Intelligent ship seakeeping forecasting method and system based on physical information neural network

Abstract

The invention belongs to the technical field of ship and ocean engineering, and discloses a ship seakeeping intelligent forecasting method and system based on a physical information neural network. According to the method, a CFD-RAO method is combined with PINNs, and rapid and high-precision prediction of ship seakeeping is realized by constructing a coupling model of a physical equation and a neural network loss function. According to the invention, rapid and high-precision prediction of ship seakeeping is realized. The method not only can effectively cope with multi-degree-of-freedom coupling motion under complex sea conditions, but also has high real-time performance and adaptability, and provides more scientific and reliable technical support for ship design and navigation safety.

Description

Technical Field

[0001] The invention belongs to the technical field of ship and ocean engineering, and in particular relates to a method and system for intelligently predicting ship seakeeping performance based on a physical information neural network. Background Art

[0002] Seakeeping is one of the key indicators of ship hydrodynamic performance, and has an important impact on the safety, comfort and economy of ships during navigation. In traditional hull design, designers usually pay more attention to the optimization of static water resistance to increase the speed of ships. However, with the complexity of the ship's navigation environment, the importance of seakeeping has gradually been recognized. Seakeeping not only affects the stability of ships in severe sea conditions, but is also directly related to the safety of crew and cargo, as well as the operating costs of ships.

[0003] Current seakeeping prediction technologies mainly rely on methods such as physical model experiments, computational fluid dynamics (CFD) simulations, and empirical formulas. Although these methods can reflect the seakeeping performance of ships to a certain extent, they are usually based on preset working conditions and conditions, which limits the coupling, accuracy, and rapid performance prediction of complex models under complex conditions. It is difficult to cope with complex and changeable actual sea conditions, and the low computational efficiency makes it difficult to quickly predict seakeeping. In addition, although the use of methods such as the response amplitude operator (RAO) alone has high computational efficiency, it has limitations in dealing with multiple degrees of freedom, nonlinearity, and coupled motion, and cannot achieve high-precision, real-time predictions under complex conditions.

[0004] With the rapid development of artificial intelligence and deep learning technologies, Physics-Informed Neural Networks (PINNs) have gradually become an emerging tool for solving complex problems involving physical laws. Physics-Informed Neural Networks are a hybrid method that combines physical equations with data-driven models. By embedding physical laws into the structure of neural networks, the generalization ability and prediction accuracy of the model are greatly improved. Compared with traditional neural networks, PINNs not only rely on data for learning, but also constrain the model through physical equations so that the prediction results conform to physical laws, thereby effectively reducing overfitting and improving prediction accuracy under unseen data conditions.

[0005] In the field of ship seakeeping prediction, the application of PINNs has gradually attracted attention. Although traditional deep learning models perform well in processing large amounts of data, the lack of consideration of physical laws often leads to prediction errors in complex sea conditions and nonlinear coupled motion prediction. PINNs uses the physical equations of ship motion as the loss function of the neural network, so that the model can not only learn data features during training, but also follow physical laws, thereby significantly improving the prediction accuracy of ship seakeeping under complex sea conditions. Therefore, a ship seakeeping prediction method based on physical information neural network is proposed to ensure fast and accurate prediction.

[0006] The problems and defects of the existing technology are as follows: the prediction accuracy of the ship's seakeeping performance in the existing technology is low, the real-time and adaptability of multi-degree-of-freedom coupled motion under complex sea conditions are poor, and it cannot provide a scientific basis for ship design and navigation safety. Summary of the invention

[0007] In order to overcome the problems existing in the related art, the disclosed embodiments of the present invention provide a method and system for intelligent prediction of ship seakeeping based on physical information neural network. In particular, it relates to the cross-technical field of applying machine learning and deep learning technology with physical process simulation. The method integrates computational fluid dynamics (CFD) and response amplitude operator (RAO) technology to construct an intelligent model that can quickly and accurately predict the motion response of a ship in waves and its seakeeping. The physical equations are combined with actual data through the loss function of the neural network, so that the model can not only achieve high-precision prediction, but also has higher adaptability and reliability, especially under complex sea conditions.

[0008] The technical solution is as follows: an intelligent prediction method for ship seakeeping based on physical information neural network, which combines CFD-RAO method with PINNs, and realizes rapid and high-precision prediction of ship seakeeping by constructing a coupling model of physical equations and neural network loss function, specifically including the following steps: S1. Use computational fluid dynamics (CFD) and response amplitude operator (RAO) method to obtain the data set required by the physical information neural network and pre-process the data set; S2, constructing a loss function constrained neural network using the prediction error equation and the physical equation of ship seakeeping; S3, apply the combinatorial optimizer to perform network training; S4. Carry out network prediction and evaluation to realize the prediction of ship's seakeeping performance.

[0009] In step S1, computational fluid dynamics (CFD) and response amplitude operator (RAO) methods are used to obtain the data set required by the physical information neural network, including: (1.1) Construct control model based on CFD-RAO method; (1.2) Establish physical models of ships and flow fields; (1.3) CFD-RAO coupling simulation.

[0010] In step (1.1), a control model is constructed based on the CFD-RAO method, including: The free surface waves during the ship's navigation are captured using the volume of fluid method VOF, and the expression is: ; ; Where: is the fluid density, for The fluid velocities in three directions, is a time series, is the fluid pressure, for The spatial coordinates in three directions, is the fluid velocity at The components in the directions, is the spatial coordinate in The components in the directions, For the The gravitational acceleration in each direction, is the dynamic viscosity coefficient of the fluid, is the turbulent pulsation component of the fluid velocity, is the fluid velocity at Turbulent pulsation components in all directions; use The turbulence model closes the RANS equations and simulates the ship resistance, position derivative, rudder derivative and rotation derivative tests. The basic equation is: ; ; In the formula, is the turbulent kinetic energy, For fluid The dynamic viscosity coefficient in the direction, is the fluid velocity, For fluid The velocity component in the direction, are diffusion coefficients, are turbulence generating terms, are turbulent dissipation terms, All are user-defined source items. is the specific dissipation rate, is an orthogonal divergence term; The hull is a rigid body, select the reference coordinate system , the coordinate origin is located in the middle of the ship, Pointing to the stern, Point to starboard, Vertically upward; the motion equation of the hull sailing attitude is: ; ; ; ; In the formula, is the structural mass matrix, Around the hull The moment of inertia, are the ship's heave acceleration, pitch acceleration, and roll acceleration, The hull edges are Axial force, The resultant moment about the axis of rotation The resultant moment of the axis, the resultant force is obtained by integrating the stress on the hull surface while solving the hull motion in real time and capturing the free liquid surface, taking into account the coupling effect between the degrees of freedom; After obtaining the data using CFD calculation, the response amplitude operator RAO method is used to obtain the dynamic response characteristics. The linear frequency domain motion equation of the 6-DOF ship in the regular wave is: ; In the formula, is the encounter frequency, is the quality coefficient, For the Kinetic Mode Induced The additional mass coefficient and damping coefficient in the direction, is the restoring force coefficient, for The complex motion amplitude of the mode, for The unit amplitude wave force in the direction, is the wave amplitude, are the 6 degrees of freedom of the ship, Representation and Corresponding degrees of freedom of the degree of freedom coupling.

[0011] In step (1.2), a physical model of the ship and flow field is established, including: A virtual wave pool with water as the fluid is built. The variable parameters of the wave are set, the free liquid surface is captured by the volume of fluid method VOF, and the overlapping network technology is used to simulate the ship's navigation posture in real time to complete the establishment of the ship's navigation model.

[0012] In step S2, a loss function constrained neural network is constructed using the prediction error equation and the physical equation of ship seakeeping, including: According to the energy conservation principle of ships moving in waves, a relationship function between energy loss and prediction error is constructed to physically constrain the network. The total loss function in degrees of freedom: ; In the formula, For The total loss function in degrees of freedom, are the weight coefficients of the two loss terms, is the energy loss equation between the ship and the waves, is the prediction error value of the motion response, expressed as: ; ; In the formula, is the number of data points in the ship seakeeping status data set within a short period of time, for Total energy of the vessel at time, is the predicted value of the network, For the The actual motion response amplitude of samples, is the linear damping coefficient, is the nonlinear damping coefficient, For the The angular velocity in degrees of freedom, For the A point in time, For the A point in time, For the Mean Absolute Error under degrees of freedom, used to evaluate the network prediction value With the true value The deviation between .

[0013] Furthermore, the energy method is used to derive the equation of energy loss, and the ship's freedom attenuation motion equation is obtained, which is expressed as follows: ; In the formula, is the total moment of inertia of the ship, They are Angle, angular velocity and angular acceleration in degrees of freedom, is the restoring force coefficient, is the linear damping coefficient, describing the The damping force is proportional to the effect on the system. is the nonlinear damping coefficient, describing the The influence of nonlinear damping force on the system is proportional to the Divide both sides of the equation by the total moment of inertia of the ship ,get: ; In the formula, For the system The natural angular frequency of the degrees of freedom; The total energy of the ship's freedom degree decay process consists of two parts: kinetic energy caused by free rotation and potential energy caused by the work done by the hydrostatic restoring torque. Total energy at the moment for: ; According to the principle of energy conservation in fluid dynamics, the energy loss equation at different times is obtained: ; In the formula, for Total energy of the ship at the moment, for Total energy of the ship at all times.

[0014] In step S3, a combinatorial optimizer is applied to perform network training, including: The seakeeping response data of the ship under different sea conditions are obtained by CFD-RAO method. The seakeeping response data includes the dynamic characteristics of pitch, roll and heave of the ship in waves. Based on the above data, a fully connected neural network model is constructed with time series as input, and the data is segmented and processed to train the fully connected neural network model. During the training process, an optimizer combination strategy is adopted. First, the Adam optimizer is used to converge quickly in the early stage and approach the global optimal solution; then it is switched to the LBFGS optimizer for multiple iterations and the automatic learning rate is used for adjustment.

[0015] In step S4, network prediction and evaluation are performed to achieve prediction of the ship's seakeeping performance, including: using the trained fully connected neural network model to predict the new data set in predicting the seakeeping response values ​​of the ship's pitch, roll and heave.

[0016] Another object of the present invention is to provide a ship seakeeping intelligent prediction system based on physical information neural network, the system implements the ship seakeeping intelligent prediction method based on physical information neural network, the system comprises: Initialize the network and data preprocessing module, used to obtain the data set required by the physical information neural network by using the computational fluid dynamics CFD and the response amplitude operator RAO method, and preprocess the data set; Construct a neural network loss function module to construct a loss function constraint neural network using the prediction error equation and physical equations related to ship seakeeping; A network training module, used to apply a combinatorial optimizer to perform network training; The network prediction and evaluation module is used to perform network prediction and evaluation to achieve the prediction of ship seakeeping performance.

[0017] Combining all the above technical solutions, the beneficial effects of the present invention are as follows: the present invention combines the CFD-RAO method with PINNs, and realizes the rapid and high-precision prediction of the ship's seakeeping performance by constructing a coupling model of physical equations and neural network loss functions. This method can not only effectively cope with multi-degree-of-freedom coupled motion under complex sea conditions, but also has strong real-time and adaptability, providing more scientific and reliable technical support for ship design and navigation safety.

[0018] The present invention proposes for the first time to obtain a data set through the CFD-RAO method and train it in a physical information neural network, as well as a specific physical information neural network algorithm containing a loss function. The introduction of RAO overcomes the shortcomings of high computational cost and long time consumption of traditional CFD methods; combined with CFD, it ensures adaptability to complex nonlinear conditions and complex environments; the combination of the two achieves faster and more accurate predictions to meet real-time decision-making needs. Secondly, the introduction of a physical information neural network (PINN, which combines physical constraints and neural networks) can complete high-precision predictions in a relatively short period of time, providing feasible technical support for real-time monitoring and prediction of ship seakeeping. The PINN method is suitable for fields such as ship design and navigation safety assessment. In addition, it can also be extended to the dynamic performance evaluation of other marine engineering structures such as offshore wind power platforms and floating production units. The proposal of the present invention not only realizes localization and breaks the monopoly of foreign ship industry simulation software, but also achieves further breakthroughs and innovations in the development level of foreign software systems, and proposes a system with better performance and real-time experimental capabilities.

[0019] The present invention solves the contradiction between real-time performance and computational efficiency, and proposes a CFD-RAO method, which reduces costs and improves computational efficiency, thereby achieving fast and high-precision predictions. The problem of difficulty in maintaining high precision in real sea conditions is solved by using the data obtained by the CFD-RAO method for model training, and using the trained physical information neural network, which can maintain high-precision predictions under complex nonlinear and multi-degree-of-freedom conditions, not only fitting observations under data-driven conditions, but also following physical laws under complex sea conditions, making predictions more reliable and accurate. The physical information neural network effectively integrates physical laws and a large amount of acquired data, so that the model can maintain prediction accuracy even in data-scarce areas. This advantage solves the contradiction between data scarcity and high-precision requirements in ship engineering, and meets people's long-standing technical needs.

[0020] The present invention effectively overcomes the technical biases of traditional methods in data dependence, nonlinear adaptability, computational cost and generalization ability by introducing physical information neural network (PINN). The PINN method embeds physical equations into the loss function of the neural network, realizing the organic combination of physical laws and data-driven models, so that the model still has high-precision prediction capabilities under conditions of scarce or unseen data. In addition, this method significantly improves the adaptability under complex nonlinear sea conditions, and meets real-time requirements at a lower computational cost, breaking through the limitations of traditional prediction methods in real-time and accuracy, and providing an efficient, reliable and innovative solution for ship seakeeping prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings herein are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the description, serve to explain the principles of the present disclosure; Figure 1 It is a flow chart of a method for intelligent prediction of ship seakeeping performance based on physical information neural network provided by an embodiment of the present invention; Figure 2 is a schematic diagram of a grid scene model of a ship's wave navigation provided by an embodiment of the present invention; Figure 3 is a structural diagram of a physical information neural network provided by an embodiment of the present invention; Figure 4 Schematic diagram of a ship seakeeping intelligent prediction system based on a physical information neural network provided by an embodiment of the present invention; In the figure: 1. Initialize the network and data preprocessing module; 2. Build the neural network loss function module; 3. Network training module; 4. Network prediction and evaluation module. DETAILED DESCRIPTION

[0022] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific implementation disclosed below.

[0023] The novelty of this invention is that it is based on the physical information neural network (PINN) trained by the CFD-RAO method to obtain the database, embed the physical laws into the neural network loss function, integrate data drive and physical constraints, and realize the real-time high-precision prediction of ship seakeeping under complex nonlinear sea conditions. This method has high efficiency, generalization ability and multi-field applicability, providing innovative solutions for ship design and navigation safety.

[0024] The present invention includes simulating the navigation of a ship under the action of waves based on the CFD-RAO method, constructing the ship energy and motion response prediction equation as the loss function of the neural network, and using the physical information neural network method to predict the ship's seakeeping performance through time series. The CFD-RAO method is used to perform frequency domain analysis on the ship's time domain response data, including the fluid phase and turbulence equations and the ship's motion frequency domain equations, to construct a physical model of ship wave navigation, and to perform coupled calculation simulation. The dynamic characteristics of the ship's hull in the next second are predicted, the prediction accuracy is improved, and the safety and comfort of the ship during navigation are ensured through a real-time detection system.

[0025] The present invention innovatively proposes a CFD-RAO combined method. The CFD method is commonly used with high accuracy, but low calculation efficiency; the RAO method is less used and has great limitations, but high calculation efficiency.

[0026] Embodiment 1, as Figure 1 As shown, the ship seakeeping intelligent prediction method based on physical information neural network provided by the embodiment of the present invention includes: S1. Use computational fluid dynamics (CFD) and response amplitude operator (RAO) method to obtain the data set required by the physical information neural network and pre-process the data set; In this step, the input parameters of the neural network are determined, which include the time series , Ship's moment of inertia , frequency and the coordinates of the ship and the mass matrix ; These parameters are closely related to the impact of waves on the ship's seakeeping performance. In order to accurately capture the ship's seakeeping response under specific wave conditions, it is necessary to first collect a large amount of raw data. These data include but are not limited to the ship's motion displacement, velocity and acceleration, and position characteristics under different wave conditions. The data can be analyzed in the frequency domain by the response data obtained through CFD simulation or model testing to obtain the response amplitude operator (RAO) at each frequency, reflecting the dynamic response characteristics of the ship in the waves, as model data. Figure 2 As shown in the grid scene model of ship wave navigation, the star ccm+ software is used as the model carrier, and the overlapping grid encryption method is adopted to ensure the accuracy and reliability of the calculation.

[0027] The preprocessing stage performs data enhancement and adversarial denoising on time series data, which has better processing results for time series data to ensure the consistency and accuracy of data input. In this way, a frequency domain response database is established to provide a reliable data basis for subsequent neural network training. In actual applications, the corresponding response results can be quickly obtained by simply searching or interpolating the RAO data in the database without repeating CFD simulation, achieving fast and accurate predictions.

[0028] S2, constructing a loss function constrained neural network using the prediction error equation and the physical equation of ship seakeeping; In this step, the output of the neural network model is set as the angular velocity of a certain degree of freedom of the ship, the damping coefficient and the predicted motion response amplitude. The construction of the loss function takes into account two main constraints: one is the physical equation constraint of the ship's motion response, and the other is the prediction accuracy of the predicted motion response amplitude. The constraint of the physical equation is based on the energy conservation theory in the ship's wavekeeping analysis to ensure that the prediction results of the network follow the laws of physics. At the same time, by internally predicting the parameters such as the ship's resistance coefficient, additional mass and damping coefficient, the model's ability to understand and simulate ship motion under complex sea conditions is further enhanced.

[0029] S3, apply the combinatorial optimizer to perform network training; During the training of the neural network, a combination strategy of optimizers was adopted to ensure the high efficiency of training and the accuracy of the model. At the beginning of training, the Adam optimizer was used, and its momentum accumulation and adaptive learning rate adjustment mechanism enabled the model to quickly converge to a region close to the global optimal solution. The large learning rate of the Adam optimizer in the early stage helped the model to quickly reduce the loss value, achieve rapid convergence, and ensure that the model achieved good performance in a short time. As the training progressed, the optimization strategy was switched to the LBFGS optimizer. LBFGS uses approximate second-order information, combines fine step size control and dynamic learning rate adjustment to further optimize the model parameters, so that it can reduce the prediction error under physical constraints. Through more accurate gradient direction and learning rate adjustment, LBFGS not only improves the numerical accuracy of the model, but also ensures that the prediction results conform to the physical laws of ship seakeeping, avoids overfitting, and maintains stability and reliability under complex sea conditions. This combination strategy effectively takes into account both training efficiency and model accuracy.

[0030] S4. Conduct network prediction and evaluation to predict the seakeeping performance of ships; After completing the neural network training, the trained model is used to predict the new data set, and the dynamic characteristics (such as pitch, roll and heave) of the new data set are predicted. Not only is it data-driven, but it also incorporates physical equations (such as energy conservation) as constraints of the loss function to ensure that the prediction results conform to the actual physical laws. In addition, the combined optimizer strategy (Adam and LBFGS) used in the training process achieves rapid convergence in the early stage and fine tuning in the later stage, so that the model has higher numerical accuracy and stability in prediction. Finally, by comparing with the physical model or experimental data, the system verifies the superior prediction performance of the model under complex sea conditions, ensuring its reliability and accuracy in practical applications.

[0031] Exemplarily, the method for constructing the neural network loss function in step 2 is specifically as follows: Aiming at the problem of ship seakeeping prediction, an advanced method combining physical information neural networks (PINNs) with ship motion response is introduced. Figure 3 The structure diagram of the physical information neural network shows that from top to bottom, it includes the input layer, hidden layer, output layer, loss function, and the mechanism of operation and output. Among the input parameters: is a time series, is the ship encounter frequency and the wave natural frequency, is the moment of inertia about the axis, is the ship coordinate position, is the structural mass matrix. Output parameters: is the angular velocity of a certain degree of freedom of the ship, are the linear damping coefficient and the nonlinear damping coefficient, respectively. is the magnitude of the motion response predicted by the neural network.

[0032] For the seakeeping indicators such as the longitudinal motion of the ship (such as heave, trim) and heave response, the model comprehensively considers multi-dimensional input variables such as time series, wave characteristics, mass matrix, etc., and innovatively embeds physical equations into the loss function to ensure that the neural network can not only learn the characteristics of the data during the training process, but also strictly follow the physical laws in ship dynamics. In the specific design of the loss function, the model not only considers the data error, but also further adds constraints based on the physical factors of energy conservation. These physical constraints are integrated into the loss function of the neural network through the formulated energy terms, so that the model not only fits the data distribution in the process of minimizing the loss, but also continuously adjusts the output to conform to the real physical response. This method effectively solves the problems of overfitting and instability of conventional neural networks in nonlinear coupling systems.

[0033] Exemplary, the total loss function is constructed: The total loss function of the PINNs model proposed in this invention consists of two parts, which are specifically expressed as follows: The data prediction loss part quantifies the error between the ship attitude predicted by the model and the experimental data.

[0034] ; The capacity loss is constructed based on the energy conservation principle of ships and waves, which further ensures the physical rationality of the model prediction.

[0035] ; In the formula, For The total loss function in degrees of freedom, are the weight coefficients of the two loss terms, is the energy loss equation between the ship and the waves, is the prediction error value of the motion response, expressed as: ; By adjusting the value of the weight coefficient, reducing the physical error and data error, the total loss function is minimized. The minimization process of the total loss function can prompt the neural network to accurately fit the data under the premise of satisfying physical constraints, so that PINN can handle complex problems with physical constraints. It can not only fit the training data well, but also ensure that the prediction results of the model conform to the laws of physics, which is more reliable and accurate in complex practical applications.

[0036] Example 2, as another implementation of the present invention, the ship seakeeping intelligent prediction method based on physical information neural network provided in the embodiment of the present invention includes: Step 1: Obtain physical information using computational fluid dynamics (CFD) and response amplitude operator (RAO) methods. The data sets required for the neural network include: (1.1) Construct a control model based on the CFD-RAO method.

[0037] When the fluid phase is used for simulation, the fluid phase is considered to be an incompressible viscous fluid and follows the laws of conservation of mass and momentum when flowing in the flow field. The continuity equation and the Navier-Stokes equation are used as the basic equations, and the volume of fluid method (VOF) is used to capture the free surface waves during the ship's navigation process.

[0038] ; ; Where: is the fluid density, for The fluid velocities in three directions, is the fluid pressure, for The spatial coordinates in three directions, is the fluid velocity at The components in the directions, is the spatial coordinate in The components in the directions, For the The gravitational acceleration in each direction, is the dynamic viscosity coefficient of the fluid, is the turbulent pulsation component of the fluid velocity, is the fluid velocity at Turbulent pulsation components in all directions; use The turbulence model closes the RANS equations and simulates the ship resistance, position derivative, rudder derivative and rotation derivative tests. The basic equation is: ; ; In the formula, is the turbulent kinetic energy, For fluid The dynamic viscosity coefficient in the direction, is the fluid velocity, For fluid The velocity component in the direction, are diffusion coefficients, are turbulence generating terms, are turbulent dissipation terms, All are user-defined source items. is the specific dissipation rate, is an orthogonal divergence term; Assuming the hull is a rigid body, select the reference coordinate system , the coordinate origin is located in the middle of the ship, Pointing to the stern, Point to starboard, Vertically upward; the motion equations of the ship's sailing attitude (heave, pitch, roll, etc.) are as follows: ; ; ; ; In the formula, is the structural mass matrix, Around the hull The moment of inertia, are the ship's heave acceleration, pitch acceleration, and roll acceleration, The hull edges are Axial force, The resultant moment about the axis of rotation The resultant moment of the axis, the resultant force is obtained by integrating the stress on the hull surface while solving the hull motion in real time and capturing the free liquid surface, taking into account the coupling effect between the degrees of freedom; The core of ship seakeeping calculation is to accurately predict the motion of the ship in waves. The dynamic characteristics of the hull under the action of waves can be described by the frequency response method within the frequency range. After the data is obtained by CFD calculation, the response amplitude operator (RAO) method is used to obtain the dynamic response characteristics. The linear frequency domain motion equation of the 6-DOF ship in regular waves is: ; In the formula, is the encounter frequency, is the quality coefficient, For the Kinetic Mode Induced The additional mass coefficient and damping coefficient in the direction, is the restoring force coefficient, for The complex motion amplitude of the mode, for The unit amplitude wave force in the direction, is the wave amplitude, are the 6 degrees of freedom of the ship, Representation and Corresponding degrees of freedom of the degree of freedom coupling.

[0039] (1.2) Establish a physical model of the ship and flow field.

[0040] The hull model uses a variety of ship types and ratios for data calculation. Here, the MOERI Container Ship (KCS) model with a scale ratio of 31.599 is taken as an example. The other parameters are shown in Table 1.

[0041] Table 1 Ship model parameters

[0042] Ship navigation model: A virtual wave pool with water as the fluid, setting the variable parameters of the wave, capturing the free liquid surface with the VOF (Volume of Fluid) method, and using overlapping network technology to simulate the ship's navigation posture in real time. The density of the model water is 997.561 kg / m³. The properties of water and waves are defined in the Euler phase, and the relative motion of the ship and the flow field is defined. Figure 2 shown.

[0043] (1.3) CFD-RAO coupling simulation.

[0044] In order to accurately simulate the hydrodynamic performance of the ship during navigation, the longitudinal and transverse lengths of the computational domain are set to 6.0 times the ship length, and the vertical height is set to 4.0 times the ship length. The ship remains at a fixed position, and the upstream inlet boundary is the wave generation area. The wave height, period, wavelength, and wave direction of the wave are defined at the inlet. Perform calculations. Set the outlet boundary to the wave absorption area and the lateral boundary to the symmetry surface. The hull surface is configured to have an anti-slip wall condition. Perform wave elimination processing on the physical model to prevent waves from reflecting, interfering or having other adverse effects in unnecessary areas. Connect the motion model to the flow solver, update the spatial position and boundary condition function of the control grid at each time step, and perform local mesh refinement in areas such as free surface, ship model, and hull Kelvin wave to ensure the accuracy of flow field calculation; generate a six-layer prismatic grid near the hull surface to simulate boundary layer flow. Perform frequency domain transformation analysis on the ship dynamic characteristics collected over time, calculate the response amplitude of the ship when the wave frequency changes, and obtain the motion response of the hull to study the hull's wave resistance. Perform multiple simulations according to different working conditions to evaluate the hull's wave resistance under different conditions.

[0045] Step 2: construct a loss function constrained neural network using physical equations related to waves and ship seakeeping; Due to the energy conservation principle of the ship moving in the waves, the relationship function between energy loss and prediction error is constructed to physically constrain the network. Loss function in degrees of freedom: ; In the formula, For The total loss function in degrees of freedom, are the weight coefficients of the two loss terms, is the energy loss equation between the ship and the waves, is the prediction error value of the motion response, expressed as: ; ; In the formula, is the number of data points in the ship seakeeping status data set within a short period of time, for Total energy of the vessel at time, is the predicted value of the network, For the The actual motion response amplitude of samples, is the linear damping coefficient, is the nonlinear damping coefficient, For the The angular velocity in degrees of freedom, For the A point in time, For the A point in time, For the The mean absolute error under degrees of freedom is used to evaluate the network prediction value With the true value The deviation between .

[0046] During the training process, the neural network not only takes into account the prediction error, but also ensures the compliance with physical laws, making the model more credible and reliable in practical applications.

[0047] Exemplarily, in step 2, the energy loss equation is derived using the energy method, and further research is conducted based on the ship freedom decay motion equation given by Jiang Yin et al. The present invention innovatively gives the ship freedom decay motion equation as follows: ; In the formula, is the total moment of inertia of the ship, They are Angle, angular velocity and angular acceleration of the degrees of freedom, is the restoring force coefficient, is the linear damping coefficient, describing the The damping force is proportional to the effect on the system. is the nonlinear damping coefficient, describing the The influence of nonlinear damping force on the system is proportional to the Divide both sides of the equation by the total moment of inertia of the ship ,get: ; In the formula, For the system The natural angular frequency of the degrees of freedom; The total energy of the ship's freedom degree decay process consists of two parts: kinetic energy caused by free rotation and potential energy caused by the work done by the hydrostatic restoring torque. Total energy at the moment for: ; According to the principle of energy conservation in fluid dynamics, the energy loss equation at different times is obtained: ; In the formula, for Total energy of the ship at the moment, for Total energy of the ship at all times.

[0048] In this way, the physical constraints in the prediction of physical neural information networks are guaranteed, which is of great significance to the authenticity and reliability of the prediction results.

[0049] Step 3 and step 4 use the physical information neural network method to predict the ship's seakeeping performance through time series and perform optimization and evaluation.

[0050] First, the CFD-RAO method is used to obtain the seakeeping response data of the ship under different sea conditions. These data include the dynamic characteristics of the ship in the waves, such as pitch, roll and heave. Based on these data, a fully connected neural network model is constructed with time series as input, and the data is segmented and processed, and then the model training begins.

[0051] During the training process, an optimizer combination strategy was adopted. First, the Adam optimizer was used for rapid convergence in the early stage to ensure that the model quickly approached the global optimal solution. As the training progressed, the LBFGS optimizer was switched to perform fine tuning to further improve the prediction accuracy of the model. The training process included tens of thousands of iterations, and the automatic learning rate was used for adjustment to ensure that the model could adapt to the complexity of different sea conditions.

[0052] After the model training is completed, the trained neural network model is used to predict the new data set, focusing on the performance of the model in predicting the seakeeping response values ​​such as pitch, roll and heave. The prediction accuracy and reliability of the model are evaluated by comparing the prediction results of the model with the existing physical model or experimental data.

[0053] For example, since the seakeeping of ships under different sea conditions is complex and changeable, each small batch of data cannot fully represent the distribution of hull characteristics during the entire voyage, which may lead to oscillations in the loss value at the end of training. However, the final neural network model can still effectively predict the seakeeping of the ship at the next moment, providing reliable technical support for ship design and navigation safety.

[0054] In the training process of deep learning models, using a large number of iterative steps and an automatic learning rate adjustment strategy is an effective way to improve model performance. By using a higher learning rate to converge quickly in the early stage, and then fine-tuning the learning rate in the later stage, the model can show stronger adaptability in complex and changing environments. This method is particularly suitable for complex tasks such as ship seakeeping prediction, because it ensures that the model can effectively learn data features and follow physical laws when processing different sea conditions to provide reliable prediction results.

[0055] This seakeeping prediction method based on physical information neural network significantly improves the rapid accuracy and practicality of prediction by combining computational fluid dynamics (CFD) and response amplitude operator (RAO) technology. The use of combined optimizer and automatic learning rate adjustment strategy not only improves the convergence speed of the model, but also ensures reliability and stability in complex sea conditions. These advantages make this method have important technical value in practical applications and provide more scientific and reliable support for ship design and navigation safety.

[0056] This invention combines CFD, RAO and physical information neural network (PINNs) methods to predict the dynamic characteristics of ships under complex sea conditions through efficient and accurate frequency domain response analysis, and verifies its superior performance in different application scenarios (such as ship stability under high-frequency waves, safety of deep-sea drilling platforms, seakeeping of military ships, attitude control of autonomous driving ships, etc.). In the simulation experiment, the sea conditions were set to include complex conditions such as large-amplitude long-period waves, high-frequency irregular waves, and multi-directional coupled wave conditions to comprehensively evaluate the adaptability and reliability of the model. The experimental results show that the prediction error of the PINNs model under these complex sea conditions is less than 9.5%, which captures the dynamic response of ships more quickly and accurately than traditional methods, and provides efficient and low-error technical support for practical applications.

[0057] The present invention has important social value, improves the safety and stability of ships in complex sea conditions, helps save energy and reduce emissions, and supports the development of smart shipping. At the same time, the technology can be extended to other industries, such as safety assessments of offshore wind power and offshore oil platforms, to ensure the operational stability of facilities in harsh environments. In addition, in the field of autonomous driving and intelligent control, the precise prediction capability of this technology can provide efficient and real-time safety protection for automated systems such as unmanned ships and unmanned platforms, perform fatigue estimation and sea condition analysis, promote the application of intelligent technology in water transportation and resource development, promote industrial transformation and upgrading, and achieve sustainable social development and higher resource utilization efficiency.

[0058] Embodiment 3, as Figure 4 As shown, an embodiment of the present invention provides a ship seakeeping intelligent prediction system based on a physical information neural network, comprising: Initialize the network and data preprocessing module 1, used to obtain the data set required by the physical information neural network by using the computational fluid dynamics CFD and the response amplitude operator RAO method, and preprocess the data set; Constructing a neural network loss function module 2, which is used to construct a loss function constraint neural network using physical equations related to waves and ship seakeeping; A network training module 3, used for applying a combinatorial optimizer to perform network training; The network prediction and evaluation module 4 is used to perform network prediction and evaluation to achieve the prediction of the ship's seakeeping performance.

[0059] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principles of the present invention should be covered within the protection scope of the present invention.

Claims

1. A method for intelligent prediction of ship seakeeping performance based on physical information neural network, characterized in that: This method combines the CFD-RAO method with PINNs, and realizes fast and high-precision prediction of ship seakeeping by constructing a coupling model of physical equations and neural network loss functions. The specific steps include: S1. Use computational fluid dynamics (CFD) and response amplitude operator (RAO) method to obtain the data set required by the physical information neural network and pre-process the data set; S2, constructing a loss function constrained neural network using the prediction error equation and the physical equation of ship seakeeping; S3, apply the combinatorial optimizer to perform network training; S4. Carry out network prediction and evaluation to realize the prediction of ship's seakeeping performance.

2. The ship seakeeping intelligent prediction method based on physical information neural network according to claim 1 is characterized in that: In step S1, the method of using computational fluid dynamics (CFD) and response amplitude operator (RAO) to obtain the data set required by the physical information neural network includes: (1.1) Construct control model based on CFD-RAO method; (1.2) Establish physical models of ships and flow fields; (1.3) CFD-RAO coupling simulation.

3. The ship seakeeping intelligent prediction method based on physical information neural network according to claim 2 is characterized in that: In step (1.1), the control model is constructed based on the CFD-RAO method, including: The free surface waves during the ship's navigation are captured using the volume of fluid method VOF, and the expression is: ； ； Where: is the fluid density, for The fluid velocities in three directions, is a time series, is the fluid pressure, for The spatial coordinates in three directions, is the fluid velocity at The components in the directions, is the spatial coordinate in The components in the directions, For the The gravitational acceleration in each direction, is the dynamic viscosity coefficient of the fluid, is the turbulent pulsation component of the fluid velocity, is the fluid velocity at Turbulent pulsation components in all directions; use The turbulence model closes the RANS equations and simulates the ship resistance, position derivative, rudder derivative and rotation derivative tests. The basic equation is: ； ； In the formula, is the turbulent kinetic energy, For fluid The dynamic viscosity coefficient in the direction, is the fluid velocity, For fluid The velocity component in the direction, are diffusion coefficients, are turbulence generating terms, are turbulent dissipation terms, All are user-defined source items. is the specific dissipation rate, is an orthogonal divergence term; The hull is a rigid body, select the reference coordinate system , the coordinate origin is located in the middle of the ship, Pointing to the stern, Point to starboard, Vertically upward; the motion equation of the hull sailing attitude is: ； ； ； ； In the formula, is the structural mass matrix, Around the hull The moment of inertia, are the ship's heave acceleration, pitch acceleration, and roll acceleration, The acting hull edges are Axial force, The resultant moment about the axis of rotation The resultant moment of the axis, the resultant force is obtained by integrating the stress on the hull surface while solving the hull motion in real time and capturing the free liquid surface, taking into account the coupling effect between the degrees of freedom; After obtaining the data using CFD calculation, the response amplitude operator RAO method is used to obtain the dynamic response characteristics. The linear frequency domain motion equation of the 6-DOF ship in the regular wave is: ； In the formula, is the encounter frequency, is the quality coefficient, For the Kinetic Mode Induced The additional mass coefficient and damping coefficient in the direction, is the restoring force coefficient, for The complex motion amplitude of the mode, for The unit amplitude wave force in the direction, is the wave amplitude, are the 6 degrees of freedom of the ship, Representation and Corresponding degrees of freedom of the degree of freedom coupling.

4. The ship seakeeping intelligent prediction method based on physical information neural network according to claim 2 is characterized in that: In step (1.2), the physical model of the ship and the flow field is established, including: A virtual wave pool with water as the fluid is built. The variable parameters of the wave are set, the free liquid surface is captured by the volume of fluid method VOF, and the overlapping network technology is used to simulate the ship's navigation posture in real time to complete the establishment of the ship's navigation model.

5. The method for intelligent prediction of ship seakeeping performance based on physical information neural network according to claim 1 is characterized in that: In step S2, the loss function constraint neural network is constructed by using the prediction error equation and the physical equation of the ship's seakeeping, including: According to the energy conservation principle of ships moving in waves, a relationship function between energy loss and prediction error is constructed to physically constrain the network. The total loss function in degrees of freedom: ； In the formula, For The total loss function in degrees of freedom, are the weight coefficients of the two loss terms, is the energy loss equation between the ship and the waves, is the prediction error value of the motion response, expressed as: ； ； In the formula, is the number of data points in the ship seakeeping status data set within a short period of time, for Total energy of the vessel at time, is the predicted value of the network, For the The actual motion response amplitude of samples, is the linear damping coefficient, is the nonlinear damping coefficient, For the The angular velocity in degrees of freedom, For the A point in time, For the A point in time, For the The mean absolute error under degrees of freedom is used to evaluate the network prediction value With the true value The deviation between .

6. The ship seakeeping intelligent prediction method based on physical information neural network according to claim 5 is characterized in that: The energy method is used to derive the equation of energy loss, and the ship's freedom attenuation motion equation is obtained, which is expressed as follows: ； In the formula, is the total moment of inertia of the ship, They are Angle, angular velocity and angular acceleration of the degrees of freedom, is the restoring force coefficient, is the linear damping coefficient, describing the The damping force is proportional to the effect on the system. is the nonlinear damping coefficient, describing the The influence of nonlinear damping force on the system is proportional to the Divide both sides of the equation by the total moment of inertia of the ship ,get: ； In the formula, For the system The natural angular frequency of the degrees of freedom; The total energy of the ship's freedom decay process consists of two parts: kinetic energy caused by free rotation and potential energy caused by the work done by the hydrostatic restoring torque. Total energy at the moment for: ； According to the principle of energy conservation in fluid dynamics, the energy loss equation at different times is obtained: ； In the formula, for Total energy of the ship at the moment, for Total energy of the ship at the moment.

7. The method for intelligent prediction of ship seakeeping performance based on physical information neural network according to claim 1 is characterized in that: In step S3, the application combination optimizer performs network training, including: The seakeeping response data of the ship under different sea conditions are obtained by CFD-RAO method. The seakeeping response data includes the dynamic characteristics of pitch, roll and heave of the ship in waves. Based on the above data, a fully connected neural network model is constructed with time series as input, and the data is segmented and processed to train the fully connected neural network model. During the training process, an optimizer combination strategy is adopted. First, the Adam optimizer is used to converge quickly in the early stage and approach the global optimal solution; then it is switched to the LBFGS optimizer for multiple iterations and the automatic learning rate is used for adjustment.

8. The method for intelligent prediction of ship seakeeping performance based on physical information neural network according to claim 1 is characterized in that: In step S4, the network prediction and evaluation are performed to realize the prediction of the ship's seakeeping performance, including: using the trained fully connected neural network model to predict the new data set in predicting the seakeeping response values ​​of the ship's pitch, roll and heave.

9. An intelligent prediction system for ship seakeeping based on physical information neural network, characterized in that: The system implements the ship seakeeping intelligent prediction method based on physical information neural network according to any one of claims 1 to 8, and the system comprises: Initializing the network and data preprocessing module (1), used to obtain the data set required by the physical information neural network by using computational fluid dynamics (CFD) and response amplitude operator (RAO) method, and preprocessing the data set; Constructing a neural network loss function module (2) for constructing a loss function constraint neural network using a prediction error equation and a physical equation related to the ship's seakeeping performance; A network training module (3), used for applying a combinatorial optimizer to perform network training; The network prediction and evaluation module (4) is used to perform network prediction and evaluation to achieve prediction of the ship's seakeeping performance.

10. The ship seakeeping intelligent prediction system based on physical information neural network according to claim 9 is characterized in that: The system is mounted on a computer device, which includes: at least one processor, a memory, and a computer program stored in the memory and running on the at least one processor. When the processor executes the computer program, the functions of the above-mentioned ship seakeeping intelligent prediction system based on physical information neural network are realized.

Images