull-fender-dock" during ship berthing.

Using the nonlinear finite element method, a port side and fender structure finite element model is established to simulate the dynamic evolution of speed, stress and energy during ship berthing.

The results show that when the fender makes contact with the dock, the ship's speed decreases to zero and the dynamic deformation and interaction force of the fender structure are at their maximum. When the ship is berthing, the fender shows strong energy absorption capacity, accounting for about 70% of the initial total kinetic energy of the ship, and the hull structure is well protected.