Airline cable dc series arc fault detection method and storage medium

Abstract

The application belongs to the technical field of electrical engineering, and discloses a fault detection method for a DC series arc of an aviation cable and a storage medium. The fault detection method comprises the following steps: obtaining the voltage power spectral density of a line before and after the occurrence of an arc, and preprocessing the voltage power spectral density through a VMD algorithm to obtain an intrinsic mode function; obtaining the optimal decomposition layer number K and the penalty factor a of the intrinsic mode function through a GA algorithm; obtaining the characteristic index of the intrinsic mode function; and obtaining the fault detection threshold value through an SVM algorithm according to the characteristic index. The application uses only a single voltage signal as an input signal, and uses a GA algorithm to obtain the VMD optimal decomposition layer number K and the penalty factor a, so that the whole algorithm is more adaptive, the arc detection efficiency and reliability are improved, and the detection precision is improved.

Description

Technical Field

[0001] This invention belongs to the field of electrical engineering technology, specifically relating to a fault detection method and storage medium for DC series arcing in aviation cables. Background Technology

[0002] DC series arcing is a common fault in aircraft power distribution systems. Loose connectors or broken cable cores in the transmission lines are the main causes of this type of arcing. During the arcing process, intense light, radiation, and a large amount of heat are generated, causing severe physical or electrical damage to surrounding or terminal components, and may even trigger a regional fire.

[0003] In existing technologies, arc detection methods are mainly classified into three categories: detection methods based on arc physical signals, detection methods based on electromagnetic radiation, and detection methods based on arc electrical signals. Among these, detection methods based on arc physical signals are unsuitable for large-scale low-voltage DC systems due to numerous interference factors and unclear arc physical characteristics in open environments. Electromagnetic radiation-based detection methods have high hardware implementation costs, hindering widespread application. All of these arc detection methods result in unintuitive fault detection, reducing detection efficiency and reliability. Summary of the Invention

[0004] In view of this, the present invention provides a fault detection method and storage medium for DC series arc faults in aviation cables, in order to solve the problem that existing arc detection methods have low efficiency and reliability in fault detection.

[0005] In a first aspect, the present invention provides a fault detection method for DC series arcing in aviation cables, comprising: acquiring the voltage power spectral density of the line before and after the arcing occurs, and preprocessing the voltage power spectral density using a VMD algorithm (Variational Modal Decomposition, or VMD for short) to obtain intrinsic mode functions; obtaining the optimal decomposition level K and penalty factor α of the intrinsic mode functions using a GA algorithm (Genetic Algorithm, or GA for short); acquiring the feature index of the intrinsic mode functions; and obtaining a fault detection threshold based on the feature index using an SVM algorithm (Support Vector Machine, or SVM for short).

[0006] In an optional real-time approach, the voltage power spectral density is preprocessed using the VMD algorithm, specifically including: setting the voltage power spectral density as pseudo-time domain data; and decomposing the pseudo-time domain data using the VMD algorithm to obtain the intrinsic mode functions.

[0007] In an optional real-time approach, the optimal decomposition level K and penalty factor α of the intrinsic mode functions are obtained using the GA algorithm. Specifically, this includes: obtaining the envelope entropy and kurtosis of the K intrinsic mode functions; calculating the average of the K envelope entropies and the average of the K kurtosis; and obtaining the comprehensive objective function based on the average of the envelope entropies and the average of the kurtosis.

[0008] C = E ave +k ave (1); where C is the objective function, E ave and k ave Let K and α represent the average envelope entropy and average kurtosis obtained from the current K and α, respectively; the fitness function is obtained according to the objective function: (2); where fit is the fitness function, C max and C min These are the maximum and minimum values ​​of the integrated objective function in the current population, respectively. The optimal decomposition level K and penalty factor α that maximize the fitness function are calculated using the GA algorithm, and are used as the optimal decomposition level K and penalty factor α for VMD.

[0009] In an optional real-time approach, the fitness function is obtained based on the objective function, specifically including: obtaining the number of chromosomes, the number of genes, and the number of iterations representing K and α based on the objective function; and constructing the fitness function based on the number of chromosomes, the number of genes, and the number of iterations.

[0010] In an optional real-time mode, the line voltage before and after the arc occurs is collected at intervals of 1000 and a sampling frequency of 2MHz to obtain the voltage power spectral density.

[0011] In one optional real-time mode, the characteristic indices of the intrinsic mode function include three categories: kurtosis, margin, and power value.

[0012] In an optional real-time approach, the fault detection threshold is obtained using an SVM algorithm based on the aforementioned feature indicators. Specifically, this includes: normalizing each acquired feature indicator; adding a first label to the feature indicator data before the arc occurs and adding a second label to the feature indicator data after the arc occurs, combining them to obtain a test sample set; dividing the test sample set into a training set and a test set according to a preset ratio; training the training set using the SVM algorithm and using recall as a model performance evaluation indicator; and converting the model's output score to obtain the fault detection threshold.

[0013] In an optional real-time mode, the first label is a normal label and is recorded as 1, and the second label is an outlier label and is recorded as -1.

[0014] In one optional real-time mode, the preset ratio is 4:1.

[0015] On the other hand, the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of the fault detection method for DC series arcing of aviation cables as described in any one of the claims.

[0016] Since the aforementioned storage medium can implement the steps of the aforementioned fault detection method for DC series arc in aviation cables, it possesses all the beneficial effects of the fault detection method for DC series arc in aviation cables, which will not be elaborated further here.

[0017] The advantages and beneficial effects of this invention are as follows:

[0018] This invention introduces the concept of pseudo-frequency domain to improve the limitations of traditional arc signal-based detection methods that only analyze from the two dimensions of time and frequency domains. This application uses the VMD algorithm to extract arc fault detection features and uses the SVM algorithm to achieve adaptive determination of the fault detection threshold, making fault detection more intuitive and improving detection efficiency and reliability. Attached Figure Description

[0019] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0020] Figure 1 This is one of the flowcharts for a fault detection method of DC series arc in aviation cables according to an embodiment of the present invention;

[0021] Figure 2 The flowchart illustrates the GA algorithm used in a fault detection method for DC series arc in aviation cables according to an embodiment of the present invention to obtain the optimal VMD decomposition layer number K and penalty factor α.

[0022] Figure 3 A flowchart illustrating the method for detecting DC series arc faults in aviation cables according to an embodiment of the present invention, using the SVM algorithm to obtain the fault detection threshold;

[0023] Figure 4 This is a series DC fault arc data diagram of a fault detection method for DC series arc in aviation cables according to an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] The following is combined with Figures 1 to 4 The following describes embodiments of the present invention.

[0026] like Figure 1 As shown, according to an embodiment of the present invention, in one aspect, a method for detecting DC series arc faults in aviation cables is provided. The method for detecting DC series arc faults in aviation cables includes the following steps:

[0027] Step S101: Obtain the voltage power spectral density of the line before and after the arc occurs, and preprocess the voltage power spectral density using the VMD algorithm to obtain the intrinsic mode function;

[0028] Step S103: Obtain the optimal decomposition level K and penalty factor α of the intrinsic mode function using the GA algorithm;

[0029] Step S105: Obtain the characteristic indices of the intrinsic mode functions;

[0030] Step S107: Based on the feature indicators, obtain the fault detection threshold using the SVM algorithm.

[0031] The aforementioned cables can be airborne cables. Power spectral density is a probabilistic statistical method that measures the mean square value of a random variable. Generally, power spectral density is used for random vibration analysis. Continuous transient responses can only be described by probability distribution functions, i.e., the probability of a certain level of response occurring.

[0032] The voltage power spectral density of the line can be obtained by calculation, and the solution of the VMD algorithm is an existing technology, which will not be elaborated here.

[0033] Considering the difference in voltage power spectral density of the line before and after an arc occurs, it is necessary to collect the voltage power spectral density before and after the arc. By inputting the optimal decomposition level K and the penalty factor α, the intrinsic mode functions (EMFs) can be obtained using the VMD algorithm. Using only a single voltage signal as input and employing the GA algorithm to obtain the optimal VMD decomposition level K and penalty factor α, the entire algorithm is more adaptive, improving the efficiency and reliability of arc detection, and enhancing detection accuracy.

[0034] In another implementation, the voltage power spectral density is preprocessed using the VMD algorithm, specifically including the following steps:

[0035] Step S201: Set the voltage power spectral density to pseudo-time domain data;

[0036] Step S203: Decompose the pseudo-time domain data using the VMD algorithm to obtain the intrinsic mode functions.

[0037] The voltage power spectral density mentioned above is frequency domain data. By treating the voltage power spectral density as pseudo-time domain data and decomposing the pseudo-time domain data using the VMD algorithm, the resulting intrinsic mode functions are regarded as pseudo-frequency domain data. This improves the limitations of traditional analysis based solely on the time and frequency domains by introducing the concept of pseudo-frequency domain, and further enhances the reliability of arc fault detection.

[0038] like Figure 2 As shown, in another embodiment, the optimal decomposition level K and penalty factor α of the intrinsic mode function are obtained through the GA algorithm, specifically including the following steps:

[0039] Step S301: Obtain the envelope entropy and kurtosis of the K intrinsic mode functions;

[0040] Step S303: Calculate the average of the K envelope entropies and the average of the K kurtosis;

[0041] Step S305: Based on the average value of the envelope entropy and the average value of the kurtosis, the comprehensive objective function is obtained as follows:

[0042] C = E ave +k ave (1)

[0043] Where C is the objective function, E ave and k ave Let K and α represent the average envelope entropy and average kurtosis, respectively.

[0044] Step S301: Obtain the fitness function based on the objective function:

[0045]

[0046] Where fit is the fitness function, C max and C min These represent the maximum and minimum values ​​of the integrated objective function in the current population, respectively.

[0047] Step S307: Calculate the optimal decomposition layer K and penalty factor α that maximize the fitness function using the GA algorithm, and use them as the optimal decomposition layer K and penalty factor α for VMD.

[0048] This can be understood as using the average of K envelope entropies and K kurtosis as indices for the GA algorithm to optimize the optimal decomposition level K and penalty factor α, establishing a comprehensive objective function, constructing a comprehensive objective function representing the optimal decomposition level K and penalty factor α to represent the fitness function, and using the GA algorithm to calculate the optimal decomposition level K and penalty factor α that maximizes the fitness function, which are then used as the optimal decomposition level K and penalty factor α for the VMD algorithm.

[0049] In another embodiment, the fitness function is obtained based on the objective function, specifically including the following steps:

[0050] Step S401: Obtain the number of chromosomes, genes, and iterations representing the optimal decomposition level K and penalty factor α based on the objective function;

[0051] Step S403: Construct a fitness function based on the number of chromosomes, the number of genes, and the number of iterations.

[0052] The line voltage before and after the arc was collected at intervals of N=1000 and a sampling frequency of 2MHz to obtain the voltage power spectral density.

[0053] In another embodiment, the characteristic indices of the intrinsic mode function include three categories: kurtosis, margin, and power value. Using the VMD algorithm to extract kurtosis, margin, and power value as detection features better reflects the essential characteristics of the electric arc.

[0054] like Figure 3 As shown, in another embodiment, the fault detection threshold is obtained using the SVM algorithm based on feature indicators, specifically including the following steps:

[0055] Step S501: Normalize each acquired feature index, add the feature index data before the arc occurs to the first label, add the feature index data after the arc occurs to the second label, and combine them to obtain the test sample set.

[0056] Step S503: Divide the test sample set into a training set and a test set according to a preset ratio;

[0057] Step S505: Train the training set using the SVM algorithm and use recall as the model performance evaluation metric.

[0058] Step S507: Convert the scores output by the model to obtain the fault detection threshold.

[0059] In this context, the first label is the normal label, which is denoted as 1, and the second label is the outlier label, which is denoted as -1.

[0060] The preset ratio is 4:1.

[0061] The aforementioned fault detection method for DC series arc in aviation cables is a fault detection method for DC series arc in airborne cables based on pseudo-frequency domain and VMD-GA-SVM. By setting the voltage sampling interval N=1000 and the sampling frequency to 2MHz, the power spectral density of the collected line voltage is calculated, and the intrinsic mode function is obtained by preprocessing with the VMD algorithm. Since the power spectral density itself is frequency domain data, it is regarded as pseudo-time domain data. The intrinsic mode function obtained by processing the pseudo-time domain data with the VMD algorithm is regarded as pseudo-frequency domain data.

[0062] The VMD algorithm, including the calculation of kurtosis, margin, and power values, is an existing technology and will not be elaborated upon here.

[0063] Each acquired feature index is normalized, and the feature index data before the arc occurs is labeled with a normal label (denoted as 1), and the feature index data after the arc occurs is labeled with an outlier label (denoted as -1). The combination is used to obtain the test sample set, which is then divided into a training set and a test set in a 4:1 ratio.

[0064] The training set is trained using the SVM algorithm, and recall is used as the performance evaluation metric for the model.

[0065] The scores output by the model are converted into thresholds for outliers / normal points, which are the fault detection thresholds.

[0066] The introduction of the pseudo-frequency domain concept improves the limitations of traditional arc signal-based detection methods that only analyze from the time and frequency domains. The VMD algorithm is used to extract arc fault detection features, and the SVM algorithm is used to adaptively determine the fault detection threshold, making fault detection more intuitive and improving detection efficiency and reliability.

[0067] like Figure 4 As shown, U1 represents the arc initiation point, which is the voltage value corresponding to the moment the electric arc is generated.

[0068] According to an embodiment of the present invention, in another aspect, a storage medium is also provided, storing a computer program, which, when executed by a processor, implements the steps of any of the fault detection methods for DC series arcs in aviation cables.

[0069] In the above embodiments, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0070] Optionally, the processor executes the method steps described in the above embodiments based on the program code stored in the storage medium.

[0071] Specific examples in this embodiment can be found in the examples described in the above embodiments and optional implementations, and will not be repeated here.

[0072] Obviously, those skilled in the art should understand that the modules or steps of the present disclosure described above can be implemented using general-purpose computing devices, which can be centralized on a single computing device or distributed on a network of multiple computing devices.

[0073] Alternatively, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device. Furthermore, in some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this disclosure is not limited to any particular combination of hardware and software.

[0074] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for detecting DC series arc faults in aviation cables, characterized in that, include: The voltage power spectral density of the line before and after the arc occurs is obtained, and the voltage power spectral density is preprocessed by the VMD algorithm to obtain the intrinsic mode function. The optimal decomposition level of the intrinsic mode function is obtained using the GA algorithm. and penalty factor ; The feature indices of the intrinsic mode function are obtained, and the feature indices of the intrinsic mode function include three categories: kurtosis, margin, and power value; Based on the aforementioned feature indicators, the fault detection threshold is obtained using the SVM algorithm; The voltage power spectral density is preprocessed using the VMD algorithm, specifically including: The voltage power spectral density is set to pseudo-time domain data; The pseudo-time domain data is decomposed using the VMD algorithm to obtain the intrinsic mode functions; The optimal decomposition level of the intrinsic mode function is obtained using the GA algorithm. and penalty factor Specifically, it includes: Obtain the envelope entropy and kurtosis of the K intrinsic mode functions; calculate The average value of the envelope entropy and The average of the kurtosis; The comprehensive objective function is obtained based on the average value of the envelope entropy and the average value of the kurtosis: （1） in, Let be the objective function. and Let K and the penalty factor be the number of the current optimal decomposition layers, respectively. The obtained average envelope entropy and average kurtosis; The fitness function is obtained based on the objective function: （2） in, For the fitness function, and These represent the maximum and minimum values ​​of the integrated objective function in the current population, respectively. The optimal decomposition level that maximizes the fitness function is calculated using the GA algorithm. and penalty factor As the optimal number of decomposition levels for the VMD algorithm and penalty factor .

2. The fault detection method for DC series arc in aviation cables according to claim 1, characterized in that, The fitness function is obtained based on the objective function, specifically including: The optimal decomposition level is obtained based on the objective function. and penalty factor The number of chromosomes, the number of genes, and the number of iterations; The fitness function is constructed based on the number of chromosomes, the number of genes, and the number of iterations.

3. The fault detection method for DC series arc in aviation cables according to any one of claims 1 to 2, characterized in that, The voltage of the line before and after the arc occurs is collected at intervals of N equal to 1000 and a sampling frequency of 2MHz to obtain the voltage power spectral density.

4. The fault detection method for DC series arc in aviation cables according to claim 3, characterized in that, Based on the aforementioned feature indicators, the fault detection threshold is obtained using the SVM algorithm, specifically including: Each of the acquired feature indicators is normalized, and the feature indicator data before the arc occurs is added to the first label, and the feature indicator data after the arc occurs is added to the second label. The combination is used to obtain the test sample set. The test sample set is divided into a training set and a test set according to a preset ratio; The training set is trained using the SVM algorithm, and the recall rate is used as the performance evaluation metric for the model. The scores output by the model are converted to obtain the fault detection threshold.

5. The fault detection method for DC series arc in aviation cables according to claim 4, characterized in that, The first label is a normal label, and is recorded as 1. The second label is an outlier label, and is recorded as -1.

6. The fault detection method for DC series arc in aviation cables according to claim 4, characterized in that, The preset ratio is 4:

1.

7. A storage medium, characterized in that, The device contains a computer program that, when executed by a processor, implements the steps of the fault detection method for DC series arcing of aviation cables as described in any one of claims 1 to 6.

Images