Cable insulation defect detection method and related equipment

Abstract

The invention relates to the technical field of power systems, and discloses a cable insulation defect detection method and related equipment, and the method comprises the steps: collecting a signal, and compressing the collected signal to obtain a compressed sensing sample; generating an excitation signal according to compressed sensing sampling, and converting the excitation signal into an analog signal; inputting the analog signal into a coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compression signals; cable fault types are defined according to the plurality of sample compression signals, and a coaxial cable database is created according to each cable fault type; and adding random Gaussian noise into the database of the coaxial cable, and carrying out cable insulation defect detection according to the signal-to-noise ratio. According to the method, the signal is collected and compressed to obtain the compressed sensing sample, so that the sampling rate and the storage requirement of the data can be remarkably reduced, and meanwhile, key information is reserved, so that the data processing efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a cable insulation defect detection method and related equipment. Background Art

[0002] Reliability, quality, economy and environmental protection are the basic requirements for the operation of modern power systems. Among them, it is very important to ensure the reliable and continuous power supply of the power system. The stable and reliable operation of the power system is an important aspect of the national economy and people's livelihood. Once a large-scale power outage occurs in the power system, the development of the national economy and even social public safety will be seriously affected. In order to ensure the stability of the power system operation, it is particularly important to regularly detect faults in power equipment.

[0003] Power equipment, such as transformers, circuit breakers, insulators, etc., rely on the integrity of their insulation systems to ensure safe and reliable operation. The main function of the insulation system is to prevent current leakage and avoid short circuits between electrical equipment. Due to the long-term operation of power equipment and the influence of environmental factors, the insulating materials in the insulation system may be damaged by aging, pollution and mechanical stress, which may cause insulation defects. Electrical insulation defects refer to the situation in which the service life of cables in the insulation system of electrical equipment is greatly reduced due to mechanical damage, insulation moisture, overvoltage, overcurrent, corrosion of the protective layer, etc., and even cause electrical short circuits and fires. This situation seriously affects the property safety of various enterprises. Therefore, the detection and analysis of electrical insulation defects are becoming increasingly important in the maintenance and management of power systems.

[0004] Cable insulation defect detection can provide early warning before the insulation system of the equipment deteriorates significantly. By monitoring and analyzing defect signals, operation and maintenance personnel can evaluate the health of the insulation system, so as to take preventive maintenance measures to extend the service life of the equipment and avoid sudden failures. Regular cable insulation defect detection and maintenance can significantly reduce the equipment failure rate, reduce unexpected downtime and repair costs, and thus achieve better economic benefits. Especially in large-scale power systems, reducing equipment failures is particularly important for the entire power network. Through effective electrical insulation defect detection, the safety, reliability and economy of equipment operation can be improved, and the impact of equipment failures on the power system can be prevented. With the continuous advancement of technology, electrical insulation defect detection will become an increasingly important means to ensure the stable operation of power systems.

[0005] However, reflection is a common technique used to monitor the health of cables, including time domain reflectometry (TDR) and frequency domain reflectometry. Traditional reflection has some limitations in cable fault detection, location and characterization. For example, although the compressed sensing paradigm has been introduced into the reflection diagnosis system to improve the accuracy of fault location and characterization, compressed sensing requires a signal reconstruction step, which requires an iterative algorithm and is computationally time-consuming and energy-consuming. At the same time, when using deep learning to analyze reflection signals, most of the existing methods are targeted at specific scenarios and signal processing methods, lacking a general and efficient cable fault diagnosis method. Some studies use machine learning models such as convolutional neural networks (CNN) and K nearest neighbor algorithms (KNN) to analyze reflection signals, but when processing compressed signals, most of them need to reconstruct the signal first, which increases the computational complexity and time cost. For different types of faults and cables, the adaptability and accuracy of existing methods need to be improved. Summary of the invention

[0006] In order to overcome the defects of the above-mentioned prior art, the purpose of the present invention is to provide a cable insulation defect detection method and related equipment to solve the technical problems of long calculation time and high power consumption in the prior art of cable insulation defect detection.

[0007] The present invention is achieved through the following technical solutions: In a first aspect, the present invention provides a method for detecting cable insulation defects, comprising: Collecting signals, and compressing the collected signals to obtain compressed sensing sampling; Generate an excitation signal according to compressed sensing sampling, and convert the excitation signal into an analog signal; Inputting an analog signal into a coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compression signals; defining cable fault types according to a number of sample compressed signals, and creating a database of coaxial cables according to each cable fault type; Random Gaussian noise is added to the coaxial cable database, and cable insulation defects are detected based on the signal-to-noise ratio.

[0008] Preferably, in the step of collecting signals and compressing the collected signals to obtain compressed sensing samples, the specific process is as follows: The relationship between the collected signal and the observation vector is determined, and compressed sampling is performed through the AIC analog-to-digital converter, in which the standard ADC analog-to-digital converter is replaced by the AIC analog-to-digital converter, and the mixer in the AIC analog-to-digital converter modulates the signal with a pseudo-random sequence, which is filtered by the filter and converted by the ADC analog-to-digital converter.

[0009] Preferably, in the step of inputting the analog signal into the coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compressed signals, the obtained feedback signal is compressed by a random demodulator at a frequency of 200 MHz with a compression factor of 2 to obtain a plurality of sample compressed signals.

[0010] Preferably, in the step of defining cable fault types according to a plurality of sample compression signals and creating a database of coaxial cables according to each cable fault type, the cable fault types include short circuit, open circuit and soft fault, and reflection coefficient modeling is performed according to each cable fault type, wherein the reflection coefficient model is as follows: Γd = Zd−Zc / Zd + Zc Where Zd is the fault impedance; Zc is the cable characteristic impedance, which is a constant value; Γd is the reflection coefficient, and the value of Γd depends directly on Zd; When Γd = −1, the fault type corresponds to short circuit, Zd = 0 Ω; When Γd = 1, the fault type corresponds to open circuit, Zd = +∞Ω; When Γd < 1, the fault type corresponds to a soft fault; When Γd = 0, there is no fault and Zd = 58.00 Ω.

[0011] Preferably, random Gaussian noise is added to the database of the coaxial cable, and in the step of performing cable insulation defect detection according to the signal-to-noise ratio, cable insulation defect detection is performed according to the signal-to-noise ratio, wherein when the signal-to-noise ratio ranges from 0 to 15 dB, the KNN algorithm is used for cable insulation defect detection; when the signal-to-noise ratio ranges from 15 to 60 dB, the CNN model is used for cable insulation defect detection.

[0012] Furthermore, when the KNN algorithm is used for cable insulation defect detection, fault impedance estimation and fault location prediction are performed respectively, and the fault impedance estimation and fault location prediction respectively determine the optimal parameters through GridSearchCV; Among them, the optimal hyperparameters are determined by using GridSearchCV through fork validation in fault impedance estimation, including the number of neighbors as 1, weight as uniform, algorithm as ball tree, leaf size as 10, and metric methods as euclidean and minkowski.

[0013] Furthermore, when the CNN model is used for cable insulation defect detection, fault impedance estimation and fault location prediction are performed respectively; The fault impedance estimation and fault location prediction are trained on 80% of the generated data, 10% is used for validation, and 10% is used for testing. Leave-one-out cross validation is used during training, and the Adam optimizer is used with MAE as the loss function. During testing, the optimal parameters are evaluated in 10% of the database to obtain the optimal parameters.

[0014] In a second aspect, the present invention further provides a cable insulation defect detection system, comprising: A signal compression module is used to collect signals and compress the collected signals to obtain compressed sensing samples; A signal conversion module, used for generating an excitation signal according to compressed sensing sampling, and converting the excitation signal into an analog signal; A feedback signal compression module, used for inputting an analog signal into a coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compression signals; A database creation module, used to define cable fault types according to a number of sample compression signals, and to create a coaxial cable database according to each cable fault type; The defect detection module is used to add random Gaussian noise to the database of the coaxial cable and perform cable insulation defect detection based on the signal-to-noise ratio.

[0015] In a third aspect, the present invention further provides a mobile terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the cable insulation defect detection method as described above when executing the computer program.

[0016] In a fourth aspect, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the cable insulation defect detection method as described above are implemented.

[0017] Compared with the prior art, the present invention has the following beneficial technical effects: The present invention provides a cable insulation defect detection method. By collecting and compressing signals, compressed sensing sampling is obtained, which can significantly reduce the sampling rate and storage requirements of data, while retaining key information, thereby improving the efficiency of data processing. Compressed sensing technology utilizes the sparsity of signals to reconstruct signals under conditions far lower than traditional sampling rates, which can greatly reduce the burden of data transmission and storage. An excitation signal is generated according to compressed sensing sampling, and converted into an analog signal, so that the signal can adapt to the transmission characteristics of a coaxial cable. The analog signal is input into the coaxial cable and a feedback signal is obtained. Through the collection of the feedback signal, the state of the cable can be monitored in real time, and potential insulation defects can be discovered in time. The cable fault type is defined according to a number of sample compressed signals. A database of coaxial cables is created according to each cable fault type. These databases can store characteristic information of different fault types. Random Gaussian noise is added to the database of the coaxial cable, which can simulate noise interference in an actual environment and improve the robustness and accuracy of insulation defect detection. Cable insulation defect detection is performed according to the signal-to-noise ratio. By calculating the signal-to-noise ratio between the feedback signal and the noise, the health of the cable insulation layer can be evaluated, and potential defects and faults can be discovered in time.

[0018] Furthermore, the present invention provides a cable discontinuity detection, location and characterization method based on compressed Chirp-OMTDR acquisition and machine learning, including predicting fault impedance, enabling fault detection and predicting fault location when a fault is detected. The KNN algorithm and the CNN model are compared to evaluate their performance in impedance estimation and fault location. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Flow chart of a cable insulation defect detection method according to an embodiment of the present invention; Figure 2 Schematic diagram of the architecture of OMTDR measurement compressed data in an embodiment of the present invention; Figure 3 Schematic diagram of compression reflection signal spectrum of different types of faults in an embodiment of the present invention; Figure 4 is a CNN architecture for fault impedance estimation in an embodiment of the present invention; Figure 5 is a CNN architecture used for fault location estimation in an embodiment of the present invention; Figure 6 This is a schematic diagram of the structure of a cable insulation defect detection system in an embodiment of the present invention; In the figure: 1. Signal compression module; 2. Signal conversion module; 3. Feedback signal compression module; 4. Database creation module; 5. Defect detection module. DETAILED DESCRIPTION

[0020] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0021] The object of the present invention is to provide a cable insulation defect detection method and related equipment to solve the technical problems of long calculation time and high power consumption in the prior art of cable insulation defect detection.

[0022] The present invention is further described in detail below in conjunction with the accompanying drawings: Example 1 See also Figure 1 In one embodiment of the present invention, a method for detecting cable insulation defects is provided, comprising: Step 1, collecting signals, and compressing the collected signals to obtain compressed sensing samples; Specifically, the relationship between the collected signal and the observation vector is determined, and compressed sampling is performed through an AIC analog-to-digital converter, wherein the standard ADC analog-to-digital converter is replaced by an AIC analog-to-digital converter, and the mixer in the AIC analog-to-digital converter modulates the signal with a pseudo-random sequence, which is filtered by a filter and converted by the ADC analog-to-digital converter.

[0023] Let X∈RN represent the signal of interest and the observation vector Y∈RM, where M< <N; The sampling operation is modeled according to Y = ϕ x , where φ denotes the measurement matrix ∈RNxM. A signal X is compressible if it can be represented by a basis ψ, such as X = ψs, where s contains a small number of nonzero coefficients.

[0024] Step 2, generating an excitation signal according to compressed sensing sampling, and converting the excitation signal into an analog signal; Step 3, inputting the analog signal into the coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compressed signals; Specifically, the obtained feedback signal is compressed by a random demodulator at a frequency of 200 MHz with a compression factor of 2 to obtain a plurality of sample compressed signals.

[0025] Step 4, defining cable fault types according to a number of sample compression signals, and creating a coaxial cable database according to each cable fault type; Specifically, the cable fault types include short circuit, open circuit and soft fault, and reflection coefficient modeling is performed according to each cable fault type, wherein the reflection coefficient model is as follows: Γd = Zd−Zc / Zd + Zc Where Zd is the fault impedance; Zc is the cable characteristic impedance, which is a constant value; Γd is the reflection coefficient, and the value of Γd depends directly on Zd; When Γd = −1, the fault type corresponds to a short circuit, and Zd = 0 Ω.

[0026] When Γd = 1, the fault type corresponds to an open circuit, and Zd = +∞Ω.

[0027] When Γd = 0, there is no fault, Zd = 58.00 Ω The specifications corresponding to different fault types in this embodiment are shown in Table 1.

[0028]

[0029] Table 1 Specifications corresponding to different fault types Step 5: Add random Gaussian noise to the coaxial cable database and perform cable insulation defect detection based on the signal-to-noise ratio.

[0030] Specifically, the cable insulation defect detection is performed according to the signal-to-noise ratio. When the signal-to-noise ratio is between 0-15dB, the KNN algorithm is used for cable insulation defect detection; when the signal-to-noise ratio is between 15-60dB, the CNN model is used for cable insulation defect detection.

[0031] When the KNN algorithm is used for cable insulation defect detection, fault impedance estimation and fault location prediction are performed respectively, and the optimal parameters of the fault impedance estimation and fault location prediction are determined by GridSearchCV respectively.

[0032] Among them, when using the CNN model for cable insulation defect detection, fault impedance estimation and fault location prediction are performed respectively; The fault impedance estimation and fault location prediction are trained on 80% of the generated data, 10% is used for validation, and 10% is used for testing. Leave-one-out cross validation is used during training, and the Adam optimizer is used with MAE as the loss function. During testing, the optimal parameters are evaluated in 10% of the database to obtain the optimal parameters.

[0033] The impedance conversion formula in this embodiment is as follows:

[0034] Where N represents the minibatch dimension, Yn is the true value, and y_yn is the model prediction value; The custom formula loss function expression of the mean square error equation (MSE) is as follows:

[0035] Where N represents the minibatch dimension, Yn is the real label, y_yn is the model prediction, and wn is the weight associated with the prediction. For example, the description of N is as follows:

[0036] Where λ is a parameter learned by the model during the training phase. By introducing the weighted MSE function wn, the model can penalize high error predictions above 1.0 m.

[0037] This embodiment studies the cases where the compression factors are 4, 8, and 16, generates pseudo-random sequences of corresponding lengths (e.g., compression factor 4 corresponds to 2048 samples, corresponding to 800MHz; compression factor 8 corresponds to 4096 samples, corresponding to 1.6GHz; compression factor 16 corresponds to 8192 samples, corresponding to 3.2GHz), and generates corresponding training databases. The KNN algorithm and CNN model are tested on the test data sets under these new compression factors to evaluate the impact on the fault location accuracy.

[0038] In the test k-nearest neighbor (KNN) is different from CNN, the KNN algorithm does not directly predict the fault impedance through the reflection coefficient. The two methods are equivalent. The KNN algorithm is a non-parametric model that can be used to perform classification or regression tasks. In the case of this work, it is used to perform regression. The performance of the model depends on the number of neighbors K, which must be selected based on the intrinsic characteristics of the training data. Smaller values ​​of K will reduce the bias but increase the variance, making the model sensitive to noise. In contrast, larger values ​​of K will reduce the variance at the cost of higher bias, which will make the prediction too smooth but may lead to a loss in accuracy. Therefore, it is crucial to find an optimal K that balances the bias and variance. In order to find the best hyperparameters for the KNN algorithm, a 10-fold cross validation was performed using GridSearchCV to explore the grid of hyperparameter combinations to determine the best parameterization that maximizes the performance of the KNN algorithm. Table 2 gives the optimal parameterization of KNN obtained.

[0039]

[0040] Table 2 Hyperparameters to be considered in optimization Test results of the performance evaluation model During the testing phase, the optimal parameters were obtained. The optimal model was tested on 10% of the simulated database, i.e., 3960 samples (990 samples for each fault type). The performance obtained during the test is shown in Table 3. Two cases are distinguished: samples with a signal-to-noise ratio ∈ [0,60] dB (the entire test set) and samples with a signal-to-noise ratio ∈ [15,60] dB (the number of samples equals 3016). These results show the impact of noise on the performance of the KNN algorithm. The root mean square error (RMSE) < 1.42 Ω, the mean absolute percentage error (MAPE) < 0.009%, and the mean absolute error (MAE) < 0.50 Ω for the entire test indicate that the impedance estimation performance of the KNN algorithm is satisfactory.

[0041]

[0042] Table 3 Estimation error of KNN test data set 1 In the case of SNR>0 dB, the values ​​of RMSE = 1.27 Ω, MAPE = 0.0064%, and MAE = 0.37 Ω obtained on the test set show that CNN has excellent performance in predicting impedance. In the case of SNR>15 dB, the accuracy of the model is higher than that of the KNN algorithm and is very close to the results of the KNN algorithm, as shown in Table 4.

[0043]

[0044] Table 4 Estimation error of CNN test dataset2 In the case of SNR > 15 dB, the results of CNN far exceed those of KNN algorithm using MAE and MAPE, both of which have more than twice the SNR of the former; in terms of soft fault accuracy, CNN's MAE is 0.79 Ω instead of 1.4931 Ω.

[0045] But in the fault-free case, the KNN algorithm is more accurate than the CNN model.

[0046] For fault location prediction, only samples with faults are considered. It is assumed that fault detection is done through impedance prediction. Considering three fault types: short circuit, open circuit, and soft fault, there is a database of 29,700 samples (9,900 for each fault type). The training database accounts for 80%, the validation database accounts for 10%, and the test database accounts for 10%. These three databases are the same for both machine learning models. For the location prediction of the KNN algorithm, the same process as the impedance prediction is followed. GridSearchCV is used to determine the best parameterization to maximize the performance of the KNN algorithm in fault location prediction. The super-optimal parameters obtained through this optimization are shown in Table 5.

[0047]

[0048] Table 5 KNN optimal hyperparameters Test results of the performance evaluation model The optimal model obtained in the training phase was tested on 2970 samples, that is, 990 samples for each fault type. The performance obtained on the test data set according to the signal-to-noise ratio ∈ [0,60]dB (the entire test set) and the signal-to-noise ratio ∈ [15,60]dB (the number of samples is equal to 2032) is shown in Table 6.

[0049]

[0050] Table 6 Estimation error of KNN test data set 3 When the signal-to-noise ratio is > 0 dB, the RMSE, MAPE, and MEA values ​​show that the KNN algorithm has good fault location prediction performance, with an RMSE of 1.34 cm and a MAE of 5.00 mm. The KNN algorithm is more accurate in a less noisy environment, with an RMSE of 3.30 mm and a MAE of 2.30 mm, which is twice lower than the noisy sample case.

[0051] The convolutional neural network CNN is trained on 80% of the simulated data and robustness is evaluated using the leave-one-out cross-validation technique. The performance of CNN on the test dataset is shown in Table 7.

[0052]

[0053] Table 7 Estimation error of CNN test dataset 4 For the case of signal-to-noise ratio ∈ [0,60] dB, the RMSE is 15.06 cm and the MAE is 2.30 cm, which are lower than the results obtained using the KNN algorithm.

[0054] For low-noise samples with SNR ∈ [15,60]dB, RMSE = 1.21 cm is reduced by 12 times, and MAE = 9.3 mm is reduced by 2 times compared to the case where the SNR is in the range [0,60]dB. This shows that CNN is more sensitive to noise than KNN. KNN algorithm significantly outperforms CNN, achieving prediction accuracy within a few millimeters in both SNR cases.

[0055] Influence of compression factor on positioning accuracy in tests In this work, it is proposed to increase the compression factor by increasing the frequency of the reflected signal to improve the spatial resolution. This is achieved by multiplying the reflected signal by a pseudo-random sequence operating at a frequency higher than the Nyquist frequency by a factor equal to the value of the compression factor. For each value of the compression factor, a different pseudo-random sequence is generated, whose length is proportional to the value of the compression factor. For example, in the case of a compression factor of 2, a pseudo-random sequence with 1024 samples is used. Three new compression factors are studied: 4, 8, and 16. For these three compression factors, pseudo-random sequences with lengths of 2048 (corresponding to 800 MHz), 4096 (corresponding to 1.6 GHz), and 8192 (corresponding to 3.2 GHz) are generated, respectively. A training database is also generated for each compression factor. Finally, the optimal machine learning model for a compression factor of 2 is trained in the case of these new compression factors. Tests are performed on data with compression factors of 4, 8, and 16, and the results are shown in Tables 8 and 9.

[0056]

[0057] Table 8 Estimation error of KNN test data set 5

[0058] Table 9 Estimation error of CNN test dataset 6 In the case of SNR>0 dB, increasing the compression factor can slightly improve the results. For example, increasing the compression factor from 2 to 16 reduces the RMSE value from 15.06 cm to 9.98 cm and the MAE value from 2.30 cm to 2.05 cm. However, when the SNR is >15 dB, increasing the compression factor does not provide more accurate prediction results. The MAE value of the compression factor is close to 9 mm on average, and when the compression factor is 8, the model accuracy is slightly higher, with an RMSE of 1.06 cm and a MAE of 8.20 mm.

[0059] The results show that both evaluated machine learning models, the KNN algorithm and the CNN, provide satisfactory performance. However, it must be emphasized that the two approaches have different advantages and challenges. The final choice will mainly depend on the specific requirements of the target application, especially in terms of available computational and memory resources, but also the accuracy and noise level required to characterize such tasks. In this case, the CNN is more effective for fault impedance estimation, while the KNN is more effective for fault location estimation. The KNN algorithm has the advantage of being relatively computationally inefficient during the model optimization phase. However, its main disadvantage lies in the need to store the entire training dataset, which quickly becomes an issue in terms of memory consumption. In contrast, the KNN requires a lot of computing power during the training process to derive the optimal model. However, once trained, the model can be deployed immediately to make new predictions without access to the initial training data. In addition, the trained CNN can serve as the basis for transfer learning, which can effectively adapt to similar but different configurations. On the other hand, in order to handle a completely new scenario, it is necessary to retrain the model from scratch using an appropriate dataset.

[0060] In summary, this embodiment proposes a cable insulation defect detection method, a cable discontinuity point detection, location and characterization method based on compressed Chirp-OMTDR acquisition and machine learning. The method includes predicting fault impedance, enabling fault detection and predicting fault location when a fault is detected. In this work, two machine learning models, the KNN algorithm and the CNN model, were compared to evaluate their performance in impedance estimation and fault location. Both models have high accuracy in impedance estimation and fault location. Therefore, the developed fault diagnosis method can be used for any type of fault (hard or soft) and any type of cable. In addition, the study also showed interesting results in terms of compression factor: increasing the compression factor can improve the fault location accuracy, especially for the KNN algorithm, better performance can be obtained. This study shows that faults can be detected, located and characterized from compressed Chirp-OMTDR data without reconstruction.

[0061] Example 2 according to Figure 6 As shown, this embodiment provides a cable insulation defect detection system, including: The signal compression module 1 is used to collect signals and compress the collected signals to obtain compressed sensing samples; A signal conversion module 2, used to generate an excitation signal according to compressed sensing sampling, and convert the excitation signal into an analog signal; A feedback signal compression module 3 is used to input the analog signal into the coaxial cable to obtain a feedback signal, and compress the feedback signal to obtain a plurality of sample compression signals; A database creation module 4 is used to define the cable fault type according to a number of sample compression signals, and to create a coaxial cable database according to each cable fault type; The defect detection module 5 is used to add random Gaussian noise into the database of the coaxial cable and perform cable insulation defect detection according to the signal-to-noise ratio.

[0062] Example 3 The present invention also provides a mobile terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, such as a cable insulation defect detection program.

[0063] When the processor executes the computer program, the steps of the above-mentioned cable insulation defect detection method are implemented, for example: Collecting signals, and compressing the collected signals to obtain compressed sensing sampling; Generate an excitation signal according to compressed sensing sampling, and convert the excitation signal into an analog signal; Inputting an analog signal into a coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compression signals; defining cable fault types according to a number of sample compressed signals, and creating a database of coaxial cables according to each cable fault type; Random Gaussian noise is added to the coaxial cable database, and cable insulation defects are detected based on the signal-to-noise ratio.

[0064] Alternatively, when the processor executes the computer program, the functions of each module in the above system are realized, for example: The signal compression module 1 is used to collect signals and compress the collected signals to obtain compressed sensing samples; A signal conversion module 2, used to generate an excitation signal according to compressed sensing sampling, and convert the excitation signal into an analog signal; A feedback signal compression module 3 is used to input the analog signal into the coaxial cable to obtain a feedback signal, and compress the feedback signal to obtain a plurality of sample compression signals; A database creation module 4 is used to define the cable fault type according to a number of sample compression signals, and to create a coaxial cable database according to each cable fault type; The defect detection module 5 is used to add random Gaussian noise into the database of the coaxial cable and perform cable insulation defect detection according to the signal-to-noise ratio.

[0065] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the mobile terminal.

[0066] For example, the computer program may be divided into a signal compression module 1, a signal conversion module 2, a feedback signal compression module 3, a feedback signal compression module 4, and a defect detection module 5; The specific functions of each module are as follows: The signal compression module 1 is used to collect signals and compress the collected signals to obtain compressed sensing samples; A signal conversion module 2, used to generate an excitation signal according to compressed sensing sampling, and convert the excitation signal into an analog signal; A feedback signal compression module 3 is used to input the analog signal into the coaxial cable to obtain a feedback signal, and compress the feedback signal to obtain a plurality of sample compression signals; A database creation module 4 is used to define the cable fault type according to a number of sample compression signals, and to create a coaxial cable database according to each cable fault type; The defect detection module 5 is used to add random Gaussian noise into the database of the coaxial cable and perform cable insulation defect detection according to the signal-to-noise ratio.

[0067] The mobile terminal may be a computing device such as a desktop computer, a notebook, a palm computer, a cloud server, etc. The mobile terminal may include, but is not limited to, a processor and a memory.

[0068] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the mobile terminal, and uses various interfaces and lines to connect various parts of the entire mobile terminal.

[0069] The memory may be used to store the computer programs and / or modules, and the processor implements various functions of the mobile terminal by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory.

[0070] The memory may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, a phone book, etc.), etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (SmartMediaCard, SMC), a secure digital (SecureDigital, SD) card, a flash card (FlashCard), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0071] Example 4 The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the cable insulation defect detection method are implemented.

[0072] If the module / unit integrated in the mobile terminal is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0073] Based on this understanding, the present invention implements all or part of the processes in the above method, and can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, the steps of the above-mentioned aggregate reinforcement learning resource scheduling method can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc.

[0074] The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc.

[0075] It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media does not include electrical carrier signals and telecommunication signals.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting cable insulation defects, characterized in that: include: Collecting signals, and compressing the collected signals to obtain compressed sensing sampling; Generate an excitation signal according to compressed sensing sampling, and convert the excitation signal into an analog signal; Inputting an analog signal into a coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compression signals; defining cable fault types according to a number of sample compressed signals, and creating a database of coaxial cables according to each cable fault type; Random Gaussian noise is added to the coaxial cable database, and cable insulation defects are detected based on the signal-to-noise ratio.

2. A cable insulation defect detection method according to claim 1, characterized in that: In the step of collecting signals and compressing the collected signals to obtain compressed sensing samples, the specific process is as follows: The relationship between the collected signal and the observation vector is determined, and compressed sampling is performed through the AIC analog-to-digital converter, in which the standard ADC analog-to-digital converter is replaced by the AIC analog-to-digital converter, and the mixer in the AIC analog-to-digital converter modulates the signal with a pseudo-random sequence, which is filtered by the filter and converted by the ADC analog-to-digital converter.

3. A cable insulation defect detection method according to claim 1, characterized in that: In the step of inputting the analog signal into the coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compressed signals, the obtained feedback signal is compressed by a random demodulator at a frequency of 200 MHz with a compression factor of 2 to obtain a plurality of sample compressed signals.

4. A cable insulation defect detection method according to claim 1, characterized in that: In the step of defining the cable fault type according to a plurality of sample compression signals and creating a coaxial cable database according to each cable fault type, the cable fault type includes short circuit, open circuit and soft fault, and reflection coefficient modeling is performed according to each cable fault type, wherein the reflection coefficient model is as follows: Γd = Zd−Zc / Zd + Zc Where Zd is the fault impedance; Zc is the cable characteristic impedance, which is a constant value; Γd is the reflection coefficient, and the value of Γd depends directly on Zd; When Γd = −1, the fault type corresponds to short circuit, Zd = 0 Ω; When Γd = 1, the fault type corresponds to open circuit, Zd = +∞Ω; When Γd < 1, the fault type corresponds to a soft fault; When Γd = 0, there is no fault and Zd = 58.00 Ω.

5. A cable insulation defect detection method according to claim 1, characterized in that: In the step of adding random Gaussian noise to the database of the coaxial cable and performing cable insulation defect detection according to the signal-to-noise ratio, cable insulation defect detection is performed according to the signal-to-noise ratio, wherein when the signal-to-noise ratio ranges from 0 to 15 dB, the KNN algorithm is used to detect cable insulation defects; when the signal-to-noise ratio ranges from 15 to 60 dB, the CNN model is used to detect cable insulation defects.

6. A cable insulation defect detection method according to claim 5, characterized in that: When the KNN algorithm is used for cable insulation defect detection, fault impedance estimation and fault location prediction are performed respectively, and the fault impedance estimation and fault location prediction respectively determine the optimal parameters through GridSearchCV; Among them, the optimal hyperparameters are determined by using GridSearchCV through fork validation in fault impedance estimation, including the number of neighbors as 1, the weight as uniform, the algorithm as ball tree, the leaf size as 10, and the metrics as euclidean and minkowski.

7. A cable insulation defect detection method according to claim 5, characterized in that: When using the CNN model for cable insulation defect detection, fault impedance estimation and fault location prediction are performed respectively; The fault impedance estimation and fault location prediction are trained on 80% of the generated data, 10% is used for validation, and 10% is used for testing. Leave-one-out cross validation is used during training, and the Adam optimizer is used with MAE as the loss function. During testing, the optimal parameters are evaluated in 10% of the database to obtain the optimal parameters.

8. A cable insulation defect detection system, characterized in that: include: A signal compression module is used to collect signals and compress the collected signals to obtain compressed sensing samples; A signal conversion module, used for generating an excitation signal according to compressed sensing sampling, and converting the excitation signal into an analog signal; A feedback signal compression module, used for inputting an analog signal into a coaxial cable to obtain a feedback signal, and compressing the feedback signal to obtain a plurality of sample compression signals; A database creation module, used to define cable fault types according to a number of sample compression signals, and to create a coaxial cable database according to each cable fault type; The defect detection module is used to add random Gaussian noise to the database of the coaxial cable and perform cable insulation defect detection based on the signal-to-noise ratio.

9. A mobile terminal comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the cable insulation defect detection method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the cable insulation defect detection method according to any one of claims 1 to 7 are implemented.

Images