Power distribution network permanent fault identification method and system

Abstract

The invention provides a power distribution network permanent fault identification method and system, and is suitable for three-phase reclosing permanent fault discrimination of a photovoltaic access power distribution network, and the method comprises the steps: collecting voltage and current data of a photovoltaic access power distribution network line through an online monitoring device, and transmitting the voltage and current data to a data storage module in real time; performing data preprocessing by adopting Kalman filtering to form a database; a line instantaneous fault is used as a reference model, an improved Bergeron model of a line is established, and photovoltaic low-voltage ride-through capability and nonlinear characteristics of power electronic equipment are considered; identifying the inter-phase capacitance parameter of the line by adopting a least square algorithm, and obtaining a capacitance parameter value of the line in the current state; according to the error between the identified capacitance parameter value and the actual capacitance value, the line fault type is judged, when the error is small, the fault is judged to be an instantaneous fault, and when the error is large, the fault is judged to be a permanent fault; according to the identified fault type, a corresponding protection action is automatically triggered, reclosing is allowed for an instantaneous fault, a line is cut off for a permanent fault, and the occurrence of a secondary fault is avoided. According to the fault identification method provided by the invention, a stable fault identification effect can still be kept in an environment with relatively high photovoltaic permeability, and the fault identification method can adapt to a power distribution network scene containing a large number of distributed power supplies in the future.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of permanent fault identification of power distribution network, and particularly relates to a permanent fault identification method and system of power distribution network. BACKGROUND

[0002] The reclosing of permanent fault of power distribution network will seriously affect the safe and stable operation of power grid. The transient fault is usually caused by external temporary disturbance (such as lightning, wind blowing tree branches contacting the line, etc.), and the line can recover itself after a short time, which will not cause long-term impact on the system. However, the permanent fault is usually caused by line breakage, equipment damage, etc. (such as lightning caused by phase loss), which cannot be blindly reclosed and needs manual intervention. However, the access of photovoltaic power generation and other distributed power sources changes the electrical characteristics of the power distribution network, which significantly increases the difficulty of permanent fault identification of the power distribution network.

[0003] In recent years, researchers have carried out a lot of research on adaptive reclosing and fault type diagnosis. These methods mainly focus on two aspects: one is based on mathematical model; the other is to introduce intelligent algorithm and data-driven method, through the analysis of a large number of electrical data before and after the fault, the fault type is distinguished by using machine learning or optimization algorithm. The above research improves the accuracy of fault identification to a certain extent, but also has problems such as high computational complexity and poor real-time performance. SUMMARY

[0004] The technical problem to be solved by the present application is how to provide a fault identification method with high real-time performance and accuracy for the complex environment of photovoltaic access to power distribution network.

[0005] The present application solves the above technical problems by the following technical means: a permanent fault identification method of power distribution network, comprising the following steps:

[0006] S1, collecting voltage and current data of the line of photovoltaic access to power distribution network through an online monitoring device, and transmitting the data to a data processing center in real time, and adopting Kalman filtering for data preprocessing to form a database;

[0007] S2, taking the line transient fault as a reference model, establishing an improved Bergeron model of the line, and considering the low voltage ride-through capability of photovoltaic and the nonlinear characteristics of power electronic equipment;

[0008] S3, using least square algorithm to identify the inter-phase capacitance parameters of the line to obtain the capacitance parameter value of the line in the current state;

[0009] S4, judging the line fault type according to the error between the identified capacitance parameter value and the actual capacitance value, and determining as transient fault when the error is small, and determining as permanent fault when the error is large;

[0010] S5, according to the identified fault type, automatically trigger the corresponding protection action, for transient fault allows reclosing, for permanent fault cut off the line.

[0011] As a further optimized technical solution, the step S1 is specifically:

[0012] S11, install voltage and current sensors on the line where the photovoltaic system accesses the power distribution network, and collect real-time three-phase voltage and current data;

[0013] S12, the collected data is transmitted to the data storage module through the communication module, and the distributed file system Apache Hadoop is used to integrate and store the collected data;

[0014] S13, the collected data is cleaned by using Kalman filtering method to eliminate noise and error;

[0015] S14, using the latest state estimation of Kalman filter as the cleaned data, all types of information of the same line are classified to form a database.

[0016] As a further optimized technical solution, the step S13 uses Kalman filtering method to clean the collected data to eliminate noise and error, which is specifically:

[0017] Set the estimated value of the initial data Corresponding to voltage data and current data respectively; calculate the covariance matrix P0 of the initialization estimation error; data state prediction: The estimated value of the tth type of data for the kth iteration, F t The state matrix of the tth type of data, B t The control matrix of the tth type of data, The control variable of the tth type of data for the kth iteration; error covariance prediction: The error covariance of the tth type of data for the kth iteration, Q t The covariance matrix of the observation noise of the tth type of data, (F t ) T The transpose of the state matrix of the tth type of data; calculate the Kalman gain: The Kalman gain of the tth type of data for the kth iteration, H t The measurement matrix of the tth type of data, R t The covariance matrix of the measurement noise; update the state estimation: The actual value of the tth type of data for the kth iteration; update the error covariance:

[0018] As a further optimized technical solution, the step S2 is specifically:

[0019] The basic principle of the Bergeon model is represented by the wave equation:

[0020] V(x, t) = V(x - ct, t) + Z c · I(x - ct, t)

[0021]

[0022] Where V(x, t) and I(x, t) are the voltage and current of the line at position x, c is the wave speed, Z c is the characteristic impedance of the line, the propagation speed v of the line and the characteristic impedance Z c is represented as:

[0023]

[0024] Where L is the distributed inductance of the line, and C is the interphase capacitance.

[0025] The improved representation considering the low voltage ride-through capability of photovoltaics is:

[0026] V LVRT (t) = f(V(t), t LVRT , a)

[0027] Where V LVRT (t) represents the voltage response considering low voltage ride-through, t LVRT is the ride-through time parameter, and a is the adjustment factor. The introduction of low voltage ride-through can better simulate the dynamic behavior of photovoltaic systems under low voltage conditions.

[0028] The improved representation considering the nonlinear characteristics of power electronic devices is, in the traditional Bergeon model, by introducing a nonlinear impedance Z nonlinear (I) to simulate the nonlinearity of power electronic devices:

[0029] Z nonlinear (I) = Z0 + k · I n

[0030] Where Z0 is the initial impedance; k and n are nonlinear parameters.

[0031] As a further optimized technical solution, the step S3 is specifically:

[0032] The least squares algorithm is used to identify the interphase capacitance parameters of the line:

[0033]

[0034] Where C ijCij is the line-to-line capacitance between lines i and j, V ij (t) and I ij (t) are the line-to-line voltage and current, respectively, and the line-to-line capacitance is estimated effectively by least square fitting of data at multiple time points; the objective function of the least square method is:

[0035]

[0036] wherein, is the voltage estimate, and N is the number of data points.

[0037] As a further optimized technical solution, the step S4 is specifically:

[0038] The error ò and the permanent fault criterion are defined as:

[0039]

[0040] If ò>threshold, it is determined as a permanent fault, wherein threshold is a pre-set error threshold.

[0041] The application also provides a power distribution network permanent fault identification system, comprising the following modules:

[0042] A data processing module is used to collect voltage and current data of a photovoltaic access power distribution network line through an online monitoring device, and transmit the data to a data processing center in real time, and perform data preprocessing by using Kalman filtering to form a database;

[0043] An improved Bergeron model establishment module is used to establish an improved Bergeron model of the line by taking a line transient fault as a reference model, and consider the low voltage ride-through capability of the photovoltaic and the nonlinear characteristics of the power electronic equipment;

[0044] An identification module is used to identify the line-to-line capacitance parameters of the line by using a least square algorithm, and obtain the capacitance parameter value of the line in the current state;

[0045] A comparison module is used to determine the line fault type according to the error between the identified capacitance parameter value and the actual capacitance value, and determine a transient fault when the error is small, and determine a permanent fault when the error is large;

[0046] A triggering module is used to automatically trigger a corresponding protection action according to the identified fault type, and allow reclosing for a transient fault and cut off the line for a permanent fault.

[0047] As a further optimized technical solution, the processing process of the data processing module specifically includes:

[0048] S11, install voltage and current sensors on the line where the photovoltaic system is connected to the power distribution network, and collect three-phase voltage and current data in real time;

[0049] S12, transmit the collected data to the data storage module through the communication module, and use the distributed file system Apache Hadoop to integrate and store the collected data;

[0050] S13, use the Kalman filter method to clean the collected data and eliminate noise and errors;

[0051] S14, use the latest state estimation of the Kalman filter as the cleaned data, and classify all types of information of the same line to form a database.

[0052] As a further optimized technical solution, the step S13 uses the Kalman filter method to clean the collected data and eliminate noise and errors, which specifically comprises:

[0053] Set the estimated value of the initial data Corresponding to the voltage data and the current data respectively; calculate the covariance matrix P0 of the initialization estimation error; data state prediction: The estimated value of the kth iteration of the tth data, F t The state matrix of the tth data, B t The control matrix of the tth data, The control variable of the kth iteration of the tth data; error covariance prediction: The error covariance of the kth iteration of the tth data, Q t The covariance matrix of the observation noise of the tth data, (F t ) T The transpose of the state matrix of the tth data; calculate the Kalman gain: The Kalman gain of the kth iteration of the tth data, H t The measurement matrix of the tth data, R t The covariance matrix of the measurement noise; update the state estimation: The actual value of the kth iteration of the tth data; update the error covariance:

[0054] As a further optimized technical solution, the specific process of establishing the improved Bergeron model in the improved Bergeron model establishment module comprises:

[0055] The basic principle of the Bergeron model is represented by the wave equation:

[0056] V(x, t) = V(x - ct, t) + Z c • I(x - ct, t)

[0057]

[0058] where V(x, t) and I(x, t) are the voltage and current of the line at position x, c is the wave speed, Z c is the characteristic impedance of the line, the propagation speed v of the line and the characteristic impedance Z c are expressed as:

[0059]

[0060] where L is the distributed inductance of the line, C is the inter-phase capacitance;

[0061] The improved representation considering the low voltage ride-through capability of photovoltaic is:

[0062] V LVRT (t) = f(V(t), t LVRT , a)

[0063] where V LVRT (t) represents the voltage response considering low voltage ride-through, t LVRT is the ride-through time parameter, and a is the adjustment factor. The introduction of low voltage ride-through can better simulate the dynamic behavior of photovoltaic system under low voltage conditions;

[0064] The improved representation considering the nonlinear characteristics of power electronic equipment is, in the traditional Bergeon model, by introducing the nonlinear impedance Z nonlinear (I) to simulate the nonlinear characteristics of power electronic equipment:

[0065] Z nonlinear (I) = Z0 + k · I n

[0066] where Z0 is the initial impedance; k and n are nonlinear parameters.

[0067] As a further optimized technical solution, in the step identification module:

[0068] The least square algorithm is used to identify the inter-phase capacitance parameters of the line:

[0069]

[0070] where C ij is the capacitance between phase i and j of the line, V ij (t) and I ij(t) the voltage and current between phases, respectively, the line-to-line capacitance of the line is effectively estimated by least square fitting of data at multiple time points; the objective function of the least square method is:

[0071]

[0072] wherein, is the voltage estimate, and N is the number of data points;

[0073] In the comparison module:

[0074] The error ò and the permanent fault criterion are defined as:

[0075]

[0076] If ò>threshold, it is determined as a permanent fault, wherein the threshold is a pre-set error threshold.

[0077] The advantages of the present application are:

[0078] (1) The pre-processing of the original data can highly restore the real data, has strong robustness in the case of partial data anomaly, and can deal with the problem of incomplete data in actual engineering;

[0079] (2) The improved Bergeron model effectively calculates the transient data of the transmission line, and provides a basis for the subsequent calculation of the capacitance;

[0080] (3) The least square algorithm is used to identify the line-to-line capacitance parameter, which can effectively reduce the influence of noise on the identification result, and ensure that the identification result of the line-to-line capacitance is more accurate;

[0081] (4) The results show that the method proposed in the present application can effectively identify transient and permanent faults under different fault positions (such as near the transformer outlet or near the end of the line), and has good generalization ability; BRIEF DESCRIPTION OF DRAWINGS

[0082] Figure 1 is a flow chart of a power distribution network permanent fault identification method according to an embodiment of the present application;

[0083] Figure 2 is a 10kV power distribution network model diagram. DETAILED DESCRIPTION

[0084] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0085] Referring to Figure 1 The present application provides a power distribution network permanent fault identification method, comprising the following steps:

[0086] S1, collecting voltage and current data of a photovoltaic access power distribution network line through an online monitoring device, and transmitting the data to a data processing center in real time, performing data preprocessing by using Kalman filtering, forming a database, and specifically:

[0087] S11, installing voltage and current sensors on the line of the photovoltaic system access power distribution network, and collecting three-phase voltage and current data in real time, with a collection frequency of 0.1 seconds / time;

[0088] S12, transmitting the collected data to a data storage module through a communication module, and using a distributed file system Apache Hadoop to integrate and store the collected data;

[0089] S13, using Kalman filtering method to clean the collected data, and eliminating noise and errors, and specifically:

[0090] Setting the estimated value of the initial data corresponding to the voltage data and the current data respectively; calculating the covariance matrix P0 of the initialization estimation error; data state prediction: is the estimated value of the kth iteration of the tth data, F t is the state matrix of the tth data, B t is the control matrix of the tth data, is the control variable of the kth iteration of the tth data; error covariance prediction: is the error covariance of the kth iteration of the tth data, Q t is the covariance matrix of the observation noise of the tth data, (F t ) T transpose of the state matrix of the tth data; calculating the Kalman gain: is the Kalman gain of the kth iteration of the tth data, H t is the measurement matrix of the tth data, R t is the covariance matrix of the measurement noise; updating the state estimation: Actual value of the kth iteration of the tth type of data; update error covariance:

[0091] S14, adopt the latest state estimation of Kalman filter as the cleaned data, and classify all types of information of the same line to form a database.

[0092] Through the step of preprocessing the original data, the real data can be highly restored.

[0093] S2, taking the line transient fault as the reference model, an improved Bergeron model of the line is established, considering the low voltage ride-through capability of photovoltaic and the nonlinear characteristics of power electronic equipment:

[0094] Bergeron model is a transient analysis model of transmission line based on wave theory, which describes the propagation behavior of voltage and current wave in transmission line. Bergeron model is based on transmission line equation, assuming that the electrical characteristics of transmission line are uniformly distributed. By modeling the propagation of voltage and current wave on transmission line, Bergeron model can calculate the evolution of wave in time and space, and describe how voltage and current propagate after fault occurs. Its basic principle can be represented by wave equation:

[0095] V(x,t)=V(x-ct,t)+Z c ·I(x-ct,t)

[0096]

[0097] Where, V(x,t) and I(x,t) are the voltage and current of the line at position x, c is the wave speed, Z c is the characteristic impedance of the line. The propagation speed v and characteristic impedance Z c of the line are represented as:

[0098]

[0099] Where, L is the distributed inductance of the line, and C is the interphase capacitance; traditional Bergeron model can well describe the voltage and current fluctuation on the line, but its accuracy will be affected under the condition of distributed photovoltaic system connected to distribution network; the low voltage ride-through capability of photovoltaic system and the nonlinear characteristics of power electronic equipment may cause large error in fault analysis of traditional Bergeron model;

[0100] The improved representation considering the low voltage ride-through capability of photovoltaic is:

[0101] V LVRT (x,t)=f(V(x,t),t LVRT ,α)

[0102] where V LVRT (x,t) represents the voltage response considering low voltage ride through, t LVRT is the ride through time parameter, and a is the adjustment factor. The introduction of low voltage ride through can better simulate the dynamic behavior of photovoltaic systems under low voltage conditions.

[0103] The improvement considering the nonlinear characteristics of power electronic devices is represented as follows: in the traditional Bergeon model, a nonlinear impedance Z nonlinear (I) is introduced to simulate the nonlinearity of power electronic devices.

[0104] Z nonlinear (I) = Z c +k·I n

[0105] where k and n are nonlinear parameters.

[0106] S3, the least squares algorithm is used to identify the line-to-line capacitance parameters to obtain the capacitance parameter value of the line in the current state. In actual power distribution network operation, the measurement data of voltage and current will be disturbed by external environment, such as noise, equipment precision limitation, etc., resulting in errors or deviations in measurement data. The least squares method can effectively reduce the influence of noise on the identification result by minimizing the sum of squares of errors between the measured value and the model calculated value, ensuring that the identification result of line-to-line capacitance is more accurate.

[0107]

[0108] where C ij is the capacitance between lines i and j, V ij (t) and I ij (t) are the voltages and currents between lines, respectively. By least squares fitting of data at multiple time points, the line-to-line capacitance can be effectively estimated; the objective function of the least squares method is:

[0109]

[0110] where V is the voltage estimate, and N is the number of data points.

[0111] S4, according to the error between the identified capacitance parameter value and the actual capacitance value , the line fault type is judged. When the identified capacitance parameter value is close to the actual capacitance value , it is determined that the line has a transient fault, and when the identified capacitance parameter value deviates from the actual capacitance value by a large margin, it is determined that the line has a permanent fault. The error ò and the permanent fault criterion are defined as follows:

[0112]

[0113] If ò>threshold, it is determined as a permanent fault, wherein the threshold is a preset error threshold.

[0114] S5, according to the identified fault type, automatically triggering the corresponding protection action, for transient fault, allowing reclosing, for permanent fault, cutting off the line, avoiding the occurrence of secondary fault.

[0115] Embodiments:

[0116] The present application uses PSCAD simulation platform to build a 10kV distribution network model, the overhead line impedance is 0.12+j0.45Ω / km, the cable impedance is 0.10+j0.17Ω / km, the transition resistance is set to 5Ω, the model is as Figure 2 , the permanent fault is simulated to be persistent, the transient fault is persistent for 0.3s, in order to verify the effectiveness and generalization ability of the model, the results of four kinds of situations of different photovoltaic penetration, different fault positions, part of the collected data abnormal, different fault types are analyzed.

[0117] 3.1 Different PV penetration

[0118] With the construction of new power systems, the penetration rate of photovoltaic in the future will gradually increase, therefore, the present application sets five different photovoltaic penetration scenarios to verify the effectiveness and generalization ability of the model, the photovoltaic penetration rates are: 0%, 30%, 70%, 100% and 150%, the fault is set at the transformer outlet 0.5km, and the fault type is set as AB two-phase short circuit. The results are shown in Table 1.

[0119] Table 1 Identification results of different PV penetration

[0120]

[0121]

[0122] From the table, it can be seen that as the photovoltaic penetration rate increases from 0% to 150%, the fault identification model proposed by the present application can accurately identify at each penetration rate. The proposed model is not sensitive to the change of photovoltaic access ratio, and can adapt to the trend of increasing photovoltaic penetration rate in the future.

[0123] 3.2 Different fault positions

[0124] Since the fault location is uncertain, in order to verify the accuracy of the method in identifying different fault locations, four different fault locations are set, which are 0.5km, 1km, 5km and the end of the line from the transformer outlet, the photovoltaic penetration rate is 70%, and the fault type is set as AB two-phase short circuit. The results are shown in Table 2.

[0125] Table 2 Identification results of different fault locations

[0126]

[0127] It can be seen that in the case of different fault locations (such as near the transformer outlet or near the end of the line), the proposed model can effectively identify transient and permanent faults, verifying that the model has good generalization ability.

[0128] 3.3 Partial data anomaly

[0129] Further, considering that the actual voltage and current acquisition data may be missing or inaccurate, 10% of the abnormal data is set, including three cases: 10% of the data missing (case 1), 10% of the data inaccurate (case 2), 5% of the data missing + 5% of the data inaccurate (case 3). The photovoltaic penetration rate is 70%, and the fault is set at the transformer outlet 0.5km, AB two-phase short circuit, and the results are shown in Table 3.

[0130] Table 3 Identification results of partial data anomaly

[0131]

[0132] In the case of 10% data missing or data anomaly, the model can still accurately identify the fault type, indicating that the proposed model can reliably identify in the environment of incomplete or inaccurate data. This has important significance for improving the practical application value of fault identification.

[0133] 3.4 Different fault types

[0134] In addition, the present application sets four fault types: A-phase grounding fault (AG), AB two-phase short circuit (ABG), ABC three-phase short circuit (ABCG), and A-phase open-phase (AN), to verify the reliability of the model, the photovoltaic penetration rate is 70%, and the fault is set at the transformer outlet 0.5km, and the results are shown in Table 4.

[0135] Table 3 Identification results of different fault types

[0136]

[0137]

[0138] It can be seen that the model can accurately identify different types of faults such as single-phase grounding, two-phase short circuit, three-phase short circuit and open phase. However, for ABC three-phase short circuit fault, the identification effect of the model is not good. When the three-phase grounding fault occurs, the voltage and current change dramatically, but the change of the capacitor is not obvious, so further optimization should be carried out in this respect in future research. By improving the extraction accuracy of the three-phase short circuit fault characteristics and improving the adaptability of the model, the practicability of the model will be further enhanced.

[0139] The above examples are only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for identifying permanent faults in an electrical distribution network, characterized in that: The method comprises the following steps: S1, collecting voltage and current data of a line of a photovoltaic (PV) system connected to a power distribution network through an online monitoring device and transmitting the data to a data processing center in real time, performing data preprocessing by using Kalman filtering, and forming a database; S2, taking a transient fault of the line as a reference model, establishing an improved Bergeron model of the line, and considering low voltage ride-through capability of the PV system and nonlinear characteristics of power electronic equipment; S3, identifying inter-phase capacitance parameters of the line by using a least square algorithm to obtain capacitance parameter values of the line in a current state; S4, judging a fault type of the line according to an error between the identified capacitance parameter values and actual capacitance values, determining that the fault is a transient fault when the error is small, and determining that the fault is a permanent fault when the error is large; S5, automatically triggering a corresponding protection action according to the identified fault type, allowing reclosing for the transient fault, and cutting off the line for the permanent fault.

2. The method of claim 1, wherein: The step S1 specifically comprises: S11, installing voltage and current sensors on the line of the PV system connected to the power distribution network, and collecting three-phase voltage and current data in real time; S12, transmitting the collected data to a data storage module through a communication module, and integrating and storing the collected data by using a distributed file system Apache Hadoop; S13, cleaning the collected data by using a Kalman filtering method to eliminate noise and errors; S14, taking a latest state estimation of the Kalman filter as the cleaned data, and classifying all types of information of the same line to form a database.

3. The method of claim 2, wherein: The step S13 of cleaning the collected data by using the Kalman filtering method to eliminate noise and errors specifically comprises: Setting an initial data estimate , t = 1, 2 respectively correspond to voltage data and current data; Calculate the covariance matrix of the initial estimation error. P 0; Data status prediction: , For the first t Class Data k The estimated value of the next iteration. F t For the first t The state matrix of class data, B t For the first t Control matrix for class data, For the first t Class Data k Control variables for the next iteration; Error covariance prediction: , For the first t Class Data k Error covariance of the next iteration For the first t The covariance matrix of observation noise for class data. No. t Transpose of the state matrix of the class data; Calculate the Kalman gain: , For the first t Class Data k Kalman gain in the next iteration No. t Measurement matrix of class data, Measure the covariance matrix of the noise; update the state estimate: , No. t Class Data k The actual value of the next iteration; update the error covariance: .

4. The method of claim 1, wherein: The step S2 specifically comprises: The basic principle of the Bergeron model is represented by a wave equation: wherein V ( x , t ) and I ( x , t ) are the voltage and current of the line at positions x , c is the wave speed, Z c is the characteristic impedance of the line, the propagation speed v and the characteristic impedance Z c is given by: wherein, L L is the distributed inductance of the line, C C is the inter-phase capacitance; The improved representation considering the low voltage ride-through capability of the PV system is: wherein, V LVRT ( t ) represents the voltage response considering low voltage ride through, t LVRT is a ride through time parameter, α is an adjustment factor, the introduction of low voltage ride through enables a better simulation of the dynamic behavior of the photovoltaic system under low voltage conditions; An improved representation taking into account the non-linear characteristics of the power electronic device is introduced in the traditional Bergeon model by introducing a non-linear impedance Z nonlinear ( I ) to model the non-linearities of the power electronic device. wherein, Z 0 is the initial impedance; k and n is a non-linear parameter.

5. The method of claim 1, wherein: The step S3 specifically comprises: The inter-phase capacitance parameters of the line are identified by using the least square algorithm: in, C ij For the line i and j Interphase capacitors, V ij ( t )and I ij ( t The phase-to-phase voltage and current are represented by , respectively. By performing least-squares fitting on data from multiple time points, the phase-to-phase capacitance of the line is effectively estimated. The objective function of the least-squares method is: wherein is a voltage estimate, N is the number of data points.

6. The method of claim 1, wherein: The step S4 specifically comprises: Error and the permanent fault criterion is defined as: If >threshold then a permanent fault is determined, wherein threshold is a pre-defined error threshold.

7. A power distribution grid permanent fault identification system characterized by: The method comprises the following modules: A data processing module, configured to collect voltage and current data of a line of a photovoltaic (PV) system connected to a power distribution network through an online monitoring device, transmit the data to a data processing center in real time, perform data preprocessing by using Kalman filtering, and form a database; An improved Bergeron model establishing module, configured to take a transient fault of the line as a reference model, establish an improved Bergeron model of the line, and consider low voltage ride-through capability of the PV system and nonlinear characteristics of power electronic equipment; An identifying module, configured to identify inter-phase capacitance parameters of the line by using a least square algorithm to obtain capacitance parameter values of the line in a current state; A comparing module, configured to judge a fault type of the line according to an error between the identified capacitance parameter values and actual capacitance values, determine that the fault is a transient fault when the error is small, and determine that the fault is a permanent fault when the error is large; A triggering module, configured to automatically trigger a corresponding protection action according to the identified fault type, allow reclosing for the transient fault, and cut off the line for the permanent fault.

8. The power distribution grid permanent fault identification system of claim 7, wherein: The processing process of the data processing module specifically comprises: S11, installing voltage and current sensors on the line of the PV system connected to the power distribution network, and collecting three-phase voltage and current data in real time; S12, the collected data is transmitted to the data storage module through the communication module, and the distributed file system Apache Hadoop is used to integrate and store the collected data; S13, the collected data is cleaned by using the Kalman filtering method to eliminate noise and error; S14, the latest state estimation of the Kalman filter is used as the cleaned data, and all types of information of the same line are classified to form a database.

9. The power distribution grid permanent fault identification system of claim 8, wherein: The step S13 adopts the Kalman filtering method to clean the collected data, and specifically eliminates noise and error. Setting an initial data estimate , t = 1, 2 respectively correspond to voltage data and current data; Computing a covariance matrix of the initialization estimation error P 0; data state prediction: is the estimate of the state of the data of the first t class at the first iteration, k is the estimate of the state of the data of the first F class at the second iteration, t is the state matrix of the data of the first t class, B t is the control matrix of the data of the first t class, is the control variable of the data of the first t class at the first iteration, k error covariance prediction: is the error covariance of the data of the first t class at the first iteration, k is the error covariance of the data of the first class at the second iteration, t is the covariance matrix of the observation noise of the data of the first class, t is the transpose of the state matrix of the data of the first​ Computing the Kalman gain: , for the first iteration of the Kalman gain for the first class of data, t for the first iteration of the Kalman gain for the first class of data, k for the first iteration of the Kalman gain for the first class of data, for the first iteration of the Kalman gain for the first class of data, t for the first iteration of the Kalman gain for the first class of data, for the first iteration of the Kalman gain for the first class of data, , for the first iteration of the Kalman gain for the first class of data, t for the first iteration of the Kalman gain for the first class of data, k for the first iteration of the Kalman gain for the first class of data, for the first iteration of the Kalman gain for the first class of data.

10. The power distribution grid permanent fault identification system of claim 7, wherein: The specific process of establishing the improved Bergeon model in the improved Bergeon model establishing module includes: The basic principle of the Bergeon model is represented by a fluctuation equation: wherein, V x t I x t are the voltage and current of the line at position x c is the wave speed, Z c is the characteristic impedance of the line, the propagation speed v and the characteristic impedance Z c is given by:​​​​​​ wherein, L L is the distributed inductance of the line, C C is the inter-phase capacitance; The improved representation considering the low voltage ride-through capability of photovoltaic is: in, V LVRT ( t This indicates the voltage response considering low-voltage ride-through. t LVRT For time travel parameters, α To adjust for the low voltage ride-through factor, the introduction of low voltage ride-through can better simulate the dynamic behavior of photovoltaic systems under low voltage conditions; An improved representation taking into account the non-linear characteristics of the power electronic device is introduced in the traditional Bergeon model by introducing a non-linear impedance Z nonlinear ( I ) to model the non-linearities of the power electronic device. wherein, Z 0 is the initial impedance; k and n is a non-linear parameter.

11. The power distribution grid permanent fault identification system of claim 7, wherein: In the step identification module: The least square algorithm is used to identify the line-to-line capacitance parameters: in, C ij For the line i and j Interphase capacitors, V ij ( t )and I ij ( t The phase-to-phase voltage and current are represented by , respectively. By performing least-squares fitting on data from multiple time points, the phase-to-phase capacitance of the line is effectively estimated. The objective function of the least-squares method is: wherein is a voltage estimate, N is the number of data points; In the comparison module: Error and the permanent fault criterion is defined as: If >threshold then a permanent fault is determined, wherein threshold is a pre-defined error threshold.

Images