Ring network box with historical data intelligent storage function

Abstract

The invention relates to the technical field of ring main unit data acquisition, in particular to a ring main unit with a historical data intelligent storage function. The ring main unit comprises a difference analysis module used for segmenting power grid parameter data collected by the ring main unit to obtain data segments; obtaining a difference distance between every two data segments; the clustering analysis module is used for clustering the data segments to obtain an importance degree coefficient of each cluster; the in-cluster analysis module is used for acquiring the relative importance coefficient of each data segment in the clustering cluster; the fusion analysis module is used for obtaining the final importance coefficient of the data segment according to the relative importance coefficient of the data segment and the importance degree coefficient of the cluster where the data segment is located; the compression storage module is used for processing the component signal of each data segment based on the importance coefficient to obtain an updated data segment corresponding to each data segment; and compressing and storing the updated data segment. The pressure of the storage space of the ring main unit can be reduced, and key information of data is reserved.

Description

Technical Field

[0001] The present invention relates to the technical field of data acquisition of ring main units, and particularly to a ring main unit with an intelligent storage function for historical data. Background Art

[0002] With the continuous development of distribution network automation systems and technologies, the market's requirements for ring main unit (cabinet type switchgear) products are also getting higher and higher. The original monitoring equipment only monitors the operation status of the power grid, and it may not meet the needs of the practical application of on-site function requirements and the integrated development of monitoring and protection. Therefore, a more comprehensive ring main unit (cabinet type switchgear) is needed. This ring main unit is installed in places such as distribution rooms and switchgear rooms. Its main function is to collect the power grid parameter data of the line where the monitored switch is located, and has the functions of data acquisition and storage, real-time monitoring of the line operation status, reporting in time when a line fault occurs, and waiting for instructions from the superior system to control the opening / closing of the switch. However, as the stored data collected by the ring main unit accumulates over time, it also poses a certain challenge to the storage space of the device. Therefore, how to effectively reduce the storage space of these collected data without affecting the accuracy of its monitoring and analysis is of crucial importance and significance for the ring main unit.

[0003] During the process of the ring main unit collecting, storing, and monitoring and analyzing data in the power grid (such as voltage, current, etc. data), although the data storage method can be compressed using traditional compression algorithms, such as compression storage using ZIP, RLE and other compression algorithms, due to the particularity of the data in the power grid, when compressing the data, it may ignore the time correlation of these collected data and lack of scoring of the importance of these data, and often uses the default method of treating all equally for compressing and storing the power grid collected data. If the set lossy compression ratio for the collected data is set too small, it may lead to an increase in storage redundancy and cause a certain pressure on the storage space. If the lossy compression ratio is set too large, it may lead to the loss of some details, thus affecting the accuracy of subsequent monitoring and other analyses. Summary of the Invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a ring main unit with an intelligent storage function for historical data, and the specific technical solution adopted is as follows: An embodiment of the present invention provides a ring main unit with an intelligent storage function for historical data, and the ring main unit includes: A difference analysis module, configured to segment the power grid parameter data collected by the ring main unit to obtain data segments; and obtain the difference distance between every two data segments according to the spectrum data corresponding to each data segment; The clustering analysis module is used to cluster all data segments using the differential distance to obtain clustering clusters; and obtain the importance coefficient of a clustering cluster based on the number of data points in a clustering cluster and the distance between it and other clustering clusters; The intra-cluster analysis module is used to obtain the relative importance coefficient of a data segment within a clustering cluster according to the distance between the data segment and the current moment in time series and the differential distance between the data segment and the clustering cluster center; The fusion analysis module is used to obtain the final importance coefficient of a data segment according to the relative importance coefficient of the data segment and the importance coefficient of the clustering cluster where it is located; The compression storage module is used to obtain the component signals of each data segment; process the component signals of each data segment based on the importance coefficient to obtain the updated data segment corresponding to each data segment; and compressively store the updated data segments.

[0005] Preferably, obtaining the differential distance between every two data segments according to the spectral data corresponding to each data segment includes: Comparing the amplitude corresponding to a non-zero amplitude frequency point in the frequency data corresponding to a data segment with the sum of the amplitudes corresponding to all non-zero amplitude frequency points in the frequency data corresponding to the data segment to obtain the contribution degree of this frequency point; obtaining the differential distance between two data segments according to the difference between the contribution degrees of the non-zero amplitude frequency points in the spectral data corresponding to the two data segments.

[0006] Preferably, obtaining the differential distance between two data segments according to the difference between the contribution degrees of the non-zero amplitude frequency points in the spectral data corresponding to the two data segments includes: Finding the union of the non-zero amplitude frequency points in the spectral data corresponding to the two data segments; using the Euclidean distance calculation formula to calculate the contribution degrees corresponding to each frequency point in the union in the spectral data of the two data segments to obtain the differential distance between the two data segments.

[0007] Preferably, obtaining the importance coefficient of a clustering cluster based on the number of data points in a clustering cluster and the distance between it and other clustering clusters includes: Calculating the average value of the differential distances between the clustering center of a clustering cluster and the clustering centers of other clustering clusters, denoted as the average distance of this clustering cluster; comparing the number of data segments in this clustering cluster with the number of data segments in all clustering clusters to obtain the proportion of the number of data segments in this clustering cluster; comparing the average distance of this clustering cluster with the average distances of all clustering clusters to obtain the distance proportion of this clustering cluster; using a first preset value to perform a negative correlation mapping on the proportion of the number of data segments in this clustering cluster and multiplying it by the distance proportion to obtain the importance coefficient of this clustering cluster.

[0008] Preferably, obtaining the relative importance coefficient of a data segment within a clustering cluster according to the distance of the data segment from the current moment in time series and the difference distance between the data segment and the clustering cluster center includes: Taking the time length of the middle moment of a data segment within a clustering cluster from the current moment, and using the exponential function with the natural constant as the base to perform a negative correlation mapping on the time length to obtain the time series related eigenvalue of the data segment; using the exponential function with the natural constant as the base to map the difference distance between the data segment and the clustering cluster center of the clustering cluster where the data segment is located to obtain the distance distribution eigenvalue; normalizing and weighted summing the time series related eigenvalue and the distance distribution eigenvalue to obtain the relative importance coefficient of the data segment.

[0009] Preferably, obtaining the final importance coefficient of a data segment according to the relative importance coefficient of the data segment and the importance degree coefficient of the clustering cluster where the data segment is located includes: Comparing the relative importance coefficient of a data segment with the relative importance coefficient of the data segment closest to the clustering cluster center within the clustering cluster where the data segment is located to obtain the importance ratio corresponding to the data segment; multiplying the configuration constant term, the importance degree coefficient of the clustering cluster where the data segment is located, and the importance ratio corresponding to the data segment and normalizing to obtain the final importance coefficient of the data segment.

[0010] Preferably, obtaining the component signal of each data segment includes: Decomposing each data segment using EMD to respectively obtain the component signals corresponding to each data segment.

[0011] Preferably, processing the component signals of each data segment based on the importance coefficient to obtain the updated data segment corresponding to each data segment includes: Arranging the component signals of a data segment in the order of obtaining the component signals during decomposition to obtain a component signal sequence; taking the component signals from the last component signal in the component signal sequence forward until the ratio of the number of component signals taken out to the number of component signals in the component signal sequence is equal to the final importance coefficient of the data segment, stopping the extraction of component signals, and the extracted component signals are the retained component signals of the data segment. Performing an inverse transformation on the retained component signals to obtain the updated data segment of the data segment.

[0012] The embodiments of the present invention have at least the following beneficial effects: In this application, the power grid parameter data collected by the ring network cabinet is segmented to obtain data segments, and then the Fourier transform is performed on different data segments to obtain spectrum data. Then, the difference distance between any two data segments is obtained based on the spectrum data, and then clustering analysis is performed on the analyzed data segments according to the difference distance between the data segments to obtain the importance degree coefficient of each clustering cluster. However, since the density and the like of different clustering clusters may be different, for the specific data within the clustering cluster, its relative importance is further analyzed, and then the relative importance coefficient of the data segment is obtained. Then, in combination with the importance degree coefficient of the clustering cluster, the final importance coefficient of each data segment is obtained. Then, EMD decomposition is performed on each data segment to obtain different component signals, and then the component signals of the data segment are retained according to the importance coefficient, so as to obtain the updated data segment corresponding to each data segment. Then, the data segment processed by the compression algorithm is compressed to achieve adaptive lossy compression of different data segments, which not only reduces the pressure on the storage space of the ring network cabinet, but also reduces the deviation influence on the accuracy during subsequent system monitoring and analysis caused by the loss of data compression. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] 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, without creative efforts, other drawings can be obtained according to these drawings.

[0014] Figure 1 It is a structural block diagram of a ring network cabinet with a historical data intelligent storage function provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] In order to further elaborate on the technical means and effects adopted by the present invention to achieve the intended invention purpose, the following, in combination with the drawings and preferred embodiments, details the specific implementation manner, structure, features and effects of a ring network cabinet with a historical data intelligent storage function 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.

[0016] 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.

[0017] The following specifically describes the specific solution of a ring network cabinet with a historical data intelligent storage function provided by the present invention in combination with the drawings.

[0018] Embodiment: The main application scenario of the present invention is as follows: during the process of a ring main unit collecting, storing, and monitoring data in the power grid (such as voltage, current, etc. data), the existing compression algorithms ignore the analysis of the time correlation and importance of these data when compressing and storing them, and compress and store all the data, resulting in a waste of storage resources. Therefore, it is necessary to analyze the collected data to determine the compressed data.

[0019] Please refer to Figure 1 , which shows a structural block diagram of a ring main unit with an intelligent storage function for historical data provided by an embodiment of the present invention. The ring main unit includes the following modules: A difference analysis module, configured to segment the power grid parameter data collected by the ring main unit to obtain data segments; and obtain the difference distance between every two data segments according to the spectrum data corresponding to each data segment.

[0020] The main function of the ring main unit is to collect power grid parameter data in the power grid, such as voltage, current, etc. data, and then use a large amount of historical data as training sample data for decision-making models such as an anomaly monitoring model and power energy regulation for training and analysis. However, when the traditional algorithm stores the electrical parameter data collected by the ring main unit, such as voltage, current, etc. data, due to the passage of time, it also brings certain challenges to the storage space pressure of the ring main unit.

[0021] Therefore, it is necessary to analyze the power grid parameter data collected by the ring main unit, so as to minimize the impact on the accuracy of functions such as monitoring of the ring main unit as much as possible, effectively improve the compression ratio of these data, and further reduce the pressure on the storage space of the ring main unit and extend the service life of the device, etc.

[0022] The power grid parameter data collected by the ring main unit are time series data such as voltage data and current data. Furthermore, analyze the two types of power grid parameter data, and then process them to reduce the pressure on the storage space of the ring main unit when storing them. Therefore, it is necessary to use data acquisition modules such as voltage and current to collect voltage, current, etc. data in the power grid in real time. These data are processed by an analysis and processing module, and then the processed data are compressed and stored using the compression algorithm of the compression module. Among them, the power grid parameter data are current data and voltage data. The acquisition frequency of the power grid parameter data uses the acquisition frequency of a conventional ring main unit.

[0023] Since there are certain dynamic and random changes in the current, voltage, etc. in the power grid during use, and their change characteristics are different in different time periods, then their importance levels may also be different. Therefore, it is necessary to minimize the space occupied when storing data as much as possible, and at the same time minimize the impact on the accuracy of analysis such as monitoring of the processed data.

[0024] Further, taking a kind of power grid parameter data for analysis, in the embodiments of the present application, current data is taken as an example for analysis. Taking the historical current data two hours before the current moment as an example for analysis, first, the currently collected current data is divided into small segments according to a certain time period. Taking 50 Hz alternating current as an example, its period is 20 milliseconds. Therefore, when dividing it into small segments, at least one period of this alternating current is used as the lower limit, and a suitable length is selected. In this case, the selected length is 60 milliseconds as a preset segmentation length (specifically selected according to the space and computing power of the device. The smaller the segmentation length, the more detailed the analysis and the better the compression effect may be. However, if you want a fast calculation speed and low complexity, a longer segmentation is required. Therefore, the selection of the segmentation is specifically configured and decided according to the storage space size and computing performance of the device).

[0025] Use the preset segmentation length to segment the power grid parameter data (current data) collected by the ring main unit to obtain data segments. For each data segment, use the fast Fourier transform to obtain the spectral data corresponding to each data segment.

[0026] For the arbitrarily segmented current data, the amplitude corresponding to any frequency represents its contribution degree to the current signal. By comparing the segmented data after introducing the frequency contribution degree, the actual influence degree of different frequency components in the signal can be more accurately reflected, avoiding the problem of masking key feature changes due to simple amplitude differences. Especially for power grid signals mainly featuring main frequency and low-frequency characteristics, through the difference calculation based on the contribution degree, the changes in the main frequency energy distribution of the signal can be more sensitively captured, thereby improving the rationality and accuracy of segmentation classification and subsequent compression decision-making.

[0027] Obtain the difference distance between every two data segments according to the spectral data corresponding to each data segment. Specifically, divide the amplitude corresponding to a non-zero amplitude frequency point in the frequency data corresponding to a data segment by the sum of the amplitudes corresponding to all non-zero amplitude frequency points in the frequency data corresponding to this data segment to obtain the contribution degree of this frequency point.

[0028] The calculation model of the contribution degree is specifically as follows: , Among them, represents the contribution degree corresponding to the i-th frequency point with non-zero amplitude in the frequency data corresponding to the a-th data segment, represents the amplitude corresponding to the i-th frequency point with non-zero amplitude in the frequency data corresponding to the a-th data segment (if the amplitude is 0, its contribution degree is 0 and the calculation is directly skipped), the number of frequency points with non-zero amplitude in the frequency data corresponding to the a-th data segment; It represents the ratio of the amplitude of the \(i\)-th non-zero amplitude frequency point in the frequency data corresponding to the \(a\)-th data segment to the sum of the amplitudes of all non-zero amplitude frequency points in the frequency data corresponding to the \(a\)-th data segment, that is, it represents the contribution degree of this frequency point to the current signal segment.

[0029] Furthermore, the difference distance between two data segments is obtained according to the difference between the contribution degrees of the non-zero amplitude frequency points in the spectrum data corresponding to the two data segments.

[0030] Specifically, find the union of the non-zero amplitude frequency points in the spectrum data corresponding to the two data segments; use the Euclidean distance calculation formula to calculate the contribution degrees corresponding to each frequency point in the spectrum data of the two data segments in the union to obtain the difference distance between the two data segments.

[0031] The calculation model of its difference distance is: , where, represents the difference distance between the \(a\)-th and \(b\)-th data segments currently collected, 、 respectively represent the number of non-zero amplitude frequency points in the spectrum data corresponding to the \(a\)-th and \(b\)-th data segments, represents the number of frequency points in the union of the non-zero amplitude frequency points in the spectrum data corresponding to the \(a\)-th and \(b\)-th data segments, and respectively represent the contribution degrees corresponding to the \(i\)-th frequency point in the spectrum data of the \(a\)-th and \(b\)-th data segments in the union; represents the contribution degree difference of the \(i\)-th non-zero amplitude frequency point in the spectrum data of the \(a\)-th and \(b\)-th data segments. The larger this value is, the greater the difference between the two data segments. This calculation method is equivalent to calculating the Euclidean distance between two points in a multi-dimensional space using the Euclidean distance calculation formula. and represent two values under one dimension. Thus, the difference distance between every two data segments can be obtained.

[0032] The clustering analysis module is used to cluster all data segments using the difference distance to obtain clustering clusters; based on the number of data points in a clustering cluster and its distance from other clustering clusters, obtain the importance degree coefficient of this clustering cluster.

[0033] Through the above, the difference distance between any two data segments in the currently collected power grid parameter data can be obtained, and then the data segments of the currently collected power grid parameter data are clustered using density peak clustering (Density peaks clustering, DPC). This algorithm is a well-known technology and will not be elaborated here. That is, clustering is performed with one data segment as one element.

[0034] Compared with the data in the power grid, the proportion of normal parameter data in the overall scale of this data is relatively large, and it usually has general similarity. After clustering, if the distance between a certain cluster and other clusters is farther, it indicates that the data in this cluster is more different from the data in other clusters, that is, the difference between the data in this cluster and the data in other clusters is greater, and the importance of this cluster may be higher. Similarly, if the scale of a certain cluster is relatively small, it means that compared with other data, these data features are relatively rare over the entire time period, and they may play an important role in the monitoring and other analysis processes, and their importance may also be relatively high.

[0035] Obtain the importance degree coefficient of a cluster based on the number of data points in the cluster and its distance from other clusters. Specifically, calculate the average value of the difference distances between the cluster center of a cluster and the cluster centers of other clusters, which is denoted as the average distance of this cluster; compare the number of data segments in this cluster with the number of data segments in all clusters to obtain the proportion of the number of data segments in this cluster; compare the average distance of this cluster with the average distances of all clusters to obtain the distance proportion of this cluster; use the first preset value to perform a negative correlation mapping on the proportion of the number of data segments in this cluster and multiply it by the distance proportion to obtain the importance degree coefficient of this cluster.

[0036] The calculation model of the average distance is: , where represents the average value of the difference distances between the s-th cluster and other clusters, that is, the average distance of the s-th cluster; represents the difference distance between the cluster center of the s-th cluster and the cluster center of the r-th cluster, represents the average value of the sum of the difference distances between the cluster center of the s-th cluster and the cluster centers of each of the remaining clusters. The larger this value is, the farther the distance of this cluster compared with other clusters.

[0037] After obtaining the distance of any cluster compared with other clusters, combined with the number of data segments in the cluster, the importance degree coefficient of the cluster can be calculated. The specific calculation model of the importance degree coefficient is: , In the formula represents the importance degree coefficient of the s-th cluster in the currently collected data, represents the number of elements in the s-th cluster (that is, the number of data segments in the cluster); represents the number of all elements in all current clusters (that is, the number of all divided data segments); represents the average distance of the s-th cluster, and Z represents the number of clusters. is the proportion of the number of data segments in the s-th cluster, which represents the ratio of the number of elements in the s-th cluster to the number of all elements in all current clusters (i.e., segmented data). The smaller this value is, that is the larger the value, the smaller the scale of this cluster compared to other clusters, and then the higher the importance of this cluster may be; is the distance proportion of the s-th cluster, which represents the ratio of the distance of the s-th cluster compared to other clusters to the total distance of all clusters compared to other clusters. The larger this value is, the farther this cluster is compared to other clusters, and then its importance may be higher.

[0038] Thus, the importance coefficient of each cluster can be obtained.

[0039] The intra-cluster analysis module is used to obtain the relative importance coefficient of a data segment within a cluster based on the distance of the data segment from the current moment in time series and the difference distance between the data segment and the cluster center.

[0040] Through the above, the importance coefficient of any cluster is obtained. Although for the cluster center of the same cluster, it can overall reflect the basic characteristics of the data of this cluster, due to the different distances between each element (i.e., segmented data) within each cluster and the cluster center, and the different time periods during actual acquisition. For example, during monitoring, the closer some data segments are to the monitoring moment (the current moment), the greater their reference significance during monitoring or analysis, and then the retention degree of the information collected during this time period may need to be higher.

[0041] Therefore, it is necessary to further analyze each data segment within the cluster, so as to more finely evaluate its importance, achieve more accurate and flexible lossy compression, and effectively balance data redundancy reduction and key information retention.

[0042] Obtain the relative importance coefficient of a data segment within a cluster based on the distance of the data segment from the current moment in time series and the difference distance between the data segment and the cluster center.

[0043] Specifically, obtain the time length between the middle moment of a data segment within a cluster and the current moment, and use the exponential function with the natural constant as the base to perform a negative correlation mapping on the time length to obtain the time series related eigenvalue of the data segment; use the exponential function with the natural constant as the base to map the difference distance between the data segment and the cluster center of the cluster where the data segment is located to obtain the distance distribution eigenvalue; normalize and weighted sum the time series related eigenvalue and the distance distribution eigenvalue to obtain the relative importance coefficient of the data segment.

[0044] The specific calculation model of the relative importance coefficient is as follows:

[0045] In the formula represents the relative importance coefficient of the j-th element (data segment) in the s-th clustering cluster; and represent the weight coefficients, with values of 0.4 and 0.6 here, which are empirical values (specifically set according to the actual requirements for historical data. If the dependence on earlier historical data is smaller during analysis, then a larger weight can be set, and vice versa, a smaller weight can be set). represents the linear normalization function, and e represents the natural constant; represents the time distance between the middle moment of the j-th data segment in the s-th clustering cluster and the latest sampling moment (i.e., the current moment). The smaller this value is, the larger the value of the time series correlation eigenvalue . Then it indicates that the closer this data segment is to the current time, the higher its importance may be; represents the distance between the j-th data segment in the s-th clustering cluster and the center of this clustering cluster. The larger this value is, the larger the value of the distance distribution eigenvalue . Then it indicates that the farther this data segment is from the center of this clustering cluster. For this clustering cluster, the difference between this data segment and the representative data segment closest to the center of this clustering cluster may be larger, and its relative importance degree in the current clustering cluster may be higher.

[0046] The fusion analysis module is used to obtain the final importance coefficient of a data segment according to the relative importance coefficient of the data segment and the importance degree coefficient of the clustering cluster where it is located.

[0047] Through the above, the importance degree coefficient of any clustering cluster and the relative importance coefficient of each clustering cluster can be obtained. Furthermore, by fusing and analyzing the two, the final importance coefficient of each data segment can be obtained.

[0048] Obtain the final importance coefficient of a data segment according to the relative importance coefficient of the data segment and the importance degree coefficient of the clustering cluster where it is located. Specifically, compare the relative importance coefficient of a data segment with the relative importance coefficient of the data segment closest to the center of the clustering cluster within the clustering cluster where this data segment is located to obtain the corresponding importance ratio of this data segment; multiply the configuration constant term, the importance degree coefficient of the clustering cluster where this data segment is located, and the corresponding importance ratio of this data segment and normalize them to obtain the final importance coefficient of this data segment.

[0049] The calculation model of the final importance coefficient is as follows: , In the formula Denotes the final importance coefficient of the j-th data segment in the s-th clustering cluster. and respectively denote the importance degree coefficient of the s-th clustering cluster and the relative importance coefficient of the j-th data segment in the s-th clustering cluster; G represents the configuration constant term, and the value range here is , by configuring this parameter, the importance of each segmented signal can be artificially controlled, thereby indirectly controlling the information loss degree of each segmented information. Here, the default value is 1 (selected according to the actual environment). The smaller this value is, the higher the information loss degree during subsequent compression of this segment; Denotes the relative importance coefficient corresponding to the data segment closest to the clustering center point in the s-th clustering cluster. Denotes the ratio of the relative importance coefficient of the j-th data segment in the s-th clustering cluster to the relative importance coefficient corresponding to the data segment closest to the clustering center point in the s-th clustering cluster, that is, the importance ratio corresponding to this data segment. If this value is larger, it indicates that this data segment may have a relatively higher importance coefficient relative to this clustering cluster, and then this data segment may have a relatively higher importance degree relative to the initial estimate of this clustering. Conversely, it is lower, thereby more accurately evaluating the importance of this data segment.

[0050] The compression storage module is used to obtain the component signals of each data segment; process the component signals of each data segment based on the importance coefficient to obtain the updated data segment corresponding to each data segment; and compress and store the updated data segment.

[0051] Through the above, the final importance coefficients of different data segments in the current collector current data collected by the ring network cabinet can be obtained , and then each data segment is decomposed using EMD (Empirical Mode Decomposition), and different decomposed component signals are obtained, that is, each data segment can obtain its corresponding component signal. According to the order of signal decomposition by EMD, the frequencies of the decomposed component signals are from high to low in sequence. High-frequency signals often contain large fluctuations and noises, while low-frequency signals represent the main frequency of the entire signal and the overall change characteristics.

[0052] Therefore, retaining the proportion of component signals (low-frequency part) is more conducive to signal reconstruction and improvement of compression efficiency, because the low-frequency components contain the core features of the signal, can effectively reduce redundant data, and maintain the main information of the signal. While the first 1 - proportion of components (high-frequency components) usually contains more noises and tiny fluctuations, and contributes less to the main features of the signal, so they can be ignored.

[0053] Process the component signals of each data segment based on the importance coefficient to obtain the updated data segment corresponding to each data segment. Specifically, arrange the component signals of a data segment in the order in which the component signals are obtained during decomposition to obtain a component signal sequence; take component signals forward from the last component signal in the component signal sequence until the ratio of the number of component signals taken out to the number of component signals in the component signal sequence is equal to the final importance coefficient of this data segment, and stop extracting component signals. The extracted component signals are the retained component signals of this data segment. Perform inverse transformation on the retained component signals to obtain the updated data segment of this data segment.

[0054] In addition, it should be noted that empirical mode decomposition and the inverse transformation of component signals are prior arts and will not be elaborated here. In the actual application process, the ratio of the number of component signals taken out to the number of component signals in the component signal sequence may not be equal to the final importance coefficient of this data segment. At this time, it is necessary to make the ratio of the number of component signals taken out to the number of component signals in the component signal sequence slightly greater than the final importance coefficient. That is, when the ratio is slightly less than the final importance coefficient, continue to take component signals until the ratio is greater than the final importance coefficient, and then stop extracting component signals.

[0055] The updated data segment obtained by performing inverse transformation on the retained component signals can replace the original data of this data segment, and the same applies to any other data segment. For the processed data, then use compression algorithms such as ZIP to compress the updated data segments of each data segment, and then store them, so as to achieve adaptive lossy compression processing of the collected data to reduce the pressure on the data storage space. Similarly, perform the same processing on the voltage data and then compress and store it.

[0056] In summary, this application segments the grid parameter data collected by the ring main unit, then obtains the differences between different data segments, and then performs clustering based on the differences between different data segments. Then, analyze outside and inside the clustering clusters to obtain the importance degree coefficient of the clustering cluster and the relative importance coefficient of the data segments inside the clustering cluster respectively, and then obtain the final importance coefficient of each data segment. Then use EMD to decompose the data segment to obtain component signals, determine the component signals to be retained based on the final importance coefficient of the data segment, and then perform inverse transformation to obtain the updated data segment corresponding to the data segment, and then perform compression storage to reduce the space occupied by its storage and reduce the pressure on the storage of the ring main unit.

[0057] It should be noted that the above order of the embodiments of the present invention is only for description and does not represent the superiority or inferiority of the embodiments. In addition, the specific embodiments of this specification have been described. Further, 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.

[0058] 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, and the key point of each embodiment is to illustrate the differences from other embodiments.

[0059] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A ring main cabinet with an intelligent storage function for historical data, characterized in that, The ring network cabinet includes: A difference analysis module, configured to segment the power grid parameter data collected by the ring network cabinet to obtain data segments; and obtain the difference distance between every two data segments according to the spectrum data corresponding to each data segment; A clustering analysis module, configured to cluster all data segments by using the difference distance to obtain clustering clusters; and obtain the importance degree coefficient of a clustering cluster based on the number of data points in a clustering cluster and the distance between the clustering cluster and other clustering clusters; An intra-cluster analysis module, configured to obtain the relative importance coefficient of a data segment within a clustering cluster according to the distance between the data segment and the current moment in time series and the difference distance between the data segment and the clustering cluster center; A fusion analysis module, configured to obtain the final importance coefficient of a data segment according to the relative importance coefficient of the data segment and the importance degree coefficient of the clustering cluster where the data segment is located; A compression storage module, configured to obtain the component signals of each data segment; process the component signals of each data segment based on the importance coefficient to obtain the updated data segment corresponding to each data segment; and compress and store the updated data segments.

2. The loop network cabinet with the intelligent storage function of historical data according to claim 1, characterized in that, The obtaining the difference distance between every two data segments according to the spectrum data corresponding to each data segment includes: Comparing the amplitude corresponding to a frequency point with a non-zero amplitude in the frequency data corresponding to a data segment with the sum of the amplitudes corresponding to all frequency points with non-zero amplitudes in the frequency data corresponding to the data segment to obtain the contribution degree of the frequency point; and obtaining the difference distance between two data segments according to the difference between the contribution degrees of the frequency points with non-zero amplitudes in the spectrum data corresponding to the two data segments.

3. The loop network cabinet with the intelligent storage function for historical data according to claim 2, wherein, The obtaining the difference distance between two data segments according to the difference between the contribution degrees of the frequency points with non-zero amplitudes in the spectrum data corresponding to the two data segments includes: Obtaining the union of the frequency points with non-zero amplitudes in the spectrum data corresponding to the two data segments; and calculating the contribution degrees corresponding to each frequency point in the union in the spectrum data of the two data segments by using the Euclidean distance calculation formula to obtain the difference distance between the two data segments.

4. A loop network cabinet with a historical data intelligent storage function according to claim 1, characterized in that, The obtaining the importance degree coefficient of a clustering cluster based on the number of data points in a clustering cluster and the distance between the clustering cluster and other clustering clusters includes: Calculating the average value of the difference distances between the clustering center of a clustering cluster and the clustering centers of other clustering clusters, and denoting it as the average distance of the clustering cluster; comparing the number of data segments in the clustering cluster with the number of data segments in all clustering clusters to obtain the proportion of the number of data segments in the clustering cluster; comparing the average distance of the clustering cluster with the average distances of all clustering clusters to obtain the distance proportion of the clustering cluster; and multiplying the proportion of the number of data segments in the clustering cluster after negative correlation mapping by a first preset value and the distance proportion to obtain the importance degree coefficient of the clustering cluster.

5. A ring main cabinet with a historical data intelligent storage function according to claim 1, characterized in that, The obtaining the relative importance coefficient of a data segment within a clustering cluster according to the distance between the data segment and the current moment in time series and the difference distance between the data segment and the clustering cluster center includes: Obtain the time length between the middle moment of a data segment within a cluster and the current moment, and use the exponential function with the natural constant as the base to perform a negative correlation mapping on the time length to obtain the time series related eigenvalue of the data segment; use the exponential function with the natural constant as the base to map the difference distance between the data segment and the cluster center of the cluster where the data segment is located to obtain the distance distribution eigenvalue; normalize and weighted sum the time series related eigenvalue and the distance distribution eigenvalue to obtain the relative importance coefficient of the data segment.

6. The loop network cabinet with the intelligent storage function for historical data according to claim 1, wherein, The obtaining of the final importance coefficient of a data segment according to the relative importance coefficient of the data segment and the importance degree coefficient of the cluster where it is located includes: Compare the relative importance coefficient of a data segment with the relative importance coefficient of the data segment closest to the cluster center within the cluster where the data segment is located to obtain the importance ratio corresponding to the data segment; multiply the configuration constant term, the importance degree coefficient of the cluster where the data segment is located, and the importance ratio corresponding to the data segment and normalize to obtain the final importance coefficient of the data segment.

7. The loop network cabinet with the intelligent storage function of historical data according to claim 1, characterized in that, The obtaining of the component signal of each data segment includes: Decompose each data segment using EMD to respectively obtain the component signals corresponding to each data segment.

8. A loop network cabinet with a historical data intelligent storage function according to claim 1, characterized in that, The processing of the component signal of each data segment based on the importance coefficient to obtain the updated data segment corresponding to each data segment includes: Arrange the component signals of a data segment in the order of obtaining the component signals during decomposition to obtain a component signal sequence; take the component signals from the last component signal in the component signal sequence forward until the ratio of the number of component signals taken out to the number of component signals in the component signal sequence is equal to the final importance coefficient of the data segment, stop extracting the component signals, the extracted component signals are the retained component signals of the data segment, and perform an inverse transform on the retained component signals to obtain the updated data segment of the data segment.

Images