A power supply monitoring method and system for microgrid power distribution

Abstract

The present invention relates to the technical field of power grid data monitoring, and particularly relates to a power supply monitoring method and system for microgrid power distribution; obtaining a characteristic autocorrelation coefficient and an autocorrelation coefficient sequence according to the data difference characteristics of the power supply characteristic monitoring sequence at intervals of a preset time period; obtaining an optimal time window length according to the data change characteristics of the autocorrelation coefficient sequence; obtaining different dimension change characteristic values and dimension anomaly degrees according to the data change characteristics within the optimal time window length of the target data point in the power supply characteristic monitoring sequence; obtaining an overall anomaly degree and an anomaly sequence segment according to the dimension anomaly degree. The present invention obtains anomaly data points according to the data distribution characteristics in the anomaly sequence segment; monitors the power supply state of the microgrid according to the anomaly data points, thereby improving the accuracy of power supply monitoring.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid data monitoring, and in particular to a power supply monitoring method and system for microgrid power distribution. Background Art

[0002] With the rapid development of renewable energy and the continuous advancement of intelligent grid technology, microgrids, as an important part of distributed energy, play an increasingly important role in power supply. Due to the diverse types of power sources and complex network structures in microgrids, various problems are prone to occur during the power supply process, such as power supply fluctuations and load imbalance; therefore, monitoring the power distribution of microgrids is of great significance to ensure the stability, safety and reliability of power supply.

[0003] Traditional power supply monitoring is based on a single threshold for fault identification. This method can only detect monitoring data with obvious abnormalities. As for monitoring data that has a risk of failure but has not reached the set threshold, a large amount of normal data in the time series can easily overwhelm a small amount of data with a risk of failure, making it difficult to discover and screen the monitoring data with a risk of failure, thus affecting the timeliness and accuracy of power supply abnormality monitoring. Summary of the invention

[0004] In order to solve the above-mentioned technical problem that the traditional setting of thresholds easily leads to the difficulty in discovering monitoring data with fault risks, affecting the timeliness and accuracy of power supply abnormality monitoring, the purpose of the present invention is to provide a power supply monitoring method and system for microgrid power distribution, and the technical solutions adopted are as follows:

[0005] Acquire a power supply feature monitoring sequence for monitoring a microgrid; obtain a feature autocorrelation coefficient according to data difference characteristics separated by a preset time period in the power supply feature monitoring sequence;

[0006] Obtain an autocorrelation coefficient sequence according to characteristic autocorrelation coefficients corresponding to different preset time periods; obtain an optimal time window length according to the data change characteristics of the autocorrelation coefficient sequence; obtain different dimensional change characteristic values ​​according to the data change characteristics within the optimal time window length range of the target data point in the power supply characteristic monitoring sequence;

[0007] Obtain the dimensional anomaly degree of the target data point according to the distribution difference characteristics between the dimensional change feature values ​​of the same type; obtain the overall anomaly degree of the data segment within the optimal time window length range of the target data point according to different dimensional anomaly degrees; obtain the abnormal sequence segment in the power supply feature monitoring sequence according to the overall anomaly degree;

[0008] Abnormal data points are obtained according to the data distribution characteristics in the abnormal sequence segment; and the power supply status of the microgrid is monitored according to the abnormal data points.

[0009] Further, the step of obtaining the characteristic autocorrelation coefficient according to the data difference characteristics of the power supply characteristic monitoring sequence separated by a preset time period includes:

[0010] Calculate the average value of the data point values in the power supply characteristic monitoring sequence to obtain the overall data mean value; calculate the variance of the data point values in the power supply characteristic monitoring sequence to obtain the overall fluctuation value;

[0011] Delete a preset number of data points from the end in the power supply characteristic monitoring sequence to obtain a first sub-power supply sequence; delete a preset number of data points from the beginning in the power supply characteristic monitoring sequence to obtain a second sub-power supply sequence; the size of the preset time period does not exceed half of the number of data points in the power supply characteristic monitoring sequence; obtain the autocovariance corresponding to the preset time period according to the first sub-power supply sequence, the second sub-power supply sequence and the overall data mean value;

[0012] Calculate the ratio of the autocovariance to the overall fluctuation value to obtain the characteristic autocorrelation coefficient corresponding to the preset time period.

[0013] Further, the step of obtaining the autocorrelation coefficient sequence according to the characteristic autocorrelation coefficients corresponding to different preset time periods includes:

[0014] Sort the characteristic autocorrelation coefficients corresponding to the preset time periods in ascending order of the preset time periods to obtain the autocorrelation coefficient sequence.

[0015] Further, the step of obtaining the optimal time window length according to the data change characteristics of the autocorrelation coefficient sequence includes:

[0016] Construct a rectangular coordinate system for the autocorrelation coefficient sequence, where the horizontal axis of the rectangular coordinate system is different preset time periods and the vertical axis is the value of the characteristic autocorrelation coefficient corresponding to the preset time period; calculate the absolute value of the difference between the tangent slopes of the characteristic autocorrelation coefficient and the other adjacent characteristic autocorrelation coefficients before and after in the rectangular coordinate system to obtain the change speed value of the characteristic autocorrelation coefficient; calculate the sum value of the change speed values to obtain the key characteristic value of the characteristic autocorrelation coefficient; calculate the product of the value of the characteristic autocorrelation coefficient and the key characteristic value to obtain the optimal characteristic value of the characteristic autocorrelation coefficient; use the preset time period corresponding to the maximum value of the optimal characteristic value as the optimal time window length.

[0017] Further, the step of obtaining different dimensional change characteristic values according to the data change characteristics within the optimal time window length of the target data point in the power supply characteristic monitoring sequence includes:

[0018] The dimension change characteristic values include mean characteristic values, standard deviation characteristic values, median characteristic values, and interquartile range characteristic values;

[0019] Calculate the average value of the data point values within the optimal time window length of the target data point to obtain the mean characteristic value of the target data point; calculate the standard deviation of the data point values within the optimal time window length of the target data point to obtain the standard deviation characteristic value of the target data point; calculate the median of the data point values within the optimal time window length of the target data point to obtain the median characteristic value of the target data point; calculate the interquartile range of the data point values within the optimal time window length of the target data point to obtain the interquartile range characteristic value of the target data point.

[0020] Further, the step of obtaining the dimension anomaly degree of the target data point according to the distribution difference characteristics between the same type of dimension change characteristic values includes:

[0021] Construct a distribution histogram of the same type of dimension change characteristic values, where the horizontal axis of the distribution histogram is the different values in the same type of dimension change characteristic values, and the vertical axis is the number of target data points corresponding to different dimension change characteristic values; take the maximum value corresponding to the vertical axis of the distribution histogram as the high-frequency characterization value; calculate the absolute value of the difference between the vertical axis value corresponding to the dimension change characteristic value of any target data point in the distribution histogram and the high-frequency characterization value and normalize it to obtain the dimension anomaly degree of the any target data point.

[0022] Further, the step of obtaining the overall anomaly degree of the data segment within the optimal time window length of the target data point according to different dimension anomaly degrees includes:

[0023] Calculate the average value of different dimension anomaly degrees of the target data point; obtain the overall anomaly degree of the data segment within the optimal time window length of the target data point.

[0024] Further, the step of obtaining the abnormal sequence segment in the power supply characteristic monitoring sequence according to the overall anomaly degree includes:

[0025] When the overall anomaly degree exceeds the preset anomaly threshold, mark the data segment corresponding to the overall anomaly degree to obtain the marked data segment; calculate the union of the marked data segments to obtain the abnormal sequence segment in the power supply characteristic monitoring sequence.

[0026] Further, the step of obtaining the abnormal data point according to the data distribution characteristics in the abnormal sequence segment includes:

[0027] Construct a normal distribution curve based on the data point values in the abnormal sequence segment, and use the 3-sigma rule to take the data points exceeding the preset judgment interval as abnormal data points.

[0028] The present invention also provides a power supply monitoring system for microgrid power distribution, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the steps of any one of the power supply monitoring methods for microgrid power distribution.

[0029] The present invention has the following beneficial effects:

[0030] In the present invention, obtaining the feature autocorrelation coefficient can characterize the similarity degree of data in the power supply feature monitoring sequence at different preset time intervals; obtaining the autocorrelation coefficient sequence can reflect the differences in the feature autocorrelation coefficients corresponding to different preset time intervals; obtaining the optimal time window length can characterize the length of the change period of the power supply feature monitoring sequence, thereby improving the accuracy of obtaining abnormal sequence segments and abnormal data points. Obtaining the dimension change feature value can reflect the data features within the neighborhood of the target data point, and then the dimension abnormality degree and the overall abnormality degree can be obtained based on the differences in the dimension change feature values. Obtaining the overall abnormality degree can avoid the error of the dimension abnormality degree and improve the accuracy of obtaining abnormal sequence segments. Finally, the power supply of the microgrid is monitored according to the abnormal data points, improving the monitoring accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0032] Figure 1 It is a flowchart of a power supply monitoring method for microgrid power distribution provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following, in conjunction with the drawings and preferred embodiments, details the specific implementation manners, structures, features, and effects of a power supply monitoring method and system for microgrid power distribution proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs.

[0035] The following specifically describes the specific solutions of a power supply monitoring method and system for microgrid power distribution provided by the present invention in conjunction with the accompanying drawings.

[0036] Please refer to Figure 1 , which shows a flowchart of a power supply monitoring method for microgrid power distribution provided by an embodiment of the present invention. The method includes the following steps:

[0037] Step S1, obtaining a power supply feature monitoring sequence for monitoring the microgrid; obtaining a feature autocorrelation coefficient according to the data difference feature between data separated by a preset time period in the power supply feature monitoring sequence.

[0038] In the embodiment of the present invention, the implementation scenario is the power supply monitoring of the microgrid power distribution; first, a power supply feature monitoring sequence for monitoring the microgrid is obtained. The power supply data of the microgrid power distribution mainly covers various data such as monitoring the renewable energy power generation and the performance of the energy storage system. In the embodiment of the present invention, the voltage time series data of the energy storage system is collected to obtain a power supply feature monitoring sequence of the voltage of the energy storage system. It should be noted that the implementer can determine the feature object and the collection frequency for monitoring the power supply of the microgrid according to the implementation scenario.

[0039] Traditional power supply monitoring identifies faults based on a single threshold. This method can only detect significantly abnormal monitoring data. For monitoring data with a fault risk but not reaching the set threshold, since a large amount of normal data in the time series is likely to submerge a small amount of data with a fault risk, it is difficult to discover and screen the monitoring data with a fault risk. Therefore, it is necessary to split and screen the power supply feature monitoring sequence, obtain sequence segments that may be abnormal for analysis, and obtain accurate abnormal values. The length of the sequence segment is particularly important for the analysis of abnormal data. An overly large sequence segment is difficult to capture the changes in abnormal data, and an overly small sequence segment is difficult to represent the complete change trend of abnormal data. Therefore, by analyzing the autocorrelation coefficient of the power supply feature monitoring sequence, the optimal window size is obtained according to the autocorrelation coefficient, and data analysis is performed according to the optimal window to obtain abnormal data.

[0040] Further, when the microgrid system is operating normally, there is a strong correlation between the time-series data of the monitored characteristic objects before and after. The data at different acquisition times in the power supply characteristic monitoring sequence are relatively similar, and the time-series data shows a certain periodic correlation, with a large similarity between the current value and the historical value. In the power supply characteristic monitoring sequence, the more similar the data segment at the current moment is to the data segment at a certain historical moment, the greater the autocorrelation coefficient. The similarity between the data segment at the current moment and the data segments at different historical moments is different, and the corresponding autocorrelation coefficients are also different; when the autocorrelation coefficient is larger, the interval length between the current moment and the historical moment is closer to the complete change period of the power supply characteristic data; therefore, the characteristic autocorrelation coefficient is obtained according to the data difference characteristics of the data separated by a preset time period in the power supply characteristic monitoring sequence.

[0041] Preferably, in an embodiment of the present invention, obtaining the characteristic autocorrelation coefficient includes: calculating the average value of the data point values in the power supply characteristic monitoring sequence to obtain the overall data mean value; calculating the variance of the data point values in the power supply characteristic monitoring sequence to obtain the overall fluctuation value; deleting a preset number of data points from the end in the power supply characteristic monitoring sequence to obtain a first sub-power supply sequence; deleting a preset number of data points from the beginning in the power supply characteristic monitoring sequence to obtain a second sub-power supply sequence; the size of the preset time period does not exceed half of the number of data points in the power supply characteristic monitoring sequence, otherwise it will affect the calculation accuracy of the autocorrelation coefficient; obtaining the autocovariance corresponding to the preset time period according to the first sub-power supply sequence, the second sub-power supply sequence and the overall data mean value; calculating the ratio of the autocovariance to the overall fluctuation value to obtain the characteristic autocorrelation coefficient corresponding to the preset time period. It should be noted that the acquisition of the autocorrelation coefficient belongs to the prior art, and the specific steps will not be elaborated; when the first sub-power supply sequence and the second sub-power supply sequence are more similar, the characteristic autocorrelation coefficient is larger; the preset time period is the number of data points between the two sub-power supply sequences. For example, if the positions of the data points in the power supply characteristic monitoring sequence are from 1 to 100, the length range of the preset time period is from 1 to 50; when the length of the preset time period is 10, the data point positions of the first sub-power supply sequence are from 1 to 90, and the data point positions of the second sub-power supply sequence are from 11 to 100.

[0042] Step S2, obtaining an autocorrelation coefficient sequence according to the characteristic autocorrelation coefficients corresponding to different preset time periods; obtaining the optimal time window length according to the data change characteristics of the autocorrelation coefficient sequence; obtaining different dimension change characteristic values according to the data change characteristics within the optimal time window length of the target data point in the power supply characteristic monitoring sequence.

[0043] The characteristic autocorrelation coefficients obtained at different preset time periods are different. When the length of the preset time period is one data change cycle of the power supply characteristic monitoring sequence, it means that the first sub-power supply sequence and the second sub-power supply sequence are separated by one data change cycle. In this case, under normal circumstances, the first sub-power supply sequence and the second sub-power supply sequence are the most similar, and the obtained characteristic autocorrelation coefficient is the largest. Therefore, an autocorrelation coefficient sequence is obtained based on the characteristic autocorrelation coefficients corresponding to different preset time periods. Preferably, in an embodiment of the present invention, obtaining the autocorrelation coefficient sequence includes: sorting the characteristic autocorrelation coefficients corresponding to the preset time periods in ascending order of the preset time periods to obtain the autocorrelation coefficient sequence.

[0044] Furthermore, the larger the characteristic autocorrelation coefficient is, the closer the length of the preset time period is to one data change cycle in the power supply characteristic monitoring sequence. Therefore, the optimal time window length can be obtained according to the data change characteristics of the autocorrelation coefficient sequence. Preferably, in an embodiment of the present invention, obtaining the optimal time window length includes: constructing a rectangular coordinate system for the autocorrelation coefficient sequence, where the horizontal axis of the rectangular coordinate system is different preset time periods, and the vertical axis is the value of the characteristic autocorrelation coefficient corresponding to the preset time period; calculating the absolute value of the difference between the tangent slopes of the characteristic autocorrelation coefficient and the other adjacent characteristic autocorrelation coefficients before and after in the rectangular coordinate system to obtain the change speed value of the characteristic autocorrelation coefficient; the larger the change speed value is, the greater the change of the characteristic autocorrelation coefficient. Calculate the sum value of the change speed values to obtain the key characteristic value of the characteristic autocorrelation coefficient; the larger the key characteristic value is, the more likely the characteristic autocorrelation coefficient corresponding to the preset time period is at the inflection point of the peak and valley in the rectangular coordinate system, and the trend of the characteristic autocorrelation coefficient has a reverse change. Calculate the product of the value of the characteristic autocorrelation coefficient and the key characteristic value to obtain the optimal characteristic value of the characteristic autocorrelation coefficient; the larger the optimal characteristic value is, the more the length of the preset time period conforms to one change cycle of the power supply characteristic monitoring sequence. Take the preset time period corresponding to the maximum value of the optimal characteristic value as the optimal time window length. The optimal time window length is the length of one data change cycle of the power supply characteristic monitoring sequence, and the optimal time window length can better reflect the data change trend within the complete cycle, thereby improving the accuracy of abnormal data identification and screening.

[0045] After obtaining the optimal time window length of the power supply feature monitoring sequence, different dimensional change feature values can be obtained according to the data change characteristics within the optimal time window length of the target data point in the power supply feature monitoring sequence. Preferably, in an embodiment of the present invention, obtaining different dimensional change feature values includes: the dimensional change feature values include mean feature values, standard deviation feature values, median feature values, and interquartile range feature values; the dimensional change feature values of these four dimensions can all reflect the data characteristics and data change characteristics within the optimal time window length of the target data point. The target data point is any data point in the power supply feature monitoring sequence, and the target data point is located at the central position within the optimal time window length of the target data point. Further, calculate the average value of the data point values within the optimal time window length of the target data point to obtain the mean feature value of the target data point; calculate the standard deviation of the data point values within the optimal time window length of the target data point to obtain the standard deviation feature value of the target data point; calculate the median of the data point values within the optimal time window length of the target data point to obtain the median feature value of the target data point; calculate the interquartile range of the data point values within the optimal time window length of the target data point to obtain the interquartile range feature value of the target data point.

[0046] Step S3, obtain the dimensional abnormality degree of the target data point according to the distribution difference characteristics between the dimensional change feature values of the same type; obtain the overall abnormality degree of the data segment within the optimal time window length of the target data point according to different dimensional abnormality degrees; obtain the abnormal sequence segment in the power supply feature monitoring sequence according to the overall abnormality degree.

[0047] When there is a fault risk in the sequence segment where the target data point is located, the dimensional change feature value corresponding to the target data point is quite different from the dimensional change feature values of other target data points; since the data points with fault risks are in the minority, among all the dimensional change feature values of the same type, the dimensional change feature values of most normal target data points are similar, while the dimensional change feature values of a small number of target data points with abnormal risks are relatively abnormal. Therefore, obtain the dimensional abnormality degree of the target data point according to the distribution difference characteristics between the dimensional change feature values of the same type.

[0048] Preferably, in an embodiment of the present invention, the step of obtaining the dimension anomaly degree includes: constructing a distribution histogram of the dimension change eigenvalue of the same category, and there is only one category of dimension change eigenvalue in each distribution histogram. The horizontal axis of the distribution histogram is different values in the dimension change eigenvalue of the same category, and the vertical axis is the number of target data points corresponding to different dimension change eigenvalues; taking the maximum value corresponding to the vertical axis of the distribution histogram as the high-frequency characterization value; the high-frequency characterization value means that the number of target data points corresponding to this dimension change eigenvalue is the largest, and this dimension change eigenvalue is more normal. Calculate the absolute value of the difference between the vertical axis value corresponding to the dimension change eigenvalue of any target data point in the distribution histogram and the high-frequency characterization value and normalize it to obtain the dimension anomaly degree of any target data point. When the dimension anomaly degree is larger, it means that the dimension change eigenvalue of this arbitrary target data point is less common, and this arbitrary target data point is more likely to be in an abnormal sequence segment; the formula for obtaining the dimension anomaly degree includes:

[0049] G i = 1 - exp(-|N i - M|)

[0050] In the formula, G i represents the dimension anomaly degree of the i-th target data point, N i represents the vertical axis value corresponding to the dimension change eigenvalue of the i-th target data point in the distribution histogram, M represents the high-frequency characterization value, and exp( ) represents the exponential function with the natural constant as the base.

[0051] Furthermore, if the dimension anomaly degree of each category of the target data point is larger, it means that the possibility of this target data point in the abnormal sequence segment is greater, and the accuracy of judging the abnormal characteristics of the target data point based only on the dimension anomaly degree of a certain category is higher; therefore, the overall anomaly degree of the data segment within the optimal time window length of the target data point is obtained according to different dimension anomaly degrees.

[0052] Preferably, in an embodiment of the present invention, obtaining the overall anomaly degree includes: calculating the average value of the different dimension anomaly degrees of the target data point; obtaining the overall anomaly degree of the data segment within the optimal time window length of the target data point. When the overall anomaly degree of the target data point is larger, it means that this target data point is more likely to be in an abnormal sequence segment. Therefore, the abnormal sequence segment in the power supply characteristic monitoring sequence is obtained according to the overall anomaly degree, specifically including: when the overall anomaly degree exceeds the preset anomaly threshold, marking the data segment corresponding to the overall anomaly degree to obtain the marked data segment; calculating the union of the marked data segments to obtain the abnormal sequence segment in the power supply characteristic monitoring sequence; in the embodiment of the present invention, the preset anomaly threshold is 0.6, and the implementer can determine it according to the implementation scenario; the data characteristics of some data points in the abnormal sequence segment characterize the fault risk, and the data in the abnormal sequence segment needs to be analyzed key points during monitoring.

[0053] Step S4: Obtain abnormal data points according to the data distribution characteristics in the abnormal sequence segment; monitor the power supply status of the microgrid based on the abnormal data points.

[0054] Obtaining the abnormal sequence segment in the power supply characteristic monitoring sequence can eliminate a large number of normal data points, greatly reducing the interference of normal data points on abnormal data points and avoiding the problem that abnormal data points with potential fault risks are submerged. Since the abnormal data points only account for a small proportion in the abnormal sequence segment, the distribution of the data point values in the obtained abnormal sequence segment is more in line with the characteristics of the normal distribution. The values of most normal data points are similar, and the values of a small number of abnormal data points are relatively similar. In the normal distribution curve constructed based on the data point values in the abnormal sequence segment, most normal data points are in the middle region, and abnormal data points are at both ends. Then, abnormal data points are obtained according to the data distribution characteristics in the abnormal sequence segment, specifically including: constructing a normal distribution curve based on the data point values in the abnormal sequence segment, and taking the data points exceeding the preset judgment interval as abnormal data points according to the 3σ criterion; the 3σ criterion believes that the data beyond three standard deviations from the average value in the normal distribution curve is abnormal data with gross error. In the embodiment of the present invention, the preset judgment interval is the interval within three standard deviations above and below the average value in the normal distribution curve; the data points outside the preset judgment interval are abnormal data points.

[0055] After obtaining the abnormal data points with potential fault risks in the power supply characteristic monitoring sequence, the power supply status of the microgrid can be monitored based on the abnormal data points. When the number of abnormal data points per unit time exceeds the preset quantity threshold, a warning is issued; the implementer can set the warning rule by himself / herself, which is not limited here. Thus, by obtaining the abnormal sequence segment and abnormal data points in the power supply characteristic monitoring sequence, the power supply monitoring of the microgrid power distribution is carried out, avoiding the problem that abnormal data points with potential fault risks are difficult to detect and improving the monitoring accuracy.

[0056] In summary, the embodiment of the present invention provides a power supply monitoring method for the microgrid power distribution. The characteristic autocorrelation coefficient and the autocorrelation coefficient sequence are obtained according to the data difference characteristics of the data separated by the preset time period in the power supply characteristic monitoring sequence; the optimal time window length is obtained according to the data change characteristics of the autocorrelation coefficient sequence; different dimension change characteristic values and dimension abnormality degrees are obtained according to the data change characteristics of the target data points within the optimal time window length in the power supply characteristic monitoring sequence; the overall abnormality degree and the abnormal sequence segment are obtained according to the dimension abnormality degree. The present invention obtains abnormal data points according to the data distribution characteristics in the abnormal sequence segment; monitors the power supply status of the microgrid based on the abnormal data points, improving the power supply monitoring accuracy.

[0057] The present invention also provides a power supply monitoring system for microgrid power distribution, including a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the computer program to implement the steps of any one of the power supply monitoring methods for microgrid power distribution.

[0058] It should be noted that the above-mentioned sequence of embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0059] Each embodiment in this specification is described in a progressive manner. For the same or similar parts among the embodiments, reference can be made to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A power supply monitoring method for microgrid power distribution, characterized in that: The method comprises the following steps: Acquire a power supply feature monitoring sequence for monitoring a microgrid; obtain a feature autocorrelation coefficient according to data difference characteristics separated by a preset time period in the power supply feature monitoring sequence; Obtain an autocorrelation coefficient sequence according to characteristic autocorrelation coefficients corresponding to different preset time periods; obtain an optimal time window length according to the data change characteristics of the autocorrelation coefficient sequence; obtain different dimensional change characteristic values ​​according to the data change characteristics within the optimal time window length range of the target data point in the power supply characteristic monitoring sequence; Obtain the dimensional anomaly degree of the target data point according to the distribution difference characteristics between the dimensional change feature values ​​of the same type; obtain the overall anomaly degree of the data segment within the optimal time window length range of the target data point according to different dimensional anomaly degrees; obtain the abnormal sequence segment in the power supply feature monitoring sequence according to the overall anomaly degree; Obtaining abnormal data points according to the data distribution characteristics in the abnormal sequence segment; monitoring the power supply status of the microgrid according to the abnormal data points; The step of obtaining a characteristic autocorrelation coefficient according to the data difference characteristics separated by a preset time period in the power supply characteristic monitoring sequence comprises: Calculate the average value of the data point values ​​in the power supply feature monitoring sequence to obtain the overall data mean; calculate the variance of the data point values ​​in the power supply feature monitoring sequence to obtain the overall fluctuation value; A first sub-power supply sequence is obtained by deleting a preset time period number of data points from the last position in the power supply feature monitoring sequence; a second sub-power supply sequence is obtained by deleting a preset time period number of data points from the first position in the power supply feature monitoring sequence; the size of the preset time period does not exceed half of the number of data points in the power supply feature monitoring sequence; the autocovariance corresponding to the preset time period is obtained according to the first sub-power supply sequence, the second sub-power supply sequence and the overall data mean; Calculating the ratio of the autocovariance to the overall fluctuation value to obtain a characteristic autocorrelation coefficient corresponding to a preset time period; The step of obtaining the optimal time window length according to the data variation characteristics of the autocorrelation coefficient sequence comprises: Constructing a rectangular coordinate system about the autocorrelation coefficient sequence, wherein the horizontal axis of the rectangular coordinate system is different preset time periods, and the vertical axis is the value of the characteristic autocorrelation coefficient corresponding to the preset time period; calculating the absolute value of the difference between the characteristic autocorrelation coefficient and the tangent slope of other adjacent characteristic autocorrelation coefficients in the rectangular coordinate system, respectively, to obtain the change rate value of the characteristic autocorrelation coefficient; calculating the sum of the change rate values, to obtain the key characteristic value of the characteristic autocorrelation coefficient; calculating the product of the value of the characteristic autocorrelation coefficient and the key characteristic value, to obtain the optimal characteristic value of the characteristic autocorrelation coefficient; and taking the preset time period corresponding to the maximum value of the optimal characteristic value as the optimal time window length; The step of obtaining different dimensional change feature values ​​according to the data change characteristics within the optimal time window length range of the target data point in the power supply feature monitoring sequence comprises: The dimension change characteristic values ​​include mean characteristic value, standard deviation characteristic value, median characteristic value and interquartile range characteristic value; Calculate the average value of the data point values ​​within the optimal time window length of the target data point to obtain the mean characteristic value of the target data point; calculate the standard deviation of the data point values ​​within the optimal time window length of the target data point to obtain the standard deviation characteristic value of the target data point; calculate the median of the data point values ​​within the optimal time window length of the target data point to obtain the median characteristic value of the target data point; calculate the interquartile range of the data point values ​​within the optimal time window length of the target data point to obtain the interquartile range characteristic value of the target data point.

2. A power supply monitoring method for microgrid power distribution according to claim 1, characterized in that: The step of obtaining an autocorrelation coefficient sequence according to characteristic autocorrelation coefficients corresponding to different preset time periods comprises: The characteristic autocorrelation coefficients corresponding to the preset time periods are sorted in order from small to large to obtain the autocorrelation coefficient sequence.

3. A power supply monitoring method for microgrid power distribution according to claim 1, characterized in that: The step of obtaining the dimensional anomaly degree of the target data point according to the distribution difference characteristics between the dimensional change characteristic values ​​of the same type includes: Construct a distribution histogram of the same type of dimensional change characteristic values, where the horizontal axis of the distribution histogram is the different numerical values ​​in the same type of dimensional change characteristic values, and the vertical axis is the number of target data points corresponding to different dimensional change characteristic values; use the maximum value corresponding to the vertical axis of the distribution histogram as the high-frequency characterization value; calculate the absolute value of the difference between the vertical axis value corresponding to the dimensional change characteristic value of any target data point in the distribution histogram and the high-frequency characterization value and normalize them to obtain the dimensional anomaly degree of the arbitrary target data point.

4. A power supply monitoring method for microgrid power distribution according to claim 1, characterized in that: The step of obtaining the overall abnormality degree of the data segment within the optimal time window length of the target data point according to the abnormality degrees of different dimensions includes: Calculate the average value of the abnormality degree of different dimensions of the target data point; obtain the overall abnormality degree of the data segment within the optimal time window length range of the target data point.

5. The power supply monitoring method for microgrid power distribution according to claim 1, characterized in that: The step of obtaining an abnormal sequence segment in the power supply feature monitoring sequence according to the overall abnormality degree comprises: When the overall abnormality level exceeds a preset abnormality threshold, the data segment corresponding to the overall abnormality level is marked to obtain a marked data segment; and the union of the marked data segments is calculated to obtain an abnormal sequence segment in the power supply feature monitoring sequence.

6. A power supply monitoring method for microgrid power distribution according to claim 1, characterized in that: The step of obtaining abnormal data points according to the data distribution characteristics in the abnormal sequence segment comprises: A normal distribution curve is constructed according to the data point values ​​in the abnormal sequence segment, and the data points exceeding the preset judgment interval are regarded as abnormal data points through the Laida criterion.

7. A power supply monitoring system for microgrid power distribution, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

Images