A method and device for diagnosing moisture defect of cable intermediate joint

Abstract

The application discloses a cable intermediate joint moisture defect diagnosis method and device, which comprises the following steps: obtaining a one-dimensional time domain signal of a cable intermediate joint; adopting three one-dimensional convolution layers in series to perform defect positioning according to the one-dimensional time domain signal, and determining a cable defect coordinate; adopting a fully connected network to perform defect diagnosis according to the cable defect coordinate, and generating a moisture defect diagnosis result; and the technical problem of low diagnosis efficiency caused by the existing cable intermediate joint moisture defect diagnosis technology is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of cable defect diagnosis, and in particular to a cable intermediate joint moisture defect diagnosis method and device. BACKGROUND

[0002] Power distribution cables are basically laid underground, and the operating environment is usually harsh, especially in the rainy south. In this environment, power cables are long-term affected by water, electric field, temperature and mechanical stress, etc., which affects the health status and normal operation of the cable.

[0003] Compared with the cable body, the cable intermediate joint structure is more complex, and there are multiple composite interfaces composed of different media. These composite interfaces are prone to water intrusion under the long-term action of various environmental factors, which seriously affects the insulation performance of the cable intermediate joint, and even causes breakdown accidents. Therefore, it is of great significance to carry out moisture defect positioning and detection of the cable intermediate joint for maintaining the stable operation of the cable system.

[0004] Most of the existing moisture defect diagnosis techniques for cable intermediate joints are based on the TFDR (Time-Frequency Domain Reflectometry) method for cable defect positioning and diagnosis. It needs to be manually judged whether there is a defect in the positioning curve obtained, how many defects there are, the specific location of each defect, and the severity of the defect, etc. The above process relies too much on the knowledge and experience of the operator, and the degree of automation is low, resulting in poor diagnosis efficiency. SUMMARY

[0005] The present application provides a cable intermediate joint moisture defect diagnosis method and device to solve the technical problem of poor diagnosis efficiency caused by the existing moisture defect diagnosis techniques for cable intermediate joints.

[0006] The first aspect of the present application provides a cable intermediate joint moisture defect diagnosis method, comprising:

[0007] Obtaining a one-dimensional time domain signal of the cable intermediate joint;

[0008] Using three one-dimensional convolution layers in series to perform defect positioning according to the one-dimensional time domain signal and determine the cable defect coordinates;

[0009] Using a fully connected network to perform defect diagnosis according to the cable defect coordinates and generate a moisture defect diagnosis result.

[0010] Optionally, the step of using three one-dimensional convolution layers in series to perform defect positioning according to the one-dimensional time domain signal and determine the cable defect coordinates comprises:

[0011] inputting the one-dimensional time domain signal into the three one-dimensional convolution layers in series for feature extraction, and outputting a probability distribution graph;

[0012] determining a time difference based on a probability peak value in the probability distribution graph;

[0013] calculating cable defect coordinates according to the time difference by using a preset coordinate function.

[0014] Optionally, the step of adopting the full connection network to perform defect diagnosis according to the cable defect coordinates and generate a damp defect diagnosis result comprises:

[0015] performing data interception at a cable defect position corresponding to the cable defect coordinates, and determining a time domain positioning signal local maximum horizontal and vertical coordinates, an upper envelope line local maximum horizontal and vertical coordinates, and a phase feature at the cable defect position;

[0016] inputting the time domain positioning signal local maximum horizontal and vertical coordinates, the upper envelope line local maximum horizontal and vertical coordinates, and the phase feature at the cable defect position into the full connection network for prediction, and outputting a defect prediction value;

[0017] generating a damp defect diagnosis result according to a comparison result of the defect prediction value and a preset comparison threshold.

[0018] Optionally, the damp defect diagnosis result comprises a first damp defect diagnosis result and a second damp defect diagnosis result; and the step of generating a damp defect diagnosis result according to a comparison result of the defect prediction value and a preset comparison threshold comprises:

[0019] comparing the defect prediction value and the preset comparison threshold;

[0020] if the defect prediction value is greater than the preset comparison threshold, generating a first damp defect diagnosis result; and the first damp defect diagnosis result is that a damp defect exists in the cable intermediate joint;

[0021] if the defect prediction value is less than or equal to the preset comparison threshold, generating a second damp defect diagnosis result; and the second damp defect diagnosis result is that a damp defect does not exist in the cable intermediate joint.

[0022] Optionally, the method further comprises:

[0023] obtaining a to-be-trained simulation one-dimensional time domain signal;

[0024] adopting three one-dimensional convolution layers in series to perform defect positioning according to the to-be-trained simulation one-dimensional time domain signal, and determining to-be-trained cable defect coordinates;

[0025] An initial full connection network is adopted to perform prediction according to the cable defect coordinates to be trained, and an output is obtained as a defect prediction value to be trained.

[0026] Based on the defect prediction value to be trained, a model gradient is determined, the model parameter of the initial full connection network is updated by using the model gradient, an intermediate full connection network is determined, and the number of model updates is counted in real time.

[0027] The intermediate full connection network is adopted to perform prediction according to the cable defect coordinates to be trained, and an intermediate defect prediction value is outputted, and a target loss value is calculated according to the intermediate defect prediction value.

[0028] It is judged whether the target loss value converges or the number of model updates reaches a preset update threshold.

[0029] If the target loss value converges or the number of model updates reaches the preset update threshold, the intermediate full connection network is taken as the trained full connection network.

[0030] Optionally, a preset coordinate function is provided, and specifically,

[0031] ;

[0032] Wherein, d is the cable defect coordinate, v represents the wave speed of the electromagnetic wave in the cable, and Δt is the time difference between the reflected wave and the incident wave.

[0033] The second aspect of the present application provides a cable intermediate joint moisture defect diagnosis device, comprising:

[0034] An acquisition module is configured to acquire a one-dimensional time domain signal of a cable intermediate joint.

[0035] A positioning module is configured to perform defect positioning according to the one-dimensional time domain signal by using three one-dimensional convolution layers connected in series, and determine a cable defect coordinate.

[0036] A diagnosis module is configured to perform defect diagnosis according to the cable defect coordinate by using a full connection network, and generate a moisture defect diagnosis result.

[0037] The third aspect of the present application provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the computer program is executed by the processor to make the processor execute the steps of the cable intermediate joint moisture defect diagnosis method according to any one of the above.

[0038] The fourth aspect of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed to realize the steps of the cable intermediate joint moisture defect diagnosis method according to any one of the above.

[0039] The fifth aspect of the present application provides a computer program product, the computer program product comprises a computer program stored on a non-transitory computer readable storage medium, the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the steps of the cable intermediate joint moisture defect diagnosis method according to any one of the above.

[0040] From the above technical solutions, the present application has the following advantages:

[0041] The above technical solutions of the present application provide a cable intermediate joint moisture defect diagnosis method, first acquire a one-dimensional time domain signal of the cable intermediate joint; then, three one-dimensional convolution layers in series are used to locate defects according to the one-dimensional time domain signal, and determine the cable defect coordinates; finally, a full connection network is used to diagnose defects according to the cable defect coordinates, and generate a moisture defect diagnosis result; based on the above scheme, the three one-dimensional convolution layers in series are used to locate defects according to the acquired one-dimensional time domain signal, and the full connection network is used to diagnose defects according to the generated cable defect coordinates, and the process of outputting the moisture defect diagnosis result, without human judgment of whether there is a defect in the final obtained positioning curve, how many defects exist, the specific position of each defect and the severity of the defect, etc., saves the process of human judgment, and further improves the diagnosis efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0043] Figure 1 A step flow chart of a cable intermediate joint moisture defect diagnosis method provided by the embodiment of the present application;

[0044] Figure 2 A structural schematic diagram of three one-dimensional convolution layers in series provided by the embodiment of the present application;

[0045] Figure 3 A structural schematic diagram of a full connection network provided by the embodiment of the present application;

[0046] Figure 4 A step flow chart of another cable intermediate joint moisture defect diagnosis method provided by the embodiment of the present application;

[0047] Figure 5 A circuit structural schematic diagram of a cable distributed parameter model provided by the embodiment of the present application;

[0048] Figure 6 A structural schematic diagram of a signal transmission model of a single-defect cable provided for an embodiment of the present application is shown in FIG. 1.

[0049] Figure 7 A schematic diagram of a TFDR data waveform provided for an embodiment of the present application is shown in FIG. 2.

[0050] Figure 8 A schematic diagram of a defect position probability distribution provided for an embodiment of the present application is shown in FIG. 3.

[0051] Figure 9 A structural block diagram of a cable intermediate joint moisture defect diagnosis device provided for an embodiment of the present application is shown in FIG. 4. DETAILED DESCRIPTION

[0052] Embodiments of the present application provide a cable intermediate joint moisture defect diagnosis method and device, which are used to solve the technical problem of poor diagnosis efficiency caused by the existing cable intermediate joint moisture defect diagnosis technology.

[0053] In order to make the technical features and advantages of the present application more obvious and easy to understand, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the embodiments described below are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0054] Please refer to Figure 1 , Figure 1 A step flowchart of a cable intermediate joint moisture defect diagnosis method provided for an embodiment of the present application is shown in FIG. 5.

[0055] The cable intermediate joint moisture defect diagnosis method provided by the present application comprises the following steps.

[0056] Step 101: Obtain a one-dimensional time domain signal of a cable intermediate joint.

[0057] It should be noted that the obtained one-dimensional time domain signal of the cable intermediate joint is a one-dimensional time domain signal directly obtained through TFDR testing, which can avoid the interference of cross terms in time-frequency analysis, does not exist signal distortion, and has smaller data dimension.

[0058] In this embodiment, a one-dimensional time domain signal of a cable intermediate joint is obtained.

[0059] Step 102: Adopt three one-dimensional convolution layers connected in series to perform defect positioning according to the one-dimensional time domain signal, and determine the cable defect coordinates.

[0060] Specifically, a one-dimensional time domain signal is input to three one-dimensional convolution layers connected in series for feature extraction, and a probability distribution map is output; based on the probability peak value in the probability distribution map, the time difference is determined; and the cable defect coordinates are calculated according to the time difference by using a preset coordinate function.

[0061] It is worth mentioning that, please refer to Figure 2 In order to realize the positioning of the defect position, the application introduces a knowledge-driven idea, converts the original classification problem into a probability distribution problem for processing, that is, judges the probability of the existence of a defect at a certain point in the cable and its vicinity. When the label of the defect position is a specific position, the probability distribution is highly concentrated on the defect position, so this task is unreasonable and not easy to learn. Considering that the time domain positioning signal has the characteristic of time width, a Gaussian distribution is superimposed around the defect position as the probability distribution of the occurrence of the defect, which makes the probability distribution around the defect very smooth and more conducive to network learning. The physical meaning is: the farther away from the defect position, the lower the probability; the closer to the defect position, the higher the probability. The same operation is performed on the cable head and the cable tail, and finally the defect positioning probability distribution spectrum containing the head and tail is obtained; wherein, Figure 2 D in the formula is the data size (i.e. the number of sampling points) of the one-dimensional time domain signal.

[0062] Further, for one-dimensional data (one-dimensional time domain signal), three one-dimensional convolution layers connected in series are used for feature extraction, and finally the probability distribution of the defect position is obtained. With each increase in defects, a peak value of the probability distribution in the spectrum will be added. The maximum value of each peak value of the obtained probability distribution spectrum is taken to obtain the corresponding point of each defect, that is, the time difference is determined according to the probability peak value in the probability distribution map, and then the position of each defect can be calculated according to the time difference by using a preset coordinate function; wherein, the time difference is the time difference between the reflected wave and the incident wave, and even if it is a normal joint (without defects), a peak value will appear after obtaining the probability distribution map, that is, the positioning of the joint is realized; the preset coordinate function is specifically:

[0063] ;

[0064] Wherein, d is the cable defect coordinate; v represents the wave speed of the electromagnetic wave in the cable; and Δt is the time difference between the reflected wave and the incident wave.

[0065] In this embodiment, three one-dimensional convolution layers connected in series are used to perform defect positioning according to the one-dimensional time domain signal, and the cable defect coordinates are determined.

[0066] Step 103, using a fully connected network to perform defect diagnosis according to the cable defect coordinates, and generating a damp defect diagnosis result.

[0067] The moisture defect diagnosis result includes a first moisture defect diagnosis result and a second moisture defect diagnosis result.

[0068] It should be noted that after obtaining the coordinates corresponding to the defect points in the one-dimensional data, the time-domain waveform at the cable defect position corresponding to the cable defect coordinates can be intercepted for extracting the time delay, peak voltage and phase information, and the extracted data is input into the full connection network. Specifically, data is intercepted at the cable defect position corresponding to the cable defect coordinates to determine the local maximum horizontal and vertical coordinates of the time-domain positioning signal, the local maximum horizontal and vertical coordinates of the upper envelope line and the phase feature at the cable defect position. The local maximum horizontal and vertical coordinates of the time-domain positioning signal are the horizontal and vertical coordinates corresponding to the local maximum of the defect time-domain positioning signal, which include the horizontal coordinate t r and the vertical coordinate V m corresponding to the local maximum of the defect time-domain positioning signal. The local maximum horizontal and vertical coordinates of the upper envelope line are the horizontal and vertical coordinates corresponding to the local maximum of the upper envelope line, which include the horizontal coordinate t up and the vertical coordinate V up corresponding to the local maximum of the upper envelope line. The phase feature at the cable defect position is the phase feature φ at the defect position.

[0069] Further, in order to intuitively reflect the phase feature φ of the time-domain waveform, the phase is [1, 0, -1] when the maximum value of the time-domain waveform is located to the left of the maximum value of the envelope line and the maximum value of the time-domain waveform is smaller than the maximum value of the envelope line. The phase is [0, 0, 0] when neither the envelope line nor the time-domain waveform has a positioning peak value. The phase is [-1, 0, 1] when the maximum value of the time-domain waveform is located to the right of the maximum value of the envelope line and the maximum value of the time-domain waveform is smaller than the maximum value of the envelope line.

[0070] Further, please refer to Figure 3, after the above steps, the horizontal and vertical coordinates of the local maximum of the time-domain positioning signal, the horizontal and vertical coordinates of the local maximum of the upper envelope line, and the phase characteristics at the cable defect position are obtained, which are input into the full connection network for learning and prediction, and the output result is the ratio of the capacitance at the defect to the capacitance of the cable body (defect prediction value). That is, the horizontal and vertical coordinates of the local maximum of the time-domain positioning signal, the horizontal and vertical coordinates of the local maximum of the upper envelope line, and the phase characteristics at the cable defect position are input into the full connection network for prediction, and the defect prediction value is output. Then, the defect prediction value and the preset comparison threshold are compared. If the defect prediction value is greater than the preset comparison threshold, a first moisture defect diagnosis result is generated. The first moisture defect diagnosis result is that the intermediate joint of the cable has a moisture defect. If the defect prediction value is less than or equal to the preset comparison threshold, a second moisture defect diagnosis result is generated. The second moisture defect diagnosis result is that the intermediate joint of the cable does not have a moisture defect. The full connection network includes three hidden layers, the first hidden layer includes 32 nodes, the second hidden layer includes 64 nodes, and the third hidden layer includes 128 nodes. The preset comparison threshold can be set as needed, and the present application is not limited thereto.

[0071] For example, assuming that the preset comparison threshold is 0.9, the capacitance of a normal intermediate joint is about 0.85 of the capacitance of the cable body. When the value is greater than 0.9, it means that the cable has a moisture defect, and the greater the value, the more serious the moisture defect. If it is less than or equal to 0.9, it means that the cable does not have a moisture defect.

[0072] In this embodiment, a full connection network is used to diagnose defects according to the cable defect coordinates to generate a moisture defect diagnosis result.

[0073] In the embodiment of the present application, a cable intermediate joint moisture defect diagnosis method is provided. First, a one-dimensional time-domain signal of the cable intermediate joint is obtained. Then, three one-dimensional convolution layers in series are used to locate defects according to the one-dimensional time-domain signal to determine the cable defect coordinates. Finally, a full connection network is used to diagnose defects according to the cable defect coordinates to generate a moisture defect diagnosis result. Based on the above scheme, the process of locating defects according to the obtained one-dimensional time-domain signal by three one-dimensional convolution layers in series, and diagnosing defects according to the generated cable defect coordinates by the full connection network to output the moisture defect diagnosis result, does not require manual judgment of whether there is a defect in the final obtained positioning curve, how many defects there are, the specific position of each defect, and how serious the defect is, etc. The process of manual judgment is saved, and the diagnosis efficiency is further improved.

[0074] Please refer to Figure 4 , Figure 4 The step flow chart of another cable intermediate joint moisture defect diagnosis method provided in the embodiment of the present application.

[0075] The application provides a cable intermediate joint damp defect diagnosis method, which comprises the following steps.

[0076] Step 401, obtaining a one-dimensional time domain signal of a cable intermediate joint.

[0077] In this embodiment, a one-dimensional time domain signal of a cable intermediate joint is obtained.

[0078] Step 402, inputting the one-dimensional time domain signal into three one-dimensional convolution layers in series for feature extraction, and outputting a probability distribution graph.

[0079] It should be noted that, please refer to Figure 5 According to the transmission line theory, under high frequency conditions, a coaxial structure power cable line can be equivalent to a cascade of multiple unit distributed parameter equivalent circuits, wherein R (Ω / m), L (H / m), G (S / m) and C (F / m) are the resistance, inductance, conductance and capacitance per unit length of the power cable, and the characteristic impedance Z0 of the power cable per unit length is determined according to the above four parameters, and the calculation formula can be expressed as:

[0080] ;

[0081] Wherein, R is the resistance per unit length of the power cable; L is the inductance per unit length of the power cable; G is the conductance per unit length of the power cable; Z0 is the characteristic impedance per unit length of the cable; C is the capacitance per unit length of the power cable; and j is an imaginary unit.

[0082] Further, please refer to Figure 6 According to the transmission line theory, when a defect occurs in the cable, the characteristic impedance and propagation constant of the defect section change, and when the signal transmission encounters the defect section, the impedance will suddenly change and the reflection phenomenon will occur, and at the same time, the change of the cable head transfer function will also occur, denoted as Figure 6 There are three signal reflection points, and the positions of the three reflection points are respectively x=l 11 , x=l 12 , x=l, x=l 11 , x=l 12 , x=l. According to the principle of traveling wave reflection, the expression of the reflection coefficient is as follows:

[0083] ;

[0084] Wherein, is the reflection coefficient at a distance of l 11 from the cable head; is the reflection coefficient at a distance of l 12 from the cable head, and The same size and opposite signs; is the reflection coefficient at the cable end l; Z l is the characteristic impedance of the defect; Z0 is the characteristic impedance of the intact cable; Z0 is the end load impedance; Z L is the characteristic impedance of the end load.

[0085] Further, based on the single-defect cable signal transmission model, the reflection signal tested by the head-end contains two parts, respectively, the reflection signal at the defect and the cable end reflection signal. Among them, the reflection at the defect consists of the reflection U 11 caused by the defect head-end and the reflection U 12 caused by the defect end, and the reflection signal H1 of the total reflection signal at the defect can be represented by the following formula:

[0086] ;

[0087] Among them, H1 is the reflection signal of the total reflection signal at the defect; U 11 is the reflection caused by the defect head-end; U 12 is the reflection caused by the defect end; U0 is the incident signal; γ0 is the propagation coefficient of the signal in the intact cable section; γ1 is the propagation coefficient of the signal in the defective cable section.

[0088] Based on the above basis, the cable end reflection signal H2 can be represented as:

[0089] ;

[0090] Among them, U represents the reflection caused by the cable end; U0 is the incident signal; is the reflection coefficient at the cable end l; Z is the reflection coefficient at the left side of the impedance mismatch section, i.e. the distance from the cable head-end l 11 ; The size is the same as but the sign is opposite; γ0 is the propagation coefficient of the signal in the intact cable section; γ1 is the propagation coefficient of the signal in the defective cable section.

[0091] Further, according to the expression of the cable end reflection signal H2 and the reflection signal H1 of the total reflection signal at the defect, the single-defect cable model transfer function G(ω) can be represented as:

[0092] ;

[0093] Among them, G(ω) is the single-defect cable model transfer function; H2 is the cable end reflection signal; H1 is the reflection signal of the total reflection signal at the defect; U0 is the incident signal.

[0094] Based on the above, the defect position in the cable and the reflection information at the defect are contained in the transfer function of the cable, and are also reflected in the time domain waveform; therefore, the one-dimensional time domain signal is input to three one-dimensional convolution layers connected in series for feature extraction, and a probability distribution map is output, and the corresponding cable defect coordinates are determined according to the peak value in the probability distribution map, so that the position of the weak reflection signal in the original signal can be better observed, and the positioning of the defect is realized.

[0095] In this embodiment, the one-dimensional time domain signal is input to three one-dimensional convolution layers connected in series for feature extraction, and a probability distribution map is output.

[0096] Step 403, determining the time difference based on the probability peak value in the probability distribution map.

[0097] In this embodiment, the time difference is determined based on the probability peak value in the probability distribution map.

[0098] Step 404, calculating the cable defect coordinates according to the time difference by using a preset coordinate function.

[0099] In this embodiment, the cable defect coordinates are calculated according to the time difference by using a preset coordinate function.

[0100] Step 405, performing defect diagnosis according to the cable defect coordinates by using a full connection network, and generating a damp defect diagnosis result.

[0101] Further, step 405 can include the following sub-steps:

[0102] S51, data interception is performed on the cable defect position corresponding to the cable defect coordinates, and the local maximum value horizontal and vertical coordinates of the time domain positioning signal, the local maximum value horizontal and vertical coordinates of the upper envelope line, and the phase feature at the cable defect position are determined;

[0103] S52, the local maximum value horizontal and vertical coordinates of the time domain positioning signal, the local maximum value horizontal and vertical coordinates of the upper envelope line, and the phase feature at the cable defect position are input to the full connection network for prediction, and a defect prediction value is output;

[0104] S53, generating a damp defect diagnosis result according to the comparison result of the defect prediction value and the preset comparison threshold.

[0105] The damp defect diagnosis result includes a first damp defect diagnosis result and a second damp defect diagnosis result.

[0106] Further, S53 can include the following sub-steps:

[0107] S531, comparing the defect prediction value and the preset comparison threshold;

[0108] S532, if the defect prediction value is greater than the preset comparison threshold, a first moisture defect diagnosis result is generated; the first moisture defect diagnosis result is that the cable intermediate joint has a moisture defect;

[0109] S533, if the defect prediction value is less than or equal to the preset comparison threshold, a second moisture defect diagnosis result is generated; the second moisture defect diagnosis result is that the cable intermediate joint does not have a moisture defect.

[0110] It should be noted that, in the case where the propagation coefficients of the incident signal, the intact cable segment and the defect cable segment are unchanged, the characteristics of the reflected signal H are related to the position l of the defect. The traditional TFDR method uses the Wigner-Ville distribution to locate the defect, but when the signal is composed of multiple components, the Wigner-Ville distribution may not satisfy the linear superposition principle, resulting in serious cross-term interference, which adversely affects the positioning accuracy. Therefore, the envelope line is used to directly extract the time-domain characteristics of the original defect reflected signal, avoiding the influence of cross-terms in time-frequency domain analysis, and the state evaluation of the defect is completed according to these characteristics.

[0111] Further, for the cable intermediate joint that has not been affected by moisture, the reflection coefficient is positive, and the time-domain waveform reflection peak is positive; when the intermediate joint is affected by moisture but the moisture level is relatively mild, the reflection coefficient will decrease with the increase of the moisture level, but it is still positive, and the polarity of the time-domain waveform reflection peak is also positive; as the moisture level increases, the impedance at the intermediate joint will approach the impedance of the cable body, at which time the reflection coefficient tends to 0, and the time-domain waveform reflection peak disappears; when the moisture reaches a serious level, the polarity of the reflection coefficient reverses, and the time-domain waveform reflection peak is negative, and its size will increase with the increase of the moisture level. Therefore, the polarity and amplitude of the time-domain waveform reflection peak can be used as a feature to judge the moisture level. Secondly, the amplitude of the defect and the position of the defect are also related, and the amplitude of the time-domain reflection peak will decrease with the increase of the propagation distance. Therefore, the time delay of the time-domain waveform reflection peak also needs to be used as an important feature to judge the moisture level.

[0112] In summary, the time delay, peak voltage size and polarity (phase) on the envelope line in the time domain are selected as the TFDR feature parameters reflecting the moisture intermediate position and severity for corresponding defect diagnosis, which can avoid the case that when the signal is composed of multiple components, the Wigner-Ville distribution may not satisfy the linear superposition principle, resulting in serious cross-term interference, which adversely affects the positioning accuracy,

[0113] In this embodiment, a full connection network is used to perform defect diagnosis according to the cable defect coordinates, and generate a moisture defect diagnosis result.

[0114] Optionally, the training process of the full connection network comprises:

[0115] Obtaining a to-be-trained simulated one-dimensional time domain signal;

[0116] Using three one-dimensional convolution layers in series to locate defects according to the to-be-trained simulated one-dimensional time domain signal, and determine the to-be-trained cable defect coordinates;

[0117] Using the initial full connection network to predict according to the to-be-trained cable defect coordinates, and output the to-be-trained defect prediction value;

[0118] Based on the to-be-trained defect prediction value, determine the model gradient, and update the model parameters of the initial full connection network using the model gradient, determine the intermediate full connection network, and real-time statistics model update times;

[0119] Using the intermediate full connection network to predict according to the to-be-trained cable defect coordinates, and output the intermediate defect prediction value, and calculate the target loss value according to the intermediate defect prediction value;

[0120] Determine whether the target loss value converges or the model update times reaches the preset update threshold;

[0121] If the target loss value converges or the model update times reaches the preset update threshold, the intermediate full connection network is used as the trained full connection network.

[0122] It should be noted that if the target loss value does not converge and the model update times does not reach the preset update threshold, the intermediate full connection network is used as a new initial full connection network, and the step of using the initial full connection network to predict according to the to-be-trained cable defect coordinates and output the to-be-trained defect prediction value is executed, until the target loss value converges or the model update times reaches the preset update threshold, the intermediate full connection network determined when the target loss value converges or the model update times reaches the preset update threshold is used as the trained full connection network.

[0123] Further, please refer to Figures 7-8, the initial full connection network to be trained is trained by using simulation data (to be trained simulation one-dimensional time domain signal), and after the training is completed, verification is performed on real cable data; specifically, the TFDR data waveform actually measured on a 40-meter cable and the defect distribution probability obtained by the positioning network, the 40-meter cable has an intermediate joint at 10m and 30m from the test end, and the intermediate joint at 30m is a damp intermediate joint. There are two defects on the 40-meter cable, which are 9.32m and 31.1m, which shows that the network structure designed in the application can effectively position the short cable containing multiple defects. Compared with the simulation data positioning result, the actual measurement data positioning error is within 1.1m, and the error increases. At this time, the defect degree prediction values corresponding to the defects at 10m and 30m are 0.8637 and 0.9533 respectively, that is, the joint at 10m is a normal joint, and the joint at 30m is a damp joint, which corresponds to the known cable damp condition.

[0124] As a comparison of technical effects, reference can be made in combination with the prior art. When a defect occurs in the cable, the characteristic impedance at the defect will change. According to the existing method, the high-frequency electromagnetic wave will be folded and reflected when passing through the impedance mutation point according to the principle of traveling wave reflection. The traveling wave method is to realize the positioning of the cable defect by observing the folding and reflection behavior of the high-frequency electromagnetic wave in the cable. This method is widely used in the current cable defect positioning field because of its simple operation and the fact that it does not require high-voltage excitation. According to the difference of the injected reference signal and the detection method, the traveling wave method can be mainly divided into time domain reflectometry (TDR), frequency domain reflectometry (FDR) and time-frequency domain reflectometry (TFDR). Among them, the TFDR method uses a parameter-adjustable Gaussian envelope linear frequency modulation (LFM) signal as the incident signal, which can carry more information in the time domain and the frequency domain. The positioning of the cable defect can be realized by calculating the time-frequency domain correlation degree of the reflected signal using the time-frequency cross-correlation function. Since the frequency band of the signal injected by the TFDR method is higher and the energy is higher, compared with TDR and FDR, it has a higher detection rate for defects, and can more accurately locate the defect position, thereby improving the efficiency and reliability of the cable detection. However, the application of the cable defect positioning method based on the TFDR method depends on the knowledge and experience of the operator. Whether there is a defect in the positioning curve obtained by human judgment, how many defects exist, the specific position of each defect, and how serious the defect is, etc. The positioning result obtained has a certain subjectivity, which is not conducive to popularization. That is, the application of the cable defect positioning method based on the TFDR method depends on the knowledge and experience of the operator. Whether there is a defect in the positioning curve obtained by human judgment, how many defects exist, the specific position of each defect, and how serious the defect is, etc. The positioning result obtained has a certain subjectivity, which is not conducive to popularization. At the same time, the traditional TFDR method uses Wigner distribution to position the defect. However, when the signal is composed of multiple components, the Wigner distribution may not satisfy the linear superposition principle, which may cause serious cross-term interference, thereby adversely affecting the positioning accuracy.

[0125] To solve the above problems, the application provides a cable intermediate joint moisture defect diagnosis method, which directly takes TFDR data as input, avoids serious cross term interference caused by the fact that the Wigner distribution does not satisfy the linear superposition principle when time-frequency processing is performed, and can better observe the position of a weak reflection signal in the original signal based on a defect probability distribution map (probability distribution atlas) output by three one-dimensional convolution layers connected in series, so that the positioning of the defect is realized; after the position of the defect is obtained, the time delay, peak voltage and phase information in the envelope at the time of the defect position are intercepted as input of the diagnosis network, the severity of the defect is judged by predicting the size of the capacitance at the defect compared with the capacitance of the cable body, and the process of manual judgment is omitted.

[0126] In the embodiment of the application, the application provides a cable intermediate joint moisture defect diagnosis method, which first acquires a one-dimensional time domain signal of a cable intermediate joint, then adopts three one-dimensional convolution layers connected in series to perform defect positioning based on the one-dimensional time domain signal, determines the cable defect coordinates, finally adopts a fully connected network to perform defect diagnosis based on the cable defect coordinates, and generates a moisture defect diagnosis result; based on the above scheme, the three one-dimensional convolution layers connected in series are used to perform defect positioning based on the acquired one-dimensional time domain signal, and the fully connected network is used to perform defect diagnosis based on the generated cable defect coordinates, so that the process of outputting the moisture defect diagnosis result is realized, the presence or absence of a defect in the finally obtained positioning curve, the number of defects, the specific position of each defect and the severity of the defect are not required to be judged manually, the process of manual judgment is omitted, and the diagnosis efficiency is further improved.

[0127] Please refer to Figure 9 , Figure 9 The application provides a structure block diagram of a cable intermediate joint moisture defect diagnosis device.

[0128] The application provides a cable intermediate joint moisture defect diagnosis device, which comprises:

[0129] The acquisition module 901 is used to acquire a one-dimensional time domain signal of a cable intermediate joint.

[0130] The positioning module 902 is used to adopt three one-dimensional convolution layers connected in series to perform defect positioning based on the one-dimensional time domain signal, and determine the cable defect coordinates.

[0131] The diagnosis module 903 is used to adopt a fully connected network to perform defect diagnosis based on the cable defect coordinates, and generate a moisture defect diagnosis result.

[0132] Further, the positioning module 902 is specifically used for:

[0133] inputting the one-dimensional time domain signal into the three one-dimensional convolution layers connected in series for feature extraction, and outputting a probability distribution atlas;

[0134] determine the time difference based on a probability peak in the probability distribution map;

[0135] calculate the cable defect coordinate based on the time difference using the preset coordinate function.

[0136] Further, the diagnosis module 903 comprises:

[0137] The first sub-module is configured to perform data interception at the cable defect position corresponding to the cable defect coordinate, to determine the local maximum horizontal and vertical coordinates of the time-domain positioning signal, the local maximum horizontal and vertical coordinates of the upper envelope line, and the phase feature at the cable defect position.

[0138] The second sub-module is configured to input the local maximum horizontal and vertical coordinates of the time-domain positioning signal, the local maximum horizontal and vertical coordinates of the upper envelope line, and the phase feature at the cable defect position into the full connection network for prediction, and output a defect prediction value.

[0139] The third sub-module is configured to generate a damp defect diagnosis result based on a comparison result of the defect prediction value and a preset comparison threshold.

[0140] Further, the damp defect diagnosis result comprises a first damp defect diagnosis result and a second damp defect diagnosis result; and the third sub-module is specifically configured to:

[0141] compare the defect prediction value with the preset comparison threshold;

[0142] If the defect prediction value is greater than the preset comparison threshold, the first damp defect diagnosis result is generated; and the first damp defect diagnosis result indicates that the damp defect exists in the cable intermediate joint.

[0143] If the defect prediction value is less than or equal to the preset comparison threshold, the second damp defect diagnosis result is generated; and the second damp defect diagnosis result indicates that the damp defect does not exist in the cable intermediate joint.

[0144] In an alternative embodiment, the device further comprises:

[0145] The first module is configured to acquire a to-be-trained simulation one-dimensional time-domain signal.

[0146] The second module is configured to perform defect positioning based on the to-be-trained simulation one-dimensional time-domain signal using three one-dimensional convolution layers connected in series, to determine a to-be-trained cable defect coordinate.

[0147] The third module is configured to perform prediction based on the to-be-trained cable defect coordinate using an initial full connection network, to output a to-be-trained defect prediction value.

[0148] The fourth module is configured to determine a model gradient based on the to-be-trained defect prediction value, to update a model parameter of the initial full connection network using the model gradient, to determine an intermediate full connection network, and to statistically determine a model update number in real time.

[0149] a fifth module configured to employ an intermediate fully connected network to predict according to the cable defect coordinates to be trained, output an intermediate defect prediction value, and calculate a target loss value according to the intermediate defect prediction value;

[0150] a sixth module configured to determine whether the target loss value converges or whether the number of model updates reaches a preset update threshold;

[0151] a seventh module configured to, if the target loss value converges or the number of model updates reaches the preset update threshold, take the intermediate fully connected network as a trained fully connected network.

[0152] Further, the preset coordinate function is specifically:

[0153]

[0154] wherein d is the cable defect coordinate, v represents the wave speed of the electromagnetic wave in the cable, and At is the time difference between the reflected wave and the incident wave.

[0155] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices, modules and sub-modules can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein.

[0156] The embodiment of the present application also provides a computer device, which comprises a memory and a processor, and the memory stores a computer program; the computer program is executed by the processor to make the processor execute the steps of the cable intermediate joint moisture defect diagnosis method according to any one of the foregoing embodiments.

[0157] The embodiment of the present application also provides a computer readable storage medium, which stores a computer program / instruction, and the computer program / instruction is executed by a processor to implement the steps of the cable intermediate joint moisture defect diagnosis method according to any one of the foregoing embodiments.

[0158] The embodiment of the present application also provides a computer program product, which comprises a computer program / instruction, and the computer program / instruction is executed by a processor to implement the steps of the cable intermediate joint moisture defect diagnosis method according to any one of the foregoing embodiments.

[0159] ​In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other manners. For example, the division of the units is only a logical function division, and there can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between the units can be indirect couplings or communication connections through some interfaces, devices or units, and can be in electrical, mechanical or other forms.

[0160] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, can be located in one place, or can be distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0161] The above description and the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of diagnosing a moisture defect of a cable intermediate joint, characterized by, The method comprises the following steps: acquiring a one-dimensional time domain signal of a cable intermediate joint; performing defect positioning according to the one-dimensional time domain signal by using three one-dimensional convolution layers connected in series to determine cable defect coordinates; performing defect diagnosis according to the cable defect coordinates by using a fully connected network to generate a damp defect diagnosis result, comprising: performing data interception at a cable defect position corresponding to the cable defect coordinates to determine a local maximum value horizontal and vertical coordinate of a time domain positioning signal, a local maximum value horizontal and vertical coordinate of an upper envelope line, and a phase feature at the cable defect position; inputting the local maximum value horizontal and vertical coordinate of the time domain positioning signal, the local maximum value horizontal and vertical coordinate of the upper envelope line, and the phase feature at the cable defect position into the fully connected network for prediction to output a defect prediction value; generating a damp defect diagnosis result according to a comparison result of the defect prediction value and a preset comparison threshold.

2. The moisture defect diagnosis method for cable intermediate joint according to claim 1, characterized by, The step of performing defect positioning according to the one-dimensional time domain signal by using three one-dimensional convolution layers connected in series to determine cable defect coordinates comprises: inputting the one-dimensional time domain signal into the three one-dimensional convolution layers connected in series for feature extraction to output a probability distribution map; determining a time difference based on a probability peak value in the probability distribution map; calculating cable defect coordinates according to the time difference by using a preset coordinate function; wherein the time difference is a time difference between a reflected wave and an incident wave.

3. The moisture defect diagnosis method for cable intermediate joint according to claim 1, characterized by, The damp defect diagnosis result comprises a first damp defect diagnosis result and a second damp defect diagnosis result; the step of generating a damp defect diagnosis result according to a comparison result of the defect prediction value and a preset comparison threshold comprises: comparing the defect prediction value and the preset comparison threshold; if the defect prediction value is greater than the preset comparison threshold, generating a first damp defect diagnosis result; the first damp defect diagnosis result is that the cable intermediate joint has a damp defect; if the defect prediction value is less than or equal to the preset comparison threshold, generating a second damp defect diagnosis result; the second damp defect diagnosis result is that the cable intermediate joint does not have a damp defect.

4. The moisture defect diagnosis method for cable intermediate joint according to claim 1, characterized by, The method further comprises the following steps: acquiring a to-be-trained simulated one-dimensional time domain signal; performing defect positioning according to the to-be-trained simulated one-dimensional time domain signal by using three one-dimensional convolution layers connected in series to determine to-be-trained cable defect coordinates; performing prediction according to the to-be-trained cable defect coordinates by using an initial fully connected network to output a to-be-trained defect prediction value; determining a model gradient based on the to-be-trained defect prediction value, updating model parameters of the initial fully connected network by using the model gradient to determine an intermediate fully connected network, and statistically determining a model update number in real time; performing prediction according to the to-be-trained cable defect coordinates by using the intermediate fully connected network to output an intermediate defect prediction value, and calculating a target loss value according to the intermediate defect prediction value; determining whether the target loss value converges or the model update number reaches a preset update threshold; if the target loss value converges or the model update number reaches the preset update threshold, taking the intermediate fully connected network as the trained fully connected network.

5. The moisture defect diagnosis method for cable intermediate joint according to claim 2, characterized by, The preset coordinate function is specifically: ； Wherein, d is the cable defect coordinate; v represents the wave speed of electromagnetic wave in the cable; Δt is the time difference between the reflected wave and the incident wave.

6. A cable joint intermediate moisture defect diagnosis device for use in the cable joint intermediate moisture defect diagnosis method according to claim 1, characterized by The method comprises the steps of: acquiring a one-dimensional time domain signal of the cable intermediate joint; positioning the defect by using three one-dimensional convolution layers connected in series according to the one-dimensional time domain signal to determine the cable defect coordinate; diagnosing the defect by using a full connection network according to the cable defect coordinate to generate a damp defect diagnosis result.

7. A computer device, comprising: The computer program is executed to realize the damp defect diagnosis method of the cable intermediate joint according to any one of claims 1-5.

8. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed to realize the damp defect diagnosis method of the cable intermediate joint according to any one of claims 1-5.

9. A computer program product, characterised in that, The computer program product comprises a computer program stored on a non-transitory computer readable storage medium, and the computer program comprises program instructions, wherein when the program instructions are executed by a computer, the computer executes the damp defect diagnosis method of the cable intermediate joint according to any one of claims 1-5.

Images