Traveling wave transmission characteristic fault location prediction method for multi-type cable architecture

Abstract

The invention discloses a traveling wave transmission characteristic fault positioning prediction method for a multi-type cable architecture, and relates to the technical field of fault positioning, and the method comprises the steps: carrying out the real-time monitoring of a multi-type cable network, and obtaining a state data flow containing traveling wave characteristics; establishing a cable traveling wave transmission characteristic dynamic reconstruction model, performing time-frequency domain conjoint analysis on the state data flow, extracting a fault feature vector group, and constructing a fault mapping network according to the fault feature vector group; the fault mapping network comprises a fault type feature matrix and a position feature matrix; and carrying out cooperative operation on the fault type feature matrix and the position feature matrix, completing joint prediction of the fault type and the position, and outputting fault early warning information. According to the invention, accurate positioning and timely early warning of faults in a multi-type cable architecture are realized, the accuracy and reliability of fault prediction are improved, and effective technical support is provided for preventive maintenance of a cable system.

Description

Technical Field

[0001] The present invention relates to the technical field of fault location, and in particular to a method for locating and predicting traveling wave transmission characteristics faults of multiple types of cable architectures. Background Art

[0002] Cable fault location prediction technology is mainly based on traveling wave theory and fault feature analysis. Existing technologies usually use sensors arranged at fixed intervals to collect signals, and use traditional signal processing methods (such as Fourier transform, wavelet transform, etc.) to extract fault features. In terms of feature analysis, a single threshold judgment or simple statistical analysis method is often used to identify fault features, and some methods introduce machine learning algorithms to identify fault patterns. For the early warning system, it mainly relies on setting a fixed alarm threshold, and triggers an alarm when the monitoring parameter exceeds the threshold. At the same time, some systems have begun to try to take operating parameters such as temperature and load into consideration, but a complete dynamic response mechanism has not yet been formed. These technologies can achieve basic fault detection functions under a single working condition, but there are still major technical limitations in complex multi-type cable architectures.

[0003] At present, the following technical problems exist in the field of fault location prediction of multi-type cable architectures: there is a lack of basis for sensor layout, resulting in redundancy or blind spots in signal acquisition; the existing fault feature extraction methods do not respond sufficiently to the dynamic changes in the cable operating status and cannot accurately reflect the real-time changes in cable parameters with load and temperature; traditional feature extraction technologies often only focus on one aspect of feature information, resulting in incomplete feature extraction, and the feature fusion method is relatively simple and fails to fully utilize the correlation information between various features; in terms of fault warning, due to the relatively fragmented feature analysis, there is a lack of spatiotemporal correlation analysis of the fault development process, and the warning mechanism is too single, making it difficult to achieve accurate quantification and dynamic tracking of the degree of fault development. These problems seriously restrict the accuracy and timeliness of fault location prediction. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the present invention provides a method for locating and predicting faults based on traveling wave transmission characteristics of multiple types of cable architectures, which can solve the problems mentioned in the background technology.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: A method for locating and predicting faults based on traveling wave transmission characteristics of a multi-type cable architecture, comprising: real-time monitoring of a multi-type cable network to obtain a state data stream containing traveling wave characteristics;

[0007] Establish a dynamic reconstruction model of cable traveling wave transmission characteristics, perform time-frequency domain joint analysis on the state data stream, extract a fault feature vector group, and construct a fault mapping network based on the fault feature vector group; the fault mapping network includes a fault type feature matrix and a position feature matrix;

[0008] The fault type feature matrix and the position feature matrix are collaboratively calculated to complete the joint prediction of the fault type and the position, and output the fault warning information.

[0009] As a preferred solution of the method for locating and predicting faults of traveling wave transmission characteristics of multi-type cable architectures described in the present invention, real-time monitoring of multi-type cable networks is performed to obtain a state data stream containing traveling wave characteristics, including the following steps:

[0010] Traveling wave sensors are arranged at key nodes of multi-type cable networks to collect voltage and current signals in the cable networks;

[0011] The voltage and current signals are digitally sampled and converted to generate a state data stream containing traveling wave characteristics.

[0012] As a preferred solution of the method for locating and predicting the fault of the traveling wave transmission characteristics of the multi-type cable architecture described in the present invention, wherein: the key nodes include the cable joints, the branch connections and the terminal equipment connections;

[0013] The arrangement of the key nodes satisfies that the cable length between two adjacent traveling wave sensors is not greater than the product of the propagation speed of the traveling wave in the cable and half of the sampling period;

[0014] Determining the signal transmission coefficient between the key nodes according to the ratio of the transmission power between the key nodes to the system reference power and the impedance matching characteristics between the key nodes; wherein the impedance matching characteristics include the characteristic impedance and the equivalent impedance between the key nodes;

[0015] Determine the connection state coefficient between the key nodes according to the impedance difference between the key nodes, specifically, when the key nodes are directly connected and the absolute value of the impedance difference is not greater than the system characteristic impedance, the connection state coefficient is the ratio of the difference between the system characteristic impedance and the absolute value of the impedance difference to the system characteristic impedance; when the nodes are directly connected and the absolute value of the impedance difference is greater than the system characteristic impedance, the connection state coefficient is the ratio of the system characteristic impedance to the absolute value of the impedance difference; when the nodes are not directly connected, the connection state coefficient is zero;

[0016] Determine the coupling degree between the key nodes according to the signal transmission coefficient and the connection state coefficient, and classify the key nodes with a coupling degree greater than a first preset threshold into a node group according to the coupling degree between the key nodes; for the node group, calculate the sensing sensitivity of the node group based on the minimum coupling degree and the maximum coupling degree between the key nodes in the node group and the number of nodes in the node group;

[0017] The sensing sensitivity is used to characterize the detection capability of the node group to the fault signal.

[0018] As a preferred solution of the method for locating and predicting the traveling wave transmission characteristics fault of the multi-type cable architecture described in the present invention, a dynamic reconstruction model of the traveling wave transmission characteristics of the cable is established, the state data stream is jointly analyzed in the time and frequency domains, a fault feature vector group is extracted, and a fault mapping network is constructed according to the fault feature vector group, including the following steps:

[0019] Establishing a cable traveling wave transmission characteristic reconstruction model according to the state data stream, mapping the cable parameter changes into a transmission characteristic change matrix;

[0020] Using hyperbolic wavelet transform to perform time-frequency domain joint analysis on the state data stream to obtain time-frequency characteristic components;

[0021] Extracting a fault feature vector group according to the time-frequency feature components and constructing a fault mapping network;

[0022] A fault type feature matrix and a position feature matrix are established based on the fault feature vector group, and a fault mapping relationship is established through matrix operations.

[0023] As a preferred solution of the method for locating and predicting faults of traveling wave transmission characteristics of multi-type cable architectures described in the present invention, the cable traveling wave transmission characteristic reconstruction model is established by:

[0024] When the cable load change rate exceeds the first dynamic coefficient, the load influence factor is calculated and the corresponding element of the transmission characteristic change matrix is ​​updated;

[0025] When the temperature change exceeds the second dynamic coefficient, a temperature compensation factor is calculated and corresponding elements of the transmission characteristic change matrix are updated;

[0026] Performing a convolution operation on the updated transmission characteristic change matrix and the original transmission characteristic matrix to obtain a reconstruction model of the cable traveling wave transmission characteristic;

[0027] Acquiring the time-frequency characteristic component comprises the following steps:

[0028] The transmission characteristics of the model are reconstructed according to the cable traveling wave transmission characteristics, and the scale parameters of the hyperbolic wavelet transform are set;

[0029] Determine the number of decomposition layers according to the maximum point of signal energy distribution;

[0030] The modulus sequence and phase sequence of the wavelet coefficients of each layer are extracted to obtain the time-frequency characteristic components.

[0031] As a preferred solution of the method for locating and predicting the traveling wave transmission characteristics fault of the multi-type cable architecture described in the present invention, wherein: extracting the fault feature vector group according to the time-frequency feature component and constructing the fault mapping network comprises the following steps:

[0032] Performing Fourier transform on the modulus sequence and phase sequence in the time-frequency characteristic component to extract amplitude spectrum and phase spectrum features;

[0033] Extract characteristic frequencies and characteristic frequency bands based on multi-scale dynamic threshold criteria;

[0034] A hierarchical weighted combination strategy is used to construct a fault feature vector group;

[0035] The attention mechanism is used for feature fusion to obtain the fault mapping network, which is as follows:

[0036] Calculate feature attention weights through the interaction between query matrix, key matrix and value matrix;

[0037] Using the feature attention weights, weighted fusion is performed on the fault feature vector group, and the fused fault feature vector group is reorganized according to the fault type and the fault position, respectively, to construct a fault type feature matrix and a position feature matrix;

[0038] Performing singular value decomposition on the fault type feature matrix and the position feature matrix to extract main feature components;

[0039] Based on the main characteristic components, respectively calculating the mapping matrix between the fault type characteristic matrix and the fault type label, and the mapping matrix between the position characteristic matrix and the fault position label, and establishing a fault mapping relationship;

[0040] The fault mapping network is constructed based on a mapping matrix between the fault type feature matrix and the fault type label, and a mapping matrix between the position feature matrix and the fault position label.

[0041] As a preferred solution of the method for locating and predicting the traveling wave transmission characteristics of multi-type cable architectures described in the present invention, the fault type feature matrix and the position feature matrix are collaboratively calculated to complete the joint prediction of the fault type and position, and output the fault warning information, including the following steps:

[0042] Based on the fault mapping network, performing a tensor product operation on the fault type feature matrix and the position feature matrix to obtain a collaborative feature matrix;

[0043] Calculating the cosine similarity of each pair of features in the collaborative feature matrix to form a feature similarity matrix;

[0044] If the feature similarity is greater than a first preset similarity threshold, the corresponding feature combination is marked as a high-correlation feature pair;

[0045] If the feature similarity is less than the first preset similarity threshold but greater than the second preset similarity threshold, it is marked as a medium-correlated feature pair;

[0046] If the feature similarity is less than a second preset similarity threshold, it is marked as a low-correlation feature pair; the high-correlation feature pair is associated with the sensor sensitivity of the node group to establish a feature spatiotemporal mapping;

[0047] Tracking the feature similarity change trend of the highly correlated feature pair based on the feature spatiotemporal mapping, if the feature similarity shows a continuous upward trend and the change rate exceeds the normal fluctuation range, determining the warning level in combination with the time series information of the fault feature vector group;

[0048] Perform location analysis on the position with the highest feature similarity. Specifically, match the position with the highest feature similarity with the sensor sensitivity of the corresponding node group. If the sensor sensitivity is greater than the average sensitivity of the node group, it is judged as the core position of the fault. Otherwise, find the position with the highest sensor sensitivity in the adjacent node group as the possible fault position.

[0049] Integrate the feature spatiotemporal mapping results, output warning information including fault time, location and development trend, and update the verified feature pattern to the fault mapping network.

[0050] To further solve the above technical problems, the present invention provides the following technical solutions: A traveling wave transmission characteristic fault location prediction system for a multi-type cable architecture, comprising: a data acquisition module for real-time monitoring of a multi-type cable network and obtaining a state data stream containing traveling wave characteristics;

[0051] Constructing an analysis module, which is used to establish a dynamic reconstruction model of cable traveling wave transmission characteristics, perform a joint analysis of the state data stream in the time and frequency domains, extract a fault feature vector group, and construct a fault mapping network according to the fault feature vector group;

[0052] The prediction and warning module is used to perform collaborative operations on the fault type feature matrix and the position feature matrix to complete the joint prediction of the fault type and position and output fault warning information.

[0053] A computer device includes a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the steps of the method for locating and predicting faults of traveling wave transmission characteristics of multi-type cable architectures as described above are implemented.

[0054] 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 locating and predicting faults of traveling wave transmission characteristics of multi-type cable architectures as described above are implemented.

[0055] Beneficial effects of the present invention: The present invention introduces the concepts of coupling degree and sensor sensitivity to establish a node group division mechanism, thereby overcoming the problems of blind sensor arrangement and redundant signal acquisition in traditional methods; the present invention establishes a traveling wave transmission characteristic reconstruction model based on the dynamic changes of cable parameters, adopts a hyperbolic wavelet transform and a dynamically weighted frequency band feature extraction method, and combines the feature fusion strategy of the attention mechanism to effectively solve the problems of insufficient response to the dynamic changes of the cable operation status, incomplete feature extraction, and suboptimal feature fusion in the prior art; the present invention organically combines the time evolution characteristics of fault development with the spatial distribution characteristics by establishing a feature space-time mapping, adopts a dual-threshold feature similarity evaluation mechanism to achieve refined classification of the degree of fault correlation, and realizes dynamic quantification of the degree of fault development based on a warning classification method based on multi-dimensional evaluation, thereby overcoming the technical bottlenecks of feature analysis fragmentation, lack of correlation, and single warning mechanism in traditional methods. The present invention realizes the precise positioning and timely warning of faults in multi-type cable architectures, improves the accuracy and reliability of fault prediction, and provides effective technical support for the preventive maintenance of cable systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0057] Figure 1 This is a schematic diagram of the overall process of a method for locating and predicting traveling wave transmission characteristics faults of multiple types of cable architectures proposed by the present invention;

[0058] Figure 2 A diagram of computer equipment in a method for locating and predicting faults based on traveling wave transmission characteristics of a multi-type cable architecture proposed by the present invention. DETAILED DESCRIPTION

[0059] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0060] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0061] Example 1, reference Figure 1 , as an embodiment of the present invention, provides a method for locating and predicting faults based on traveling wave transmission characteristics of multi-type cable architectures.

[0062] In related technologies, there is a lack of basis for sensor layout, resulting in redundancy or blind spots in signal acquisition; existing fault feature extraction methods do not respond sufficiently to the dynamic changes in the cable operating status and cannot accurately reflect the real-time changes in cable parameters with load and temperature; traditional feature extraction techniques often only focus on one aspect of feature information, resulting in incomplete feature extraction, and the feature fusion method is relatively simple and fails to fully utilize the correlation information between various features; in terms of fault warning, due to the relatively fragmented feature analysis, there is a lack of spatiotemporal correlation analysis of the fault development process, and the warning mechanism is too single, making it difficult to achieve accurate quantification and dynamic tracking of the degree of fault development. These problems seriously restrict the accuracy and timeliness of fault location prediction.

[0063] The present application provides an effective solution to the above-mentioned problems. Next, a plurality of embodiments will be combined to elaborate on how to implement the traveling wave transmission characteristic fault location prediction method of the multi-type cable architecture.

[0064] Figure 1 The overall process diagram of a traveling wave transmission characteristic fault location prediction method for multiple types of cable architectures is shown, including the following steps:

[0065] S1: Distributed traveling wave sensors are used to monitor multiple types of cable networks in real time to obtain status data streams containing traveling wave characteristics.

[0066] S1.1: Traveling wave sensors are arranged at key nodes of multi-type cable networks to collect voltage and current signals in the cable networks.

[0067] Specifically, the key nodes include cable joints, branch connections, and terminal equipment connections. The arrangement of key nodes satisfies that the cable length between two adjacent traveling wave sensors is not greater than the product of the propagation speed of the traveling wave in the cable and half of the sampling period. The formula is expressed as:

[0068]

[0069] Among them, L i,j is the cable length between two adjacent traveling wave sensors, v is the propagation speed of the traveling wave in the cable, T s is the sampling period.

[0070] The signal transmission coefficient between key nodes is determined based on the ratio of the transmission power between key nodes to the system reference power and the impedance matching characteristics between key nodes. The impedance matching characteristics include the characteristic impedance and the equivalent impedance between key nodes. The calculation formula of the signal transmission coefficient is:

[0071]

[0072] Among them, W ij is the signal transmission coefficient between key node i and key node j, Z c is the characteristic impedance, Z i and Z j are the equivalent impedances at key nodes i and j, respectively, ij is the transmission power between key nodes, and P0 is the system reference power.

[0073] The connection state coefficient between the key nodes is determined according to the impedance difference between the key nodes, specifically:

[0074] When the key nodes are directly connected and the absolute value of the impedance difference is not greater than the system characteristic impedance, the connection state coefficient is the ratio of the difference between the system characteristic impedance and the absolute value of the impedance difference to the system characteristic impedance;

[0075] When the nodes are directly connected and the absolute value of the impedance difference is greater than the system characteristic impedance, the connection state coefficient is the ratio of the system characteristic impedance to the absolute value of the impedance difference;

[0076] When the nodes are not directly connected, the connection state coefficient is zero.

[0077] The calculation formula of the connection state coefficient is:

[0078]

[0079] Among them, S ij is the connection state coefficient between key node i and key node j, ΔZ ijis the impedance difference between key node i and key node j, and Z0 is the system characteristic impedance.

[0080] The coupling degree between the key nodes is determined according to the signal transmission coefficient and the connection state coefficient, and the key nodes whose coupling degree is greater than a first preset threshold are classified into a node group.

[0081] The calculation formula of coupling degree is:

[0082] T ij =W ij ·S ij ;

[0083] The sensing sensitivity of the node group is calculated based on the minimum coupling degree, maximum coupling degree and number of nodes in the node group between the key nodes in the node group, and the sensing sensitivity is used to characterize the detection ability of the node group for fault signals. The calculation formula of sensing sensitivity is:

[0084]

[0085] Among them, G is the node group set, and n is the number of key nodes in the node group.

[0086] It should be noted that the first preset threshold is a coupling degree threshold value set based on the signal transmission characteristics between key nodes in the cable network. In the present invention, the first preset threshold is set by comprehensively considering the topological structure characteristics, signal attenuation characteristics and fault location accuracy requirements of the cable network. In practical applications, when the coupling degree between two key nodes exceeds the first preset threshold, it indicates that there is a strong signal transmission correlation between the two key nodes, and they can be classified into the same node group for collaborative fault detection. The method for determining the first preset threshold is: first, through statistical analysis of historical operation data, the coupling degree distribution characteristics between key nodes under normal working conditions are obtained; then, in combination with electromagnetic transient simulation, a coupling degree change model under different types of faults is established; finally, based on the reliability requirements of fault detection, a suitable threshold is selected to balance the detection sensitivity and false alarm rate. According to a large amount of engineering practice experience, when the cable network is stable and the noise interference is small, the first preset threshold is preferably 0.75-0.85; when the external environment interference is large, it can be appropriately increased to 0.85-0.95 to improve the reliability of fault detection. In addition, the first preset threshold can also be dynamically adjusted according to the actual operation situation to adapt to changes in network status.

[0087] S1.2: Digitally sample and convert the voltage and current signals to generate a state data stream containing traveling wave characteristics.

[0088] The state data flow can be expressed as follows:

[0089] X d (t) = {xd,1 (t),x d,2 (t),...,x d,n (t)};

[0090] Among them, X d (t) is the state data stream after digital sampling conversion at time t, which contains the sampling data of all key nodes; x d,i (t) represents the sampled data vector at the i-th key node, and n is the total number of key nodes. The sampled data vector at each key node can be expressed as:

[0091]

[0092] Among them, v d,i (t) represents the voltage sampling value at the i-th key node, i d,i (t) represents the current sampling value at the i-th key node. For the voltage sampling value at each node, its expression is:

[0093] v d,i (t) = v i (kΔt),k=1,2,...,N s ;

[0094] v i (kΔt) represents the voltage digital value at the i-th key node at the k-th sampling moment; Δt is the sampling interval; N s is the number of sampling points. Similarly, for the current sampling value at each node, its expression is:

[0095] i d,i (t) = i i (kΔt),k=1,2,...,N s ;

[0096] In the formula, i i (kΔt) represents the digital value of the current at the i-th key node at the k-th sampling moment; the sampling interval Δt and the number of sampling points N s Keep consistent with voltage sampling to ensure the synchronization of voltage and current sampling. After the above digital sampling conversion, the state data stream contains complete traveling wave feature information, laying the foundation for subsequent time-frequency domain analysis and fault feature extraction.

[0097] Preferably, the present invention first proposes a sensor layout strategy based on key nodes in the fault location prediction method of traveling wave transmission characteristics of multi-type cable architectures. Different from the traditional fixed-distance layout scheme, the present invention closely combines the sensor layout with the physical characteristics of the cable network, and optimizes the layout at the cable joints, branch connections and terminal equipment connections. By associating the cable length between adjacent sensors with the traveling wave propagation characteristics, the effective capture of the fault traveling wave signal is ensured. This layout overcomes the problem of missing fault signals in traditional schemes and significantly improves the accuracy of fault location. At the same time, through the synchronous sampling and digital conversion of voltage and current signals, a complete state data stream acquisition system is established, laying a reliable data foundation for subsequent analysis.

[0098] The present invention introduces a node grouping mechanism based on coupling degree. By calculating the signal transmission coefficient and connection state coefficient between key nodes, an evaluation system for the coupling degree between nodes is established. Nodes with coupling degrees exceeding a preset threshold are grouped into the same group, and the detection capability of the group for fault signals is evaluated by sensor sensitivity. The present invention fully considers the power transmission characteristics and impedance matching characteristics between nodes, making data collection more targeted, and while improving the accuracy of fault location, it also enhances the anti-interference ability of the system. Compared with the simple sensor layout and data collection scheme in the prior art, the present invention achieves more efficient and reliable fault signal collection and preprocessing through the optimized layout of key nodes and the node grouping mechanism based on coupling degree.

[0099] S2: Establish a dynamic reconstruction model of cable traveling wave transmission characteristics, perform a joint analysis of the state data stream in the time and frequency domains, extract a fault feature vector group, and construct a fault mapping network with time-varying characteristics based on the fault feature vector group.

[0100] Specifically, the fault mapping network includes a fault type feature matrix and a location feature matrix.

[0101] S2.1: A cable traveling wave transmission characteristic reconstruction model is established based on the state data stream, and the cable parameter changes are mapped into a transmission characteristic change matrix.

[0102] Specifically, the cable traveling wave transmission characteristic reconstruction model is established in the following way:

[0103] When the cable load change rate exceeds the first dynamic coefficient, the load influence factor is calculated and the corresponding element of the transmission characteristic change matrix is ​​updated;

[0104] When the temperature variation exceeds the second dynamic coefficient, a temperature compensation factor is calculated and corresponding elements of the transmission characteristic variation matrix are updated;

[0105] The updated transmission characteristic change matrix is ​​convolved with the original transmission characteristic matrix to obtain the cable traveling wave transmission characteristic reconstruction model.

[0106] It should be noted that the cable load change rate refers to the percentage change of the cable load current per unit time. In the power cable system, the current value of each phase of the cable is obtained through the online monitoring device, and the ratio of the current change in two adjacent sampling cycles to the initial current is calculated. This parameter reflects the sudden change of the cable load and is closely related to the degree of distortion of the fault waveform. The load influence factor is a correction coefficient that describes the degree of influence of the load change on the traveling wave transmission characteristics. It is calculated by combining the impedance-temperature characteristic curve of the cable (the impedance-temperature characteristic curve is a functional relationship curve that describes the law of change of the characteristic impedance of the power cable under different temperature conditions, reflecting the impedance change caused by the increase of the cable conductor resistance and the change of the tangent value of the insulation medium loss angle when the temperature rises). The load influence factor is used to correct the load-related elements in the transmission characteristic change matrix to ensure the accuracy of the reconstruction of the traveling wave transmission characteristics. The first dynamic coefficient refers to the reference value for determining whether the cable load change is significant, and its value is obtained based on historical operation data statistics. When the cable load change rate exceeds the first dynamic coefficient, it indicates that the load change may cause a significant change in the traveling wave transmission characteristics, and the transmission characteristic change matrix needs to be updated. The first dynamic coefficient is related to parameters such as cable type and rated capacity.

[0107] The second dynamic coefficient refers to the reference value for determining whether the temperature change is significant, and is determined based on the temperature characteristics of the cable insulation material. When the temperature change exceeds the second dynamic coefficient, it indicates that the temperature change will cause a significant change in the cable dielectric parameters, and the transmission characteristic change matrix needs to be updated. The second dynamic coefficient is closely related to the type of cable insulation material. The temperature compensation factor is a correction factor that describes the degree of influence of temperature change on the cable dielectric parameters. It is calculated based on the temperature data obtained by the distributed optical fiber temperature measurement system and the temperature-dielectric constant characteristic curve of the cable medium (a functional relationship curve that describes the change law of the dielectric constant of the power cable insulation material under different temperature conditions, which reflects the change in the dielectric constant caused by the change in the molecular polarization characteristics of the insulation material due to temperature change). The temperature compensation factor is used to correct the temperature-related elements in the transmission characteristic change matrix and improve the accuracy of the reconstruction of the traveling wave transmission characteristics.

[0108] S2.2: Use hyperbolic wavelet transform to perform time-frequency domain joint analysis on the state data stream to obtain time-frequency feature components.

[0109] Specifically, the transmission characteristics of the cable traveling wave transmission characteristic reconstruction model are used to set the scale parameters of the hyperbolic wavelet transform; the number of decomposition layers is determined according to the maximum point of the signal energy distribution; the modulus sequence and phase sequence of the wavelet coefficients of each layer are extracted to obtain the time-frequency characteristic components.

[0110] In the present invention, the hyperbolic wavelet transform is selected instead of the traditional wavelet transform or short-time Fourier transform because the hyperbolic wavelet transform has a frequency-dependent time resolution characteristic and is more suitable for analyzing the traveling wave propagation characteristics of cable faults. Its scale parameter is set according to the transmission characteristics output by the cable traveling wave transmission characteristic reconstruction model, which includes parameter information such as impedance and dielectric constant of the cable in the current operating state. The corresponding relationship between the scale parameter and the transmission characteristic is: Where v is the speed of traveling wave propagation, and f is the main frequency of the signal. This correspondence is derived from the theory of electromagnetic wave propagation in cables.

[0111] The specific implementation process is as follows: first, input the state data stream into the hyperbolic wavelet transform module, and determine the initial scale parameters based on the transmission characteristics; then, the energy entropy minimization criterion is used to calculate the energy distribution of the signal at different decomposition scales, and the local maximum point of the energy distribution is found through an iterative optimization algorithm. The scale corresponding to this point is used as the optimal decomposition layer number. The optimal decomposition layer number determines the degree of fineness of the signal decomposition. Generally, the optimal decomposition layer number N satisfies: E(N)>E(N-1) and E(N)>E(N+1), where E(N) is the energy value of the Nth layer. Then, the wavelet coefficients are extracted from each layer of the decomposition results, and the Hilbert transform is used to calculate its modulus sequence (characterizing the amplitude characteristics of the signal) and phase sequence (characterizing the phase characteristics of the signal). Modulus sequence M k (t) Calculated by the instantaneous amplitude of the wavelet coefficient: M k (t)=|W f (a,t)|, where W f (a, t) is the wavelet transform coefficient; the phase sequence φ k (t) is obtained by calculating the argument of the wavelet coefficient: k (t) = arg[W f (a, t)]. These sequences together constitute the time-frequency feature components, providing basic data for subsequent fault feature extraction. k represents the kth time analysis window.

[0112] Preferably, this method of setting parameters based on transmission characteristics ensures that the time-frequency analysis results can accurately reflect the characteristic information of cable faults by establishing a mapping relationship between scale parameters and physical characteristics. At the same time, the adaptive determination of the optimal number of decomposition layers and the mathematical description of sequence extraction make the feature extraction process repeatable and reliable.

[0113] S2.3: Extracting a fault feature vector group according to the time-frequency feature components and constructing a fault mapping network.

[0114] Specifically, the modulus sequence and phase sequence in the time-frequency feature component are Fourier transformed to extract the amplitude spectrum and phase spectrum features; the characteristic frequency and characteristic frequency band are extracted based on the multi-scale dynamic threshold criterion; and the hierarchical weighted combination strategy is adopted to construct the fault feature vector group.

[0115] In the present invention, a dynamic weighted frequency band feature extraction method is introduced for the time-frequency feature component. First, the modulus sequence M obtained in S2.2 is k (t) and the phase sequence φ k (t) Perform Fourier transform to obtain the corresponding amplitude spectrum A k (f) and phase spectrum P k (f).

[0116] The characteristic frequency band is extracted using a dynamic decision criterion based on the signal-to-noise ratio:

[0117]

[0118] Among them, R k (f) is the frequency band characteristic response function, ΔP k (f) is the local difference of the phase spectrum, is the average amplitude in the local frequency band, β is the frequency band attenuation coefficient, and Δf is the bandwidth of the frequency band to be evaluated. k (f) When the dynamic threshold is exceeded, the corresponding frequency band is marked as a characteristic frequency band. The dynamic threshold is determined by statistically analyzing the frequency band characteristic response function in each analysis window. Specifically, it is the mean of the frequency band characteristic response function in the window plus α times the standard deviation, where α is an adjustable coefficient (usually 2 to 3). This adaptive threshold setting method can automatically adjust according to changes in signal characteristics, increase the threshold when the noise is large to avoid misjudgment, and reduce the threshold when the signal is weak to ensure detection sensitivity, thereby improving the accuracy and robustness of characteristic frequency band identification.

[0119] It should be noted that in fault diagnosis, the frequency band characteristic response function R k The design of (f) specifically solves the limitations of traditional methods. Traditional methods often only focus on the sudden change of amplitude spectrum or phase spectrum, which makes it difficult to fully capture the fault characteristics. k (f)|Characterizes the local variation characteristics of the phase spectrum and reflects the phase mutation caused by the fault; through the denominator Amplitude information is introduced, but the square root form is used to avoid excessive dominance of amplitude in feature extraction; at the same time, the exp(-β|Δf|) term imposes a penalty on wide-band features and gives priority to narrow-band mutation features. This design is more in line with the physical characteristics of mechanical faults. This multi-dimensional response function design can more accurately identify and extract fault feature frequency bands.

[0120] Furthermore, a hierarchical weighted combination strategy is used to construct a fault feature vector group. First, the feature frequency weight w is defined f (i)

[0121]

[0122] Among them, A i is the amplitude of the i-th characteristic frequency point, is its local variance, γ1 and γ2 are weight adjustment parameters.

[0123] It should be noted that the characteristic frequency weight w in the present invention is f The calculation of (i) reflects the evaluation strategy of the importance of frequency features in fault diagnosis. The characteristic frequency weight w f (i) Evaluate from two dimensions: On the one hand, the amplitude of the characteristic frequency point A is examined. i , a larger amplitude often means a more significant fault feature; on the other hand, the local variance Smaller variance indicates better stability of the feature. The dual Softmax structure is adopted, and the influence of the two dimensions is adjusted respectively through the two weight adjustment parameters γ1 and γ2, which not only ensures the prominent features but also ensures the reliability of the features, thus effectively improving the recognition accuracy of fault features.

[0124] Similarly, define the feature band weight w b (i)

[0125]

[0126] Among them, R i is the characteristic response value of the i-th frequency band, B i is its standardized bandwidth, and η1 and η2 are the frequency band weight adjustment parameters.

[0127] It should be noted that the characteristic band weight w b The calculation of (i) designs a similar dual evaluation mechanism for the characteristic frequency band. i reflects the characteristic response intensity of the frequency band, while B i The standardized bandwidth characteristics of the frequency band are characterized. By adjusting η1 and η2, an accurate assessment of the importance of the characteristic frequency band is achieved. Since faults often show significant characteristics within a specific frequency band, the design adopted by the present invention is suitable for extracting frequency band features in mechanical fault diagnosis, and the calculation method of the characteristic frequency band weight can effectively capture such characteristics.

[0128] Finally, the fault feature vector group V is combined and expressed as:

[0129] V=[α1F T W f,α2B T W b ];

[0130] Among them, F is the frequency feature matrix, B is the frequency band feature matrix, and W f and W b are the corresponding weight matrices respectively, and α1 and α2 are the feature type balance coefficients.

[0131] It should be noted that the calculation formula of the fault feature vector group V realizes the unified expression of frequency characteristics and frequency band characteristics. T W f and B T W b The weighted frequency features and frequency band features are obtained respectively, and then the relative importance of the two types of features is adjusted by the balance coefficients α1 and α2. This is not only concise in mathematical form, but also efficient in computational implementation, and can make full use of the complementarity of the two types of features to provide a more comprehensive fault feature expression.

[0132] The attention mechanism is used for feature fusion to obtain the fault mapping network, which includes the following steps:

[0133] First, define the feature attention weights:

[0134]

[0135] Among them, λ ε is the attention weight output of the kth feature, Q ε is the query matrix of the ε-th layer feature, K is the key matrix, V' is the value matrix, and L is the number of feature layers.

[0136] It should be noted that the attention weight mechanism is mainly used for adaptive fusion of features in the present invention. ε The interaction of the key matrix K and the value matrix V' can automatically learn the association between features at different levels. This mechanism is particularly suitable for dealing with complex feature dependencies in fault diagnosis, because different types of faults may show different feature combination patterns at different feature levels, and the attention mechanism can adaptively capture this feature combination relationship.

[0137] Preferably, since traditional methods mostly use fixed window Fourier transform for spectrum analysis, spectrum leakage is prone to occur when processing transient fault signals. The present invention can more accurately capture the characteristic information of the fault moment by analyzing the time-frequency characteristic components of the modulus sequence and the phase sequence. Existing characteristic frequency band extraction methods are usually based on fixed thresholds or simple amplitude comparisons, which are difficult to cope with complex noise environments. The dynamic judgment criterion based on signal-to-noise ratio (R k (f)), by comprehensively considering the phase mutation (ΔPk (f)), local amplitude and bandwidth (exp(-β|Δf|)), thereby improving the accuracy of feature band identification. Existing methods often directly combine various features when constructing feature vectors, without fully considering the differences in feature importance. The hierarchical weighted combination strategy designed by the present invention processes frequency features and band features respectively by defining feature frequency weights and feature band weights, thereby achieving optimal feature selection. In addition, traditional neural networks usually use simple full connections or pooling operations in the feature fusion layer, which can easily lose important feature information. The present invention uses an attention mechanism to allocate feature attention weights according to the interaction between the query matrix, the key matrix, and the value matrix, thereby enhancing the fault mapping network's ability to recognize key features.

[0138] Secondly, the fault feature vector group is weightedly fused using the calculated feature attention weights, and the fused fault feature vector group is reorganized according to the fault type and fault location to construct the fault type feature matrix and position feature matrix.

[0139] Then, singular value decomposition is performed on the fault type feature matrix and the location feature matrix to extract the main feature components.

[0140] Finally, based on the main feature components, the mapping matrix between the fault type feature matrix and the fault type label, as well as the mapping matrix between the position feature matrix and the fault position label are calculated to establish the fault mapping relationship. These mapping matrices realize the correspondence between the fault features and the fault type and location, forming a complete fault mapping network.

[0141] Among them, the number of rows of the fault type feature matrix is ​​equal to the dimension of the feature vector, and the number of columns is equal to the number of fault types; the number of rows of the position feature matrix is ​​equal to the dimension of the feature vector, and the number of columns is equal to the number of fault locations. When performing singular value decomposition, a series of singular values ​​can be obtained. These singular values ​​are arranged from large to small to reflect the importance of the corresponding characteristic components. The main characteristic components refer to the characteristic components corresponding to the larger singular values, which contain the most important information or patterns in the data. By retaining only these main characteristic components, most of the effective information can be retained while reducing the data dimension, and the secondary or noise information can be removed, thereby achieving data dimensionality reduction and feature extraction.

[0142] S3: Perform collaborative calculations on the fault type feature matrix and the position feature matrix to complete joint prediction of the fault type and position, and output fault warning information.

[0143] S3.1: Based on the fault mapping network obtained in S2, the fault type feature matrix and the position feature matrix are tensor-producted to obtain a collaborative feature matrix.

[0144] Specifically, the cosine similarity of each pair of features in the collaborative feature matrix is ​​calculated to form a feature similarity matrix. If the feature similarity is greater than the first preset similarity threshold, the corresponding feature combination is marked as a high-correlation feature pair; if the feature similarity is less than the first preset similarity threshold but greater than the second preset similarity threshold, it is marked as a medium-correlation feature pair; if the feature similarity is less than the second preset similarity threshold, it is marked as a low-correlation feature pair. The high-correlation feature pair is associated with the node group sensor sensitivity in S1 to establish a feature spatiotemporal mapping.

[0145] It should be noted that in order to achieve accurate division of fault correlations of different degrees, the present invention sets two preset similarity thresholds. The first preset similarity threshold is preferably 0.85-0.95, which is used to identify highly correlated feature combinations, which usually represent different manifestations of the same fault mode. The second preset similarity threshold is preferably 0.6-0.7, which is used to screen feature combinations with potential correlations. The setting of these two thresholds is based on statistical analysis of a large amount of historical fault data and is determined through iterative optimization, which can effectively balance the sensitivity and reliability of fault detection. Feature spatiotemporal mapping is a data structure that combines the time evolution characteristics of highly correlated feature pairs with the spatial distribution characteristics. Feature spatiotemporal mapping forms a spatiotemporal evolution model of fault development by establishing a corresponding relationship between feature similarity, node position and sensor sensitivity. In practical applications, feature spatiotemporal mapping can be expressed as a three-dimensional tensor, in which the time dimension reflects the dynamic changes of feature correlation, the space dimension describes the physical distribution of fault features, and the feature dimension characterizes the degree of correlation between different types of faults.

[0146] S3.2: Track the similarity change trend of highly correlated feature pairs based on feature spatiotemporal mapping.

[0147] Specifically, if the feature similarity shows a continuous upward trend and the rate of change exceeds the normal fluctuation range, the warning level is determined in combination with the time series information of the fault feature vector group in S2. At the same time, the location analysis is performed on the position with the highest feature similarity: the position is matched with the sensor sensitivity of the corresponding node group. If the sensor sensitivity of the position is greater than the average sensitivity of the node group, it is determined as the core position of the fault; otherwise, the position with the highest sensor sensitivity in the adjacent node group is found as the possible fault position.

[0148] It should be noted that the normal fluctuation range refers to the natural variation range of feature similarity under normal operation of the equipment. This range is obtained through statistical analysis of long-term operation data of healthy equipment, and its upper and lower limits are generally determined by the 3σ principle. Specifically, after calculating the standard deviation σ of feature similarity, the mean ±3σ is used as the fluctuation range boundary. Changes beyond this range are considered to be early signs of potential failures.

[0149] The process of determining the warning level by combining the time series information of the fault feature vector group is a multi-dimensional evaluation mechanism. First, the rate of change of feature similarity is calculated and compared with the pre-defined rate threshold (such as 0.1 / hour, 0.3 / hour, 0.5 / hour); secondly, the amplitude change trend and spectral characteristics of the fault feature vector in S2 are analyzed; finally, considering the severity of these indicators, the warning level is divided into four levels: observation level (feature similarity rises slowly, no obvious abnormality), attention level (feature similarity rises at a medium speed, with slight abnormality), warning level (feature similarity rises rapidly, accompanied by obvious abnormality) and emergency level (feature similarity rises sharply, and multiple indicators are seriously abnormal). This grading method can achieve accurate quantification of the degree of fault development and provide a reliable basis for equipment maintenance decisions.

[0150] S3.3: Integrate the feature spatiotemporal mapping results, output warning information including fault time, location and development trend, and update the verified feature pattern to the fault mapping network.

[0151] It should be noted that the present invention realizes the joint prediction of fault type and location through the collaborative feature analysis and early warning mechanism in S3. The present invention deeply integrates the fault type feature and the location feature through tensor product operation, avoiding the problem of feature analysis fragmentation and lack of correlation in traditional methods; adopting a dual-threshold feature similarity evaluation mechanism, it can finely classify the degree of fault correlation and improve the accuracy of fault identification; establishing a feature spatiotemporal mapping to organically combine the time evolution characteristics of fault development with the spatial distribution characteristics, and enhance the reliability of fault location; based on the early warning classification method of feature similarity change trend and multi-dimensional evaluation mechanism, it realizes the dynamic quantification of the degree of fault development, and can timely discover potential faults and divide risk levels. This technical solution not only improves the accuracy and timeliness of fault prediction, but also provides a more comprehensive and reliable basis for equipment maintenance decisions, effectively reduces the risk of unplanned equipment downtime, and improves the safety and reliability of system operation.

[0152] In summary, the present invention introduces the concepts of coupling degree and sensor sensitivity to establish a node group division mechanism, thus overcoming the problems of blind sensor arrangement and redundant signal acquisition in traditional methods; the present invention establishes a traveling wave transmission characteristic reconstruction model based on the dynamic changes of cable parameters, adopts a hyperbolic wavelet transform and a dynamically weighted frequency band feature extraction method, and combines the feature fusion strategy of the attention mechanism to effectively solve the problems of insufficient response to the dynamic changes of cable operation status, incomplete feature extraction, and suboptimal feature fusion in the prior art; the present invention organically combines the temporal evolution characteristics of fault development with the spatial distribution characteristics by establishing a feature spatiotemporal mapping, adopts a dual-threshold feature similarity evaluation mechanism to achieve refined classification of the degree of fault correlation, and realizes dynamic quantification of the degree of fault development based on a warning classification method based on multi-dimensional evaluation, thus overcoming the technical bottlenecks of feature analysis fragmentation, lack of correlation, and single warning mechanism in traditional methods. The present invention realizes the precise positioning and timely warning of faults in multi-type cable architectures, improves the accuracy and reliability of fault prediction, and provides effective technical support for the preventive maintenance of cable systems.

[0153] Embodiment 2 is an embodiment of the present invention, which provides a traveling wave transmission characteristic fault location prediction system for multiple types of cable architectures, including:

[0154] Data acquisition module, used to monitor multiple types of cable networks in real time and obtain status data streams containing traveling wave characteristics;

[0155] Construct an analysis module to establish a dynamic reconstruction model of cable traveling wave transmission characteristics, perform a joint analysis of the state data stream in the time and frequency domains, extract a fault feature vector group, and construct a fault mapping network based on the fault feature vector group;

[0156] The prediction and warning module is used to coordinate the fault type feature matrix and the location feature matrix to complete the joint prediction of fault type and location and output fault warning information.

[0157] Example 3, reference Figure 2, is an embodiment of the present invention, which is different from the previous embodiment in that: if the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0158] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0159] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0160] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0161] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for locating and predicting faults based on traveling wave transmission characteristics of multi-type cable architectures, characterized in that: include: Real-time monitoring of multi-type cable networks to obtain status data streams containing traveling wave characteristics; Establish a dynamic reconstruction model of cable traveling wave transmission characteristics, perform time-frequency domain joint analysis on the state data stream, extract a fault feature vector group, and construct a fault mapping network based on the fault feature vector group; the fault mapping network includes a fault type feature matrix and a position feature matrix; The fault type feature matrix and the position feature matrix are collaboratively calculated to complete the joint prediction of the fault type and the position, and output the fault warning information.

2. The method for locating and predicting traveling wave transmission characteristics faults of a multi-type cable architecture according to claim 1, characterized in that: Real-time monitoring of multi-type cable networks to obtain status data streams containing traveling wave characteristics includes the following steps: Traveling wave sensors are arranged at key nodes of multi-type cable networks to collect voltage and current signals in the cable networks; The voltage and current signals are digitally sampled and converted to generate a state data stream containing traveling wave characteristics.

3. The method for locating and predicting traveling wave transmission characteristics faults of a multi-type cable architecture according to claim 2, characterized in that: The key nodes include cable joints, branch connections and terminal equipment connections; The arrangement of the key nodes satisfies that the cable length between two adjacent traveling wave sensors is not greater than the product of the propagation speed of the traveling wave in the cable and half of the sampling period; Determining the signal transmission coefficient between the key nodes according to the ratio of the transmission power between the key nodes to the system reference power and the impedance matching characteristics between the key nodes; wherein the impedance matching characteristics include the characteristic impedance and the equivalent impedance between the key nodes; Determine the connection state coefficient between the key nodes according to the impedance difference between the key nodes, specifically, when the key nodes are directly connected and the absolute value of the impedance difference is not greater than the system characteristic impedance, the connection state coefficient is the ratio of the difference between the system characteristic impedance and the absolute value of the impedance difference to the system characteristic impedance; when the nodes are directly connected and the absolute value of the impedance difference is greater than the system characteristic impedance, the connection state coefficient is the ratio of the system characteristic impedance to the absolute value of the impedance difference; when the nodes are not directly connected, the connection state coefficient is zero; Determine the coupling degree between the key nodes according to the signal transmission coefficient and the connection state coefficient, and classify the key nodes with a coupling degree greater than a first preset threshold into a node group according to the coupling degree between the key nodes; for the node group, calculate the sensing sensitivity of the node group based on the minimum coupling degree and the maximum coupling degree between the key nodes in the node group and the number of nodes in the node group; The sensing sensitivity is used to characterize the detection capability of the node group to the fault signal.

4. The method for locating and predicting traveling wave transmission characteristics faults of multiple types of cable structures according to claim 3, characterized in that: Establishing a dynamic reconstruction model of cable traveling wave transmission characteristics, performing a joint analysis of the state data stream in the time and frequency domains, extracting a fault feature vector group, and constructing a fault mapping network according to the fault feature vector group, including the following steps: Establishing a cable traveling wave transmission characteristic reconstruction model according to the state data stream, mapping the cable parameter changes into a transmission characteristic change matrix; Using hyperbolic wavelet transform to perform time-frequency domain joint analysis on the state data stream to obtain time-frequency characteristic components; Extracting a fault feature vector group according to the time-frequency feature components and constructing a fault mapping network; A fault type feature matrix and a position feature matrix are established based on the fault feature vector group, and a fault mapping relationship is established through matrix operations.

5. The method for locating and predicting traveling wave transmission characteristics faults of multi-type cable structures according to claim 4, characterized in that: The cable traveling wave transmission characteristic reconstruction model is established in the following way: When the cable load change rate exceeds the first dynamic coefficient, the load influence factor is calculated and the corresponding element of the transmission characteristic change matrix is ​​updated; When the temperature change exceeds the second dynamic coefficient, a temperature compensation factor is calculated and corresponding elements of the transmission characteristic change matrix are updated; Performing a convolution operation on the updated transmission characteristic change matrix and the original transmission characteristic matrix to obtain a reconstruction model of the cable traveling wave transmission characteristic; Acquiring the time-frequency characteristic component comprises the following steps: The transmission characteristics of the model are reconstructed according to the cable traveling wave transmission characteristics, and the scale parameters of the hyperbolic wavelet transform are set; Determine the number of decomposition layers according to the maximum point of signal energy distribution; The modulus sequence and phase sequence of the wavelet coefficients of each layer are extracted to obtain the time-frequency characteristic components.

6. The method for locating and predicting traveling wave transmission characteristics faults of a multi-type cable architecture according to claim 5, characterized in that: Extracting a fault feature vector group according to the time-frequency feature components and constructing a fault mapping network includes the following steps: Performing Fourier transform on the modulus sequence and phase sequence in the time-frequency characteristic component to extract amplitude spectrum and phase spectrum features; Extract characteristic frequencies and characteristic frequency bands based on multi-scale dynamic threshold criteria; A hierarchical weighted combination strategy is used to construct a fault feature vector group; The attention mechanism is used for feature fusion to obtain the fault mapping network, which is as follows: Calculate feature attention weights through the interaction between query matrix, key matrix and value matrix; Using the feature attention weights, weighted fusion is performed on the fault feature vector group, and the fused fault feature vector group is reorganized according to the fault type and the fault position, respectively, to construct a fault type feature matrix and a position feature matrix; Performing singular value decomposition on the fault type feature matrix and the position feature matrix to extract main feature components; Based on the main characteristic components, respectively calculating the mapping matrix between the fault type characteristic matrix and the fault type label, and the mapping matrix between the position characteristic matrix and the fault position label, and establishing a fault mapping relationship; The fault mapping network is constructed based on a mapping matrix between the fault type feature matrix and the fault type label, and a mapping matrix between the position feature matrix and the fault position label.

7. The method for locating and predicting traveling wave transmission characteristics faults of a multi-type cable architecture according to claim 6, wherein: The fault type feature matrix and the position feature matrix are collaboratively calculated to complete the joint prediction of the fault type and the position, and the fault warning information is output, including the following steps: Based on the fault mapping network, performing a tensor product operation on the fault type feature matrix and the position feature matrix to obtain a collaborative feature matrix; Calculating the cosine similarity of each pair of features in the collaborative feature matrix to form a feature similarity matrix; If the feature similarity is greater than a first preset similarity threshold, the corresponding feature combination is marked as a high-correlation feature pair; If the feature similarity is less than the first preset similarity threshold but greater than the second preset similarity threshold, it is marked as a medium-correlated feature pair; If the feature similarity is less than a second preset similarity threshold, it is marked as a low-correlation feature pair; the high-correlation feature pair is associated with the sensor sensitivity of the node group to establish a feature spatiotemporal mapping; Tracking the feature similarity change trend of the highly correlated feature pair based on the feature spatiotemporal mapping, if the feature similarity shows a continuous upward trend and the change rate exceeds the normal fluctuation range, determining the warning level in combination with the time series information of the fault feature vector group; Perform location analysis on the position with the highest feature similarity. Specifically, match the position with the highest feature similarity with the sensor sensitivity of the corresponding node group. If the sensor sensitivity is greater than the average sensitivity of the node group, it is judged as the core position of the fault. Otherwise, find the position with the highest sensor sensitivity in the adjacent node group as the possible fault position. Integrate the feature spatiotemporal mapping results, output warning information including fault time, location and development trend, and update the verified feature pattern to the fault mapping network.

8. A system for locating and predicting traveling wave transmission characteristics faults of multiple types of cable structures, based on the method for locating and predicting traveling wave transmission characteristics faults of multiple types of cable structures according to any one of claims 1 to 7, characterized in that: include, Data acquisition module, used to monitor multiple types of cable networks in real time and obtain status data streams containing traveling wave characteristics; Constructing an analysis module, which is used to establish a dynamic reconstruction model of cable traveling wave transmission characteristics, perform a joint analysis of the state data stream in the time and frequency domains, extract a fault feature vector group, and construct a fault mapping network according to the fault feature vector group; The prediction and warning module is used to perform collaborative operations on the fault type feature matrix and the position feature matrix to complete the joint prediction of the fault type and position and output fault warning information.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for locating and predicting faults of traveling wave transmission characteristics of multi-type cable architectures according to any one of claims 1 to 7 are implemented.

10. 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 locating and predicting traveling wave transmission characteristics faults of multi-type cable architectures according to any one of claims 1 to 7 are implemented.

Images