Ship pipe network pressure and flow prediction method and system, medium and terminal

Abstract

According to the pressure and flow prediction method for the ship pipe network, firstly, a pipeline of the ship pipe network is divided into a plurality of pipe sections and a plurality of nodes, and the nodes comprise internal nodes and reserve nodes; secondly, collecting flow data and pressure data of each pipe section and each node in a preset time period, and obtaining a corresponding relation between time parameters in the preset time period and the flow data and the pressure data; training in a neural network according to the corresponding relation between the time parameter data and the flow data and the pressure data to obtain a neural network model; and finally, inputting instantaneous time parameters of the ship pipe network to be predicted into the neural network model to obtain instantaneous flow data and instantaneous pressure data of the ship pipe network corresponding to the instantaneous time parameters. The pressure data and the flow data of the ship pipe network can be obtained by only depending on less measurement data, and finally the global prediction of the transient pressure and the instantaneous flow of the ship pipe network is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of ship design, and particularly to a method, system, medium and terminal for predicting the pressure and flow rate of a ship pipe network. Background Art

[0002] In the ship industrial system, the ship pipeline network, as a key facility for transporting various media such as water, fuel oil, lubricating oil, and gas, is of self-evident importance. This complex pipe network system not only concerns the normal operation of the ship, but also directly relates to the safety performance of the ship. In order to ensure the operation efficiency and safety of the pipe network, operators must precisely control the pumps and valves of the system and closely monitor the changes in system pressure. However, it is not easy to achieve a comprehensive monitoring of the instantaneous pressure of the ship pipe network.

[0003] Currently, the measurement of the instantaneous pressure of the ship pipe network mainly relies on sensing devices. However, limited by the high cost and physical space constraints, the sensing devices cannot be installed in every area of the pipe network and can only be deployed at specific key positions. This results in blind spots in the monitoring of the overall pressure state of the pipe network and cannot comprehensively and real-time reflect the actual pressure distribution of the pipe network.

[0004] Due to the intricate layout of the ship pipe network, which includes a large number of structural parameters such as elbows, reducers, and tees, the diversity and complexity of these parameters pose great challenges to the manual calculation of the instantaneous pressure of the pipe network system. Traditional fluid calculation methods, such as the finite element method, finite difference method, and finite volume method, although to a certain extent can simulate the fluid flow and pressure distribution in the pipe network, these methods require discretizing the computational domain into a set of grid points and solving complex fluid mechanics problems at the grid points. This not only requires establishing detailed models of the pipeline structure parameters and the performance parameters of pumps and valves, but also needs to calculate the pressure and flow rate of each part in the system by iteratively solving the coefficient matrix.

[0005] However, in practical applications, due to the large variety and quantity of the structural parameters involved in the ship pipe network, the modeling process is often time-consuming and laborious. In addition, fine grid division is required when dealing with complex partial differential equation systems, which leads to a sharp increase in the amount of computational grids. Multiple iterative calculations not only take a long time and are slow, but also, limited by the approximation of numerical methods and the accumulation of calculation errors, the accuracy is often difficult to guarantee. Summary of the Invention

[0006] In view of the problems existing in the above-mentioned prior art, the present application provides a method, system, medium and terminal for predicting the pressure and flow rate of a ship pipe network, which can accurately predict the instantaneous flow rate and instantaneous pressure of the ship pipe network.

[0007] To achieve the above object and other related objects, on the one hand, the present invention provides a method for predicting the pressure and flow rate of a ship pipeline network, including the following steps:

[0008] Divide the pipeline of the ship pipeline network into multiple pipe segments and multiple nodes, and the nodes include internal nodes and reserve nodes;

[0009] During a predetermined time period, collect the flow rate data and pressure data of each pipe segment and node, and obtain the corresponding relationship between the time parameters within the predetermined time period and the flow rate data and the pressure data;

[0010] Train a neural network model based on the corresponding relationship between the time parameter data and the flow rate data and the pressure data in the neural network;

[0011] Input the instantaneous time parameters of the ship pipeline network to be predicted into the neural network model to obtain the instantaneous flow rate data and instantaneous pressure data of the ship pipeline network corresponding to the instantaneous time parameters.

[0012] Optionally, dividing the pipeline of the ship pipeline network into multiple pipe segments and multiple nodes further includes:

[0013] Obtain the pipe length and pipe diameter of each pipe segment;

[0014] Obtain the height of each internal node and reserve node. The internal node is the node where different pipe segments in the ship pipeline network are connected to each other, and the reserve node is the node in the ship pipeline network connected to the external environment;

[0015] Obtain the friction coefficient between the pipeline and water.

[0016] Optionally, before training a neural network model based on the corresponding relationship between the time parameter data and the flow rate data and the pressure data in the neural network, it includes:

[0017] Embed an elastic water column model into the neural network. The elastic water column model includes the ordinary differential equation of the fluid momentum equation and the ordinary differential equation of the continuity equation. The ordinary differential equation of the fluid motion equation is:

[0018]

[0019] The ordinary differential equation of the continuity equation is:

[0020]

[0021] where t is the time parameter, h I is the pressure data of the internal node, h I =[h1, h2, …, h n T h R is the pressure data of the reserve node, h​R = [h n+1 , h n+2 , …, h n+R T , where q is the flow rate data, q = [q1, q2, …, q m T , L is the inductance matrix, L(j, j) = L j (j = 1, 2, …, m), R is the resistance matrix, R(j, j) = R j (j = 1, 2, …, m), C is the capacitance vector matrix, C(j, j) = C j (j = 1, 2, …, m), D I is the discharge matrix, D I (i, i) = d i (i = 1, 2, …, n), A I and A R are the incidence matrices of the ship pipeline network, A (i,j) represents the topological structure relationship between the pipe segment and the node at the i-th row and j-th column of the incidence matrix. When the i-th node is at the starting point of the j-th pipe segment, the value of A (i,j) is 1; when the i-th node is at the end point of the j-th pipe segment, the value of A (i,j) is -1; in other cases, the value of A (i,j) is 0;

[0022] The calculation formulas for the inductance matrix L, the resistance matrix R, the capacitance vector matrix C, and the discharge matrix D I are as follows:

[0023]

[0024] where l is the pipeline length, D is the pipeline diameter, a is the medium flow velocity, A is the pipeline cross-sectional area, f is the friction coefficient between the pipeline and water, C d is the pipeline outflow efficiency, A d is the effective area when the fluid flows out of the pipeline, and g is the acceleration due to gravity.

[0025] Optionally, within a predetermined time period, collect the flow rate data and pressure data of each pipe segment and node, and obtain the corresponding relationship between the time parameters and the flow rate data and pressure data within the predetermined time period. It further includes:

[0026] Establish a training set, a test set, and a validation set based on the time parameters and the corresponding flow rate data and pressure data. The training set is used to train the neural network model, the validation set is used to ensure that the neural network model can be generalized to different working conditions, and the test set is used to finally evaluate the performance of the neural network model.

[0027] ​​Optionally, a neural network model is obtained through training in a neural network according to the correspondence relationship between time parameter data, flow rate data, and pressure data, including the following steps:

[0028] Taking the time parameter data as the input value and the flow rate data and pressure data as the output values, a physics-informed neural network model is established in the neural network;

[0029] Substituting the training set, test set, and validation set into the physics-informed neural network model and training until the physics-informed neural network model converges to obtain the neural network model.

[0030] Optionally, a training set, a test set, and a validation set are established according to the time parameter and the corresponding flow rate data and pressure data, including:

[0031] Converting the time parameter, flow rate data, and pressure data into matrix form to form a total data set;

[0032] Randomly sampling from the total data set to establish a training set, a test set, and a validation set.

[0033] Optionally, substituting the training set, test set, and validation set into the physics-informed neural network model and training until the physics-informed neural network model converges to obtain the neural network model, including the following steps:

[0034] Taking the time parameter as the input and the flow rate data and pressure data as the output, a loss function is constructed

[0035] LOSS = MSE u +MSE f , where MSE u is the data loss term and MSE f is the differential equation loss;

[0036] Using the gradient descent algorithm to adjust the weight parameters θ of the neural network, and iterating multiple times until convergence to obtain the neural network model.

[0037] On the other hand, the present invention provides a pressure and flow rate prediction system for a ship pipe network, including:

[0038] A pipeline division module for dividing the pipelines of the ship pipe network into multiple pipe segments and multiple nodes, where the nodes include internal nodes and reserve nodes;

[0039] An acquisition module for collecting the flow rate data and pressure data of each pipe segment and node within a predetermined time period, and obtaining the correspondence relationship between the time parameter and the flow rate data and pressure data within the predetermined time period;

[0040] A model training module, which is used to train a neural network model in a neural network according to the correspondence relationship between time parameter data, flow data, and pressure data.

[0041] An input inference module, which is used to input the instantaneous time parameters of the ship pipe network to be predicted into the neural network model to obtain the instantaneous flow data and instantaneous pressure data of the ship pipe network corresponding to the instantaneous time parameters.

[0042] On the other hand, the present invention provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, the pressure and flow prediction method of the ship pipe network described above is implemented.

[0043] On the other hand, the present invention provides a terminal, including:

[0044] A memory, which is used to store a computer program;

[0045] A processor, which is used to execute the computer program stored in the memory so that the terminal executes the pressure and flow prediction method of the ship pipe network described above.

[0046] As described above, a pressure and flow prediction method, system, medium, and terminal for a ship pipe network provided by the present invention at least have the following beneficial technical effects:

[0047] In the pressure and flow prediction method for a ship pipe network of the present invention, first, the pipeline of the ship pipe network is divided into multiple pipe segments and multiple nodes, and the nodes include internal nodes and reserve nodes. Then, within a predetermined time period, the flow data and pressure data of each pipe segment and node are collected, and the correspondence relationship between the time parameters within the predetermined time period and the flow data and the pressure data is obtained. Next, a neural network model is trained in a neural network according to the correspondence relationship between the time parameter data, the flow data, and the pressure data. Finally, the instantaneous time parameters of the ship pipe network to be predicted are input into the neural network model to obtain the instantaneous flow data and instantaneous pressure data of the ship pipe network corresponding to the instantaneous time parameters. It realizes obtaining the pressure data and flow data of the ship pipe network only relying on fewer measurement data, and finally realizes the global prediction of the transient pressure and instantaneous flow of the ship pipe network. This method not only has a simple calculation method and high calculation efficiency, but also has strong scalability. Only by modifying the corresponding instantaneous time parameters, it can realize the accurate prediction of the instantaneous pressure and instantaneous flow of pipelines with different ship types, different systems, and different media, greatly improving the reliability and efficiency of ship pipeline design. Description of the Drawings

[0048] Figure 1 It shows a schematic flow chart of the pressure and flow prediction method for a ship pipe network provided in Embodiment 1 of the present invention.

[0049] Figure 2 It shows a schematic diagram of the modules of the pressure and flow prediction system for a ship pipeline network provided in the second embodiment of the present invention.

[0050] Figure 3 It shows a schematic diagram of the structure of a terminal provided in the fourth embodiment of the present invention. Specific embodiments

[0051] The following uses specific specific examples to illustrate the implementation manners of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific implementation manners. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention.

[0052] It should be noted that the diagrams provided in this embodiment only illustrate the basic concept of the present invention in a schematic manner. Although only the components related to the present invention are shown in the diagrams and are not drawn according to the number, shape, and size of the components in actual implementation, the forms, quantities, positional relationships, and proportions of the components in actual implementation can be arbitrarily changed on the premise of implementing the technical solution of the present invention, and the component layout form may also be more complex.

[0053] Embodiment 1

[0054] This embodiment provides a method for predicting the pressure and flow rate of a ship pipeline network. Referring to Figure 1 , the method includes the following steps:

[0055] S100: Divide the pipeline of the ship pipeline network into multiple pipe segments and multiple nodes, where the nodes include internal nodes and reserve nodes;

[0056] Specifically, the pipelines of the ship pipeline network are divided into multiple pipe segments and multiple nodes. Nodes are key points in the pipeline network system, usually representing connection points, branch points, intersection points or terminal points of the pipelines. For example, positions such as the starting point, ending point, tee joint, valve, pump, etc. of the pipeline can be regarded as nodes. A pipe segment is the pipeline part between two nodes. In an alternative embodiment of this embodiment, when dividing nodes and pipe segments, first, according to the actual layout of the pipeline network system, all key points (such as connection points, branch points, valves, pumps, etc.) are determined as nodes. After all nodes are determined, the adjacent nodes are connected by pipe segments. For example, the pipeline between node 1 and node 2 can be divided into one pipe segment, and the pipeline between node 2 and node 3 can be divided into another pipe segment, and so on. After dividing the pipe segments and nodes, obtain the pipeline length, pipeline diameter of each pipe segment of the ship pipeline network, the height of each node, and the friction coefficient between the pipeline and water. Nodes include internal nodes and reserve nodes. An internal node is a node of different pipe segments in the ship pipeline network. The internal node is located inside the ship pipeline network and is used to describe the flow and pressure transmission of the fluid between the pipelines. A reserve node refers to a node in the ship pipeline network that is connected to the external environment. The reserve node includes at least an inlet node, an outlet node of the fluid, and a node connected to the atmosphere.

[0057] Taking the cooling water pipeline network system as an example, the pipelines of the cooling water pipeline network system are divided into 27 pipe segments and 18 nodes. Respectively obtain the pipeline length, pipeline diameter of each pipe segment of the ship pipeline network, the height of each node, and the friction coefficient between the pipeline and water. There are sensors at the 8th node and the 16th node in the cooling water pipeline network system, which can detect the pressure and flow rate of the medium at the node positions.

[0058] S200: During a predetermined time period, collect the flow rate data and pressure data of each of the pipe segments and the nodes, and obtain the correspondence relationship between the time parameter within the predetermined time period and the flow rate data and the pressure data;

[0059] Specifically, further, time parameters during the operation of the ship pipe network for a period of time, as well as the corresponding flow rate data and pressure data, are obtained to obtain the pressure data matrix M and the flow rate data matrix N of each pipe section and node at different times. The matrices are preprocessed to form the total data set Φ, and then random sampling is performed to establish a training set, a test set, and a validation set. The training set is used to train the neural network model, the validation set is used to ensure that the neural network model can be generalized to different working conditions, and the test set is used to finally evaluate the performance of the neural network model. Still taking the cooling water pipe network system as an example, first, the cooling water pipe network system is turned on, and the relationship between the flow rate data and the pressure data of 18 nodes of the pipeline of the cooling water pipe network system changing with the time parameter within 10 s is recorded. The flow rate data, the pressure data, and the time parameter are converted into matrix form to obtain the pressure number matrix M and the flow rate value matrix N of each node at different times. The matrices are preprocessed to form the total data set Φ, and then random sampling is performed to establish a training set, a test set, and a validation set. Due to the limitations of detection equipment, in the actual pipe network layout, limited by factors such as construction costs, the number of sensors arranged is small, so only limited flow rate data and pressure data can be obtained through measurement.

[0060] S300: Train a neural network model in the neural network according to the corresponding relationship between the time parameter data, the flow rate data, and the pressure data;

[0061] First, embed the elastic water column model into the neural network. The elastic water column model includes the ordinary differential equation of the fluid momentum equation and the ordinary differential equation of the continuity equation. The ordinary differential equation of the fluid motion equation is:

[0062]

[0063] The ordinary differential equation of the continuity equation is:

[0064]

[0065] where t is the time parameter, h I is the pressure data of the internal node, h I =[h1, h2, …, h n T , h R is the pressure data of the reserve node, h R =[h n+1 , h n+2 , …, h n+R T , q is the flow rate data, q = [q1, q2, …, q m T .

[0066] In the fluid momentum equation and the continuity equation, L is the inductance matrix, L(j, j) = L​​​j (j = 1, 2, …, m), R is a resistance matrix, R(j, j) = R j (j = 1, 2, …, m), C is a capacitance vector matrix, C(j, j) = C j (j = 1, 2, …, m), D I is a discharge matrix, D I (i, i) = d i (i = 1, 2, …, n), A I and A R represent the incidence matrix of the ship pipeline network, A I is an n×m matrix, A R is an r×m matrix, A (i,j) represents the topological structure relationship between the pipe segment and the node at the i-th row and j-th column of the incidence matrix, A (i,j) The value of A (i,j) is 1, -1 or 0. Specifically, when the i-th node is at the starting point of the j-th pipe segment, A (i,j) takes the value of 1; when the i-th node is at the end point of the j-th pipe segment, A (i,j) takes the value of -1; in other cases, A (i,j) takes the value of 0. The calculation formulas for the inductance matrix, resistance matrix, capacitance vector matrix and discharge matrix are:

[0067]

[0068] where, l is the pipeline length, D is the pipeline diameter, a is the medium flow velocity, A is the pipeline cross-sectional area, f is the friction coefficient between the pipeline and water, C d is the pipeline outflow efficiency, A d is the effective area when the fluid flows out of the pipeline, and g is the acceleration due to gravity.

[0069] After embedding the elastic water column model into the neural network, with the time parameter as the input and the flow rate data and pressure data corresponding to the time parameter as the output, a physics-informed neural network model is established in the neural network. Since the elastic water column model is embedded in the neural network, the physics-informed neural network model not only learns data but also learns physical laws, enabling the physics-informed neural network model to maintain a high prediction accuracy even when the data is insufficient. According to the elastic water column model, the loss function LOSS = MSE u + MSE f , MSE u represents the data loss term, which is used to measure the difference between the predicted value of the physics-informed neural network model and the actual measurement data. MSE fIt represents the loss term of the differential equation, which is used to measure whether the model prediction values satisfy the fluid momentum equation and the continuity equation of the elastic water column model. Further, the training set data, the test set data, and the validation set data are substituted into the physics-informed neural network model for training. With the help of the automatic differentiation technique, the initial data loss and the differential equation loss are calculated, and the gradient descent algorithm is used to adjust the weight parameters θ of the neural network. The training process is repeated multiple times, and the weight parameters are adjusted each time to minimize the loss function. When the loss function reaches the preset threshold or the change rate is less than a certain value, it is considered that the physics-informed neural network model converges, and finally, a neural network model regarding the time parameter data, the pipe segment flow rate data, and the pressure data is obtained.

[0070] S400: Input the instantaneous time parameter of the ship pipe network to be predicted into the neural network model to obtain the instantaneous flow rate data and the instantaneous pressure data of the ship pipe network corresponding to the instantaneous time parameter.

[0071] When the corresponding instantaneous time parameter is input into the physics-informed neural network model, the instantaneous flow rate data and the instantaneous pressure data of the ship pipe network corresponding to the instantaneous time parameter can be obtained.

[0072] The pressure and flow rate prediction method for the ship pipe network of the present application integrates the complex topological structure of the pipe network system into the neural network, and then combines the physics-informed neural network model with the elastic water column model. The training process of the neural network model is completed through the flow rate data and the pressure data of a finite number of pipelines. The finally generated neural network model is based on the measured data of the real flow field. It realizes obtaining the pressure data and the flow rate data of the ship pipe network only relying on less measurement data, and finally realizes the global prediction of the transient pressure and the instantaneous flow rate of the ship pipe network. This method not only has a simple calculation method and high calculation efficiency, but also has strong scalability. Only by modifying the corresponding instantaneous time parameter, it can accurately predict the instantaneous pressure and the instantaneous flow rate of the pipelines of different ship types, different systems, and different media, greatly improving the reliability and efficiency of the ship pipeline design.

[0073] Embodiment 2

[0074] This embodiment provides a pressure and flow rate prediction system for a ship pipe network. Refer to Figure 2, the pressure and flow prediction system of the ship pipe network includes a pipeline differentiation module, a collection module, a model training module, and an input inference module. The pipeline differentiation module is used to divide the pipelines of the ship pipe network into multiple pipe segments and multiple nodes, and the nodes include internal nodes and reserve nodes. The collection module is used to collect the flow data and pressure data of each pipe segment and node within a predetermined time period, and obtain the corresponding relationship between the time parameters and the flow data and pressure data within the predetermined time period. The model training module is used to train a neural network model in the neural network according to the corresponding relationship between the time parameter data and the flow data and pressure data. The input inference module is used to input the instantaneous time parameters of the ship pipe network to be predicted into the neural network model to obtain the instantaneous flow data and instantaneous pressure data of the ship pipe network corresponding to the instantaneous time parameters.

[0075] Embodiment III

[0076] This embodiment provides a storage medium. A computer program is stored on the storage medium of this embodiment, and when the program is executed by a processor, it implements the pressure and flow prediction method of the ship pipe network described in Embodiment I. The storage medium includes: various media such as ROM, RAM, magnetic disk, USB flash drive, memory card, or optical disc that can store program codes.

[0077] Embodiment IV

[0078] This embodiment provides a terminal, as Figure 3 shown. The terminal of this embodiment includes a memory and a processor. The memory is used to store a computer program. Preferably, the memory includes: various media such as ROM, RAM, magnetic disk, USB flash drive, memory card, or optical disc that can store program codes. The processor is connected to the memory and is used to execute the computer program stored in the memory so that the terminal executes the pressure and flow prediction method of the ship pipe network described in Embodiment I. Preferably, the processor can be a general-purpose processor, including a central processing unit (Central Processing Unit, abbreviated as CPU), a network processor (Network Processor, abbreviated as NP), etc.; it can also be a digital signal processor (Digital Signal Processor, abbreviated as DSP), an application specific integrated circuit (Application Specific Integrated Circuit, abbreviated as ASIC), a field programmable gate array (Field Programmable Gate Array, abbreviated as FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0079] The above embodiments are only illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical idea disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. A method for predicting pressure and flow of a ship pipe network, characterized in that: The following steps are involved: Dividing the pipeline of the ship pipeline network into a plurality of pipeline sections and a plurality of nodes, wherein the nodes include internal nodes and reserve nodes; Collecting flow data and pressure data of each pipe segment and each node within a predetermined time period, and obtaining a corresponding relationship between a time parameter within the predetermined time period and the flow data and the pressure data; Training a neural network model in a neural network according to the corresponding relationship between the time parameter data and the flow data and the pressure data; The instantaneous time parameters of the ship pipeline network that need to be predicted are input into the neural network model to obtain instantaneous flow data and instantaneous pressure data of the ship pipeline network corresponding to the instantaneous time parameters.

2. The method for predicting pressure and flow of a ship pipe network according to claim 1, characterized in that: The pipeline of the ship pipeline network is divided into a plurality of pipeline sections and a plurality of nodes, and further includes: Obtaining the pipeline length and pipeline diameter of each of the pipeline sections; Acquire the height of each of the internal nodes and the reserve nodes, the internal nodes being nodes where different pipe sections in the ship pipe network are connected to each other, and the reserve nodes being nodes in the ship pipe network connected to the external environment; The friction coefficient between the pipeline and water is obtained.

3. The method for predicting pressure and flow of a ship pipe network according to claim 1, characterized in that: Before training a neural network model according to the corresponding relationship between the time parameter data and the flow data and the pressure data, the method includes: The elastic water column model is embedded into the neural network. The elastic water column model includes ordinary differential equations of fluid momentum equation and ordinary differential equations of continuity equation. The ordinary differential equation of fluid motion equation is: The ordinary differential equation of the continuity equation is: Among them, t is the time parameter, h I is the pressure data of the internal node, h I =[h1,h2,…,h n ] T ,h R is the pressure data of the reserve node, h R =[h n+1 ,h n+2 ,…,h n+R ] T , q is the flow data, q=[q1,q2,…,q m ] T , L is the inductance matrix, L(j,j)=L j (j=1,2,…,m), R is the resistance matrix, R(j,j)=R j (j=1,2,…,m), C is the capacitance vector matrix, C(j,j)=C j (j=1,2,…,m),D I is the discharge matrix, D I (i,i)=d i (i=1,2,…,n), A I and A R is the association matrix of the ship pipe network, A (i,j) Represents the topological relationship between the pipe segment and the node in the i-th row and j-th column of the association matrix. When the i-th node is at the starting point of the j-th pipe segment, A (i,j) The value of is 1; when the i-th node is at the end of the j-th pipe segment, A (i,j) The value of is -1; in other cases A (i,j) The value of is 0; The inductance matrix L, the resistance matrix R, the capacitance vector matrix C and the discharge matrix D I The calculation formula is: Where, l is the length of the pipeline, D is the diameter of the pipeline, a is the flow rate of the medium, A is the cross-sectional area of ​​the pipeline, f is the friction coefficient between the pipeline and water, C d is the pipeline outflow efficiency, A d is the effective area when the fluid flows out of the pipe, and g is the acceleration due to gravity.

4. The method for predicting pressure and flow of a ship pipe network according to claim 1, characterized in that: In a predetermined time period, the flow data and the pressure data of each pipe section and each node are collected, and the corresponding relationship between the time parameter in the predetermined time period and the flow data and the pressure data is obtained, and further comprising: A training set, a test set and a validation set are established according to the time parameters and the flow data and pressure data corresponding to the time parameters. The training set is used to train the neural network model, the validation set is used to ensure that the neural network model can be generalized to different working conditions, and the test set is used to ultimately evaluate the performance of the neural network model.

5. The method for predicting pressure and flow rate of a ship pipe network according to claim 4, characterized in that: The neural network model is obtained by training in a neural network according to the corresponding relationship between the time parameter data and the flow data and the pressure data, comprising the following steps: Taking the time parameter data as input values ​​and the flow data and the pressure data as output values, a physical information neural network model is established in the neural network; The training set, the test set and the validation set are substituted into the physical information neural network model and trained until the physical information neural network model converges to obtain a neural network model.

6. The method for predicting pressure and flow rate of a ship pipe network according to claim 4, characterized in that: A training set, a test set and a validation set are established according to the time parameter and the flow data and pressure data corresponding to the time parameter, including: Converting the time parameter, the flow data and the pressure data into a matrix form to form a total data set; Random sampling is performed in the total data set to establish the training set, the test set and the validation set.

7. The method for predicting pressure and flow rate of a ship pipe network according to claim 5, characterized in that: Substituting the training set, the test set and the validation set into the physical information neural network model for training until the physical information neural network model converges to obtain a neural network model, comprising the following steps: Construct a loss function with time parameters as input and flow data and pressure data as output LOSS=MSE u +MSE f , where MSE u is the data loss term, MSE f is the loss term of the differential equation; The weight parameter θ of the neural network is adjusted using a gradient descent algorithm, and multiple iterations are performed until convergence to obtain the neural network model.

8. A pressure and flow prediction system for a ship pipe network, characterized in that: include: A pipeline distinguishing module, used for dividing the pipeline of the ship pipeline network into a plurality of pipeline sections and a plurality of nodes, wherein the nodes include internal nodes and reserve nodes; A collection module, used for collecting flow data and pressure data of each pipe segment and each node within a predetermined time period, and obtaining a corresponding relationship between a time parameter within the predetermined time period and the flow data and the pressure data; A model training module, used for training a neural network model in a neural network according to the corresponding relationship between the time parameter data and the flow data and the pressure data; The input reasoning module is used to input the instantaneous time parameters of the ship pipeline network that need to be predicted into the neural network model to obtain the instantaneous flow data and instantaneous pressure data of the ship pipeline network corresponding to the instantaneous time parameters.

9. A storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the pressure and flow prediction method of the ship pipe network according to any one of claims 1 to 7 is implemented.

10. A terminal, characterized in that: include: Memory for storing computer programs; A processor is used to execute the computer program stored in the memory so that the terminal executes the pressure and flow prediction method for the ship pipeline network according to any one of claims 1 to 7.

Images