Cable fault detection system and detection method thereof

Abstract

The invention discloses a cable fault detection system and a detection method thereof, and particularly relates to the technical field of cable fault detection.The cable fault detection system comprises a sensor array module, a signal processing module, a fault analysis module, a positioning module and an alarm module; the signal processing module carries out wavelet transform preprocessing, the fault analysis module carries out classification based on a convolutional neural network, the positioning module calculates a fault distance by using a time reversal algorithm, and the alarm module carries out grading early warning according to fault grades and starts a self-healing strategy; the detection method comprises the steps of data acquisition, signal processing, fault identification, positioning and graded response. The system can comprehensively and accurately collect data, efficiently process signals, accurately classify faults, accurately determine fault positions and timely perform graded response, effectively improve the operation reliability of the cable, reduce fault loss and guarantee stable operation of a power system.

Description

Technical Field

[0001] The present invention relates to the technical field of cable fault detection, and more particularly, to a cable fault detection system and a detection method thereof. Background Art

[0002] In modern power transmission systems, cables are key carriers of electrical energy. Their operational stability and reliability are crucial to the entire power system. With the rapid development of urban construction and the growing demand for industrial electricity, the use of cables is increasing, and their application scenarios are becoming increasingly complex. This has significantly increased the probability of cable failures. A cable failure not only disrupts power supply, impacting residents' lives and industrial production, but can also cause serious safety incidents and result in significant economic losses.

[0003] Currently, existing cable fault detection technologies have numerous limitations. Some traditional detection methods rely on a single type of sensor, capturing only a single operating parameter of the cable. This makes it difficult to fully reflect the cable's actual operating status and can easily miss potential faults. For example, monitoring only current or voltage signals cannot promptly detect insulation degradation caused by cable heating. In terms of signal processing, traditional methods often fail to effectively remove noise interference, making the extracted fault signature unclear and affecting the accuracy of fault diagnosis.

[0004] Some fault analysis methods based on simple algorithms suffer from low classification accuracy and high misclassification rates when faced with complex cable fault patterns. When a cable simultaneously presents multiple potential faults, it's difficult to accurately distinguish the fault type, leading to a lack of targeted repair and remediation measures. Traditional fault location methods have limited accuracy and are unable to quickly and accurately pinpoint the specific location of the fault, extending troubleshooting and repair time and further expanding the scope of the fault's impact.

[0005] In addition, the existing fault alarm mechanism is usually relatively simple and lacks the ability to respond to the severity of the fault in a graded manner. It is impossible to take effective countermeasures at different stages of the fault, making it difficult to quickly handle cable faults and self-repair the power system.

[0006] Therefore, a cable fault detection system and a detection method thereof are proposed. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art, the present invention provides a cable fault detection system and a detection method thereof to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above object, the present invention provides the following technical solution: a cable fault detection system, comprising: The sensor array module is distributed along the length of the cable and is used to collect multi-modal signal data of the cable in real time, including voltage, current and temperature signals; The signal processing module receives sensor data and performs wavelet transform preprocessing to extract signal time-frequency characteristics; The fault analysis module classifies features based on a convolutional neural network model. The network structure meets the following requirements: Input layer → 3 convolutional layers → max pooling layer → fully connected layer → Softmax output layer; Positioning module, which uses time reversal algorithm to calculate the distance to the fault point; The alarm module generates multi-level warning signals according to the fault level and initiates the self-healing control strategy.

[0009] Preferably, the signal processing module performs the wavelet transform preprocessing processing formula as follows:

[0010] in, is the Morlet wavelet basis function, is the scale factor, is the translation factor.

[0011] Preferably, the positioning module uses the time reversal algorithm to calculate the distance to the fault point using the formula:

[0012] in, is the speed of light, is the delay between the reflected wave and the incident wave, is the dielectric constant of the cable.

[0013] Preferably, the sensor array comprises: High-frequency current transformer, sampling frequency ≥ 100kHz; Distributed fiber optic temperature sensor, spatial resolution ≤ 0.5m; Ultra-high frequency partial discharge sensor, detection bandwidth 300MHz-3GHz; The sensors are arranged at intervals of λ / 4, where λ is the wavelength of the electromagnetic wave in the cable.

[0014] Preferably, the convolutional neural network introduces an attention mechanism, and adds a channel attention submodule after the third convolutional layer, and its output feature map weight is calculated as:

[0015] in, is the sigmoid function, For the channel feature maps.

[0016] Preferably, the time reversal algorithm adopts a frequency domain compensation method, and the reflection signal processing formula is:

[0017] in, is the original signal spectrum, is the channel transfer function.

[0018] Preferably, the alarm module adopts a multi-objective optimization strategy to coordinate the balance between response speed and false alarm rate.

[0019] A cable fault detection method, comprising: Step S1: collecting cable operation data through a multi-source sensor array; Step S2: using improved wavelet transform to perform signal denoising and feature extraction; Step S3: construct a deep convolutional neural network for fault pattern recognition; Step S4: Calculate the fault location based on time reversal theory; Step S5: Start a hierarchical response mechanism according to the fault level.

[0020] Preferably, step S5 includes: Level 1 warning: When the partial discharge amount is greater than 50pC, the insulation aging warning is activated; Secondary response: When the temperature abnormality area is greater than 2m, the cooling system is triggered to increase pressure; Level 3 protection: When the short-circuit current exceeds the threshold, the solid-state circuit breaker is triggered within 0.1ms.

[0021] Technical effects and advantages of the present invention: 1. In the sensor array module, the high-frequency current transformer has a sampling frequency of ≥100kHz, accurately capturing subtle changes in cable current and promptly detecting early-stage partial discharge faults. The distributed fiber-optic temperature sensor has a spatial resolution of ≤0.5m, accurately determining the location of abnormal cable temperature. The ultra-high frequency partial discharge sensor has a detection bandwidth of 300MHz-3GHz, providing high-precision monitoring of cable insulation faults. The sensors are arranged at a λ / 4 spacing to comprehensively acquire multimodal signal data, providing rich and accurate data support for fault analysis.

[0022] 2. Wavelet transform preprocessing is used in the signal processing module to effectively remove noise, extract time-frequency features, and highlight fault characteristic components. Compared with traditional methods, it can better suppress noise interference and make subsequent fault analysis more accurate.

[0023] 3. By introducing an attention mechanism into the convolutional neural network of the fault analysis module and adding a channel attention submodule after the third convolutional layer, the network pays more attention to important channel features, improves the ability to identify fault characteristics, and can more accurately classify complex faults, reducing the misjudgment rate.

[0024] 4. By combining the time reversal algorithm of the positioning module with the frequency domain compensation method and utilizing the propagation characteristics of electromagnetic waves, the distance to the fault point can be accurately calculated. Compared with traditional positioning methods, it has higher accuracy and facilitates maintenance personnel to quickly find and repair faults.

[0025] 5. The alarm module generates multi-level warning signals based on the fault level and initiates a self-healing control strategy. A multi-objective optimization strategy balances response speed and false alarm rate. Level 1 warning monitors partial discharge to prevent insulation aging; level 2 responds to temperature anomalies to prevent fault progression; and level 3 protection rapidly disconnects the circuit in the event of a short circuit, protecting equipment and improving power system safety and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of the cable fault detection method of the present invention. DETAILED DESCRIPTION

[0027] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0028] like Figure 1 As shown, the present invention provides a cable fault detection system and a cable fault detection method using the system. First, the sensor array modules are installed and data is collected. Specifically, the sensor array modules are distributed along the length of the cable. The sampling frequency of the high-frequency current transformer is set to ≥100kHz. This high sampling frequency can accurately capture subtle changes in the cable current signal. For example, when an early partial discharge fault occurs in the cable, the high-frequency current transformer can collect weak high-frequency current fluctuations, providing a detailed data foundation for subsequent fault diagnosis. Compared with traditional sensors with lower sampling frequencies, potential faults can be discovered more promptly and accurately.

[0029] The spatial resolution of the distributed fiber optic temperature sensor is set to ≤0.5m. Such a high spatial resolution can accurately determine the specific location of the cable temperature anomaly. When the cable becomes locally hot due to overload or poor contact, the hot area can be quickly located, which is conducive to taking timely measures to prevent the fault from expanding. Ordinary temperature sensors cannot achieve such precise positioning.

[0030] The detection bandwidth of the UHF partial discharge sensor is set at 300MHz-3GHz. This frequency band can effectively detect partial discharge signals inside the cable. It has a highly sensitive detection capability for partial discharge phenomena caused by the degradation of cable insulation performance, greatly improving the monitoring accuracy of insulation faults.

[0031] The sensors are arranged at λ / 4 intervals (λ is the wavelength of the electromagnetic wave in the cable). This arrangement optimizes the distance between sensors, ensures the complementarity and synergy of the data collected by different sensors, and can comprehensively obtain multimodal signal data on the cable's operating status, including voltage, current, and temperature signals, providing rich and comprehensive data support for subsequent accurate fault analysis.

[0032] After the signal processing module receives the data collected by the sensor, it performs wavelet transform preprocessing. According to the formula

[0033] in, is the Morlet wavelet basis function, is the scale factor, is the translation factor, by choosing an appropriate scale factor and translation factors , performing time-frequency analysis on the collected signals. Wavelet transforms can effectively remove noise from signals while extracting their time-frequency characteristics. For example, they can decompose complex voltage, current, and temperature signals into components of different frequencies and time scales, highlighting the fault characteristics in the signals. This allows subsequent fault analysis modules to more accurately identify fault modes. Compared to traditional signal processing methods, wavelet transforms can better suppress noise interference while preserving signal characteristics.

[0034] The fault analysis module classifies the features extracted by the signal processing module based on the convolutional neural network model. The convolutional neural network structure is input layer → 3 convolution layers → maximum pooling layer → fully connected layer → Softmax output layer, and a channel attention submodule is added after the third convolution layer. The weight calculation method of the output feature map of the channel attention submodule is ,in, is the sigmoid function, For the channel feature maps.

[0035] Specifically, the convolutional layer automatically extracts signal features, with the convolution kernel sliding across the feature map to learn feature representations at different levels. The max pooling layer downsamples the feature map, preserving key features while reducing data volume and improving computational efficiency. The fully connected layer integrates the features processed by convolution and pooling, ultimately outputting the fault classification results through the softmax output layer. The introduction of the attention mechanism allows the network to focus more on channel features that contribute significantly to fault classification, improving the model's ability to identify fault features. Compared to networks without the attention mechanism, it can more accurately classify different types of cable faults in complex fault scenarios, reducing the false positive rate.

[0036] The positioning module uses the time reversal algorithm to calculate the distance to the fault point. The calculation formula is: ,in, is the speed of light, is the delay between the reflected wave and the incident wave, is the dielectric constant of the cable. The time reversal algorithm uses the propagation characteristics of electromagnetic waves in the cable. When a cable fails, the transmitted detection signal will generate a reflected wave at the fault point. By measuring the time delay between the reflected wave and the incident wave , combined with the known speed of light and cable dielectric constant , the distance to the fault point can be accurately calculated; In addition, the time reversal algorithm adopts the frequency domain compensation method, and the reflection signal processing formula is:

[0037] in, is the original signal spectrum, is the channel transfer function. The frequency domain compensation method can effectively compensate for the attenuation and distortion of the signal during transmission, improve the quality of the reflected signal, and thus more accurately measure the time delay. ,Compared with traditional positioning methods, this algorithm has higher ,positioning accuracy, and can locate the fault point at a more ,precise position, making it easier for maintenance personnel to ,quickly find and repair the fault.

[0038] The alarm module generates multi-level warning signals based on the fault level and initiates a self-healing control strategy, using a multi-objective optimization strategy to coordinate the balance between response speed and false alarm rate. The specific hierarchical response mechanism is as follows: Level 1 Warning: When the partial discharge exceeds 50pC, the insulation aging warning is activated. Partial discharge is a key indicator of cable insulation degradation. By monitoring partial discharge in real time, a warning is issued when it exceeds the threshold, alerting personnel to the cable insulation condition and taking appropriate measures, such as increasing monitoring frequency and arranging preventive maintenance, to prevent further insulation aging and serious failures, effectively extending the cable's service life.

[0039] Level 2 response: When the temperature abnormality zone is greater than 2 meters, the cooling system pressurization is triggered. Abnormally high cable temperatures can cause serious problems such as insulation damage. Promptly initiating the cooling system pressurization can quickly reduce the cable temperature, preventing further deterioration of the fault, ensuring safe and stable operation of the cable, and reducing the probability of failures caused by overheating.

[0040] Level 3 protection: When the short-circuit current exceeds the threshold, the solid-state circuit breaker is triggered within 0.1ms. Short-circuit faults generate extremely high currents, which can cause serious damage to cables and related equipment. The rapid triggering of the solid-state circuit breaker can quickly cut the circuit, protecting cables and other equipment from the impact of the short-circuit current, greatly reducing the risk of equipment damage and improving the safety and reliability of the power system.

[0041] In summary, the present invention realizes the accurate detection, positioning and graded response of cable faults through the coordinated work of various modules, which has significant improvements in improving cable operation reliability, reducing fault losses, and ensuring stable operation of the power system.

[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A cable fault detection system, characterized in that: include: The sensor array module is distributed along the length of the cable and is used to collect multi-modal signal data of the cable in real time, including voltage, current and temperature signals; The signal processing module receives sensor data and performs wavelet transform preprocessing to extract signal time-frequency characteristics; The fault analysis module classifies features based on a convolutional neural network model. The network structure meets the following requirements: Input layer → 3 convolutional layers → max pooling layer → fully connected layer → Softmax output layer; Positioning module, which uses time reversal algorithm to calculate the distance to the fault point; The alarm module generates multi-level warning signals according to the fault level and initiates the self-healing control strategy.

2. A cable fault detection system according to claim 1, characterized in that: The signal processing module performs the wavelet transform preprocessing processing formula as follows: in, is the Morlet wavelet basis function, is the scale factor, is the translation factor.

3. A cable fault detection system according to claim 1, characterized in that: The positioning module uses the time reversal algorithm to calculate the distance to the fault point: in, is the speed of light, is the delay between the reflected wave and the incident wave, is the dielectric constant of the cable.

4. A cable fault detection system according to claim 1, characterized in that: The sensor array comprises: High-frequency current transformer, sampling frequency ≥ 100kHz; Distributed fiber optic temperature sensor, spatial resolution ≤ 0.5m; Ultra-high frequency partial discharge sensor, detection bandwidth 300MHz-3GHz; The sensors are arranged at intervals of λ / 4, where λ is the wavelength of the electromagnetic wave in the cable.

5. A cable fault detection system according to claim 1, characterized in that: The convolutional neural network introduces an attention mechanism and adds a channel attention submodule after the third convolutional layer. The output feature map weight is calculated as: in, is the sigmoid function, For the channel feature maps.

6. A cable fault detection system according to claim 1, characterized in that: The time reversal algorithm adopts the frequency domain compensation method, and the reflection signal processing formula is: in, is the original signal spectrum, is the channel transfer function.

7. A cable fault detection system according to claim 1, characterized in that: The alarm module adopts a multi-objective optimization strategy to coordinate the balance between response speed and false alarm rate.

8. A method for fault detection using a cable fault detection system according to any one of claims 1 to 7, characterized in that: include: Step S1: collecting cable operation data through a multi-source sensor array; Step S2: using improved wavelet transform to perform signal denoising and feature extraction; Step S3: construct a deep convolutional neural network for fault pattern recognition; Step S4: Calculate the fault location based on time reversal theory; Step S5: Start a hierarchical response mechanism according to the fault level.

9. A cable fault detection method according to claim 8, characterized in that: Step S5 includes: Level 1 warning: When the partial discharge amount is greater than 50pC, the insulation aging warning is activated; Secondary response: When the temperature abnormality area is greater than 2m, the cooling system is triggered to increase pressure; Level 3 protection: When the short-circuit current exceeds the threshold, the solid-state circuit breaker is triggered within 0.1ms.

Images