Intelligent fusion terminal with efficient data acquisition function

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent fusion terminal with an efficient data acquisition function, which comprises a processor and a memory, and is characterized in that the processor executes a computer program of the memory to realize the following steps: acquiring an electric power data sequence and dividing the electric power data sequence into at least two subsequences; obtaining a template sub-sequence of each sub-sequence, a matching degree between each sub-sequence and the template sub-sequence thereof, and a difference sequence corresponding to each sub-sequence of which the matching degree is not 1; according to the matching degree between each sub-sequence and the template sub-sequence thereof and the data fluctuation degree of each difference sequence, obtaining the attention degree of each difference sequence; and according to the matching degree between each sub-sequence and the template sub-sequence thereof and the attention degree of each difference sequence, obtaining a compression matrix and an adaptive singular value retention proportionality coefficient, and performing data compression on the electric power data sequence by using a singular value decomposition algorithm to complete efficient acquisition of the electric power data in the current compression period.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to an intelligent fusion terminal with efficient data collection function. BACKGROUND

[0002] With the rapid development of the Internet of Things, industrial Internet and other fields, a large number of sensors are often deployed in the application environment of power equipment for real-time monitoring of current, voltage and other data, so that the power controller and collection terminal can realize the functions of automatic monitoring and control of the power system to a certain extent. However, the traditional data collection mode mostly adopts the mode of continuous collection and uploading, which may cause a large amount of redundant and invalid data to occupy network bandwidth and storage space, thereby increasing the processing pressure of the backend and making it difficult to meet the requirements of real-time and high efficiency.

[0003] In order to solve the above problems, an intelligent fusion terminal with efficient data collection function is needed, which can complete the redundant filtering of collected data on the terminal side and realize efficient collection and other functions, thereby reducing storage and transmission overhead, ensuring the accuracy of monitoring and analysis, and improving the data utilization efficiency and operation intelligence level of the power system.

[0004] The existing intelligent fusion terminal usually uses singular value decomposition (SVD) algorithm to compress the collected data to a certain extent to solve the problem of data redundancy. However, if the original data characteristics are not fully analyzed and the matrix construction is not optimized when using the singular value decomposition algorithm to compress data, the singular value decomposition algorithm may not be able to adaptively retain key information according to the characteristics of the data, resulting in loss of key information or retention of redundant data, and the compression rate and information retention degree are difficult to effectively balance.

[0005] Therefore, how to fully analyze the characteristics of the original data, obtain a compression matrix and an adaptive singular value retention ratio that are more suitable for singular value decomposition algorithm compression has become a problem to be solved. SUMMARY

[0006] Therefore, the embodiments of the present application provide an intelligent fusion terminal with efficient data collection function to solve the problem of how to fully analyze the characteristics of the original data and obtain a compression matrix and an adaptive singular value retention ratio that are more suitable for singular value decomposition algorithm compression.

[0007] An intelligent fusion terminal with efficient data collection function is provided in the embodiments of the present application, which includes a memory, a processor and a computer program stored in the memory and running on the processor. The processor implements the following steps when executing the computer program: Obtaining power data of any one of the power equipment at each time, obtaining a power data sequence of the current compression period, and dividing the power data sequence into at least two subsequences according to a preset time period; Obtaining a power data template sequence, sliding matching each subsequence with the power data template sequence in the power data template sequence, obtaining a template subsequence of each subsequence and a matching degree between each subsequence and the template subsequence thereof, and if there is any subsequence and the template subsequence thereof with a matching degree not equal to 1, then obtaining a difference sequence corresponding to the any subsequence by subtracting the template subsequence from the any subsequence; According to the matching degree between each subsequence and the template subsequence thereof and the data fluctuation degree of the difference sequence corresponding to each subsequence, obtaining the attention degree of each difference sequence; According to the matching degree between each subsequence and the template subsequence thereof and the attention degree of each difference sequence, obtaining a compression matrix and an adaptive singular value retention proportion coefficient when the power data sequence is compressed by using a singular value decomposition algorithm, and compressing the power data sequence based on the compression matrix and the adaptive singular value retention proportion coefficient, thereby completing efficient collection of power data in the current compression period.

[0008] Preferably, the power data template sequence is obtained by: Obtaining a reference signal of any one of the power equipment under a standard power grid fundamental frequency, obtaining reference power data at each time in a preset time period, and composing a power data template sequence.

[0009] Preferably, the sliding matching of each subsequence with the power data template sequence in the power data template sequence, the obtaining of the template subsequence of each subsequence and the matching degree between each subsequence and the template subsequence thereof, comprises: For any subsequence, a window equal in length to the any subsequence is established in the power data template sequence with the first reference power data in the power data template sequence as the starting data, and the first power data of the any subsequence is aligned with each reference power data in the window, thereby obtaining at least two matching sequences of the any subsequence; For any matching sequence, the matching degree between the any subsequence and the any matching sequence is obtained according to the data difference between the any subsequence and the any matching sequence; The matching degree between the any subsequence and each matching sequence is obtained, the matching sequence corresponding to the maximum matching degree is taken as the template subsequence of the any subsequence, and the maximum matching degree is taken as the matching degree between the any subsequence and the template subsequence thereof.

[0010] Preferably, the obtaining the matching degree between the any sub-sequence and the any matching sequence according to the data difference between the any sub-sequence and the any matching sequence comprises: obtaining the mean square error between the any sub-sequence and the any matching sequence, and substituting the inverse of the mean square error into an exponential function with a natural constant as the base to obtain the matching degree between the any sub-sequence and the any matching sequence.

[0011] Preferably, the obtaining the attention degree of each difference sequence according to the matching degree between each sub-sequence and its template sub-sequence and the data fluctuation degree of the difference sequence corresponding to each sub-sequence comprises: for any difference sequence, obtaining the change stability of the any difference sequence according to the data fluctuation feature of the any difference sequence; obtaining at least two reference difference sequences of the any difference sequence, for any reference difference sequence, taking the matching degree between the sub-sequence corresponding to the any reference difference sequence and its template sub-sequence as a first matching degree, taking the matching degree between the sub-sequence corresponding to the any difference sequence and its template sub-sequence as a second matching degree, obtaining the ratio of the first matching degree to the second matching degree to obtain a matching degree ratio; obtaining the ratio of the change stability of the any reference difference sequence to the change stability of the any difference sequence to obtain a change stability ratio, and obtaining the product between the matching degree ratio and the change stability ratio to obtain the attention value of the any difference sequence compared with the any reference difference sequence; obtaining the time interval between the any difference sequence and the any reference difference sequence, substituting the inverse of the time interval into an exponential function with a natural constant as the base to obtain a reference degree of the any reference difference sequence; obtaining the attention value of the any difference sequence compared with each reference difference sequence, obtaining the reference degree of each reference difference sequence, taking the reference degree of each reference difference sequence as a weight coefficient of the attention value of the any difference sequence compared with each reference difference sequence, obtaining the weighted average value of the attention value of the any difference sequence compared with each reference difference sequence to obtain the attention degree of the any difference sequence.

[0012] Preferably, the obtaining the change stability of the any difference sequence according to the data fluctuation feature of the any difference sequence comprises: For any difference data except the first difference data and the last difference data in any difference sequence, obtaining the mean between the left adjacent difference data and the right adjacent difference data of the any difference data, calculating the absolute value of the difference between the any difference data and the mean to obtain a change value of the any difference data, obtaining the ratio of the change value of the any difference data to the absolute value of the mean to obtain a degree of difference of the any difference data; The degree of difference of each difference data is obtained, and the corresponding mean value of the degree of difference is obtained. The opposite number of the mean value of the degree of difference is substituted into an exponential function with a natural constant as the base to obtain the change stability of any difference sequence.

[0013] Preferably, obtaining at least two reference difference sequences of any one of the difference sequences comprises: The difference sequences corresponding to each subsequence are combined into a sequence to be analyzed. In the sequence to be analyzed, if the number of difference sequences before any difference sequence and the number of difference sequences after any difference sequence are both greater than or equal to a preset number, the preset number of difference sequences before any difference sequence and the preset number of difference sequences after any difference sequence are used as reference difference sequences for the any difference sequence; If the number of difference sequences before any difference sequence is less than a preset number and greater than 0, all difference sequences before any difference sequence and a preset number of difference sequences after any difference sequence are used as reference difference sequences for any difference sequence; If the number of difference sequences after any difference sequence is less than a preset number and greater than 0, the preset number of difference sequences before any difference sequence and all difference sequences after any difference sequence are used as reference difference sequences for any difference sequence; If the number of difference sequences before any difference sequence or the number of difference sequences after any difference sequence is 0, the preset number of difference sequences after any difference sequence or the preset number of difference sequences before any difference sequence are used as reference difference sequences for any difference sequence.

[0014] Preferably, the method of obtaining a compression matrix and an adaptive singular value retention ratio coefficient when compressing the power data sequence using a singular value decomposition algorithm based on the matching degree between each subsequence and its template subsequence and the attention degree of each difference sequence includes: The attention degree of each difference sequence is normalized to obtain a normalized attention degree corresponding to each difference sequence, the matching degree between the subsequence corresponding to each difference sequence and the template subsequence thereof is normalized to obtain a normalized matching degree corresponding to each difference sequence, the position sequence number of the first reference power data in the template subsequence of the subsequence corresponding to each difference sequence is obtained in the power data template sequence, the position sequence number of the first reference power data in the template subsequence of the subsequence corresponding to each difference sequence is normalized to obtain a normalized position sequence number corresponding to each difference sequence, and a three-dimensional space coordinate system is established with the normalized attention degree corresponding to each difference sequence as the x-axis, the normalized matching degree corresponding to each difference sequence as the y-axis, and the normalized position sequence number corresponding to each difference sequence as the z-axis. In the three-dimensional space coordinate system, a data point corresponding to each difference sequence is obtained, and the data points in the three-dimensional space coordinate system are clustered to obtain at least one cluster. For any cluster, the difference data in the difference sequence corresponding to the data points in the any cluster form a compression matrix, the number of rows of the compression matrix is the number of data points in the any cluster, and the number of columns of the compression matrix is the number of difference data in each difference sequence. According to the positions of the data points in the any cluster and the attention degrees of the difference sequences corresponding to the data points in the any cluster, an adaptive singular value retention proportion coefficient corresponding to the compression matrix is obtained.

[0015] Preferably, the adaptive singular value retention proportion coefficient corresponding to the compression matrix is obtained according to the positions of the data points in the any cluster and the attention degrees of the difference sequences corresponding to the data points in the any cluster, and includes: For any data point in the any cluster, the data points in the any cluster except the any data point are regarded as reference data points, the distance mean value between the any data point and each reference data point is obtained to obtain the distance characteristic value of the any data point. The distance characteristic value of each reference data point is obtained, the average value of the absolute value of the difference between the distance characteristic value of the any data point and the distance characteristic value of each reference data point is calculated to obtain the distance characteristic difference mean value, the average value of the distance characteristic difference mean value of each data point in the any cluster is obtained to obtain the distance difference average value of the any cluster. The attention degree mean value of the data points in the any cluster is obtained, the distance characteristic value mean value of the data points in the any cluster is obtained, and the product between the attention degree mean value, the distance characteristic value mean value and the distance difference average value of the any cluster is normalized to obtain the adaptive singular value retention proportion coefficient corresponding to the compression matrix.

[0016] Preferably, after obtaining the template sub-sequence of each sub-sequence and the matching degree between each sub-sequence and its template sub-sequence, the method further comprises: For any sub-sequence, if the matching degree between the any sub-sequence and its template sub-sequence is 1, then in the power data template sequence, the position sequence number of the first reference power data in the template sub-sequence of the any sub-sequence is obtained, and the data compression of the any sub-sequence is completed.

[0017] Compared with the prior art, the embodiment of the present application has the following beneficial effects: The present application obtains the power data of any kind of data in the power equipment at each time, obtains the power data sequence of the current compression period, divides the power data sequence into at least two sub-sequences according to a preset time period, obtains a power data template sequence, respectively matches each sub-sequence with the power data template sequence in the power data template sequence, obtains the template sub-sequence of each sub-sequence and the matching degree between each sub-sequence and its template sub-sequence, if there is any sub-sequence and its template sub-sequence between which the matching degree is not 1, then the any sub-sequence and its template sub-sequence are subtracted to obtain the difference sequence corresponding to the any sub-sequence, according to the matching degree between each sub-sequence and its template sub-sequence and the data fluctuation degree of each difference sequence, the attention degree of each difference sequence is obtained, according to the matching degree between each sub-sequence and its template sub-sequence and the attention degree of each difference sequence, the compression matrix and the adaptive singular value retention proportion coefficient when the power data sequence is compressed by using the singular value decomposition algorithm are obtained, the power data sequence is compressed based on the compression matrix and the adaptive singular value retention proportion coefficient, and the efficient collection of the power data in the current compression period is completed. If there is any sub-sequence and its template sub-sequence between which the matching degree is not 1, then the any sub-sequence and its template sub-sequence are subtracted to obtain the difference sequence corresponding to the any sub-sequence, and the difference sequence in the power data sequence that needs to be compressed by using the singular value decomposition algorithm is screened out, according to the matching degree between each sub-sequence and its template sub-sequence and the attention degree of each difference sequence, the compression matrix and the adaptive singular value retention proportion coefficient when the power data sequence is compressed by using the singular value decomposition algorithm are obtained, the power data sequence is compressed based on the compression matrix and the adaptive singular value retention proportion coefficient, the efficient compression of the power data sequence is realized, the burden of the intelligent fusion terminal when collecting data for subsequent transmission and storage is reduced, and solid data support is provided for the high reliability and high precision of the intelligent fusion terminal in the power grid state monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative effort based on these drawings.

[0019] Figure 1 is a flow chart of a data efficient collection method applied to an intelligent fusion terminal provided by the first embodiment of the present application. DETAILED DESCRIPTION

[0020] The embodiments of the present disclosure will be described in detail below, and examples of the embodiments are shown in the drawings. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present disclosure, and should not be understood as a limitation of the present disclosure.

[0021] It should be noted that the terms "first", "second" and the like in the specification of the present disclosure and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or a chronological sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present disclosure described herein can be implemented in an order other than that illustrated or described herein. The implementation described in the following exemplary embodiments does not represent all the implementations consistent with the present disclosure. Rather, they are only examples of devices and methods consistent with some aspects of the present disclosure.

[0022] In order to illustrate the technical solutions of the present application, the following will be described by specific embodiments.

[0023] The embodiments of the present application provide an intelligent fusion terminal with data efficient collection function, comprising a processor and a memory, the processor executes the computer program of the memory to realize a data efficient collection method applied to an intelligent fusion terminal, as shown in Figure 1 The method comprises the following steps: Step S101, obtaining the power data of any kind of data in the power equipment at each time, obtaining the power data sequence of the current compression period, and dividing the power data sequence into at least two subsequences according to the preset time period.

[0024] The power grid intelligent fusion terminal mainly collects key power data such as voltage and current in real time, and is used for dynamic monitoring and regulation of power quality, load distribution, and power consumption anomalies. In actual application, in order to ensure the continuity and real-time performance of monitoring, the power grid intelligent fusion terminal usually records a large amount of data at a high sampling frequency, however, there is often a lot of redundancy in these time series data, especially the voltage and current signals have strong periodicity and regularity, which makes part of the information redundant in time domain, causing large storage pressure and high transmission overhead, and then affecting the terminal processing efficiency and system response capability. Therefore, an intelligent fusion terminal with efficient data acquisition function is needed, which can meet the efficient compression of collected data while also realizing the retention of effective information, which is particularly important for subsequent power grid monitoring and power system control.

[0025] Since the analysis method of each power data is the same, in this embodiment, the power data of any kind of data in the power equipment at each time is obtained according to the acquisition frequency of 4KHz. 4KHz is the common acquisition frequency of power data, which can meet the accurate recording demand of power frequency signal, which is not limited here and can be set according to the specific implementation scene. The power data sequence of the current compression period is obtained, and then the power data sequence is divided into at least two subsequences according to the preset time period for subsequent analysis. Since the power grid voltage, current and other signals may generally fluctuate and have short-time disturbance on the order of seconds, too long sampling interval may easily lose instantaneous abnormal information. Therefore, the preset time period is set to one second in this embodiment. In order to ensure the integrity of the overall trend of current data while avoiding too large data volume in storage and transmission, the current compression period is set to one minute in this embodiment, which is not limited here and can be set according to the specific implementation scene. In actual application, the power grid operation characteristics, monitoring targets and subsequent analysis requirements need to be combined to determine: if fast power quality disturbance needs to be captured, the preset time period and compression period should be shortened; if the long-term load trend is mainly concerned, the preset time period and compression period can be appropriately lengthened.

[0026] The existing intelligent fusion terminal usually uses singular value decomposition (SVD) algorithm to compress the collected data to a certain extent to solve the problem of data redundancy. However, if the original data characteristics are not fully analyzed and the matrix construction is not optimized when using singular value decomposition algorithm to compress data, the singular value decomposition algorithm may not be able to retain key information according to the characteristics of the data, resulting in loss of key information or retention of redundant data, and the compression rate and information retention degree are difficult to effectively balance.

[0027] Therefore, the embodiment first screens out the difference sequence in the power data sequence which needs to be compressed by using the singular value decomposition algorithm, then obtains the compression matrix and the adaptive singular value reservation proportion coefficient, and uses the singular value decomposition algorithm to perform data compression on the power data sequence, thereby completing efficient collection of the power data in the current compression period.

[0028] In step S102, the power data template sequence is obtained. In the power data template sequence, each sub-sequence is respectively matched with the power data template sequence in a sliding manner to obtain a template sub-sequence of each sub-sequence and a matching degree between each sub-sequence and its template sub-sequence. If the matching degree between any sub-sequence and its template sub-sequence is not 1, the difference sequence corresponding to the any sub-sequence is obtained by subtracting the template sub-sequence from the any sub-sequence.

[0029] Since the voltage and current data in the power grid change at a frequency of a base frequency, for example, a 50 Hz or 60 Hz sine wave signal, under the access of electrical equipment, the data of different sub-sequences may fluctuate and change, which may cause the original power data close to the sine wave to change. These changed power data reflect the power consumption conditions and other characteristics of the electrical equipment, and therefore need to be screened out.

[0030] Therefore, the reference signal of any kind of data in the power equipment under the standard power grid base frequency is obtained to obtain the reference power data at each time in a preset time period to form the power data template sequence. Since the power data has strong periodicity, in order to enable each sub-sequence to obtain a template sub-sequence in the power data template sequence, the preset time period is set to 2 seconds (i.e., 2 times of the preset time period) in the embodiment, which is not limited here and can be set according to a specific implementation scenario. In the power data template sequence, each sub-sequence is respectively matched with the power data template sequence in a sliding manner to obtain a template sub-sequence of each sub-sequence and a matching degree between each sub-sequence and its template sub-sequence. Whether the power data in each sub-sequence changes is determined according to the matching degree between each sub-sequence and its template sub-sequence.

[0031] The method for obtaining the template sub-sequence of each sub-sequence and the matching degree between each sub-sequence and its template sub-sequence is as follows: (1) Each sub-sequence is respectively matched with the power data template sequence in a sliding manner to obtain a matching sequence of each sub-sequence.

[0032] Specifically, for any sub-sequence, a window equal to the length of the any sub-sequence is established in the power data template sequence, taking the first reference power data in the power data template sequence as the starting data, and the first power data of the any sub-sequence is aligned with each reference power data in the window to obtain at least two matching sequences of the any sub-sequence.

[0033] For example, in the power data template sequence, a window with a time length of 1 second is established, taking the first reference power data as the starting data, for the athsub-sequence, the first power data of the athsub-sequence is aligned with the first reference power data in the window to obtain the first matching sequence of the athsub-sequence, the first power data of the athsub-sequence is aligned with the second reference power data in the window to obtain the second matching sequence of the athsub-sequence, and so on, the first power data of the athsub-sequence is aligned with the last reference power data in the window to obtain the last matching sequence of the athsub-sequence.

[0034] (2) For any matching sequence, the matching degree between the any sub-sequence and the any matching sequence is obtained according to the data difference between the any sub-sequence and the any matching sequence.

[0035] Specifically, the mean square error between the any sub-sequence and the any matching sequence is obtained, and the reciprocal of the mean square error is substituted into the exponential function with the natural constant as the base to obtain the matching degree between the any sub-sequence and the any matching sequence.

[0036] In an embodiment, taking the athsub-sequence and its cthmatching sequence as an example, the calculation formula of the matching degree between the athsub-sequence and its cthmatching sequence is:

[0037] Wherein, is the matching degree between the athsub-sequence and its cthmatching sequence; e is the natural constant; is the mean square error between the athsub-sequence and its cthmatching sequence.

[0038] It should be noted that, The greater the matching degree between the athsub-sequence and its cthmatching sequence, the higher the coincidence degree of the data between the athsub-sequence and its cthmatching sequence, The greater the matching degree.

[0039] (3) The matching degree between any sub-sequence and its template sub-sequence is obtained.

[0040] Specifically, a matching degree between the any sub-sequence and each matching sequence is obtained, a matching sequence corresponding to the maximum matching degree is taken as a template sub-sequence of the any sub-sequence, and the maximum matching degree is taken as the matching degree between the any sub-sequence and the template sub-sequence.

[0041] If the matching degree between the any sub-sequence and the template sub-sequence is 1, it is indicated that the power data in the any sub-sequence is the power data of the normal operation of the power equipment under the ideal condition (i.e., the standard power grid fundamental frequency), which is the same as the reference power data in the power data template sequence, and can be directly simulated by the data in the power data template sequence. Therefore, in the power data template sequence, a position sequence number of a first reference power data in the template sub-sequence of the any sub-sequence is obtained and stored, and the data compression of the any sub-sequence is completed.

[0042] In the process of grid application, more attention is often paid to the waveform and relatively abnormal data when collecting and analyzing voltage and current data. Because in the monitoring scene of the intelligent fusion terminal of the power grid, the abnormal waveform (such as sudden rise, sudden drop, harmonic distortion, etc.) of voltage and current is often directly related to power grid failure, equipment hidden danger or power quality problem, so the relatively abnormal power data has more analysis value than the smooth power data. For example, in the application of distribution network fault location, new energy grid-connected harmonic monitoring or industrial sensitive load protection, the intelligent fusion terminal will screen abnormal waveforms (such as threshold detection, fault recording) in real time through edge computing, and preferentially store and upload such data to support rapid diagnosis and decision-making. Greater information retention of the “abnormal priority” data can ensure subsequent monitoring efficiency, thereby meeting the core needs of the power system for fault early warning and active operation and maintenance.

[0043] Therefore, if the matching degree between the any sub-sequence and the template sub-sequence is not 1, it is indicated that the power data in the any sub-sequence changes. Therefore, the difference between the any sub-sequence and the template sub-sequence is obtained, and a difference sequence corresponding to the any sub-sequence is obtained, that is, a difference sequence that needs to be compressed by using the singular value decomposition algorithm.

[0044] At this point, all difference sequences that need to be compressed by using the singular value decomposition algorithm are obtained.

[0045] In step S103, the attention degree of each difference sequence is obtained according to the matching degree between each sub-sequence and the template sub-sequence and the data fluctuation degree of each difference sequence.

[0046] Since the fluctuation degree of the difference data in each difference sequence is different, if the change stability of the difference data in the difference sequence is better, the abnormality may be relatively low, and the required attention degree in data compression is smaller. Therefore, the attention degree of each difference sequence can be obtained according to the matching degree between each subsequence and the template subsequence and the data fluctuation degree of the difference sequence corresponding to each subsequence. Then, for any difference sequence, the method for obtaining the attention degree of any difference sequence is as follows: (1) According to the data fluctuation characteristics of the any difference sequence, the change stability of the any difference sequence is obtained.

[0047] Specifically, for any difference data in the any difference sequence except the first difference data and the last difference data, the mean value between the left adjacent difference data and the right adjacent difference data of the any difference data is obtained, the absolute value of the difference between the any difference data and the mean value is calculated to obtain the change value of the any difference data, and the ratio of the absolute value of the change value of the any difference data to the mean value is obtained to obtain the difference degree of the any difference data. The difference degree of each difference data is obtained, the mean value of the difference degree is obtained, the inverse of the mean value is substituted into the exponential function with the natural constant as the base to obtain the change stability of the any difference sequence.

[0048] In an embodiment, taking the s-th difference sequence as an example, the calculation formula of the change stability of the s-th difference sequence is as follows:

[0049] Wherein, is the change stability of the s-th difference sequence; is the i-th difference data in the s-th difference sequence; is the i-1-th difference data in the s-th difference sequence, that is, the left adjacent difference data of the i-th difference data; is the i+1-th difference data in the s-th difference sequence, that is, the right adjacent difference data of the i-th difference data; m is the number of difference data in the s-th difference sequence; is the absolute value symbol; is the exponential function with the natural constant as the base, which is used for inverse proportional normalization.

[0050] It should be noted that, is the difference degree of the i-th difference data in the difference sequence, the smaller, the smaller the difference between the i-th difference data and its left and right adjacent data in the difference sequence, that is, the more stable the change of the i-th difference data, the greater.

[0051] (2) Obtain at least two reference difference sequences of the any difference sequence.

[0052] Specifically, the difference sequence corresponding to each sub-sequence forms a to-be-analyzed sequence. In the to-be-analyzed sequence, if the number of difference sequences before the any difference sequence and the number of difference sequences after the any difference sequence are both greater than or equal to a preset number, the preset number of difference sequences before the any difference sequence and the preset number of difference sequences after the any difference sequence are reference difference sequences of the any difference sequence. If the number of difference sequences before the any difference sequence is less than the preset number and greater than 0, all difference sequences before the any difference sequence and the preset number of difference sequences after the any difference sequence are reference difference sequences of the any difference sequence. If the number of difference sequences after the any difference sequence is less than the preset number and greater than 0, the preset number of difference sequences before the any difference sequence and all difference sequences after the any difference sequence are reference difference sequences of the any difference sequence. If the number of difference sequences before the any difference sequence or the number of difference sequences after the any difference sequence is 0, the preset number of difference sequences after the any difference sequence or the preset number of difference sequences before the any difference sequence are reference difference sequences of the any difference sequence. Since power data such as voltage and current usually change continuously rather than suddenly, based on the time correlation and smoothness of power grid signals, the preset number in this embodiment is set to 4. If higher sensitivity is required, the preset number can be reduced. If the overall trend is more concerned, the preset number can be appropriately increased. Here, no limitation is made, and the preset number can be set according to the specific implementation scenario.

[0053] (3) Obtain the attention degree of the any difference sequence.

[0054] Specifically, for any reference difference sequence, the matching degree between the sub-sequence corresponding to the any reference difference sequence and its template sub-sequence is recorded as a first matching degree, and the matching degree between the sub-sequence corresponding to the any difference sequence and its template sub-sequence is recorded as a second matching degree. The ratio of the first matching degree to the second matching degree is obtained to obtain a matching degree ratio. The ratio of the change stability of the any reference difference sequence to the change stability of the any difference sequence is obtained to obtain a change stability ratio. The product between the matching degree ratio and the change stability ratio is obtained to obtain the attention value of the any difference sequence compared with the any reference difference sequence. Obtaining a time interval between any one of the difference sequences and any one of the reference difference sequences, substituting the inverse of the time interval into an exponential function with a natural constant as a base, to obtain a reference degree of the any one of the reference difference sequences; Obtain the attention value of any difference sequence compared to each reference difference sequence, obtain the reference degree of each reference difference sequence, use the reference degree of each reference difference sequence as the weight coefficient of the attention value of any difference sequence compared to each reference difference sequence, obtain the weighted average of the attention values ​​of any difference sequence compared to each reference difference sequence, and obtain the attention degree of any difference sequence.

[0055] In one embodiment, taking the sth difference sequence as an example, the calculation formula for the attention degree of the sth difference sequence is:

[0056] in, is the attention of the s-th difference sequence; is the sth difference sequence The matching degree between the subsequence corresponding to the reference difference sequence and its template subsequence, i.e., the first matching degree; is the matching degree between the subsequence corresponding to the sth difference sequence and its template subsequence, i.e., the second matching degree; is the sth difference sequence The stability of the reference difference sequence; is the change stability of the sth difference sequence; is the difference sequence between the sth and The time interval between reference difference sequences; e is a natural constant; k is the number of reference difference sequences.

[0057] It should be noted that The sth difference sequence compared to the The attention value of the reference difference sequence, The larger the value, the greater the difference between the sth difference sequence and its The smaller the matching degree corresponding to the sth reference difference sequence, that is, the greater the difference between the subsequence corresponding to the sth difference sequence and its template subsequence, the greater the possibility of abnormal information, and the more attention should be paid to the sth difference sequence. The bigger it is, the the bigger it is; The larger the value, the greater the difference between the sth difference sequence and its The smaller the change stability corresponding to the sth reference difference sequence, that is, the sth difference sequence is smaller than its sth The worse the stability of the sth reference difference sequence, the more attention should be paid to the sth difference sequence. The bigger it is, the the bigger it is; The smaller the value, the more significant the difference between the sth sequence and its The closer the time between the reference difference sequences, the The higher the credibility.

[0058] At this point, the attention level of each difference sequence is obtained.

[0059] Step S104: According to the matching degree between each subsequence and its template subsequence and the attention degree of each difference sequence, a compression matrix and an adaptive singular value retention ratio coefficient are obtained when the power data sequence is compressed using a singular value decomposition algorithm; based on the compression matrix and the adaptive singular value retention ratio coefficient, the power data sequence is compressed to complete the efficient collection of power data in the current compression cycle.

[0060] For the difference sequences obtained in the current compression cycle, if there are multiple difference sequences with similar attention levels and similarity levels of data in the difference sequences, the compression efficiency will be greatly improved when data is compressed on these difference sequences.

[0061] Therefore, the attention of each difference sequence is normalized to obtain the normalized attention corresponding to each difference sequence, the matching degree between the subsequence corresponding to each difference sequence and its template subsequence is normalized to obtain the normalized matching degree corresponding to each difference sequence, in the power data template sequence, the position number of the first reference power data in the template subsequence of the subsequence corresponding to each difference sequence is obtained, the position number of the first reference power data in the template subsequence of the subsequence corresponding to each difference sequence is normalized to obtain the normalized position number corresponding to each difference sequence, and the normalized attention corresponding to each difference sequence is used as the x-axis, the normalized matching degree corresponding to each difference sequence is used as the y-axis, and the normalized position number corresponding to each difference sequence is used as the z-axis to establish a three-dimensional space coordinate system. Different data are linearly normalized to avoid the influence of different dimensional dimensions of different difference sequences during clustering.

[0062] In the three-dimensional spatial coordinate system, data points corresponding to each difference sequence are obtained, and HDBSCAN clustering is performed on the data points in the three-dimensional spatial coordinate system to obtain at least one cluster. Data points with similar characteristics are grouped into the same cluster, that is, difference sequences with similar characteristics are grouped into the same cluster, thereby improving data compression efficiency. The HDBSCAN clustering algorithm is prior art and will not be further described here.

[0063] Further, for any one cluster, difference data in a difference sequence corresponding to a data point in the any one cluster is composed into a compression matrix, a number of rows of the compression matrix is a number of data points in the any one cluster, and a number of columns of the compression matrix is a number of difference data in each difference sequence, and then an adaptive singular value reserved proportion coefficient corresponding to the compression matrix is obtained according to a position of the data point in the any one cluster and attention degrees of the difference sequence corresponding to the data point in the any one cluster.

[0064] Wherein, the method for obtaining the adaptive singular value reserved proportion coefficient corresponding to the compression matrix according to the position of the data point in the any one cluster and the attention degrees of the difference sequence corresponding to the data point in the any one cluster is as follows: For any one data point in the any one cluster, data points in the any one cluster except the any one data point are regarded as reference data points, a distance mean value between the any one data point and each reference data point is obtained, and a distance feature value of any one data point is obtained. The distance feature value of each reference data point is obtained, an average value of absolute values of differences between the distance feature value of the any one data point and the distance feature value of each reference data point is calculated, a distance feature difference mean value is obtained, an average value of the distance feature difference mean values of each data point in the any one cluster is obtained, and a distance difference average value of the any one cluster is obtained. An attention degree mean value of the data points in the any one cluster is obtained, a distance feature value mean value of the data points in the any one cluster is obtained, and a product between the attention degree mean value, the distance feature value mean value and the distance difference average value of the any one cluster is normalized to obtain the adaptive singular value reserved proportion coefficient corresponding to the compression matrix.

[0065] In an embodiment, taking the jth cluster as an example, the calculation formula of the adaptive singular value reserved proportion coefficient corresponding to the compression matrix of the jth cluster is as follows:

[0066] Wherein, is the adaptive singular value reserved proportion coefficient corresponding to the compression matrix of the jth cluster; is the attention degree mean value of the data points in the jth cluster; is the distance feature value mean value of the data points in the jth cluster; is the distance feature value of the gth data point in the jth cluster; is the distance feature value of the wth reference data point of the jth data point in the jth cluster; is the number of data points in the jth cluster; is an absolute value symbol; is a normalization function.

[0067] It should be noted that, The greater, the greater the attention of the difference sequence corresponding to the data points in the jth cluster, the higher the adaptive singular value retention ratio coefficient corresponding to the compression matrix should be, The greater, The greater, the greater the distance between data points in the jth cluster, the higher the adaptive singular value retention ratio coefficient is needed to ensure the degree of information restoration in the difference sequence corresponding to these data points, The greater, The distance difference average of the jth cluster, The greater, the more uneven the distribution of data points in the jth cluster, the higher the adaptive singular value retention ratio coefficient is needed to consider the information retention degree of all data points in the jth cluster, The greater.

[0068] After obtaining the adaptive singular value retention ratio coefficient corresponding to the compression matrix of the jth cluster, As a retention ratio coefficient in the singular value decomposition algorithm, the singular value decomposition algorithm is used to compress the data of the compression matrix of the jth cluster, and the singular values whose proportion is are selected, and at the same time of compression, the position sequence number of the first reference power data in the template subsequence corresponding to each difference sequence in the jth cluster is stored.

[0069] According to the compression method of the compression matrix of the jth cluster described above, the compression matrices of all clusters obtained in the current compression period are compressed, and the efficient acquisition and decompression of power data in the current compression period is completed. When using the singular value decomposition algorithm to restore the original matrix to the singular values retained, then according to the position sequence and difference sequence retained, combined with the power data template sequence, the original data is obtained in reverse. Among them, the data compression and decompression using the singular value decomposition algorithm belong to the prior art, which will not be described here.

[0070] In summary, the embodiment of the present application obtains power data of any kind of data in the power equipment at each moment, obtains the power data sequence of the current compression period, divides the power data sequence into at least two subsequences according to the preset time period, obtains the power data template sequence, respectively matches each subsequence with the power data template sequence in the power data template sequence, obtains the template subsequence of each subsequence and the matching degree between each subsequence and its template subsequence, if there is any subsequence and its template subsequence between the matching degree is not 1, then the difference between the any subsequence and its template subsequence is obtained, and the difference sequence corresponding to the any subsequence is obtained. According to the matching degree between each subsequence and its template subsequence and the data fluctuation degree of each difference sequence, the attention degree of each difference sequence is obtained. According to the matching degree between each subsequence and its template subsequence and the attention degree of each difference sequence, the compression matrix and the adaptive singular value retention ratio coefficient when the singular value decomposition algorithm is used to compress the power data sequence are obtained, the power data sequence is compressed based on the compression matrix and the adaptive singular value retention ratio coefficient, and the efficient collection of the power data in the current compression period is completed. If there is any subsequence and its template subsequence between the matching degree is not 1, then the difference between the any subsequence and its template subsequence is obtained, and the difference sequence corresponding to the any subsequence is obtained, and the difference sequence that needs to be compressed by the singular value decomposition algorithm in the power data sequence is screened out. According to the matching degree between each subsequence and its template subsequence and the attention degree of each difference sequence, the compression matrix and the adaptive singular value retention ratio coefficient when the singular value decomposition algorithm is used to compress the power data sequence are obtained, the power data sequence is compressed based on the compression matrix and the adaptive singular value retention ratio coefficient, the efficient compression of the power data sequence is realized, the burden of the intelligent fusion terminal when collecting data for subsequent transmission and storage is reduced, and the high reliability and high precision of the intelligent fusion terminal in the power grid state monitoring are provided with solid data support.

[0071] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. An intelligent fusion terminal with efficient data collection function, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the following steps are implemented: Obtaining power data of any type of data in the power equipment at each moment to obtain a power data sequence of a current compression period, and dividing the power data sequence into at least two subsequences according to a preset time period; Obtaining a power data template sequence, performing sliding matching on each subsequence in the power data template sequence with the power data template sequence, obtaining a template subsequence for each subsequence and a matching degree between each subsequence and its template subsequence; if the matching degree between any subsequence and its template subsequence is not 1, subtracting the subsequence from its template subsequence to obtain a difference sequence corresponding to the subsequence; Obtain the attention level of each difference sequence based on the matching degree between each subsequence and its template subsequence and the data fluctuation degree of the difference sequence corresponding to each subsequence; According to the matching degree between each subsequence and its template subsequence and the attention degree of each difference sequence, the compression matrix and the adaptive singular value retention ratio coefficient when the power data sequence is compressed using the singular value decomposition algorithm are obtained. Based on the compression matrix and the adaptive singular value retention ratio coefficient, the power data sequence is compressed to complete the efficient collection of power data in the current compression cycle.

2. The intelligent fusion terminal with efficient data collection function according to claim 1, characterized in that: The obtaining of the power data template sequence includes: Obtain the reference signal of any data in the power equipment at the standard power grid base frequency, obtain the reference power data at each moment in the preset time period, and form a power data template sequence.

3. The intelligent fusion terminal with efficient data collection function according to claim 2, characterized in that: In the power data template sequence, sliding matching is performed on each subsequence with the power data template sequence to obtain a template subsequence of each subsequence and a matching degree between each subsequence and its template subsequence, including: For any subsequence, in the power data template sequence, starting with the first reference power data in the power data template sequence, establishing a window equal to the length of the subsequence, aligning the first power data of the subsequence with each reference power data in the window, and obtaining at least two matching sequences of the subsequence; For any matching sequence, obtaining a matching degree between the any subsequence and the any matching sequence based on a data difference between the any subsequence and the any matching sequence; The degree of matching between any subsequence and each matching sequence is obtained, and the matching sequence corresponding to the maximum matching degree is used as the template subsequence of any subsequence, and the maximum matching degree is used as the matching degree between any subsequence and its template subsequence.

4. The intelligent fusion terminal with efficient data collection function according to claim 3, characterized in that: Obtaining the degree of matching between the any subsequence and the any matching sequence based on the data difference between the any subsequence and the any matching sequence includes: Obtain a mean square error between any subsequence and any matching sequence, substitute the inverse of the mean square error into an exponential function with a natural constant as a base, and obtain a degree of matching between any subsequence and any matching sequence.

5. The intelligent fusion terminal with efficient data collection function according to claim 1, characterized in that: The obtaining of the attention level of each difference sequence according to the matching degree between each subsequence and its template subsequence and the data fluctuation degree of the difference sequence corresponding to each subsequence includes: For any difference sequence, obtaining a change stability of the difference sequence according to a data fluctuation characteristic of the difference sequence; Obtaining at least two reference difference sequences of any one of the difference sequences, and for any one of the reference difference sequences, recording a degree of match between a subsequence corresponding to the any one of the reference difference sequences and its template subsequence as a first matching degree, and recording a degree of match between a subsequence corresponding to the any one of the difference sequences and its template subsequence as a second matching degree, and obtaining a ratio of the first matching degree to the second matching degree to obtain a matching degree ratio; Obtaining a ratio of the change stability of any reference difference sequence to the change stability of any difference sequence to obtain a change stability ratio, obtaining a product of the matching ratio and the change stability ratio to obtain a focus value of any difference sequence compared to any reference difference sequence; Obtaining a time interval between any one of the difference sequences and any one of the reference difference sequences, substituting the inverse of the time interval into an exponential function with a natural constant as a base, to obtain a reference degree of the any one of the reference difference sequences; Obtain the attention value of any difference sequence compared to each reference difference sequence, obtain the reference degree of each reference difference sequence, use the reference degree of each reference difference sequence as the weight coefficient of the attention value of any difference sequence compared to each reference difference sequence, obtain the weighted average of the attention values ​​of any difference sequence compared to each reference difference sequence, and obtain the attention degree of any difference sequence.

6. The intelligent fusion terminal with efficient data collection function according to claim 5, characterized in that: The obtaining, based on the data fluctuation characteristics of the any difference sequence, the change stability of the any difference sequence includes: For any difference data except the first difference data and the last difference data in any difference sequence, obtaining the mean between the left adjacent difference data and the right adjacent difference data of the any difference data, calculating the absolute value of the difference between the any difference data and the mean to obtain a change value of the any difference data, obtaining the ratio of the change value of the any difference data to the absolute value of the mean to obtain a degree of difference of the any difference data; The degree of difference of each difference data is obtained, and the corresponding mean value of the degree of difference is obtained. The opposite number of the mean value of the degree of difference is substituted into an exponential function with a natural constant as the base to obtain the change stability of any difference sequence.

7. The intelligent fusion terminal with efficient data collection function according to claim 5, characterized in that: The obtaining of at least two reference difference sequences of any one of the difference sequences comprises: The difference sequences corresponding to each subsequence are combined into a sequence to be analyzed. In the sequence to be analyzed, if the number of difference sequences before any difference sequence and the number of difference sequences after any difference sequence are both greater than or equal to a preset number, the preset number of difference sequences before any difference sequence and the preset number of difference sequences after any difference sequence are used as reference difference sequences for the any difference sequence; If the number of difference sequences before any difference sequence is less than a preset number and greater than 0, all difference sequences before any difference sequence and a preset number of difference sequences after any difference sequence are used as reference difference sequences for any difference sequence; If the number of difference sequences after any difference sequence is less than a preset number and greater than 0, the preset number of difference sequences before any difference sequence and all difference sequences after any difference sequence are used as reference difference sequences for any difference sequence; If the number of difference sequences before any difference sequence or the number of difference sequences after any difference sequence is 0, the preset number of difference sequences after any difference sequence or the preset number of difference sequences before any difference sequence are used as reference difference sequences for any difference sequence.

8. The intelligent fusion terminal with efficient data collection function according to claim 1, characterized in that: The method of obtaining a compression matrix and an adaptive singular value retention ratio coefficient when compressing the power data sequence using a singular value decomposition algorithm based on the matching degree between each subsequence and its template subsequence and the attention degree of each difference sequence includes: Normalizing the attention of each difference sequence to obtain the normalized attention corresponding to each difference sequence, normalizing the matching degree between the subsequence corresponding to each difference sequence and its template subsequence to obtain the normalized matching degree corresponding to each difference sequence, obtaining the position number of the first reference power data in the template subsequence of the subsequence corresponding to each difference sequence in the power data template sequence, normalizing the position number of the first reference power data in the template subsequence of the subsequence corresponding to each difference sequence to obtain the normalized position number corresponding to each difference sequence, establishing a three-dimensional space coordinate system with the normalized attention corresponding to each difference sequence as the x-axis, the normalized matching degree corresponding to each difference sequence as the y-axis, and the normalized position number corresponding to each difference sequence as the z-axis; In the three-dimensional space coordinate system, obtaining data points corresponding to each difference sequence, and clustering the data points in the three-dimensional space coordinate system to obtain at least one cluster; For any cluster, the difference data in the difference sequence corresponding to the data points in the cluster form a compression matrix, the number of rows of the compression matrix is ​​the number of data points in the cluster, and the number of columns of the compression matrix is ​​the number of difference data in each difference sequence; According to the position of the data point in any cluster and the attention degree of the difference sequence corresponding to the data point in any cluster, the adaptive singular value retention proportional coefficient corresponding to the compression matrix is ​​obtained.

9. The intelligent fusion terminal with efficient data collection function according to claim 8, characterized in that: The step of obtaining the adaptive singular value retention proportional coefficient corresponding to the compression matrix according to the position of the data point in any cluster and the attention degree of the difference sequence corresponding to the data point in any cluster includes: For any data point in any cluster, record the data points in any cluster except the data point as reference data points, obtain the mean distance between the data point and each reference data point, and obtain the distance feature value of any data point; Obtaining a distance feature value for each reference data point, calculating the average of the absolute values ​​of the differences between the distance feature values ​​of any data point and each reference data point to obtain a distance feature difference mean, obtaining the average of the distance feature difference means for each data point in any cluster to obtain a distance difference mean for any cluster; Obtain the mean attention value of the data points in any cluster, obtain the mean distance eigenvalue of the data points in any cluster, normalize the product of the mean attention value, the average distance eigenvalue and the average distance difference of any cluster, and obtain the adaptive singular value retention proportional coefficient corresponding to the compression matrix.

10. The intelligent fusion terminal with efficient data collection function according to claim 1, characterized in that: After obtaining the template subsequence of each subsequence and the matching degree between each subsequence and its template subsequence, the method further includes: For any subsequence, if the matching degree between the subsequence and its template subsequence is 1, then in the power data template sequence, the position number of the first reference power data in the template subsequence of the subsequence is obtained to complete the data compression of the subsequence.

Images