A power distribution network fault detection method and device based on variational mode decomposition

Abstract

The embodiment of the disclosure discloses a power distribution network fault detection method and device based on variational mode decomposition, wherein the method comprises the following steps: according to the negative sequence current suppression control strategy, the transient part of the zero sequence voltage of the power distribution network is collected and calculated, then according to the variational mode decomposition algorithm, the transient signal is decomposed, the energy distribution proportion of the mode component in the transient signal is calculated, and the energy entropy value is calculated according to the calculation result, and the amplitude mean value of the zero sequence voltage time domain signal is calculated; the characteristic vector is constructed by using the calculated result, and the support vector machine model is evaluated; the characteristics of the zero sequence voltage are identified by using the evaluated support vector machine model, and the detection of the power distribution network fault is completed. The embodiment of the disclosure realizes the identification of the high resistance fault by using the zero sequence voltage signal, ensures the accuracy of identifying the fault type, and improves the adaptability and interpretability of the system.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of power distribution network fault detection, and in particular, to a power distribution network fault detection method and device based on variational mode decomposition. BACKGROUND

[0002] With the wide application of distributed power supply in the power distribution network, the scale of the power distribution network system is more massive and the operating environment is more complex, which makes the detection of high-resistance faults an increasingly important problem. High-resistance faults are usually caused by non-conductive media such as branches and soil. Due to the high transition resistance, the amplitude of the fault current and voltage signal is small, resulting in very weak fault characteristics, and it is easy to be disturbed by harmonics, noise and other disturbances. Therefore, high-resistance ground faults are more difficult to detect, resulting in the persistence of the fault, and ultimately causing electric shock, fire and other serious accidents.

[0003] Traditional fault detection methods, such as time domain, frequency domain and time-frequency domain methods, all extract feature quantities of fault signals and set threshold values for judgment. However, these methods have certain limitations, especially in high-resistance faults. Due to the small signal amplitude and the easy disturbance from the outside world, the above-mentioned methods have poor anti-interference ability and low adaptability, and are prone to miss or misjudge. In addition, other researches also combine deep learning methods such as convolutional neural network (CNN) to further improve the accuracy of feature extraction. However, although the neural network-based algorithm can improve the accuracy of the model, it requires a large amount of training data, which may lead to a decline in the performance of the model in a small sample environment for this relatively rare fault type of high-resistance fault.

[0004] Therefore, one or more methods are needed to solve the above problems.

[0005] It should be noted that the information disclosed in the above background section is only used to strengthen the understanding of the background of the present disclosure, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0006] The purpose of the present disclosure is to provide a power distribution network fault detection method and device based on variational mode decomposition, thereby at least partially overcoming one or more problems caused by the limitations and defects of the related art.

[0007] According to one aspect of the present disclosure, a power distribution network fault detection method based on variational mode decomposition is provided, comprising:

[0008] According to the negative sequence current suppression control strategy, the transient signal is generated by collecting and calculating the transient part of the zero sequence voltage of the power distribution network;

[0009] According to the variational mode decomposition algorithm, the mode component is generated by performing variational mode decomposition on the transient signal;

[0010] According to the orthogonality of variational mode decomposition, the variational mode energy distribution is generated by calculating the energy distribution proportion of the mode component in the transient signal;

[0011] According to the mode energy distribution, the variational mode energy entropy is generated by calculating the energy entropy value of the variational mode decomposition;

[0012] According to the transient signal, the transient signal amplitude mean value is generated by calculating the amplitude mean value of the time domain signal of the zero sequence voltage;

[0013] The feature vector is constructed using the variational mode energy distribution, the variational mode energy entropy, and the transient signal amplitude mean value, and the support vector machine model is evaluated through the feature vector;

[0014] The transient characteristics of the zero sequence voltage are identified using the evaluated support vector machine model, and the detection of the power distribution network fault is completed.

[0015] In an exemplary embodiment of the present disclosure, the transient part of the zero sequence voltage of the power distribution network is collected and calculated, including:

[0016] Based on the negative sequence current suppression control strategy, the circuit topology analysis of the zero sequence component during fault is completed by calculating the reference value of the positive sequence current component on the direct axis and the cross axis of the inverter when the single high-impedance grounding fault occurs in the power distribution network;

[0017] Based on the circuit topology analysis of the zero sequence component during fault, the high-impedance fault equivalent circuit is constructed by virtually setting the fault point as an equivalent power source;

[0018] Based on the high-impedance fault equivalent circuit, the zero sequence voltage signal is generated by collecting and calculating the zero sequence voltage in the circuit;

[0019] Based on the zero sequence voltage signal, the transient signal is generated by collecting the transient part of the zero sequence voltage in a preset time window.

[0020] In an exemplary embodiment of the present disclosure, the transient signal is decomposed by variational mode decomposition, including:

[0021] Based on the variational mode decomposition algorithm, the transient signal constraint variational problem is generated by constructing the constraint variational problem of the transient signal;

[0022] Based on the preset distortion degree limit, the Lagrange operator is solved by introducing noise tolerance, and the Lagrange operator is generated;

[0023] In an example embodiment of the present disclosure, the transient signal is decomposed into modal signals by variational mode decomposition, including:

[0024] Based on the Lagrange algorithm, the constrained variational problem of the transient signal is transformed by introducing the Lagrange operator to generate an unconstrained variational problem;

[0025] Using an iterative algorithm, the unconstrained variational problem is Fourier equidistant transformed by introducing a quadratic penalty factor, and iteration of the components of the modal signal decomposition is started;

[0026] When the component of the modal signal decomposition is not less than the preset accuracy, it is determined that the iteration does not terminate;

[0027] When the component of the modal signal decomposition is less than the preset accuracy, it is determined that the iteration terminates and outputs the iteration result to generate the modal component.

[0028] In an example embodiment of the present disclosure, a feature vector is constructed using the variational modal energy distribution, the variational modal energy entropy, and the transient signal amplitude mean, including:

[0029] Using the variational modal energy distribution, the variational modal energy entropy, and the transient signal amplitude mean, the features of the support vector machine input sample are constructed to generate a feature vector;

[0030] Based on the feature vector, a training set of a support vector machine model is prepared to generate a training data set;

[0031] Based on the feature vector, a test set of a support vector machine model is prepared to generate a test data set;

[0032] The ratio of the training data set to the test data set is 2:1.

[0033] In an example embodiment of the present disclosure, the support vector machine model is evaluated by the feature vector, including:

[0034] Based on the support vector machine model, the training data set is mapped in a high-dimensional space, and the classification interval of the training data set is classified in the mapped high-dimensional space by using a hyperplane to generate a classification interval constraint problem;

[0035] Based on the Lagrange algorithm, the classification interval constraint problem is transformed by introducing a Lagrange multiplier to generate a classification interval optimization problem;

[0036] Based on the classification interval optimization problem, the partial derivative of the Lagrange function is obtained and the result is set to zero to obtain a dual problem.

[0037] In an example embodiment of the present disclosure, the support vector machine model is evaluated by the feature vector, including:

[0038] The parameters of the support vector machine model are calculated by solving the dual problem, to generate support vector machine model parameters;

[0039] Based on the support vector machine model parameters, the decision function is constructed, and the test data set is introduced into the decision function, and the evaluation of the support vector machine model is completed by solving the decision function.

[0040] In an example embodiment of the present disclosure, the transient characteristics of the zero sequence voltage are identified by using the evaluated support vector machine model, and the detection of the power distribution network fault is completed, including:

[0041] The transient characteristics of the zero sequence voltage are input into the evaluated support vector machine model as input data, and the decision function is solved to generate a decision result;

[0042] When the decision result falls into a preset high-resistance fault decision interval, it is determined that the power distribution network has a high-resistance ground fault;

[0043] When the decision result falls into a preset line interference decision interval, it is determined that the power distribution network is interfered;

[0044] When the decision result falls into a preset line normal decision interval, it is determined that the line of the power distribution network is in normal operation.

[0045] In an example embodiment of the present disclosure, it further includes:

[0046] Using a local interpretability analysis algorithm, the trained support vector machine model is analyzed for interpretability to generate a preset variable analysis report of the working condition recognition support vector machine model;

[0047] Using a local interpretability analysis algorithm, the adaptive analysis of the fault scenario is performed to generate an analysis report of the influence of the fault scenario on the detection effect.

[0048] In one aspect of the present disclosure, a power distribution network fault detection device based on variational mode decomposition is provided, including:

[0049] The transient signal acquisition module is configured to acquire and calculate the transient part of the zero sequence voltage of the power distribution network according to the negative sequence current suppression control strategy, to generate a transient signal;

[0050] The variational mode decomposition module is configured to perform variational mode decomposition on the transient signal according to a variational mode decomposition algorithm to generate a modal component;

[0051] An energy distribution proportion calculation module is configured to calculate the energy distribution proportion of the modal components in the transient signal according to the orthogonality of the variational modal decomposition, and generate variational modal energy distribution.

[0052] An energy entropy value calculation module is configured to calculate the energy entropy value of the variational modal decomposition according to the modal energy distribution, and generate variational modal energy entropy.

[0053] A time-domain signal amplitude mean value calculation module is configured to calculate the amplitude mean value of the time-domain signal of the zero sequence voltage according to the transient signal, and generate a transient signal amplitude mean value.

[0054] A feature vector construction module is configured to construct a feature vector by using the variational modal energy distribution, the variational modal energy entropy, and the transient signal amplitude mean value, and evaluate a support vector machine model by using the feature vector.

[0055] A fault detection module is configured to identify the transient feature of the zero sequence voltage by using the evaluated support vector machine model, and complete the detection of the power distribution network fault.

[0056] One of the exemplary embodiments of the present disclosure is a power distribution network fault detection method based on variational modal decomposition. First, the transient part of the zero sequence voltage of the power distribution network is calculated according to the negative sequence current control strategy. Then, the transient signal is decomposed according to the variational modal decomposition algorithm, and the energy distribution proportion of the modal components in the transient signal is calculated. The energy entropy value is calculated according to the calculation result. The amplitude mean value of the time-domain signal of the zero sequence voltage is calculated. Then, the feature vector is constructed by using the calculated results, and the support vector machine model is evaluated. Finally, the feature of the zero sequence voltage is identified by using the evaluated support vector machine model, and the detection of the power distribution network fault is completed. The embodiment of the present disclosure provides a scheme of decomposing the zero sequence voltage based on the variational modal decomposition algorithm, taking the decomposed parameters as the feature vector to evaluate the support vector machine model, and detecting the fault. The scheme realizes the identification of the high-resistance fault by using the zero sequence voltage signal, ensures the accuracy of identifying the fault type, and improves the adaptability and interpretability of the system.

[0057] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0058] The above and other features and advantages of the present disclosure will become more apparent by describing in detail exemplary embodiments thereof with reference to the attached drawings.

[0059] Figure 1A flow chart of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0060] Figure 2 A fault detection flow chart of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0061] Figure 3 A high resistance fault equivalent circuit of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0062] Figure 4 A variational modal decomposition algorithm flow chart of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0063] Figure 5 A distribution diagram of VMD energy under each working condition of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0064] Figure 6 A feature category comparison diagram of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0065] Figure 7 A support vector machine principle diagram of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0066] Figure 8 A neutral point ungrounded power distribution network topology diagram of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0067] Figure 9 A support vector machine model simulation fault determination confusion matrix diagram of a power distribution network fault detection method based on variational modal decomposition according to an example embodiment of the present disclosure is shown;

[0068] Figure 10 A schematic block diagram of a power distribution network fault detection device based on variational modal decomposition according to an example embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0069] Example embodiments now will be described more fully hereinafter with reference to the accompanying drawings; however, they can be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the example embodiments to those skilled in the art. Like reference numerals refer to like elements throughout the several views, and descriptions of the same elements can be omitted from the descriptions of the other views.

[0070] Moreover, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the

[0071] The block diagrams in the drawings show only the functionality of the features and can not imply a physical or architectural arrangement of the devices. No inference should be made regarding databases (i.e., information stored in memory or other storage), logical connections, physical connections, or relationships of the devices to each other. Functional aspects can be implemented in software, hardware, firmware, or a combination thereof. Various functions performed by the devices can be performed in one or more of the devices shown or in a different device.

[0072] In the present example embodiment, firstly, a power distribution network fault detection method based on variational mode decomposition is provided; with reference to the drawings shown in Figure 1 The power distribution network fault detection method based on variational mode decomposition can include the following steps:

[0073] Step S110, according to the negative sequence current control strategy, by collecting and calculating the transient part of the power distribution network zero sequence voltage, a transient signal is generated;

[0074] Step S12, according to the variational mode decomposition algorithm, by variational mode decomposition of the transient signal, a modal component is generated;

[0075] Step S130, according to the orthogonality of variational mode decomposition, by calculating the energy distribution proportion of the modal component in the transient signal, a variational mode energy distribution is generated;

[0076] Step S140, according to the modal energy distribution, by calculating the energy entropy value of the variational mode decomposition, a variational mode energy entropy is generated;

[0077] Step S150, according to the transient signal, by calculating the amplitude mean value of the time domain signal of the zero sequence voltage, a transient signal amplitude mean value is generated;

[0078] Step S160, a feature vector is constructed using the variational modal energy distribution, variational modal energy entropy, and transient signal amplitude mean value, and a support vector machine model is evaluated through the feature vector;

[0079] Step S170, the transient characteristics of the zero sequence voltage are identified using the evaluated support vector machine model, and the detection of the power distribution network fault is completed.

[0080] One of the exemplary embodiments of the present disclosure is a power distribution network fault detection method based on variational modal decomposition and support vector machine: first, according to the negative sequence current suppression control strategy, the transient part of the zero sequence voltage of the power distribution network is collected and calculated. Then, according to the variational modal decomposition algorithm, the transient signal is decomposed, and then the energy distribution proportion of the modal component in the transient signal is calculated, and the energy entropy value is calculated according to the calculation result. At the same time, the amplitude mean value of the zero sequence voltage time domain signal is calculated. Then, the feature vector is constructed using the calculated results, and the support vector machine model is evaluated. Finally, the features of the zero sequence voltage are identified using the evaluated support vector machine model, and the detection of the power distribution network fault is completed. The present embodiment realizes the identification of high resistance fault using the zero sequence voltage signal, to ensure the accuracy of identifying the fault type, and improves the adaptability and interpretability of the system.

[0081] Next, as shown in Figure 2 , a power distribution network fault detection method based on variational modal decomposition in the present exemplary embodiment will be further described.

[0082] In the template configuration step S110, according to the negative sequence current suppression control strategy, the transient part of the zero sequence voltage of the power distribution network is collected and calculated to generate a transient signal.

[0083] Nowadays, the voltage transformer of the power distribution network is relatively sensitive. When a single-phase high-resistance grounding fault occurs in the line, the voltage transformer can collect the transient signal of the zero sequence voltage which is relatively obvious. Therefore, the transient characteristics of the zero sequence voltage after the single-phase high-resistance grounding fault of the power distribution network containing distributed power sources can be analyzed, and the theoretical basis for the high-resistance fault detection method can be established according to the characteristics.

[0084] In the present exemplary embodiment, the distributed power source adopting PQ control (P and Q in PQ control respectively refer to active power and reactive power) usually adopts the negative sequence current suppression control strategy during power grid fault. Therefore, based on this control strategy, it is considered that the inverter-type power source only outputs positive sequence current during fault.

[0085] Then, the fault period differential equation of the inverter dq axis (direct axis and cross axis) current can be expressed as:

[0086]

[0087] The direct axis and cross axis current expressions of formula (1) are obtained by solving differential equations:

[0088]

[0089] Where A1, A2, B1, B2 are undetermined coefficients, i d *and i q *are the reference values of the direct axis and cross axis components of positive sequence current.

[0090] Thus, the circuit topology analysis of zero sequence component when fault occurs is completed. That is, when single-phase high resistance grounding fault occurs in the distribution network, the distributed voltage in the microgrid does not output zero sequence component for the power grid.

[0091] As Figure 3 shown, according to the results of the above circuit topology analysis, the high resistance fault equivalent circuit of the neutral point ungrounded system is constructed. In order to simplify the model of the neutral point ungrounded system, the fault point can be equivalent to a virtual power source Where U m is the phase voltage amplitude of the fault phase during normal operation, ω is the power frequency angular frequency, is the initial phase angle of the fault. And in Figure 3 , R h is the neutral point grounding resistance; I 0n is the zero sequence current flowing out of the fault line, I 0f is the zero sequence current flowing through the fault point; C j (j=1, 2,...n) is the zero sequence distributed capacitance of the jth feeder to ground; I cj represents the current flowing through the zero sequence distributed capacitance Cj of the jth feeder to ground; u c is the bus zero sequence voltage, u 0f is the zero sequence voltage of the fault point.

[0092] The collection of zero sequence voltage signal can be completed by the equivalent circuit, and the specific steps are as follows:

[0093] First, the first order differential equation expression of the virtual power source is obtained as:

[0094] u f = u c + R(i cΣ +i 0z ) (3)

[0095] Where i c∑ is the sum of the currents of all feeder capacitances to ground, i 0z is the zero sequence current flowing into the ground through the neutral point.

[0096] Secondly, for simplifying the calculation, it is assumed that the neutral point to ground resistance R h is approximately equal to 0, then i 0z tends to 0, and the following equation can be obtained:

[0097]

[0098] where C ∑ is the sum of the zero sequence distributed capacitance of all feeders to ground.

[0099] Thirdly, according to equation (4), equation (3) is transformed as follows:

[0100]

[0101] Fourthly, equation (5) is solved, and the bus zero sequence voltage u c can be obtained:

[0102]

[0103] where e is the natural logarithm, ω is the power frequency angular frequency, and the parameters θ and δ in equation (6) are respectively:

[0104]

[0105] Finally, through the above steps, it is concluded that when the high resistance fault occurs in the neutral point ungrounded system, the zero sequence voltage and the zero sequence current are both superimposed by the DC component and the steady-state sinusoidal component. And with the increase of the transition resistance R, the initial amplitude of the steady-state voltage, the transient voltage and the current component will decrease. According to equation (5), with the increase of the transition resistance, the decay factor δ will continuously decrease, and thus the decay speed will become slower and slower.

[0106] Then in this example, a 500us time window is set to obtain the transient signal of the zero sequence voltage signal, and the time domain characteristics and the frequency domain characteristics are extracted and compared, so as to realize the accurate identification of the high resistance fault.

[0107] In the template configuration step S120, according to the variational mode decomposition algorithm, the modal components are generated by performing variational mode decomposition on the transient signal.

[0108] In the embodiment of the present example, as shown in Figure 4 , it is intended to extract the time domain characteristics and the frequency domain characteristics of the transient signal, so it is necessary to perform variational mode decomposition (VMD, Variational Mode Decomposition) on the original transient signal, and decompose it into K intrinsic mode function (IMF, intrinsic mode function) components.

[0109] First step, the constrained variational problem can be constructed as shown in equation (8):

[0110]

[0111] Wherein, K is the total number of modes to be decomposed, The partial derivative of t is taken, {u k} is the kth component after decomposition, {ω k} is the center frequency of the kth component, s.t. is the constraint condition that needs to be met all {u k} sum is f.

[0112] Second step, to solve equation (8) by introducing Lagrange multiplier λ to convert it into an unconstrained variational problem, which can be obtained:

[0113]

[0114] Wherein, in order to limit the distortion of the decomposed signal within a certain range, noise tolerance γ is introduced, so the solving process of Lagrange multiplier λ is shown in equation (10):

[0115]

[0116] Third step, in equation (9) introduce quadratic penalty factor α to reduce the Gaussian noise interference existing in the signal. Again, the Fourier equidistant transform is performed on u k (t), ω(t) and λ(t) (i.e. at time t, the Fourier equidistant transform is performed on the kth component, the k+1th component, the power frequency angular frequency, and the Lagrange multiplier), and then the optimal component and center frequency of modal signal decomposition are calculated using an iterative algorithm, and the solving process is shown in equations (11) and (12):

[0117]

[0118]

[0119] Fourth step, set a preset accuracy ε, as shown in equation (13):

[0120]

[0121] Fifth step, compare the calculation results after iteration in the third step with the preset accuracy ε, if the accuracy requirement of equation (13) is met, the iteration is terminated, and the results are output for plotting to generate modal components. If the accuracy requirement is not met, return to step (11) to continue the next iteration calculation.

[0122] In the template configuration steps S130 and S140, according to the orthogonality of the variational modal decomposition, the energy distribution of the variational modal is generated by calculating the energy distribution proportion of the modal component in the transient signal. And according to the modal energy distribution, the energy entropy of the variational modal decomposition is generated by calculating the energy entropy value.

[0123] In the embodiment of the present example, when the line is faulty or disturbed, the frequency distribution of the zero sequence voltage will change, and therefore the energy distribution of its signal will also change accordingly, so the concept of energy entropy can be introduced to study the energy distribution of each IMF obtained after VMD decomposition.

[0124] According to the K IMF components obtained by the variational modal decomposition of the original transient signal, the energy of the i th IMF component can be expressed as:

[0125]

[0126] K is the total number of sampling points of the i th IMF signal, and k(j) is the j th sampling point of the i th IMF signal.

[0127] Since the VMD decomposition has orthogonality, the sum of the energies of the k IMFs is equal to the total energy E of the original transient signal. Thus the definition of the VMD energy entropy value can be obtained:

[0128]

[0129] In the formula, p i is the proportion of the energy of the i th IMF signal in the total energy E, that is, the energy distribution of the i th modal component in the total signal after VMD decomposition, which can be expressed as:

[0130]

[0131] In the template configuration step S150, according to the transient signal, the transient signal amplitude mean is generated by calculating the time domain signal amplitude mean of the zero sequence voltage.

[0132] In the embodiment of the present example, as shown in Figure 5 , the VMD modal energy distribution of the distribution network zero sequence voltage is different under the conditions of normal operation, high resistance fault, noise and harmonic interference, asymmetric load switching, etc.

[0133] Among them, the VMD energy distribution characteristics of high resistance fault, asymmetric load and normal operation are similar, especially the VMD energy distribution characteristics of high resistance fault and asymmetric load are highly similar, so it can be seen that using only the VMD energy distribution as a feature vector may lead to low accuracy of SVM classification and recognition.

[0134] At the same time, since the zero sequence voltage amplitude of high resistance fault and asymmetric load switching is much larger than that in normal operation, in order to better distinguish the transient characteristics of the above working conditions, the transient signal amplitude mean value of the time domain signal of the zero sequence voltage can be calculated:

[0135]

[0136] Wherein, K is the total number of sampling points in the 500us time window, and v0(k) is the kth sampling point. At the same time, the capacitive factor is dominant in the transient process of high resistance fault, while the inductive factor is dominant in the transient process of asymmetric load, so there is a big difference in the VMD energy entropy characteristics of their transient processes.

[0137] In the template configuration step S160, the feature vector is constructed by using the variational modal energy distribution, the variational modal energy entropy and the transient signal amplitude mean value, and the support vector machine model is evaluated through the feature vector.

[0138] In the embodiment of the present example, as shown in Figure 6 The capacitive factor is dominant in the transient process of high resistance fault, while the inductive factor is dominant in the transient process of asymmetric load, so there is a big difference in the VMD energy entropy characteristics of their transient processes.

[0139] Therefore, the variational modal energy entropy H EVMD of the transient signal, the variational modal energy distribution p i of each modal in the signal and the transient signal amplitude mean value of the time domain waveform are calculated respectively by using formula (15), formula (16) and formula (17), and the feature vector T is constructed according to these values.

[0140] Then, the training data set and the test data set are constituted by using the feature vector T, the SVM (Support Vector Machine) is trained by using the training data set, and the accuracy of the trained model is verified by using the test data set. The ratio of the training data set to the test data set is 2:1.

[0141] In a specific example, as shown in Figure 7 The support vector machine model divides the samples into two types of samples, H is the optimal hyperplane of classification, H1 and H2 are respectively the hyperplanes passing through the samples closest to the classification line in each type and parallel to H, and the optimal classification hyperplane requires that the classification surface not only correctly separates the two types (the training error rate is 0), but also maximizes the classification margin.

[0142] If the sample set (x i , y i ), i = 1, 2,...l is linearly separable and satisfies:

[0143] y i [(ω·x i )+b]-1≥0(18)

[0144] The classification interval of the sample can be set as 2 / |ω|, and then in order to get the maximum classification interval, the value of |ω| 2 needs to be minimized, so the optimal classification hyperplane can be found by solving the classification interval constraint problem (19).

[0145]

[0146] x.t.y i (ω·x i +b)≥1,i=1,2,...,l(19)

[0147] Then, in order to solve the above constraint problem, the Lagrange multipliers ai can be introduced to transform the classification interval constraint problem, construct the Lagrange function, and establish the classification interval optimization problem:

[0148]

[0149] In order to solve ω and b, the optimization problem is transformed into a dual problem by taking the partial derivative of the Lagrange function and setting the result to zero:

[0150]

[0151] By solving the dual problem, the Lagrange multipliers a i are obtained, and then the optimal ω and b can be further solved:

[0152]

[0153] Then, by obtaining the optimal ω and b, the final decision function is constructed:

[0154]

[0155] Finally, the evaluation of the support vector machine model is completed by calculating the constructed decision function using the test data set (i.e., verifying the accuracy of the trained support vector machine model).

[0156] In a specific example, through matrix laboratory (matlab) calculation, when y = 1, the fault of the power distribution network is a high-impedance fault, when y = 0, the power distribution network is in normal operation, and other operating modes need to be comprehensively judged by multiple decision functions.

[0157] In the template configuration step S170, the transient characteristics of the zero sequence voltage are identified by using the evaluated support vector machine model, and detection of the fault of the distribution network is completed.

[0158] In the embodiment of the present example, a high-resistance fault simulation model of a neutral non-grounded distribution network is built based on a PSCAD / EMTDC platform, as shown in FIG. 2, and a distributed power supply is connected to the end of the feeder 1 and the feeder 3. In the simulation model, the distribution network is composed of overhead lines and power cables, and eight fault points are set on typical lines such as the beginning of the line, the overhead line, the power cable, the end of the line, and the branch feeder, so as to collect high-resistance fault characteristics at different line positions. Meanwhile, the Emanuel high-resistance fault model is used to simulate the high-resistance fault. Figure 8

[0159] In a specific example, 900 groups of feature vectors are collected in a time window of 500 us for each working condition. These feature vectors are input into the evaluated support vector machine model as input data to determine the fault. When the decision result calculated by the decision function falls into the preset high-resistance fault decision interval, it is determined that the distribution network has a high-resistance grounding fault; when the decision result falls into the preset line interference decision interval, it is determined that the distribution network is interfered; and when the decision result falls into the preset line normal decision interval, it is determined that the line of the distribution network is in normal operation.

[0160] Figure 9 As can be seen, whether the collected feature vectors are put into the training set or the test set, the recognition rate of the method for the high-resistance fault reaches 100%, and only 5 recognition errors are made in 3600 samples, and the total accuracy of the model reaches 99.86%, which shows that the evaluated support vector machine model has a high recognition accuracy for the high-resistance fault.

[0161] In the embodiment of the present example, the evaluated support vector machine model is regarded as a complex classification model or a black box model. For a black box model, it is usually difficult to fit it with a simple and interpretable model from a global perspective, but some local samples in the model can be fitted with a simple model, so the local part of the black box model can be explained according to the local interpretability analysis algorithm.

[0162] First, the original sample features that need to be explained are sampled after N perturbations.

[0163] Second, the sampling points and the distance of the points to be explained are weighted.

[0164] Third, a simple classification model is constructed to classify the samples obtained by N perturbations.

[0165] ​​Step 4. Then, the estimated importance of the predictive variables is displayed in the simple model using the object function.

[0166] Among different working condition identification, it can be seen from the preset variable analysis report of the support vector machine model that the black box model has high importance of the predictive variables, whether identifying high resistance faults or various disturbance working conditions, the time domain features (transient signal amplitude mean ) and the frequency domain features (variational modal energy entropy H EVMD , variational modal energy distribution p i ) are all predictive variables with high importance, and the prediction results of the simple decision tree model and the black box model are consistent. In the analysis report of the influence of fault scenarios on detection effect, it can be seen that when high resistance faults occur at different positions of power cables and overhead lines, the importance of each predictive variable in the SVM model for identifying high resistance faults does not change significantly, indicating that different fault positions of different lines have no significant effect on the detection effect.

[0167] It should be noted that although the steps of the method in the present disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in this specific order, or that all the steps shown must be performed to achieve the desired results. In addition or alternatively, some steps can be omitted, multiple steps can be combined into one step, and / or one step can be divided into multiple steps, etc.

[0168] In addition, in the present example embodiment, a power distribution network fault detection device based on variational modal decomposition is also provided. As shown in Figure 10 , the power distribution network fault detection device 400 based on variational modal decomposition can include a transient signal acquisition module 410, a variational modal decomposition module 420, an energy distribution proportion calculation module 430, an energy entropy value calculation module 440, a time domain signal amplitude mean calculation module 450, a feature vector construction module 460, and a fault detection module 470. Wherein:

[0169] The transient signal acquisition module 410 is configured to generate a transient signal by acquiring and calculating the transient part of the zero sequence voltage of the power distribution network according to the negative sequence current suppression control strategy;

[0170] The variational modal decomposition module 420 is configured to generate a modal component by performing variational modal decomposition on the transient signal according to a variational modal decomposition algorithm;

[0171] The energy distribution proportion calculation module 430 is configured to generate a variational modal energy distribution by calculating the energy distribution proportion of the modal component in the transient signal according to the orthogonality of variational modal decomposition;

[0172] The energy entropy calculation module 440 is used to generate the variational mode energy entropy by calculating the energy entropy value of the variational mode decomposition based on the modal energy distribution.

[0173] The time-domain signal amplitude mean calculation module 450 is used to generate the transient signal amplitude mean by calculating the amplitude mean of the time-domain signal of the zero-sequence voltage based on the transient signal.

[0174] The feature vector construction module 460 is used to construct feature vectors using the variational mode energy distribution, variational mode energy entropy, and transient signal amplitude mean, and to evaluate the support vector machine model using the feature vectors.

[0175] The fault detection module 470 is used to identify the transient characteristics of zero-sequence voltage using the evaluated support vector machine model, thereby completing the detection of distribution network faults.

[0176] The specific details of each of the above-mentioned distribution network fault detection device modules based on variational mode decomposition have been described in detail in the corresponding distribution network fault detection method based on variational mode decomposition, so they will not be repeated here.

[0177] It should be noted that although several modules or units of a distribution network fault detection device 400 based on variational mode decomposition have been mentioned in the detailed description above, this division is not mandatory. In fact, according to the embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0178] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0179] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0180] It should be understood that the present disclosure is not limited to the precise construction that has been described above and shown in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the present disclosure. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for detecting distribution network faults based on variational mode decomposition, characterized in that, include: Based on the negative sequence current suppression control strategy, transient signals are generated by collecting and calculating the transient portion of the zero sequence voltage in the distribution network. According to the variational mode decomposition algorithm, the transient signal is subjected to variational mode decomposition to generate modal components; Based on the orthogonality of variational mode decomposition, variational mode energy distribution is generated by calculating the proportion of energy distribution of the modal components in the transient signal; Based on the modal energy distribution, the variational modal energy entropy is generated by calculating the energy entropy value of the variational mode decomposition. Based on the transient signal, the average amplitude of the transient signal is generated by calculating the average amplitude of the time-domain signal of the zero-sequence voltage. Feature vectors are constructed using the variational mode energy distribution, variational mode energy entropy, and mean transient signal amplitude, and the support vector machine model is evaluated using these feature vectors. The evaluated support vector machine model is used to identify the transient characteristics of zero-sequence voltage, thereby enabling the detection of faults in the distribution network.

2. The method according to claim 1, characterized in that, By collecting and calculating the transient portion of the zero-sequence voltage in the distribution network, including: Based on the negative sequence current suppression control strategy, the circuit topology analysis of the zero sequence component during a fault is completed by calculating the reference values ​​of the positive sequence current components on the direct axis and cross axis of the inverter when a single-phase high-resistance ground fault occurs in the distribution network. Based on the circuit topology analysis of the zero-sequence component during the fault, a high-impedance fault equivalent circuit is constructed by setting the fault point as a virtual power supply. Based on the high-impedance fault equivalent circuit, a zero-sequence voltage signal is generated by acquiring and calculating the zero-sequence voltage in the circuit. Based on the zero-sequence voltage signal, the transient portion of the zero-sequence voltage is acquired through a preset time window to generate a transient signal.

3. The method according to claim 1, characterized in that, Variational mode decomposition of the transient signal includes: Based on the variational mode decomposition algorithm, a constraint variational problem for the transient signal is generated by constructing the constraint variational problem of the transient signal. Based on a preset distortion level limit, the Lagrange operator is solved by introducing a noise tolerance to generate the Lagrange operator.

4. The method according to claim 3, characterized in that, Variational mode decomposition of the transient signal includes: Based on the Lagrange algorithm, the constrained variational problem of the transient signal is transformed by introducing the Lagrange operator to generate an unconstrained variational problem; Using an iterative algorithm, a quadratic penalty factor is introduced to perform a Fourier isometry transform on the unconstrained variational problem, and the components of the modal signal decomposition are iterated. When the components of the modal signal decomposition are not less than the preset accuracy, the iteration is determined not to terminate; When the component of the modal signal decomposition is less than the preset precision, the iteration is terminated and the iteration result is output, generating modal components.

5. The method according to claim 1, characterized in that, A feature vector is constructed using the variational mode energy distribution, variational mode energy entropy, and mean transient signal amplitude, including: Using the variational mode energy distribution, variational mode energy entropy, and mean transient signal amplitude, feature vectors are generated by constructing features from the input samples of the support vector machine. Based on the feature vectors, the training set for the support vector machine model is prepared to generate a training dataset. Based on the feature vectors, a test set for the support vector machine model is prepared to generate a test dataset. The ratio of the training dataset to the test dataset is 2:

1.

6. The method according to claim 5, characterized in that, The support vector machine model is evaluated using the feature vectors, including: Based on the support vector machine model, the training dataset is mapped to a high-dimensional space, and a classification margin constraint problem is generated by using a hyperplane to classify the training dataset in the mapped high-dimensional space. Based on the Lagrange algorithm, the classification margin constraint problem is transformed by introducing Lagrange multipliers to generate a classification margin optimization problem; Based on the aforementioned classification margin optimization problem, the dual problem is obtained by taking the partial derivative of the Lagrange function and setting the result to zero.

7. The method according to claim 6, characterized in that, The support vector machine model is evaluated using the feature vectors, including: The parameters of the support vector machine model are calculated by solving the dual problem, thereby generating the support vector machine model parameters. Based on the parameters of the support vector machine model, the decision function is constructed, and the test dataset is introduced into the decision function. By solving the decision function, the support vector machine model is evaluated.

8. The method according to claim 1, characterized in that, The evaluated support vector machine model is used to identify transient characteristics of zero-sequence voltage, thereby enabling the detection of distribution network faults, including: The transient characteristics of the zero-sequence voltage are used as input data and input into the evaluated support vector machine model. The decision function is then solved to generate a decision result. When the decision result falls within the preset high-resistance fault decision range, it is determined that a high-resistance grounding fault has occurred in the distribution network. When the decision result falls within the preset line interference decision range, it is determined that the distribution network is interfered with; When the decision result falls within the preset normal decision range for the line, the distribution network line is determined to be operating normally.

9. The method according to any one of claims 1-8, characterized in that, Also includes: By utilizing the local interpretability analysis algorithm, an interpretability analysis of the trained support vector machine model is performed to generate a pre-defined variable analysis report for the support vector machine model for work condition identification. By utilizing the local interpretability analysis algorithm, an analysis report on the impact of fault scenarios on detection effectiveness is generated through adaptive analysis of fault scenarios.

10. A distribution network fault detection device based on variational mode decomposition, characterized in that, include: The transient signal acquisition module is used to generate transient signals by acquiring and calculating the transient portion of the zero-sequence voltage of the distribution network according to the negative sequence current suppression control strategy. The variational mode decomposition module is used to generate modal components by performing variational mode decomposition on the transient signal according to the variational mode decomposition algorithm. The energy distribution weight calculation module is used to generate the variational mode energy distribution by calculating the energy distribution weight of the modal components in the transient signal based on the orthogonality of the variational mode decomposition. The energy entropy calculation module is used to generate the variational mode energy entropy by calculating the energy entropy value of the variational mode decomposition based on the modal energy distribution. The time-domain signal amplitude mean calculation module is used to generate the transient signal amplitude mean by calculating the amplitude mean of the time-domain signal of the zero-sequence voltage based on the transient signal. The feature vector construction module is used to construct feature vectors using the variational mode energy distribution, variational mode energy entropy, and transient signal amplitude mean, and to evaluate the support vector machine model using the feature vectors. The fault detection module is used to identify the transient characteristics of zero-sequence voltage using the evaluated support vector machine model, thereby completing the detection of faults in the distribution network.

Images