Method for identifying a fault arc based on correlation of model parameters

Abstract

The method for identifying fault arc based on model parameter correlation degree comprises the following steps: detecting current information of the fault arc and inputting to an arc feature identification module; the arc feature identification module identifies valid information and inputs to a fault arc parameter reconstruction module, which reconstructs parameters based on an adaptive fault arc model; the obtained reconstructed parameter set is input to a parameter correlation degree processing module for parameter correlation degree processing; the obtained reconstructed parameters and obtained correlation degree information are compared with an arc feature library caused by a fault type set in a power fault type processing module; and the power fault caused by the fault arc is classified according to the comparison result. The method classifies the electrical fault type according to the difference of the fault arc feature and the correlation degree relationship of the reconstructed model parameters, and takes appropriate measures based on the fault type category to improve the operation and management level of the distribution network.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fault arc identification, and particularly relates to a fault arc identification method based on model parameter correlation degree. BACKGROUND

[0002] Fault arcs are mostly caused by the intrusion of humid air into the working environment of electrical equipment, loose connection of the post plug, sharp increase of voltage and current, and insulation damage or aging of electrical equipment or lines. Arcs can cause air gas discharge, the center of the arc is close to 3000℃, and there is high-temperature metal spatter, which is extremely easy to ignite surrounding combustible materials, thereby causing a fire, and even causing personnel injury accidents. Low-voltage distribution lines in densely populated areas have complex operating conditions, and the scale and length are very long, so it is difficult to implement timely and effective operation and maintenance management. Electrical accidents caused by fault arcs occur frequently.

[0003] Although various reasons can cause fault arcs, different electrical faults cause different fault arcs, and different electrical fault arcs can cause different electrical accident consequences and take different measures in accident management. Therefore, if different electrical faults, such as line aging or line insulation damage, poor electrical connection, and long-time overload operation of the line, can be accurately distinguished, corresponding early warning and processing measures can be implemented, that is, according to the monitored fault arc characteristics, the electrical fault category causing the fault arc can be distinguished, the expansion or occurrence of the electrical fault can be prevented, the operation risk of the power system can be reduced, and the operation and maintenance level of the entire power system can be improved.

[0004] The relatively accurate arc model of the fault arc includes the "control theory" model, the Schwarz model, and the Mayr model. Based on the fault arc model, the arc characteristics can be obtained by detecting the arc current or the arc voltage, thereby completing the identification of the fault arc. However, the measurement of the arc current or the arc voltage is relatively complex, and the fault arc is dynamic and easily affected by complex environmental factors, so it is difficult to identify. The existing arc detection method and technology have low identification accuracy of the fault arc, and there is no application of distinguishing the corresponding power fault through the fault arc characteristics, classifying the power fault, and taking measures to improve the operation and maintenance management level of the distribution network. SUMMARY

[0005] The application provides a fault arc recognition method based on model parameter correlation, which is based on a relatively accurate fault arc model, and the difference of fault arc characteristics is recognized and distinguished according to the correlation of main parameters of the fault arc model, so that the recognition accuracy of the fault arc is not high. At the same time, the method classifies the electrical fault type according to the difference of the fault arc characteristics and the correlation of the reconstructed model parameters, that is, the fault type is counted in the form of fault arc, and appropriate measures are taken based on the fault type to improve the operation and management level of the distribution network.

[0006] The technical scheme adopted by the application is:

[0007] The fault arc recognition method based on model parameter correlation comprises the following steps:

[0008] Step 1: A current sensor detects the current information of the fault arc and inputs the information to an arc characteristic recognition module through a coupler.

[0009] Step 2: The arc characteristic recognition module recognizes the effective information and inputs the information to a fault arc parameter reconstruction module, and the fault arc parameter reconstruction module reconstructs parameters based on a self-adaptive fault arc model.

[0010] Step 3: The reconstructed parameter set obtained in step 2 is input to a parameter correlation processing module for parameter correlation processing.

[0011] Step 4: The reconstructed parameters obtained in step 2 and the correlation information obtained in step 3 are compared with the arc characteristic library caused by the fault type set in the power fault type processing module.

[0012] Step 5: According to the comparison result in step 4, the power fault caused by the fault arc is classified.

[0013] In step 1, a current sensor is arranged to detect the effective information data of the fault arc, and the effective information data is normalized to eliminate the adverse effects of abnormal data.

[0014] The specific normalization method is shown in formula (1):

[0015]

[0016] In the formula, x * is the normalized information data, x is the original information data to be normalized, x max is the maximum value of the original information data, and x min is the minimum value of the original information data.

[0017] In step 2, the normalized information data x *The parameters are input into the fault arc parameter reconstruction module and reconstructed based on the adaptive fault arc model.

[0018] Includes the following steps:

[0019] S2.1: First, the normalized information data x * The sequence is decomposed into multiple groups according to its center frequency, i.e., variable center frequency decomposition, thereby transforming the information data x * It can be decomposed into multiple variable center frequency components, as shown in the following equation:

[0020] x * (x i )=k i J i (x i -s i (1);

[0021] In the formula: k i J is the reconstruction function of the i-th frequency function. i The coefficient of (x), s i Let be the center wavelength of the i-th frequency function, and be the difference between the upper and lower limits of the frequency function.

[0022] In this way, the characteristics of the original information data can be preserved, while the adverse effects of aliasing of multi-frequency component characteristics on the prediction results can be reduced.

[0023] S2.2: Selecting an appropriate number of variable center frequency component decompositions, n, is crucial for the algorithm. Too few decompositions result in aliasing of multi-frequency component characteristics, while too many decompositions increase noise disturbances. In the method of this invention, the value of n ranges from 3 to 7.

[0024] S2.3: When constructing the prediction model results, since redundant noise has been discarded, the fault arc reconstruction model can be obtained by directly summing the results according to the unit coefficients.

[0025] The unit coefficient is designed as follows:

[0026]

[0027] In the formula, k i Let be the reconstruction coefficients of the i-th frequency function.

[0028] The general model for fault arc reconstruction is as follows:

[0029]

[0030] In the formula, the general model for fault arc reconstruction is the reconstruction function J for all frequency bands. i (x) is a weighted sum, with weighting coefficients k. i .

[0031] The adaptive fault arc model is:

[0032]

[0033] In the formula, the adaptive fault arc model is a frequency segment reconstruction function J with a unit coefficient i (x) weighted superposition and.

[0034] In step 3, parameter correlation degree processing is performed, and the specific process is as follows:

[0035] According to the correlation function, the fitting degree value D of the reconstructed parameter and the existing parameter is calculated 2 to determine the correlation degree between the sampled current fitting model and the existing fault arc model, D 2 The smaller the value is, the higher the correlation degree is, and when D 2 is less than 1, it is determined that the sampled current fitting model has occurred consistent fault with the existing fault arc model.

[0036]

[0037] In formula (5), k i represents the number of measured current parameter samples, k r represents the fitting model parameter value, k 2 represents the existing actual model parameter value, and k 2 represents the existing actual model parameter average value.

[0038] In step 4, the arc feature library is mainly the fault features contained in the existing fault arc model. The fault features contained in the existing fault arc model mainly include: ① sharp pulse; ② multi-peak becomes single-peak, and the peak value of the single-peak is larger than that of the multi-peak; ③ frequent fluctuation of fault current amplitude, etc.

[0039] In step 4, the measured current fitting model formed by the reconstructed parameters obtained in step 2 and the correlation degree information obtained in step 3 is compared with the existing power fault arc model, and the parameter fitting degree value D 2 is calculated, and the smaller the value is, the higher the correlation degree is, and when D 2 is less than 1, it is determined that the sampled current fitting model has occurred consistent fault with the existing fault arc model.

[0040] The fault arc recognition method based on model parameter correlation degree has the following technical effects:

[0041] The present application is based on a relatively accurate fault arc model, and the existing fault arc model contains relatively comprehensive arc features. Then, the measured current is fitted, and the fitted model is compared with the existing power fault arc model to calculate the fitting degree value of the main parameters, so that the characteristics of the fault arc current are converted into the fitting degree value D2 the sampled current fitting model and the existing fault arc model is D 2 The smaller the value is, the higher the correlation is, and when D 2 is less than 1, it is determined that the sampled current fitting model has a fault consistent with the existing fault arc model. Therefore, the identification of the fault arc no longer relies on the calculation and selection of a complex feature set, thereby improving the identification accuracy of the fault arc current. BRIEF DESCRIPTION OF DRAWINGS

[0042] Figure 1 is a flow chart of the present application.

[0043] Figure 2 is a flow chart of the fault arc reconstruction model.

[0044] Figure 3 is a flow chart of the adaptive fault arc model reconstruction. DETAILED DESCRIPTION

[0045] The reconstruction flow chart is shown as above Figure 2 :

[0046] From the data in Table 1, the information data x * The center frequencies of the components are slightly different but the overall trend is basically the same. When the decomposition number n≥5, the center frequencies of the latter variable center frequency components are close, such as i4 and i5 when n=5, which are easily mixed and unclear, and therefore the n value is selected to be 4.

[0047] Table 1 Center frequencies of each component when the decomposition number n is different

[0048]

[0049]

[0050] According to the above algorithm flow, after obtaining the fault arc reconstruction model, the reconstruction parameter set is selected and input to the parameter correlation processing module for parameter correlation processing, and then fault arc determination is performed:

[0051] First, the measured current is model fitted;

[0052] Then, the fitting model and the existing power fault arc model are fitted for the main parameter fitting degree value calculation, thereby converting the fault arc current feature discrimination into the calculation of the fitting degree value D 2 .

[0053] Finally, fault arc determination is performed. When the correlation D 2 between the sampled current fitting model and the existing fault arc model is D 2If the DTSJWQ value is less than 1, it is determined that the sampled current fitting model has occurred consistent with the existing fault arc model of the fault.

[0054] The DTSJWQ value is calculated by using the DTSJWQ algorithm to analyze the correlation of the parameter set corresponding to the existing fault arc library model and the fault arc reconstruction model. 2 Numerical value.

[0055] The fault arc reconstruction uses formula (3) and formula (4);

[0056] The correlation analysis is performed, and the DTSJWQ value is calculated, that is, D 2 The numerical value is calculated by using formula (5)

[0057] Then, according to the size of the DTSJWQ value, the correlation degree of the fault arc reconstruction model and the existing fault arc library model is compared. The specific determination is as follows:

[0058] When the correlation degree D 2 The smaller the numerical value is, the higher the correlation degree between the two models is. When D 2 If the DTSJWQ value is less than 1, it is determined that the sampled current fitting model has occurred consistent with the existing fault arc model; when there are multiple D 2 values less than 1, the smallest D 2 value is used to determine that the fitting model is consistent with the existing fault arc model, that is, the fault consistent with the existing fault arc model has occurred at the current sampling point.

[0059] Finally, fault identification is performed: the one with the smallest DTSJWQ value is the fault arc reconstruction model with the highest correlation degree with the existing fault arc, and the closest feature. When the DTSJWQ value is less than 1, it is determined that a fault arc has occurred.

[0060] The DTSJWQ example value is calculated as shown in Table 2:

[0061] Table 2 DTSJWQ value of fault arc reconstruction model

[0062]

[0063]

[0064] From the fault arc reconstruction model in Table 2 and the existing fault arc library model, a total of m fault models, in actual application, the representative fault model can be selected according to the application scene. The correlation DTSJWQ value of each fault model and the fault arc reconstruction model is calculated, and then the required value is selected for judgment. For example, as shown in Table 2: after comparison, it is found that the DTSJWQ value calculated by the correlation of library 1 is the smallest, so the fault arc belongs to the type 1 in the existing fault arc library model. After processing by the parameter correlation processing module, the existing fault arc type is matched for the fault arc, and then transmitted to the power fault type processing module.

Claims

1. A fault arc identification method based on model parameter correlation, characterized in that... Includes the following steps: Step 1: Detect the current information of the fault arc and input it into the arc feature recognition module; Step 2: The arc feature recognition module identifies valid information and inputs it into the fault arc parameter reconstruction module. The fault arc parameter reconstruction module reconstructs parameters based on the adaptive fault arc model. Step 3: Input the reconstructed parameter set obtained in Step 2 into the parameter correlation processing module for parameter correlation processing; Step 4: Compare the reconstruction parameters obtained in Step 2 and the correlation information obtained in Step 3 with the arc feature library caused by the fault type set in the power fault type processing module; Step 5: Based on the comparison results in Step 4, classify the power faults caused by the fault arc; In step 3, parameter correlation processing is performed, as follows: The goodness-of-fit value between the reconstructed parameters and the existing parameters is determined using the relevant function. To determine the correlation between the sampled current fitting model and the existing fault arc model, The smaller the value, the higher the correlation. ； In the formula, represents the number of samples of the measured current parameter. This represents the parameter values ​​of the fitted model. This represents the existing actual model parameter values. This represents the average value of the existing actual model parameters.

2. The fault arc identification method based on model parameter correlation according to claim 1, characterized in that: In step 1, a current sensor is set up to detect valid information data of the fault arc, and the valid information data is normalized to eliminate the adverse effects of outlier data. The specific normalization method is shown in formula (1): ； In the formula, For normalized information data; For the raw information data that needs to be normalized, The maximum value in the original information data. It is the minimum value in the original information data.

3. The fault arc identification method based on model parameter correlation according to claim 2, characterized in that: In step 2, the information data after normalization processing The parameters are input into the fault arc parameter reconstruction module and reconstructed based on the adaptive fault arc model. Includes the following steps: S2.1: First, the normalized information data The sequence is decomposed into multiple groups according to the center frequency, thereby separating the information data. It can be decomposed into multiple variable center frequency components, as shown in the following equation: （1）； In the formula: For the first Reconstruction function of a frequency function coefficient, For the first The center wavelength of a frequency function is the difference between the upper and lower limits of the frequency function; S2.2: Select an appropriate number of variable center frequency component decompositions n ; S2.3: The fault arc reconstruction model can be obtained by summing the results using unit coefficients; The unit coefficient is designed as follows: （2）； In the formula, For the first Reconstruction coefficients of a frequency function; The general model for fault arc reconstruction is as follows: （3）； In the formula, the general model for fault arc reconstruction is the reconstruction function for all frequency bands. Weighted summation, with weighting coefficients as follows: ; The adaptive fault arc model is as follows: （4）； In the formula, the adaptive fault arc model is a frequency band reconstruction function with unit coefficients. Weighted summation.

4. The fault arc identification method based on model parameter correlation according to claim 1, characterized in that: In step 4, the measurement current fitting model formed by the reconstructed parameters obtained in step 2 and the correlation information obtained in step 3 is compared with the existing power fault arc model to calculate the degree of parameter fitting. The smaller the value, the higher the correlation. If the value is less than 1, it is determined that the sampled current fitting model has a fault consistent with the existing fault arc model.

Images