User portrait analysis method and system based on big data model, and medium

Abstract

The invention relates to a user portrait analysis method and system based on a big data model and a medium, and the method comprises the following steps: collecting voltage data, and integrating user basic information and power grid topological structure data; constructing a user voltage prediction model; classifying according to the voltage data of the users in four seasons in one year; establishing an association relationship among temperature change, workday and weekend multi-dimensional association factors and user voltage, and obtaining an influence weight coefficient of each factor; carrying out k-means clustering analysis; generating a detailed voltage portrait for each user or region according to the analysis result; according to the voltage portrait, a power grid enterprise takes measures in a targeted manner to improve the voltage quality of a user and the power supply reliability; through continuous data collection and analysis, the user voltage portrait is updated in real time. According to the method, the user voltage is decomposed into the basic voltage and the sensitive voltage according to seasons, clustering analysis is carried out on the basic voltage and the sensitive voltage, and the portrait analysis result obtained through clustering is more accurate.

Description

Technical Field

[0001] This application relates to the technical field of power grids, and in particular to a user portrait analysis method, system and medium based on a big data model. Background Art

[0002] User portrait analysis relies on user voltage data, constructs comprehensive tags to accurately describe user voltage characteristics, and supports the refined management and optimized services of power grid companies based on this. Under the background of the power market reform, carrying out user behavior portrait analysis of power users can not only help power companies solve voltage quality problems and improve user experience, but also improve the overall operation efficiency of the power grid. At present, with the growth of the power grid intelligence level and the popularization of intelligent metering terminal equipment, a large amount of user-side voltage-related data has been brought to power enterprises, providing a solid data foundation for carrying out user portrait analysis. Therefore, a large number of researchers have carried out relevant research on power user portrait analysis based on data mining technology in combination with user-side voltage-related data, including feature analysis, clustering analysis, etc.

[0003] The current user portrait analysis methods mainly establish a tag system indicating user-side voltage characteristics through big data analysis methods such as k-means clustering, and analyze the voltage quality level of each type of user according to the user voltage change curve, which can realize the corresponding power user portrait, but there are still some deficiencies. One is that many studies mostly portrait the voltages of different users as a whole, and few studies decompose the user voltage, strip out the basic voltage and sensitive voltage of the user, and then analyze the basic voltage situation of the user in different seasons and the voltage fluctuation situation under different influences. The other is the lack of predictive analysis of the "future state" user voltage, unable to predict the voltage change trend of users in the target area in the future time period, and it is difficult to judge in advance the possible voltage quality problems. Summary of the Invention

[0004] The purpose of the embodiments of this application is to provide a user portrait analysis method, system and medium based on a big data model, which decomposes the user voltage into basic voltage and sensitive voltage according to seasons, and conducts clustering analysis respectively, and the portrait analysis results obtained by clustering are more accurate.

[0005] To achieve the above object, this application provides the following technical solutions:

[0006] In the first aspect, the embodiments of this application provide a user portrait analysis method based on a big data model, including the following steps:

[0007] Collect voltage data from the data center and user complaint record channels, and at the same time integrate user basic information and power grid topology structure data to form a comprehensive database;

[0008] Build a user voltage prediction model to achieve the prediction of the voltage values of all users in the target area at future moments;

[0009] Classify the voltage data of users according to the four seasons in a year;

[0010] Establish the correlation relationship between multi-dimensional correlation factors such as temperature change, weekdays and weekends and user voltage, and obtain the influence weight coefficients of each factor;

[0011] k-means clustering analysis;

[0012] Based on the above analysis results, generate a detailed voltage profile for each user or area;

[0013] According to the voltage profile, power grid enterprises take targeted measures to improve the user voltage quality and power supply reliability;

[0014] Through continuous data collection and analysis, the user voltage profile is updated in real time.

[0015] The specific construction of the user voltage prediction model is as follows:

[0016] Assume that there are X users in the target area, and use u x (t) to represent the historical voltage data of the x-th user;

[0017] Combined with the long short-term memory network and the historical voltage data of the x-th user, predict the voltage change trend at its future moment, and the prediction result is represented by u x '(t);

[0018]

[0019] In the above formula (1), f x (t), i x (t), c x (t) represent the forget gate, input gate and output gate respectively, h x (t) represents the t-th output, h x (t - 1) represents the (t - 1)-th output, c x (t) represents the cell state, represents the candidate cell state, w and b represent the weight vector and bias vector, and σ and tanh represent the activation functions.

[0020] The specific classification of the voltage data of users according to the four seasons in a year is as follows:

[0021] Assume that for a certain user x in the target area in the i-th season, there are m + 1 days in this season, then the voltage set U of this user in this season xi can be represented by the following matrix:

[0022]

[0023] In the above formula, is the voltage value of user x at the h-th hour on the d-th day in the i-th season;

[0024] Assume that the daily basic voltage of the x-th user in the i-th season is

[0025]

[0026] Taking the shortest distance of all coordinates between curves as the objective function:

[0027]

[0028] In the formula: S xi is the objective function value, and m is the number of days in the i-th season;

[0029] Meanwhile, the values at each moment cannot exceed the upper and lower limits of the corresponding moment:

[0030]

[0031] Combining formula (4) and formula (5), using optimization algorithms such as ant colony optimization / particle swarm optimization / simulated annealing to solve the objective function, and obtaining the daily basic voltage value;

[0032] Performing a difference operation on U xi and to obtain the daily voltage affected by other factors. Specifically as follows:

[0033]

[0034] The specific k-means clustering analysis is as follows:

[0035] Seasonal daily basic voltage clustering analysis

[0036] a. Calculate the 24-hour daily average basic voltage of the x-th user in the target area in the same season

[0037] In the above formula, represents the daily average basic voltage of user x at the i-th hour (i ∈ {1, 2,..., 24}) in a day,

[0038] b. Repeat the above step a for all users to obtain the 24-hour daily average basic voltage of all users in the target area in the same season, and establish X 24-dimensional row vectors based on this,

[0039] c. Randomly select K row vectors from the row vectors as the initial cluster centers.

[0040] d. Divide each row vector into the cluster represented by the row vector closest to it.

[0041] e. Replace the original central row vector with the central row vector of all samples in each cluster.

[0042] f. Repeat steps d and e until the central row vector remains unchanged or reaches the predetermined number of iterations, then the algorithm terminates.

[0043] The portrait for generating a detailed voltage portrait for each user or region includes a voltage quality profile, a problem hot spot area, a sensitive user identifier, a future trend prediction, and improvement suggestions, providing an intuitive basis for power grid planning and operation and maintenance decision-making.

[0044] In a second aspect, an embodiment of the present application provides a user portrait analysis system based on a big data model, including a memory and a processor. The memory includes a program of the user portrait analysis method based on the big data model. When the program of the user portrait analysis method based on the big data model is executed by the processor, the following steps are implemented: Collect voltage data from the data center and the user complaint record channel, and at the same time integrate the user's basic information and the power grid topology structure data to form a comprehensive database; Build a user voltage prediction model to realize the prediction of the voltage values of all users in the target area at future moments; Classify the voltage data of users in the four seasons of a year; Establish the correlation relationship between the multi-dimensional correlation factors of temperature change, weekdays, and weekends and the user's voltage to obtain the influence weight coefficients of each factor; k-means clustering analysis; Based on the above analysis results, generate a detailed voltage portrait for each user or region; According to the voltage portrait, the power grid enterprise takes targeted measures to improve the user voltage quality and power supply reliability; Through continuous data collection and analysis, the user voltage portrait is updated in real time.

[0045] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the user portrait analysis method based on the big data model as described above are implemented.

[0046] Compared with the prior art, the beneficial effects of the present invention are:

[0047] 1. The prediction of the future user voltage change trend is realized by combining the long short-term memory network.

[0048] 2. Decompose the user's voltage into the basic voltage and the sensitive voltage affected by other factors according to the different seasonal voltages of the user, and conduct clustering analysis separately. The final comprehensive user portrait can not only reflect the changes in the basic voltage of the user in different seasons, but also reflect the voltage sensitivity of the user affected by different sensitive factors.

[0049] 3. When extracting the basic voltage, a data method is introduced for decomposition, which also ensures the objectivity of the basic voltage extraction.

[0050] 4. Decompose the user's voltage into the basic voltage and the sensitive voltage according to seasons, and conduct clustering analysis separately. The analysis results of the portraits obtained by clustering are more accurate. Description of the Drawings

[0051] To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can be obtained based on these drawings without creative efforts.

[0052] Figure 1 It is a flowchart of a user portrait analysis method based on a big data model provided by the present application. Detailed Embodiments

[0053] The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application. It should be noted that: similar reference numerals and letters denote similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0054] The term "comprising", "including" or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "comprising a..." does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0055] Terms such as "first", "second", etc. are only used to distinguish one entity or operation from another entity or operation, and cannot be understood as indicating or implying relative importance, nor can they be understood as requiring or implying any actual relationship or order between these entities or operations.

[0056] Such as Figure 1As shown in the figure, a user portrait analysis method based on a big data model provided by the present invention includes the following steps:

[0057] 1. Collect voltage data from channels such as the data middle platform and user complaint records, including real-time voltage values, voltage fluctuation frequencies, voltage qualification rates, voltage anomaly event records, etc., and perform data cleaning. At the same time, integrate user basic information (such as geographical location, user type, load characteristics) and power grid topology structure data to form a comprehensive database.

[0058] 2. "Future state" user voltage prediction.

[0059] (1) Assume that there are X users in the target area, and use u x (t) to represent the historical voltage data of the xth user.

[0060] (2) Combine the long short-term memory network with the historical voltage data of the xth user to predict the voltage change trend at its future moment, and the prediction result is represented by u x '(t).

[0061] f x (t) = σ(w x (t) * [h x (t - 1), u x (t)] + b f )

[0062] i x (t) = σ(w i * [h x (t - 1), u x (t)] + b i )

[0063]

[0064] o x (t) = σ(w o * [h x (t - 1), u x (t)] + b o )

[0065]

[0066] In the above formula (1), f x (t), i x (t), c x (t) respectively represent the forgetting gate, input gate and output gate, h x (t) represents the tth output, h x (t - 1) represents the (t - 1)th output, c x (t) represents the cell state, represents the candidate cell state, w and b represent the weight vector and bias vector, and σ and tanh represent the activation function.

[0067] In this step, a five-layer LSTM neural network model is constructed. The first layer of this model is an LSTM layer, equipped with 100 neurons; followed by the second layer, the Dropout layer, whose parameter is set to 0.3; the third layer is another LSTM layer with 120 neurons; followed by the fourth layer, also a Dropout layer, with the parameter also set to 0.3; the final fifth layer is a Dense layer, which is only configured with one neuron because the output requirement is only a single dimension.

[0068] In terms of data input, a loop structure is used to process sequence data. Specifically, the first 100 data points of each subsequence are used as the first set of input data, and the 101th data point is used as the corresponding label for this set of input. Then, the second set of input data consists of the 2nd to 101th data points, and its label is the 102th data point. This process continues until the last input group, which consists of the T-100th to T-1th data points, and its label is the Tth data point. When the loop completes all these steps, the number of iterations of the model is set to 500.

[0069] The above parameter settings are used to train the LSTM model. Input a subsequence u x (t), we can automatically get a subsequence u x The prediction result u of (t) x '(t).

[0070] This can be deduced to other users in the same way to achieve the prediction of voltage values ​​of all users in the target area at future times.

[0071] 3. Data Decomposition

[0072] (1) Classify users’ voltage data according to the four seasons of the year.

[0073] (2) Assume that a user x in the target area is in the i-th season, and there are m+1 days in this season. Then the voltage set U of this user in this season is xi It can be represented by the following matrix:

[0074]

[0075] In the above formula, is the voltage value of user x at the hth hour on the dth day in the i-th season.

[0076] Assume that the base voltage of the xth user on the ith day of the season is

[0077]

[0078] Taking the shortest coordinate distance between curves as the objective function:

[0079]

[0080] Where: S xi is the objective function value, and m is the number of days in the i-th season.

[0081] Meanwhile, the values at each moment cannot exceed the upper and lower limits corresponding to that moment:

[0082]

[0083] Combining Equation (4) and Equation (5), using optimization algorithms such as ant colony optimization / particle swarm optimization / simulated annealing to solve the objective function, and obtaining the basic daily voltage of the season value.

[0084] Performing a difference operation on U xi and to obtain the daily voltage affected by other factors Specifically as follows:

[0085]

[0086] 4. Correlation analysis

[0087] Combining the Pearson product-moment correlation coefficient, establishing the correlation relationship between multi-dimensional correlation factors such as temperature change, weekdays and weekends and the user voltage, obtaining the influence weight coefficients of each factor, and using this to support the subsequent clustering analysis.

[0088] 5. k-means clustering analysis

[0089] (1) Seasonal daily basic voltage clustering analysis

[0090] a. Calculate the 24-point daily average basic voltage of the x-th user in the target area in the same season

[0091] In the above formula, represents the daily average basic voltage of user x at the i-th hour (i ∈ {1, 2,..., 24}) of a day.

[0092] b. Repeat the above step a for all users to obtain the 24-point daily average basic voltage of all users in the target area in the same season, and use this to establish X 24-dimensional row vectors.

[0093] c. Randomly select K row vectors from the row vectors as the initial cluster centers.

[0094] d. Divide each row vector into the cluster represented by the row vector closest to it.

[0095] e. Replace the original central row vector with the central row vector of all samples in each cluster.

[0096] f. Repeat steps d and e until the central row vector remains unchanged or reaches a predetermined number of iterations, at which point the algorithm terminates.

[0097] (2) Voltage clustering analysis affected by other factors

[0098] The steps are similar to the seasonal daily-based voltage clustering analysis. It should be noted that when establishing the row vector, multi-dimensional correlation factors such as temperature changes, weekdays, and weekends need to be considered.

[0099] 6. Voltage portrait generation: Based on the above analysis results, generate a detailed voltage portrait for each user or area. The portrait includes the voltage quality profile, problem hotspots, sensitive user identification, future trend prediction, and improvement suggestions, providing an intuitive basis for power grid planning and operation and maintenance decisions.

[0100] 7. Application and optimization: Based on the voltage portrait, power grid enterprises can take targeted measures, such as adjusting the reactive power compensation strategy, optimizing the power grid layout, and implementing dynamic voltage control, to improve the user voltage quality and power supply reliability.

[0101] 8. Real-time monitoring and dynamic update: Through continuous data collection and analysis, the user voltage portrait is updated in real time to ensure that the analysis results keep up with the actual situation and effectively respond to new changes in power grid operation.

[0102] The embodiment of this application provides a user portrait analysis system based on a big data model, including a memory and a processor. The memory includes a program of the user portrait analysis method based on the big data model. When the program of the user portrait analysis method based on the big data model is executed by the processor, the following steps are implemented: collect voltage data from the data center and the user complaint record channel, and at the same time integrate the user basic information and the power grid topology structure data to form a comprehensive database; construct a user voltage prediction model to realize the prediction of the voltage values of all users in the target area at future moments; classify the voltage data of users in the four seasons of a year; establish the correlation relationship between multi-dimensional correlation factors such as temperature changes, weekdays, and weekends and the user voltage to obtain the influence weight coefficients of each factor; k-means clustering analysis; based on the above analysis results, generate a detailed voltage portrait for each user or area; based on the voltage portrait, power grid enterprises take targeted measures to improve the user voltage quality and power supply reliability; through continuous data collection and analysis, the user voltage portrait is updated in real time.

[0103] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the user portrait analysis method based on a big data model as described above are implemented.

[0104] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0106] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that implements the specified functions in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0107] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the specified functions in Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0108] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0109] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM) and / or non-volatile memory such as read-only memory (ROM) or flash RAM. The memory is an example of computer-readable media.

[0110] Computer-readable media includes both permanent and non-permanent, removable and non-removable media and can store information by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile discs (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0111] The above are only examples of the embodiments of the present application and are not intended to limit the scope of protection of the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for user portrait analysis based on a big data model, characterized in that, It includes the following steps: Collect voltage data from the data center and user complaint record channels, and at the same time integrate user basic information and power grid topology structure data to form a comprehensive database; Build a user voltage prediction model to achieve the prediction of voltage values of all users in the target area at future times; Classify the voltage data of users according to the four seasons in a year; Establish the correlation relationship between multi-dimensional correlation factors such as temperature change, weekdays and weekends and user voltage, and obtain the influence weight coefficients of each factor; k-means clustering analysis; Based on the above analysis results, generate a detailed voltage profile for each user or area; According to the voltage profile, the power grid enterprise takes targeted measures to improve the user voltage quality and power supply reliability; Through continuous data collection and analysis, the user voltage profile is updated in real time.

2. The user portrait analysis method based on a big data model according to claim 1, characterized in that The specific construction of the user voltage prediction model is as follows: Assume there are X users in the target area, and use u x (t) to represent the historical voltage data of the x-th user; Combined with the historical voltage data of the x-th user, the long short-term memory network is used to predict the future voltage change trend, and the prediction result is represented by u x '(t), In the above formula (1), f x (t), i x (t), c x (t) represent the forget gate, input gate, and output gate respectively, h x (t) represents the t-th output, h x (t - 1) represents the (t - 1)-th output, c x (t) represents the cell state, represents the candidate cell state, w and b represent the weight vector and bias vector, and σ and tanh represent the activation functions.

3. The method for analyzing user portraits based on a big data model according to claim 1, characterized in that The specific classification of the voltage data of users according to the four seasons in a year is as follows: Suppose that for a user x in a target area in the i-th season, there are m + 1 days in this season, then the voltage set U of this user in this season xi can be represented by the following matrix: In the above formula, is the voltage value of user x at the h-th hour on the d-th day in the i-th season; Assume that the base voltage of the x-th user on the i-th seasonal day is Take the shortest distance between all coordinates of the curves as the objective function: Where: S xi is the objective function value, and m is the number of days in the i-th season; Meanwhile, the value at each moment cannot exceed the upper and lower limits corresponding to that moment: Combining formula (4) and formula (5), using optimization algorithms such as ant colony optimization / particle swarm optimization / simulated annealing to solve the objective function, the seasonal daily base voltage value is obtained; For U xi and perform a difference operation to obtain the daily voltage affected by other factors Specifically as follows:

4. The user portrait analysis method based on a big data model according to claim 1, characterized in that The specific k-means clustering analysis is as follows: Seasonal daily base voltage clustering analysis a. Calculate the 24 - hour daily average basic voltage of the x - th user in the target area in the same season In the above formula, represents the daily average basic voltage of user x at the i-th hour of a day (i ∈ {1, 2,..., 24}), b. Repeat the above step a for all users to obtain the 24-point daily average basic voltage of all users in the target area in the same season, and establish X 24-dimensional row vectors based on this; c. Randomly select K row vectors from the row vectors as the initial cluster centers; d. Divide each row vector into the cluster represented by the row vector closest to it; e. Replace the original central row vector with the central row vector of all samples in each cluster; f. Repeat steps d and e until the central row vector remains unchanged or reaches the predetermined number of iterations, and the algorithm terminates.

5. A user portrait analysis method based on a big data model according to claim 1, characterized in that The profile generated for each user or area includes a voltage quality overview, problem hot spots, sensitive user identification, future trend prediction and improvement suggestions, providing an intuitive basis for power grid planning and operation and maintenance decision-making.

6. A user portrait analysis system based on a big data model, characterized in that, It includes a memory and a processor. The memory includes a program of the user profile analysis method based on the big data model. When the program of the user profile analysis method based on the big data model is executed by the processor, the following steps are implemented: collect voltage data from the data center and user complaint record channels, and at the same time integrate user basic information and power grid topology structure data to form a comprehensive database; build a user voltage prediction model to achieve the prediction of voltage values of all users in the target area at future times; Classify the voltage data of users according to the four seasons in a year; establish the correlation relationship between multi-dimensional correlation factors such as temperature change, weekdays and weekends and user voltage, and obtain the influence weight coefficients of each factor; k-means clustering analysis; based on the above analysis results, generate a detailed voltage profile for each user or area; According to the voltage profile, the power grid enterprise takes targeted measures to improve the user voltage quality and power supply reliability; Through continuous data collection and analysis, the user voltage profile is updated in real time.

7. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the user profile analysis method based on the big data model according to any one of claims 1-5 are implemented.

Images