Method and system for predicting transient overvoltage of direct-current submarine cable
By optimizing hyperparameters using the CNN-BiLSTM-Attention model and the symbiotic organism search algorithm, the problem of multi-dimensional influencing factors in the prediction of transient overvoltage of DC submarine cables was solved, achieving high-precision and stable prediction results, which are applicable to complex engineering practices of different submarine cable types and marine environments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-15
- Publication Date
- 2026-03-27
AI Technical Summary
Existing methods for predicting transient overvoltages in DC submarine cables suffer from large computational loads, poor adaptability, limited prediction accuracy, and poor real-time performance. They also fail to fully consider multi-dimensional influencing factors, making it difficult to meet the needs of practical engineering applications.
By employing a CNN-BiLSTM-Attention model combined with a symbiotic organism search algorithm, multi-dimensional key features are extracted, a dataset is constructed, and hyperparameters are optimized to predict transient overvoltages in DC submarine cables.
It improves prediction accuracy and environmental adaptability, enhances the model's generalization ability and robustness, and can output stable and reliable results in complex scenarios.
Smart Images
Figure CN121743873A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of power system overvoltage prediction, and mainly relates to a prediction method and system for transient overvoltage of a DC submarine cable. BACKGROUND
[0002] With large-scale development and grid connection of offshore wind power and other new energy, the safe and stable operation of the DC submarine cable, as a key carrier for long-distance and large-capacity power transmission, is crucial to the reliability of the power system. During the operation of the DC submarine cable, due to the inherent characteristics of the DC transmission system, transient overvoltage often occurs when various faults or operations occur. The transient overvoltage of the DC submarine cable mainly occurs in the following three situations: first, when the DC system fails to commutate, second, when the DC system fails to lock, and third, when the AC fault occurs in the vicinity of the DC system. Due to the special structure and parameter characteristics of the DC submarine cable system, the waveform and amplitude of the transient overvoltage in these situations are significantly different from those of the transient overvoltage of the ultra-high voltage DC converter station.
[0003] The existing prediction methods for transient overvoltage of the DC submarine cable mainly include: 1. Electromagnetic transient simulation method: based on electromagnetic transient theory, a refined model of the DC submarine cable system is established, and simulation tools are used to simulate the overvoltage generation and propagation process under the conditions of commutation failure, lock failure, and AC side fault, and then the overvoltage amplitude, waveform, and other prediction results are obtained; 2. Analytical method: based on circuit theory and transmission line equations, the structure of the DC submarine cable system is simplified, and the analytical expression of the transient overvoltage is derived, and the prediction value is calculated by substituting the known parameters; 3. Traditional machine learning method: based on statistical learning theory, basic characteristic quantities of the DC submarine cable operation are extracted, a mapping relationship between the characteristic quantities and the transient overvoltage is established, and the prediction is realized through a machine learning model; 4. Prediction method based on simple neural network: the characteristic extraction or time series modeling capability of the basic neural network is used, the submarine cable operation data is input, the overvoltage variation law is fitted through network training, and the prediction result is output; however, these methods have problems such as large calculation amount, poor adaptability, limited prediction accuracy, poor real-time performance, weak generalization ability, and insufficient consideration of multi-dimensional influencing factors, which are difficult to meet the needs of actual engineering applications.
[0004] Therefore, there is an urgent need for a method that can accurately predict the transient overvoltage of the DC submarine cable in real time, and provide technical support for the safe and stable operation of the DC submarine cable system. SUMMARY
[0005] In order to solve the problems existing in the prior art, the present application provides a prediction method and system for transient overvoltage of a DC submarine cable.
[0006] The technical scheme of the present application is as follows: On the one hand, the present application provides a prediction method for transient overvoltage of a DC submarine cable, which comprises: The operation data of DC submarine cable is acquired, and multiple key features are extracted from the operation data based on the influencing factors of transient overvoltage of DC submarine cable and preprocessed. A dataset is constructed based on the preprocessed key features and divided into training set and test set according to a preset ratio. A transient overvoltage prediction model is constructed, and the hyperparameters of the transient overvoltage prediction model are optimized using a symbiotic biological search algorithm to obtain the optimal hyperparameter set. The transient overvoltage prediction model with the optimal hyperparameter set is trained using a training set, and the trained transient overvoltage prediction model is evaluated based on a test set and a preset performance evaluation index to obtain the final transient overvoltage prediction model. The real-time operating data of the DC submarine cable is obtained and input into the final transient overvoltage prediction model to obtain the prediction results of the transient overvoltage of the DC submarine cable.
[0007] Preferably, the preprocessing includes data normalization and outlier handling, specifically: The current key feature quantity is normalized based on the maximum and minimum values to eliminate the influence of different dimensions; The 3σ principle is used to identify and handle outliers, specifically: if If the current key feature is considered an outlier, then... Indicates the first Key feature quantities, This represents the mean of key feature quantities. The standard deviation of the key feature is represented; otherwise, it is a normal value. The outlier is replaced by the median of the current key feature.
[0008] Preferably, the key characteristic quantities include DC system parameter characteristics, marine environment parameter characteristics, and submarine cable laying status characteristics, specifically: The DC system parameter characteristics include DC power characteristics, DC submarine cable parameter characteristics, DC submarine cable-to-ground capacitance characteristics, and fault characteristics. The marine environmental parameters include seawater temperature characteristics, seawater salinity characteristics, and ocean current characteristics; The characteristics of the submarine cable laying status include the characteristics of the submarine cable burial depth and the characteristics of the submarine cable curvature.
[0009] Preferably, the transient overvoltage prediction model is constructed based on CNN-BiLSTM-Attention, including an input layer, a CNN layer, a BiLSTM layer, an Attention layer, and an output layer, wherein: The input layer is used to receive input data; The CNN layer uses a one-dimensional convolutional kernel to extract local features from the input data and outputs a feature vector for each time step. The BiLSTM layer performs bidirectional propagation of the feature vector at each time step to obtain the hidden state at each time step; The Attention layer calculates the score at each time step based on the hidden state, and uses the softmax function to convert the score into a probability distribution as attention weights. The hidden state is then weighted based on the attention weights to obtain the weighted result. The output layer uses a fully connected layer to linearly map the weighted results to obtain the prediction results of transient overvoltage of DC submarine cable.
[0010] Preferably, the symbiotic organism search algorithm includes an initialization phase, a mutualistic symbiosis phase, a symbiotic bias phase, and a parasitic phase, with the following specific steps: In the initialization phase, each set of hyperparameters to be optimized in the transient overvoltage prediction model is regarded as a biological individual, and an initial ecosystem is randomly generated within a preset parameter range. The root mean square error of the transient overvoltage prediction model is used as the fitness function. Each biological individual is substituted into the fitness function to calculate the corresponding fitness value. The biological individual with the maximum fitness value is selected as the current optimal biological individual. In the mutualistic symbiosis stage, for each individual in the initial ecosystem, another individual is randomly selected to simulate the mutualistic relationship between the two, and the two individuals are updated. The calculation method is as follows: ; ; ; In the formula, Indicates the first An updated biological individual; Indicates the first Individual organisms; Represents the first random number; Indicates the second random number; This represents the current optimal biological individual; Represents a reciprocal vector; This represents the pre-defined first benefit factor; Indicates the first An updated biological individual; Indicates the first Individual organisms; This represents the pre-defined second benefit factor; calculate and The fitness value, if The fitness value is greater than fitness value or The fitness value is greater than The fitness value will then be Replace with or Replace with Otherwise, retain or No replacement will be made; In the symbiotic phase, the current individual randomly selects another individual and benefits from it, calculated as follows: ; calculate The fitness value, if The fitness value is greater than The fitness value will then be Replace with Otherwise, retain No replacement will be made; The parasitic stage utilizes the current biological individual Generate a mutation vector The calculation method is as follows: ; In the formula, Indicates the location of the mutated gene; Represents a random number function; express Length; express The upper bound of the search; express The lower bound of the search; Randomly select a biological individual As the host of the mutation vector, calculate and The fitness value, if The fitness value is greater than The fitness value will then be Replace with Otherwise, retain No replacement will be made; The above optimization process is iterated until the preset maximum number of iterations is reached or the fitness function converges. Then the iteration is stopped, and the current best biological individual, i.e., the best hyperparameter set, is output.
[0011] Preferably, the preset performance evaluation indicators include mean absolute error, root mean square error, and prediction accuracy.
[0012] Preferred options also include: When the predicted transient overvoltage of the DC submarine cable exceeds the preset threshold, an early warning signal is issued to remind operators to take measures. Based on the prediction results of transient overvoltage of DC submarine cable, the control parameters of DC submarine cable are adjusted to suppress transient overvoltage.
[0013] On the other hand, the present invention also provides a prediction system for transient overvoltage of DC submarine cables, the system comprising: The data acquisition module acquires the operational data of the DC submarine cable, extracts multiple key features from the operational data based on the influencing factors of transient overvoltage of the DC submarine cable, and performs preprocessing; it constructs a dataset based on the preprocessed key features and divides it into training set and test set according to a preset ratio; The model training module has a built-in transient overvoltage prediction model. It uses a symbiotic biological search algorithm to optimize the hyperparameters of the transient overvoltage prediction model to obtain the optimal hyperparameter set. The transient overvoltage prediction model with the optimal hyperparameter set is trained using the training set. The trained transient overvoltage prediction model is evaluated based on the test set and preset performance evaluation indicators to obtain the final transient overvoltage prediction model. The prediction module acquires real-time operating data of the DC submarine cable and inputs it into the final transient overvoltage prediction model to obtain the prediction results of the transient overvoltage of the DC submarine cable.
[0014] In another aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in the present invention.
[0015] In another aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the method described in the present invention.
[0016] The present invention has the following beneficial effects: 1. This invention provides a method and system for predicting transient overvoltages in DC submarine cables, covering multiple key features and considering the influencing factors of transient overvoltages in DC submarine cables, breaking through the limitation of focusing only on a single system parameter; it solves the problems of poor adaptability and weak generalization ability caused by single feature dimensions and poor data quality in existing methods, allowing the model to adapt to complex engineering realities of different submarine cable types, different marine environments, and different fault scenarios, thereby enhancing the environmental adaptability of the model and the generalization ability of the prediction model; 2. This invention provides a method and system for predicting transient overvoltages in DC submarine cables. It employs a CNN-BiLSTM-Attention model, which, through this hierarchical design, achieves a complete processing flow from local feature extraction and temporal dependency modeling to focusing on important features. The CNN layer effectively captures the local fluctuation patterns and short-term abrupt changes in the transient overvoltage signal of the DC submarine cable. The BiLSTM layer fully utilizes the bidirectional temporal information of the transient overvoltage sequence, not only capturing the impact of historical voltage changes on the current state but also comprehensively analyzing potential future trends, making it particularly suitable for modeling non-stationary transient signals. The Attention layer further highlights the temporal segments most influential on the prediction target. The combination of BiLSTM's bidirectional information fusion capability and the adaptive feature weighting strategy of the Attention mechanism enables the model to exhibit excellent robustness and adaptability when processing highly non-stationary and multivariable transient overvoltage signals of DC submarine cables, improving the prediction accuracy of transient overvoltages in DC submarine cables and enhancing the feature capture capability in complex scenarios. 3. This invention provides a method and system for predicting transient overvoltages in DC submarine cables. It employs a symbiotic biological search algorithm to optimize hyperparameters. Through multi-stage iterations of mutualistic symbiosis (bidirectional optimization of two individuals), biased symbiosis (unidirectional borrowing of the advantages of the best individual), and parasitism (mutational iterative optimization), it avoids the subjectivity of manual parameter tuning and the trap of local optima, thereby improving the efficiency of hyperparameter optimization. It also enhances the stability of model performance, avoids fluctuations in prediction accuracy caused by improper hyperparameter settings, and enables the model to output stable and reliable results under different datasets and different operating scenarios. Attached Figure Description
[0017] Figure 1 This is a detailed flowchart of an embodiment of the present invention. Detailed Implementation
[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. 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 are within the scope of protection of the present invention.
[0019] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.
[0020] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0021] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.
[0022] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.
[0023] Example 1: See Figure 1 This invention provides a method for predicting transient overvoltages in DC submarine cables, the method comprising: S1. Obtain the operational data of the DC submarine cable; S2. Based on the influencing factors of transient overvoltage of DC submarine cable, extract several key feature quantities from the operating data and perform preprocessing; S21. The preprocessing includes data normalization and outlier handling, specifically: The current key feature quantities are normalized based on their maximum and minimum values to eliminate the influence of different dimensions. The calculation method is as follows: ; In the formula, This represents the key feature quantity after normalization; Represents key feature quantities; This represents the minimum value of a key feature quantity; This represents the maximum value of a key feature. The 3σ principle is used to identify and handle outliers, specifically: if If the current key feature is considered an outlier, then... Indicates the first Key feature quantities, This represents the mean of key feature quantities. This represents the standard deviation of the key feature; otherwise, it is a normal value. Outliers are replaced with the median of the current key feature. S22. The key characteristic quantities include DC system parameter characteristics, marine environment parameter characteristics, and submarine cable laying status characteristics, specifically: S221, The DC system parameter characteristics include DC power characteristics, DC submarine cable parameter characteristics, DC submarine cable-to-ground capacitance characteristics, and fault characteristics; The DC power characteristic quantity is a core parameter reflecting the operating status of the DC submarine cable system, and its magnitude directly determines the flow and distribution of energy in the system. The characteristic parameters of the DC submarine cable include the cable's inductance and resistance. Considering the significant differences in parameters among different types of cables, a cable type coefficient is introduced. Its value is determined based on the cable model, and the calculation method is as follows: ; In the formula, Indicates the characteristic parameters of DC submarine cables; Indicates cable inductance; Indicates cable resistance; Indicates the first cable type coefficient; Indicates the second cable type coefficient; The specific capacitance characteristic of the DC submarine cable to the ground is the distributed capacitance formed between the conductor of the DC submarine cable, the metal sheath, and the ground. The charging and discharging behavior of the capacitance will cause significant voltage fluctuations and energy surges. The fault characteristic quantities include three types of faults: continuous commutation failure, single-pole blockage, and double-pole blockage. S222, The marine environmental parameters include seawater temperature characteristics, seawater salinity characteristics, and ocean current characteristics; The seawater temperature characteristics affect the thermal and insulation properties of the cable, and thus affect transient overvoltage. The seawater salinity characteristic quantity mentioned above affects the conductivity around the cable, which in turn affects the propagation and attenuation of transient overvoltages; The ocean current characteristics mentioned above can affect the stress and heat dissipation of the cable, and thus affect the transient overvoltage. S223, The characteristics of the submarine cable laying status include the characteristics of the submarine cable burial depth and the characteristics of the submarine cable curvature; The burial depth characteristics of the submarine cable affect the thermal and mechanical properties of the cable, and thus affect the transient overvoltage. The bending characteristics of the submarine cable affect the cable's mechanical stress and electrical performance, which in turn affect transient overvoltage. S3. Construct a dataset based on the preprocessed key features and divide it into a training set and a test set according to a preset ratio; S4. Construct a transient overvoltage prediction model based on CNN-BiLSTM-Attention, including an input layer, CNN layer, BiLSTM layer, Attention layer, and output layer, wherein: The input layer is used to receive input data. ,in Indicates the total number of time steps. Indicates the number of input data; The CNN layer uses a size of One-dimensional convolution kernel Local feature extraction is performed on the input data, and the feature vector at each time step is output. For each output channel, a convolution operation is performed at each time step, calculated as follows: ; ; In the formula, Indicates the first The first time step Feature vectors of each output channel; Indicates the activation function; Indicates the first In the first step The first input data One-dimensional convolutional kernel weights for each output channel; Indicates the first The first time step One input data; Indicates the first The bias of each output channel; Indicates the number of output channels; The BiLSTM layer consists of two sub-networks, a forward LSTM and a backward LSTM, which propagate the feature vector at each time step bidirectionally to obtain the hidden state at each time step. The calculation method is as follows: ; ; ; In the formula, Indicates the first Feature vectors after forward propagation at each time step; Indicates an LSTM network; Indicates the first Feature vectors after forward propagation at each time step; Indicates the first Feature vectors after backward propagation at each time step; Indicates the first Feature vectors after backward propagation at each time step; Indicates the first The hidden state at each time step; The Attention layer calculates the score at each time step based on the hidden state, and the calculation method is as follows: ; In the formula, Indicates the first Score for each time step; express function; Indicates the weights of the Attention layer; This indicates the bias of the Attention layer; The scores are converted into a probability distribution using the softmax function and used as attention weights. The hidden states are then weighted based on these attention weights to obtain a weighted result. The calculation method is as follows: ; ; In the formula, Indicates the first Attention weights for each time step; Indicates the weighted result; The output layer uses a fully connected layer to linearly map the weighted results to obtain the prediction results of the transient overvoltage of the DC submarine cable. The calculation method is as follows: ; In the formula, This indicates the prediction results for transient overvoltages in DC submarine cables; Indicates the weights of the fully connected layer; Indicates the bias of the fully connected layer; S5. Use the symbiotic organism search algorithm to optimize the hyperparameters of the transient overvoltage prediction model and obtain the optimal hyperparameter set. The symbiotic organism search algorithm includes an initialization phase, a mutualistic symbiosis phase, a symbiotic-parasitic phase, and a parasitic phase. S51. In the initialization phase, each set of hyperparameters to be optimized in the transient overvoltage prediction model is regarded as a biological individual, and an initial ecosystem is randomly generated within a preset parameter range. The root mean square error of the transient overvoltage prediction model is used as the fitness function. Each biological individual is substituted into the fitness function to calculate the corresponding fitness value. The biological individual with the maximum fitness value is selected as the current optimal biological individual. S52. In the mutualistic symbiosis stage, for each individual in the initial ecosystem, another individual is randomly selected to simulate the mutually beneficial relationship between the two, and the two individuals are updated. The calculation method is as follows: ; ; ; In the formula, Indicates the first An updated biological individual; Indicates the first Individual organisms; This represents the first random number, which is in the range [0,1]. This represents the second random number, which is in the range [0,1]. This represents the current optimal biological individual; Represents a reciprocal vector; This represents the pre-defined first benefit factor; Indicates the first An updated biological individual; Indicates the first Individual organisms; This represents the pre-defined second benefit factor; calculate and The fitness value, if The fitness value is greater than fitness value or The fitness value is greater than The fitness value will then be Replace with or Replace with Otherwise, retain or No replacement will be made; S53. In the symbiotic phase, the current organism randomly selects another individual and benefits from it, calculated as follows: ; calculate The fitness value, if The fitness value is greater than The fitness value will then be Replace with Otherwise, retain No replacement will be made; S54, The parasitic stage utilizes the current biological individual Generate a mutation vector The calculation method is as follows: ; In the formula, Indicates the location of the mutated gene; Represents a random number function; express Length; express The upper bound of the search; express The lower bound of the search; Randomly select a biological individual As the host of the mutation vector, calculate and The fitness value, if The fitness value is greater than The fitness value will then be Replace with Otherwise, retain No replacement will be made; S55. Iterate the above optimization process until the preset maximum number of iterations is reached or the fitness function converges, then stop the iteration and output the current best biological individual, i.e. the best hyperparameter set. S6. Train the transient overvoltage prediction model with the optimal hyperparameter set using the training set; The Adam optimization algorithm based on mini-batch gradient descent is adopted, with a batch size of 32, a learning rate of 0.001, 500 training epochs, and mean squared error as the loss function. During training, an early stopping strategy was used to prevent overfitting, i.e., training was stopped when the test set loss did not decrease for 20 consecutive rounds; in addition, Dropout technology (with a dropout rate of 0.2) was used to further prevent overfitting. S7. Evaluate the trained transient overvoltage prediction model based on the test set and preset performance evaluation indicators to obtain the final transient overvoltage prediction model. The preset performance evaluation indicators include mean absolute error, root mean square error, and prediction accuracy. The mean absolute error is calculated as follows: ; In the formula, Indicates the mean absolute error; Indicates the number of samples in the test set; Indicates the first Prediction results for each test set sample; Indicates the first The true results of each test set sample; The root mean square error The calculation method is as follows: ; The prediction accuracy The calculation method is as follows: ; S8. Obtain the real-time operating data of the DC submarine cable and input it into the final transient overvoltage prediction model to obtain the prediction result of the transient overvoltage of the DC submarine cable, that is, the specific quantitative prediction value of the transient overvoltage of the DC submarine cable. S9 also includes: When the predicted transient overvoltage of the DC submarine cable exceeds the preset threshold, an early warning signal is issued to remind operators to take measures. Based on the prediction results of transient overvoltage of DC submarine cable, the control parameters of DC submarine cable are adjusted to suppress transient overvoltage.
[0024] Example 2: This embodiment provides a system for predicting transient overvoltages in DC submarine cables, the system comprising: The data acquisition module acquires the operational data of the DC submarine cable, extracts multiple key features from the operational data based on the influencing factors of transient overvoltage of the DC submarine cable, and performs preprocessing; it constructs a dataset based on the preprocessed key features and divides it into training set and test set according to a preset ratio; The model training module has a built-in transient overvoltage prediction model. It uses a symbiotic biological search algorithm to optimize the hyperparameters of the transient overvoltage prediction model to obtain the optimal hyperparameter set. The transient overvoltage prediction model with the optimal hyperparameter set is trained using the training set. The trained transient overvoltage prediction model is evaluated based on the test set and preset performance evaluation indicators to obtain the final transient overvoltage prediction model. The prediction module acquires real-time operating data of the DC submarine cable and inputs it into the final transient overvoltage prediction model to obtain the prediction results of the transient overvoltage of the DC submarine cable.
[0025] Example 3: This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a method for predicting transient overvoltages of a DC submarine cable as described in any one of Embodiment 1.
[0026] Example 4: This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a method for predicting transient overvoltages of a DC submarine cable as described in any one of Embodiment 1.
[0027] In this application embodiment, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.
[0028] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented using electronic hardware, computer software, or a combination of electronic hardware and software. 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 implementation should not be considered beyond the scope of this application.
[0029] 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.
[0030] In the several embodiments provided in this application, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0031] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.
Claims
1. A method for predicting transient overvoltages in DC submarine cables, characterized in that, The method includes: The operation data of DC submarine cable is acquired, and multiple key features are extracted from the operation data based on the influencing factors of transient overvoltage of DC submarine cable and preprocessed. A dataset is constructed based on the preprocessed key features and divided into training set and test set according to a preset ratio. A transient overvoltage prediction model is constructed, and the hyperparameters of the transient overvoltage prediction model are optimized using a symbiotic biological search algorithm to obtain the optimal hyperparameter set. The transient overvoltage prediction model with the optimal hyperparameter set is trained using a training set, and the trained transient overvoltage prediction model is evaluated based on a test set and a preset performance evaluation index to obtain the final transient overvoltage prediction model. The real-time operating data of the DC submarine cable is obtained and input into the final transient overvoltage prediction model to obtain the prediction results of the transient overvoltage of the DC submarine cable.
2. The method for predicting transient overvoltages in a DC submarine cable according to claim 1, characterized in that, The preprocessing includes data normalization and outlier handling, specifically: The current key feature quantity is normalized based on the maximum and minimum values to eliminate the influence of different dimensions; The 3σ principle is used to identify and handle outliers, specifically: if If the current key feature is considered an outlier, then... Indicates the first Key feature quantities, This represents the mean of key feature quantities. The standard deviation of the key feature is represented; otherwise, it is a normal value. The outlier is replaced by the median of the current key feature.
3. The method for predicting transient overvoltages in a DC submarine cable according to claim 1, characterized in that, The key characteristic quantities include DC system parameter characteristics, marine environment parameter characteristics, and submarine cable laying status characteristics, specifically: The DC system parameter characteristics include DC power characteristics, DC submarine cable parameter characteristics, DC submarine cable-to-ground capacitance characteristics, and fault characteristics. The marine environmental parameters include seawater temperature characteristics, seawater salinity characteristics, and ocean current characteristics; The characteristics of the submarine cable laying status include the characteristics of the submarine cable burial depth and the characteristics of the submarine cable curvature.
4. The method for predicting transient overvoltages in a DC submarine cable according to claim 1, characterized in that, The transient overvoltage prediction model is built based on CNN-BiLSTM-Attention, and includes an input layer, a CNN layer, a BiLSTM layer, an Attention layer, and an output layer, wherein: The input layer is used to receive input data; The CNN layer uses a one-dimensional convolutional kernel to extract local features from the input data and outputs a feature vector for each time step. The BiLSTM layer performs bidirectional propagation of the feature vector at each time step to obtain the hidden state at each time step; The Attention layer calculates the score at each time step based on the hidden state, and uses the softmax function to convert the score into a probability distribution as attention weights. The hidden state is then weighted based on the attention weights to obtain the weighted result. The output layer uses a fully connected layer to linearly map the weighted results to obtain the prediction results of transient overvoltage of DC submarine cable.
5. The method for predicting transient overvoltages in a DC submarine cable according to claim 1, characterized in that, The symbiotic organism search algorithm includes an initialization phase, a mutualistic symbiosis phase, a symbiotic-parasitic phase, and a parasitic phase. The specific steps are as follows: In the initialization phase, each set of hyperparameters to be optimized in the transient overvoltage prediction model is regarded as a biological individual, and an initial ecosystem is randomly generated within a preset parameter range. The root mean square error of the transient overvoltage prediction model is used as the fitness function, and each biological individual is substituted into the fitness function to calculate the corresponding fitness value. Select the organism with the highest fitness value as the current best organism. In the mutualistic symbiosis stage, for each individual in the initial ecosystem, another individual is randomly selected to simulate the mutualistic relationship between the two, and the two individuals are updated. The calculation method is as follows: ； ； ； In the formula, Indicates the first An updated biological individual; Indicates the first Individual organisms; Represents the first random number; Indicates the second random number; This represents the current optimal biological individual; Represents a reciprocal vector; This represents the pre-defined first benefit factor; Indicates the first An updated biological individual; Indicates the first Individual organisms; This represents the pre-defined second benefit factor; calculate and The fitness value, if The fitness value is greater than fitness value or The fitness value is greater than The fitness value will then be Replace with or Replace with ; Otherwise, keep or No replacement will be made; In the symbiotic phase, the current individual randomly selects another individual and benefits from it, calculated as follows: ； calculate The fitness value, if The fitness value is greater than The fitness value will then be Replace with ; Otherwise, keep No replacement will be made; The parasitic stage utilizes the current biological individual Generate a mutation vector The calculation method is as follows: ； In the formula, Indicates the location of the mutated gene; Represents a random number function; express Length; express The upper bound of the search; express The lower bound of the search; Randomly select a biological individual As the host of the mutation vector, calculate and The fitness value, if The fitness value is greater than The fitness value will then be Replace with ; Otherwise, keep No replacement will be made; The above optimization process is iterated until the preset maximum number of iterations is reached or the fitness function converges. Then the iteration is stopped, and the current best biological individual, i.e., the best hyperparameter set, is output.
6. The method for predicting transient overvoltages in a DC submarine cable according to claim 1, characterized in that, The preset performance evaluation indicators include mean absolute error, root mean square error, and prediction accuracy.
7. The method for predicting transient overvoltages in a DC submarine cable according to claim 1, characterized in that, Also includes: When the predicted transient overvoltage of the DC submarine cable exceeds the preset threshold, an early warning signal is issued to remind operators to take measures. Based on the prediction results of transient overvoltage of DC submarine cable, the control parameters of DC submarine cable are adjusted to suppress transient overvoltage.
8. A prediction system for transient overvoltage in a DC submarine cable, characterized in that, The system includes: The data acquisition module acquires the operational data of the DC submarine cable, extracts multiple key features from the operational data based on the influencing factors of transient overvoltage of the DC submarine cable, and performs preprocessing; it constructs a dataset based on the preprocessed key features and divides it into training set and test set according to a preset ratio; The model training module has a built-in transient overvoltage prediction model. It uses a symbiotic biological search algorithm to optimize the hyperparameters of the transient overvoltage prediction model to obtain the optimal hyperparameter set. The transient overvoltage prediction model with the optimal hyperparameter set is trained using the training set. The trained transient overvoltage prediction model is evaluated based on the test set and preset performance evaluation indicators to obtain the final transient overvoltage prediction model. The prediction module acquires real-time operating data of the DC submarine cable and inputs it into the final transient overvoltage prediction model to obtain the prediction results of the transient overvoltage of the DC submarine cable.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1 to 7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1 to 7.