Method, device and equipment for scour prediction of submarine cable landing end and storage medium

Abstract

The application discloses a scour prediction method, device and equipment for a submarine cable mud entry end and a storage medium. The submarine cable mud entry end prediction model is trained by taking the flow rate, attack angle and mud entry angle of the mud entry end as input parameters and taking the scour extension length and time scale as output parameters. After collecting the real-time flow rate, attack angle and mud entry angle of the submarine cable mud entry end, the collected data is input into the submarine cable mud entry end prediction model, and the prediction values of the scour extension length and time scale of the submarine cable mud entry end can be obtained. On the one hand, after training, the submarine cable mud entry end prediction model can capture the complex nonlinear relationship between the flow rate, attack angle, mud entry angle and scour extension length and time scale. Compared with fluid mechanics formula or empirical relationship, the prediction result is more accurate. On the other hand, the scour extension length and time scale can be synchronously output, the inconsistency and calculation complexity of training models are avoided, and the prediction efficiency and engineering application convenience are improved.

Description

Technical Field

[0001] This invention relates to the field of marine engineering technology, and in particular to a method, apparatus, equipment, and storage medium for predicting scour at the mud-entry end of a submarine cable. Background Technology

[0002] Submarine cables are usually buried deep in the seabed. When they reach the vicinity of wind turbine piles or offshore substations, they need to be gradually raised from the seabed to connect to the electrical circuit of the piles, thus forming a section of the cable that enters the mud. Once the section of the cable that enters the mud is suspended due to scouring, it will sway and dynamic response under the load of complex marine environment, resulting in wear, fatigue damage and even insulation layer damage to the cable.

[0003] Currently, existing technologies for predicting scour at the mud end of submarine cables mainly focus on predicting the scour depth of horizontally laid pipelines in shallow sea areas. They can only predict a single scour parameter (such as scour depth) and cannot meet the requirement of simultaneous prediction of multiple target parameters such as scour extension length and time scale at the mud end of the submarine cable. Moreover, the models mostly use simplified fluid dynamics formulas or empirical relationships, and do not fully consider the nonlinear coupling effect between multiple variables such as flow velocity, angle of attack, and mud entry angle, resulting in insufficient prediction accuracy. Summary of the Invention

[0004] This invention provides a method, apparatus, equipment, and storage medium for predicting the scour at the mud-entry end of a submarine cable, so as to achieve high-precision, multi-target synchronous prediction of the scour extension length and time scale of the mud-entry end of the submarine cable.

[0005] In a first aspect, the present invention provides a method for predicting scour at the mud-entry end of a submarine cable, comprising: Acquire scour experimental data and construct a training dataset. Each set of data in the training dataset includes input parameters and corresponding output parameters. The input parameters include flow velocity, angle of attack, and angle of entry into the mud. The output parameters include scour extension length and time scale. A dual feedforward neural network model is constructed, and the training dataset is used to train the dual feedforward neural network model to obtain a trained prediction model for the mud entry point of the submarine cable. The prediction model for the mud entry point of the submarine cable is used to output the scour extension length and time scale. Collect real-time scour impact parameters at the entry point of the submarine cable into the mud, including flow velocity, angle of attack, and angle of entry into the mud; The actual scour impact parameters at the site are input into the prediction model of the submarine cable entry point in the mud to obtain the predicted values ​​of the scour propagation length and time scale of the submarine cable entry point in the mud.

[0006] Secondly, the present invention provides a device for predicting scour at the mud-entry end of a submarine cable, comprising: The training dataset construction module is used to acquire scour experimental data and construct a training dataset. Each set of data in the training dataset includes input parameters and corresponding output parameters. The input parameters include flow velocity, angle of attack, and mud penetration angle. The output parameters include scour extension length and time scale. The training module is used to construct a dual feedforward neural network model. The dual feedforward neural network model is trained using the training dataset to obtain a trained prediction model for the mud entry point of the submarine cable. The prediction model for the mud entry point of the submarine cable is used to output the scour extension length and time scale. The on-site real data acquisition module is used to collect on-site real scour influence parameters at the mud entry end of the submarine cable. The on-site real scour influence parameters include flow velocity, angle of attack, and mud entry angle. The prediction module is used to input the actual scour impact parameters from the field into the prediction model of the submarine cable entry point in the mud to obtain the predicted values ​​of the scour extension length and time scale of the submarine cable entry point in the mud.

[0007] Thirdly, the present invention provides an electronic device, the electronic device comprising: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the scour prediction method for the mud-entry end of the submarine cable as described in the first aspect of the invention.

[0008] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement the scour prediction method for the mud-entry end of a submarine cable as described in the first aspect of the present invention.

[0009] This invention trains a prediction model for submarine cable entry points in mud using the flow velocity, angle of attack, and entry angle at the entry point as input parameters, and the scour propagation length and time scale as output parameters. This allows the prediction model to learn to predict the scour propagation length and time scale based on the flow velocity, angle of attack, and entry angle at the entry point. After collecting the actual flow velocity, angle of attack, and entry angle at the entry point, the collected data is input into the prediction model to obtain predicted values ​​for the scour propagation length and time scale. On one hand, through training with data, the prediction model can capture the complex nonlinear relationships between flow velocity, angle of attack, entry angle, and scour propagation length and time scale. Compared to traditional techniques that use fluid dynamics formulas or empirical relationships, the prediction results are more accurate. On the other hand, it can simultaneously output the scour propagation length and time scale, avoiding the inconsistencies and computational complexity of training models separately, thus improving prediction efficiency and ease of engineering application.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of a method for predicting the scour of the mud-entry end of a submarine cable provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a submarine cable; Figure 3 This is a schematic diagram of a dual feedforward neural network model; Figure 4 This is a schematic diagram of the structure of a scour prediction device for the mud entry end of a submarine cable provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0014] Figure 1 This is a flowchart illustrating a method for predicting scour at the mud-entry end of a submarine cable according to an embodiment of the present invention. This embodiment is applicable to scenarios involving scour prediction at the mud-entry end of a submarine cable. The method can be executed by a scour prediction device for the mud-entry end of a submarine cable. This device can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the scour prediction method for the mud-entry end of the submarine cable includes: S101. Obtain scour experimental data and construct a training dataset. Each set of data in the training dataset includes input parameters and corresponding output parameters. The input parameters include flow velocity, angle of attack, and mud entry angle. The output parameters include scour extension length and time scale.

[0015] like Figure 2 As shown, the mud entry end refers to the transition section where the submarine cable 1 transitions from the seabed buried state (deep buried section 2) to the suspended state and connects to the wind turbine pile foundation 3 or the offshore substation. In this section, the submarine cable gradually rises from the seabed and connects to the pile foundation electrical circuit via a J-shaped pipe or a bell mouth.

[0016] In one embodiment, multiple sets of scour test data can be obtained. Each set of scour test data includes flow velocity U, angle of attack α, angle of penetration β, scour propagation length Le, and time scale Ts. Flow velocity U is the velocity of the ocean current near the penetration end, measured in meters per second (m / s), and is the main hydrodynamic parameter affecting cable scour. Angle of attack α is the angle between the direction of the ocean current and the direction of the cable axis, measured in degrees (°), reflecting the direction of the ocean current's effect on the cable. Angle of penetration β is the angle at which the cable emerges from the seabed, measured in degrees (°), reflecting the cable's burial pattern.

[0017] Scour extension length Le refers to the maximum axial extension of the suspended area formed by scour at the end of the submarine cable entering the mud, which is the largest extension of the cable along the cable axis. It is a key indicator for assessing the safety status of the submarine cable. Time scale Ts refers to the time required for scour to develop from the beginning to reach a certain stable state or a specific suspended length. It is a unit of seconds (s) and is used to assess the speed of scour development.

[0018] The scouring test data comes from the water tank scouring test. For example, 60 sets of scouring test data can be obtained through the water tank test. In the test, by controlling the water flow speed in the water tank, adjusting the angle between the submarine cable model and the water flow direction, and the angle at which the submarine cable passes through the mud surface, the scouring extension length under different working conditions and the time scale required to reach the scouring state are recorded.

[0019] In an optional embodiment of the present invention, after obtaining the scouring test data, the data is further standardized, specifically including the following sub-steps: S1011, calculate the mean μ and standard deviation σ of each type of parameter in multiple sets of scour experimental data.

[0020] The mean values ​​of the input parameters flow velocity U, angle of attack α, and mud penetration angle β are denoted as μ. X The standard deviation is denoted as σ. X The mean values ​​of the output parameters, scour extension length Le and time scale Ts, are denoted as μ. Y The standard deviation is denoted as σ. Y .

[0021] It should be noted that the mean and standard deviation of the flow velocity U, angle of attack α, and angle of entry into the mud β are calculated respectively to obtain the mean and standard deviation of the flow velocity U, the angle of attack α, and the angle of entry into the mud β. Similarly, the mean and standard deviation of the scour extension length Le and the time scale Ts are calculated respectively to obtain the mean and standard deviation of the scour extension length Le and the time scale Ts.

[0022] S1012, using the Z-score standardization method, the input and output parameters of each set of scour experimental data are converted into standardized data with a mean of 0 and a standard deviation of 1, respectively, according to the following formula: ; ; Where X represents the input parameters (flow velocity, angle of attack, angle of entry into mud), and Y represents the output parameters (scour extension length Le, time scale Ts).

[0023] S1013 uses standardized multiple sets of scouring experimental data to construct a training dataset.

[0024] The input parameters for each training dataset include standardized flow velocity, angle of attack, and angle of entry into the mud, while the output parameters include standardized scour spread length and time scale.

[0025] This embodiment uses Z-score standardization to transform all data into a distribution with a mean of 0 and a standard deviation of 1. During model training, this accelerates convergence and makes the distribution of inputs to each layer of the network relatively stable, which is beneficial for gradient optimization algorithms to quickly find the optimal solution. At the same time, it avoids large numbers of parameters dominating the loss function and ensures that the model learns the relationship between all input features and the output target equally. In addition, standardization is based on the statistics of the training set, ensuring that the model learns relative change patterns rather than absolute values, thereby enhancing its adaptability to new data.

[0026] S102. Construct a dual feedforward neural network model and train the dual feedforward neural network model using the training dataset to obtain a trained prediction model for the entry point of the submarine cable into the mud. The prediction model for the entry point of the submarine cable into the mud is used to output the scour extension length and time scale.

[0027] In an optional embodiment, step S102 may include the following sub-steps: S1021, Construct a dual feedforward neural network model consisting of an input layer, a first hidden layer, a second hidden layer, and an output layer connected in sequence.

[0028] like Figure 3 As shown, the dual feedforward neural network model includes an input layer L0, a first hidden layer L1, a second hidden layer L2, and an output layer L3 connected in sequence. The input layer contains 3 neurons, corresponding to the three input parameters of flow velocity, angle of attack, and mud entry angle, respectively. The first hidden layer contains 10 neurons, the second hidden layer contains 5 neurons, and the output layer contains 2 neurons, corresponding to the two output parameters of scour extension length and time scale, respectively.

[0029] The first and second hidden layers use the hyperbolic tangent activation function, while the output layer uses a linear activation function. The range of the hyperbolic tangent activation function is (-1, 1), which can effectively introduce nonlinear mapping capabilities, enabling the model to fit complex nonlinear relationships. The output layer uses a linear activation function (purelin) to ensure that the output value can cover any real number range and is not limited by the range of the activation function.

[0030] Compared to single-hidden-layer networks, the dual-hidden-layer structure has a stronger nonlinear fitting ability and can capture the complex coupling relationship between flow velocity, angle of attack, mud penetration angle and scour extension length and time scale. Compared to deep networks with more than three layers, this structure has a moderate number of parameters (total weights are 3×10+10×5+5×2=90), and can effectively avoid overfitting under limited experimental data (e.g., 60 sets).

[0031] S1022, the standardized training dataset is randomly divided into a training set and a validation set according to a preset ratio.

[0032] In this embodiment, the standardized training dataset is randomly divided into a training set, a validation set, and a test set according to a preset ratio. For example, the dataset is divided into training, validation, and test sets in a ratio of 75%, 10%, and 15%, respectively. The training set is used to update the model's weights and biases. The validation set is used to evaluate the model's generalization ability in real time and independently during training. It does not participate in weight updates but is used to monitor the training status and trigger the early stopping mechanism. The test set is used to evaluate the model's final prediction performance after training is completed.

[0033] S1023, the standardized training set is input into the dual feedforward neural network model, and after propagation through the input layer, the first hidden layer, and the second hidden layer, the output layer outputs the standardized predicted values ​​of the scour extension length and time scale.

[0034] For each training data point, the standardized flow velocity U, angle of attack α, and angle of entry into the mud β are input into the neurons of the input layer L0, and the forward propagation calculation process is as follows: The input to the first hidden layer L1 is the standardized input parameter X. norm (Flow velocity U, angle of attack α, angle of entry β), after linear transformation and nonlinear activation, the output is: H1=tanh(W1×X norm +b1); Where W1 is the weight matrix of the first hidden layer and b1 is the bias vector of the first hidden layer.

[0035] The output of the second hidden layer L2 is: H2 = tanh(W2 × H1 + b2); Where W2 is the weight matrix of the second hidden layer and b2 is the bias vector of the second hidden layer.

[0036] The output of output layer L3 is: Y pred,norm =W3×H2+b3; Where W3 is the weight matrix of the output layer and b3 is the bias vector of the output layer.

[0037] S1024 calculates the error between the predicted value and the standardized output parameters in the training set, which is used as the loss value.

[0038] In this embodiment, the loss function is the mean squared error (MSE): ; Where n represents the sample data in the training set, Y pred,norm,i Y is the standardized predicted value for the i-th sample. true,norm,iThe standardized true value of the i-th sample can be used as an example. Alternatively, the standardized predicted value can be destandardized and then the mean squared error loss function can be used to calculate the loss value.

[0039] In addition, since the model outputs two parameters simultaneously, namely the scour extension length and the time scale, the mean square error of the two output parameters can be calculated separately, and then the average or weighted sum of the two can be taken as the final total loss value.

[0040] S1025 employs the Levenberg-Marquardt backpropagation algorithm to update the model parameters of the dual feedforward neural network model based on the loss value, and performs iterative training with the goal of minimizing the loss value.

[0041] The Levenberg-Marquardt algorithm is an improvement on Newton's method, and its parameter update formula is as follows: Δw=-(J T J+λI) -1 J T e; Where w is the model parameter vector, J is the Jacobian matrix (containing the partial derivatives of the loss function with respect to each parameter), e is the error vector, λ is the damping factor, and I is the identity matrix. The Levenberg-Marquardt algorithm switches between gradient descent and Gauss-Newton methods by dynamically adjusting the damping factor λ. When the current solution is far from the optimal solution (λ is large), the algorithm behaves like gradient descent, robustly approaching the optimal region. When it is close to the optimal solution (λ is small), the algorithm switches to Gauss-Newton method to achieve fast convergence.

[0042] The Levenberg-Marquardt algorithm is suitable for fast and accurate training on small to medium-sized datasets (such as the 60 sets of experimental data in this embodiment). Its convergence speed is much better than that of traditional gradient descent algorithms, and it can achieve high prediction accuracy in fewer iterations.

[0043] S1026 After each iteration of training is completed, the validation set is input into the dual feedforward neural network model after the current iteration of training is completed, and the loss value of the dual feedforward neural network model on the validation set is calculated.

[0044] Specifically, the validation set is input into the dual feedforward neural network model after the current iteration training is completed to obtain the predicted value, and the loss value is calculated by using the above mean squared error loss calculation method through the predicted value and the true value.

[0045] S1027 When the loss value of the dual feedforward neural network model on the validation set no longer decreases or begins to increase in consecutive iterations, the early stopping mechanism is triggered, the training is terminated and the model parameters corresponding to the minimum loss value on the validation set are saved, and the current dual feedforward neural network model is used as the trained prediction model for the mud entry end of the submarine cable.

[0046] During training, the training error usually decreases continuously, but the validation error shows a typical curve of decreasing rapidly at first, then leveling off, and finally possibly increasing slightly. When the validation error starts to rise, it indicates that the model may begin to fit the noise in the training set, i.e., overfitting has occurred. The early stopping mechanism automatically stops training at the point where the validation error is at its lowest and saves the optimal model, ensuring that the final model has the best generalization performance.

[0047] In an optional embodiment, a test set is also provided. After training, the test set data is input into the trained model to evaluate the model's predictive performance. The evaluation metrics include the correlation coefficient R and the root mean square error RMSE. ; ; Y actual Y represents the actual value of the test set. pred These are the model's predicted values. and These are the mean of the actual value and the predicted value, respectively, where n is the number of samples in the test set, and Y is the mean of the predicted value. actual,i Y is the actual value of the i-th sample. pred,i Let be the predicted value of the i-th sample. The correlation coefficient R and root mean square error RMSE mentioned above are calculated based on the numerical values ​​of the physical dimensions after inverse standardization.

[0048] The following table shows the results of testing the model using the test set: As shown in the table above, the correlation coefficient and root mean square error of the two parameters predicted by the model, namely the scour extension length and the scour time scale, are 0.98 and 0.89, and 2.94 and 487.4, respectively. The predicted values ​​and the measured values ​​have a very strong correlation and the error is also very small, which proves that the model works well.

[0049] S103. Collect the actual scour impact parameters at the entry point of the submarine cable into the mud. The actual scour impact parameters include flow velocity, angle of attack, and angle of entry into the mud.

[0050] Specifically, the actual scouring parameters at the entry point of the submarine cable into the mud are collected, including the current velocity U (m / s), angle of attack α (°), and entry angle β (°). For example, the current velocity can be obtained by on-site measurement in the sea area near the entry point using an acoustic Doppler current profiler (ADCP), the angle of attack can be obtained by measuring the angle between the direction of the ocean current and the direction of the cable axis, and the entry angle can be calculated by obtaining the topographic data of the cable exit point through a multibeam echo sounder.

[0051] S104. Input the actual scour impact parameters from the field into the prediction model of the submarine cable entry point into the mud to obtain the predicted values ​​of the scour propagation length and time scale of the submarine cable entry point into the mud.

[0052] First, based on the mean and standard deviation of each type of parameter, the flow velocity, angle of attack, and angle of penetration in the actual field scour parameters are standardized to obtain the standardized actual flow velocity U. norm Angle of attack α norm and the angle of entry into the mud β norm For specific methods of standardization processing, please refer to step S1012.

[0053] Secondly, the standardized on-site actual flow velocity U norm Angle of attack α norm and the angle of entry into the mud β norm Inputting the data into the prediction model for the subsea cable entry point into the mud yields the scour propagation length Le at the entry point of the subsea cable. pred,norm and time scale Ts pred,norm .

[0054] Finally, the predicted values ​​of scour propagation length and time scale are denormalized to obtain the actual scour propagation length and actual time scale at the mud entry point of the submarine cable.

[0055] The formula for destandardization is as follows: ; Among them, Y pred This is the actual predicted value, Y pred,norm It is the standardized predicted value output by the model, σ train μ train These are the standard deviation and mean of each parameter in the training data, respectively: Le pred =Le pred,norm ×σ Le +μ Le ; Ts pred =Ts pred,norm ×σ Ts +μ Ts ; σ Le and μLe These represent the standard deviation and mean of the flush spread length in the training dataset, respectively, σ Ts and μ Ts These represent the standard deviation and mean of the time scale in the training dataset, respectively.

[0056] The final actual scour extension length Le at the mud entry point of the submarine cable was obtained. pred (Unit: m) and actual time scale Ts pred Predicted values ​​(unit: s).

[0057] In this embodiment, the standardized predicted value is a dimensionless abstract value that engineers cannot directly understand and use. Through destandardization, the predicted value is restored to an actual value with physical units (meters and seconds), so that the prediction results can be directly applied to engineering design and risk assessment. For example, based on the predicted scour extension length, engineers can determine whether the suspended span of the submarine cable exceeds the safety threshold and whether protective measures such as rock dumping or installing bend limiters are required. Based on the predicted time scale, engineers can assess the speed of scour development and determine whether there is a sufficient safety window for protective construction before the typhoon season or ocean current changes.

[0058] In a specific application example, in a certain offshore wind farm project, before the flood season each year, the operation and maintenance personnel collect data on the flow velocity, angle of attack, and angle of entry into the mud near the mud entry point of each wind turbine pile foundation. The data are then input into the prediction model trained in this embodiment of the invention to obtain the predicted values ​​of the scour propagation length and time scale of the submarine cable at each mud entry point. For submarine cable sections where the prediction results show that the scour propagation length is close to the safety threshold, rock-filling protection construction is arranged in advance. For submarine cable sections where the prediction results show that the time scale is short (the scour development is rapid), the monitoring frequency is increased.

[0059] This invention trains a prediction model for submarine cable entry points in mud using the flow velocity, angle of attack, and entry angle at the entry point as input parameters, and the scour propagation length and time scale as output parameters. This allows the model to learn the mapping relationship between the three input parameters and the two output parameters. After collecting the actual flow velocity, angle of attack, and entry angle at the entry point of the submarine cable, these values ​​are input into the prediction model to obtain predicted values ​​for the scour propagation length and time scale. On one hand, through training with data, the prediction model can capture the complex nonlinear relationship between flow velocity, angle of attack, entry angle, scour propagation length, and time scale. Compared to traditional techniques that use fluid dynamics formulas or empirical relationships, the prediction results are more accurate. On the other hand, it can simultaneously output the scour propagation length and time scale, avoiding the inconsistency and computational complexity of training models separately, thus improving prediction efficiency and ease of engineering application.

[0060] Figure 4This is a schematic diagram of a scour prediction device for the mud-entry end of a submarine cable, provided as an embodiment of the present invention. Figure 4 As shown, the scour prediction device for the mud-entry end of the submarine cable includes: The training dataset construction module 401 is used to acquire scour experimental data and construct a training dataset. Each set of data in the training dataset includes input parameters and corresponding output parameters. The input parameters include flow velocity, angle of attack and mud entry angle. The output parameters include scour extension length and time scale. Training module 402 is used to construct a dual feedforward neural network model. The dual feedforward neural network model is trained using the training dataset to obtain a trained prediction model for the mud entry point of the submarine cable. The prediction model for the mud entry point of the submarine cable is used to output the scour extension length and time scale. The on-site real data acquisition module 403 is used to collect on-site real scour influence parameters of the submarine cable entry point into the mud. The on-site real scour influence parameters include flow velocity, angle of attack and angle of entry into the mud. The prediction module 404 is used to input the actual scour impact parameters on site into the prediction model of the submarine cable entry point in the mud to obtain the predicted values ​​of the scour extension length and time scale of the submarine cable entry point in the mud.

[0061] The scour prediction device for the mud-entry end of the submarine cable provided in this embodiment of the invention can execute the scour prediction method for the mud-entry end of the submarine cable provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0062] Example 4 Figure 5 A schematic diagram of an electronic device 40 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0063] like Figure 5As shown, the electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory (ROM) 42 or a random access memory (RAM) 43, communicatively connected to the at least one processor 41. The memory stores computer programs executable by the at least one processor. The processor 41 can perform various appropriate actions and processes based on the computer program stored in the ROM 42 or loaded from storage unit 48 into the RAM 43. The RAM 43 may also store various programs and data required for the operation of the electronic device 40. The processor 41, ROM 42, and RAM 43 are interconnected via a bus 44. An input / output (I / O) interface 45 is also connected to the bus 44.

[0064] Multiple components in electronic device 40 are connected to I / O interface 45, including: input unit 46, such as keyboard, mouse, etc.; output unit 47, such as various types of monitors, speakers, etc.; storage unit 48, such as disk, optical disk, etc.; and communication unit 49, such as network card, modem, wireless transceiver, etc. Communication unit 49 allows electronic device 40 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0065] Processor 41 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 41 performs the various methods and processes described above, such as the scour prediction method for the mud-entry end of a submarine cable.

[0066] In some embodiments, the method for predicting scour at the entry point of a submarine cable can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 40 via ROM 42 and / or communication unit 49. When the computer program is loaded into RAM 43 and executed by processor 41, one or more steps of the method for predicting scour at the entry point of a submarine cable described above can be performed. Alternatively, in other embodiments, processor 41 can be configured to perform the method for predicting scour at the entry point of a submarine cable by any other suitable means (e.g., by means of firmware).

[0067] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0068] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0069] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0070] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0071] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0072] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0073] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0074] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for predicting scour at the mud-entry end of a submarine cable, characterized in that, include: Acquire scour experimental data and construct a training dataset. Each set of data in the training dataset includes input parameters and corresponding output parameters. The input parameters include flow velocity, angle of attack, and angle of entry into the mud. The output parameters include scour extension length and time scale. A dual feedforward neural network model is constructed, and the training dataset is used to train the dual feedforward neural network model to obtain a trained prediction model for the mud entry point of the submarine cable. The prediction model for the mud entry point of the submarine cable is used to output the scour extension length and time scale. Collect real-time scour impact parameters at the entry point of the submarine cable into the mud, including flow velocity, angle of attack, and angle of entry into the mud; The actual scour impact parameters at the site are input into the prediction model of the submarine cable entry point in the mud to obtain the predicted values ​​of the scour propagation length and time scale of the submarine cable entry point in the mud.

2. The method according to claim 1, characterized in that, Obtain scouring experiment data and construct a training dataset, including: Multiple sets of scour experimental data were obtained, each set of scour experimental data including flow velocity, angle of attack, angle of entry into mud, scour propagation length and time scale; Calculate the mean and standard deviation of each type of parameter in multiple sets of scouring experimental data; Based on the mean and standard deviation of each type of parameter, the parameters of the corresponding type in each set of scour experimental data are Z-score standardized to obtain standardized scour experimental data. A training dataset is constructed using the standardized multiple sets of scour experimental data. The input parameters of each set of training data include the standardized flow velocity, angle of attack, and angle of entry into the mud, and the output parameters include the standardized scour extension length and time scale.

3. The method according to claim 2, characterized in that, Constructing a dual feedforward neural network model includes: Construct a dual feedforward neural network model consisting of an input layer, a first hidden layer, a second hidden layer, and an output layer connected in sequence; The input layer contains three neurons, which are used to input three input parameters: flow velocity, angle of attack, and angle of entry into the mud. The first hidden layer contains 10 neurons, the second hidden layer contains 5 neurons, and the output layer contains 2 neurons, which are used to output two output parameters: the scour extension length and the time scale, respectively. The first and second hidden layers use the hyperbolic tangent activation function, and the output layer uses the linear activation function.

4. The method according to claim 3, characterized in that, The dual feedforward neural network model is trained using the training dataset to obtain a trained prediction model for the entry point of the submarine cable into the mud, including: The standardized training dataset is randomly divided into a training set and a validation set according to a preset ratio; The standardized training set is input into the dual feedforward neural network model. After propagation through the input layer, the first hidden layer, and the second hidden layer, the output layer outputs the standardized predicted values ​​of the scour extension length and time scale. The error between the predicted value and the standardized output parameters in the training set is calculated and used as the loss value. The Levenberg-Marquardt backpropagation algorithm is used to update the model parameters of the dual feedforward neural network model based on the loss value, and iterative training is performed with the goal of minimizing the loss value.

5. The method according to claim 4, characterized in that, The dual feedforward neural network model is trained using the training dataset to obtain a trained prediction model for the entry point of the submarine cable into the mud, which also includes: After each iteration of training is completed, the validation set is input into the dual feedforward neural network model after the current iteration of training is completed, and the loss value of the dual feedforward neural network model on the validation set is calculated. When the loss value of the dual feedforward neural network model on the validation set no longer decreases or begins to increase in consecutive iterations, the early stopping mechanism is triggered, the training is terminated and the model parameters corresponding to the minimum loss value on the validation set are saved, and the current dual feedforward neural network model is used as the trained prediction model for the mud entry point of the submarine cable.

6. The method according to any one of claims 2-5, characterized in that, The actual scour impact parameters at the site are input into the prediction model for the subsea cable entry point into the mud to obtain predicted values ​​for the scour propagation length and time scale of the subsea cable entry point into the mud, including: Based on the mean and standard deviation of each type of parameter, the flow velocity, angle of attack and angle of entry into the actual field scour influence parameters are standardized to obtain the standardized actual flow velocity, angle of attack and angle of entry into the field. The standardized on-site flow velocity, angle of attack, and angle of entry into the mud are input into the prediction model of the mud entry end of the submarine cable to obtain the predicted values ​​of the scour propagation length and time scale of the mud entry end of the submarine cable. The predicted values ​​of the scour extension length and time scale are denormalized to obtain the actual scour extension length and actual time scale of the submarine cable at the mud entry point.

7. The method according to claim 6, characterized in that, The formula for destandardization is as follows: ； Among them, Y pred This is the actual predicted value, Y pred,norm It is the standardized predicted value output by the model, σ train μ train These are the standard deviation and mean of each parameter in the training data, respectively.

8. A device for predicting scour at the mud-entry end of a submarine cable, characterized in that, include: The training dataset construction module is used to acquire scour experimental data and construct a training dataset. Each set of data in the training dataset includes input parameters and corresponding output parameters. The input parameters include flow velocity, angle of attack, and mud penetration angle. The output parameters include scour extension length and time scale. The training module is used to construct a dual feedforward neural network model. The dual feedforward neural network model is trained using the training dataset to obtain a trained prediction model for the mud entry point of the submarine cable. The prediction model for the mud entry point of the submarine cable is used to output the scour extension length and time scale. The on-site real data acquisition module is used to collect on-site real scour influence parameters at the mud entry end of the submarine cable. The on-site real scour influence parameters include flow velocity, angle of attack, and mud entry angle. The prediction module is used to input the actual scour impact parameters from the field into the prediction model of the submarine cable entry point in the mud to obtain the predicted values ​​of the scour extension length and time scale of the submarine cable entry point in the mud.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the scour prediction method for the mud-entry end of the submarine cable as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for predicting the scour at the mud-entry end of a submarine cable as described in any one of claims 1-7.

Images