Mining cable fault diagnosis system and method based on multiband leakage current characteristics

Abstract

The invention belongs to the technical field of mining cable fault diagnosis, and particularly discloses a mining cable fault diagnosis system and method based on multiband leakage current characteristics. According to the method, a flexible dynamic threshold value denoising method is provided based on the multi-band leakage current characteristics of the mining cable, the multi-band leakage current signals can be preprocessed, and the multi-band leakage current characteristics are subjected to multi-level mining and analysis in combination with the deep convolutional neural network. The method shows relatively high diagnosis precision in a complex environment, and can effectively improve the timeliness and accuracy of fault diagnosis of the mining cable. In addition, in order to meet the requirement for collecting the multi-band leakage current, the system can independently collect the low-frequency leakage current and the high-frequency ground wire leakage current in the mining cable, and multi-band leakage current collection is achieved; the technical problems that an existing zero-sequence current sensor is not sensitive to high-frequency response, a high-frequency sensor is not sensitive to low-frequency current, and a special multi-band leakage current collecting device does not exist are solved.

Description

Technical Field

[0001] The invention belongs to the technical field of mining cable fault diagnosis, and in particular relates to a mining cable fault diagnosis system and method based on multi-band leakage current characteristics. Background Art

[0002] The complex environment, harsh working conditions and frequent load changes of the mines have put forward extremely high requirements on the operation of mining cables, which are the core components of the coal mine power supply system. The air humidity in coal mines is high, and the temperature difference between different areas is large. The insulation of mining cables is very easy to age, resulting in insulation degradation. Mining cables are prone to overheating, insulation aging and other problems when subjected to harsh conditions such as high temperature and high humidity, and when facing long-term high-load operation. In addition, the space underground in coal mines is narrow, and mining cables are easily damaged by sudden situations such as smashing, collision and dragging, causing grounding or leakage faults. Once a mining cable fails, it will cause power outages in the mine, production interruptions, and may even cause safety accidents. Therefore, accurate monitoring and fault diagnosis of the operating status of mining cables are of great significance to maintaining equipment safety and ensuring stable mine production.

[0003] In recent years, with the development of artificial intelligence and signal processing technology, the fault diagnosis method of mining cables based on feature recognition and machine learning algorithms has gradually become a new means of detecting defects and faults in mining cables. When a fault occurs, the multi-band leakage current signal in the bottom line or shielding layer of the mining cable contains rich feature information, which can effectively reflect the defects or fault status inside the mining cable. However, due to the complex and changeable environment of the mine, the frequency, wavelength, and intensity of the multi-band leakage current signal of the mining cable will fluctuate accordingly, making it difficult for traditional judgments that rely on experience to effectively identify the operating status of the mining cable. Therefore, it is urgent to propose an intelligent diagnosis technology based on multi-band leakage current signals to achieve accurate identification of mining cable anomalies, provide important guarantees for the safe and stable operation of mine power equipment, and promote the improvement of the intelligent management level of mine power systems. Summary of the invention

[0004] The purpose of the present invention is to propose a mining cable fault diagnosis method based on multi-band leakage current characteristics. The method mines the multi-band leakage current characteristics of mining cables and proposes a flexible dynamic threshold denoising method. The multi-band leakage current signals collected by the system can be pre-processed. At the same time, the multi-band leakage current characteristics are combined with a deep convolutional neural network to perform multi-level mining and analysis, which is conducive to the accurate identification of mining cable abnormalities.

[0005] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a mining cable fault diagnosis method based on multi-band leakage current characteristics, comprising the following steps: step 1. performing noise reduction preprocessing on the acquired mining cable multi-band leakage current signal; wherein the mining cable multi-band leakage current signal is obtained by combining the synchronously collected high-frequency leakage current signal and the low-frequency leakage current signal; wherein the high-frequency leakage current signal and the low-frequency leakage current signal are two homologous data of the same original signal; step 2. performing feature extraction on the mining cable multi-band leakage current signal preprocessed by step 1, and marking the extracted mining cable multi-band leakage current characteristics to obtain a training data set; step 3. building a mining cable fault diagnosis model based on a convolutional neural network, training the mining cable fault diagnosis model based on the training data set obtained in step 2, and obtaining a trained mining cable fault diagnosis model; performing noise reduction preprocessing on the mining cable multi-band leakage current signal to be diagnosed according to the method of step 1, and after feature extraction in step 2, inputting the trained mining cable fault diagnosis model to obtain a mining cable fault diagnosis result.

[0006] The purpose of the present invention is to propose a mining cable fault diagnosis system based on multi-band leakage current characteristics, which can separately collect low-frequency leakage current in the cable and high-frequency grounding wire leakage current to achieve multi-band leakage current collection.

[0007] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: a mining cable fault diagnosis system based on multi-band leakage current characteristics, including a multi-band leakage current mining cable fault information acquisition device and a host computer; the multi-band leakage current mining cable fault information acquisition device is used to collect multi-band leakage current signals of mining cables, and transmit the collected multi-band leakage current signals of mining cables to the host computer; the host computer includes a memory and one or more processors; executable codes are stored in the memory; when the processor executes the executable code, the mining cable fault diagnosis method based on multi-band leakage current characteristics as described above is implemented.

[0008] The multi-band leakage current mining cable fault information collection device comprises an equipment bracket, a grounding wire guide channel, a multi-band leakage current data collection unit for synchronously collecting multi-band currents, and a collection data processing unit; the grounding wire guide channel is arranged on the equipment bracket, and has two ports for the mining cable grounding wire to pass through; the collection data processing unit is also installed on the equipment bracket; the multi-band leakage current data collection unit comprises a low-frequency leakage current collection unit and a high-frequency leakage current collection unit; wherein, the low-frequency leakage current collection unit and the high-frequency leakage current collection unit are both arranged on the equipment bracket, and the low-frequency leakage current collection unit is located above the high-frequency leakage current collection unit and the grounding wire guide channel; the multi-band leakage current mining cable fault information collection device is arranged at the end position of the mining cable; wherein, the ring of the low-frequency leakage current collection unit The collecting part is arranged in a surrounding manner on the mining cable; the grounding wire led out from the end of the mining cable first passes back in the opposite direction through the gap formed between the mining cable and the annular collecting part of the low-frequency leakage current collecting unit, and then is led out through the grounding wire guide channel and grounded; the opposite direction here refers to the direction from the end of the mining cable where the multi-band leakage current mining cable fault information collecting device is located to the other opposite end of the mining cable; the annular collecting part of the high-frequency leakage current collecting unit is arranged in a surrounding manner on the grounding wire guide channel; the low-frequency leakage current collecting unit and the high-frequency leakage current collecting unit are both connected to the collecting data processing unit, and are respectively used to send the corresponding collected low-frequency leakage signal and high-frequency leakage signal to the collecting data processing unit; the collecting data processing unit of the multi-band leakage current mining cable fault information collecting device is connected to the host computer.

[0009] The present invention has the following advantages: As mentioned above, the present invention relates to a mining cable fault diagnosis system and method based on multi-band leakage current characteristics. The method of the present invention is based on the multi-band leakage current characteristics of mining cables, and proposes a flexible dynamic threshold denoising method, which can pre-process the multi-band leakage current signal, and combine the deep convolutional neural network to perform multi-level mining and analysis on the multi-band leakage current characteristics. First, the multi-band leakage current signal collected by the high-frequency leakage sensor is denoised, and then the main frequency and fault frequency of the cable multi-band leakage current are extracted under normal and fault conditions, and feature extraction and deep convolutional neural network parameter setting are performed. This method shows high diagnostic accuracy in complex environments, can effectively improve the timeliness and accuracy of mining cable fault diagnosis, and has good practical application value and promotion prospects. In addition, in order to meet the requirements of the above-mentioned method for collecting multi-band leakage current, the system of the present invention can separately collect low-frequency leakage current in mining cables and high-frequency grounding wire leakage current, thereby realizing multi-band leakage current collection, which well solves the technical problems of the existing zero-sequence current sensor being insensitive to high-frequency response and the high-frequency sensor being insensitive to low-frequency current, and the lack of a dedicated multi-band leakage current collection device. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 The present invention is a flowchart of a method for diagnosing faults of mining cables based on multi-band leakage current characteristics in an embodiment of the present invention.

[0011] Figure 2 This is a structural diagram of the mining cable fault diagnosis model built in an embodiment of the present invention.

[0012] Figure 3 It is a structural schematic diagram of a multi-band leakage current mining cable fault information collection device in an embodiment of the present invention.

[0013] Figure 4 It is a side sectional view of a multi-band leakage current mining cable fault information collection device in an embodiment of the present invention.

[0014] Figure 5 This is an installation structure diagram of a multi-band leakage current mining cable fault information collection device in an embodiment of the present invention.

[0015] Among them, 1-equipment bracket, 2-grounding wire guide channel, 3-acquisition data processing unit, 4-low-frequency leakage current acquisition unit, 5-high-frequency leakage current acquisition unit, 6-mining cable, 7-grounding wire, 8-clamp handle; 9-first shell, 10-first coil, 11-first magnetic core, 12-first output circuit, 13-second shell, 14-second coil, 15-second magnetic core, 16-second output circuit, 17-insulating fixed bracket, 18-adjustable rolling belt. DETAILED DESCRIPTION

[0016] The present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods: Example 1 In view of the long-term operation of mining cables in complex mine environments, they are easily affected by factors such as moisture, dust, and high load, which leads to the problem of aging of mining cables. This Example 1 proposes a mining cable fault diagnosis method based on multi-band leakage current characteristics. The method is based on multi-band leakage current characteristics for mining, establishes a flexible dynamic threshold denoising method, and proposes a mining cable fault diagnosis model based on a convolutional neural network. The method makes full use of the advantages of the flexible dynamic threshold algorithm in removing noise interference and retaining signal characteristics. By decomposing and transforming the collected multi-band leakage current signal, the background noise in complex working conditions such as coal mines is removed, and then the time-frequency features after denoising are extracted and input into the convolutional neural network model for feature learning and fault identification. Experiments have proved that the improved model of the present invention can effectively identify the fault state of mining cables under complex working conditions, and realize efficient and accurate diagnosis and evaluation of the operating status of mining cables.

[0017] like Figure 1As shown, a mining cable fault diagnosis method based on multi-band leakage current characteristics includes the following steps: Step 1. Pre-processing the acquired multi-band leakage current signal of the mining cable to reduce noise; wherein the multi-band leakage current signal of the mining cable is obtained by combining the synchronously collected high-frequency leakage current signal and the low-frequency leakage current signal.

[0018] The high-frequency leakage current signal and the low-frequency leakage current signal are two homologous data of the same original signal.

[0019] In the present embodiment, the high-frequency leakage current signal and the low-frequency leakage current signal can generally be used as two signal input models for feature mining and diagnosis in fault diagnosis. Of course, the data fusion principle can also be adopted, that is, the two signals are fused.

[0020] Specifically, there are at least two ways to fuse the high-frequency leakage current signal and the low-frequency leakage current signal, namely, cutoff frequency fusion and weighted fusion according to the different response characteristics of the magnetic core to each frequency band.

[0021] From at least two fusion methods, one can be selected as the fusion method of the high-frequency leakage current signal and the low-frequency leakage current signal. Of course, the above two fusion methods are only examples, and a specific fusion method can also be used as needed.

[0022] Step 2. Extract features of the multi-band leakage current signal of the mining cable preprocessed in step 1, and annotate the extracted multi-band leakage current features of the mining cable to obtain a training data set.

[0023] The acoustic signal frequency of a normally operating mining cable is usually concentrated below 1000Hz, mainly manifested as n frequency components with 50Hz as the base frequency. However, when a mining cable fails, the frequency range of the zero-sequence leakage current signal will be significantly extended, usually to within 4kHz. Noise interference exceeding 4kHz can be regarded as a minor factor and can be almost ignored.

[0024] In order to more effectively analyze and diagnose different types of mining cable faults, it is necessary to systematically classify noise signals. Through this classification method, the characteristics of each type of noise can be more accurately extracted, providing a basis for subsequent fault type identification.

[0025] Due to the complexity and time-varying characteristics of non-stationary signals, it is extremely difficult to directly analyze and apply these signals for fault diagnosis. In order to pre-process the extracted acoustic signal, the present invention adopts a newly improved microwave threshold denoising algorithm to implement the pre-processing process of multi-band leakage current signals. Microwave threshold denoising uses a decomposition transformation method to decompose the signal into different scales, and processes the high-frequency coefficients by setting the threshold value to remove noise, and finally reconstructs the effective signal with noise removed.

[0026] Microwave threshold denoising includes three steps, namely, original signal transformation and decomposition, threshold setting and denoising, and signal reconstruction. The following introduces the process of using the improved microwave threshold denoising algorithm to implement denoising preprocessing of multi-band leakage current signals.

[0027] Step 1.1. Decomposition of original signal.

[0028] First, the signal x(t) is decomposed into high-frequency coefficients representing high-frequency components and low-frequency coefficients representing low-frequency components.

[0029] (1).

[0030] Where x(t) represents the obtained multi-band leakage current signal of mining cable.

[0031] W j,k is the high-frequency coefficient of the jth layer and position k, representing the high-frequency part of the signal x(t); V J,k is the low-frequency coefficient at the Jth layer and position k, representing the low-frequency part of the signal x(t); is the basis function, is a quantity function.

[0032] Step 1.2. Flexible dynamic threshold denoising.

[0033] The high-frequency coefficients decomposed in step 1.1 above are The following processing is performed, as shown in formulas (2) and (3).

[0034] (2).

[0035] (3).

[0036] Formula (2) and Formula (3) are the denoising threshold formula and the optimization threshold formula respectively, and λ is the threshold parameter.

[0037] Traditional threshold functions are prone to introduce fixed deviations when processing high-frequency decomposition coefficients, which in turn introduce errors when reconstructing signals, resulting in reduced accuracy and significant distortion. However, the method of simply setting the denoising threshold value sets the part of the decomposition coefficient that exceeds the set value to zero, which causes unnecessary oscillations in the reconstructed signal and discontinuities at the threshold value position. This processing method is not conducive to maintaining the continuity and stability of the signal.

[0038] In order to solve the above problem, the present invention proposes a flexible dynamic denoising threshold function, the formula of which is as follows: (4).

[0039] (5).

[0040] In the formula, x represents the high frequency coefficient , T(x) is the denoising threshold output, The dynamic threshold value is used to determine which coefficients are compressed or retained. It is a small positive number used to avoid division by zero. It is usually set to 0.01 or other sufficiently small values ​​to ensure calculation stability. N represents the selected signal length, and a represents the number of signal data points.

[0041] Find After that, it is substituted into formula (2) as the high-frequency coefficient of signal reconstruction. Through the above steps, the flexible dynamic denoising threshold coefficient can be obtained through formula (4), and the optimized threshold coefficient can be obtained through formula (3).

[0042] Step 1.3. Signal reconstruction.

[0043] The processed decomposition coefficients are used to reconstruct the signal to obtain a denoised signal. In the signal reconstruction process of step 1.3, it is usually necessary to select one of the threshold coefficients according to the signal characteristics and needs.

[0044] Specifically, if the noise amplitude in the signal is large and a pure filtered signal is required, the optimized threshold coefficient calculated by formula (3) is used; if the noise amplitude in the signal is low and the details in the signal need to be retained, the flexible dynamic denoising threshold coefficient calculated by formula (4) is used.

[0045] After the flexible dynamic threshold value denoising obtained by formula (4) or the optimized threshold value denoising by formula (3), the new coefficients are substituted into formula (1), and the inverse operation of formula (1) is performed to obtain the denoised signal S(i); (6).

[0046] In the feature extraction of multi-band leakage current signal of mining cables, the time-frequency analysis method has a strong ability to capture non-stationary signals and can effectively monitor the state parameters of mining cables. Based on the structural characteristics of mining cables, the optimization scheme of sensor layout, actual operating conditions and multiple sets of fault data, the present invention systematically extracts the following features to characterize the multi-band leakage current characteristics of mining cable status by combining the comprehensive analysis method of time domain and frequency domain.

[0047] Among them, the extracted multi-band leakage current characteristics of mining cables include empirical characteristics corresponding to various faults, main frequency characteristics, odd-even ratio, harmonic ratio, spectrum center of gravity, total harmonic distortion, instantaneous frequency variation and energy entropy.

[0048] 1. Empirical characteristics corresponding to major faults.

[0049] The working conditions of mining cables under different load conditions, environmental noise and various fault conditions (such as partial discharge and single-phase grounding fault) are marked according to the empirical parameter characteristics corresponding to each type of fault.

[0050] For example, single-phase to ground fault characteristics.

[0051] When a single-phase grounding occurs, a low-frequency zero-sequence current is generated, and its characteristics are different in different operating line states. For a neutral point ungrounded system, when a single-phase grounding fault occurs, the single-phase grounding zero-sequence current Phase line to ground capacitance current when the system is operating normally 3 times.

[0052] (7).

[0053] The low-frequency zero-sequence current of a power system with a directly grounded neutral point is equal to the short-circuit current.

[0054] For example, locally amplify fault characteristics.

[0055] When partial discharge occurs, the high-frequency partial discharge pulse current can be measured by the pulse current method. The center frequency of the current is about 20k, and the threshold can be set , when the pulse current value When it is greater than this threshold, it is regarded as a partial discharge current characteristic.

[0056] (8).

[0057] This part is a commonly used fault judgment method. In the present invention, it is only regarded as a feature as an input of the diagnosis model.

[0058] 2. Main frequency characteristics.

[0059] Main frequency F m It indicates the main frequency point where the energy in the signal spectrum is concentrated, and is the core frequency feature of the multi-band leakage current signal. This feature is used to reveal the main components of the electrical fundamental frequency in the operation of mining cables.

[0060] Under normal circumstances, the main frequency of mining cables is usually equal to the power frequency (50 Hz). When an abnormal deviation occurs, it indicates that the system is in an abnormal state or is subject to external interference. The calculation formula is: (9).

[0061] In the formula, X f is the signal amplitude at frequency f.

[0062] 3. Odd-even ratio.

[0063] Odd-even ratio P mIt is used to measure the energy distribution of odd and even multiples in the signal. In normal operation, odd multiples tend to dominate. When the energy of even multiples increases significantly, the mining cable may have non-metallic grounding or arc grounding discharge problems. The calculation formula is: (10).

[0064] Where n is the upper limit of the multiple within the statistical range.

[0065] 4. Harmonic ratio.

[0066] Harmonic ratio H m Evaluate the relative strength of the fundamental frequency and its harmonic components. A high harmonic ratio often means an increase in the nonlinear characteristics of the system, which can be used to determine the health of mining cables. The calculation formula is: (11).

[0067] In the formula, is the amplitude of the nth harmonic, f0 is the fundamental frequency, represents the fundamental amplitude, and N represents the total number of harmonics.

[0068] 5. Spectral center of gravity.

[0069] Spectrum center C f Represents the center of gravity of the signal spectrum. It reflects the dynamic characteristics of the system by measuring the deviation of the frequency distribution. The displacement of the center of gravity usually indicates the change of frequency components in the operation of the system. The calculation formula is: (12).

[0070] Where, X f is the frequency amplitude in the vibration signal spectrum, represents the maximum frequency, Indicates the minimum frequency. The frequency spectrum center of the noise at different measuring points and under different working conditions of mining cables reflects the degree of concentrated distribution of frequency components. Its overall trend is similar to the frequency complexity.

[0071] 6. Total harmonic distortion.

[0072] Total harmonic distortion (THD) quantitatively evaluates the degree of distortion of harmonic components in a signal relative to the fundamental frequency. It is a key indicator for describing harmonic pollution in power systems and nonlinearity of equipment. A high THD value means that the equipment is subject to greater nonlinear effects. The calculation formula is: (13).

[0073] in, represents the amplitude of the nth harmonic, Indicates the fundamental amplitude.

[0074] 7. Instantaneous frequency variation.

[0075] The instantaneous frequency variation IFD reflects the dynamic characteristics of the frequency change in the signal. It is used to capture the possible frequency mutation and instability in the system. It is an important feature for identifying sudden faults. The calculation formula is: (14).

[0076] In the formula, represents the instantaneous phase, and T represents the transient time difference.

[0077] 8. Energy entropy.

[0078] Energy entropy E s It describes the complexity of energy distribution of signal spectrum and measures the effective characteristics of random components and irregular behaviors in the signal. High energy entropy usually reflects that the system is subject to more random interference or a mixture of multi-frequency components. The calculation formula is: (15).

[0079] in, represents the energy of sample i.

[0080] Through the leakage current characteristics, the mine cable fault diagnosis results that can be achieved are the output of the mine cable fault diagnosis model, including no fault, single-phase grounding fault, high-resistance single-phase grounding fault, partial discharge and two-phase short circuit.

[0081] Step 3. Build a mining cable fault diagnosis model based on convolutional neural network, such as Figure 2 As shown, the mining cable fault diagnosis model is trained based on the training data set obtained in step 2 to obtain a trained mining cable fault diagnosis model.

[0082] As an important model of deep learning, convolutional neural network has shown great superiority in processing data with spatial structure such as images, videos, and speech. Through local connection and parameter sharing mechanism, convolutional neural network reduces the redundancy of a large number of parameters in traditional fully connected networks, enabling it to efficiently extract local and global features from data.

[0083] like Figure 2 As shown, the convolutional neural network in this embodiment is composed of a convolutional layer, a pooling layer, an activation function and a fully connected layer, which can gradually abstract the input data and realize automatic learning and classification of complex tasks.

[0084] After extracting the multi-band leakage current characteristics of mining cables, the present invention uses this as input, uses the deep learning method of convolutional neural network to train the deep feature relationship between the input characteristics and fault classification, and forms an intelligent agent for multi-band leakage current signal fault diagnosis. After the sample is input into the convolutional neural network, the output result of each convolution layer is a code parameter without physical meaning. Until the full connection is passed through the softmax function, the output content is the probability distribution of various faults.

[0085] The main function of the convolution layer is to locally scan the input data through a set of convolution kernels (Filters) to extract local features. The essence of the convolution operation is to generate a new feature map by performing element-by-element dot product between the convolution kernel and the local area of ​​the input feature map. Each convolution kernel in the convolution layer corresponds to extracting a certain type of specific pattern in the input (such as edges, textures, etc.). By stacking layer by layer, the convolution layer can extract the feature representation of the input from low-level to high-level.

[0086] The convolution operation can be expressed in the following mathematical form: (16).

[0087] Where: is the pixel value at the output feature position (i, j) of the convolutional layer l+1, is the value input in the lth layer, the position is (i+m, j+n), and the offsets of m and n are used to correspond to different positions of the convolution kernel. M represents the total offset of i, and N1 represents the total offset of j. represents the convolution kernel, represents the integration variable.

[0088] In order to further reduce the spatial dimension of the feature map, reduce computational complexity, and enhance the model's anti-interference ability, CNN often uses a pooling layer. The pooling layer downsamples the local area, retaining important feature information and discarding unnecessary details.

[0089] The most commonly used pooling method is Max Pooling, which retains local significant features by selecting the maximum value in each pooling window. The mathematical expression of Max Pooling is: (17).

[0090] In the formula, represents the maximum pooling output, p and q are the sizes of the pooling window. Pooling can not only reduce the parameters and computational complexity of the model, but also provide a certain degree of translation invariance for the model, making it robust to small changes in the input signal.

[0091] Convolution operations are linear in nature. In order to enhance the model's ability to learn nonlinear features, CNN usually introduces an activation function after the convolution layer. The most commonly used activation function is ReLU (Rectified Linear Unit), and its formula is: (18).

[0092] ReLU can effectively avoid the gradient vanishing problem, so that the network can still maintain good learning efficiency when the training depth is deep.

[0093] After several layers of convolution and pooling, CNN will flatten the extracted features and input them into the fully connected layer (FC Layer). The fully connected layer is similar to the layers in traditional neural networks, where each neuron is connected to all neurons in the previous layer. The fully connected layer maps the previously extracted features to the final classification space and performs classification or regression prediction based on the learned weights. The calculation formula for the fully connected layer is: (19).

[0094] In the formula, represents the output of the fully connected layer, x j is the input feature, w ij is the weight, b i The final output is usually passed through the softmax function for multi-classification tasks, which converts the output value into a probability distribution: (20).

[0095] Where K is the number of categories, and P(y=i) is the probability that the input data belongs to category i.

[0096] The training process of convolutional neural networks relies on the back propagation algorithm, which optimizes the weights, biases and other parameters of the convolution kernel by calculating the gradient of the loss function. Specifically, the loss function The calculation formula is as follows: (twenty one).

[0097] In the formula, y i is the probability of the true label, is the probability predicted by the model. The mining cable fault diagnosis model gradually updates the parameters through the gradient descent algorithm to minimize the loss function and thus optimize the model performance.

[0098] The training process of the mining cable fault diagnosis model is as follows: In practical applications, the training of the mining cable fault diagnosis model is first completed, and all the feature data of the multi-band leakage current characteristics of the mining cable extracted in step 2 are used as input and input into the convolutional neural network. The convolutional neural network mines the deep feature correspondence between each input feature and the fault according to the set fault type.

[0099] By continuously adjusting the internal feature parameters of the convolutional neural network, the mining cable fault diagnosis model training is completed.

[0100] The trained mining cable fault diagnosis model is deployed on a host computer in a mining cable fault diagnosis system based on multi-band leakage current characteristics, and mining cable fault diagnosis is implemented according to the above method steps.

[0101] Specifically, the multi-band leakage current signal of the mining cable to be diagnosed is subjected to noise reduction preprocessing according to step 1, and after feature extraction in step 2, it is input into the trained model to obtain the mining cable fault diagnosis result.

[0102] The method of the present invention outputs the above-mentioned mining cable fault diagnosis model, and ultimately can accurately diagnose and identify mining cable faults by detecting multi-band leakage current signals when an abnormality occurs in the mining cable.

[0103] Embodiment 2 This embodiment 2 describes a mining cable fault diagnosis system based on multi-band leakage current characteristics, which includes a multi-band leakage current mining cable fault information acquisition device and a host computer. Among them, the host computer is generally composed of a high-performance industrial-grade workstation, which is mainly used for the training of fault diagnosis algorithms and actual diagnosis calculations. The multi-band leakage current mining cable fault information acquisition device is an information acquisition device specially designed for the diagnostic method described in the present invention.

[0104] Specifically, the multi-band leakage current mining cable fault information collection device is used to collect the multi-band leakage current signal of the mining cable, and transmit the collected multi-band leakage current signal of the mining cable to the host computer.

[0105] The host computer receives the data and implements the fault identification and diagnosis of mining cables based on multi-band leakage current feature mining through the program.

[0106] In this embodiment, the host computer includes a memory and one or more processors.

[0107] Among them, executable codes are stored in the memory; when the processor executes the executable codes, the mining cable fault diagnosis method based on multi-band leakage current characteristics as described in the above-mentioned embodiment 1 is implemented.

[0108] In addition, in response to the problems that the existing zero-sequence current sensors are insensitive to high-frequency response and high-frequency sensors are insensitive to low-frequency current, and there is no dedicated multi-band leakage current collection device, the present invention provides a multi-band leakage current mining cable fault information collection device, which can separately collect low-frequency leakage current in the cable and high-frequency grounding wire leakage current, thereby realizing multi-band leakage current collection.

[0109] like Figure 3As shown, the present invention designs a new multi-band leakage current mining cable fault information acquisition device according to the operating characteristics of mining cables. When a mining cable fails, its leakage current signal contains current responses of multiple frequency bands due to different fault types. In actual engineering acquisition, due to the characteristics of sensor materials, sensor coils composed of different materials respond differently to frequency ranges. Sensors composed of only one material cannot meet the demand for obtaining multi-band signals. In addition, mining cables are generally three-core cables. There is a difference between the fault leakage current composed of the algebraic sum of the three-phase currents and the leakage current flowing through the shielding layer grounding wire. The difference can be used to realize partial fault diagnosis and fault location of mining cables.

[0110] Therefore, in order to capture all signal information of the leakage current generated by the fault, the present invention designs a dedicated multi-band leakage current data acquisition unit, which can simultaneously and synchronously acquire multi-band currents.

[0111] Specifically, the multi-band leakage current mining cable fault information collection device includes an equipment bracket 1, a grounding wire guide channel 2, a multi-band leakage current data collection unit for synchronously collecting multi-band currents, and a collection data processing unit 3.

[0112] The equipment support 1 is a rectangular parallelepiped structure. The grounding wire guide channel 2 is arranged on the equipment support 1 and has two ports for the mine cable grounding wire to pass through. The grounding wire guide channel 2 can be built into the equipment support 1, for example.

[0113] One end of the grounding wire guide channel 2 passes through the upper surface of the equipment bracket 1 (i.e., one end is open facing upward), and the other end (for example, can be defined as the front end) extends to the front end surface of the equipment bracket 1 (one end is open ahead). Therefore, the grounding wire guide channel 2 can be designed as a tubular structure with a corner (i.e., it is composed of two straight pipes and a curved pipe at one end) to facilitate the lead-out of the grounding wire.

[0114] The data acquisition processing unit 3 is also installed on the equipment bracket 1 , and is, for example, arranged at the rear end bottom of the equipment bracket 1 .

[0115] The multi-band leakage current data acquisition unit comprises a low-frequency leakage current acquisition unit 4 and a high-frequency leakage current acquisition unit 5 .

[0116] The low-frequency leakage current collection unit 4 and the high-frequency leakage current collection unit 5 are both arranged on the equipment bracket 1 , and the low-frequency leakage current collection unit 4 is located above the high-frequency leakage current collection unit 5 and the grounding wire guide channel 2 .

[0117] The multi-band leakage current mining cable fault information collection device is set at the end of the mining cable, such as Figure 3 As shown; wherein, the annular collection part of the low-frequency leakage current collection unit 4 is arranged around the mining cable 6.

[0118] The frequency response of the low-frequency leakage current acquisition unit 4 is 30-1500 Hz, and it is mainly used for collecting information of power frequency and low-order harmonics.

[0119] The grounding wire 7 led out from the end of the mining cable first passes back in the opposite direction through the gap formed between the mining cable 6 and the annular collection part of the low-frequency leakage current collection unit 4, then leads out through the grounding wire guide channel 2 and is grounded.

[0120] The reverse direction here refers to the direction from the end of the mining cable where the multi-band leakage current mining cable fault information collection device is located to the other opposite end of the mining cable (that is, the direction from the front end of the equipment bracket 1 to the rear end).

[0121] The annular collection part of the high-frequency leakage current collection unit 5 is arranged around the grounding wire guide channel 2 .

[0122] The low-frequency leakage current acquisition unit 4 and the high-frequency leakage current acquisition unit 5 are both connected to the acquisition data processing unit 3 and are used to send the corresponding low-frequency leakage current signal and high-frequency leakage current signal to the acquisition data processing unit 3 respectively.

[0123] The data collection processing unit 3 of the multi-band leakage current mining cable fault information collection device is connected to the host computer.

[0124] like Figure 2 As shown, the low-frequency leakage current collection unit is designed as a clamp-shaped structure, that is, the upper annular collection part is designed as an openable structure, and the lower part of the low-frequency leakage current collection unit is designed with a clamp-shaped handle 8 for controlling the opening and closing of the openable structure.

[0125] Through the clamp-shaped structure design, the low-frequency leakage current collection unit 4 can be directly clamped on the mining cable 6 without removing the mining cable from the system for installation, thereby simplifying the installation method of the low-frequency leakage current collection unit 4.

[0126] like Figure 2 As shown, the annular collection part of the low-frequency leakage current collection unit 4 includes a first housing 9, a first coil 10, a first magnetic core 11 and a first output circuit 12. The first housing 9 is made of insulating material, and the first magnetic core 11 is located inside the first housing.

[0127] The first shell 9 and the first magnetic core 11 have annular overall contours. To match the clamp structure of the low-frequency leakage current acquisition unit, the first shell 9 and the first magnetic core 11 are both designed with a semicircular ring structure, and are assembled to form an annular structure.

[0128] The first magnetic core 11 is made of ferrite, silicon steel sheet or alloy material with good low-frequency response effect.

[0129] A first coil 10 is wound around the first magnetic core 11 ; wherein the first coil is connected to a first output circuit 12 ; the first output circuit is used to output the induced low-frequency leakage current in a regulated voltage and send it to the acquisition data processing unit 3 through a signal line.

[0130] The process of collecting low-frequency leakage current by the low-frequency leakage current collection unit is as follows: the first magnetic core 11 inserts the entire mining cable 6, and the three phases A, B, and C of the mining cable and the ground wire are regarded as four lines. The currents in each line, namely the three-phase current of the cable and the ground wire current, all induce an alternating magnetic field on the magnetic core through the principle of electromagnetic induction.

[0131] Since the grounding wire 7 is reversed during installation, the magnetic fields generated by the currents of the grounding wire 7 cancel each other out, so that the first magnetic ring 11 only has the induced magnetic field of the three-phase current, that is, the magnetic field induced by the algebraic sum of the three-phase current.

[0132] Under the action of the alternating magnetic field, the first coil 10 wound on the first magnetic ring induces a low-frequency current; the first coil is connected to the first output circuit, and the low-frequency current is sent to the acquisition data processing unit 3 after being stabilized by the first output circuit.

[0133] The frequency response of the high-frequency leakage current acquisition unit 5 is 0.5kHz~50kHz. The high-frequency leakage current acquisition unit 5 is mainly used for medium and high frequency response. Its structural form is a rectangular parallelepiped with a through hole for the grounding wire guide channel 2 arranged inside.

[0134] The annular collection part of the high-frequency leakage current collection unit includes a second shell 13, a second coil 14, a second magnetic core 15 and a second output circuit 16; wherein the second shell is made of insulating material, and the second magnetic core 15 is located inside the second shell.

[0135] The second housing 13 and the second magnetic core 15 are both annular in overall outline.

[0136] The second magnetic core 15 is made of, for example, Permalloy which has a good high frequency response effect.

[0137] A second coil 14 is wound around the second magnetic core 15 ; the second coil 14 is connected to a second output circuit 16 ; the second output circuit is used to output the induced high-frequency leakage current in a regulated voltage and send it to the acquisition data processing unit 3 through a signal line.

[0138] The process of collecting high-frequency leakage current by the high-frequency leakage current collection unit is as follows: the second magnetic core 15 inserts the ground wire 7, and the ground wire current induces an alternating magnetic field on the magnetic core through the principle of electromagnetic induction. Since the magnetic ring is sensitive to high-frequency signals; under the action of the alternating magnetic field, the second coil wound on the second magnetic ring induces a high-frequency current.

[0139] The second coil is connected to the output circuit, and the high-frequency current is sent to the acquisition data processing unit 3 after being stabilized by the second output circuit.

[0140] Through the low-frequency leakage current acquisition unit 4 and the high-frequency leakage current acquisition unit 5, low-frequency leakage current signals and medium- and high-frequency leakage current signals can be acquired at the cable end at the same time without using multiple sensors and acquisition devices.

[0141] The data acquisition processing unit 3 is mainly composed of a housing, a bracket, a main control board, a power port, an output terminal and an antenna. The main control board contains dual-channel input terminals, a conditioning circuit, an analog-to-digital conversion chip, a CPU, an output terminal, a storage chip, and a communication circuit, and has dual-channel synchronous acquisition input, Beidou synchronous timing, data preprocessing, data storage and output functions.

[0142] There is an antenna at the rear of the shell, which is connected to the output port of the main control board for satellite synchronization and data 4G transmission. The output terminal is connected to the output port of the main control board, generally using a network cable or optical fiber interface, for data transmission with the host computer.

[0143] The function realization process of the data acquisition processing unit is as follows: the high-frequency leakage current signal and the low-frequency leakage current signal collected and output by the multi-band leakage current data acquisition unit flow into the dual-channel synchronous acquisition input terminal of the main control board through the signal line. The high-frequency and low-frequency signals pass through two channels respectively, and are conditioned and filtered by the conditioning circuit and then enter the dual-channel analog-to-digital conversion chip of the main control board. The two signals are synchronously converted into digital quantities and enter the memory, and are sent to the communication circuit by the main control board CPU and transmitted to the host computer.

[0144] In addition, the multi-band leakage current mining cable fault information collection device further includes an insulating fixing bracket 17, wherein the insulating fixing bracket is arranged on the equipment bracket 1, and an adjustable cable tie is used to realize installation and fixation with the mining cable.

[0145] The insulating fixed bracket 17 and the low-frequency leakage current collection unit 4 are located at opposite sides of the equipment bracket 1 , respectively. The low-frequency leakage current collection unit 4 is located at the front side of the equipment bracket 1 , and the insulating fixed bracket 17 is located at the rear side of the equipment bracket 1 .

[0146] The present invention can fix the collection device on the mining cable 6 through the insulating fixing bracket 17. Figure 3 The installation structure of the multi-band leakage current mining cable fault information collection device is shown.

[0147] Generally, for each mining cable line, a double-end installation method is adopted, that is, a collection device is installed at each end of the mining cable 6, and the two devices collect current signals synchronously. The installation method is similar to that of an ordinary zero-sequence current transformer, specifically: 1. Through the low-frequency leakage current collection unit 4 with a clamp-shaped structure, the collection part of the low-frequency leakage current collection unit 4 is clamped into the mining cable, and at the same time, the ground wire of the mining cable is reversed back to the collection part of the low-frequency leakage current collection unit 4.

[0148] The purpose of the low-frequency acquisition part is to collect the algebraic sum of the three-phase current. When the clamp sensor is clamped on the cable, it includes the ground wire current in addition to the three-phase current. Therefore, the ground wire needs to be led out at the end of the cable and reversely passed back to the low-frequency acquisition part. In this way, the positive and negative ground wire currents can be offset, ensuring that the low-frequency acquisition part only collects the algebraic sum of the three-phase current.

[0149] 2. Insert the ground wire into the ground wire routing channel of the device, pass it out from the front end, and then ground the ground wire.

[0150] In order to obtain high-frequency characteristics, the high-frequency collection part collects the high-frequency leakage current of the ground wire, so it needs to be collected separately. Therefore, the present invention passes the ground wire 7 through the collection magnetic core (ie, the second magnetic core 15) and then grounded.

[0151] 3. The adjustable rolled band 18 of the insulating fixing bracket 17 of the device is wrapped around the mining cable 6 and fixed to fix the device.

[0152] When in use, the multi-band leakage current mining cable fault information collection device is powered on and started, and the device's main control board has a built-in program to automatically realize functions such as synchronous timing, synchronous data collection, and data transmission.

[0153] After being collected by the multi-band leakage current mining cable fault information collection device, two synchronously collected signals are obtained, namely the high-frequency leakage current signal and the low-frequency leakage current signal. The two signals can generally be used as two signal inputs for feature mining and diagnosis in the following fault diagnosis. Of course, the data fusion principle can also be used to fuse the two signals.

[0154] There are at least two ways to fuse the high-frequency leakage current signal and the low-frequency leakage current signal, namely cutoff frequency fusion and weighted fusion according to the different response characteristics of the magnetic core to each frequency band; from the at least two fusion methods, one can be selected as the fusion method for the high-frequency leakage current signal and the low-frequency leakage current signal.

[0155] Specifically, the process of cutoff frequency fusion is as follows: because the response frequency of the above-mentioned low-frequency signal acquisition module is 30~1500Hz, the higher the frequency in the signal spectrum, the more distorted it is. The response frequency range of the above-mentioned medium and high frequency signal acquisition module is 0.5kHz~50kHz, and the lower the frequency in the signal spectrum, the more distorted it is.

[0156] Select 1k as the cutoff frequency, and only extract the low-frequency acquisition module to collect the signal below 1kHz (inclusive) and define it as x ’ (t), and extract the part x''(t) above 1kHz of the signal collected by the high-frequency acquisition module, and fuse the signal x(t) = x ’ (t)+ x''(t).

[0157] The process of weighted fusion according to the different response characteristics of the magnetic core to each frequency band is as follows: suppose the signal of a certain frequency band of the real signal is f(t), the signals collected by the low-frequency acquisition module and the high-frequency acquisition module in this frequency band are f'(t) and f''(t) respectively; since the high-frequency module and the low-frequency module have different response characteristics to the signal in this frequency band, weights α and β can be assigned respectively according to their characteristics; α+β=1; then the collected signal of this frequency band f(t)=α*f'(t)+β*f''(t).

[0158] Of course, the above two fusion methods are only examples, and you can choose the fusion method according to your needs.

[0159] The method of the present invention uses the high-frequency leakage current signal and the low-frequency leakage current signal as two homologous data of the same original signal, and extracts their respective features as the input of the diagnostic model, which is equivalent to the same signal source. Since the present invention simultaneously obtains two groups of parameter samples, each group of samples can be used for feature mining and diagnosis. Therefore, as the number of training and diagnostic samples increases, the accuracy of feature mining is improved. This method enriches the fault data input samples and improves the accuracy of model feature mining.

[0160] The multi-band leakage current mining cable fault information acquisition device of the present invention can separately collect low-frequency leakage current and high-frequency grounding wire leakage current in mining cables, realize multi-band leakage current acquisition, and solve the technical problems that the existing zero-sequence current sensor is insensitive to high-frequency response and the high-frequency sensor is insensitive to low-frequency current, and there is no dedicated multi-band leakage current acquisition device.

[0161] Of course, the above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments. It should be noted that all equivalent substitutions and obvious deformation forms made by any technician familiar with the field under the guidance of this specification fall within the essential scope of this specification and should be protected by the present invention.

Claims

1. A mining cable fault diagnosis method based on multi-band leakage current characteristics, characterized in that: The steps include: Step 1. Pre-process the acquired multi-band leakage current signal of the mining cable to reduce noise; wherein the multi-band leakage current signal of the mining cable is obtained by combining the high-frequency leakage current signal and the low-frequency leakage current signal collected synchronously; The high-frequency leakage current signal and the low-frequency leakage current signal are two homologous data of the same original signal; Step 2. Extract features of the multi-band leakage current signal of the mining cable preprocessed in step 1, and annotate the extracted multi-band leakage current features of the mining cable to obtain a training data set; Step 3. Build a mining cable fault diagnosis model based on a convolutional neural network, and train the mining cable fault diagnosis model based on the training data set obtained in step 2 to obtain a trained mining cable fault diagnosis model; The multi-band leakage current signal of the mining cable to be diagnosed is subjected to noise reduction preprocessing according to step 1, and after feature extraction in step 2, it is input into the trained mining cable fault diagnosis model to obtain the mining cable fault diagnosis result.

2. The mining cable fault diagnosis method based on multi-band leakage current characteristics according to claim 1 is characterized in that: In the step 1, an improved microwave threshold denoising algorithm is used to implement denoising preprocessing of the multi-band leakage current signal; The specific process is as follows: Step 1.

1. Decomposition of original signal; First, the signal x(t) is decomposed into high-frequency coefficients representing high-frequency components and low-frequency coefficients representing low-frequency components; (1) Where x(t) represents the obtained multi-band leakage current signal of mining cable; W j,k is the high-frequency coefficient of the jth layer and position k, representing the high-frequency part of the signal x(t); V J,k is the low-frequency coefficient at the Jth layer and position k, representing the low-frequency part of the signal x(t); is the basis function, is the magnitude function; Step 1.

2. Flexible dynamic threshold denoising; The high-frequency coefficients decomposed in step 1.1 above are The following processing is performed, as shown in formulas (2) and (3); (2) (3) Formula (2) and Formula (3) are the denoising threshold formula and the optimization threshold formula respectively, and λ is the threshold parameter; Next, a flexible dynamic denoising threshold function is proposed, and the formula is as follows: (4) (5) In the formula, x represents the high frequency coefficient , T(x) is the denoising threshold output, is the dynamic threshold value set. is a positive number used to avoid division by zero, N represents the selected signal length, and a represents the number of signal data points; Find Then, replace it with formula (2) as the high-frequency coefficient of signal reconstruction; Step 1.

3. Signal reconstruction; The processed decomposition coefficients are used to reconstruct the signal to obtain the denoised signal. The specific method is as follows: if the noise amplitude in the signal is large and a pure filtered signal is required, the optimized threshold coefficient obtained by formula (3) is used; if the noise amplitude in the signal is low and the details in the signal need to be retained, the flexible dynamic denoising threshold coefficient obtained by formula (4) is used. After the flexible dynamic threshold value denoising obtained by formula (4) or the optimized threshold value denoising by formula (3), the new coefficients are substituted into formula (1), and the inverse operation of formula (1) is performed to obtain the denoised signal S(i); (6)。 3. The mining cable fault diagnosis method based on multi-band leakage current characteristics according to claim 1 is characterized in that: In step 2, the extracted multi-band leakage current characteristics of mining cables include empirical characteristics, main frequency characteristics, odd-even ratio, harmonic ratio, spectrum center of gravity, total harmonic distortion, instantaneous frequency variation and energy entropy corresponding to various faults.

4. The mining cable fault diagnosis method based on multi-band leakage current characteristics according to claim 1 is characterized in that: In step 3, the mining cable fault diagnosis result, i.e., the output of the mining cable fault diagnosis model, is divided into five types, i.e., no fault, single-phase grounding fault, high-resistance single-phase grounding fault, partial discharge, and two-phase short circuit.

5. The mining cable fault diagnosis method based on multi-band leakage current characteristics according to claim 1 is characterized in that: In step 3, the training process of the mining cable fault diagnosis model is as follows: All feature data of the multi-band leakage current features of mining cables extracted in step 2 are used as input and input into the convolutional neural network. The convolutional neural network mines the corresponding relationship between each input feature and the deep feature of the fault according to the set fault type; By continuously adjusting the internal feature parameters of the convolutional neural network, the mining cable fault diagnosis model training is completed.

6. A mining cable fault diagnosis system based on multi-band leakage current characteristics, characterized in that: It includes a multi-band leakage current mining cable fault information collection device and a host computer; The multi-band leakage current mining cable fault information acquisition device is used to collect the multi-band leakage current signal of the mining cable, and transmit the collected multi-band leakage current signal of the mining cable to the host computer; The host computer includes a memory and one or more processors; Wherein, executable codes are stored in the memory; when the processor executes the executable codes, the mining cable fault diagnosis method based on multi-band leakage current characteristics as described in any one of claims 1 to 5 above is implemented.

7. The mining cable fault diagnosis system based on multi-band leakage current characteristics according to claim 6 is characterized in that: The multi-band leakage current mining cable fault information acquisition device comprises an equipment bracket, a grounding wire guide channel, a multi-band leakage current data acquisition unit for synchronously acquiring multi-band currents, and an acquisition data processing unit; The grounding wire guide channel is arranged on the equipment support and has two ports for the mine cable grounding wire to pass through; The acquisition data processing unit is also installed on the equipment bracket; The multi-band leakage current data acquisition unit includes a low-frequency leakage current acquisition unit and a high-frequency leakage current acquisition unit; Wherein, the low-frequency leakage current collection unit and the high-frequency leakage current collection unit are both arranged on the equipment bracket, and the low-frequency leakage current collection unit is located above the high-frequency leakage current collection unit and the grounding wire guide channel; The multi-band leakage current mining cable fault information collection device is arranged at the end of the mining cable; wherein the annular collection part of the low-frequency leakage current collection unit is arranged around the mining cable; The grounding wire led out from the end of the mining cable first passes back in the opposite direction through the gap formed between the mining cable and the annular collection part of the low-frequency leakage current collection unit, then leads out through the grounding wire guide channel and is grounded; The reverse direction here refers to the direction from the end of the mining cable where the multi-band leakage current mining cable fault information collection device is located to the other opposite end of the mining cable; The annular collection part of the high-frequency leakage current collection unit is arranged around the grounding wire guide channel; The low-frequency leakage current acquisition unit and the high-frequency leakage current acquisition unit are both connected to the acquisition data processing unit, and are used to send the corresponding low-frequency leakage current signal and high-frequency leakage current signal to the acquisition data processing unit respectively; The data collection processing unit of the multi-band leakage current mining cable fault information collection device is connected to the host computer.

8. The mining cable fault diagnosis system based on multi-band leakage current characteristics according to claim 7 is characterized in that: The low-frequency leakage current collection unit is designed as a clamp-shaped structure, that is, the annular collection part on the upper part is designed as an openable structure, and the lower part of the low-frequency leakage current collection unit is designed with a clamp-shaped handle for controlling the opening and closing of the openable structure; The multi-band leakage current mining cable fault information collection device also includes an insulating fixing bracket, wherein the insulating fixing bracket is arranged on the equipment bracket and an adjustable tie is used to realize installation and fixing with the mining cable; Wherein, the insulating fixing bracket and the low-frequency leakage current collection unit are respectively located on opposite sides of the equipment bracket.

9. The mining cable fault diagnosis system based on multi-band leakage current characteristics according to claim 7 is characterized in that: The annular collection part of the low-frequency leakage current collection unit includes a first shell, a first coil, a first magnetic core and a first output circuit; wherein the first shell is made of insulating material, and the first magnetic core is located in the first shell; The first shell and the first magnetic core are both annular in overall outline; The first magnetic core is made of ferrite, silicon steel sheet or alloy material with good low-frequency response effect; A first coil is wound on the first magnetic core; wherein the first coil is connected to a first output circuit; the first output circuit is used to output the induced low-frequency leakage current in a regulated voltage and send it to a data acquisition processing unit through a signal line; The process of low-frequency leakage current acquisition unit collecting low-frequency leakage current is as follows: The first magnetic core is used to sheath the entire mining cable. The three phases A, B, C and the ground wire of the mining cable are regarded as four lines. The currents in each line, i.e. the three-phase current of the cable and the ground wire current, induce an alternating magnetic field on the magnetic core through the principle of electromagnetic induction. Since the grounding wire is reversed during installation, the magnetic fields generated by the grounding wire currents cancel each other out. Therefore, there is only the induced magnetic field of the three-phase currents in the first magnetic ring, that is, the magnetic field induced by the algebraic sum of the three-phase currents. Under the action of the alternating magnetic field, the first coil wound on the first magnetic ring induces a low-frequency current; wherein the first coil is connected to the first output circuit, and the low-frequency current is sent to the acquisition data processing unit after being stabilized by the first output circuit.

10. The mining cable fault diagnosis system based on multi-band leakage current characteristics according to claim 7, characterized in that: The annular collection part of the high-frequency leakage current collection unit includes a second shell, a second coil, a second magnetic core and a second output circuit; wherein the second shell is made of insulating material, and the second magnetic core is located in the second shell; The overall contours of the second shell and the second magnetic core are both ring-shaped; The second magnetic core is made of Permalloy material which has good high frequency response effect; A second coil is wound on the second magnetic core; wherein the second coil is connected to a second output circuit; the second output circuit is used to output the induced high-frequency leakage current in a regulated voltage and send it to the acquisition data processing unit through a signal line; The process of high-frequency leakage current acquisition unit collecting high-frequency leakage current is as follows: The second magnetic core inserts the ground wire, and the ground wire current induces an alternating magnetic field on the magnetic core through the principle of electromagnetic induction. Since the magnetic ring is sensitive to high-frequency signals, the second coil wound on the second magnetic ring induces a high-frequency current under the action of the alternating magnetic field. The second coil is connected to the output circuit, and the high-frequency current is sent to the acquisition data processing unit after being stabilized by the second output circuit.

Images