A method for generating a marine dynamic simulation result for a target sea area

Abstract

The embodiment of the present application provides a kind of marine dynamic simulation result generation method for target sea area, by collecting the sea area information for the target sea area;Physical information neural network is constructed;Determine the neural network output function for representing water level and flow rate based on the sea area information through the physical information neural network;Based on the sea area information and the neural network output function, two-dimensional plane two-dimensional water dynamic model is constructed;Residual function is constructed by the two-dimensional plane two-dimensional water dynamic model;Determine target weight, and adjust the residual function based on the target weight, and generate marine dynamic simulation result for the target sea area, to realize the accuracy and efficiency of marine dynamic simulation result for target sea area are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of generating marine dynamic simulation results for a target sea area, and in particular, to a method for generating marine dynamic simulation results for a target sea area, a device for generating marine dynamic simulation results for a target sea area, an electronic device, and a computer-readable storage medium. BACKGROUND

[0002] The development and construction of offshore wind farms are rapidly developing. In order to rationally and orderly develop and utilize, and manage offshore resources, and promote the comprehensive development and utilization of offshore wind farm projects, it is necessary to deeply carry out numerical simulation research on the sea area where the wind farm is located, and to construct a mathematical model suitable for the characteristics of the engineering area, so as to provide a basis for the research and design and management of offshore wind farm projects. However, at the present stage, the design, construction and operation of offshore wind farms mostly use the simulation results of classical physical models. Due to the complex and changeable environmental factors such as hydrology, meteorology and geology of the sea, the data quantity is small, the existing hydrodynamic simulation results have low precision, and the real motion state of the complex hydrodynamic characteristics and offshore wind turbine systems under the influence of offshore wind power and other marine structures cannot be completely described.

[0003] Therefore, how to improve the accuracy of marine dynamic simulation is a technical problem that needs to be overcome by those skilled in the art. SUMMARY

[0004] The embodiments of the present application provide a method and device for generating marine dynamic simulation results for a target sea area, an electronic device, and a computer-readable storage medium, to solve the problem of how to generate marine dynamic simulation results for a target sea area.

[0005] The embodiments of the present application disclose a method for generating marine dynamic simulation results for a target sea area, which can include:

[0006] Collecting sea area information for the target sea area;

[0007] Constructing a physical information neural network;

[0008] Determining a neural network output function for characterizing water level and flow rate based on the sea area information through the physical information neural network;

[0009] Constructing a two-dimensional plane two-dimensional hydrodynamic model based on the sea area information and the neural network output function;

[0010] Constructing a residual function through the two-dimensional plane two-dimensional hydrodynamic model;

[0011] Determining a target weight, adjusting the residual function based on the target weight, and generating marine dynamic simulation results for the target sea area.

[0012] Optionally, the sea area information can include average flow velocity information, sea area boundary information, water level information, and total area information, and the average flow velocity information includes flow velocity information.

[0013] Optionally, the step of determining, based on the sea area information, a neural network output function for representing water level and flow velocity by using the physical information neural network can include:

[0014] determining horizontal spatial position information and vertical spatial position information for the target sea area according to the sea area boundary information and the total area information;

[0015] determining an initial weight value and a detection time of an activation function for the physical information neural network;

[0016] inputting the initial weight value, the detection time, the average flow velocity information, the horizontal spatial position information, and the vertical spatial position information into Formula 1 to determine the neural network output function, the Formula 1 being:

[0017] η(x,y,t)≈η s (x,y,t;θ u )

[0018]

[0019] wherein x is the horizontal spatial position information, y is the vertical spatial position information, t is the detection time, θ is the initial weight value, η is the water level information, η s is an approximation of the water level information, u is the flow velocity information, is average flow velocity information for the horizontal spatial position information, is an approximation of the average flow velocity information for the horizontal spatial position information, is average flow velocity information for the vertical spatial position information, is an approximation of the average flow velocity information for the vertical spatial position information.

[0020] Optionally, the step of constructing a two-dimensional plane two-dimensional hydrodynamic model based on the sea area information and the neural network output function can include:

[0021] obtaining still water depth information, total water depth information, gravitational acceleration information, rotation acceleration information, a Coriolis force parameter, and a Manning coefficient for the target sea area;

[0022] determining a horizontal eddy viscosity coefficient function for the horizontal spatial position information and the vertical spatial position information;

[0023] inputting the still water depth information, the total water depth information, the gravity acceleration information, the rotation acceleration information, the Chezy coefficient, the Coriolis force parameter, the Manning coefficient, the horizontal eddy viscosity coefficient function, the detection time, the vertical spatial position information, the lateral spatial position information and the average flow velocity information into Formula Two to determine the two-dimensional plane two-dimensional water dynamic model, the Formula Two being:

[0024]

[0025] wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η is the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is the average flow velocity information for the lateral spatial position information, is the average flow velocity information for the vertical spatial position information, is the rotation acceleration information for the lateral spatial position information, f is the Coriolis force parameter, is the rotation acceleration information for the vertical spatial position information, C is the Chezy coefficient, n is the Manning coefficient, ε x is the horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is the horizontal eddy viscosity coefficient function for the lateral spatial position information.

[0026] Optionally, the step of constructing the residual function through the two-dimensional plane two-dimensional water dynamic model can comprise:

[0027] inputting the lateral spatial position information, the vertical spatial position information, the detection time, an approximate value of the water level information, an approximate value of the average flow velocity information for the lateral spatial position information, an approximate value of the average flow velocity information for the vertical spatial position information, the Coriolis force parameter, the still water depth information, the total water depth information, the gravity acceleration information, the horizontal eddy viscosity coefficient function for the lateral spatial position information, the horizontal eddy viscosity coefficient function for the vertical spatial position information and the Chezy coefficient into Formula Three to construct the residual function, the Formula Three being:

[0028]

[0029] e = e1(θ) + e2(θ) + e3(θ)

[0030] wherein e, e1, e2 and e3 are the residual function, x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η swhere h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, where h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, where h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, x where h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, y where h is the still water depth information, H is the total water depth information, g is the gravity acceleration information.

[0031] Optionally, the step of determining the target weight value can comprise:

[0032] determining a learning rate and an iteration number;

[0033] inputting the learning rate, the residual function and the iteration number into Formula Four to determine the target weight value, the Formula Four being:

[0034]

[0035] where θ is the initial weight value, n is the iteration number, α is the learning rate, e is the residual function, θ n+1 is the target weight value.

[0036] Optionally, the physical information neural network can comprise two hidden layers.

[0037] The embodiment of the present application also discloses a device for generating a marine dynamic simulation result for a target sea area, which can comprise:

[0038] a sea area information collection module configured to collect sea area information for the target sea area;

[0039] a physical information neural network construction module configured to construct a physical information neural network;

[0040] a neural network output function determination module configured to determine, based on the sea area information, a neural network output function for representing water level and flow rate by using the physical information neural network;

[0041] a two-dimensional plane two-dimensional water dynamic model output module configured to construct a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function;

[0042] a residual function construction module configured to construct a residual function by using the two-dimensional plane two-dimensional water dynamic model;

[0043] A target weight determining module is configured to determine a target weight, adjust the residual function based on the target weight, and generate a marine dynamic simulation result for the target sea area.

[0044] The embodiment of the present application further discloses an electronic device, including a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete mutual communication through the communication bus.

[0045] The memory is used for storing a computer program.

[0046] The processor is used for executing the program stored on the memory, and realizes the method as described in the embodiment of the present application.

[0047] The embodiment of the present application further discloses a computer readable storage medium, which stores instructions, and when executed by one or more processors, causes the processor to execute the method as described in the embodiment of the present application.

[0048] The embodiment of the present application has the following advantages:

[0049] The embodiment of the present application collects sea area information for the target sea area, constructs a physical information neural network, determines a neural network output function for representing water level and flow rate based on the sea area information and the physical information neural network, constructs a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function, constructs a residual function through the two-dimensional plane two-dimensional water dynamic model, determines a target weight, adjusts the residual function based on the target weight, and generates a marine dynamic simulation result for the target sea area, thereby improving the accuracy and efficiency of the marine dynamic simulation result for the target sea area. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 is a step flow chart of a marine dynamic simulation result generation method for a target sea area provided in the embodiment of the present application;

[0051] Figure 2 is a marine water dynamic process schematic diagram provided in the embodiment of the present application;

[0052] Figure 3 is a structural block diagram of a marine dynamic simulation result generation device for a target sea area provided in the embodiment of the present application;

[0053] Figure 4 is a hardware structural block diagram of an electronic device provided in each embodiment of the present application;

[0054] Figure 5 is a schematic diagram of a computer readable medium provided in the embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0056] In order to better understand the embodiments of the present application, the technical terms related to the embodiments of the present application are described below.

[0057] PINNs: Physics-Informed Neural Networks, also known as Physics-Informed Neural Networks, is a new type of machine learning method that combines physical laws and neural networks, aiming to solve the limitations of traditional data-based machine learning methods in physical modeling.

[0058] Ocean dynamic simulation is a method that uses mathematical models and computer technology to numerically simulate various physical processes in the ocean, such as waves, tides, ocean currents, etc. Through simulation, we can better understand the movement rules of the ocean, predict changes in the marine environment, and provide scientific basis for marine engineering, fisheries, navigation, etc. Based on the simulation of ocean dynamics, we can predict changes in the marine environment, evaluate the impact of marine engineering, optimize route design, and study marine ecosystems. However, the accuracy of the simulation results of the classical physical model is often unsatisfactory when faced with complex and variable marine environments. The cost of obtaining hydrological, meteorological and geological data at sea is high, and the amount of data is relatively small, which limits the accuracy of the model. The impact of offshore wind farms and other marine structures on the hydrodynamic environment is complex and diverse, and existing simulation methods are difficult to fully describe. The embodiments of the present application provide a method for generating ocean dynamic simulation results for a target sea area, which combines physics-informed neural networks and two-dimensional plane two-dimensional hydrodynamic models, and uses real-time monitoring data to quickly obtain the hydrodynamic process under the conditions of the target sea area, in order to improve the accuracy and efficiency of the ocean dynamic simulation results for the target sea area.

[0059] Reference Figure 1 , a step flowchart of a method for generating ocean dynamic simulation results for a target sea area provided by an embodiment of the present application is shown, which can specifically include the following steps:

[0060] Step 101, collecting sea area information for the target sea area;

[0061] Step 102, constructing a physics-informed neural network;

[0062] Step 103, determining a neural network output function for characterizing water level and flow rate based on the sea area information through the physics-informed neural network;

[0063] Step 104, constructing a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function;

[0064] Step 105, constructing a residual function through the two-dimensional plane two-dimensional water dynamic model;

[0065] Step 106, determining a target weight value, adjusting the residual function based on the target weight value, and generating a marine dynamic simulation result for the target sea area.

[0066] In actual application, the target sea area in the embodiment of the present application can be a sea area that needs to be simulated for marine dynamics, for example, a sea area where a sea wind power plant is developed and constructed, in addition, the target sea area can also be a complex topography sea area, a human-affected sea area, an extreme weather-prone sea area, an ecologically sensitive sea area, or a data-deficient sea area, etc. that needs to be simulated for marine dynamics.

[0067] In specific implementation, the embodiment of the present application can collect sea area information for a target sea area; construct a physical information neural network; determine a neural network output function for representing water level and flow rate based on the sea area information through the physical information neural network; construct a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function; construct a residual function through the two-dimensional plane two-dimensional water dynamic model; determine a target weight value, adjust the residual function based on the target weight value, and generate a marine dynamic simulation result for the target sea area, for example, the sea area information for the target sea area can be collected, for example, topographic data, including but not limited to obtaining high-precision DEM (Digital Elevation Model, also known as digital elevation model) of the target sea area, which can include seabed topography and coastline information; hydrological data, including but not limited to historical water level observation data, tidal data, river flow data, etc.; meteorological data, including but not limited to wind speed, wind direction, air pressure, precipitation, etc. meteorological data, etc., marine environmental data, including but not limited to temperature, salinity tidal current, suspended matter concentration, etc. marine environmental data; human activity data, including but not limited to port operation, reclamation, river into the sea, etc. human activity data, then, a physical information neural network is established, a solver established by a physical information neural network PINNs with stronger generalization ability can be selected, PINNs is constructed based on a deep forward neural network, the forward neural network is composed of a series of nonlinear activation functions, the activation functions are finally combined into output functions, the output functions can realize approximation of any function in nonlinear function optimization, and a neural network output function for representing water level and flow rate is determined based on the sea area information through PINNs.

[0068] Preferably, the sea area information can include average flow rate information, sea area boundary information, water level information, and total area information, the average flow rate information can include flow rate information.

[0069] Then a two-dimensional plane two-dimensional hydrodynamic model can be constructed based on the sea area information and the neural network output function. The two-dimensional plane two-dimensional hydrodynamic model has a wide range of applications in the fields of hydrology, water conservancy, and ocean engineering. Compared with one-dimensional models, two-dimensional models can more accurately simulate the flow changes of water on the horizontal plane, including flow velocity distribution, water level changes, etc. They can simulate the influence of complex coastlines, islands, shoals, and other topographies on water flow, and can be used for hydrodynamic simulation of large-scale sea areas or lakes, providing support for regional-scale water environment research. At the same time, compared with three-dimensional models, two-dimensional models have relatively simple equation systems and smaller computational loads, making them suitable for parallel computing and allowing full use of the computing resources of computer clusters to improve computational efficiency. In addition, two-dimensional plane two-dimensional hydrodynamic models can simulate the propagation and amplitude of tides, the impact of extreme weather events such as typhoons and storm surges on water levels and flow velocities, the processes of river estuary water and salt intrusion, sediment transport, and the diffusion and migration of pollutants in water bodies. They are easy to couple with other models and simple to calibrate parameters, and can intuitively display the distribution of water level, flow velocity, and other hydrodynamic elements, facilitating analysis and understanding.

[0070] A residual function can be constructed through the two-dimensional plane two-dimensional hydrodynamic model, for example, the residual function can be ResidualFunction, which can be used to construct a ResNet (Residual Network) that outputs not only the learning features of the layer but also a direct input from the previous layer. The residual network can alleviate the gradient vanishing problem, as the gradient vanishing problem in deep networks makes it difficult to train the network. The residual connection directly passes the input to the output, making it easier for gradients to be backpropagated, thereby alleviating the gradient vanishing problem. At the same time, the residual network can accelerate network convergence, making it easier to find the optimal solution. In addition, the residual network can improve the performance of the network, and the residual network has shown better performance than traditional networks in many tasks, especially in image classification and object detection. The residual network can effectively solve the network degradation problem that occurs as the network depth increases, making it easier to stack together and build deeper networks without significantly increasing the training difficulty. It helps to effectively pass the features of shallow networks to deep networks, thereby improving the expression ability of the network. In PINNs, by introducing residual connections, PINNs can better fit complex physical systems, thereby improving the prediction accuracy of the model. It can help PINNs to better generalize to new data, thereby reducing the risk of overfitting and accelerating the training process of PINNs, making the model converge faster.

[0071] Further, target weights can be determined, and the residual function is adjusted based on the target weights, and the ocean dynamic simulation result for the target sea area is generated. In PINNs, the target weights can refer to the weight coefficients assigned to the residuals of different physical quantities (such as water level, flow rate, pressure, etc.). These weight coefficients determine the degree of influence of different physical quantities on the final loss function. By adjusting the target weights, the constraint conditions of different physical quantities can be emphasized or weakened, so as to better capture the specific characteristics of the target sea area. The solution output by the neural network is substituted into the model to calculate the residual of the equation, which represents the error between the neural network output and the true solution. The residuals of different physical quantities are multiplied by the corresponding weight coefficients, and then the products are added to obtain the total loss function. Through continuous iteration and optimization process, the neural network can well fit the training data and meet the physical constraint conditions. The trained neural network is applied to new input data to predict the water level, flow rate and other information of the target sea area. When the residual converges, the neural network output function is a two-dimensional plane two-dimensional water dynamic model, which is used as the ocean dynamic simulation result. Preferably, the prediction result can be visualized, such as drawing contour maps, vector diagrams, etc., to intuitively display the simulation result.

[0072] In the embodiment of the present application, the sea area information for the target sea area is collected, a physical information neural network is constructed, a neural network output function for representing water level and flow rate is determined based on the sea area information through the physical information neural network, a two-dimensional plane two-dimensional water dynamic model is constructed based on the sea area information and the neural network output function, a residual function is constructed through the two-dimensional plane two-dimensional water dynamic model, target weights are determined, and the residual function is adjusted based on the target weights, and an ocean dynamic simulation result for the target sea area is generated, thereby improving the accuracy and efficiency of the ocean dynamic simulation result for the target sea area.

[0073] On the basis of the above-mentioned embodiments, variant embodiments of the above-mentioned embodiments are proposed. It should be noted that, in order to make the description brief, only the differences between the variant embodiments and the above-mentioned embodiments are described in the variant embodiments.

[0074] In an optional embodiment of the present application, the step of determining the neural network output function for representing water level and flow rate based on the sea area information through the physical information neural network comprises:

[0075] Determining the horizontal spatial position information and the vertical spatial position information for the target sea area according to the sea area boundary information and the total area information;

[0076] Determining the initial weights and detection time of the activation function of the physical information neural network;

[0077] inputting the initial weight, the detection time, the average flow rate information, the lateral spatial position information and the vertical spatial position information into Formula I, determining the neural network output function, the Formula I being:

[0078] η(x,y,t)≈η s (x,y,t;θ u )

[0079]

[0080] wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, θ is the initial weight, η is the water level information, η s is an approximate value of the water level information, u is the flow rate information, is average flow rate information for the lateral spatial position information, is an approximate value of the average flow rate information for the lateral spatial position information, is average flow rate information for the vertical spatial position information, is an approximate value of the average flow rate information for the vertical spatial position information.

[0081] In a specific implementation, the embodiment of the present application can determine the lateral spatial position information and the vertical spatial position information for a target sea area according to sea area boundary information and total area information; determine the initial weight and the detection time of an activation function of a physical information neural network; input the initial weight, the detection time, the average flow rate information, the lateral spatial position information and the vertical spatial position information into the above Formula I, determine the neural network output function, and exemplarily, η(x,y,t) can be water level information, the water level information can refer to an actual water level, i.e., a target value about water level in a marine dynamic simulation result determined for the target sea area, η s (x,y,t;θ u ) can be an approximate value of the water level information, i.e., a predicted value for the water level, which is a function about spatial coordinates (x, y), detection time t and initial weight θ, can be average flow rate information for the lateral spatial position information, can be average flow rate information for the vertical spatial position information, and can be actual values, the neural network output function approximates the real water level and flow rate through η s (x,y,t;θ u ), θ u can be a learnable parameter of the neural network output function, and θ uThe output of the neural network output function can be made as similar as possible to the true value, and the input of the neural network output function includes spatial coordinates (x, y), detection time t and initial weight value θ. This shows that the neural network can learn the law of water level and flow rate changing with space and time, and the output value generated by formula one is used to approximate the true water level and flow rate. By introducing spatial coordinates, time and initial weight value, the information of spatial variation, time evolution and initial condition in the physical process is integrated into the neural network. The initial weight value θ is learnable, and through training, a set of optimal weight values can be found to minimize the error between the output of the neural network and the true value. The initial weight value is initialized based on average flow rate information, horizontal spatial position information, vertical spatial position information and detection time, which helps the network converge to the optimal solution faster. The neural network output function provides a basis for simulating hydrodynamic processes. By integrating physical information into the neural network and using learnable parameters, the model can effectively capture the complexity of hydrodynamic processes and provide accurate prediction results.

[0082] In the embodiment of the present application, the horizontal spatial position information and the vertical spatial position information for the target sea area are determined according to the sea area boundary information and the total area information;

[0083] The initial weight value and the detection time of the activation function for the physical information neural network are determined;

[0084] The initial weight value, the detection time, the average flow rate information, the horizontal spatial position information and the vertical spatial position information are input into formula one to determine the neural network output function, and the formula one is:

[0085] η(x,y,t)≈η s (x,y,t;θ u )

[0086]

[0087] wherein x is the horizontal spatial position information, y is the vertical spatial position information, t is the detection time, θ is the initial weight value, η is the water level information, η s is an approximate value of the water level information, u is the flow rate information, is the average flow rate information for the horizontal spatial position information, is an approximate value of the average flow rate information for the horizontal spatial position information, is the average flow rate information for the vertical spatial position information, is an approximate value of the average flow rate information for the vertical spatial position information, thereby further improving the accuracy of generating marine dynamic simulation results for the target sea area.

[0088] In an optional embodiment of the present application, the step of constructing the two-dimensional plane two-dimensional hydrodynamic model based on the sea area information and the neural network output function comprises:

[0089] obtaining static water depth information, total water depth information, gravity acceleration information, rotation acceleration information, a Coriolis force parameter, and a Manning coefficient for the target sea area;

[0090] determining a horizontal eddy viscosity coefficient function for the lateral spatial position information and the vertical spatial position information;

[0091] inputting the static water depth information, the total water depth information, the gravity acceleration information, the rotation acceleration information, the Coriolis force parameter, the Manning coefficient, the horizontal eddy viscosity coefficient function, the detection time, the vertical spatial position information, the lateral spatial position information, and the average flow velocity information into Formula Two to determine the two-dimensional plane two-dimensional hydrodynamic model, the Formula Two being:

[0092]

[0093] wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η is the water level information, h is the static water depth information, H is the total water depth information, g is the gravity acceleration information, is average flow velocity information for the lateral spatial position information, is average flow velocity information for the vertical spatial position information, is rotation acceleration information for the lateral spatial position information, f is the Coriolis force parameter, is rotation acceleration information for the vertical spatial position information, C is the Chezy coefficient, n is the Manning coefficient, ε x is a horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is a horizontal eddy viscosity coefficient function for the lateral spatial position information.

[0094] In a specific implementation, the embodiment of the present application can obtain the still water depth information, total water depth information, gravity acceleration information, rotation acceleration information, Chezy coefficient, Coriolis force parameter and Manning coefficient for the target sea area; determine the horizontal eddy viscosity coefficient function for the horizontal spatial position information and the vertical spatial position information; input the still water depth information, total water depth information, gravity acceleration information, rotation acceleration information, Chezy coefficient, Coriolis force parameter, Manning coefficient, horizontal eddy viscosity coefficient function, detection time, vertical spatial position information, horizontal spatial position information and average flow velocity information into Formula Two to determine the two-dimensional plane two-dimensional hydrodynamic model, for example, x is the horizontal spatial position information, y is the vertical spatial position information, t is the detection time, η is the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is the average flow velocity information for the horizontal spatial position information, is the average flow velocity information for the vertical spatial position information, is the rotation acceleration information for the horizontal spatial position information, f is the Coriolis force parameter, is the rotation acceleration information for the vertical spatial position information, C is the Chezy coefficient, n is the Manning coefficient, ε x is the horizontal eddy viscosity coefficient function for the horizontal spatial position information, ε y is the horizontal eddy viscosity coefficient function for the horizontal spatial position information, the Coriolis force parameter can represent the influence of the earth rotation on the flow, the Chezy coefficient can be used to calculate the friction, the Manning coefficient can be used to calculate the bottom friction, and the horizontal eddy viscosity coefficient can consider the influence of the turbulent flow on the flow. Through the two-dimensional plane two-dimensional hydrodynamic model, the output of the neural network can be strictly constrained to ensure that the prediction result of the neural network meets the physical law. Preferably, when training the neural network, the residual error of the two-dimensional plane two-dimensional hydrodynamic model is taken as part of the loss function. By minimizing the residual error, the output of the neural network can better meet the physical equation. By introducing the physical equation as a constraint, the physical consistency of the model can be significantly improved, so that the model can better capture the essence of the hydrodynamic process. Through physical parameters and physical processes, stronger physical constraints can be provided to improve the accuracy of the model. The model can be applied to a wider range of hydrodynamic problems, such as tides, storm surges, river inflow, etc., so that the physical information neural network has a stronger physical basis when simulating the hydrodynamic process. By taking the physical equation as a constraint, the accuracy and reliability of the model can be improved.

[0095] In the embodiment of the present application, the still water depth information, total water depth information, gravity acceleration information, rotation acceleration information, Chezy coefficient, Coriolis force parameter and Manning coefficient for the target sea area are obtained;

[0096] determining a horizontal eddy viscosity coefficient function for the horizontal spatial position information and the vertical spatial position information;

[0097] inputting the still water depth information, the total water depth information, the gravity acceleration information, the rotation acceleration information, the Chezy coefficient, the Coriolis force parameter, the Manning coefficient, the horizontal eddy viscosity coefficient function, the detection time, the vertical spatial position information, the horizontal spatial position information and the average flow velocity information into formula two to determine the two-dimensional plane two-dimensional hydrodynamic model, the formula two being:

[0098]

[0099] wherein x is the horizontal spatial position information, y is the vertical spatial position information, t is the detection time, η is the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is the average flow velocity information for the horizontal spatial position information, is the average flow velocity information for the vertical spatial position information, is the rotation acceleration information for the horizontal spatial position information, f is the Coriolis force parameter, is the rotation acceleration information for the vertical spatial position information, C is the Chezy coefficient, n is the Manning coefficient, ε x is the horizontal eddy viscosity coefficient function for the horizontal spatial position information, ε y is the horizontal eddy viscosity coefficient function for the horizontal spatial position information, thereby introducing more physical parameters, realizing the improvement of the physical consistency of the results, and further increasing the accuracy and effectiveness of the generation of the ocean dynamic simulation results.

[0100] In an optional embodiment of the present application, the step of constructing a residual function through the two-dimensional plane two-dimensional hydrodynamic model comprises:

[0101] inputting the horizontal spatial position information, the vertical spatial position information, the detection time, the approximate value of the water level information, the approximate value of the average flow velocity information for the horizontal spatial position information, the approximate value of the average flow velocity information for the vertical spatial position information, the Coriolis force parameter, the still water depth information, the total water depth information, the gravity acceleration information, the horizontal eddy viscosity coefficient function for the horizontal spatial position information, the horizontal eddy viscosity coefficient function for the horizontal spatial position information and the Chezy coefficient into formula three to construct the residual function, the formula three being:

[0102]

[0103] e = e1 (x) + e2 (y) + e3 (t)

[0104] wherein e, e1, e2 and e3 are the residual functions, x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η s is an approximation of the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is an approximation of the average flow velocity information for the lateral spatial position information, is an approximation of the average flow velocity information for the vertical spatial position information, f is the Coriolis force parameter, C is the Chezy coefficient, ε x is a horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is a horizontal eddy viscosity coefficient function for the vertical spatial position information.

[0105] In a specific implementation, the lateral spatial position information, the vertical spatial position information, the detection time, the approximation of the water level information, the approximation of the average flow velocity information for the lateral spatial position information, the approximation of the average flow velocity information for the vertical spatial position information, the Coriolis force parameter, the still water depth information, the total water depth information, the gravity acceleration information, the horizontal eddy viscosity coefficient function for the lateral spatial position information, the horizontal eddy viscosity coefficient function for the vertical spatial position information and the Chezy coefficient can be input into Formula Three to construct a residual function. Exemplarily, e, e1, e2 and e3 can be the residual functions, x can be the lateral spatial position information, y can be the vertical spatial position information, t can be the detection time, η s may be an approximation of the water level information, h can be the still water depth information, H can be the total water depth information, g can be the gravity acceleration information, may be an approximation of the average flow velocity information for the lateral spatial position information, is an approximation of the average flow velocity information for the vertical spatial position information, f can be the Coriolis force parameter, C can be the Chezy coefficient, ε x may be a horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y may be a horizontal eddy viscosity coefficient function for the vertical spatial position information. By substituting e1, e2 and e3 into the formula of e, the total residual value is calculated, which represents the error between the numerical solution and the true solution. The purpose of the residual function is to measure the accuracy of the numerical model. By minimizing the residual, the numerical solution can be closer to the true physical process, which represents the total error between the numerical solution and the true solution at the spatial point (x, y) and the detection time t.

[0106] The embodiment of the present application inputs the lateral spatial position information, the vertical spatial position information, the detection time, the approximation of the water level information, the approximation of the average flow velocity information for the lateral spatial position information, the approximation of the average flow velocity information for the vertical spatial position information, the Coriolis force parameter, the still water depth information, the total water depth information, the gravity acceleration information, the horizontal eddy viscosity coefficient function for the lateral spatial position information, the horizontal eddy viscosity coefficient function for the vertical spatial position information and the Chezy coefficient into formula three to construct the residual function, and the formula three is:

[0107]

[0108] e = e1 (θ) + e2 (θ) + e3 (θ)

[0109] Wherein, e, e1, e2 and e3 are the residual function, x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η s is the approximation of the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is the approximation of the average flow velocity information for the lateral spatial position information, is the approximation of the average flow velocity information for the vertical spatial position information, f is the Coriolis force parameter, C is the Chezy coefficient, ε x is the horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is the horizontal eddy viscosity coefficient function for the vertical spatial position information, so that the accuracy of the model is evaluated by using the residual function, the model parameters are optimized, and the accuracy and effectiveness of the result generation are further improved.

[0110] In an optional embodiment of the present application, the step of determining the target weight value comprises:

[0111] Determining a learning rate and an iteration number;

[0112] Inputting the learning rate, the residual function and the iteration number into formula four to determine the target weight value, and the formula four is:

[0113]

[0114] Wherein, θ is the initial weight value, n is the iteration number, α is the learning rate, e is the residual function, θ n+1 is the target weight value.

[0115] In a specific implementation, the embodiment of the present application can determine the learning rate and the number of iterations; the learning rate, the residual function and the number of iterations are input into formula four to determine the target weight, for example, theta can be the initial weight, n can be the number of iterations, alpha can be the learning rate, e can be the residual function, theta n+1 can be the target weight, the weight formula can be updated by the gradient descent method, which is used for optimizing the neural network model, theta n+1 is the target weight, which can be the weight after the n+1 iteration, theta n is the weight after the n iteration, and the step length of each update can be controlled by the alpha learning rate, in the process of updating the parameters, if the step length is too large, the model may jump in the parameter space, which leads to the failure to converge to the optimal solution, or even divergence, if the step length is too small, the model will tend to use the existing information, and gradually approach the local optimal solution, but it may fall into a local minimum, the optimal learning rate can make the model quickly converge to the optimal solution, the learning rate can be adjusted step by step through multiple experiments to find the optimal value, and the learning rate can also be gradually reduced during training to improve the stability of convergence, which represents the gradient of the residual function e(Theta) with respect to the initial weight theta, and the most steep descent direction of the loss function at the current weight point is identified, by finding a set of optimal weight theta as the target weight, the error between the output of the model and the true value is minimized, by continuously iterating, the weight of the model will gradually adjust in the direction of minimizing the loss function, thereby improving the performance of the model, when the residual converges, the neural network output function is a two-dimensional plane two-dimensional hydrodynamic process, which can be used as the ocean dynamic simulation result for the target sea area.

[0116] In the embodiment of the present application, the learning rate and the number of iterations are determined;

[0117] The learning rate, the residual function and the number of iterations are input into formula four to determine the target weight, the formula four is:

[0118]

[0119] Wherein, theta is the initial weight, n is the number of iterations, alpha is the learning rate, e is the residual function, theta n+1 is the target weight, thereby further improving the performance of the model, optimizing the model and increasing the accuracy of the ocean dynamic simulation result.

[0120] In an optional embodiment of the present application, the physical information neural network comprises two hidden layers.

[0121] In a specific implementation, the physical information neural network in the embodiment of the present application can contain two hidden layers. Exemplarily, when the physical information neural network is a PINN, the first hidden layer can be mainly responsible for extracting low-level features from the input data, such as the relationship between simple physical quantities, and can learn the relationship between basic physical quantities such as velocity, pressure and density. The second hidden layer can further extract high-level features on the basis of the features extracted in the first layer, capture more complex physical laws and nonlinear relationships, and can learn more complex fluid phenomena such as vorticity and boundary layer. The two hidden layers provide sufficient model capacity and can learn complex physical laws. Too few hidden layers can cause model underfitting and fail to capture the complexity of the data. Too many hidden layers can cause overfitting and poor generalization ability. The first hidden layer can learn features related to physical laws, and the second hidden layer can combine these features to form higher-level physical representations. In this way, the PINN can better integrate physical information and data information.

[0122] In actual application, the two hidden layers in the physical information neural network can jointly improve the performance of the model by extracting features, introducing nonlinearity, improving model capacity and integrating physical information. Reasonable design of the structure and parameters of the hidden layers is crucial for constructing an efficient and accurate PINN model.

[0123] The embodiment of the present application further improves the accuracy and efficiency of the physical information neural network by including two hidden layers.

[0124] To better enable those skilled in the art to understand the embodiments of the present application, the embodiments of the present application are described below with a complete example.

[0125] Reference Figure 2 Fig. 1 shows a schematic diagram of a marine hydrodynamic process provided in the embodiment of the present application;

[0126] With the rapid development and construction of offshore wind farms, the complexity of the sea area where the offshore wind farm is located brings great challenges to engineering design and management. Traditional numerical simulation methods often lack accuracy when dealing with the coupling of multiple factors such as sea waves, sea currents, geology, etc. In order to more accurately evaluate the performance of offshore wind farms in complex marine environments, it is urgent to establish a more perfect numerical simulation model to determine more accurate marine dynamic simulation results. The specific process is as follows:

[0127] Collect information about the target sea area, including average flow velocity, temperature and salinity tidal current data collected in the sea area, determine the boundary and total area of the sea area.

[0128] A physical information neural network (PINN) with stronger generalization ability is used to establish a solver. The PINN is based on a deep feedforward neural network, which is composed of a series of nonlinear activation functions that are ultimately combined into output functions that can approximate any function in terms of nonlinear function optimization. Here, neural network output functions are constructed for the unknown water level, vertical flow rate, and lateral flow rate in the two-dimensional water dynamic equation in a two-dimensional plane, as shown in Equation 1 above.

[0129] Based on Figure 2 A two-dimensional water dynamic model for a sea area is established based on the topology in the schematic diagram of the ocean hydrodynamic process shown in the figure. The control equation can be as shown in Equation 2 above.

[0130] By adding physical constraints based on the two-dimensional water dynamic equation to the neural network, a data-independent residual term can be constructed. Here, a continuous time format can be used, and the residual function can be constructed using the model shown in Equation 2 above, as shown in Equation 3 above.

[0131] The closer the value of the residual sum is to zero, the closer the approximate solution of the PINN is to the true solution.

[0132] Then, the value of the residual can be reduced by adjusting the weights θ, so that the output function of the neural network approximates the two-dimensional water dynamic equation in a two-dimensional plane. A gradient algorithm suitable for convex optimization can be used, and the basic form can be as shown in Equation 4 above.

[0133] Preferably, the Adam (Adaptive Moment Estimation) algorithm can be selected. The Adam algorithm uses an adaptive learning rate to improve learning efficiency and introduces momentum to enhance algorithm stability, making it more suitable for complex non-convex optimization problems. Further, when local convergence problems occur, global optimization algorithms such as GA (Genetic Algorithm) and PSO (Particle Swarm Optimization) can be used. After the residual converges, the output function of the neural network is the two-dimensional water dynamic process in a two-dimensional plane.

[0134] By introducing a physical information neural network, partial differential equations that describe physical processes are incorporated into the loss function of the neural network. Under the condition of very few data, complex physical systems can be accurately learned and simulated. Not only can the rules of the sea hydrodynamic process be quickly extracted from limited monitoring data, but the model prediction results can also be highly consistent with the actual physical process, thereby providing strong support for fields such as marine engineering and marine environmental monitoring.

[0135] It should be noted that, for the method embodiments, the series of acts / combinations thereof are described for simplicity, but those skilled in the art should understand that the present embodiments are not limited to the acts described, as some of the acts could be performed in other sequences or even at the same time in accordance with the present embodiments. In addition, those skilled in the art should understand that the embodiments described in the specification are all preferred embodiments, and the acts involved are not necessarily essential to the present embodiments.

[0136] Referring to Figure 3 , a structural block diagram of a device for generating a marine dynamic simulation result for a target sea area is shown, which can specifically include the following modules:

[0137] The sea area information collection module 301 is configured to collect sea area information for the target sea area.

[0138] The physical information neural network construction module 302 is configured to construct a physical information neural network.

[0139] The neural network output function determination module 303 is configured to determine a neural network output function for representing water level and flow rate based on the sea area information through the physical information neural network.

[0140] The two-dimensional plane two-dimensional water dynamic model output module 304 is configured to construct a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function.

[0141] The residual function construction module 305 is configured to construct a residual function through the two-dimensional plane two-dimensional water dynamic model.

[0142] The target weight determination module 306 is configured to determine a target weight, adjust the residual function based on the target weight, and generate a marine dynamic simulation result for the target sea area.

[0143] For the device embodiments, the description is relatively simple because they are basically similar to the method embodiments, and the relevant parts refer to the part of the description of the method embodiments.

[0144] In addition, the present embodiments also provide an electronic device, which includes a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements each process of the above-mentioned method for generating a marine dynamic simulation result for a target sea area, and achieves the same technical effects. To avoid repetition, no further description is given here.

[0145] The embodiment of the present application further provides a computer readable storage medium, and the computer readable storage medium stores a computer program, the computer program is executed by a processor to realize each process of the method for generating the ocean dynamic simulation result of a target sea area and achieve the same technical effects. To avoid repetition, details are not described herein. The computer readable storage medium includes a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0146] Figure 4 A hardware structure schematic diagram of an electronic device for implementing various embodiments of the present application.

[0147] The electronic device 400 includes, but is not limited to, a radio frequency unit 401, a network module 402, an audio output unit 403, an input unit 404, a sensor 405, a display unit 406, a user input unit 407, an interface unit 408, a memory 409, a processor 410, and a power supply 411, etc. Those skilled in the art can understand that the electronic device structure shown in the figure is not a limitation on the electronic device, and the electronic device can include more or fewer components than shown, or combine certain components, or different component arrangements. In the embodiments of the present application, the electronic device includes, but is not limited to, a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle terminal, a wearable device, and a pedometer, etc. Figure 4 The electronic device structure shown in the figure is not a limitation on the electronic device, and the electronic device can include more or fewer components than shown, or combine certain components, or different component arrangements. In the embodiments of the present application, the electronic device includes, but is not limited to, a mobile phone, a tablet computer, a notebook computer, a palm computer, a vehicle terminal, a wearable device, and a pedometer, etc.

[0148] It should be understood that in the embodiments of the present application, the radio frequency unit 401 can be used for receiving and sending signals in the process of information or call. Specifically, after receiving the downlink data from the base station, the processor 410 processes it. In addition, the uplink data is sent to the base station. Generally, the radio frequency unit 401 includes, but is not limited to, an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc. In addition, the radio frequency unit 401 can also communicate with the network and other devices through a wireless communication system.

[0149] The electronic device provides wireless broadband Internet access for users through the network module 402, such as helping users to send and receive emails, browse web pages and access streaming media, etc.

[0150] The audio output unit 403 can convert audio data received by the radio frequency unit 401 or the network module 402 or stored in the memory 409 into an audio signal and output it as a sound. Moreover, the audio output unit 403 can also provide audio output related to a specific function performed by the electronic device 400 (e.g., a call signal receiving sound, a message receiving sound, etc.). The audio output unit 403 includes a speaker, a buzzer, and a receiver, etc.

[0151] The input unit 404 is configured to receive audio or video signals. The input unit 404 can include a graphic processing unit (GPU) 4041 and a microphone 4042. The graphic processing unit 4041 processes image data of a still picture or a video obtained by an image capture apparatus (e.g., a camera) in a video capture mode or an image capture mode. Processed image frames can be displayed on the display unit 406. Processed image frames can be stored in the memory 409 (or other storage medium) or transmitted via the radio frequency unit 401 or the network module 402. The microphone 4042 can receive sound, and can process such sound as audio data. Processed audio data can be converted into a format that can be transmitted to a mobile communication base station via the radio frequency unit 401 in a telephone call mode.

[0152] The electronic device 400 further includes at least one sensor 405, such as a light sensor, a motion sensor, and other sensors. Specifically, the light sensor includes an ambient light sensor and a proximity sensor, wherein the ambient light sensor can adjust the brightness of the display panel 4061 according to the brightness of ambient light, and the proximity sensor can turn off the display panel 4061 and / or the backlight when the electronic device 400 is moved to the ear. As one of the motion sensors, the accelerometer sensor can detect the magnitude of acceleration in each direction (generally three axes), and can detect the magnitude and direction of gravity when at rest, and can be used to identify the electronic device posture (such as screen switching, related games, magnetometer posture calibration), vibration recognition related functions (such as pedometer, knock), and the like. The sensor 405 can also include a fingerprint sensor, a pressure sensor, an iris sensor, a molecular sensor, a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, and the like, which will not be described here.

[0153] The display unit 406 is configured to display information input by a user or information provided to the user. The display unit 406 can include a display panel 4061, which can be configured in the form of a liquid crystal display (LCD), an organic light-emitting diode (OLED), or the like.

[0154] The user input unit 407 can be used to receive inputted digital or character information, and to generate key signal input related to user settings of the electronic device and control of functions. Specifically, the user input unit 407 includes a touch panel 4071 and other input devices 4072. The touch panel 4071, also called a touch screen, can collect a user's touch operation (such as a user's operation on or near the touch panel 4071 using a finger, a stylus, or any suitable object or accessory) on or near it. The touch panel 4071 can include two parts, a touch detection device and a touch controller. The touch detection device detects the user's touch position and detects a signal caused by the touch operation, and transmits the signal to the touch controller; the touch controller receives the touch information from the touch detection device, converts it into touch coordinates, and sends it to the processor 410, receives commands from the processor 410 and executes them. In addition, the touch panel 4071 can be implemented in various types such as resistive, capacitive, infrared, and surface acoustic wave. In addition to the touch panel 4071, the user input unit 407 can also include other input devices 4072. Specifically, the other input devices 4072 can include, but are not limited to, a physical keyboard, function keys (such as volume control buttons, on / off buttons, etc.), trackballs, mice, joysticks, and the like, which will not be described here.

[0155] Further, the touch panel 4071 can be overlaid on the display panel 4061, and when the touch panel 4071 detects a touch operation on or near it, it transmits to the processor 410 to determine the type of touch event, and then the processor 410 provides corresponding visual output on the display panel 4061 according to the type of touch event. Although in the Figure 4 In some embodiments, the touch panel 4071 and the display panel 4061 can be integrated to realize the input and output functions of the electronic device, which is not limited here.

[0156] The interface unit 408 is an interface for connecting external devices to the electronic device 400. For example, the external devices can include a wired or wireless headset port, an external power supply (or battery charger) port, a wired or wireless data port, a memory card port, a port for connecting a device having an identification module, an audio input / output (I / O) port, a video I / O port, a headphone port, and the like. The interface unit 408 can be used to receive input (e.g., data information, power, etc.) from external devices and transmit the received input to one or more elements within the electronic device 400, or can be used to transmit data between the electronic device 400 and external devices.

[0157] The memory 409 can be used to store software programs and various data. The memory 409 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, application programs required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; and the data storage area can store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), and the like. In addition, the memory 409 can include a high-speed random access memory, and can also include a nonvolatile memory, for example, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state memory device.

[0158] The processor 410 is the control center of the electronic device, connects all parts of the electronic device through various interfaces and lines, performs various functions of the electronic device and processes data by running or executing software programs and / or modules stored in the memory 409 and calling data stored in the memory 409, and thus performs overall monitoring of the electronic device. The processor 410 can include one or more processing units; preferably, the processor 410 can integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, and application programs, and the like, and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor can also not be integrated into the processor 410.

[0159] The electronic device 400 can also include a power supply 411 (such as a battery) for supplying power to various components; preferably, the power supply 411 can be logically connected to the processor 410 through a power management system, so as to realize functions such as management of charging, discharging, and power consumption management through the power management system.

[0160] In addition, the electronic device 400 includes some functional modules that are not shown and will not be described here.

[0161] It should be noted that in this document, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or apparatus including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such a process, method, article or apparatus. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the process, method, article or apparatus including the element.

[0162] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0163] like Figure 5 As shown, in another embodiment of the present invention, a computer-readable storage medium 501 is also provided, which stores instructions that, when run on a computer, cause the computer to execute the method for generating marine dynamic simulation results for a target sea area as described in the above embodiment.

[0164] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

[0165] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0166] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0167] In the embodiments of the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the embodiments of the apparatus described above are merely schematic, and the division of the units is merely logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0168] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments of the present application.

[0169] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can be physically present separately, or two or more units can be integrated into one unit.

[0170] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the present application that essentially contributes to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, and various storage media that can store program codes.

[0171] The above description is merely a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for generating a result of a simulation of ocean dynamics for a target sea area, characterized by, The method comprises the following steps: collecting sea area information of the target sea area; constructing a physical information neural network; determining a neural network output function for representing water level and flow rate based on the sea area information through the physical information neural network; constructing a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function; constructing a residual function through the two-dimensional plane two-dimensional water dynamic model; determining a target weight, adjusting the residual function based on the target weight, and generating a marine dynamic simulation result for the target sea area; the sea area information comprises average flow rate information, sea area boundary information, water level information, and total area information, and the average flow rate information comprises flow rate information; the step of determining a neural network output function for representing water level and flow rate based on the sea area information through the physical information neural network comprises: determining horizontal spatial position information and vertical spatial position information of the target sea area according to the sea area boundary information and the total area information; determining an initial weight of an activation function of the physical information neural network and a detection time; inputting the initial weight, the detection time, the average flow rate information, the horizontal spatial position information, and the vertical spatial position information into Formula 1 to determine the neural network output function, wherein the Formula 1 is: η(x, y, t) ~ η s (x, y, t; θ u ) wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, Θ is the initial weight value, η is the water level information, η s is an approximation of the water level information, u is the flow velocity information, is the average flow velocity information for the lateral spatial position information, is an approximation of the average flow velocity information for the lateral spatial position information, is the average flow velocity information for the vertical spatial position information, is an approximation of the average flow velocity information for the vertical spatial position information. the step of constructing a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function comprises: obtaining static water depth information, total water depth information, gravity acceleration information, rotation acceleration information, a Charnock coefficient, a Coriolis force parameter, and a Manning coefficient of the target sea area; determining a horizontal eddy viscosity coefficient function of the horizontal spatial position information and the vertical spatial position information; inputting the static water depth information, the total water depth information, the gravity acceleration information, the rotation acceleration information, the Charnock coefficient, the Coriolis force parameter, the Manning coefficient, the horizontal eddy viscosity coefficient function, the detection time, the vertical spatial position information, the horizontal spatial position information, and the average flow rate information into Formula 2 to determine the two-dimensional plane two-dimensional water dynamic model, wherein the Formula 2 is: wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η is the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is the average flow velocity information for the lateral spatial position information, is the average flow velocity information for the vertical spatial position information, is the self-propelled acceleration information for the lateral spatial position information, f is the Coriolis force parameter, is the self-propelled acceleration information for the vertical spatial position information, C is the Chezy coefficient, n is the Manning coefficient, ε x is the horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is the horizontal eddy viscosity coefficient function for the lateral spatial position information.

2. The method of claim 1, wherein, the step of constructing a residual function through the two-dimensional plane two-dimensional water dynamic model comprises: inputting horizontal spatial position information, the vertical spatial position information, the detection time, an approximate value of the water level information, an approximate value of the average flow rate information of the horizontal spatial position information, an approximate value of the average flow rate information of the vertical spatial position information, the Coriolis force parameter, the static water depth information, the total water depth information, the gravity acceleration information, the horizontal eddy viscosity coefficient function of the horizontal spatial position information, the horizontal eddy viscosity coefficient function of the vertical spatial position information, and the Charnock coefficient into Formula 3 to construct the residual function, wherein the Formula 3 is: e=e1(θ)+e2(θ)+e3(θ) wherein e, e1, e2, and e3 are the residual functions, x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η s is an approximation of the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is an approximation of the average flow velocity information for the lateral spatial position information, is an approximation of the average flow velocity information for the vertical spatial position information, f is the Coriolis force parameter, C is the Chezy coefficient, ε x is a horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is a horizontal eddy viscosity coefficient function for the vertical spatial position information.

3. The method of claim 2, wherein, the step of determining a target weight comprises: determining a learning rate and an iteration number; inputting the learning rate, the residual function and the iteration number into formula four to determine the target weight, the formula four being: wherein θ is the initial weight, n is the iteration number, a is the learning rate, e is the residual function, θ n+1 is the target weight.

4. The method according to any one of claims 1 to 3, characterized in that, The physical information neural network comprises two hidden layers.

5. A device for generating a result of a simulation of ocean dynamics for a target sea area, characterized in that The method comprises: a sea area information collection module configured to collect sea area information for the target sea area; a physical information neural network construction module configured to construct a physical information neural network; a neural network output function determination module configured to determine, based on the sea area information, a neural network output function for representing water level and flow velocity by using the physical information neural network; a two-dimensional plane two-dimensional water dynamic model output module configured to construct a two-dimensional plane two-dimensional water dynamic model based on the sea area information and the neural network output function; a residual function construction module configured to construct a residual function by using the two-dimensional plane two-dimensional water dynamic model; a target weight determination module configured to determine a target weight, adjust the residual function based on the target weight, and generate a marine dynamic simulation result for the target sea area; The sea area information comprises average flow velocity information, sea area boundary information, water level information and total area information, and the average flow velocity information comprises flow velocity information. The neural network output function determination module is further configured to determine horizontal spatial position information and vertical spatial position information for the target sea area according to the sea area boundary information and the total area information. determine an initial weight and a detection time for an activation function of the physical information neural network; input the initial weight, the detection time, the average flow velocity information, the horizontal spatial position information and the vertical spatial position information into formula one to determine the neural network output function, the formula one being: η(x, y, t) ~ η s (x, y, t; θ u ) wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, Θ is the initial weight value, η is the water level information, η S is an approximation of the water level information, u is the flow velocity information, is the average flow velocity information for the lateral spatial position information, is an approximation of the average flow velocity information for the lateral spatial position information, is the average flow velocity information for the vertical spatial position information, is an approximation of the average flow velocity information for the vertical spatial position information. The two-dimensional plane two-dimensional water dynamic model output module is further configured to obtain still water depth information, total water depth information, gravity acceleration information, rotation acceleration information, a Chezy coefficient, a Coriolis force parameter and a Manning coefficient for the target sea area; determine a horizontal eddy viscosity coefficient function for the horizontal spatial position information and the vertical spatial position information; input the still water depth information, the total water depth information, the gravity acceleration information, the rotation acceleration information, the Chezy coefficient, the Coriolis force parameter, the Manning coefficient, the horizontal eddy viscosity coefficient function, the detection time, the vertical spatial position information, the horizontal spatial position information and the average flow velocity information into formula two to determine the two-dimensional plane two-dimensional water dynamic model, the formula two being: wherein x is the lateral spatial position information, y is the vertical spatial position information, t is the detection time, η is the water level information, h is the still water depth information, H is the total water depth information, g is the gravity acceleration information, is the average flow velocity information for the lateral spatial position information, is the average flow velocity information for the vertical spatial position information, is the self-propelled acceleration information for the lateral spatial position information, f is the Coriolis force parameter, is the self-propelled acceleration information for the vertical spatial position information, C is the Chezy coefficient, n is the Manning coefficient, ε x is the horizontal eddy viscosity coefficient function for the lateral spatial position information, ε y is the horizontal eddy viscosity coefficient function for the lateral spatial position information.

6. An electronic device, comprising: The device comprises a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory complete communication with each other through the communication bus; The memory is configured to store a computer program; The processor is configured to execute the program stored on the memory to implement the method of any one of claims 1-4.

7. A computer readable storage medium having instructions stored thereon that, when executed by one or more processors, cause the processors to perform the method of any one of claims 1-4.

Images