A user coupling analysis method and related device based on random matrix theory

Abstract

The present application discloses a user coupling analysis method and related device based on random matrix theory. By sampling the main measurement indicators and other measurement indicators to obtain corresponding measurement data, a coupling analysis matrix and a reference matrix are constructed for the sampled measurement data based on a moving data window. By analyzing the coupling relationship between the coupling analysis matrix and the reference matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be obtained. Furthermore, by analyzing different sampling times, and because the collected data can be high-dimensional space data, real-time and comprehensive analysis of user coupling is achieved, thereby mining hidden high-dimensional statistical information of users.

Description

Technical Field

[0001] The present application relates to the field of data analysis technology, and in particular to a user coupling analysis method and related devices based on random matrix theory. Background Art

[0002] With the rapid development of distributed resources, renewable energy generation, electric vehicles, and energy storage batteries have become key components of the energy supply model, significantly changing the landscape of energy production and consumption. The widespread adoption of household photovoltaic power generation, energy storage batteries, and electric vehicles has gradually transformed users into energy producers. In this emerging model, users are no longer merely energy consumers but rather energy producers—prosumers. The interaction between users and the energy system is becoming more intimate and flexible, placing higher demands on the efficient management and control of user-side resources. Therefore, addressing the issue of optimal resource allocation at the user side, strengthening understanding of user-side resources, and exploring the coupling factors that influence user electricity usage behavior are key to addressing this issue.

[0003] Currently, research on user coupling analysis has been conducted both domestically and internationally, primarily encompassing two analytical models: mechanism-driven and data-driven. Mechanism-driven coupling analysis methods are based on the causal relationships of the objective physical world and exploit the coupling relationships between environmental factors and direct electricity consumption through various subjective analyses. Data-driven coupling analysis, however, has become the mainstream of current research due to its greater generalizability and intelligence. Data-driven coupling analysis primarily employs data analysis algorithms to uncover non-completely rational user behaviors contained in statistical data sets. Common algorithms employed in data-driven coupling analysis include principal component analysis, support vector machines, mutual information analysis, grey relational analysis, and chaos theory.

[0004] The coupling factors of user electricity usage behavior can change over time. Existing user coupling analysis lacks timeliness and continuous updating capabilities, and data acquisition and processing are difficult, impacting the reliability and accuracy of analysis results. Therefore, how to fully analyze user coupling in real time and mine hidden high-dimensional statistical information about users has become an urgent problem to be solved. Summary of the Invention

[0005] The embodiments of the present application provide a user coupling analysis method and related devices based on random matrix theory. A coupling analysis matrix and a reference matrix are constructed for sampled measurement data based on a moving data window. By analyzing the coupling relationship between the coupling analysis matrix and the reference matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be obtained. Different sampling times can also be analyzed, and the collected data can be data in a high-dimensional space, thereby realizing real-time and comprehensive analysis of user coupling, and further mining hidden high-dimensional statistical information of users.

[0006] In a first aspect, an embodiment of the present application provides a user coupling analysis method based on random matrix theory, including:

[0007] Acquire a first sampling data set and a second sampling data set according to a preset sampling time interval, wherein the first sampling data set includes an electricity consumption data set of a first user, and the second sampling data set includes one of the following: an environment data set and an electricity consumption data set of a second user;

[0008] Preprocessing the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set;

[0009] Obtaining a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point;

[0010] intercepting, based on a preset moving data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set, intercepting a portion of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set, and constructing a target basic measurement matrix and a target augmented measurement matrix based on the first partial measurement data set and the second partial measurement data set, respectively;

[0011] Determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix respectively;

[0012] Determining a first linear eigenvalue of the target coupling analysis matrix and determining a second linear eigenvalue of the target reference matrix;

[0013] A difference between the first linear eigenvalue and the second linear eigenvalue is calculated to obtain a target difference; the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix.

[0014] In a second aspect, an embodiment of the present application provides a user coupling analysis device based on random matrix theory, the user coupling analysis device based on random matrix theory comprising:

[0015] a data acquisition unit, configured to acquire a first sampling data set and a second sampling data set at a preset sampling time interval, wherein the first sampling data set includes an electricity consumption data set of a first user, and the second sampling data set includes one of the following: an environmental data set and an electricity consumption data set of a second user;

[0016] a preprocessing unit, configured to preprocess the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set;

[0017] The data acquisition unit is further configured to acquire a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point;

[0018] a matrix construction unit, configured to intercept, based on a preset moving data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set, intercept a portion of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set, and construct a target basic measurement matrix and a target augmented measurement matrix based on the first partial measurement data set and the second partial measurement data set, respectively;

[0019] an analysis and determination unit, configured to respectively determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix;

[0020] The analysis and determination unit is further configured to determine a first linear eigenvalue of the target coupling analysis matrix and a second linear eigenvalue of the target reference matrix;

[0021] A calculation unit is used to calculate the difference between the first linear eigenvalue and the second linear eigenvalue to obtain a target difference; the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix.

[0022] In a third aspect, an embodiment of the present application provides an electronic device comprising: a processor, a memory, a communication interface, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, and the program includes instructions for executing the steps in the first aspect of the embodiment of the present application.

[0023] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the above-mentioned computer-readable storage medium stores a computer program for electronic data exchange, wherein the above-mentioned computer program enables a computer to execute some or all of the steps described in the first aspect of the embodiment of the present application.

[0024] In a fifth aspect, embodiments of the present application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program, wherein the computer program is operable to cause a computer to perform some or all of the steps described in the first aspect of the embodiments of the present application. The computer program product may be a software installation package.

[0025] By implementing the embodiments of the present application, a first sampling data set and a second sampling data set are acquired according to a preset sampling time interval, the first sampling data set and the second sampling data set are preprocessed respectively to obtain a first measurement data set and a second measurement data set, a sampling time point of at least one measurement data in the first measurement data set is acquired to obtain a target sampling time point, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point is intercepted based on a preset moving data window to obtain a first partial measurement data set, a portion of the measurement data in the second measurement data set corresponding to the target sampling time point is intercepted to obtain a second partial measurement data set, and a target basic measurement matrix and a target augmented measurement matrix are respectively constructed based on the first partial measurement data set and the second partial measurement data set, a target coupling analysis matrix and a target reference matrix are respectively determined based on the target basic measurement matrix and the target augmented measurement matrix, a first linear eigenvalue of the target coupling analysis matrix is ​​determined, a second linear eigenvalue of the target reference matrix is ​​determined, a difference between the first linear eigenvalue and the second linear eigenvalue is calculated to obtain a target difference, and a coupling relationship between the target coupling analysis matrix and the target reference matrix is ​​determined based on the target difference. Therefore, by analyzing different sampling times and the collected data can be high-dimensional space data, real-time and comprehensive analysis of user coupling is achieved, thereby mining hidden high-dimensional statistical information of users. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the background technology, the drawings required for use in the embodiments of the present application or the background technology will be described below.

[0027] Figure 1 This is a structural diagram of a user coupling analysis system based on random matrix theory provided in an embodiment of the present application;

[0028] Figure 2 This is a flow chart of a user coupling analysis method based on random matrix theory provided in an embodiment of the present application;

[0029] Figure 3 This is a schematic diagram of the structure of a user coupling analysis device based on random matrix theory provided in an embodiment of the present application;

[0030] Figure 4 This is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0032] The terms "first," "second," and the like in the specification and claims of this application and the accompanying drawings are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0033] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0034] The following describes the relevant contents, concepts, meanings, technical issues, technical solutions, beneficial effects, etc. involved in the embodiments of this application.

[0035] First, some of the terms involved in this application are explained:

[0036] Random Matrix Theory: Random matrices are matrices whose elements are random variables. Random matrix theory primarily studies the statistical properties of the characteristic roots (spectra) and eigenvectors of random matrices. It analyzes correlations in random data, characterizes data volatility, and maps data characteristics to physical systems.

[0037] Linear eigenvalue statistic (LES): This generally refers to a method used in statistics and data analysis to evaluate linear data structures. Specifically, it involves calculating the eigenvalues ​​of a dataset or matrix and then performing statistical analysis on these eigenvalues.

[0038] The embodiments of the present application are described below in conjunction with the drawings in the embodiments of the present application.

[0039] See also Figure 1 , Figure 1 This is a structural diagram of a user coupling analysis system based on random matrix theory provided in an embodiment of the present application. The user coupling analysis system 100 based on random matrix theory includes: an input module 110, a processing module 120, and an output module 130. The processing module 120 includes a data preprocessing module 121 and a matrix construction module 122.

[0040] Among them, the user coupling analysis system 100 based on random matrix theory mainly uses random matrix theory to analyze the coupling between users and the environment or between users, and evaluates the degree of correlation between user behavior and environmental conditions by preprocessing the collected sampling data and constructing a corresponding analysis matrix.

[0041] The first sampling data set and the second sampling data set are acquired through the input module 110 . The first sampling data set includes a first user's electricity usage data set, and the second sampling data set includes an environment data set or a second user's electricity usage data set.

[0042] Among them, the processing module 120 includes a data preprocessing module 121 and a matrix construction module 122. The preprocessing module 121 mainly preprocesses the first sampling data set and the second sampling data set respectively to obtain the first measurement data set and the second measurement data set. The matrix construction module 122 constructs the target basic measurement matrix and the target augmented measurement matrix based on the first measurement data set and the second measurement data set, and can construct the target coupling analysis matrix and the target reference matrix for the final analysis based on the target basic measurement matrix and the target augmented measurement matrix.

[0043] Among them, the output module 130 is responsible for performing coupling analysis on the target coupling analysis matrix and the target reference matrix and generating analysis results. It mainly determines the coupling relationship between the target coupling analysis matrix and the target reference matrix by calculating the linear eigenvalue difference between them.

[0044] By adopting the user coupling analysis system 100 based on random matrix theory in the embodiment of the present application, a coupling analysis matrix and a reference matrix can be constructed for the sampled measurement data based on a moving data window. Based on the analysis of the coupling relationship between the coupling analysis matrix and the reference matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be obtained. Different sampling times can also be analyzed, and the collected data can be data in a high-dimensional space, thereby realizing real-time and comprehensive analysis of user coupling, and further mining hidden high-dimensional statistical information of users.

[0045] See Figure 2 , Figure 2: This is a flow chart of a user coupling analysis method based on random matrix theory provided in an embodiment of the present application. The method includes but is not limited to the following steps:

[0046] S201 : Acquire a first sampling data set and a second sampling data set at a preset sampling time interval, where the first sampling data set includes a first user's electricity usage data set, and the second sampling data set includes one of the following: an environment data set and a second user's electricity usage data set.

[0047] The preset sampling time interval may be a standard time series, such as every minute, every hour, or every day, or a pre-set time interval. The sampling time interval may be selected according to specific needs.

[0048] The first sampled data set may include an electricity usage data set of a first user, where the first user is the primary measurement indicator analyzed in the embodiments of this application. The electricity usage data set represents multiple electricity usage data collected from the user within a preset time period. The electricity usage data may include user power, voltage, current, and other data. The electricity usage data set of the first user may be multidimensional data, i.e., it may include multiple types of data such as user power, voltage, and current, and each electricity usage data set includes its corresponding sampling time.

[0049] The second sampling data set is different from the first sampling data set. The second sampling data set may include one of the following: an environmental data set and a second user's electricity consumption data set. The environment or the second user is another measurement indicator analyzed in the embodiment of the present application. The second user is different from the first user. The first user belongs to the main measurement indicator. The embodiment of the present application analyzes the coupling relationship between the main measurement indicator and other measurement indicators. The environmental data set can be the maximum ambient temperature, the minimum ambient temperature, the air humidity, the wind speed, etc. The environmental data set is one-dimensional data. The electricity consumption data in the second user's electricity consumption data set can be user power, voltage, current and other data. The second user's electricity consumption data set is one-dimensional data, that is, only one type of data is taken for analysis with the main measurement indicator. The purpose is to ensure the accuracy of the subsequent coupling analysis of the main measurement indicator and other measurement indicators.

[0050] Specifically, the embodiment of the present application mainly analyzes the coupling relationship between the first user and the environment or the first user and the second user, and therefore obtains the first sampling data set and the second sampling data set according to the preset sampling time interval. Specifically, the user's electricity consumption data and environmental data can be collected from the power grid or other related systems, wherein the user's electricity consumption data can be collected from photovoltaic, electric vehicle charging piles, industrial and other scenarios, and the environmental data within a preset time period can be collected from the weather station. When collecting data, multiple sampling can be performed at the sampling time point to obtain multiple data at the current sampling time point.

[0051] S202: Preprocess the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set.

[0052] Specifically, the first and second sample data sets are preprocessed separately to detect and address possible outliers or erroneous data. The collected data is cleaned and normalized to ensure data quality. To analyze the coupling relationship between the primary measurement indicator and other measurement indicators at the same time point, the first and second sample data sets can be time-aligned so that the preprocessed first and second measurement data sets correspond in time, ensuring the accuracy of subsequent coupling analysis.

[0053] Furthermore, the first and second sample data sets may be time-aligned according to specific sampling time points, or multiple sampling time points may be divided into time periods, and the first and second sample data sets may be time-aligned according to the time periods.

[0054] Optionally, the above step S202 of preprocessing the first sampling data set and the second sampling data set to obtain the first measurement data set and the second measurement data set may include the following steps:

[0055] A21. Determine a first sampling time set for the first sampling data set;

[0056] A22. Divide the two timestamps at predetermined time intervals in the first sampling time set into timestamp tuples to obtain multiple timestamp tuples;

[0057] A23. Sample the first sampling data set according to the multiple timestamp tuples to obtain the first measurement data set, and sample the second sampling data set according to the multiple timestamp tuples to obtain the second measurement data set.

[0058] Specifically, each sampled data in the first sampling data set corresponds to a sampling time. A first sampling time set of the first sampling data set can be obtained, where the sampling time can be represented by a timestamp. The sampling times in the first sampling time set can be sorted in chronological order to facilitate subsequent coupling analysis in chronological order. Since the data sampled in the first sampling data set is the primary measurement indicator, the first sampling time set is used as the standard time scale to align all data in the first measurement data set and the second measurement data set according to the standard time scale.

[0059] Specifically, two timestamps at intervals of a preset time interval in the first sampling time set are divided into timestamp tuples to obtain multiple timestamp tuples, wherein the preset time interval can be a system default or pre-set, and is used to standardize and control the generation of timestamp tuples. The timestamp tuple is a tuple consisting of two timestamps, wherein the two timestamps can be represented as a first timestamp and a second timestamp, the position interval of the first timestamp and the second timestamp is a preset time interval, and the time interval between the first timestamp and the second timestamp is a time period. For example, in the embodiment of the present application, the preset time interval is set to 1, then according to the first sampling time t0 in the first sampling time set and the preset time interval, the first sampling time set can be divided into multiple timestamp tuples, and the multiple timestamp tuples can be represented as (t0, t1), (t1, t2), ..., (t n-1 , t n ), where t n Indicates the nth timestamp of the first sampling time set, where n is the size of the first sampling time set. If the preset time interval is set to 2, multiple timestamp tuples can be represented as (t0, t2), (t1, t3), ..., (t n-2 , t n The size of the preset time interval directly affects the time period size of each timestamp tuple. The larger the preset time interval is, the longer the time period formed by a timestamp tuple will be.

[0060] Specifically, data is sampled from the first sampled dataset based on multiple timestamp tuples. Specifically, data in the first sampled dataset with sampling times within the same timestamp tuple is grouped together. This means that multiple samplings can be performed within the time period corresponding to the same timestamp tuple. Ultimately, the first measurement dataset is obtained based on the multiple samplings. For ease of understanding, using an array as an example, the first measurement dataset can be imagined as a two-dimensional array, where each column represents a timestamp tuple, and each row represents measurement data, potentially containing multiple measurement data. For example, if data is collected hourly, the timestamp tuple in the first column of the first measurement dataset might represent the time period from 0:00 to 1:00, and the timestamp tuple in the second column might represent the time period from 1:00 to 2:00. The multiple measurement data corresponding to each time period are arranged in the same column, ultimately resulting in the first measurement dataset. Similarly, data is sampled from the second sampled dataset based on multiple timestamp tuples to obtain the second measurement dataset.

[0061] Among them, the first sampling time set is used as the standard time scale, and all data of the first measurement data set and the second measurement data set are aligned according to the standard time scale, that is, each timestamp tuple in the first measurement data set and the second measurement data set can be mapped to the same time axis. Regardless of whether sampling is performed according to any preset sampling time interval, by aligning all data of the first measurement data set and the second measurement data set according to the standard time scale, the data points in the first measurement data set and the second measurement data set can be accurately analyzed.

[0062] Optionally, the following steps may also be included:

[0063] B21. Select a target timestamp tuple from the multiple timestamp tuples;

[0064] B22. Determine whether the time span of the two timestamps in the target timestamp tuple is less than a preset sampling missing threshold;

[0065] B23. If the target timestamp is less than the preset sampling missing threshold, skip the target timestamp tuple;

[0066] B24. Otherwise, corresponding sampling data is collected from the first sampling data set according to the target timestamp tuple, and corresponding sampling data is collected from the second sampling data set according to the target timestamp tuple.

[0067] If the preset sampling interval is a pre-set time interval, when constructing a timestamp tuple, if the span between two adjacent timestamps in the timestamp tuple is greater than a preset sampling loss threshold, it indicates that sampling is missing or interrupted within that time period. The sampling loss threshold can be a system default or pre-set threshold and represents the maximum acceptable time span between two adjacent timestamps in the timestamp tuple.

[0068] Specifically, a target timestamp tuple is selected from multiple timestamp tuples, where the target timestamp tuple is any timestamp tuple from the multiple timestamp tuples. A determination is made as to whether the time span between two timestamps in the target timestamp tuple is less than a preset sampling missing threshold. If so, the target timestamp tuple is skipped. Otherwise, corresponding sampling data is collected from the first sampling dataset based on the target timestamp tuple, and corresponding sampling data is collected from the second sampling dataset based on the target timestamp tuple. The timestamp tuples in the multiple timestamp tuples are analyzed, and after a comprehensive analysis of all processed timestamp tuples, the first measurement dataset and the second measurement dataset are ultimately obtained based on the timestamp tuples, thereby ensuring the accuracy of subsequent coupling analysis results.

[0069] For example, if sampling is performed without a fixed time interval, with the first and second sampling times separated by 5 hours, the second and third sampling times separated by 8 hours, and the third and fourth sampling times separated by 13 hours, then a first sampling dataset can be obtained based on the preset sampling intervals and divided into multiple timestamp tuples. If the preset sampling loss threshold is set to 12 hours, and the timestamp tuples divided by the third and fourth sampling times span longer than 12 hours, it is determined that sampling was lost or interrupted during this time period, and the timestamp tuple will be skipped, discarding the data collected between the third and fourth sampling times.

[0070] S203: Acquire a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point.

[0071] Specifically, a sampling time point of at least one measurement data in the first measurement data set is obtained to obtain a target sampling time point, and the measurement data in the first measurement data set and the second measurement data set at the target sampling time point are analyzed to obtain the coupling relationship between the main measurement indicator and other measurement indicators at this time point.

[0072] Furthermore, the target sampling time point may be a time period, and the measurement data in the first measurement data set and the second measurement data set in the time period may be analyzed.

[0073] S204. Based on a preset mobile data window, intercept part of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set; intercept part of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set; and construct a target basic measurement matrix and a target augmented measurement matrix according to the first partial measurement data set and the second partial measurement data set, respectively.

[0074] The preset moving data window refers to a data window that moves based on a time series, used to intercept data from a continuous data set for further processing or analysis. The preset moving data window can be set to different sizes based on requirements. Its length represents the number of samples at the target sampling time point, and its size can be measured based on the number of data points. For example, if the target sampling time points are from 1 to 3 o'clock and the preset sampling interval is 1 hour, three measurement data sets can be obtained. If the moving data window length is set to 5, the moving data window size is 15, indicating that it can contain 15 data points.

[0075] Specifically, based on the preset mobile data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point is intercepted to obtain the first partial measurement data set. Based on the preset mobile data window, a portion of the measurement data in the second measurement data set corresponding to the target sampling time point is intercepted to obtain the second partial measurement data set. It should be noted that the size of the preset mobile data window needs to be set according to specific needs. For example, in the embodiment of the present application, three samples are required at the sampling time point, the number of measurement data of the main measurement indicator is 3, and the number of measurement data of other measurement indicators is 2. Then, the size of the mobile data window for intercepting the first measurement data set can be 9, and the size of the mobile data window for intercepting the second measurement data set can be 6.

[0076] Furthermore, the target basic measurement matrix is ​​constructed based on the first part of the measurement data set. The number of measurement data of the main measurement indicators in the first part of the measurement data set is N b Indicates that the number of samples at the target sampling time point is represented by S, and the target basic measurement matrix is ​​represented by M b Indicates that the dimension of the target basic measurement matrix is ​​N b ×S. The target augmented measurement matrix is ​​constructed based on the second part of the measurement data set. The number of measurement data of other measurement indicators in the second part of the measurement data set is N a Indicates that the number of sampling times at the target sampling time point is the same as above, represented by S, and the target augmented measurement matrix is ​​represented by M a Represented as follows, the dimension of the target augmented measurement matrix is ​​N a × S. Among them, the number of measurement data of the main measurement indicators is N b The number of measurement data N greater than other measurement indicators a .

[0077] Optionally, before the above step S204 of intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on the preset mobile data window to obtain the first partial measurement data set, the following steps may be further included:

[0078] A41. Obtain the number of data in the first measurement data set at the target sampling time point to obtain a target measurement quantity;

[0079] A42. Determine half of the size of the preset mobile data window as a first quantity threshold;

[0080] A43. When the target measurement quantity is less than the first quantity threshold, execute the step of obtaining a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point.

[0081] Specifically, the number of data in the first measurement data set at the target sampling time point is obtained to obtain the target measurement quantity. The target measurement data can be used to represent the data density at the current target sampling time point. Half of the size of the preset mobile data window is determined as a first quantity threshold. The first quantity threshold is an integer. If half of the size of the preset mobile data window is a decimal, it is rounded down according to the floor operator. When the target measurement quantity is less than the first quantity threshold, the step of obtaining the sampling time point of at least one measurement data in the first measurement data set to obtain the target sampling time point is executed. That is, the data density of the target sampling time point is less than the preset data density threshold. It can be determined that there is data missing at the current target sampling time point, and it is necessary to abandon the target sampling time point and reselect a new sampling time point to ensure that the subsequent coupling analysis can be carried out accurately.

[0082] Optionally, the above step S204, intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on a preset mobile data window to obtain the first partial measurement data set, may include the following steps:

[0083] B41. Determine the size of the preset mobile data window as a second quantity threshold;

[0084] B42. When the target measurement quantity is greater than or equal to the first quantity threshold and the target measurement quantity is less than the second quantity threshold, intercepting the first measurement data set based on the preset mobile data window to obtain a reference partial measurement data set;

[0085] B43. Fill the reference partial measurement data set using a preset difference function to obtain the first partial measurement data set.

[0086] Specifically, the size of the preset mobile data window is set to a second quantity threshold. When the target measurement quantity is greater than or equal to the first quantity threshold and less than the second quantity threshold, the first measurement data set is intercepted based on the preset mobile data window to obtain a reference partial measurement data set. Since the number of data in the reference partial measurement data set cannot completely fill the preset mobile data window, a preset difference function can be used to fill the remaining gaps in the preset mobile data window. The preset difference function can be linear interpolation, etc., which is not limited here. By filling the reference partial measurement data set with the preset difference function, the first partial measurement data set can be obtained, thereby ensuring the integrity of the data set and improving the accuracy of subsequent coupling analysis.

[0087] Optionally, the following steps may also be included:

[0088] C41. When the target measurement quantity is greater than the second quantity threshold, determining the target number of segments according to the target measurement quantity and the second quantity threshold;

[0089] C42. Segment the first measurement data set according to the target number of segments to obtain multiple segmented data sets;

[0090] C43. Randomly sample each segmented data set from the multiple segmented data sets, and fill the sample into the preset movement data window in sequence to obtain the first part of the measurement data set.

[0091] Specifically, when the target measurement quantity is greater than the second quantity threshold, data in the first measurement data set may be extracted by segmented random sampling and filled into a preset mobile data window, thereby finally obtaining a first portion of the measurement data set.

[0092] Specifically, a target number of segments is obtained by dividing the target measurement quantity by a second quantity threshold and rounding up the result according to a rounding-up operator. The first measurement dataset is segmented according to the target number of segments to obtain multiple segmented datasets, each of which contains the target number of segments. Random samples are taken from each of the multiple segmented datasets and sequentially filled into a preset mobile data window to obtain a first portion of the measurement dataset.

[0093] Furthermore, if the number of segmented data sets is less than the second threshold value, indicating that the data in the preset mobile data window is incomplete, interpolation can be used to fill the remaining gaps in the preset mobile data window, ultimately obtaining the first measurement data set. For example, if the number of data points in the first measurement data set, i.e., the target measurement quantity, is 10, and the size of the preset mobile data window, i.e., the second threshold value, is 6, then the target number of segments calculated by rounding is 2. The first measurement data set is segmented to obtain multiple data sets, each of which has 2 data points. Random sampling of each of the multiple segmented data sets results in only 5 data points. In this case, interpolation is required to fill the remaining gaps in the preset mobile data window.

[0094] In an embodiment of the present application, a portion of measurement data in a first measurement data set corresponding to a target sampling time point is intercepted based on a preset mobile data window to obtain a first partial measurement data set, and a portion of measurement data in a second measurement data set corresponding to the target sampling time point is intercepted based on the preset mobile data window to obtain a second partial measurement data set, ensuring that a target basic measurement matrix can be constructed based on the first partial measurement data set, and ensuring that a target augmented measurement matrix can be constructed based on the second partial measurement data set, so as to improve the efficiency and reliability of subsequent data analysis.

[0095] S205 . Determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix respectively.

[0096] Among them, the target basic measurement matrix reflects the characteristic information of the main measurement indicators, and the target augmented measurement matrix reflects the characteristic information of other measurement indicators. By analyzing the relationship between the target basic measurement matrix and the target augmented measurement matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be determined, that is, the degree of mutual correlation between the main measurement indicators and other measurement indicators in data characteristics or other aspects can be analyzed, so as to reveal the complex relationships and potential influencing factors within the data set.

[0097] Specifically, in order to improve the accuracy of coupling analysis, the target coupling analysis matrix and the target reference matrix can be determined according to the target basic measurement matrix and the target augmented measurement matrix respectively. By analyzing the coupling relationship between the target coupling analysis matrix and the target reference matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be better determined.

[0098] The target coupling analysis matrix can be obtained by combining the target basic measurement matrix and the optimized target augmented measurement matrix, and the target reference matrix can be obtained by introducing a white noise matrix into the target basic measurement matrix. The noise distribution of the white noise matrix is ​​a standard normal distribution, and the matrix elements of the white noise matrix, i.e., the white noise data, are random data that have no relationship with the data in the first measurement data set and are used only as a reference. For example, the closer the coupling relationship between the target coupling analysis matrix and the target reference matrix, the less correlated the target coupling analysis matrix and the target reference matrix are.

[0099] Optionally, the above step S205, determining a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix, respectively, may include the following steps:

[0100] A51. Determine a target dimension increase parameter according to the target basic measurement matrix and the target augmented measurement matrix;

[0101] A52. Perform dimension increase on the target augmented measurement matrix based on the target dimension increase parameter to obtain a dimension increase augmented matrix;

[0102] A53. Determine the target dimension of the dimension-increased augmented matrix;

[0103] A54. Obtain a white noise matrix, where the dimension of the white noise matrix is ​​the same as the target dimension;

[0104] A55. Introducing the white noise matrix into the dimension-increased augmented matrix to obtain an extended augmented matrix;

[0105] A56. Determine the target coupling analysis matrix according to the target basic measurement matrix and the extended augmented matrix;

[0106] A57. Determine the target reference matrix according to the target basic measurement matrix and the white noise matrix.

[0107] Specifically, the target dimension-increasing parameter is represented by k. The target dimension-increasing parameter can be determined based on the target basic measurement matrix and the target augmented measurement matrix. Since the dimension of the target basic measurement matrix is ​​N b ×S, and the dimension of the target augmented measurement matrix is ​​N a ×S, then the target basic measurement matrix is ​​divided by the dimension of the target augmented measurement matrix, and the target augmented dimension parameter can be obtained by rounding down according to the floor operator, that is, Denotes the floor operator. According to the target dimension-increasing parameter, the target augmented measurement matrix is ​​increased in dimension to obtain the increased dimension augmented matrix, which can be used Indicates that the increased dimension augmentation matrix can improve the accuracy of coupling analysis. Among them, the target dimension of the increased dimension augmentation matrix is ​​(k×N a )×S.

[0108] Furthermore, a white noise matrix is ​​obtained. The white noise matrix can be represented by N. The dimension of the white noise matrix is ​​the same as the target dimension, which is (k×N a )×S. The purpose of introducing the white noise matrix is ​​to prevent repeated data from interfering with the subsequent coupling analysis, so as to increase the diversity of the data. The white noise matrix is ​​introduced into the dimension-increasing augmented matrix to obtain the extended augmented matrix. The extended augmented matrix can be used to e It is expressed as and calculated by the following formula: In the above formula, α is the noise amplitude, which directly affects the results of the coupling analysis. If the noise amplitude is too small, repeated data in the expanded augmented matrix can still interfere with the coupling analysis. Conversely, if the noise amplitude is too large, the measurements in the expanded augmented matrix will be distorted, resulting in abnormal measurements and unreliable analysis results. Therefore, when selecting the noise amplitude, it is necessary to balance the characteristics of the data and the accuracy of the analysis to ensure the reliability of the coupling analysis results.

[0109] In the embodiment of the present application, in order to quantitatively evaluate and better select the noise amplitude of white noise, the signal-to-noise ratio of the extended augmented matrix is ​​defined according to the following formula:

[0110]

[0111] In the above formula, ρ(M e ) is the signal-to-noise ratio, Tr() is the trace of the matrix, M e To expand the augmented matrix, is the transpose of the expanded augmentation matrix, α is the noise amplitude, N is the white noise matrix, N T is the transpose of the white noise matrix.

[0112] To ensure the accuracy and consistency of subsequent coupling analyses, the signal-to-noise ratio (SNR) of the expanded augmented matrix must be uniform for each coupling analysis. This uniform SNR allows for the determination of the appropriate noise amplitude across different coupling analyses, ensuring consistency and comparability across the analysis. This approach helps maintain the stability of data analysis results and reduces bias caused by variations in noise amplitude.

[0113] Furthermore, the target basic measurement matrix and the extended augmented matrix are combined to obtain the target coupling analysis matrix, which is represented by C, where is the transpose of the target basic measurement matrix, is the transpose of the extended augmentation matrix. Combining the target basic measurement matrix and the white noise matrix, the target reference matrix can be obtained. The target reference matrix is ​​calculated using C r Indicates that, is the transpose of the target basic measurement matrix, N T is the transpose of the white noise matrix. The target coupling analysis matrix serves as the data source for coupling analysis, and the target reference matrix serves as the reference for coupling analysis. Constructing the target reference matrix instead of directly performing coupling analysis using the target coupling analysis matrix and the white noise matrix can reduce the interference caused by white noise to a certain extent.

[0114] S206 : Determine a first linear eigenvalue of the target coupling analysis matrix, and determine a second linear eigenvalue of the target reference matrix.

[0115] Among them, the linear eigenvalue statistic (LES) can be obtained by setting different detection functions To extract and analyze features of high-dimensional spatial data, it should be noted that the embodiment of the present application is based on random matrix theory and solves the statistical problem of high-dimensional spatial data by studying the limiting spectral distribution (LSD) of high-dimensional matrices, that is, by statistically analyzing a large amount of power data in the power system, it can be seen that the uncertainty of the user's electricity consumption behavior constitutes the randomness of the measurement data. Therefore, by studying the user's power data and analyzing and verifying it, the Ring Law and MP Law (Marcenko-Pastur Law) in the random matrix theory are verified in the application of the power system. It is further explained that the embodiment of the present application can analyze the coupling between users and the environment or between users based on linear eigenvalue statistics.

[0116] Specifically, the first linear eigenvalue of the target coupling analysis matrix and the second linear eigenvalue of the target reference matrix are calculated based on the preset linear eigenvalue function. The first linear eigenvalue and the second linear eigenvalue represent linear eigenvalue statistics. Analyzing the relationship between the first linear eigenvalue and the second linear eigenvalue can further determine the relationship between the target coupling analysis matrix and the target reference matrix.

[0117] Optionally, the above step S206, determining the first linear eigenvalue of the target coupling analysis matrix and determining the second linear eigenvalue of the target reference matrix, may include the following steps:

[0118] A61. Obtain the preset linear eigenvalue expression;

[0119] A62, obtain target detection function;

[0120] A63. Substitute the target detection function into the preset linear eigenvalue expression to obtain a linear eigenvalue function;

[0121] A64. Obtain target eigenvalues ​​of the target coupling analysis matrix;

[0122] A65. Input the target eigenvalue into the linear eigenvalue function to obtain the first linear eigenvalue;

[0123] A66. Determine the second linear eigenvalue according to the target reference matrix and the linear eigenvalue function.

[0124] Specifically, a preset linear eigenvalue expression is obtained, and a target detection function is obtained. Commonly used detection functions include average spectral radius, Chebyshev polynomials, determinants, likelihood functions, etc. Different detection functions can be selected for further analysis. The target detection function is substituted into the preset linear eigenvalue expression to obtain a linear eigenvalue function. According to the linear eigenvalue function, the linear eigenvalues ​​of the target coupling analysis matrix and the target reference matrix can be calculated.

[0125] Furthermore, the target eigenvalue of the target coupling analysis matrix is ​​obtained, and the target eigenvalue is input into the linear eigenvalue function to obtain the first linear eigenvalue of the target coupling analysis matrix. Similarly, the second linear eigenvalue of the target reference matrix can be determined according to the target reference matrix and the linear eigenvalue function.

[0126] The preset linear eigenvalue expression can be:

[0127]

[0128] In the above formula, λ is the eigenvalue of the random matrix,

[0129] Among them, the target detection function can be:

[0130]

[0131] Specifically, the linear eigenvalue function can be obtained as:

[0132]

[0133] Specifically, according to the preset linear eigenvalue expression and the target detection function, a linear eigenvalue function can be obtained, according to the target coupling analysis matrix and the linear eigenvalue function, the first linear eigenvalue can be determined, and according to the target reference matrix and the linear eigenvalue function, the second linear eigenvalue can be determined.

[0134] S207 . Calculate a difference between the first linear eigenvalue and the second linear eigenvalue to obtain a target difference; the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix.

[0135] Specifically, the difference between the first linear eigenvalue and the second linear eigenvalue is calculated to obtain the target difference. By judging the size of the target difference, the coupling relationship between the target coupling analysis matrix and the target reference matrix can be obtained. For example, the smaller the target difference, the smaller the gap between the first linear eigenvalue and the second linear eigenvalue, and the less correlated the coupling relationship between the target coupling analysis matrix and the target reference matrix.

[0136] For example, the first user is used as the primary measurement indicator, and the second user is used as an additional measurement indicator. After obtaining the measurement data of the first and second users, the coupling relationship between the first and second users at the target sampling time point is analyzed. Specifically, partial measurement data of the first user is processed to obtain a target basic measurement matrix, and partial measurement data of the second user is processed to obtain a target augmented measurement matrix. Based on the target basic measurement matrix and the target augmented measurement matrix, a target coupling analysis matrix can be obtained. Based on the target basic measurement matrix and the white noise matrix, a target reference matrix can be obtained. If the target coupling analysis matrix and the target reference matrix are closer, it means that the impact of the second user on the first user is similar to that of noise, that is, there is no impact, which can indicate that the coupling relationship between the primary measurement indicator and the other measurement indicators is unrelated.

[0137] Optionally, the following steps may also be included:

[0138] A71. Obtain a target sampling time set for the first measurement data set; the target sampling time set includes multiple sampling time points;

[0139] A72. Obtain a difference between the first linear eigenvalue and the second linear eigenvalue at each sampling time point in the target sampling time set to obtain a plurality of linear eigenvalue differences;

[0140] A73. Perform fitting based on the multiple linear eigenvalue differences and the sampling time points corresponding to each linear eigenvalue difference to obtain a target fitting curve, wherein the horizontal axis of the target fitting curve is the sampling time points, and the vertical axis of the target fitting curve is the linear eigenvalue differences;

[0141] A74. intercepting a partial fitting curve segment from the target fitting curve, and determining an average linear characteristic value difference based on the partial fitting curve segment;

[0142] A75. Determine a coupling relationship between the first measurement data set and the second measurement data set according to the average linear eigenvalue difference.

[0143] Specifically, a target sampling time set of the first measurement data set is obtained, wherein the target sampling time set includes multiple sampling time points, and the difference between the first linear eigenvalue and the second linear eigenvalue at each sampling time point in the target sampling time set is obtained to obtain multiple linear eigenvalue differences. Fitting is performed based on the multiple linear eigenvalue differences and the sampling time points corresponding to each linear eigenvalue difference to obtain a target fitting curve, wherein the horizontal axis of the target fitting curve is the sampling time point, and the vertical axis of the target fitting curve is the linear eigenvalue difference. A partial fitting curve segment is intercepted from the target fitting curve, and the partial fitting curve segment can be used to analyze the coupling relationship between the first measurement data set and the second measurement data set, and the average linear eigenvalue difference is determined based on the partial fitting curve segment. The coupling relationship between the first measurement data set and the second measurement data set is determined based on the average linear eigenvalue difference.

[0144] Furthermore, the coupling relationship between the main measurement indicators and other measurement indicators can be predicted based on the target fitting curve. For example, the predicted time can be determined. The coupling relationship between the first measurement data set and the second measurement data set at the predicted time can be determined based on the target fitting curve, and then the coupling relationship between the main measurement indicators and other measurement indicators can be judged.

[0145] In an embodiment of the present application, compared with traditional low-dimensional indicators (such as the Pearson correlation coefficient or the Spearman correlation coefficient, etc.), the LES indicator has better analytical robustness for non-continuous data and data with outliers. In the above steps, the first sampling data set and the second sampling data set are pre-processed separately, which cannot effectively remove the abnormal data values ​​in the sampling, so a calculation indicator with better robustness is needed. In an embodiment of the present application, the LES indicator can also directly process high-dimensional matrix inputs, and can more accurately grasp the nonlinear correlation in the data than low-dimensional indicators.

[0146] In summary, by implementing the embodiments of the present application, a first sampling data set and a second sampling data set are obtained according to a preset sampling time interval, the first sampling data set and the second sampling data set are preprocessed respectively to obtain a first measurement data set and a second measurement data set, a sampling time point of at least one measurement data in the first measurement data set is obtained to obtain a target sampling time point, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point is intercepted based on a preset mobile data window to obtain a first partial measurement data set, a portion of the measurement data in the second measurement data set corresponding to the target sampling time point is intercepted to obtain a second partial measurement data set, and a target basic measurement matrix and a target augmented measurement matrix are respectively constructed based on the first partial measurement data set and the second partial measurement data set, a target coupling analysis matrix and a target reference matrix are respectively determined based on the target basic measurement matrix and the target augmented measurement matrix, a first linear eigenvalue of the target coupling analysis matrix is ​​determined, a second linear eigenvalue of the target reference matrix is ​​determined, a difference between the first linear eigenvalue and the second linear eigenvalue is calculated to obtain a target difference, and a coupling relationship between the target coupling analysis matrix and the target reference matrix is ​​determined based on the target difference. Therefore, by analyzing different sampling times and the collected data can be high-dimensional space data, real-time and comprehensive analysis of user coupling is achieved, thereby mining hidden high-dimensional statistical information of users.

[0147] The above describes in detail the method of the embodiment of the present application, and the following provides an apparatus of the embodiment of the present application.

[0148] See Figure 3 , Figure 3 : is a schematic diagram of the structure of a user coupling analysis device based on random matrix theory provided in an embodiment of the present application. The user coupling analysis device 300 based on random matrix theory includes:

[0149] The data acquisition unit 301 is configured to acquire a first sampling data set and a second sampling data set according to a preset sampling time interval, wherein the first sampling data set includes an electricity consumption data set of a first user, and the second sampling data set includes one of the following: an environmental data set and an electricity consumption data set of a second user;

[0150] A preprocessing unit 302 is configured to preprocess the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set;

[0151] The data acquisition unit 301 is further configured to acquire a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point;

[0152] a matrix construction unit 303 configured to intercept, based on a preset moving data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set, intercept a portion of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set, and construct a target basic measurement matrix and a target augmented measurement matrix based on the first partial measurement data set and the second partial measurement data set, respectively;

[0153] An analysis and determination unit 304 is configured to determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix, respectively;

[0154] The analysis and determination unit 304 is further configured to determine a first linear eigenvalue of the target coupling analysis matrix and a second linear eigenvalue of the target reference matrix;

[0155] The calculation unit 305 is configured to calculate a difference between the first linear eigenvalue and the second linear eigenvalue to obtain a target difference; the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix.

[0156] Optionally, in terms of preprocessing the first sampling data set and the second sampling data set to obtain a first measurement data set and a second measurement data set, the preprocessing unit 302 is further specifically configured to:

[0157] determining a first sampling time set for the first sample data set;

[0158] Dividing two timestamps at predetermined time intervals in the first sampling time set into timestamp tuples to obtain a plurality of timestamp tuples;

[0159] Data sampling is performed on the first sampling data set according to the multiple timestamp tuples to obtain the first measurement data set, and data sampling is performed on the second sampling data set according to the multiple timestamp tuples to obtain the second measurement data set.

[0160] Optionally, the user coupling analysis device 300 based on random matrix theory is further specifically used for:

[0161] Selecting a target timestamp tuple from the multiple timestamp tuples;

[0162] Determine whether the time span of two timestamps in the target timestamp tuple is less than a preset sampling missing threshold;

[0163] If it is less than the preset sampling missing threshold, skipping the target timestamp tuple;

[0164] Otherwise, corresponding sampled data is collected from the first sampled data set according to the target timestamp tuple, and corresponding sampled data is collected from the second sampled data set according to the target timestamp tuple.

[0165] Optionally, before intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on the preset moving data window to obtain the first partial measurement data set, the matrix construction unit 303 is further specifically configured to:

[0166] Obtaining the number of data in the first measurement data set at the target sampling time point to obtain a target measurement quantity;

[0167] Determining half of the size of the preset mobile data window as a first quantity threshold;

[0168] When the target measurement quantity is less than the first quantity threshold, the step of obtaining a sampling time point of at least one measurement data in the first measurement data set is performed to obtain a target sampling time point.

[0169] Optionally, in the aspect of intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on the preset moving data window to obtain the first partial measurement data set, the matrix construction unit 303 is further specifically configured to:

[0170] Determining the size of the preset mobile data window as a second quantity threshold;

[0171] When the target measurement quantity is greater than or equal to the first quantity threshold and the target measurement quantity is less than the second quantity threshold, intercepting the first measurement data set based on the preset mobile data window to obtain a reference partial measurement data set;

[0172] The reference partial measurement data set is filled using a preset difference function to obtain the first partial measurement data set.

[0173] Optionally, the user coupling analysis device 300 based on random matrix theory is further specifically used for:

[0174] When the target measurement quantity is greater than the second quantity threshold, determining the target number of segments according to the target measurement quantity and the second quantity threshold;

[0175] Segmenting the first measurement data set according to the target number of segments to obtain multiple segmented data sets;

[0176] Each of the plurality of segmented data sets is randomly sampled and sequentially filled into the preset movement data window to obtain the first portion of the measurement data set.

[0177] Optionally, in determining a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix, the analysis and determination unit 304 is further specifically configured to:

[0178] Determining a target dimension increase parameter according to the target basic measurement matrix and the target augmented measurement matrix;

[0179] Performing dimension increase on the target augmented measurement matrix based on the target dimension increase parameter to obtain a dimension increase augmented matrix;

[0180] Determining a target dimension of the dimension-increased augmented matrix;

[0181] Obtaining a white noise matrix, where the dimension of the white noise matrix is ​​the same as the target dimension;

[0182] Introducing the white noise matrix into the dimension-increased augmented matrix to obtain an extended augmented matrix;

[0183] Determining the target coupling analysis matrix according to the target basic measurement matrix and the extended augmented matrix;

[0184] The target reference matrix is ​​determined according to the target basic measurement matrix and the white noise matrix.

[0185] Optionally, in determining the first linear eigenvalue of the target coupling analysis matrix and determining the second linear eigenvalue of the target reference matrix, the analysis and determination unit 304 is further specifically configured to:

[0186] Get the preset linear eigenvalue expression;

[0187] Get the target detection function;

[0188] Substituting the target detection function into the preset linear eigenvalue expression to obtain a linear eigenvalue function;

[0189] Obtaining target eigenvalues ​​of the target coupling analysis matrix;

[0190] Inputting the target eigenvalue into the linear eigenvalue function to obtain the first linear eigenvalue;

[0191] The second linear eigenvalue is determined according to the target reference matrix and the linear eigenvalue function.

[0192] Optionally, the user coupling analysis device 300 based on random matrix theory is further specifically used for:

[0193] Acquire a target sampling time set for the first measurement data set; the target sampling time set includes a plurality of sampling time points;

[0194] Obtaining a difference between the first linear eigenvalue and the second linear eigenvalue at each sampling time point in the target sampling time set to obtain a plurality of linear eigenvalue differences;

[0195] Fitting is performed according to the plurality of linear eigenvalue differences and the sampling time points corresponding to each linear eigenvalue difference to obtain a target fitting curve; the horizontal axis of the target fitting curve is the sampling time points, and the vertical axis of the target fitting curve is the linear eigenvalue difference;

[0196] intercepting a partial fitting curve segment from the target fitting curve, and determining an average linear characteristic value difference based on the partial fitting curve segment;

[0197] A coupling relationship between the first measurement data set and the second measurement data set is determined according to the average linear eigenvalue difference.

[0198] The user coupling analysis device 300 based on random matrix theory described in the present application can construct a coupling analysis matrix and a reference matrix for the sampled measurement data based on a mobile data window. Based on the analysis of the coupling relationship between the coupling analysis matrix and the reference matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be obtained. Therefore, by analyzing different sampling times and the collected data can be data in a high-dimensional space, real-time and comprehensive analysis of user coupling is achieved, thereby mining hidden high-dimensional statistical information of users.

[0199] See Figure 4 , Figure 4 : is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include a processor, a memory, a communication interface, and one or more programs. The processor, memory, and communication interface may be interconnected via a bus. The one or more programs are stored in the memory and configured to be executed by the processor. In the embodiment of the present application, the program includes instructions for performing the following steps:

[0200] Acquire a first sampling data set and a second sampling data set according to a preset sampling time interval, wherein the first sampling data set includes an electricity consumption data set of a first user, and the second sampling data set includes one of the following: an environment data set and an electricity consumption data set of a second user;

[0201] Preprocessing the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set;

[0202] Obtaining a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point;

[0203] intercepting, based on a preset moving data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set, intercepting a portion of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set, and constructing a target basic measurement matrix and a target augmented measurement matrix based on the first partial measurement data set and the second partial measurement data set, respectively;

[0204] Determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix respectively;

[0205] Determining a first linear eigenvalue of the target coupling analysis matrix and determining a second linear eigenvalue of the target reference matrix;

[0206] A difference between the first linear eigenvalue and the second linear eigenvalue is calculated to obtain a target difference; the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix.

[0207] The electronic device described in this application can construct a coupling analysis matrix and a reference matrix for the sampled measurement data based on a mobile data window. Based on the analysis of the coupling relationship between the coupling analysis matrix and the reference matrix, the coupling relationship between the main measurement indicators and other measurement indicators can be obtained. Therefore, by analyzing different sampling times and the collected data can be data in a high-dimensional space, real-time and comprehensive analysis of user coupling is achieved, thereby mining hidden high-dimensional statistical information of users.

[0208] An embodiment of the present application also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for electronic data exchange, and the computer program enables a computer to execute part or all of the steps of any method described in the above method embodiments, and the above computer includes a terminal device.

[0209] The present application also provides a computer program product comprising a non-transitory computer-readable storage medium storing a computer program, wherein the computer program is operable to cause a computer to perform some or all of the steps of any of the methods described in the above method embodiments. The computer program product may be a software installation package, and the computer may comprise a terminal device.

[0210] Those skilled in the art will appreciate that all or part of the process steps in the above-described method embodiments can be implemented by a computer program instructing the relevant hardware. The program can be stored in a computer-readable storage medium, and when executed, the program can include the process steps in the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0211] The steps of the method or algorithm described in the embodiments of the present application can be implemented in hardware or by a processor executing software instructions. The software instructions can be composed of corresponding software modules, which can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (EEPROM), registers, hard disks, mobile hard disks, CD-ROMs, or any other form of storage medium well known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. Of course, the storage medium can also be an integral part of the processor. The processor and storage medium can be located in an ASIC. In addition, the ASIC can be located in a terminal device or a management device. Of course, the processor and storage medium can also be present in a terminal device or a management device as discrete components.

[0212] Those skilled in the art will appreciate that in one or more of the above examples, the functions described in the embodiments of the present application can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, they can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated therein. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital video disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0213] The modules / units included in the devices and products described in the above embodiments may be software modules / units, hardware modules / units, or partly software modules / units and partly hardware modules / units. For example, for the devices and products applied to or integrated in the chip, the modules / units included therein may all be implemented in the form of hardware such as circuits, or at least part of the modules / units may be implemented in the form of software programs, which run on the processor integrated inside the chip, and the remaining (if any) modules / units may be implemented in the form of hardware such as circuits; for the devices and products applied to or integrated in the chip module, the modules / units included therein may all be implemented in the form of hardware such as circuits, and different modules / units may be located in the same component (such as chip, circuit module, etc.) or different components of the chip module, or at least part of the modules / units may be It is implemented in the form of a software program, which runs on the processor integrated inside the chip module, and the remaining (if any) modules / units can be implemented in the form of hardware such as circuits; for various devices and products applied to or integrated in the terminal equipment, the various modules / units contained therein can be implemented in the form of hardware such as circuits, and different modules / units can be located in the same component (for example, chip, circuit module, etc.) or different components in the terminal equipment, or, at least some modules / units can be implemented in the form of a software program, which runs on the processor integrated inside the terminal equipment, and the remaining (if any) modules / units can be implemented in the form of hardware such as circuits.

[0214] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above description is only a specific implementation method of the embodiments of the present application and is not intended to limit the scope of protection of the embodiments of the present application. Any modifications, equivalent replacements, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the scope of protection of the embodiments of the present application.

Claims

1. A user coupling analysis method based on random matrix theory, characterized in that: The method comprises: Acquire a first sampling data set and a second sampling data set according to a preset sampling time interval, wherein the first sampling data set includes an electricity consumption data set of a first user, and the second sampling data set includes one of the following: an environment data set and an electricity consumption data set of a second user; Preprocessing the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set; Obtaining a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point; intercepting, based on a preset moving data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set, intercepting a portion of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set, and constructing a target basic measurement matrix and a target augmented measurement matrix based on the first partial measurement data set and the second partial measurement data set, respectively; Determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix respectively; Determining a first linear eigenvalue of the target coupling analysis matrix and determining a second linear eigenvalue of the target reference matrix; Calculating a difference between the first linear eigenvalue and the second linear eigenvalue to obtain a target difference; wherein the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix; Before intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on the preset mobile data window to obtain the first partial measurement data set, the method further includes: Obtaining the number of data in the first measurement data set at the target sampling time point to obtain a target measurement quantity; Determining half of the size of the preset mobile data window as a first quantity threshold; When the target measurement quantity is less than the first quantity threshold, performing the step of obtaining a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point; The method of intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on a preset mobile data window to obtain a first partial measurement data set includes: Determining the size of the preset mobile data window as a second quantity threshold; When the target measurement quantity is greater than or equal to the first quantity threshold and the target measurement quantity is less than the second quantity threshold, intercepting the first measurement data set based on the preset mobile data window to obtain a reference partial measurement data set; Filling the reference partial measurement data set by a preset difference function to obtain the first partial measurement data set; The method further comprises: When the target measurement quantity is greater than the second quantity threshold, determining the target number of segments according to the target measurement quantity and the second quantity threshold; Segmenting the first measurement data set according to the target number of segments to obtain multiple segmented data sets; Each of the plurality of segmented data sets is randomly sampled and sequentially filled into the preset movement data window to obtain the first portion of the measurement data set.

2. The method according to claim 1, wherein The preprocessing of the first sampling data set and the second sampling data set to obtain a first measurement data set and a second measurement data set includes: determining a first sampling time set for the first sample data set; Dividing two timestamps at predetermined time intervals in the first sampling time set into timestamp tuples to obtain a plurality of timestamp tuples; Data sampling is performed on the first sampling data set according to the multiple timestamp tuples to obtain the first measurement data set, and data sampling is performed on the second sampling data set according to the multiple timestamp tuples to obtain the second measurement data set.

3. The method according to claim 2, wherein The method further comprises: Selecting a target timestamp tuple from the multiple timestamp tuples; Determine whether the time span of two timestamps in the target timestamp tuple is less than a preset sampling missing threshold; If it is less than the preset sampling missing threshold, skipping the target timestamp tuple; Otherwise, corresponding sampled data is collected from the first sampled data set according to the target timestamp tuple, and corresponding sampled data is collected from the second sampled data set according to the target timestamp tuple.

4. The method according to claim 1, wherein The determining a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix respectively includes: Determining a target dimension increase parameter according to the target basic measurement matrix and the target augmented measurement matrix; Performing dimension increase on the target augmented measurement matrix based on the target dimension increase parameter to obtain a dimension increase augmented matrix; Determining a target dimension of the dimension-increased augmented matrix; Obtaining a white noise matrix, where the dimension of the white noise matrix is ​​the same as the target dimension; Introducing the white noise matrix into the dimension-increased augmented matrix to obtain an extended augmented matrix; Determining the target coupling analysis matrix according to the target basic measurement matrix and the extended augmented matrix; The target reference matrix is ​​determined according to the target basic measurement matrix and the white noise matrix.

5. The method according to any one of claims 1 to 4, characterized in that The determining of the first linear eigenvalue of the target coupling analysis matrix and the determining of the second linear eigenvalue of the target reference matrix include: Get the preset linear eigenvalue expression; Get the target detection function; Substituting the target detection function into the preset linear eigenvalue expression to obtain a linear eigenvalue function; Obtaining target eigenvalues ​​of the target coupling analysis matrix; Inputting the target eigenvalue into the linear eigenvalue function to obtain the first linear eigenvalue; The second linear eigenvalue is determined according to the target reference matrix and the linear eigenvalue function.

6. The method according to claim 5, wherein The method further comprises: Acquire a target sampling time set for the first measurement data set; the target sampling time set includes a plurality of sampling time points; Obtaining a difference between the first linear eigenvalue and the second linear eigenvalue at each sampling time point in the target sampling time set to obtain a plurality of linear eigenvalue differences; Fitting is performed according to the plurality of linear eigenvalue differences and the sampling time points corresponding to each linear eigenvalue difference to obtain a target fitting curve; the horizontal axis of the target fitting curve is the sampling time points, and the vertical axis of the target fitting curve is the linear eigenvalue difference; intercepting a partial fitting curve segment from the target fitting curve, and determining an average linear characteristic value difference based on the partial fitting curve segment; A coupling relationship between the first measurement data set and the second measurement data set is determined according to the average linear eigenvalue difference.

7. A user coupling analysis device based on random matrix theory, characterized in that: The user coupling analysis device based on random matrix theory includes: a data acquisition unit, a preprocessing unit, a matrix construction unit, an analysis and determination unit, and a calculation unit, wherein: The data acquisition unit is configured to acquire a first sampling data set and a second sampling data set according to a preset sampling time interval, wherein the first sampling data set includes an electricity consumption data set of a first user, and the second sampling data set includes one of the following: an environmental data set and an electricity consumption data set of a second user; The preprocessing unit is configured to preprocess the first sampling data set and the second sampling data set respectively to obtain a first measurement data set and a second measurement data set; The data acquisition unit is further configured to acquire a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point; The matrix construction unit is configured to intercept, based on a preset moving data window, a portion of the measurement data in the first measurement data set corresponding to the target sampling time point to obtain a first partial measurement data set, intercept a portion of the measurement data in the second measurement data set corresponding to the target sampling time point to obtain a second partial measurement data set, and construct a target basic measurement matrix and a target augmented measurement matrix based on the first partial measurement data set and the second partial measurement data set, respectively; The analysis and determination unit is configured to respectively determine a target coupling analysis matrix and a target reference matrix based on the target basic measurement matrix and the target augmented measurement matrix; The analysis and determination unit is further configured to determine a first linear eigenvalue of the target coupling analysis matrix and a second linear eigenvalue of the target reference matrix; The calculation unit is configured to calculate a difference between the first linear eigenvalue and the second linear eigenvalue to obtain a target difference; the target difference is positively correlated with the coupling between the target coupling analysis matrix and the target reference matrix; Before intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on the preset mobile data window to obtain the first partial measurement data set, the user coupling analysis device based on random matrix theory is further specifically used to: Obtaining the number of data in the first measurement data set at the target sampling time point to obtain a target measurement quantity; Determining half of the size of the preset mobile data window as a first quantity threshold; When the target measurement quantity is less than the first quantity threshold, performing the step of obtaining a sampling time point of at least one measurement data in the first measurement data set to obtain a target sampling time point; The method of intercepting part of the measurement data in the first measurement data set corresponding to the target sampling time point based on a preset mobile data window to obtain a first partial measurement data set includes: Determining the size of the preset mobile data window as a second quantity threshold; When the target measurement quantity is greater than or equal to the first quantity threshold and the target measurement quantity is less than the second quantity threshold, intercepting the first measurement data set based on the preset mobile data window to obtain a reference partial measurement data set; Filling the reference partial measurement data set by a preset difference function to obtain the first partial measurement data set; The user coupling analysis device based on random matrix theory is further specifically used for: When the target measurement quantity is greater than the second quantity threshold, determining the target number of segments according to the target measurement quantity and the second quantity threshold; Segmenting the first measurement data set according to the target number of segments to obtain multiple segmented data sets; Each of the plurality of segmented data sets is randomly sampled and sequentially filled into the preset movement data window to obtain the first portion of the measurement data set.

Images