A precise detection and protection control method for preventing islanding effect in photovoltaic inverters

Abstract

This invention discloses a method for accurate detection and protection control of islanding effect in photovoltaic inverters, relating to the field of photovoltaic inverter islanding effect detection technology. The method includes the following steps: acquiring relevant threshold parameters of the photovoltaic inverter; collecting voltage data in real time at the grid-connected end of the photovoltaic inverter to obtain grid-connected voltage data; performing voltage anomaly analysis and processing the anomaly data to obtain reference grid-connected voltage data; performing fluctuation deviation analysis and prediction based on the reference grid-connected voltage data, and setting dynamic thresholds to obtain dynamic threshold data; detecting the islanding effect of the photovoltaic inverter based on the dynamic threshold data, and performing corresponding protection control. This invention addresses the problem that existing photovoltaic inverter islanding effect detection technologies, when using over- and under-voltage detection, cannot predict short-term voltage changes based on grid voltage variations, and simultaneously dynamically adjust over-voltage and under-voltage thresholds.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic inverter islanding effect detection technology, specifically a method for accurate detection and protection control of photovoltaic inverter anti-islanding effect. Background Technology

[0002] Photovoltaic inverter islanding detection technology refers to a technical system that combines hardware circuits and software algorithms to monitor the grid-connected operation status of photovoltaic inverters in real time and accurately identify the abnormal state in which the inverter continues to supply power to local loads, forming an independent island, after the grid is disconnected due to faults or maintenance. Its core objective is to quickly and reliably determine the islanding effect after grid loss of voltage, providing a basis for subsequent anti-islanding protection control, thereby avoiding safety risks such as electric shock to maintenance personnel, equipment overvoltage and overfrequency damage, and impact short circuits when the grid is restored.

[0003] Existing photovoltaic inverter islanding detection technologies often use preset, fixed overvoltage and undervoltage thresholds when detecting islanding effects using overvoltage and undervoltage detection. However, in actual grid operation, voltage is not constant and fluctuates dynamically due to electricity consumption periods, regional loads, and other factors. Fixed thresholds cannot adapt to these changes. For example, during off-peak hours, the grid voltage is already close to the preset overvoltage threshold; even slight voltage fluctuations can easily trigger the threshold judgment, potentially leading to unnecessary inverter shutdowns and lost power generation. Conversely, during peak hours, the voltage approaches the undervoltage threshold, which may trigger false protection due to a temporary increase in load. Furthermore, this approach exacerbates the inherent detection blind spot problem of passive detection. Preset overvoltage and undervoltage thresholds... To avoid misjudgment, a relatively broad threshold is often preset, such as when the photovoltaic power is close to the local load power. When islanding occurs, although the voltage changes, it is not easy to exceed the broad fixed threshold. At this time, the occurrence of islanding cannot be detected, causing the inverter to continuously supply power, increasing the safety hazard. Furthermore, the load characteristics of different application scenarios are significantly different, and the voltage change patterns are different. The fixed threshold lacks adaptability and it is difficult to balance detection sensitivity and accuracy. Therefore, when using over- and under-voltage detection to detect islanding, existing photovoltaic inverter islanding detection technologies cannot predict short-term voltage changes based on grid voltage changes, and cannot dynamically adjust the over-voltage and under-voltage thresholds separately. Summary of the Invention

[0004] This invention aims to at least partially solve one of the technical problems in the prior art. It obtains relevant threshold parameters of the photovoltaic inverter, collects voltage data in real time at the grid-connected end of the photovoltaic inverter to obtain grid-connected voltage data; performs voltage anomaly analysis based on the inverter's grid-connected voltage data and processes the anomaly data to obtain reference grid-connected voltage data; performs fluctuation deviation analysis and prediction based on the reference grid-connected voltage data and sets dynamic thresholds to obtain dynamic threshold data; and detects the islanding effect of the photovoltaic inverter based on the dynamic threshold data and performs corresponding protection control. This addresses the problem that existing photovoltaic inverter islanding effect detection technologies, when using over- and under-voltage detection, cannot predict short-term voltage changes based on grid voltage variations and dynamically adjust over- and under-voltage thresholds separately.

[0005] To achieve the above objectives, this application provides a method for precise detection and protection control of the islanding effect in photovoltaic inverters, comprising the following steps:

[0006] Obtain relevant threshold parameters of the photovoltaic inverter, collect voltage data in real time at the grid-connected end of the photovoltaic inverter, and obtain the grid-connected voltage data of the inverter;

[0007] Voltage anomaly analysis is performed based on inverter grid-connected voltage data, and anomaly data processing is performed to obtain reference grid-connected voltage data.

[0008] Fluctuation deviation analysis and prediction are performed based on reference grid-connected voltage data, and dynamic threshold data is obtained by setting dynamic thresholds.

[0009] The islanding effect of photovoltaic inverters is detected based on dynamic threshold data, and corresponding protection and control measures are implemented.

[0010] Voltage anomaly analysis and anomaly data processing based on inverter grid-connected voltage data to obtain reference grid-connected voltage data includes the following sub-steps:

[0011] The voltage sequence of the grid-connected section is denoted as the first voltage sequence, and any voltage value in the first voltage sequence is denoted as the first voltage value;

[0012] Set the sliding window size to k1 and the sliding step size to k2, and slide it sequentially from the starting position of the first voltage sequence. Record the window corresponding to each slide as a region window; obtain all region windows that include the first voltage value, record them as the first window, and record any first window as the first region window;

[0013] Remove the largest and smallest voltage values ​​in the first region window, and perform linear fitting on the remaining voltage values ​​to obtain the corresponding fitting function. Calculate the fitting value corresponding to the position of the first voltage value based on the corresponding fitting function, and record it as the first fitting value. Calculate the absolute difference between the first voltage value and the first fitting value, and record it as the fitting deviation of the first voltage value.

[0014] Repeatedly calculate the fitting deviation of all voltage values ​​within the first region window, and denote it as the set of fitting deviations for the first region window. Calculate the mean and standard deviation of the set of fitting deviations, and denote them as AB and AP respectively in order. Then denote AB as the mean deviation corresponding to the first region window.

[0015] Furthermore, obtaining the relevant threshold parameters of the photovoltaic inverter and acquiring voltage data in real time at the grid-connected terminal of the photovoltaic inverter to obtain the inverter's grid-connected voltage data includes the following sub-steps:

[0016] For any photovoltaic inverter to be tested, it is denoted as the first inverter; the grid connection terminal of the first inverter is denoted as the first grid connection terminal.

[0017] Obtain the overvoltage threshold and undervoltage threshold of the first inverter to determine whether islanding occurs, and denot them as the basic overvoltage threshold AV0 and the basic undervoltage threshold CV0 in sequence; and calculate BV0, denoted as the basic voltage reference, where BV0 = (AV0 + CV0) / 2.

[0018] Furthermore, obtaining the relevant threshold parameters of the photovoltaic inverter and acquiring voltage data in real time at the grid-connected terminal of the photovoltaic inverter to obtain the inverter's grid-connected voltage data includes the following sub-steps:

[0019] At the first grid-connected terminal, the effective value of the AC voltage at the first grid-connected terminal is collected at the first sampling frequency and recorded as the grid-connected segment voltage. The collection time is also recorded and arranged in chronological order as the grid-connected segment voltage sequence, which is labeled as inverter grid-connected voltage data. The first sampling frequency is t1.

[0020] Furthermore, the process of analyzing voltage anomalies based on inverter grid-connected voltage data and processing the anomaly data to obtain reference grid-connected voltage data includes the following sub-steps:

[0021] If the first voltage value is not located in [AB-k3*AP, AB+k3*AP], then the abnormal judgment result corresponding to the first voltage value in the first area window is marked as abnormal; otherwise, it is marked as normal, where k3 is the set proportional coefficient.

[0022] Repeatedly obtain the mean deviation corresponding to all first windows, and repeatedly obtain the anomaly judgment results of the first voltage value corresponding to all first windows, and obtain the total number of first windows marked as abnormal by the corresponding anomaly judgment results, denoted as the number of abnormal windows F1;

[0023] Let F0 be the total number of the first window. If F1 / F0 > k4, then mark the first voltage value as an abnormal voltage value; otherwise, mark it as a normal voltage value. Repeat the marking of all voltage values ​​in the first voltage sequence, where k4 is the set scaling factor.

[0024] Furthermore, the process of analyzing voltage anomalies based on inverter grid-connected voltage data and processing the anomaly data to obtain reference grid-connected voltage data includes the following sub-steps:

[0025] If the first voltage value is an abnormal voltage value, then obtain the first window where all abnormal judgment results for the first voltage value are abnormal, and record it as the abnormal judgment window, and record any abnormal judgment window as the first judgment window;

[0026] The first fitted value corresponding to the first voltage value in the first judgment window is denoted as RV1, and the mean deviation corresponding to the first judgment window is denoted as RC; [1 / (RC+e1)] is denoted as the basic deviation weight QC of the first judgment window; the basic deviation weights of all abnormal judgment windows are repeatedly obtained and summed, and denoted as the benchmark deviation weight HC; where e1 is a set minimum constant;

[0027] Calculate the weighted fitting value QV corresponding to the first voltage value in the first judgment window based on RV1, QC, and HC, where QV = (QC / HC) * RV1; repeat the calculation of the weighted fitting value corresponding to the first voltage value in all anomaly judgment windows, and calculate the sum of all corresponding weighted fitting values, which is recorded as the corrected voltage value corresponding to the first voltage value; replace the first voltage value with the corresponding corrected voltage value.

[0028] Repeatedly replace all abnormal voltage values ​​in the first voltage sequence to obtain the reference voltage sequence corresponding to the grid-connected segment voltage sequence, which is then marked as the reference grid-connected voltage data.

[0029] Furthermore, fluctuation deviation analysis and prediction are performed based on reference grid-connected voltage data, and dynamic thresholds are set to obtain dynamic threshold data, including the following sub-steps:

[0030] Set the first time length t2, divide the reference voltage sequence into multiple segments with a duration of the first time length according to the acquisition time, and denot them as periodic voltage segments; and denot any one of the periodic voltage segments as the first voltage segment;

[0031] Record any voltage value in the first voltage segment as the second voltage value DV, and calculate the reference voltage deviation VE corresponding to the second voltage value; where VE = DV - BV0; repeatedly obtain the reference voltage deviation corresponding to all voltage values ​​in the first voltage segment, and record the maximum reference voltage deviation and the minimum reference voltage deviation as the maximum voltage deviation and the minimum voltage deviation of the first voltage segment, respectively.

[0032] Repeatedly obtain the maximum and minimum voltage deviations of all periodic voltage segments; and sort all the maximum and minimum voltage deviations in order of time from far to near, and record them as voltage deviation sequence 1 and voltage deviation sequence 2 respectively.

[0033] Furthermore, the process of analyzing and predicting fluctuation deviations based on reference grid-connected voltage data, and setting dynamic thresholds to obtain dynamic threshold data, also includes the following sub-steps:

[0034] Set the prediction time length to k5*t2, where k5 is an integer; denote the voltage deviation sequence 1 as the first deviation sequence; set the initial size of the dynamic window to k6; and denote the data at the end of the first deviation sequence as the first end deviation;

[0035] Starting from the initial size of the dynamic window, and taking the first end deviation as the endpoint of the dynamic window, linearly fit the data within the dynamic window to obtain the corresponding fitting function, denoted as fitting function 0. Based on fitting function 0, obtain the fitting value corresponding to the position of the first voltage value, and calculate the absolute difference with the first end deviation, denoted as the fitting difference of the first end deviation. Repeat the process of obtaining the fitting difference of all data within all dynamic windows and calculating the average value, denoted as the average fitting difference 0.

[0036] Keeping the first end deviation as the endpoint of the dynamic window, expand the dynamic window by k7, and obtain the average deviation corresponding to the dynamic window at this time, denoted as average deviation 1; repeat the operation to obtain k8 average deviations, and arrange them in the order of acquisition, denoted as the average deviation sequence, where k8 is the set number;

[0037] Perform linear fitting on the average fitted sequence and obtain the corresponding slope, denoted as the fitted slope; if the fitted slope is ≥1, then the k9 smallest average fitted sequences in the average fitted sequence are denoted as the reference fitted sequence, where k9 is the set number, and k9≤k8;

[0038] If the slope of the pseudo-pseudo ...

[0039] Furthermore, the process of analyzing and predicting fluctuation deviations based on reference grid-connected voltage data, and setting dynamic thresholds to obtain dynamic threshold data, also includes the following sub-steps:

[0040] Obtain any average deviation in the reference deviation sequence and denote it as the first deviation CW. Denote [1 / (CW+e1)] as the basic deviation weight of the first deviation. Repeatedly obtain the basic deviation weights of all average deviations in the reference deviation sequence and sum them, denote it as HW. Denote [1 / (CW+e1)] / HW as the deviation weight QW of the first deviation.

[0041] The fitting function corresponding to the first fitting error is denoted as the first fitting function. The maximum voltage deviation of the future prediction time length is obtained using the first fitting function and denoted as the prediction sequence of the first fitting error. The maximum voltage deviation of any prediction in the prediction sequence of the first fitting error is denoted as YCV. QW*YCV is calculated and denoted as the weighted prediction deviation corresponding to the first fitting error. The weighted prediction deviation of all average fitting errors in the reference fitting error sequence at the time corresponding to YCV is repeatedly obtained and summed to obtain the corresponding average prediction deviation.

[0042] Repeatedly obtain the average prediction deviation corresponding to each time period of the future prediction time length, denoted as the prediction deviation sequence corresponding to voltage deviation sequence 1, and marked as the maximum deviation prediction sequence;

[0043] Repeat the acquisition of the predicted deviation sequence corresponding to voltage deviation sequence 2, and mark it as the minimum deviation prediction sequence;

[0044] Take any corresponding time period in the future prediction time length as the first time period, and obtain the corresponding maximum voltage deviation UC1 and minimum voltage deviation UC2 from the maximum deviation prediction sequence and the minimum deviation prediction sequence respectively; calculate (BV0+UC1) and (BV0+UC2), and denot them as UB1 and UB2 in order;

[0045] Let UA1 = min(AV0, e2*UB1) and UA2 = max(CV0, e2*UB2). Let [UA2, UA1] be the dynamic threshold range of the first time period. Repeatedly obtain the dynamic threshold range of all corresponding time periods in the future prediction time length and record them as dynamic threshold data. Here, e2 is the set scaling factor.

[0046] Furthermore, the detection and corresponding protection control of the islanding effect of the photovoltaic inverter based on dynamic threshold data includes the following sub-steps:

[0047] During the first time period, the effective value of the AC voltage at the first grid-connected terminal is continuously collected. If it is not within the corresponding dynamic threshold range and the duration of the non-islanding period is greater than t3, it is determined that the first inverter has experienced an islanding effect; otherwise, it is determined that the first inverter has not experienced an islanding effect. Here, t3 is the set duration threshold.

[0048] If the first inverter experiences an islanding effect, the first inverter will be disconnected from the power grid.

[0049] The beneficial effects of this invention are as follows: This invention obtains relevant threshold parameters of the photovoltaic inverter, collects voltage data in real time at the grid-connected end of the photovoltaic inverter to obtain inverter grid-connected voltage data; performs voltage anomaly analysis based on the inverter grid-connected voltage data and processes the anomaly data to obtain reference grid-connected voltage data; performs fluctuation deviation analysis and prediction based on the reference grid-connected voltage data and sets dynamic thresholds to obtain dynamic threshold data; detects the islanding effect of the photovoltaic inverter based on the dynamic threshold data and performs corresponding protection control; when using over- and under-voltage to detect the islanding effect, it can predict short-term voltage changes based on changes in grid voltage, and dynamically adjust the over-voltage threshold and under-voltage threshold respectively to improve the accuracy of over- and under-voltage detection of the islanding effect;

[0050] This invention effectively counteracts sudden extreme noise and avoids skewed fitting results by removing the maximum and minimum values ​​within the window before refitting, thus obtaining accurate fitting results. For voltage values ​​identified as abnormal, the fitted values ​​from all windows identified as abnormal are weighted, allowing for more accurate correction of outliers using the fitted values ​​from the most accurate windows, maintaining the temporal correlation and physical consistency of the sequence. By predicting the maximum and minimum deviations in future time periods, a dynamic threshold range for each prediction period is obtained. This dynamic threshold range can be widened when the power grid fluctuates significantly to avoid misjudgments, and narrowed when the power grid is stable to improve detection sensitivity. By gradually expanding the dynamic window from the initial window to obtain a series of average fitting errors, and calculating the fitting error slope to obtain a reference fitting error sequence, the most stable time scale for fitting can be automatically identified, thus using the most reliable data scale for prediction and improving prediction accuracy. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the steps of the method of the present invention;

[0052] Figure 2 This is a flowchart of the abnormal voltage value screening process of the present invention;

[0053] Figure 3 This is a flowchart illustrating the abnormal voltage value replacement process of the present invention.

[0054] Figure 4 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Example 1, please refer to Figure 1 As shown, this application provides a method for accurate detection and protection control of the islanding effect in photovoltaic inverters, including the following steps:

[0057] Step S1 involves obtaining relevant threshold parameters of the photovoltaic inverter and collecting voltage data in real time at the grid-connected terminal of the photovoltaic inverter to obtain the inverter's grid-connected voltage data. Step S1 includes the following sub-steps:

[0058] Step S101: For any photovoltaic inverter to be tested, it is denoted as the first inverter; the grid connection terminal of the first inverter is denoted as the first grid connection terminal; the grid connection terminal refers to the connection point between the AC output side of the photovoltaic inverter and the external power grid and local load. It is the final node from which electrical energy is output from the inverter to the outside, and it is also the node where the inverter senses the external power environment.

[0059] Step S102: Obtain the overvoltage threshold and undervoltage threshold of the first inverter for determining whether islanding has occurred, and record them as the basic overvoltage threshold AV0 and the basic undervoltage threshold CV0 in sequence; and calculate BV0, which is recorded as the basic voltage reference, where BV0 = (AV0 + CV0) / 2; the overvoltage threshold and undervoltage threshold are two core critical values ​​used to determine whether the grid-connected voltage exceeds the normal range. When the voltage exceeds the threshold and continues for a certain period of time, it can be determined that islanding has occurred.

[0060] Step S103: At the first grid-connected terminal, the effective value of the AC voltage at the first grid-connected terminal is collected at the first sampling frequency and recorded as the grid-connected segment voltage. The collection time is also recorded and arranged in chronological order as the grid-connected segment voltage sequence, which is labeled as inverter grid-connected voltage data. The first sampling frequency is t1, which in this embodiment is 200Hz. The grid frequency is generally 50Hz. It needs to be at least twice the grid frequency to completely restore the sine wave waveform and obtain an accurate effective value, thus avoiding sampling distortion.

[0061] In practice, the traditional fixed threshold is based on the ideal assumption of stable grid voltage and simple load characteristics. However, in real-world scenarios, the main grid voltage is not constant and fluctuates with the time of day and regional load changes. Moreover, different scenarios have different load types, and the island voltage benchmark formed after grid power failure will vary. Fixed thresholds are difficult to adapt to these changes, which can easily lead to frequent misjudgments and shutdowns when the grid voltage is close to the threshold. In islanded states, when the voltage does not exceed the threshold, a detection blind zone is formed, making it difficult to balance safety and operational stability.

[0062] Step S2 involves performing voltage anomaly analysis based on the inverter grid-connected voltage data and processing the anomaly data to obtain reference grid-connected voltage data. Step S2 includes the following sub-steps:

[0063] For step S201, please refer to... Figure 2 As shown, the voltage sequence of the grid-connected section is denoted as the first voltage sequence, and any voltage value in the first voltage sequence is denoted as the first voltage value;

[0064] Step S202: Set the sliding window size to k1 and the sliding step size to k2, and slide sequentially from the starting position of the first voltage sequence. Record the window corresponding to each slide as a region window; obtain all region windows that include the first voltage value, record them as the first window, and record any first window as the first region window; in this embodiment, k1=20 and k2=2, which can be flexibly set; using local windows instead of global statistics of the entire sequence can capture short-term series trends and is more robust to local anomalies;

[0065] Step S203: Remove the maximum and minimum voltage values ​​from the first region window, and perform linear fitting on the remaining voltage values ​​to obtain the corresponding fitting function. Calculate the fitting value corresponding to the location of the first voltage value based on the corresponding fitting function, and record it as the first fitting value. Calculate the absolute difference between the first voltage value and the first fitting value, and record it as the fitting deviation of the first voltage value. Removing the maximum and minimum points in the window is a fast preprocessing step to resist noise and avoid these extreme values ​​from skewing the fitting function. Performing linear fitting on the remaining points can capture trends within a short period of time. The fitting deviation measures the consistency between the corresponding temperature value and the local trend.

[0066] Step S204: Repeatedly calculate the fitting deviation of all voltage values ​​within the first region window, and record it as the set of fitting deviations for the first region window. Calculate the mean and standard deviation of the set of fitting deviations, and record them as AB and AP in order. Record AB as the mean deviation corresponding to the first region window. Construct a local confidence interval using the mean and standard deviation of the fitting deviations to determine whether the point is within the normal fluctuation range of the window.

[0067] Step S205: If the first voltage value is not located in [AB-k3*AP, AB+k3*AP], then mark the abnormal judgment result corresponding to the first voltage value in the first area window as abnormal; otherwise, mark it as normal. Here, k3 is a set proportional coefficient. In this embodiment, k3=3, which can be set flexibly.

[0068] Step S206: Repeatedly obtain the mean deviation corresponding to all first windows, and repeatedly obtain the abnormal judgment result of the first voltage value corresponding to all first windows, and obtain the total number of first windows marked as abnormal by the corresponding abnormal judgment result, denoted as the number of abnormal windows F1.

[0069] Step S207: Record the total number of the first window as F0. If F1 / F0>k4, mark the first voltage value as an abnormal voltage value; otherwise, mark it as a normal voltage value. Repeat the marking of all voltage values ​​in the first voltage sequence, where k4 is a set proportional coefficient. In this embodiment, k4=0.6; it can be set flexibly. Multi-window voting can minimize the occasional errors of a single window: only when most windows consider the point to be abnormal will the point be identified as abnormal.

[0070] For step S208, please refer to... Figure 3 As shown, if the first voltage value is an abnormal voltage value, then the first window containing all abnormal judgment results of the first voltage value is obtained and recorded as the abnormal judgment window, and any abnormal judgment window is recorded as the first judgment window.

[0071] Step S209: The first voltage value corresponding to the first fitted value in the first judgment window is denoted as RV1, and the mean deviation corresponding to the first judgment window is denoted as RC; [1 / (RC+e1)] is denoted as the basic deviation weight QC of the first judgment window; the basic deviation weights of all abnormal judgment windows are repeatedly obtained and summed, and denoted as the benchmark deviation weight HC; where e1 is a set minimum constant to avoid the error of the denominator being 0; QC=[1 / (RC+e1)] means that the smaller the mean deviation of the window, the more stable and reliable the fitting function of the window is, and the greater the weight is.

[0072] Step S210: Calculate the weighted fitting value QV corresponding to the first voltage value in the first judgment window based on RV1, QC, and HC, where QV = (QC / HC) * RV1; Repeat the calculation of the weighted fitting value corresponding to the first voltage value in all anomaly judgment windows, and calculate the sum of all corresponding weighted fitting values, which is recorded as the corrected voltage value corresponding to the first voltage value; Replace the first voltage value with the corresponding corrected voltage value.

[0073] Step S211: Repeat the replacement of all abnormal voltage values ​​in the first voltage sequence. After completion, the reference voltage sequence corresponding to the grid-connected segment voltage sequence is obtained and marked as reference grid-connected voltage data.

[0074] In practice, weighted fusion of local fitted values ​​from each anomaly detection window to replace anomaly data can better preserve local trends and reliable information from adjacent windows.

[0075] Step S3 involves performing fluctuation deviation analysis and prediction based on reference grid-connected voltage data, and setting dynamic thresholds to obtain dynamic threshold data. Step S3 includes the following sub-steps:

[0076] Step S301: Set the first time length t2, divide the reference voltage sequence into multiple segments with a duration of the first time length according to the acquisition time, and denot them as periodic voltage segments; and denot any one of the periodic voltage segments as the first voltage segment; in this embodiment, t2=0.04s; it can be adjusted flexibly, but should not be too large; divide the entire time series into segments, map the original high-frequency data to a time scale suitable for protection decision-making, and reduce data calculation costs;

[0077] Step S302: Record any voltage value in the first voltage segment as the second voltage value DV, and calculate the reference voltage deviation VE corresponding to the second voltage value; where VE = DV - BV0; repeatedly obtain the reference voltage deviation corresponding to all voltage values ​​in the first voltage segment, and record the largest and smallest reference voltage deviations as the maximum and minimum voltage deviations of the first voltage segment, respectively; using the deviation VE, the degree of deviation can be quantified into positive and negative offsets, and the overvoltage threshold and undervoltage threshold can be adjusted separately in the future;

[0078] Step S303: Repeatedly obtain the maximum and minimum voltage deviations of all periodic voltage segments; and sort all the maximum and minimum voltage deviations according to their corresponding time sequence from far to near, and record them as voltage deviation sequence 1 and voltage deviation sequence 2 respectively; the maximum and minimum deviations of each segment are the original data for establishing the deviation sequence. These extreme value sequences will be used as the target and input for prediction, that is, we want to predict the maximum and minimum deviations that may occur in each segment in the future, and then determine the dynamic threshold; deviation sequence 1 records the extreme cases of voltage deviating upward from the baseline in each period in history, while deviation sequence 2 records the extreme cases of downward deviation.

[0079] Step S304: Set the prediction time length to k5*t2, where k5 is an integer; denote the voltage deviation sequence 1 as the first deviation sequence; set the initial size of the dynamic window to k6; and denote the data at the end of the first deviation sequence as the first end deviation; in this embodiment, k5*t2=0.12 seconds, i.e., k5=3; k6=24, which can be flexibly set according to the actual application scenario; k5*t2 defines how many segments we want to predict in the future. Because the voltage fluctuation is relatively large and the long-term regularity is relatively weak, it is too difficult and computationally too costly to accurately predict long-term deviation changes. Therefore, in order to ensure the accuracy of the prediction, only short-term deviation changes are predicted, so k5 should not be too large, generally 1-5;

[0080] Step S305: Starting from the initial size of the dynamic window, using the first end deviation as the endpoint of the dynamic window, perform linear fitting on the data within the dynamic window to obtain the corresponding fitting function, denoted as fitting function 0. Based on fitting function 0, obtain the fitting value corresponding to the position of the first voltage value, and calculate the absolute difference with the first end deviation, denoted as the fitting difference of the first end deviation. Repeat this process to obtain the fitting difference of all data within all dynamic windows and calculate the average value, denoted as the average fitting difference 0. Choosing the end deviation of the sequence as the window endpoint is to make short-term predictions based on the latest historical data, which is more in line with real-time protection requirements.

[0081] Step S306: Keeping the first end deviation as the endpoint of the dynamic window, expand the dynamic window by k7 and obtain the average fit of the dynamic window at this time, denoted as average fit 1; repeat the operation to obtain k8 average fits and arrange them in the order of acquisition, denoted as the average fit sequence, where k8 is the set number; in this embodiment, k7=4, k8=10; can be flexibly set according to the actual application scenario; the average fit is used to detect how the fitting error changes when fitting the most recent behavior with data of different lengths; if the fit decreases or increases significantly as the window expands, it indicates that the predictability of the data for the most recent behavior is different at different scales; the average fit sequence represents the change of prediction accuracy with the window scale.

[0082] Step S307: Perform linear fitting on the average fitted sequence and obtain the corresponding slope, denoted as the fitted slope; if the fitted slope ≥ 1, then the smallest k9 average fitted sequences in the average fitted sequence are denoted as the reference fitted sequence, where k9 is the set number, k9 ≤ k8; if the slope ≥ 1, that is, the average fitted error increases or remains flat instead of decreasing, it indicates that expanding the window cannot reduce the fitting error, indicating that the recent data has high uncertainty. Therefore, it is more reasonable to select a smaller average fitted error as the reference scale. In this embodiment, k9 = 3;

[0083] Step S308: If the fitting slope is less than 1, the dynamic window is expanded to obtain the corresponding average fitting until the average fitting increases, i.e., the average fitting of the current window is greater than the average fitting of the previous window; and the k9 smallest average fittings among all the obtained average fittings are recorded as the reference fitting sequence. A slope of less than 1 indicates that using a longer window is often beneficial for fitting, but if the fitting rebounds after expanding to a certain point, it indicates that there is a critical scale, and choosing the smallest fitting point before the rebound is still a more reliable reference.

[0084] Step S309: Obtain any average fit in the reference fit sequence and denote it as the first fit CW. Denote [1 / (CW+e1)] as the basic fit weight of the first fit. Repeatedly obtain the basic deviation weights of all average fits in the reference fit sequence and sum them, denote it as HW. Denote [1 / (CW+e1)] / HW as the fit weight QW of the first fit. The smaller the average fit, the more accurate the fitting result of this scale is, so its prediction result is given a larger weight.

[0085] Step S310: The fitting function corresponding to the first fitting error is denoted as the first fitting function. The maximum voltage deviation for the future prediction time length is obtained using the first fitting function and denoted as the prediction sequence of the first fitting error. The maximum voltage deviation of any prediction in the prediction sequence of the first fitting error is denoted as YCV. QW*YCV is calculated and denoted as the weighted prediction deviation corresponding to the first fitting error. The weighted prediction deviation of all average fitting errors in the reference fitting error sequence at the corresponding time period of YCV is repeatedly obtained and summed to obtain the corresponding average prediction deviation. The fitting functions for different historical windows correspond to different reference scales. By weighted fusion of the fitting results of multiple window scales, a robust prediction result can be obtained.

[0086] Step S311: Repeatedly obtain the average prediction deviation corresponding to each time period of the future prediction time length, record it as the prediction deviation sequence corresponding to voltage deviation sequence 1, and mark it as the maximum deviation prediction sequence.

[0087] Step S312: Repeatedly obtain the predicted deviation sequence corresponding to voltage deviation sequence 2, and mark it as the minimum deviation prediction sequence;

[0088] Step S313: Denote any corresponding time period in the future prediction time length as the first time period, and obtain the corresponding maximum voltage deviation UC1 and minimum voltage deviation UC2 from the maximum deviation prediction sequence and the minimum deviation prediction sequence, respectively; calculate (BV0+UC1) and (BV0+UC2), and denote them as UB1 and UB2 in order; the predicted UC1 and UC2 are values ​​relative to BV0, and adding them back to BV0 gives the upper and lower bounds of the normal voltage fluctuation, which is a direct estimate of the future voltage fluctuation;

[0089] Step S314: Let UA1 = min(AV0, e2*UB1) to ensure that the upper limit threshold does not relax to exceed the basic overvoltage threshold, and UA2 = max(CV0, e2*UB2) to ensure that the lower limit threshold does not fall below the basic undervoltage threshold; record [UA2, UA1] as the dynamic threshold range of the first time period; repeatedly obtain the dynamic threshold range of all corresponding time periods in the future prediction time length and record it as dynamic threshold data; where e2 is the set proportional coefficient. In this embodiment, e2 = 1.1 to leave a judgment margin for normal fluctuations and reduce misjudgment;

[0090] In practice, a small mean fit indicates that the data within the window scale is highly consistent with the fitted function, meaning that the recent extreme values ​​at this scale are more regular and more suitable for prediction; a large mean fit indicates that the data at this scale is dominated by fluctuations or has nonlinear components, making the prediction unreliable.

[0091] Step S4 involves detecting the islanding effect of the photovoltaic inverter based on dynamic threshold data and implementing corresponding protection controls. Step S4 includes the following sub-steps:

[0092] Step S401: During the first time period, continuously collect the effective value of the AC voltage at the first grid-connected terminal. If it is not within the corresponding dynamic threshold range, and the duration of the non-islanding period is greater than t3, then it is determined that the first inverter has experienced an islanding effect; otherwise, it is determined that the first inverter has not experienced an islanding effect. Here, t3 is the set duration threshold. t3 can be set according to relevant standards. The dynamic threshold can widen the range when the grid fluctuates greatly to avoid false tripping, and tighten the range when the grid is stable to improve detection sensitivity.

[0093] Step S402: If the first inverter experiences an islanding effect, then control the first inverter to disconnect from the grid.

[0094] In practice, real power grids may experience transient pulses, short-term fluctuations, or measurement noise. If a single instantaneous boundary crossing is immediately identified as an islanding effect and the connection to the power grid is immediately disconnected, it will cause a large number of false disconnections. By requiring the boundary crossing to last for more than t3, transient disturbances can be distinguished from real islanding effects.

[0095] Example 2, please refer to Figure 4 As shown, Figure 4A schematic diagram of an electronic device is provided, which may include a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory stores computer-readable instructions, and the processor can call these instructions. When the processor executes a computer-readable instruction, it performs steps as described in a method for accurate detection and protection control of anti-islanding effects in photovoltaic inverters, to achieve the following functions: acquiring relevant threshold parameters of the photovoltaic inverter; real-time acquisition of voltage data at the grid-connected end of the photovoltaic inverter to obtain grid-connected voltage data; voltage anomaly analysis based on the grid-connected voltage data, and anomaly data processing to obtain reference grid-connected voltage data; fluctuation deviation analysis and prediction based on the reference grid-connected voltage data, and dynamic threshold setting to obtain dynamic threshold data; and detection of islanding effects in the photovoltaic inverter based on the dynamic threshold data, and corresponding protection control.

[0096] Furthermore, when the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] Example 3: This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program performs the steps of a method for accurate detection and protection control of anti-islanding effect in a photovoltaic inverter, to achieve the following functions: acquiring relevant threshold parameters of the photovoltaic inverter; collecting voltage data in real time at the grid-connected end of the photovoltaic inverter to obtain grid-connected voltage data; performing voltage anomaly analysis based on the grid-connected voltage data and processing the anomaly data to obtain reference grid-connected voltage data; performing fluctuation deviation analysis and prediction based on the reference grid-connected voltage data and setting dynamic thresholds to obtain dynamic threshold data; and detecting the islanding effect of the photovoltaic inverter based on the dynamic threshold data and performing corresponding protection control.

[0098] Based on the above description of the embodiments, the embodiments of the present invention can be provided as methods, systems, or computer program products. Based on this understanding, the above technical solutions, in essence or in terms of their contribution to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or certain parts of the embodiments.

[0099] In the embodiments provided in this application, it should be understood that the disclosed system or method can be implemented in other ways. The embodiments described above are merely illustrative. For example, the division of modules or units is only a logical functional division, and there may be other division methods in actual implementation. Furthermore, multiple modules or units may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces. The indirect coupling or communication connection between systems, modules, and units may be electrical, mechanical, or other forms.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for precise detection and protection control of the islanding effect in photovoltaic inverters, characterized in that, Includes the following steps: Obtain relevant threshold parameters of the photovoltaic inverter, collect voltage data in real time at the grid-connected end of the photovoltaic inverter, and obtain the grid-connected voltage data of the inverter; Voltage anomaly analysis is performed based on inverter grid-connected voltage data, and anomaly data processing is performed to obtain reference grid-connected voltage data. Fluctuation deviation analysis and prediction are performed based on reference grid-connected voltage data, and dynamic threshold data is obtained by setting dynamic thresholds. The islanding effect of photovoltaic inverters is detected based on dynamic threshold data, and corresponding protection and control measures are implemented. Voltage anomaly analysis and anomaly data processing based on inverter grid-connected voltage data to obtain reference grid-connected voltage data includes the following sub-steps: The voltage sequence of the grid-connected section is denoted as the first voltage sequence, and any voltage value in the first voltage sequence is denoted as the first voltage value; Set the sliding window size to k1 and the sliding step size to k2, and slide it sequentially from the starting position of the first voltage sequence. Record the window corresponding to each slide as a region window; obtain all region windows that include the first voltage value, record them as the first window, and record any first window as the first region window; Remove the largest and smallest voltage values ​​in the first region window, and perform linear fitting on the remaining voltage values ​​to obtain the corresponding fitting function. Calculate the fitting value corresponding to the position of the first voltage value based on the corresponding fitting function, and record it as the first fitting value. Calculate the absolute difference between the first voltage value and the first fitting value, and record it as the fitting deviation of the first voltage value. Repeatedly calculate the fitting deviation of all voltage values ​​within the first region window, and denote it as the set of fitting deviations for the first region window. Calculate the mean and standard deviation of the set of fitting deviations, and denote them as AB and AP respectively in order. Then denote AB as the mean deviation corresponding to the first region window.

2. The method for precise detection and protection control of photovoltaic inverter anti-islanding effect according to claim 1, characterized in that, Obtaining relevant threshold parameters for the photovoltaic inverter and acquiring real-time voltage data at the grid-connected terminal of the photovoltaic inverter involves the following sub-steps: For any photovoltaic inverter to be tested, it is denoted as the first inverter; the grid connection terminal of the first inverter is denoted as the first grid connection terminal. Obtain the overvoltage threshold and undervoltage threshold of the first inverter to determine whether islanding occurs, and denot them as the basic overvoltage threshold AV0 and the basic undervoltage threshold CV0 in sequence; and calculate BV0, denoted as the basic voltage reference, where BV0 = (AV0 + CV0) / 2.

3. The method for precise detection and protection control of photovoltaic inverter anti-islanding effect according to claim 2, characterized in that, Obtaining relevant threshold parameters for the photovoltaic inverter and acquiring real-time voltage data at the grid-connected terminal of the photovoltaic inverter involves the following sub-steps: At the first grid-connected terminal, the effective value of the AC voltage at the first grid-connected terminal is collected at the first sampling frequency and recorded as the grid-connected segment voltage. The collection time is also recorded and arranged in chronological order as the grid-connected segment voltage sequence, which is labeled as inverter grid-connected voltage data. The first sampling frequency is t1.

4. The method for precise detection and protection control of anti-islanding effect in photovoltaic inverters according to claim 3, characterized in that, The process of analyzing voltage anomalies based on inverter grid-connected voltage data and processing the anomalies to obtain reference grid-connected voltage data also includes the following sub-steps: If the first voltage value is not located in [AB-k3*AP, AB+k3*AP], then the abnormal judgment result corresponding to the first voltage value in the first area window is marked as abnormal; otherwise, it is marked as normal, where k3 is the set proportional coefficient. Repeatedly obtain the mean deviation corresponding to all first windows, and repeatedly obtain the anomaly judgment results of the first voltage value corresponding to all first windows, and obtain the total number of first windows marked as abnormal by the corresponding anomaly judgment results, denoted as the number of abnormal windows F1; Let F0 be the total number of the first window. If F1 / F0 > k4, then mark the first voltage value as an abnormal voltage value; otherwise, mark it as a normal voltage value. Repeat the marking of all voltage values ​​in the first voltage sequence, where k4 is the set scaling factor.

5. The method for precise detection and protection control of anti-islanding effect in photovoltaic inverters according to claim 4, characterized in that, The process of analyzing voltage anomalies based on inverter grid-connected voltage data and processing the anomalies to obtain reference grid-connected voltage data also includes the following sub-steps: If the first voltage value is an abnormal voltage value, then obtain the first window where all abnormal judgment results for the first voltage value are abnormal, and record it as the abnormal judgment window, and record any abnormal judgment window as the first judgment window; The first fitted value corresponding to the first voltage value in the first judgment window is denoted as RV1, and the mean deviation corresponding to the first judgment window is denoted as RC; [1 / (RC+e1)] is denoted as the basic deviation weight QC of the first judgment window; the basic deviation weights of all abnormal judgment windows are repeatedly obtained and summed, and denoted as the benchmark deviation weight HC; where e1 is a set minimum constant; Calculate the weighted fitting value QV corresponding to the first judgment window based on RV1, QC, and HC, where QV = (QC / HC) * RV1; Repeatedly calculate the weighted fitting value corresponding to the first voltage value in all anomaly detection windows, and calculate the sum of all corresponding weighted fitting values, which is recorded as the corrected voltage value corresponding to the first voltage value; replace the first voltage value with the corresponding corrected voltage value; Repeatedly replace all abnormal voltage values ​​in the first voltage sequence to obtain the reference voltage sequence corresponding to the grid-connected segment voltage sequence, which is then marked as the reference grid-connected voltage data.

6. The method for precise detection and protection control of anti-islanding effect in photovoltaic inverters according to claim 5, characterized in that, Fluctuation deviation analysis and prediction are performed based on reference grid-connected voltage data, and dynamic thresholds are set. The process for obtaining dynamic threshold data includes the following sub-steps: Set the first time length t2, divide the reference voltage sequence into multiple segments with a duration of the first time length according to the acquisition time, and denot them as periodic voltage segments; and denot any one of the periodic voltage segments as the first voltage segment; Record any voltage value in the first voltage segment as the second voltage value DV, and calculate the reference voltage deviation VE corresponding to the second voltage value; where VE = DV - BV0; repeatedly obtain the reference voltage deviation corresponding to all voltage values ​​in the first voltage segment, and record the maximum reference voltage deviation and the minimum reference voltage deviation as the maximum voltage deviation and the minimum voltage deviation of the first voltage segment, respectively. Repeatedly obtain the maximum and minimum voltage deviations of all periodic voltage segments; and sort all the maximum and minimum voltage deviations in order of time from far to near, and record them as voltage deviation sequence 1 and voltage deviation sequence 2 respectively.

7. The method for precise detection and protection control of anti-islanding effect in photovoltaic inverters according to claim 6, characterized in that, Fluctuation deviation analysis and prediction based on reference grid-connected voltage data, and dynamic threshold setting, further include the following sub-steps: Set the prediction time length to k5*t2, where k5 is an integer; denote the voltage deviation sequence 1 as the first deviation sequence; set the initial size of the dynamic window to k6; and denote the data at the end of the first deviation sequence as the first end deviation; Starting from the initial size of the dynamic window, and taking the first end deviation as the endpoint of the dynamic window, linearly fit the data within the dynamic window to obtain the corresponding fitting function, denoted as fitting function 0. Based on fitting function 0, obtain the fitting value corresponding to the position of the first voltage value, and calculate the absolute difference with the first end deviation, denoted as the fitting difference of the first end deviation. Repeat the process of obtaining the fitting difference of all data within all dynamic windows and calculating the average value, denoted as the average fitting difference 0. Keeping the first end deviation as the endpoint of the dynamic window, expand the dynamic window by k7, and obtain the average deviation corresponding to the dynamic window at this time, denoted as average deviation 1; repeat the operation to obtain k8 average deviations, and arrange them in the order of acquisition, denoted as the average deviation sequence, where k8 is the set number; Perform linear fitting on the average fitted sequence and obtain the corresponding slope, denoted as the fitted slope; if the fitted slope is ≥1, then the k9 smallest average fitted sequences in the average fitted sequence are denoted as the reference fitted sequence, where k9 is the set number, and k9≤k8; If the slope of the pseudo-pseudo ...

8. The method for precise detection and protection control of anti-islanding effect in photovoltaic inverters according to claim 7, characterized in that, Fluctuation deviation analysis and prediction based on reference grid-connected voltage data, and dynamic threshold setting, further include the following sub-steps: Take any average deviation in the reference fitted sequence and denote it as the first fitted deviation CW. Denote [1 / (CW+e1)] as the basic fitted deviation weight of the first fitted deviation. Repeatedly obtain the basic deviation weights of all average deviations in the reference deviation sequence, sum them, and denote them as HW; denote [1 / (CW+e1)] / HW as the deviation weight QW of the first deviation; The fitting function corresponding to the first fitting error is denoted as the first fitting function. The maximum voltage deviation of the future prediction time length is obtained using the first fitting function and denoted as the prediction sequence of the first fitting error. The maximum voltage deviation of any prediction in the prediction sequence of the first fitting error is denoted as YCV. QW*YCV is calculated and denoted as the weighted prediction deviation corresponding to the first fitting error. The weighted prediction deviation of all average fitting errors in the reference fitting error sequence at the time corresponding to YCV is repeatedly obtained and summed to obtain the corresponding average prediction deviation. Repeatedly obtain the average prediction deviation corresponding to each time period of the future prediction time length, denoted as the prediction deviation sequence corresponding to voltage deviation sequence 1, and marked as the maximum deviation prediction sequence; Repeat the acquisition of the predicted deviation sequence corresponding to voltage deviation sequence 2, and mark it as the minimum deviation prediction sequence; Take any corresponding time period in the future prediction time length as the first time period, and obtain the corresponding maximum voltage deviation UC1 and minimum voltage deviation UC2 from the maximum deviation prediction sequence and the minimum deviation prediction sequence respectively; calculate (BV0+UC1) and (BV0+UC2), and denot them as UB1 and UB2 in order; Let UA1 = min(AV0, e2*UB1) and UA2 = max(CV0, e2*UB2). Let [UA2, UA1] be the dynamic threshold range of the first time period. Repeatedly obtain the dynamic threshold range of all corresponding time periods in the future prediction time length and record them as dynamic threshold data. Here, e2 is the set scaling factor.

9. The method for precise detection and protection control of anti-islanding effect in photovoltaic inverters according to claim 8, characterized in that, Detecting the islanding effect of photovoltaic inverters based on dynamic threshold data and implementing corresponding protection controls includes the following sub-steps: During the first time period, the effective value of the AC voltage at the first grid-connected terminal is continuously collected. If it is not within the corresponding dynamic threshold range and the duration of the non-islanding period is greater than t3, it is determined that the first inverter has experienced an islanding effect; otherwise, it is determined that the first inverter has not experienced an islanding effect. Here, t3 is the set duration threshold. If the first inverter experiences an islanding effect, the first inverter will be disconnected from the power grid.

Images