A fault detection and diagnosis system and method for a distributed photovoltaic grid-connected system

Abstract

The present application relates to the technical field of distributed photovoltaic grid-connected fault diagnosis, in particular to a fault detection and diagnosis system and method for a distributed photovoltaic grid-connected system. For short lines with approximately uniform series parameters and characterized by a centralized series impedance matrix, a positioning model with only one shunt fault point in a section is adopted. Quasi-steady state synchronous phasors are obtained at both ends of the section without internal access, and a direct group is formed by the current channel of the line itself, and a check group is formed by the independent channel of the remaining ports of the end points. The detectability and consistency of the difference flow of the two groups are tested, and the position is solved by the direct group current, the voltage at both ends and the line impedance. The position is kept unchanged, and the voltage difference of the fault point is tested by the current of the check group; when the error and the identifiable condition are met, the section position under the model is output and the phase is checked. The system and method verify the fixed position with an independent current source, and constrain the compensation between the current error and the position fitting.

Description

Technical Field

[0001] This invention relates to the field of distributed photovoltaic grid-connected fault diagnosis technology, specifically to a fault detection and diagnosis system and method for distributed photovoltaic grid-connected systems. Background Technology

[0002] After distributed photovoltaic (PV) power is connected to the distribution network via a converter, the PV output, load, and switch status all affect the line current, resulting in reverse power flow during certain periods. Converter current limiting also means that the current increment during a short circuit may be less than expected in a traditional unidirectional power supply network. Therefore, relying solely on sudden changes in current amplitude can easily confuse normal output fluctuations with faults; predicting the current after a fault based solely on the branch distribution ratio before the fault may also fail during load migration.

[0003] Synchronous phasors can place voltages and currents at different locations under the same time reference, and combine this with line impedance to calculate the fault location; the current conservation of each branch at the endpoints can also provide correlation constraints. However, measurements are subject to transformer errors, clock deviations, and channel anomalies. If a set of biased currents is used to solve for the fault location, and only the residuals of the equations involved in the solution are checked, the location parameters may absorb some measurement errors, resulting in a result with small residuals but an incorrect location.

[0004] Especially in multi-port access distribution networks, the currents of other branches at the endpoints can provide additional observations, but their sampling sources, symbols, and section boundaries must be clearly defined. If the so-called additional observations are still obtained by reverse calculation from the current path being examined, no additional information can be added for verification; if an access branch is missed within the section or an unmetered shunt occurs at the endpoint, the same current conservation relationship will also be violated. Therefore, it is necessary to ensure that the location solution and verification have traceable and separate current sources, and to limit the diagnosis to sections where both the observation and the model are valid. Summary of the Invention

[0005] This invention provides a fault detection and diagnosis system and method for distributed photovoltaic grid-connected systems, which solves the problem that the low fitting residual of a single measurement is insufficient to support the determination of the fault location when there are errors in multi-port access and current measurement.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: a fault detection and diagnosis system for a distributed photovoltaic grid-connected system, comprising endpoint acquisition devices respectively configured at both ends of the line segment to be tested, a synchronous clock source connected to each endpoint acquisition device, and a diagnostic host communicating with each endpoint acquisition device. The line segment to be tested is a short line with no photovoltaic, load, or other branch access, and whose series parameters are approximately uniform along the length direction and can be characterized by a lumped series impedance matrix. Each endpoint acquisition device acquires the three-phase voltage at its endpoint and the three-phase current at each electrical port connected to that endpoint, and forms a synchronous phasor. The diagnostic host establishes a single fault location model based on the assumption that there is only one parallel fault point within the segment, and performs the following processing using only the synchronous phasor obtained when the segment is in a quasi-steady state. The diagnostic host forms a direct group from the current channels at both ends of the line section itself, and a check group from the current channels of the remaining electrical ports at both ends. The two groups are obtained by different current sensing channels. The reference direction from the top to the bottom is fixed. The current flowing into the section from the top end and the current flowing out of the section from the bottom end are taken as the currents at both ends of the direct group. The sum of the currents flowing out of the remaining ports at the top end is negative, and the sum of the currents flowing out of the remaining ports at the bottom end is kept in its original sign. These are taken as the currents at both ends of the check group. The diagnostic host obtains two sets of differential currents by subtracting the lower current from the upper current of each group. When the magnitude of the two sets of differential currents exceeds their respective detection thresholds and the magnitude of the difference between the two sets of differential currents does not exceed the current consistency threshold, the location ratio within the section that minimizes the magnitude of the difference between the fault point voltages calculated from the two ends is determined based on the measured voltages at both ends, the line series impedance matrix, and the currents at both ends of the direct group. The host also verifies that the magnitude of the minimum voltage difference of the direct group does not exceed the first voltage error boundary. The diagnostic host keeps the required location ratio unchanged, replaces the current at both ends of the direct group with the current at both ends of the verification group, and recalculates the difference in voltage at the fault point estimated at both ends. Only when the location is identifiable and located within the section, the first voltage error boundary test is passed, and the magnitude of the recalculated voltage difference does not exceed the second voltage error boundary, the internal parallel fault and location of the section under the single fault location model are output, and the relevant phases are obtained by phase-by-phase verification using the two sets of differential currents in the section. The verification group does not use the current channel of the line section itself to generate the current at both ends, nor does it reduce the voltage difference used for the test by refitting the location.

[0007] Preferably, the diagnostic host determines the segment endpoints and port set according to the switch status, and binds the topology version, channel correspondence and impedance parameter version to the same phasor time window; when the switch status changes, any participating channel is missing or saturated, or the phasor time deviation exceeds the preset allowable value within the time window, the segment positioning of that time window is stopped and the corresponding invalid reason is output. The calculation is resumed after a time window with consistent topology and all participating channels are valid is obtained again.

[0008] Preferably, the diagnostic host determines the detection threshold and current consistency threshold based on the upper bound of the error of the two sets of current channels, and determines the first voltage error boundary and the second voltage error boundary based on the voltage error at both ends, the current error, the impedance parameter error and the short-circuit model error. Before calculating the fault location, the above thresholds and error boundaries are fixed, and the upper bound of the line charging current and the non-fault shunt current that is not collected separately at the endpoint is included in the error budget. When the budget exceeds the allowable detection range, the section is marked as inapplicable. The thresholds and error boundaries are not relaxed based on the fitting results of the candidate location.

[0009] As a preferred option, the diagnostic host further uses the current at both ends of the verification group to separately calculate the second position ratio, keeps the second position ratio unchanged, and substitutes it into the current at both ends of the direct group to calculate the voltage difference; only when the magnitude of the voltage difference does not exceed the first voltage error limit, the second position ratio meets the identifiable condition within the section, and the absolute value of the difference between the two position ratios does not exceed the sum of their respective position error upper limits, is the bidirectional cross-verification of the two groups confirmed to be successful.

[0010] Preferably, the diagnostic host uses the product of the line series impedance matrix and the direct group differential current as the ranging excitation quantity. When the magnitude of the excitation quantity does not exceed the preset minimum value, the position is not output. When it exceeds the minimum value, the upper limit of the position error is determined according to the ratio of the first voltage error boundary to the magnitude of the excitation quantity. The position within the segment is output only when the error range of the required position ratio falls completely between zero and one. Otherwise, the position near the endpoint or the position is unidentifiable.

[0011] Preferably, in a three-phase system with an effectively grounded neutral point, the diagnostic host compares the amplitude of each phase of the two sets of differential currents with the corresponding phase threshold. When both sets show that the same phase exceeds the phase threshold and the other two phases do not exceed the phase threshold, a single-phase ground fault is output for that phase. When both sets show that the same two phases exceed the phase threshold, the third phase does not exceed the phase threshold, and the magnitude of the vector sum of the differential currents of the two phases does not exceed the corresponding sum current threshold, a phase-to-phase fault is output for those two phases. In other cases, internal parallel faults and those involving phases are retained, and the output type is pending verification.

[0012] Preferably, the diagnostic host performs the aforementioned calculations and checks on each tested line segment under the same topology version; when only one segment passes the check, it outputs the segment identifier; when multiple segments pass the check, it simultaneously outputs the set of segments that passed the check and the segment distinction status; when no segment passes the check and there is an abnormality in differential current or measurement consistency, it outputs the measurement or endpoint connection pending verification status and sends the time window used, topology version, and failure conditions to the maintenance terminal connected to it.

[0013] A fault detection and diagnosis method for a distributed photovoltaic grid-connected system employs a single fault location model assuming that there is only one parallel fault point within the detected line segment. The segment is a short line with approximately uniform series parameters along its length and characterizable by a lumped series impedance matrix. The method uses the synchronous phasor obtained when the segment is in a quasi-steady state for fault location, and includes the following steps: Step 1: At both ends of the line section under test, where there are no photovoltaic, loads or other branches connected internally, simultaneously obtain the three-phase voltage at the endpoints and the three-phase current phasors of each electrical port connected to each endpoint; take the current channels at both ends of the line section itself as the direct group, and take the current channels of the other electrical ports at both ends as the verification group, and use different current sensing channels for the two groups. Step 2: Fix the reference direction from the top to the bottom. Use the current flowing into the section at the top and the current flowing out of the section at the bottom as the current at both ends of the direct group. Use the negative value of the sum of the currents flowing out of the remaining ports at the top and the sum of the currents flowing out of the remaining ports at the bottom as the currents at both ends of the verification group. The verification group does not use the current channel of the line section itself or the value deduced from the channel. Step 3: Calculate the differential current between the upper and lower currents for each of the two groups. When the magnitude of the differential current in both groups exceeds their respective detection thresholds and the magnitude of the difference between the two differential currents does not exceed the current consistency threshold, based on the measured voltages at both ends, the line series impedance matrix, and the currents at both ends of the direct group, determine the position ratio that minimizes the magnitude of the difference between the fault point voltages calculated at both ends, and verify that the magnitude of the minimum voltage difference does not exceed the first voltage error boundary. Step 4: Keep the required position ratio unchanged, recalculate the difference between the fault point voltages estimated at both ends of the verification group based on the current at both ends of the verification group. When the position is identifiable and located inside the section, the first voltage error boundary test is passed, and the magnitude of the recalculated voltage difference does not exceed the second voltage error boundary, the internal parallel fault and position of the section under the single fault location model are obtained. The voltage difference of the verification group used for this test is not reduced by refitting the position. Step 5: For the same section that passes the test, check the phases involved by phase using two sets of differential current and output the diagnostic results; for sections that do not meet the conditions, retain the failure condition and do not output the determined fault location of the section.

[0014] In summary, the present invention has the following advantages: by obtaining the position using the line's own channel and verifying it at a fixed position using the measured channel of other ports at the endpoint, the verification group does not automatically lower the voltage difference used for confirmation due to refitting, thus constraining the compensation between current error and position parameters.

[0015] By fixing the segment and current direction, and simultaneously verifying the detectability and consistency of two sets of differential currents, normal photovoltaic and load changes are incorporated into the current flow, without assuming that a certain fault boundary is still in a healthy state. The location error range, measurement validity, and multiple candidate states allow situations that lack identification conditions to be reserved for verification, avoiding the forced identification of a single segment or an unsupported fault type. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the system's signal connections.

[0017] Figure 2 This is a verification diagram of the current source and fixed location.

[0018] Figure 3 This is a flowchart for section fault detection and diagnosis.

[0019] Explanation of reference numerals in the attached diagram: 10, endpoint acquisition device; 20, synchronous clock source; 30, diagnostic host; 40, operation and maintenance terminal. Detailed Implementation

[0020] like Figure 1 As shown, this embodiment sets up an endpoint acquisition device 10, a synchronous clock source 20, a diagnostic host 30, and an operation and maintenance terminal 40 in an AC distribution network with distributed photovoltaic access. The analog input of the endpoint acquisition device 10 is connected to the voltage transformer at its local end and the current transformer or equivalent independent current sensor of each electrical port to complete synchronous sampling, anti-aliasing filtering, and fundamental phasor calculation; the synchronous clock source 20 provides a unified time reference to different endpoints; the diagnostic host 30 receives data with timestamps, channel quality, and switch status and performs the following calculations; and the operation and maintenance terminal 40 receives diagnostic records. Figure 1 The lines indicate timing or communication connections, not primary electrical wiring.

[0021] First, divide the network into observable line segments. Each segment is a section of line between two acquisition endpoints, with no photovoltaic, load, or other branch connections within it. If an intermediate connection point exists, an endpoint acquisition point is set up at that connection point to split the segment; branch currents cannot be directly ignored. Other lines, upstream power sources, photovoltaic converters, and load connection branches at each endpoint are all treated as independent electrical ports for current acquisition. The sampling location should surround the defined endpoint busbar; unmetered branches between two buses should not be omitted from the summation range. Non-faulty shunts such as endpoint bypasses and grounding capacitors are only allowed to continue locating if they have already been metered or their upper bound is covered by the error budget.

[0022] The following describes a short three-phase line that can be characterized by a lumped series impedance matrix and whose series parameters are approximately uniform along its length. The two ends are named upper end u and lower end v, with the distance increasing from u to v. The upper and lower ends do not swap due to the actual power flow reversal. All three-phase currents and voltages are complex column vectors with the same frequency, phase sequence, and time reference. Voltage is measured in volts, and current in amperes. The total line length is L, measured in meters; the total series impedance matrix is ​​Z, measured in ohms, which includes the phase impedance and phase-to-phase impedance, and accounts for the equivalent effects of the neutral or ground loop. A consistent grounding reference is used for the voltages at both ends; neutral point displacement that cannot be covered by this impedance model or error budget makes this time window unsuitable for positioning. Z is obtained from the line length, conductors, and installation parameters, or calibrated on known fault-free and valid measurement records, maintained according to temperature and equipment configuration. This implementation requires that the series parameters be approximately uniform along the section; long lines with significant distributed parameter effects cannot be directly substituted into the short line relationship.

[0023] The location model in this implementation assumes that there is only one parallel fault point within the section, using data where transient components have decayed and the fundamental phasor is approximately stable within the selected time window. This assumption is a prerequisite for the applicability of the location equation and cannot be derived from verification by two sets of currents being consistent or voltages at fixed locations. If there are multiple parallel fault points in the same section, the additional voltage drop in the quantities at both ends may correspond to a weighted combination of each fault location and fault current. Even if the single-fault model obtains a small residual, it may still output an equivalent location that does not correspond to any actual fault point. Therefore, the following fault locations and their error ranges are interpreted under the condition that the single-fault model holds true; this location conclusion is not applicable to multi-point faults.

[0024] Figure 2 This shows the source and usage order of two sets of currents. Direct group A uses the actual current channels at both ends of the line itself. The upper current Iu is positive when it flows into the section from u, and the lower current Iv is positive when it flows into v from the section. Verification group B uses the actual current channels of the remaining electrical ports at both ends; the current at each port is uniformly defined as positive when it leaves the busbar at this endpoint, denoted as iu,k and iv,k respectively, where k represents other ports at this endpoint. When photovoltaic or power is injected into the busbar, the corresponding port current carries a negative sign; when the load draws power from the busbar, it carries a positive sign. The currents Ju and Jv at both ends of the verification group are obtained using the following formula: Formula (1) The summation in equation (1) only traverses the other ports of this endpoint, excluding the channel of the detected section itself. The current conservation of the upper node gives Ju and Iu, which should match; the direction of the current in the lower line is opposite to the outward flow direction of the port, so Jv is the positive value of the sum of the currents of the remaining ports. Open ends without other energized ports can use the sum of zero of an empty set, but must be confirmed by the switch status and port list to ensure that no connection is missing. Groups A and B do not share the current sensing channel. Software copying of the same phasor, back-calculation of other ports according to the direct group, or using the state estimate value of the direct group to replace the actual port sampling does not constitute the verification group of this embodiment.

[0025] The differential currents of the two groups are as follows: Formula (2) When there is no internal connection within the section and the model error is controllable, normal photovoltaic and load changes can alter the current flowing through it, but this cannot generate a continuous leakage current within the section. When a single parallel fault occurs within the section, the differential current corresponds to the three-phase current flowing into the faulty branch. Both sets of differential currents should detect this differential current and be consistent within their respective error allowable ranges. Therefore, first compare the magnitudes of the two sets of differential currents with their respective detection thresholds, and then compare the magnitude of the difference between the two sets of differential currents with the current consistency threshold; if either is not satisfied, the location is not confirmed simply because the subsequent voltage fitting residual is small. Here, the magnitude is the L2 norm of the complex three-phase vector; when determining the type phase by phase, the amplitude of the complex vector for each phase is used separately.

[0026] Let x be the distance from the fault to the upper end, and x be a real number between zero and one. The measured voltage at the upper end is Uu, and the measured voltage at the lower end is Uv. The voltage calculated from the upper end towards the fault point is Uu minus xZIu, and the voltage calculated from the lower end towards the fault point in the reverse direction is Uv plus (one minus x)ZIv. Therefore, the voltage difference at the fault point of the direct group can be written as: Formula (3) Both voltages are measured boundary conditions of the current fault window, without requiring one end to be in a healthy state, and without using historical healthy photovoltaic output to replace the current injection. For the case of a single internal fault and the short-circuit relationship being valid, the potentials on both sides of the fault point are equal, making equation (3) close to zero near the actual location. Let a and b represent respectively: Formula (4) When the modulus of a is large enough, the proportion of real numbers that minimize the modulus of equation (3) is: Formula (5) In equation (5), the superscript H indicates the conjugate transpose, Re indicates taking the real part, and the denominator is a non-negative real number. The result must not be truncated to zero or repeatedly reported as a section fault if it does not fall within the section. The minimum voltage difference of the direct group only checks whether the group conforms to the line model and cannot be used alone as proof of reliable location. In particular, current bias and location parameters may compensate for each other, so even if the voltage difference is very small, the obtained location may be incorrect.

[0027] With xA obtained from equation (5) fixed, and without changing the voltage and Z at both ends, only the current at both ends is replaced with the current obtained from the verification group, and the following calculation is performed: Formula (6) The voltage verification is passed only if the magnitude of the voltage difference between the direct groups does not exceed the first voltage error limit and the magnitude of the voltage difference in equation (6) does not exceed the second voltage error limit. The current of the verification group is given by the measured channels of different ports, but the position follows the result of the direct group, so that the verification quantity is not suppressed due to its own refitting. The two groups share the endpoint voltage and time reference, so this verification does not exclude common voltage error, common cause timing error or mutually canceling multi-channel error; when there is a corresponding quality alarm, it is treated as an invalid time window.

[0028] For enhanced verification, the two sets of roles can be swapped: xB is obtained separately for the verification group, and then the voltage difference of the direct group is calculated while fixing xB. At this time, two positions are retained respectively, and the absolute value of their difference is compared with the sum of the error boundaries of their respective positions, and the cross voltage difference is checked. The minimum voltage difference of the verification group at xB cannot be used to impersonate the voltage difference of equation (6) at xA. Bidirectional verification uses the same physical segment without adding virtual nodes or changing the support range of differential current comparison.

[0029] Each threshold should be determined before processing fault candidates. For example, for the upper bounds of the current given vector errors εIu and εIv at both ends of the direct group, the upper bound of the differential current error δDA is the sum of the two; the upper bound of the current error at each endpoint of the check group is the sum of the upper bounds of the errors of the channels involved in the summation, and then the two endpoints are added together to obtain δDB. The current consistency threshold should at least cover the sum of δDA and δDB. The detection threshold should be calculated by adding the minimum fault current margin to be detected to δDA and δDB respectively; the detection threshold should not be temporarily lowered or the consistency threshold increased because the residual of a certain line is too large.

[0030] The voltage error budget should cover both measurements and the model. Let g represent A or B, εIg be the larger of the upper bounds of the current errors at the two endpoints of the set, Mg be the larger of the measured current magnitudes at both endpoints of the set; εUu and εUv are the upper bounds of the voltage errors at both endpoints, εZ is the upper bound of the operator norm 2 of the impedance matrix error, z is the operator norm 2 of Z, and εm is the upper bound of the residual voltage drop not represented by the series model. A conservative voltage error bound τg uniformly used for the zero-to-one position range is: Formula (7) All terms in equation (7) are in volts. The first and second voltage error boundaries are τA and τB, respectively; the current amplitude Mg can be obtained according to the current effective time window, but the error parameters and budget rules are fixed in advance, and the fitting residual is not used for reverse adjustment. The charging current can be estimated from the line-to-ground capacitance, frequency and terminal voltage upper limit, and its influence on differential current and voltage drop is included in the current and voltage budgets, respectively; the non-metering shunt at the endpoint is also included according to the known upper limit. If a reliable upper limit cannot be given, or the budget has exceeded the target detection range, port measurements should be supplemented, the section shortened or the section location should be stopped, and it cannot be claimed that the zero differential current approximation still holds.

[0031] Position identifiability also depends on the magnitude of 'a'. First, a minimum ranging excitation value 'a0' greater than zero is given; division is only performed if the magnitude of 'a' is greater than 'a0'. For the linear relationship established by the above single-fault model and unified voltage error bound, a conservative upper bound for the position error is taken: Formula (8) The upper bound is dimensionless. The reported position within the segment is only given when xA minus δx is greater than zero and xA plus δx is less than one. The actual distance is LxA, and the distance error interval is L multiplied by the position error interval. Otherwise, the reported position is near the endpoint or is unidentifiable, and is not forcibly assigned to an adjacent segment. a0 can be determined by the maximum allowable position error and voltage budget. Equation (8) contains a conservative margin, indicating that even if a low-current fault has a solution in noiseless calculation, it does not necessarily have an engineering-usable ranging accuracy.

[0032] Figure 3 The actual processing flow is shown. The endpoint acquisition device 10 can synchronously sample at a fixed number of points per cycle, for example, 80 points per cycle for a 50Hz system, forming the fundamental phasor of one cycle; the diagnostic host 30 selects the start and end times of the same window using timestamps, and only uses the quasi-steady-state time window where the filtering transient has subsided and the phasor change rate is within the preset range. The allowable time deviation is determined by the current amplitude and phase angle error budget. For example, the upper limit of the fundamental angle error can be converted to the phasor error by multiplying the angular frequency and the upper limit of the time deviation.

[0033] At the start of each time window, the switch status, port set, channel direction, and impedance version are frozen. The time window becomes invalid if a switch action, port measurement failure, transformer saturation flag, timing lockout failure, or phasor quality defect occurs. Effective time windows are accumulated again after a stable topology is re-established. If multiple consecutive time window confirmations are required in the implementation, all conditions must be met separately under the same topology version; different port sets before and after switching cannot be mixed. Diagnostic calculations do not assume the instantaneous tripping function of a single protection operation.

[0034] After the validity check, two sets of differential current detection and consistency checks, direct group ranging and model checks, and fixed-position verification group voltage checks are performed in sequence. Figure 3The "Pending Review" field indicates that the actual conditions for failure are being saved: Inconsistent differential currents could indicate an anomaly in a specific channel, a lack of metering at an endpoint branch, or leakage at the endpoint busbar; this result alone cannot be used to conclude that the current sensor is damaged. When both differential currents are below the detection threshold, it means that no locatable internal parallel fault was detected in this time window, but it does not mean the system is completely fault-free. Each segment uses the same rules independently, without any additional advantages given to segments containing more nodes or residual terms.

[0035] After section location is confirmed, the fault-affected phases are determined phase by phase using DA and DB within the same time window. Each phase threshold covers the measurement error of the differential current and allows for detection margin. In systems where the neutral point is effectively grounded and the short-line model is applicable, if both groups show only phase a is significant, and phases b and c do not exceed the threshold, record phase a as a single-phase ground fault; other phases are handled similarly. If both groups show the same two phases are significant, the third phase is not significant, and the vector sum of the differential currents of these two phases does not exceed the corresponding current error limits in each group, record the corresponding two-phase inter-phase fault. The current threshold is obtained by adding the upper bounds of the two-phase errors and adding the detection margin.

[0036] In cases of inconsistent phase patterns, significant faults in all three phases, or significant faults in two phases with a significant vector sum, only the verified internal parallel fault and phase information are reported; the type is subject to further verification. Since different multi-point grounding combinations may also create similar current patterns, the above types are used for preliminary maintenance classification under the assumption of a single parallel fault, and do not directly deduce the location of the fault resistor or damaged device. The above internal parallel fault criteria do not apply to open circuits, islanding, and internal inverter failures.

[0037] The diagnostic host 30 can sequentially calculate all detected segments within the same topology version. When only one segment passes, it sends the segment identifier, location, error range, phase, and identifiable type under the single fault location model to the maintenance terminal 40; when multiple segments pass, it sends a candidate set and the segment status to be distinguished, without forcibly selecting the one with the smallest residual; when no segment passes but an anomaly exists, it sends the measurement or endpoint connection status to be verified. The record simultaneously saves the single fault model assumptions, time window, topology and impedance versions, two sets of channel sets, and the results of each criterion, allowing maintenance personnel to trace the actual measurement source.

[0038] The following example of linear phasor calculation illustrates the criteria; the values ​​are for calculation purposes. The effective value of the three-phase voltage is 230V, with phase angles of zero, -120 degrees, and +120 degrees respectively. The diagonal term of Z is 0.20 + j0.10Ω, and the off-diagonal term is 0.02 + j0.03Ω, where j is the imaginary unit. The lower-end current is a balancing current with an amplitude of 30A and the same phase sequence as the voltage. Assume phase a is grounded at x = 0.40 via a 5Ω resistor. Consistent terminal voltages and direct currents are generated from the line and fault point equations. Other port currents are then configured according to the principle of terminal current conservation. For ease of comparison, the differential current detection threshold and current consistency threshold for both sets are set to 2A, the voltage error limits for both sets are set to 2V, and the minimum ranging excitation value a0 is set to 1V. These values ​​are only for this example; actual devices must be obtained according to the aforementioned budget.

[0039] The calculated differential current of phase a is approximately 44.846 - j0.518A, while the other two phases are zero. The position recovery ratio of the direct group is 0.4000, and the voltage difference between the two groups is close to the numerical calculation accuracy. At this time, the magnitude of the ranging excitation is approximately 10.29V, and the upper limit of the position error obtained according to formula (8) is approximately 0.389, corresponding to a position range of approximately 0.011 to 0.789, which is generally located within the section. When the normal photovoltaic increase causes the line balance current to decrease from 30A to 5A, and there is no fault within the section, both differential currents are zero, and ranging is not started.

[0040] In the fault calculation example, a 20A real bias is artificially added to phase a at the upper end of the direct group. Keeping the verification group unchanged, the minimum voltage difference of the direct group is approximately 0.0147V, but the calculated position becomes 0.2766. The difference in differential current between the two groups is 20A, which is rejected because it exceeds the current consistency threshold. If a 10A real bias is added when phase a is the same at both ends of the direct group, the differential current remains the same, and the minimum voltage difference of the direct group is approximately 0.0265V, with the calculated position becoming 0.1770. The voltage difference of the verification group calculated at this fixed position is approximately 2.2933V, exceeding the 2V error limit in this example. The position error range also does not meet the internal conditions of the segment, so no definite position is output. This indicates that a small fitted value for the direct group and consistent differential current are insufficient to complete the verification independently.

[0041] When both differential currents are only 0.1A, even if the noiseless equation calculates a position of 0.40, it will not report the position because it is below the 2A detection threshold. If a port that causes a 20A change in the current at the upper end of the check set is missed, the two differential currents will be inconsistent, and the endpoint connection and measurement should be checked.

Claims

1. A fault detection and diagnosis system for a distributed photovoltaic grid-connected system, characterized in that: The system includes endpoint acquisition devices (10) respectively configured at both ends of the line section under test, a synchronous clock source (20) connected to each endpoint acquisition device (10), and a diagnostic host (30) communicating with each endpoint acquisition device (10). The line section under test is a short line with no photovoltaic, load and other branch access, and whose series parameters are approximately uniform along the length direction and can be characterized by a concentrated series impedance matrix. Each endpoint acquisition device (10) acquires the three-phase voltage of its endpoint and the three-phase current of each electrical port connected to that endpoint, and forms a synchronous phasor. The diagnostic host (30) establishes a single fault location model based on the assumption that there is only one parallel fault point in the section, and only uses the synchronous phasor obtained when the section is in a quasi-steady state to perform the following processing. The diagnostic host (30) forms a direct group of the current channels at both ends of the line section itself and a check group of the current channels of the remaining electrical ports at both ends. The two groups are obtained by different current sensing channels. The reference direction from the upper end to the lower end is fixed. The current flowing into the section from the upper end and the current flowing out of the section from the lower end are taken as the currents at both ends of the direct group. The sum of the currents flowing out of the remaining ports at the upper end is negative and the sum of the currents flowing out of the remaining ports at the lower end is kept in its original sign. These are respectively taken as the currents at both ends of the check group. The diagnostic host (30) subtracts the lower current from the upper current of each group to obtain two sets of differential currents. When the magnitude of the two sets of differential currents exceeds their respective detection thresholds and the magnitude of the difference between the two sets of differential currents does not exceed the current consistency threshold, the proportion of the section that minimizes the magnitude of the difference between the fault point voltages calculated at both ends is determined based on the measured voltages at both ends, the line series impedance matrix, and the currents at both ends of the direct group. The minimum voltage difference of the direct group is checked to ensure that the magnitude does not exceed the first voltage error limit. The diagnostic host (30) keeps the required position ratio unchanged, replaces the current at both ends of the direct group with the current at both ends of the verification group, recalculates the difference in voltage at the fault point estimated at both ends, and outputs the internal parallel fault and location of the section under the single fault location model only when the location is identifiable and located inside the section, the first voltage error boundary test is passed, and the magnitude of the recalculated voltage difference does not exceed the second voltage error boundary. The relevant phases are obtained by phase-by-phase verification of the two groups of differential currents in the section. The verification group does not use the current channel of the line section itself to generate the current at both ends, nor does it reduce the voltage difference used for the test by refitting the position.

2. The fault detection and diagnosis system for a distributed photovoltaic grid-connected system according to claim 1, characterized in that: The diagnostic host (30) determines the segment endpoints and port sets according to the switch status, and binds the topology version, channel correspondence and impedance parameter version to the same phasor time window; If a switch state changes, any participating channel is missing or saturated, or the phasor time deviation exceeds the preset allowable value within a time window, the segment positioning of that time window is stopped and the corresponding invalid reason is output. The calculation is resumed after a time window with topological consistency and all participating channels are valid is obtained again.

3. The fault detection and diagnosis system for a distributed photovoltaic grid-connected system according to claim 1, characterized in that: The diagnostic host (30) determines the detection threshold and current consistency threshold based on the upper bound of the error of the two sets of current channels, and determines the first voltage error boundary and the second voltage error boundary based on the voltage error at both ends, the current error, the impedance parameter error and the short circuit model error. Before calculating the fault location, the above threshold and error boundary are fixed. The upper bound of the line charging current and the non-fault shunt that is not collected separately at the endpoint is included in the error budget. When the budget exceeds the allowable detection range, the segment is marked as inapplicable. The threshold and error boundary are not relaxed based on the fitting result of the candidate location.

4. The fault detection and diagnosis system for a distributed photovoltaic grid-connected system according to claim 1, characterized in that: The diagnostic host (30) further uses the current at both ends of the verification group to separately obtain the second position ratio, keeps the second position ratio unchanged, and substitutes it into the current at both ends of the direct group to calculate the voltage difference; only when the magnitude of the voltage difference does not exceed the first voltage error limit, the second position ratio meets the condition that the section is identifiable, and the absolute value of the difference between the two position ratios does not exceed the sum of their respective position error upper limits, the two groups are confirmed to have passed the bidirectional cross-verification.

5. The fault detection and diagnosis system for a distributed photovoltaic grid-connected system according to claim 1, characterized in that: The diagnostic host (30) uses the product of the line series impedance matrix and the direct group differential current as the ranging excitation quantity. When the magnitude of the excitation quantity does not exceed the preset minimum value, the position is not output. When it exceeds the minimum value, the upper limit of the position error is determined according to the ratio of the first voltage error boundary to the magnitude of the excitation quantity. The position within the segment is output only when the error range of the required position ratio falls completely between zero and one. Otherwise, the position near the endpoint or the position is not identifiable.

6. The fault detection and diagnosis system for a distributed photovoltaic grid-connected system according to claim 1, characterized in that: In a three-phase system with an effective neutral ground, the diagnostic host (30) compares the amplitude of each phase of the differential current in two groups with the corresponding phase threshold. When both groups show that the same phase exceeds the phase threshold and the other two phases do not exceed the phase threshold, the host outputs a single-phase ground fault in that phase. When both groups show that the same two phases exceed the phase threshold, the third phase does not exceed the phase threshold, and the magnitude of the vector sum of the differential currents of the two phases does not exceed the corresponding current threshold, the host outputs a phase-to-phase fault in those two phases. In other cases, internal parallel faults and phases involved are retained, and the output type is pending verification.

7. The fault detection and diagnosis system for a distributed photovoltaic grid-connected system according to claim 1, characterized in that: The diagnostic host (30) performs the calculation and inspection on each line segment under the same topology version; when only one segment passes the inspection, it outputs the segment identification of the segment; when multiple segments pass the inspection, it outputs the set of segments that passed the inspection and the segment status to be distinguished; when no segment passes the inspection and there is differential current or measurement consistency abnormality, it outputs the measurement or endpoint connection status to be reviewed, and sends the time window used, topology version and failure conditions to the maintenance terminal (40) connected to it.

8. A fault detection and diagnosis method for a distributed photovoltaic grid-connected system, characterized in that: A single-fault location model is adopted, assuming that there is only one parallel fault point within the tested line section. The section is a short line with approximately uniform series parameters along its length and which can be characterized by a lumped series impedance matrix. The location is performed using the synchronous phasor obtained when the section is in a quasi-steady state, including the following steps: Step 1: At both ends of the line section under test, where there are no photovoltaic, loads or other branches connected internally, simultaneously obtain the three-phase voltage at the endpoints and the three-phase current phasors of each electrical port connected to each endpoint; take the current channels at both ends of the line section itself as the direct group, and take the current channels of the other electrical ports at both ends as the verification group, and use different current sensing channels for the two groups. Step 2: Fix the reference direction from the top to the bottom. Use the current flowing into the section at the top and the current flowing out of the section at the bottom as the current at both ends of the direct group. Use the negative value of the sum of the currents flowing out of the remaining ports at the top and the sum of the currents flowing out of the remaining ports at the bottom as the currents at both ends of the verification group. The verification group does not use the current channel of the line section itself or the value deduced from the channel. Step 3: Calculate the differential current between the upper and lower currents for each of the two groups. When the magnitude of the differential current in both groups exceeds their respective detection thresholds and the magnitude of the difference between the two differential currents does not exceed the current consistency threshold, based on the measured voltages at both ends, the line series impedance matrix, and the currents at both ends of the direct group, determine the position ratio that minimizes the magnitude of the difference between the fault point voltages calculated at both ends, and verify that the magnitude of the minimum voltage difference does not exceed the first voltage error boundary. Step 4: Keep the required position ratio unchanged, recalculate the difference between the fault point voltages estimated at both ends of the verification group based on the current at both ends of the verification group. When the position is identifiable and located inside the section, the first voltage error boundary test is passed, and the magnitude of the recalculated voltage difference does not exceed the second voltage error boundary, the internal parallel fault and position of the section under the single fault location model are obtained. The voltage difference of the verification group used for this test is not reduced by refitting the position. Step 5: For the same section that passes the test, check the phases involved by phase using two sets of differential current and output the diagnostic results; for sections that do not meet the conditions, retain the failure condition and do not output the determined fault location of the section.

Images