Cable sheath grounding topology identification method based on BP neural network
Through the cable sheath grounding topology identification method based on BP neural network, the problem of lag in the grounding fault identification of high-voltage cable sheath is solved, achieving fast and accurate fault identification and efficient maintenance efficiency.
Patent Information
- Application Number
- CN202510087852.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-11-26
- Filing Date
- 2025-01-20
- Publication Date
- 2025-05-27
AI Technical Summary
The prior art is difficult to quickly and accurately identify the grounding fault of the high-voltage cable sheath, resulting in overheating of the cable and reducing the load-bearing capacity. The traditional inspection methods are lagging behind, making it difficult to meet the needs of rapid fault identification.
The cable shelf grounding topology identification method based on BP neural network is adopted. By establishing a three-phase nine-stage cross-connected high-voltage cable model, the shelf current is simulated, and the state quantity is trained by using the BP neural network to establish an identification model.
It realizes the rapid and accurate identification of grounding faults of high-voltage cable cover, improves the efficiency of fault maintenance of cover, and meets the needs of rapid fault identification.
Smart Images
Figure CN120046474A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of high-voltage cable monitoring, and particularly to a method for identifying the grounding topology of a cable sheath based on a BP neural network. Background Art
[0002] High-voltage cables (110 kV and above) have become an important part of urban power transmission and distribution networks. However, defects in the cable sheath can lead to grounding faults in the sheath, and high sheath current can cause the cable to overheat, reduce the cable's load-carrying capacity, and have a negative impact on the operation and service life of high-voltage cables. Therefore, it is necessary to monitor the status of high-voltage cable sheaths and identify early defects. Traditional monitoring methods mostly rely on periodic inspections and maintenance of cables, which are often lagging and difficult to meet the requirements of rapid fault identification. Intelligent monitoring methods based on modern information technology, especially methods based on artificial neural networks (such as BP neural networks), can achieve automatic identification and diagnosis of sheath grounding faults by learning and analyzing data on cable defect states. Summary of the Invention
[0003] The purpose of the embodiments of this application is to provide a method for identifying the grounding topology of a cable sheath based on a BP neural network, so as to achieve rapid and accurate identification of grounding faults in high-voltage cable sheaths.
[0004] To achieve the above purpose, this application provides the following technical solutions:
[0005] The embodiments of this application provide a method for identifying the grounding topology of a cable sheath based on a BP neural network, including the following specific steps:
[0006] Step 1: Establish a three-phase nine-section cross-connected high-voltage cable model;
[0007] Step 2: Based on the established model, simulate and calculate the sheath current of the high-voltage cable;
[0008] Step 3: Sample construction and BP neural network optimization;
[0009] Step 4: Use the BP neural network to train the state variables and establish a cable sheath grounding topology recognition model based on the BP neural network.
[0010] In the three-phase nine-section cross-connected high-voltage cable model, the cable sheath is divided into nine small sections A1 - A3, B1 - B3, and C1 - C3, two direct grounding boxes G1 and G2, two cross-connected grounding boxes C1 and C2, the cable sheath is connected with phase conversion in the two cross-connected grounding boxes, and six intermediate joints J1 to J6.
[0011] The high-voltage cable sheath current is the superposition of the sheath circulating current and the leakage current. The sheath circulating current is formed due to inductive coupling, and the leakage current is determined by the capacitive component.
[0012] The sample construction and BP neural network optimization are specifically as follows:
[0013] Randomly select a type of sheath defect status number to determine the required cable structure parameters;
[0014] Call the corresponding cable parameter value range;
[0015] According to the selected status, determine the corresponding topological structure and select the corresponding sheath current calculation formula according to the circuit topological structure to obtain the sheath currents at the head and tail ends;
[0016] The characteristic quantities are the amplitudes and phases of the sheath currents at the head and tail ends, and the amplitude ratios of the sheath currents of the three circuits. After obtaining the characteristic quantities, ensure that 1000 cases are generated for each status number, a total of 9000 cases are generated, and the samples are divided into a training set and a test set at a ratio of 4:1. Input the input features in the training set, the amplitudes, phases, and amplitude ratios of the sheath currents, into the BP neural network, and at the same time use the corresponding defect status number as the target output;
[0017] During the training process, the neural network will continuously adjust its internal parameters to minimize the output error. The training process may require multiple iterations until the performance of the network reaches the expected value.
[0018] The sheath current calculation formula is:
[0019]
[0020] U sm =[-E A1 0 E B2 0 E C3 (0.12)
[0021] U m =[-E A1 0 E B2 0 E C3 (0.13)
[0022]
[0023] Similarly, the leakage current calculation is as follows:
[0024]
[0025] U l =[0 0 0 0 0 U A 0 U B 0 U C0] (0.17)
[0026]
[0027] Wherein, A m is the incidence matrix, B f is the fundamental loop matrix, U sm is the voltage source vector, Z m is the branch impedance matrix, I bm is the induced current vector, A l is the leakage current circuit incidence matrix, U l is the voltage source vector, Z l is the branch impedance matrix, I bl is the leakage current vector.
[0028] Compared with the prior art, the beneficial effects of the present application are as follows: a high-voltage cable sheath grounding fault model is proposed, and a cable sheath grounding topology identification method based on a BP neural network is designed, which can improve the efficiency of sheath fault maintenance at the present stage. BRIEF DESCRIPTION OF THE DRAWINGS
[0029] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required to be used in the embodiments of the present application. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can also be obtained based on these drawings without creative efforts.
[0030] Figure 1 is a schematic diagram of a three-phase nine-section cross-connected high-voltage cable of the present application;
[0031] Figure 2 is an equivalent schematic diagram of capacitive coupling of the present application;
[0032] Figure 3 is an equivalent schematic diagram of inductive coupling of the present application;
[0033] Figure 4 is a flow chart of sample generation and model training of the present application;
[0034] Figure 5 is a flow chart of the GA optimization algorithm of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0035] The following will describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.
[0036] The term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element qualified by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the said element.
[0037] The terms "first", "second", etc. are used only to distinguish one entity or operation from another entity or operation, and cannot be construed as indicating or implying relative importance, nor can it be construed as requiring or implying any such actual relationship or order between these entities or operations.
[0038] The present invention intends to adopt the amplitude and phase of the sheath current at the head end, the amplitude and phase of the phasor difference between the sheath currents at the head and tail ends, and the amplitude ratio of the sheath currents of the three circuits as characteristic quantities, and different cable sheath grounding defects as output quantities.
[0039] A method for identifying the grounding topology of a cable sheath based on a BP neural network includes the following specific steps:
[0040] Step 1, establish a three-phase nine-section cross-connected high-voltage cable model;
[0041] Step 2, based on the established model, simulate and calculate the sheath current of the high-voltage cable;
[0042] Step 3, sample construction and BP neural network optimization;
[0043] Step 4, use the BP neural network to train the state variables and establish a cable sheath grounding topology recognition model based on the BP neural network.
[0044] As Figure 1 shown, the sheath of the high-voltage cable is divided into nine small sections A1 - C3, G1 and G2 are two direct grounding boxes, C1 and C2 are two cross-connected grounding boxes, the sheath of the high-voltage cable is transposed and connected inside the two cross-connected grounding boxes, and J1 to J6 are 6 intermediate joints.
[0045] 2) Sheath current calculation
[0046] The sheath current of the high-voltage cable is the superposition of the sheath circulating current and the leakage current. The sheath circulating current is formed due to inductive coupling, and the leakage current is mainly determined by its capacitive component. Figure 1 and Figure 2 are respectively the equivalent schematic diagrams of capacitive coupling and inductive coupling of the cable circuits A1 - B2 - C3. Among them, Z L1 and ZL5 is the impedance of the sheath protector, U A , U B , U C are the core voltages, E A1 , E B2 , E C3 are the induced voltages, Z iA1 , Z iB2 , Z iC3 is the cable insulation impedance, Z SA1 , Z SB2 , Z SC3 is the cable sheath impedance. According to the relevant theory of the electrical network, the induced current is calculated by listing matrices, and the formulas are shown in 1.1 - 1.5.
[0047]
[0048] U sm = [-E A1 0 E B2 0 E C3 (0.21)
[0049] U m = [-E A1 0E B2 0E C3 (0.22)
[0050]
[0051] Similarly, the leakage current is calculated as shown in formulas 1.6 - 1.9.
[0052]
[0053]
[0054] U l = [0 0 0 0 0U A 0U B 0U C 0](0.26)
[0055]
[0056] In the formula, A m is the incidence matrix, B f is the fundamental loop matrix, U sm is the voltage source vector, Z m is the branch impedance matrix, I bm is the induced current vector, A l is the incidence matrix of the leakage current circuit, U l is the voltage source vector, Z lis the branch impedance matrix, and I bl is the leakage current vector.
[0057] The cable-related parameters are shown in Table 1.
[0058] Table 1 Cable-related parameters
[0059]
[0060] 3) Sample construction and BP neural network optimization
[0061] First, randomly select a class of sheath defect status numbers to determine the required cable structure parameters. Next, call the corresponding cable parameter value ranges. Subsequently, according to the selected status, determine the corresponding topological structure and select the corresponding sheath current calculation formula according to the circuit topological structure to obtain characteristic quantities. The characteristic quantities are the amplitude and phase of the sheath current at the head end, and the amplitude and phase of the phasor difference between the sheath currents at the head and tail ends. In addition, it is also necessary to calculate the amplitude ratio of the sheath currents of the three loops to comprehensively understand the current characteristics of the cable. These characteristic quantities will provide an important basis for subsequent analysis and help to deeply understand the operating state of the cable. The flow chart is as Figure 4 shown.
[0062] The simple BP neural network algorithm has poor identification effect on the grounding topology of high-voltage cable sheaths. Here, GA-BP and SA-BP are introduced for training, and the training result indicators are compared.
[0063] The steps of the GA-BP neural network are as follows.
[0064] Step 1: Convert the variables into chromosomes and select a suitable coding scheme. Select a suitable population size and k1, k2, k3, k4, determine the fitness function, and randomly generate an initial population (containing M individuals). Step 2: Selection. Calculate the cumulative probability qi and selection probability Pi of each string using Equation 1.10. Calculate the fitness through the fitness proportion method, and put the individuals with large fitness back into the population.
[0065]
[0066] where: f i is the fitness of each string of chromosomes; F is the total fitness; P i is the selection probability of each string of chromosomes; q i is the cumulative probability; M is the number of individuals in the population.
[0067] Step 3: Crossover. Select two individuals from the population and randomly select one or more chromosome positions for exchange and combination to generate a better individual. Step 4: Mutation. Select an individual from the population for mutation to generate a better individual.
[0068] The flowchart of the GA-BP neural network is as Figure 5 shown. Since the GA optimization algorithm uses fixed genetic variation parameters and performs the same genetic operations on individuals with different fitness values, its accuracy is poor and it cannot well solve the problem of excessive computational load.
[0069] Simulated annealing is a stochastic optimization algorithm inspired by the annealing process in physics. It avoids getting trapped in local optima by allowing the occasional acceptance of worse solutions, thus having the possibility of finding the global optimum. In SA-BP, simulated annealing is used to optimize the weights and biases of the BP network. The SA-BP algorithm combines simulated annealing and backpropagation. By randomly initializing the weights and biases of the BP network, new solutions are iteratively generated and their fitness is evaluated. New solutions can be accepted with a certain probability even if their fitness is worse, thus avoiding getting trapped in local optima. After each iteration, the temperature gradually decreases, controlling the probability of accepting worse solutions, and finally, when the termination condition is met, the optimized weights and biases are output.
[0070] 4) Result comparison
[0071] The prediction results of various algorithms are compared as shown in Table 2.
[0072] Table 2 Accuracy of various algorithms
[0073]
[0074] The above are only the embodiments of the present application and are not used to limit the protection scope of the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A cable sheath grounding topology identification method based on BP neural network, characterized in that: The specific steps include: Step 1: Establish a three-phase nine-section cross-connected high-voltage cable model; Step 2: Based on the established model, simulate and calculate the sheath current of the high-voltage cable; Step 3: Sample construction and BP neural network optimization; Step 4: Use BP neural network to train the state quantity and establish a cable sheath grounding topology recognition model based on BP neural network.
2. The cable sheath grounding topology identification method based on BP neural network according to claim 1 is characterized in that: In the three-phase nine-section cross-interconnected high-voltage cable model, the high-voltage cable sheath is divided into nine small sections, namely A1-A3, B1-B3, and C1-C3, two direct grounding boxes G1 and G2, and two cross-interconnected grounding boxes C1 and C2. The high-voltage cable sheath is transposed and connected in the two cross-interconnected grounding boxes, with 6 intermediate joints J1 to J6.
3. The cable sheath grounding topology identification method based on BP neural network according to claim 1 is characterized in that: The sheath current of the high-voltage cable is the superposition of the sheath circulating current and the leakage current. The sheath circulating current is formed due to inductive coupling, and the leakage current is determined by the capacitive component.
4. The cable sheath grounding topology identification method based on BP neural network according to claim 1 is characterized in that: The sample construction and BP neural network optimization are specifically as follows: Randomly select a type of sheath defect state number to determine the required cable structural parameters; Call the corresponding cable parameter value range; According to the selected state, the corresponding topological structure is determined and the corresponding sheath current calculation formula is selected according to the circuit topological structure to obtain the sheath current at the first and the end; The characteristic quantities are the amplitude and phase of the sheath current at the head and end, and the amplitude ratio of the sheath current of the three loops. After obtaining the characteristic quantities, ensure that 1000 cases of each state number are generated, and a total of 9000 cases are generated. The samples are divided into a training set and a test set at a ratio of 4:
1. The input features in the training set, the amplitude, phase, and amplitude ratio of the sheath current are input into the BP neural network, and the corresponding defect state number is used as the target output; During the training process, the neural network will continuously adjust its internal parameters to minimize the output error. The training process may require multiple iterations until the network's performance reaches the expected level.
5. The cable sheath grounding topology identification method based on BP neural network according to claim 4 is characterized in that: The sheath current calculation formula is: IN sm =[-E A1 0E B2 0E C3 ](0.3) Similarly, the leakage current is calculated as follows: U l = [0 0 0 0 0U A 0U B 0U C 0] (0.8) In the formula, A m is the incidence matrix, B f is the basic loop matrix, U sm is the voltage source vector, Z m is the branch impedance matrix, I bm is the induced current vector, A l is the leakage current circuit correlation matrix, U l is the voltage source vector, Z l is the branch impedance matrix, I bl is the leakage current vector.