A submarine cable status monitoring method, system and fault recovery method

Abstract

The present invention discloses a submarine cable status monitoring method, system, and fault recovery method, comprising: collecting real-time data of various submarine cable operating parameters and performing preprocessing; loading the preprocessed real-time data into a pre-built and trained BP neural network model to obtain submarine cable fault monitoring results; wherein, during the BP neural network model training process, a multi-objective optimization algorithm is used to optimize the model parameters of the BP neural network model; after the system inputs the collected data into the BP neural network model, it ultimately outputs the fault type and makes dynamic adjustments or alarms for different fault types. The present invention can quickly and accurately monitor the fault type of the submarine cable, so that countermeasures can be taken early, thereby improving the safety and reliability of submarine cable operation.

Description

Technical Field

[0001] The present invention relates to the technical field of submarine cable monitoring, and in particular to a submarine cable online status monitoring method, system and fault recovery method. Background Art

[0002] Driven by the general trend of renewable energy development, a number of typical distributed energy sources, such as wind and nuclear power, are experiencing rapid growth. The vigorous development of offshore energy has also led to a continuous increase in the load on submarine cable lines. In practice, an increasing number of submarine cables are adopting a parallel operation of multiple cables in phase, and this approach is also being considered for the renovation and upgrade of some medium and low voltage submarine cable transmission lines. However, due to a lack of understanding of the operating mechanism of multiple cables in phase and parallel operation, and limited field experience, problems such as excessive circulating currents in the cables due to insulation aging or damage, as well as ground wire burnout and abnormal ground currents caused by excessive sheath temperature rise, are common in actual operation. Therefore, to improve line transmission efficiency, ensure submarine cable reliability, and extend its service life, online condition monitoring and dynamic adjustment of submarine cables during operation have significant technical and economic value.

[0003] Currently, common submarine cable monitoring methods include manual observation, patrols, ocean "fishing" (fish towing), diver diving, underwater robot deployment, and coastal monitoring stations. However, these existing monitoring methods have numerous drawbacks: manual observation relies on experience and is prone to false detections and missed detections; the ocean "fish towing" method is susceptible to accuracy impacted by currents, the towing vessel, and the length of the towing cable; diver monitoring requires ensuring the safety of divers, and is limited by the long duration, long distances, and deep depths of submarine cable inspections; underwater robot detection systems are complex, and autonomous detection systems are not technologically mature; and coastal monitoring stations have low fault location accuracy, high costs, and lack visibility into the cable's surrounding environment, making it impossible to accurately predict the cable's operational status. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the above-mentioned prior art and to provide a method, system and fault recovery method for online status monitoring of submarine cables, which can quickly and accurately predict the fault type of the submarine cable so as to take countermeasures for different faults as early as possible and improve the safety of submarine cable operation.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a method for monitoring the online status of a submarine cable, the method comprising:

[0007] Collect real-time data of various operating parameters of submarine cables and perform pre-processing;

[0008] Load the pre-processed real-time data into the pre-built and trained BP neural network model to obtain the monitoring results of submarine cable faults;

[0009] The training process of the BP neural network model includes the following steps:

[0010] Acquire historical data of various operating parameters that can reflect the type of submarine cable fault, and pre-process them as input samples for submarine cable fault monitoring. Encode the type of submarine cable fault as output samples. The input and output samples constitute training samples for establishing a neural network model.

[0011] Construct a BP neural network model based on input variables and output fault types, and determine the topology of the BP neural network;

[0012] The training samples are input into the constructed BP neural network model for training, and during the training process, a multi-objective optimization algorithm is used to optimize the model parameters of the BP neural network model to obtain the optimal BP neural network model.

[0013] In combination with the first aspect, further, the operating parameters include submarine cable grounding current, submarine cable temperature rise and submarine cable circulating current; the fault types of the submarine cable include single-phase grounding fault, phase-to-phase short circuit fault, circuit breaker fault and other faults.

[0014] In combination with the first aspect, further, the input samples obtained after preprocessing the acquired historical data of each operating parameter are expressed by the matrix X as follows:

[0015] x=(I1,I2,…I p ,…ΔT1,ΔT2,…ΔT p ,…,I s1 , I s2 ,…I sp ) T (1)

[0016] Where p represents the total number of monitored submarine cables, I p is the grounding current of the pth submarine cable, ΔT p is the temperature rise of the pth submarine cable, I sp is the circulation of the p-th submarine cable, T represents the transposition operation; the elements I1 to I in the matrix X are sp Number them in the order of 1 to n and express them as X=(x1,x2,…x q ,…x n ) T , n=3p;x q Represents the qth element in the matrix X, where q = 1, 2, …, n.

[0017] In combination with the first aspect, the constructed BP neural network model further includes an input layer, a hidden layer, and an output layer. Each layer is composed of a number of neurons, and the neurons are connected in a fully connected manner. The number of layers h of the hidden layer is determined based on the matrix X and the fault type output by the output layer, and the calculation formula is as follows:

[0018]

[0019] Where n is the number of neurons in the input layer, n=3p; m is the number of neurons in the output layer, and a is a constant between 1 and 10.

[0020] In combination with the first aspect, further, the step of optimizing the structure and parameters of the BP neural network model using a multi-objective optimization algorithm during the training process includes:

[0021] Initialize the constructed BP neural network model and randomly generate a set of weight matrices V and W, which are expressed as follows:

[0022]

[0023]

[0024] Where V is the weight matrix between the input layer and the hidden layer, v ij is the element in the i-th row and j-th column of the weight matrix V, i = 1, 2, ..., 3p; j = 1, 2, ..., m; W is the weight matrix between the hidden layer and the output layer, w j1 Represents the element in the jth row and first column of the weight matrix W;

[0025] Based on the initialized BP neural network model, the training samples are forward propagated from the input layer to the hidden layer, and the output result Y of the hidden layer is expressed as follows:

[0026] y=(Y1,Y2,…Y j …Y m ) T (4)

[0027] Where Y j Represents the jth element in the matrix of the output result Y, j = 1, 2, ... m, f is the activation function;

[0028] The output result O of the output layer is expressed as follows:

[0029] O=(O1,O2,O3,O4) T (5)

[0030] Where O1, O2, O3, and O4 are single-phase grounding fault, interphase short circuit fault, open circuit fault, and other faults, respectively. The calculation formulas for O1, O2, O3, and O4 are as follows:

[0031]

[0032] Where w jk is an element in the weight matrix W, k = 1, 2, 3, 4;

[0033] Calculate the mean error E between the output result O and the expected output D;

[0034] The objective function is established by combining the number of connection weights and the mean error E of the BP neural network;

[0035] Based on the objective function, the number of hidden layers and neurons is repeatedly adjusted, and the weight matrices V and W are iteratively updated through the gradient descent method to reduce the error between the output result and the expected output until the value error E meets the set network accuracy requirements.

[0036] In combination with the first aspect, further, the step of iteratively updating the weight matrices V and W by the gradient descent method includes:

[0037] According to the output result O, calculate the adjustment amount Δv of the weight matrix V jk and the weight matrix W adjustment Δw jk They are:

[0038]

[0039] ΔW jk =η(d k -o k )o k (1-o k )Y j (7)

[0040] Where, is the error signal of the output layer: d k The expected output D = (d1, d2, e3, d4) T The kth element in , k = 1, 2, 3, 4; η is the learning rate, 0 < η < 1;

[0041] Based on the adjustment amount Δv jk , Δw jk , the weight matrices V and W are calculated and updated using formula (8):

[0042] v ij (l+1)=v ij (l)+Δv jk

[0043] w jk (l+1)=w jk (l)+Δw jk (8)

[0044] Where l represents the number of updates; v ij (l+1) and v ij (l) represents the value of the l+1th and lth iterations in the weight matrix between the input layer and the hidden layer, respectively, w jk (l+1) and w jk (l) represents the values ​​of the l+1th and lth iterations in the weight matrix between the hidden layer and the output layer respectively;

[0045] Repeat the above steps to perform iterative calculation until the preset termination condition is reached, and finally update the optimal weight matrix of the BP neural network.

[0046] Combined with the first aspect, further, the established objective function model is:

[0047] Min F(x)=(f1(x), f2(x)) (9)

[0048] Where f2(x) is the objective function of the calculation error, f2(x) = E; f1(x) represents the objective function of the number of connection weights of the BP neural network:

[0049]

[0050] Where c e and c e+1 They represent the number of nodes in the e-th layer and the e+1-th layer in the hidden layer, and h is the number of hidden layers;

[0051] The constraints set for the objective functions f1(x) and f2(x) include:

[0052] Time constraint: t≤T (11)

[0053] Where t represents the time for iteratively updating the model parameters of the BP neural network model, and T is the maximum iteration time set according to actual needs;

[0054] Regional constraints:

[0055] Where Z q Represents a parameter. If the acquired data x q If it is within the detection range, then Z q The value is 1, otherwise the value is 0.

[0056] In a second aspect, the present invention provides a submarine cable online status monitoring system, the system comprising:

[0057] Data acquisition module, used to collect real-time data of various operating parameters of submarine cables and perform pre-processing;

[0058] The fault monitoring module is used to load the pre-processed real-time data into the pre-built and trained BP neural network model to obtain the monitoring results of submarine cable faults;

[0059] The data acquisition module includes a current transformer, a sheath circulating current sensor and a temperature sensor; the current transformer is used to collect the monitored submarine cable grounding current, the sheath circulating current sensor is used to collect the monitored submarine cable circulating current, and the temperature sensor is used to collect the monitored submarine cable circulating current;

[0060] The fault monitoring module includes a BP neural network model and a display and alarm unit; the BP neural network model includes an input layer, a hidden layer and an output layer, and the model parameters of the BP neural network model are optimized using a multi-objective optimization algorithm; the display and alarm unit is used to perform different displays and alarms for different fault types output by the output layer.

[0061] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for monitoring the online status of a submarine cable as described in any one of the first aspects are implemented.

[0062] In a fourth aspect, the present invention provides a submarine cable fault recovery method, characterized in that the method comprises:

[0063] Monitoring results of submarine cable faults using the submarine cable online status monitoring method according to any one of the first aspects;

[0064] If the fault type of the monitored submarine cable is a phase-to-phase short circuit fault, the sheath circulating current of the submarine cable is controlled within a safe range by adjusting the impedance series connection.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] The present invention detects the operating status of the submarine cable by optimizing the BP neural network model through a multi-objective optimization algorithm. After the BP neural network model determines the relationship between the input quantity and the output vector, it will save it in the network. When performing fault diagnosis, it will make correct judgments even if it encounters fault input information that has never been learned. In addition, the BP neural network will not make wrong judgments due to error interference of individual input sample data. It can operate normally even in the case of local neuron damage, and can output accurate fault information even when the measurement value of a certain data in the perception layer is inaccurate. It has strong self-learning ability, generalization ability and fault tolerance ability, so that it can accurately identify faults in the shortest time, so that countermeasures can be taken early for different faults, thereby improving the safety of submarine cable operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a flow chart of a method for monitoring the online status of a submarine cable provided by an embodiment of the present invention;

[0068] Figure 2 The present invention provides a block diagram of the structure and principle of a submarine cable online status monitoring system. DETAILED DESCRIPTION

[0069] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. Unless there is a conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0070] The term "and / or" in this document simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, or B exists alone. Additionally, the character " / " in this document generally indicates an "or" relationship between the related objects.

[0071] Example 1:

[0072] like Figure 1 As shown, this embodiment introduces a method for monitoring the online status of a submarine cable, which specifically includes the following steps:

[0073] Step 1: Collect real-time data of various operating parameters of the submarine cable and perform preprocessing;

[0074] Step 2: Load the pre-processed real-time data into the pre-built and trained BP neural network model to obtain the monitoring results of the submarine cable fault.

[0075] Among them, the collected operating parameters include submarine cable grounding current, submarine cable temperature rise and submarine cable circulating current; the detected submarine cable fault types include single-phase grounding fault, phase-to-phase short circuit fault, circuit breaker fault and other faults.

[0076] As an embodiment of the present invention, the training process of the BP neural network model in step 2 includes the following steps:

[0077] Step 2.1: Obtain historical data of various operating parameters that can reflect the type of submarine cable fault, pre-process them and use them as input samples for submarine cable fault monitoring. Encode the fault type of the submarine cable and use them as output samples. The input and output samples constitute the training samples for establishing the neural network model.

[0078] It is further explained that for the historical data obtained, part of its training samples is used as a training data set. Through self-learning, the weight matrices V and W are modified to continuously improve the neural network, thereby establishing a neural network with high accuracy, strong fault tolerance and fast computing speed; the other part is used as a test data set, which is input into the existing constructed neural network to output the fault type;

[0079] In this embodiment of the present invention, the operating status of p parallel submarine cables is monitored. The input samples obtained after preprocessing the historical data of each operating parameter are expressed as follows using a matrix X:

[0080] x=(I1,I2,…I p ,…ΔT1,ΔT2,…ΔT p ,…,I s1 , I s2 ,…I sp ) T (1)

[0081] Where p represents the total number of monitored submarine cables, I p is the grounding current of the pth submarine cable, ΔT p is the temperature rise of the pth submarine cable, I sp is the circulation of the p-th submarine cable, T represents the transposition operation; the elements I1 to I in the matrix X are sp Number them in the order of 1 to n and express them as X=(x1,x2,…x q ,…x n ) T , n=3p;x q Represents the qth element in the matrix X, where q = 1, 2, …, n.

[0082] Step 2.2: Construct a BP neural network model based on the input variables and output fault types, and determine the topology of the BP neural network;

[0083] Specifically, the BP neural network model constructed in the embodiment of the present invention includes an input layer, a hidden layer, and an output layer, wherein each layer is composed of a number of nodes, each node represents a neuron, and neurons are connected in a fully connected manner; data input to the neural network is transmitted unidirectionally between each layer via the input layer, and reaches the output layer of the network after passing through each neuron node; wherein the number of hidden layers h is determined based on the obtained matrix X and the fault type output by the output layer, and the calculation formula of h is as follows:

[0084]

[0085] Where n is the number of neurons in the input layer, n=3p; m is the number of neurons in the output layer, and a is a constant between 1 and 10.

[0086] Step 2.3: Input the training samples into the constructed BP neural network model for training, and use a multi-objective optimization algorithm to optimize the model parameters of the BP neural network model during the training process to obtain the optimal BP neural network model.

[0087] As an embodiment of the present invention, the specific process of optimizing the structure and parameters of the BP neural network model using a multi-objective optimization algorithm during the training process in step 2.3 includes:

[0088] Step 2.3.1: Initialize the constructed BP neural network model and randomly generate a set of weight matrices V and W as follows:

[0089]

[0090]

[0091] Where V is the weight matrix between the input layer and the hidden layer, v ij is the element in the i-th row and j-th column of the weight matrix V, i = 1, 2, ..., 3p; j = 1, 2, ..., m; W is the weight matrix between the hidden layer and the output layer, w j1 Represents the element in the jth row and first column of the weight matrix W;

[0092] Step 2.3.2: Based on the initialized BP neural network model, the training samples are forward propagated from the input layer to the hidden layer, and the output result Y of the hidden layer is expressed as follows:

[0093] Y=(Y1,Y2,…Y j …Y m ) T (4)

[0094] Where Y j Represents the jth element in the matrix of the output result Y, j = 1, 2, ... m, f is the activation function;

[0095] Step 2.3.3: The output result O of the output layer is expressed as follows:

[0096] O=(O1,O2,O3,O4) T (5)

[0097] Where O1, O2, O3, and O4 are single-phase grounding fault, interphase short circuit fault, open circuit fault, and other faults, respectively; O1, O2, O3, and O4 are all calculated using the general formula (6):

[0098]

[0099] Where w jk is an element in the weight matrix W, k = 1, 2, 3, 4;

[0100] Step 2.3.4: Calculate the mean error E between the output O and the expected output D. The calculation formula is:

[0101]

[0102] Where D is the expected output, D=(d1,d2,d3,d4) T It should be noted that D is the actual output submarine cable fault type corresponding to the historical data of each operating parameter, and d1, d2, d3, and d4 represent the actual output phase-to-ground fault, phase-to-phase short circuit fault, open circuit fault, and other faults corresponding to the historical data of each operating parameter, respectively.

[0103] Step 2.3.5: Establish the objective function by combining the number of connection weights and the mean error E of the BP neural network;

[0104] Step 2.3.6: Based on the objective function, repeatedly adjust the number of hidden layers and neurons, and iteratively update the weight matrices V and W using the gradient descent method to reduce the error between the output result and the expected output until the value error E meets the set network accuracy requirements.

[0105] As an embodiment of the present invention, the specific process of iteratively updating the weight matrices V and W by the gradient descent method in step 2.3.6 includes:

[0106] Step a: Calculate the adjustment value Δv of the weight matrix V according to the output result O jk and the weight matrix W adjustment Δw jk They are:

[0107]

[0108] Δw jk=n(d k -o k )o k (1-o k )Y j (7)

[0109] Where, is the error signal of the output layer: d k The expected output D = (d1, d2, d3, d4) T The kth element in , k = 1, 2, 3, 4; η is the learning rate, 0 < η < 1;

[0110] Step b: Based on the adjustment amount Δv jk , Δw jk , the weight matrices V and W are calculated and updated using formula (8):

[0111] v ij (l+1)=v ii (l)+Δv jk

[0112] w jk (l+1)=w jk (l)+Δw jk (8)

[0113] Where l represents the number of updates; v ij (l+1) and v ij (l) represents the value of the l+1th and lth iterations in the weight matrix between the input layer and the hidden layer, respectively, w jk (l+1) and w jk (l) represents the values ​​of the l+1th and lth iterations in the weight matrix between the hidden layer and the output layer respectively;

[0114] Step d: Repeat the above steps to perform iterative calculations until the preset termination condition is reached, and finally update the optimal weight matrix of the BP neural network.

[0115] It is further explained that the input and output of the BP neural network in the embodiment of the present invention are fixed, and what needs to be determined is the number of hidden layers and nodes in the BP neural network. The more complex the hidden layer structure, the stronger the neural network's ability to solve problems, including generalization, but it will also lead to more complex calculations and longer time. In the embodiment of the present invention, the complexity of fault identification is limited. Therefore, a multi-objective optimization algorithm is used to optimize the structure of the hidden layer to accurately identify faults in the shortest time.

[0116] Specifically, the model of the objective function established in step 2.3.5 in the embodiment of the present invention is:

[0117] Min F(x)=(f1(x), f2(x)) (9)

[0118] Where f2(x) is the objective function of the calculation error, f2(x) = E; f1(x) represents the objective function of the number of connection weights of the BP neural network:

[0119]

[0120] Where c e and c e+1 They represent the number of nodes in the e-th layer and the e+1-th layer in the hidden layer, and h is the number of hidden layers;

[0121] The constraints set for the objective functions f1(x) and f2(x) include:

[0122] Time constraint: t≤T (11)

[0123] Where t represents the time for iteratively updating the model parameters of the BP neural network model, and T is the maximum iteration time set according to actual needs;

[0124] Regional constraints:

[0125] Where Z q Represents a parameter. If the acquired data x q If it is within the detection range, then Z q The value of is 1, otherwise the value is 0;

[0126] The above objective function model is used to constrain and adjust the number of layers, number of nodes and weight matrix of the hidden layer of the BP neural network.

[0127] The submarine cable online detection method provided by the embodiment of the present invention can realize the self-monitoring and self-repair functions of submarine cable faults, making up for the shortcomings of the current submarine cable maintenance method based on manual detection and repair, which is low in efficiency, low in accuracy, and high in risk, and improves the safety of submarine cable operation.

[0128] Example 2:

[0129] like Figure 2 As shown, the embodiment of the present invention provides a submarine cable online status monitoring system, which can be used to implement the method described in the first embodiment, specifically including:

[0130] Data acquisition module, used to collect real-time data of various operating parameters of submarine cables and perform pre-processing;

[0131] The fault monitoring module is used to load the pre-processed real-time data into the pre-built and trained BP neural network model to obtain the monitoring results of submarine cable faults;

[0132] As an embodiment of the present invention, the data acquisition module includes a current transformer, a sheath circulating current sensor and a temperature sensor; the current transformer is used to collect the monitored submarine cable grounding current, the sheath circulating current sensor is used to collect the monitored submarine cable circulating current, and the temperature sensor is used to collect the monitored submarine cable circulating current; wherein, the temperature sensor used in the embodiment of the present invention is preferably a multi-mode optical fiber temperature sensor based on BOTDR.

[0133] The fault monitoring module in the embodiment of the present invention includes a BP neural network model and a display alarm unit; the BP neural network model includes an input layer, a hidden layer and an output layer, and the model parameters of the BP neural network model are optimized using a multi-objective optimization algorithm; the display alarm unit is used to perform different displays and alarms for different fault types output by the output layer.

[0134] The submarine cable online status monitoring system of the present invention adopts advanced software and hardware design concepts, fully considering the application environment and reliability requirements of submarine cable laying scenarios, and realizes an integrated monitoring system with high integration, high electromagnetic compatibility, and long endurance, which has great promotion and application value.

[0135] The submarine cable online status monitoring system provided in the embodiment of the present invention and the submarine cable online status monitoring method provided in Example 1 are based on the same technical concept and can produce the beneficial effects described in Example 1. For matters not fully described in this embodiment, please refer to Example 1.

[0136] Example 3:

[0137] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the computer program implements the steps of any one of the methods in the first embodiment.

[0138] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0139] Example 4:

[0140] An embodiment of the present invention provides a submarine cable fault recovery method, the method comprising:

[0141] A monitoring result of a submarine cable fault according to the submarine cable online status monitoring method according to any one of the embodiments;

[0142] If the fault type of the monitored submarine cable is a phase-to-phase short circuit fault, the sheath circulating current of the submarine cable is controlled within a safe range by adjusting the impedance series connection.

[0143] The specific method for adjusting the impedance series connection in the embodiment of the present invention is as follows: first, connecting the ground wire to the power supply connector to ensure a stable connection; second, rolling the primary coil ground wire into a threaded conduit to increase the ground wire's magnetic field area and maximize coil utilization; third, connecting the impedance in series to the circuit in the form of a secondary coil to form magnetic field coupling, thereby effectively controlling the circulation current of the submarine cable sheath. The embodiment of the present invention operates the impedance in series, which improves the effect of the impedance. During the series connection process, the impedance value can be adjusted to match the current submarine cable operation data and increase the correlation fit. The impedance value can also be set in the primary coil, and electromagnetic coupling can ensure that the impedance setting remains scientific.

[0144] The submarine cable fault recovery method provided by the embodiment of the present invention provides different early warnings for submarine cable single-phase grounding faults, phase-to-phase short circuit faults, circuit break faults, and other faults. Based on the monitoring results of the submarine cable faults, monitoring personnel can fully judge the faults existing in the submarine cable, and can promptly provide corresponding maintenance, inspection, emergency treatment measures and other guiding information for submarine cables with potential faults to ensure the safe operation of the equipment.

[0145] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0146] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0147] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0148] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for monitoring the online status of a submarine cable, characterized in that: The method comprises: Collect real-time data of various operating parameters of submarine cables and perform pre-processing; Load the pre-processed real-time data into the pre-built and trained BP neural network model to obtain the monitoring results of submarine cable faults; The training process of the BP neural network model includes the following steps: Historical data of various operating parameters that can reflect the type of submarine cable fault is obtained and pre-processed as input samples for submarine cable fault monitoring. The fault type of the submarine cable is encoded and used as output samples. The input and output samples constitute training samples for establishing a neural network model. The submarine cable fault types include single-phase grounding faults, phase-to-phase short circuit faults, open circuit faults, and other faults. A BP neural network model is constructed based on the input variables and the output fault type, and the topological structure of the BP neural network is determined; wherein the constructed BP neural network model includes an input layer, a hidden layer, and an output layer, each layer is composed of a number of neurons, and the neurons are connected in a fully connected manner; The training samples are input into the constructed BP neural network model for training, and a multi-objective optimization algorithm is used to optimize the structure and parameters of the BP neural network model during the training process to obtain the optimal BP neural network model; Among them, a multi-objective optimization algorithm is used to optimize the structure and parameters of the BP neural network model during the training process, including: Initialize the constructed BP neural network model and randomly generate a set of weight matrices V and W, which are expressed as follows: Where V is the weight matrix between the input layer and the hidden layer, v ij is the element in the i-th row and j-th column of the weight matrix V, i = 1, 2, ..., 3p; j = 1, 2, ..., m; W is the weight matrix between the hidden layer and the output layer, w j1 Represents the element in the jth row and first column of the weight matrix W; Based on the initialized BP neural network model, the training samples are forward propagated from the input layer to the hidden layer, and the output result Y of the hidden layer is expressed as follows: Y=(Y1,Y2,…Y j …Y m ) T (4) Where Y j Represents the jth element in the matrix of the output result Y, j = 1, 2, ... m, f is the activation function; The output result O of the output layer is expressed as follows: O=(O1,O2,O3,O4) T (5) Where O1, O2, O3, and O4 are single-phase grounding fault, interphase short circuit fault, open circuit fault, and other faults, respectively. The calculation formulas for O1, O2, O3, and O4 are as follows: Where w jk is an element in the weight matrix W, k = 1, 2, 3, 4; Calculate the mean error E between the output result O and the expected output D; The objective function is established by combining the number of connection weights and the mean error E of the BP neural network; Based on the objective function, the number of hidden layers and neurons is repeatedly adjusted, and the weight matrices V and W are iteratively updated through the gradient descent method to reduce the error between the output result and the expected output until the value error E meets the set network accuracy requirements.

2. The method for monitoring the online status of a submarine cable according to claim 1, characterized in that: The operating parameters include submarine cable grounding current, submarine cable temperature rise and submarine cable circulating current.

3. The method for monitoring the online status of a submarine cable according to claim 2, characterized in that: The input samples obtained after preprocessing the historical data of each operating parameter are expressed as follows using matrix X: X=(I1,I2,…I p ,…ΔT1,ΔT2,…ΔT p ,…,I s1 ,I s2 ,…I sp ) T (1) Where p represents the total number of monitored submarine cables, I p is the grounding current of the pth submarine cable, ΔT p is the temperature rise of the pth submarine cable, I sp is the circulation of the p-th submarine cable, T represents the transposition operation; the elements I1 to I in the matrix X are sp Number them in the order of 1 to n and express them as X=(x1,x2,…x q ,…x n ) T , n=3p;x q Represents the qth element in the matrix X, where q = 1, 2, …, n.

4. The method for monitoring the online status of a submarine cable according to claim 1, wherein: The number of hidden layers h is determined based on the matrix X and the fault type output by the output layer, and the calculation formula is as follows: Where n is the number of neurons in the input layer, n=3p; m is the number of neurons in the output layer, and a is a constant between 1 and 10.

5. The method for monitoring submarine cable status online according to claim 1, characterized in that: The step of iteratively updating the weight matrices V and W by the gradient descent method includes: According to the output result O, calculate the adjustment amount Δv of the weight matrix V jk and the weight matrix W adjustment Δw jk They are: Where, is the error signal of the output layer: d k The expected output D = (d1, d2, d3, d4) T The kth element in , k = 1, 2, 3, 4; η is the learning rate, 0 < η < 1; Based on the adjustment amount Δv jk , Δw jk , the weight matrices V and W are calculated and updated using formula (8): Where l represents the number of updates; v ij (l+1) and v ij (l) represents the value of the l+1th and lth iterations in the weight matrix between the input layer and the hidden layer, respectively, w jk (l+1) and w jk (l) represents the values ​​of the l+1th and lth iterations in the weight matrix between the hidden layer and the output layer, respectively; Repeat the above steps to perform iterative calculation until the preset termination condition is reached, and finally update the optimal weight matrix of the BP neural network.

6. The method for monitoring the online status of a submarine cable according to claim 1, characterized in that: The established objective function model is: Min F(x)=(f1(x),f2(x)) (9) Where f2(x) is the objective function of the calculation error, f2(x) = E; f1(x) represents the objective function of the number of connection weights of the BP neural network: Where c e and c e+1 They represent the number of nodes in the e-th layer and the e+1-th layer in the hidden layer, and h is the number of hidden layers; The constraints set for the objective functions f1(x) and f2(x) include: Time constraint: t≤T(11) Where t represents the time for iteratively updating the model parameters of the BP neural network model, and T is the maximum iteration time set according to actual needs; Where Z q Represents a parameter. If the acquired data x q If it is within the detection range, then Z q The value is 1, otherwise the value is 0.

7. A submarine cable online status monitoring system, characterized in that: The system comprises: Data acquisition module, used to collect real-time data of various operating parameters of submarine cables and perform pre-processing; The fault monitoring module is used to load the pre-processed real-time data into the pre-built and trained BP neural network model to obtain the monitoring results of submarine cable faults; The training process of the BP neural network model includes the following steps: Historical data of various operating parameters that can reflect the type of submarine cable fault is obtained and pre-processed as input samples for submarine cable fault monitoring. The fault type of the submarine cable is encoded and used as output samples. The input and output samples constitute training samples for establishing a neural network model. The submarine cable fault types include single-phase grounding faults, phase-to-phase short circuit faults, open circuit faults, and other faults. A BP neural network model is constructed based on the input variables and the output fault type, and the topological structure of the BP neural network is determined; wherein the constructed BP neural network model includes an input layer, a hidden layer, and an output layer, each layer is composed of a number of neurons, and the neurons are connected in a fully connected manner; The training samples are input into the constructed BP neural network model for training, and a multi-objective optimization algorithm is used to optimize the structure and parameters of the BP neural network model during the training process to obtain the optimal BP neural network model; Among them, a multi-objective optimization algorithm is used to optimize the structure and parameters of the BP neural network model during the training process, including: Initialize the constructed BP neural network model and randomly generate a set of weight matrices V and W, which are expressed as follows: Where V is the weight matrix between the input layer and the hidden layer, v ij is the element in the i-th row and j-th column of the weight matrix V, i = 1, 2, ..., 3p; j = 1, 2, ..., m; W is the weight matrix between the hidden layer and the output layer, w j1 Represents the element in the jth row and first column of the weight matrix W; Based on the initialized BP neural network model, the training samples are forward propagated from the input layer to the hidden layer, and the output result Y of the hidden layer is expressed as follows: Y=(Y1,Y2,…Y j …Y m ) T (4) Where Y j Represents the jth element in the matrix of the output result Y, j = 1, 2, ... m, f is the activation function; The output result O of the output layer is expressed as follows: O=(O1,O2,O3,O4) T (5) Where O1, O2, O3, and O4 are single-phase grounding fault, interphase short circuit fault, open circuit fault, and other faults, respectively. The calculation formulas for O1, O2, O3, and O4 are as follows: Where w jk is an element in the weight matrix W, k = 1, 2, 3, 4; Calculate the mean error E between the output result O and the expected output D; The objective function is established by combining the number of connection weights and the mean error E of the BP neural network; Based on the objective function, the number of hidden layers and neurons is repeatedly adjusted, and the weight matrices V and W are iteratively updated through the gradient descent method to reduce the error between the output result and the expected output until the value error E meets the set network accuracy requirements.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for monitoring the online status of a submarine cable according to any one of claims 1 to 6 are implemented.

9. A method for recovering a submarine cable fault, characterized in that: The method comprises: A monitoring result of a submarine cable fault according to the submarine cable online status monitoring method according to any one of claims 1 to 6; If the fault type of the monitored submarine cable is a phase-to-phase short circuit fault, the sheath circulating current of the submarine cable is controlled within a safe range by adjusting the impedance series connection.

Images