A multi-channel data fusion method and system

Abstract

The present invention provides a multi-channel data fusion method and system, which relates to data processing technologies. The method is based on a data acquisition unit to capture data from data sources corresponding to each user terminal, obtain acquisition data of multiple dimension features corresponding to each user terminal, construct an initial data column corresponding to each user terminal, perform arrangement and update on the initial data column according to the acquisition data corresponding to each dimension feature to obtain a fusion data column, acquire fusion metrics of multiple dimension features corresponding to the user terminal in the fusion data column, calculate user portrait data corresponding to each user terminal according to the fusion metrics, generate push data corresponding to each user terminal according to the user portrait data, and send the push data to the corresponding user terminal.

Description

Technical Field

[0001] The present invention relates to data processing technologies, and in particular, to a multi-channel data fusion method and system. Background Art

[0002] In today's information society, a large amount of power data is generated during the operation of enterprises. These power data are distributed in different databases, forming a complex power data ecosystem.

[0003] Currently, due to the differences and large volumes of power data of different enterprises, when there is a recommendation requirement for power data, such as a recommendation requirement for power equipment, it is often necessary to traverse a large amount of data of different enterprises to generate user portraits, and then perform data recommendations based on the generated user portraits. This method greatly reduces the efficiency of data recommendation.

[0004] Therefore, how to quickly generate user portraits through the power data of different enterprises and perform customized data recommendations in combination with the user portraits has become an urgent problem to be solved today. Summary of the Invention

[0005] The present invention provides a multi-channel data fusion method and system, which can quickly generate user portraits through the power data of different enterprises and perform customized data recommendations in combination with the user portraits.

[0006] In the first aspect of the present invention, a multi-channel data fusion method is provided, including:

[0007] Performing data scraping from the data sources corresponding to each user terminal based on a data acquisition unit to obtain acquisition data of multiple dimensional features corresponding to each user terminal;

[0008] Constructing an initial data column corresponding to each user terminal, and arranging and updating the initial data column according to the acquisition data corresponding to each dimensional feature to obtain a fusion data column;

[0009] Obtaining fusion metrics of multiple dimensional features corresponding to the user terminal in the fusion data column, and calculating user portrait data corresponding to each user terminal according to the fusion metrics;

[0010] Generating push data corresponding to each user terminal according to the user portrait data, and sending the push data to the corresponding user terminal.

[0011] Optionally, in a possible implementation manner of the first aspect, constructing an initial data column corresponding to each user terminal, and arranging and updating the initial data column according to the acquisition data corresponding to each dimensional feature to obtain a fusion data column includes:

[0012] Count the number of users of the client, determine the length of the data column corresponding to the number of users, and construct an initial data axis according to the length of the data column;

[0013] Construct a corresponding number of data slots on the initial data axis according to the number of users, and bind the collected data corresponding to each data slot and the corresponding client to obtain an initial data column;

[0014] Generate a dimension display table based on the dimension features corresponding to each client and send it to the push management end, and obtain the target dimension selected by the push management end based on the dimension display table;

[0015] Obtain the data similarity coefficients of each client according to the collected data corresponding to the target dimension, and perform permutation and update on each data slot in the initial data column based on the data similarity coefficients to obtain a fused data column.

[0016] Optionally, in a possible implementation manner of the first aspect, obtaining the data similarity coefficients of each client according to the collected data corresponding to the target dimension, and performing permutation and update on each data slot in the initial data column based on the data similarity coefficients to obtain a fused data column includes:

[0017] Parse the collected data to obtain the power consumption corresponding to each target dimension, and count the power consumption corresponding to multiple target dimensions of each client to obtain the total power consumption of each client;

[0018] Determine the client with the least total power consumption as the target user, and obtain the data similarity coefficients between the remaining clients and the target user according to the power consumption corresponding to each target dimension;

[0019] Obtain the client with the largest data similarity coefficient as the next target user, and repeat the steps of obtaining the data similarity coefficient and the target user until the next target user is the last client;

[0020] Arrange and update each data slot according to the determination order of each target user to obtain the fused data column.

[0021] Optionally, in a possible implementation manner of the first aspect, obtaining the data similarity coefficients between the remaining clients and the target user according to the power consumption corresponding to each target dimension includes:

[0022] Obtain the power consumption differences of each target dimension corresponding to the target user and the remaining clients;

[0023] Retrieve the similarity adjustment weights corresponding to each of the target dimensions, and obtain the sub-similarity coefficients corresponding to each of the target dimensions according to the product of the reciprocal of the power consumption difference and the similarity adjustment weights;

[0024] Statistically analyze the sub-similarity coefficients corresponding to each of the target dimensions to obtain the data similarity coefficient between the target user and each of the other user terminals.

[0025] Optionally, in a possible implementation manner of the first aspect, after arranging and updating each of the data slots according to the determination order of each of the target users to obtain the fusion data column, it further includes:

[0026] Receive the gap display mode selected by the push management terminal, where the gap display mode includes a slot display mode and a spacing display mode;

[0027] Obtain the number of interval slots between adjacent data slots in the fusion data column according to the slot display mode, and insert a corresponding number of interval slots between adjacent data slots to obtain an adjusted data column; or,

[0028] Obtain the similarity difference between adjacent data slots in the fusion data column based on the spacing display mode, and adjust the spacing between adjacent data slots according to the similarity difference to obtain an adjusted data column.

[0029] Optionally, in a possible implementation manner of the first aspect, obtaining the number of interval slots between adjacent data slots in the fusion data column according to the slot display mode, and inserting a corresponding number of interval slots between adjacent data slots to obtain an adjusted data column includes:

[0030] Determine that the user terminals corresponding to adjacent data slots in the fusion data column are in the same user group, and obtain the similarity difference between the data similarity coefficients corresponding to the user group;

[0031] Compare the similarity difference with the reference coefficient difference, and determine the user group with the similarity difference greater than the reference coefficient difference as the adjustment group;

[0032] Determine the number of interval slots for the adjustment group pair according to the ratio of the similarity difference to the reference coefficient difference;

[0033] Insert the number of interval slots between the data slots corresponding to the adjustment group to obtain an adjusted data column.

[0034] Optionally, in a possible implementation manner of the first aspect, after inserting the number of interval slots between the data slots corresponding to the adjustment group to obtain an adjusted data column, it further includes:

[0035] Obtain the slot pitch between adjacent spaced slots, and determine the spaced slots with the slot pitch less than the reference gap pitch as the target slots;

[0036] Among them, the slot pitches between multiple adjacent ones of the spaced slots corresponding to the same adjustment group are equal;

[0037] Calculate the spacing difference between the slot pitch and the reference gap pitch, and obtain the size offset coefficient according to the ratio of the spacing difference to the reference spacing difference;

[0038] Obtain the size adjustment value according to the product of the size offset coefficient and the size adjustment weight of the spaced slot, and obtain the adjusted size based on the difference between the standard size of the target slot and the size adjustment value;

[0039] Based on the adjusted size, perform a reduction adjustment on the slot size of the target slot to obtain an updated adjusted data column.

[0040] Optionally, in a possible implementation manner of the first aspect, obtain the similarity difference between adjacent data slots in the fusion data column based on the spacing display mode, and adjust the spacing between adjacent data slots according to the similarity difference to obtain an adjusted data column, including:

[0041] Obtain the similarity difference between the corresponding data similarity coefficients of adjacent data slots in the fusion data column;

[0042] Obtain the spacing adjustment coefficient according to the ratio of the similarity difference to the reference coefficient difference, and obtain the adjusted spacing corresponding to the adjacent data slots based on the product of the spacing adjustment coefficient and the standard spacing;

[0043] Based on the adjusted spacing, adjust the spacing between adjacent data slots to obtain an adjusted data column.

[0044] Optionally, in a possible implementation manner of the first aspect, obtain the fusion index of multiple dimensional features corresponding to the user end in the fusion data column, and calculate the user portrait data corresponding to each user end according to the fusion index, including:

[0045] Respond to the portrait prediction request of the push management end, and screen the multiple user ends corresponding to the fusion data column to obtain reference users;

[0046] Obtain the fusion index of the multiple dimensional features corresponding to the reference users, and the fusion index at least includes the power consumption of the corresponding dimensional feature;

[0047] Calculate the reference portrait data corresponding to each reference user according to the fusion index corresponding to each reference user;

[0048] Obtain the data difference coefficients between each of the remaining client devices and the reference user, and calculate the predicted portrait data corresponding to each of the remaining client devices based on the reference portrait data and the data difference coefficients. The user portrait data includes reference portrait data and predicted portrait data.

[0049] Optionally, in a possible implementation manner of the first aspect, in response to the portrait prediction request of the push management end, screening the multiple client devices corresponding to the fusion data column to obtain a reference user includes:

[0050] In response to the portrait prediction request, obtain the spacing calculation amount according to the sum of the preset reference quantity and the spacing offset value;

[0051] Obtain the data column length of the fusion data column, and obtain the unit spacing according to the ratio of the data column length to the spacing calculation amount;

[0052] Determine the preset number of reference points in the fusion data column according to the unit spacing, obtain the reference slot corresponding to the data slot closest to the reference point, and determine the client device corresponding to the reference slot as the reference user.

[0053] Optionally, in a possible implementation manner of the first aspect, calculating the reference portrait data corresponding to each of the reference users according to the fusion indexes corresponding to each of the reference users includes:

[0054] Determine the benchmark index corresponding to each dimension feature according to the mean value of the fusion indexes of each dimension feature corresponding to each of the reference users;

[0055] Compare the fusion indexes of each dimension feature corresponding to each of the reference users with the benchmark index, determine that the dimension feature with the fusion index greater than the benchmark index corresponds to the high energy consumption attribute, and determine that the dimension feature with the fusion index less than the benchmark index corresponds to the low energy consumption attribute;

[0056] Determine the energy consumption demand corresponding to the power consumption of the fusion index according to the high energy consumption attribute or the low energy consumption attribute, and obtain the reference portrait data of each of the reference users based on the energy consumption demands corresponding to each dimension feature.

[0057] Optionally, in a possible implementation manner of the first aspect, obtaining the data difference coefficients between each of the remaining client devices and the reference user, and calculating the predicted portrait data corresponding to each of the remaining client devices based on the reference portrait data and the data difference coefficients includes:

[0058] Determine that the fusion data column with an interval slot is a type of data column, and obtain the reference user closest in distance as the type of user corresponding to each of the remaining corresponding client devices;

[0059] Obtain the number of slot positions of the interval slot between the client and the first type of user, obtain a first type of offset coefficient according to the ratio of the number of slot positions to the reference number of slot positions, and obtain the data discrimination coefficient of the client based on the product of the first type of offset coefficient and the first type of weight;

[0060] Determine that the fusion data column without an interval slot is a second type of data column, and obtain the second type of user corresponding to the remaining corresponding clients for the reference user with the closest distance;

[0061] Obtain the interval distance between the client and the second type of user, obtain a second type of offset coefficient according to the ratio of the interval distance to the reference interval distance, and obtain the data discrimination coefficient of the client based on the product of the second type of offset coefficient and the second type of weight;

[0062] Obtain the adjustment demand of each dimension feature corresponding to the corresponding client according to the product of the data discrimination coefficient and the energy consumption demand of each dimension feature corresponding to the first type of user or the second type of user;

[0063] Determine the adjustment trend of each client, determine the predicted demand of each client according to the adjustment trend and the adjustment demand, and obtain the predicted portrait data corresponding to the client based on the predicted demand corresponding to each dimension feature.

[0064] Optionally, in a possible implementation manner of the first aspect, determining the adjustment trend of each client and determining the predicted demand of each client according to the adjustment trend and the adjustment demand includes:

[0065] Obtain the slot side of the data slot corresponding to the first type of user or the second type of user in the data slot of the client, where the slot side includes a decreasing side or an increasing side;

[0066] Determine that the dimension feature corresponding to the decreasing side is a decreasing trend, and determine that the dimension feature corresponding to the increasing side is an increasing trend. The adjustment trend includes a decreasing trend and an increasing trend;

[0067] Obtain the energy consumption demand of the decreasing trend corresponding to the reference user as the first type of demand, and obtain the predicted demand corresponding to the decreasing trend according to the difference between the first type of demand and the adjustment demand;

[0068] Determine the energy consumption demand of the increasing trend corresponding to the reference user as the second type of demand, and obtain the predicted demand corresponding to the increasing trend according to the sum of the second type of demand and the adjustment demand.

[0069] Optionally, in a possible implementation manner of the first aspect, generating the push data corresponding to each client according to the user portrait data and sending the push data to the corresponding client includes:

[0070] Determine the high-demand devices corresponding to the dimensional features of the high energy consumption attributes, and determine the low-demand devices corresponding to the dimensional features of the low energy consumption attributes;

[0071] Determine the device capacity corresponding to the high-demand device or low-demand device according to the energy consumption demand, and generate sub-push data corresponding to the dimensional features based on the high-demand device or low-demand device and the device capacity;

[0072] Generate push data corresponding to each client based on the sub-push data, and send the push data to the corresponding client.

[0073] In a second aspect of the present invention, a multi-channel data fusion system is provided, including:

[0074] A scraping module, configured to scrape data from data sources corresponding to each client based on a data collection unit to obtain collection data of multiple dimensional features corresponding to each client;

[0075] A construction module, configured to construct an initial data column corresponding to each client, and arrange and update the initial data column according to the collection data corresponding to each dimensional feature to obtain a fusion data column;

[0076] An acquisition module, configured to acquire fusion metrics of multiple dimensional features corresponding to the client in the fusion data column, and calculate user portrait data corresponding to each client according to the fusion metrics;

[0077] A generation module, configured to generate push data corresponding to each client according to the user portrait data, and send the push data to the corresponding client.

[0078] The beneficial effects of the present invention are as follows:

[0079] 1. The present invention can converge and synchronize the power data of multiple enterprises. By efficiently fusing the power data of multiple enterprises, a unified and easily accessible data column is formed, thereby realizing the sharing of business data, improving the efficiency during data access, and constructing a unique user portrait for each user according to the user portrait construction technology. Corresponding data is pushed in combination with the user portrait data corresponding to the client, thereby realizing customized recommendations for the client, and when making recommendations, the actual electricity consumption needs of the client can be fully combined to recommend more suitable devices for it, which can ensure the accuracy during recommendations.

[0080] 2. The present invention can construct initial data columns corresponding to each client, and can fuse the data of multiple enterprises in the form of data columns. According to the fused data columns, different-dimensional data of multiple enterprises can be quickly viewed, so as to realize the quick access and analysis of multiple data sources corresponding to multiple enterprises, and can improve the efficiency when viewing data. When the present invention fuses the data columns corresponding to multiple enterprise users, it can arrange and update multiple initial data columns according to the similarity between the collected data with different-dimensional characteristics corresponding to each client, so that users with higher data similarity are arranged together, which can improve the efficiency of data fusion.

[0081] 3. After obtaining the fused data columns, the present invention can make corresponding adjustments to the fused data columns in combination with the data similarity coefficient, so as to display the similarity gap between adjacent data slots according to the data similarity coefficient, so that the adjusted fused data columns can more intuitively display the similarity difference between adjacent data slots, and can effectively reflect the similarity and difference between data. When the present invention performs data push according to user portrait data, it can screen from multiple clients, and predict the portrait data of the remaining each client through the portrait data corresponding to the screened client, which can reduce the data processing volume and improve the data processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 is a schematic flowchart of a multi-channel data fusion method provided by an embodiment of the present invention;

[0083] Figure 2 is a schematic diagram of determining an adjustment data column in a slot display mode provided by an embodiment of the present invention;

[0084] Figure 3 is a schematic diagram of determining an adjustment data column in a spacing display mode provided by an embodiment of the present invention;

[0085] Figure 4 is a schematic structural diagram of a multi-channel data fusion system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0086] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0087] The technical solution of the present invention will be described in detail below with specific embodiments. These several specific embodiments below can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0088] See Figure 1 , which is a schematic diagram of a multi-channel data fusion method provided by an embodiment of the present invention. Figure 2 The execution subject of the method shown can be a software and / or hardware device. The execution subject of the present application may include, but is not limited to, at least one of the following: user equipment, network equipment, etc. Among them, the user equipment may include, but is not limited to, a computer, a smart phone, a personal digital assistant (Personal Digital Assistant, abbreviated as: PDA), and the above-mentioned electronic devices, etc. The network equipment may include, but is not limited to, a single network server, a server group composed of multiple network servers, or a cloud composed of a large number of computers or network servers based on cloud computing. Among them, cloud computing is a type of distributed computing, which consists of a group of loosely coupled computers forming a super virtual computer. This embodiment does not make any restrictions on this. It includes steps S1 to S4, specifically as follows:

[0089] S1, based on the data acquisition unit, grab data from the data sources corresponding to each user terminal, and obtain the acquisition data of multiple dimensional features corresponding to each user terminal.

[0090] Among them, the data acquisition unit refers to a unit that can grab data. The user terminal refers to the terminal corresponding to an enterprise, such as a computer corresponding to an electric power company, etc. The data source refers to the database that stores various types of data in the enterprise terminal. Each enterprise may have multiple databases. The dimensional feature refers to different attributes that can be used to describe the electricity consumption characteristics of the user terminal, such as residential users, commercial users, etc. The acquisition data refers to the data grabbed from multiple data sources of the enterprise.

[0091] It can be understood that in the current field of electric power data management and application, many comprehensive enterprises each have a huge data resource system. These data resources are scattered and stored in multiple databases, covering data in multiple dimensions such as residential, commercial, and industrial. However, when it is necessary to conduct data access and analysis across enterprises and databases, the operation may be relatively complex and the efficiency is relatively low.

[0092] This solution can converge and synchronize the electric power data of multiple enterprises. By efficiently fusing the electric power data of multiple enterprises, a unified and easily accessible data column is formed, thereby realizing the sharing of business data. And according to the user portrait construction technology, a unique user portrait can be constructed for each enterprise. According to the user portrait of the enterprise, the actual situation and electricity consumption needs of the enterprise can be fully combined, so that the most suitable power generation equipment can be recommended for it, and the customization of the recommendation can be ensured.

[0093] Specifically, the data acquisition unit can access multiple data sources corresponding to each client and extract the required data from them. Since each data source contains data in different dimensions, the data acquisition unit can collect data in multiple dimensions corresponding to each client, obtaining acquisition data corresponding to multiple dimensional features. For example, for residential users, the data acquisition unit can extract the household electricity consumption data corresponding to the residential dimension, and for commercial users, it can extract the commercial electricity consumption data corresponding to the commercial dimension, etc.

[0094] S2. Construct initial data columns corresponding to each of the clients, and perform permutation and update on the initial data columns according to the acquisition data corresponding to each of the dimensional features to obtain fused data columns.

[0095] It can be understood that, in order to quickly access and analyze multiple data sources corresponding to multiple enterprises, initial data columns corresponding to each client can be constructed, and the data of multiple enterprises can be fused in the form of data columns. According to the fused data columns, different dimensional data of multiple enterprises can be quickly viewed, which can improve the efficiency when viewing data.

[0096] In practical applications, since the higher the similarity between data, the higher the possible correlation between data, in order to arrange users with relatively high data similarity together, which is convenient for subsequent user profiling and data recommendation, the initial data columns can be permuted and updated according to the acquisition data of different dimensional features corresponding to each client, so that users with relatively high data similarity are arranged together, obtaining fused data columns containing acquisition data of multiple enterprises.

[0097] Among them, the initial data column refers to the data column in the initial state corresponding to each client, and the fused data column refers to the data column obtained after permutation and update of the initial data column.

[0098] In some embodiments, step S2 includes S21 to S24, which are specifically as follows:

[0099] S21. Count the number of users of the client, determine the data column length corresponding to the number of users, and construct an initial data axis according to the data column length.

[0100] Specifically, the number of all clients can be counted, that is, the number of users. According to the number of users, the corresponding data column length can be determined. Different numbers of users correspond to different data column lengths. The more the number of users, the longer the data column length can be. According to the data column length, a data axis with the corresponding length can be generated, that is, the initial data axis.

[0101] Among them, the number of users refers to the number of user terminals, the data column length refers to the length of the data column corresponding to the number of users, and the initial data axis refers to a basic data representation axis set according to the number of users.

[0102] S22. Construct a corresponding number of data slots on the initial data axis according to the number of users, and bind the collected data corresponding to each data slot and the corresponding user terminal to obtain an initial data column.

[0103] Specifically, data slots equal to the number of users can be constructed on the initial data axis, and the interval distance between adjacent data slots can be equal. The collected data corresponding to each data slot and each user terminal can be bound in the form of a list to obtain an initial data column in the initial state. Each data slot can bind multiple collected data corresponding to one user terminal, and the dimensional characteristics corresponding to each collected data are different. By clicking on the data slot, a data list composed of the collected data of different dimensions of the enterprise can be popped up, and the collected data of each dimension can be viewed through the data list. Among them, the data slot refers to a slot that can display the data of the enterprise.

[0104] S23. Generate a dimension display table based on the dimensional characteristics corresponding to each user terminal and send it to the push management terminal, and obtain the target dimension selected by the push management terminal based on the dimension display table.

[0105] In practical applications, the enterprise's database contains data with various dimensional characteristics, but in subsequent data recommendations, data corresponding to the dimensions selected by the management personnel can be recommended.

[0106] Among them, the dimension display table refers to a list composed of multiple dimensional characteristics corresponding to the user terminal, the recommendation management terminal refers to the terminal corresponding to the management personnel responsible for data recommendation, and the target dimension refers to the dimensional characteristic selected by the management personnel responsible for data recommendation from the dimension display table in combination with the requirements during equipment recommendation.

[0107] Specifically, multiple dimensional characteristics corresponding to each user terminal can be generated into a corresponding dimension display table, and the dimension display table is sent to the recommendation management terminal. The management personnel responsible for data recommendation can select the dimensional characteristics for data recommendation from the dimension display table in combination with the requirements of the user terminal, and determine the selected dimensional characteristics as the target dimension.

[0108] S24. Obtain the data similarity coefficients of each user terminal according to the collected data corresponding to the target dimension, and perform permutation and update on each data slot in the initial data column based on the data similarity coefficients to obtain a fusion data column.

[0109] Specifically, by parsing and comparing the collected data corresponding to the target dimension, the data similarity coefficients corresponding to each client can be determined. For example, when the target dimension is the resident dimension, the corresponding collected data can be the household electricity consumption data. By comparing the household electricity consumption data corresponding to each client, the data similarity coefficients between each client can be obtained. According to the data similarity coefficients, the data collections in the initial data column are arranged and updated, and a fused data column can be obtained. The data corresponding to adjacent data slots in the fused data column are relatively similar.

[0110] Among them, the data similarity coefficient is a coefficient that can be used to measure the similarity degree between the collected data.

[0111] Through the above implementation, the efficiency during data fusion can be improved.

[0112] On the basis of the above embodiment, the specific implementation manner of step S24 can be:

[0113] S241, parse the collected data to obtain the electricity consumption corresponding to each target dimension, and count the electricity consumption corresponding to multiple target dimensions of each client to obtain the total electricity consumption of each client.

[0114] Specifically, there may be multiple target dimensions selected in combination with user requirements. By parsing the collected data, the electricity consumption corresponding to each target dimension can be obtained. By counting the electricity consumption corresponding to each target dimension, the total electricity consumption corresponding to each client can be obtained.

[0115] Among them, the electricity consumption is the electricity consumed by the client corresponding to the target dimension, and the total electricity consumption refers to the electricity obtained by statistically summarizing the electricity consumption under each target dimension.

[0116] S242, determine the client with the least total electricity consumption as the target user, and obtain the data similarity coefficients between the remaining clients and the target user according to the electricity consumption corresponding to each target dimension.

[0117] Specifically, when obtaining the data similarity coefficient, the client with the least total electricity consumption can be determined as the target user, and then the electricity consumption of each target dimension corresponding to the remaining clients can be obtained. For example, when the target dimensions are the resident dimension and the commercial dimension, the electricity consumption of the resident dimension and the commercial dimension of the target user can be obtained, as well as the electricity consumption of the resident dimension and the commercial dimension corresponding to the remaining clients. The data similarity coefficients between the remaining clients and the target user can be determined through the electricity consumption of the resident dimension and the electricity consumption of the commercial dimension.

[0118] Among them, the target user refers to the client with the least total electricity consumption.

[0119] In some embodiments, "obtaining the data similarity coefficients between the remaining client devices and the target user according to the power consumption corresponding to each of the target dimensions" in step S242 includes the following steps:

[0120] S2421, obtaining the power consumption differences of each of the target dimensions corresponding to the target user and the remaining client devices.

[0121] For example, when the target dimensions are the residential dimension and the commercial dimension, the difference in power consumption of the residential dimension between the target user and the power consumption of the residential dimension corresponding to each of the remaining client devices, as well as the difference in power consumption of the commercial dimension, can be obtained.

[0122] Wherein, the power consumption difference refers to the difference between the power consumption of the target dimension corresponding to the target user and the power consumption of the target dimension corresponding to each of the remaining client devices.

[0123] S2422, retrieving the similarity adjustment weights corresponding to each of the target dimensions, and obtaining the sub-similarity coefficients corresponding to each of the target dimensions according to the product of the reciprocal of the power consumption difference and the similarity adjustment weight.

[0124] Wherein, the similarity adjustment weight is a data index that can be used to measure the importance of different dimension features in evaluating similarity. The more important the dimension feature, the larger the corresponding similarity adjustment weight can be. The reciprocal of the power consumption difference can be used to measure the difference in power consumption between the target user and other client devices in the target dimension. The smaller the difference, the closer the power consumption of the two users in this dimension, so its reciprocal is larger, which can reflect a higher similarity. The sub-similarity coefficient refers to the similarity coefficient corresponding to each target dimension.

[0125] In practical applications, different dimension features may have different impacts on the evaluation of similarity. For example, in some cases, the power consumption of the residential dimension may be more important than other dimensions. Therefore, a similarity adjustment weight can be assigned to each dimension feature to reflect its relative importance in the similarity evaluation. By multiplying the reciprocal of the power consumption difference by the similarity adjustment weight, a corresponding sub-similarity coefficient can be obtained, and the sub-similarity coefficient can reflect the similarity between the target user and other client devices in the target dimension.

[0126] S2423, statistically calculating the sub-similarity coefficients corresponding to each of the target dimensions to obtain the data similarity coefficients between the target user and the remaining client devices.

[0127] Specifically, after obtaining the sub-similarity coefficients corresponding to each target dimension, by statistically calculating each sub-similarity coefficient, the data similarity coefficients between the remaining client devices and the target user can be obtained. The larger the data similarity coefficient, the greater the similarity between the corresponding client device and the target user.

[0128] S243. Obtain the client with the largest data similarity coefficient as the next target user, and repeat the steps of obtaining the data similarity coefficient and the target user until the next target user is the last client.

[0129] Specifically, the client with the highest similarity to the first target user can be used as the next target user. Therefore, the client with the largest data similarity coefficient can be determined as the next target user. Obtain the power consumption of the target dimension corresponding to this target user, and obtain the data similarity coefficients between the remaining clients and this target user according to the same steps as above. Then, use the client with the largest data similarity coefficient as the next next target user, and repeat the above steps of obtaining the data similarity coefficient and the target user until the next target user determined is the last client, and then stop the above repeated steps.

[0130] S244. Arrange and update each data slot according to the determination order of each target user to obtain the fused data column.

[0131] Specifically, when determining the target user according to the data similarity coefficient, the determination order of each target user can be obtained through the magnitude of the data similarity coefficient. The data slots in the initial data column can be arranged and updated according to the determination order of each target user to obtain the fused data column.

[0132] Among them, the determination order refers to the order when determining the target user according to the magnitude of the data similarity coefficient.

[0133] Through the above embodiments, the importance of different dimensions can be adjusted in combination with weights, making the evaluation of similarity more comprehensive and accurate.

[0134] Based on the above steps, this solution further includes the following embodiments:

[0135] A1. Receive the gap display mode selected by the push management end, where the gap display mode includes a slot display mode and a spacing display mode.

[0136] It can be understood that the similarity between each adjacent data slot is the highest, but the similarity between each adjacent data slot may be different. The spacing between adjacent data slots on the obtained fused data column is the same, and it may not be possible to display the different similarities between each adjacent data slot through the fused data column. In order to be able to more intuitively display the similarity between each adjacent data slot, the fused data column can be adjusted accordingly in combination with the data similarity coefficient, so that the adjusted fused data column can more intuitively display the different similarities between adjacent data slots.

[0137] In practical applications, the greater the similarity between adjacent data slots, the greater the difference between the data corresponding to the adjacent data slots can be considered. Therefore, after obtaining the fused data column, the similarity difference between adjacent data slots can be displayed in combination with the data similarity coefficient. First, the recommendation manager responsible for data recommendation can select the corresponding difference display mode. After receiving the difference display mode selected by the push management end, the fused data column can be adjusted in combination with the corresponding display mode.

[0138] Among them, the difference display mode refers to the mode of displaying the similarity difference between adjacent data slots. The slot display mode refers to the mode of displaying the similarity difference by inserting interval slots between adjacent data slots. The spacing display mode refers to the mode of displaying the similarity difference by adjusting the interval distance between adjacent data slots.

[0139] A2. Obtain the number of interval slots between adjacent data slots in the fused data column according to the slot display mode, and insert the corresponding number of interval slots between adjacent data slots to obtain an adjusted data column.

[0140] When the slot display mode is selected by the push management end, the difference between data can be displayed by inserting interval slots between adjacent data slots. The greater the difference, the more interval slots are inserted. Specifically, first, the number of interval slots to be inserted between adjacent data slots can be obtained according to the slot display mode, and the corresponding number of interval slots can be inserted between adjacent data slots to adjust the fused data column and obtain the corresponding adjusted data column.

[0141] In some embodiments, the specific implementation manner of step A2 may be:

[0142] A21. Determine that the user ends corresponding to adjacent data slots in the fused data column belong to the same user group, and obtain the similarity difference between the corresponding data similarity coefficients of the user group.

[0143] Among them, the user group refers to the display group composed of the user ends corresponding to adjacent data slots in the fused data column, and the similarity difference refers to the difference between the corresponding data similarity coefficients of adjacent user ends in the user group.

[0144] A22. Compare the similarity difference with the reference coefficient difference, and determine the user group with the similarity difference greater than the reference coefficient difference as the adjustment group.

[0145] Specifically, after obtaining the similarity difference, this solution can pre-configure a reference coefficient difference that can measure the magnitude of the similarity difference between adjacent client terminals. If the similarity difference is greater than the reference coefficient difference, it can be considered that the gap between the data corresponding to adjacent client terminals is too large. To clearly display the large gap, it can be shown by inserting spacer slots. The more spacer slots inserted, the greater the gap between the data corresponding to adjacent client terminals. At this time, the corresponding user group can be determined as the adjustment group.

[0146] Among them, the reference coefficient difference refers to the difference between pre-configured standard similarity coefficients, and the adjustment group refers to the user group with a similarity difference greater than the reference coefficient difference.

[0147] A23. Determine the number of spacer slots for the adjustment group pair according to the ratio of the similarity difference to the reference coefficient difference.

[0148] Specifically, by calculating the ratio of the similarity difference to the reference coefficient difference, the number of spacer slots corresponding to the adjustment group can be obtained. To ensure that enough spacer slots are inserted between the adjustment groups, so as to more accurately display the data gap between adjacent client terminals, the calculated ratio can be rounded up. Among them, the number of spacer slots refers to the number of spacer slots inserted between the adjustment groups.

[0149] A24. Insert the number of spacer slots between the data slots corresponding to the adjustment group to obtain an adjusted data column.

[0150] See Figure 2 , which is a schematic diagram for determining an adjusted data column in the slot display mode provided by an embodiment of the present invention. As shown in Figure 2 , in the initial data column, no spacer slots are inserted between the data slots, and the similarity gap between adjacent data slots cannot be judged. After corresponding calculations, the number of spacer slots to be inserted between data slot 1 and data slot 2 is 1, and the number of spacer slots to be inserted between spacer slot 2 and spacer slot 3 is 2. After obtaining the number of spacer slots, insert the corresponding number of spacer slots between the data slots corresponding to each adjustment group. The spacing between the inserted spacer slots is the same, and the inserted spacer slots can be displayed using pixel values different from those of the data slots, so as to obtain the corresponding adjusted data column.

[0151] Based on the above steps, this solution further includes the following embodiments:

[0152] B1. Obtain the slot spacing between adjacent spacer slots, and determine the spacer slots with a slot spacing less than the reference gap spacing as the target slots.

[0153] Among them, the slot spacing between multiple adjacent spacer slots corresponding to the same adjustment group is equal.

[0154] In practical applications, after inserting a corresponding number of spacer slots between adjacent data slots, when the data gap between adjacent data slots is too large, the number of inserted spacer slots will also be excessive. At this time, when displaying, there may be a situation of overcrowding or overlapping between multiple spacer slots. In this case, the size of the spacer slots can be adjusted in combination with the spacing distance between the spacer slots. Specifically, the spacing distance between adjacent spacer slots, that is, the slot pitch, can be obtained. When the slot pitch is less than the pre-configured reference gap spacing, it can be considered that adjacent spacer slots overlap. At this time, the adjacent spacer slots can be determined as target slots, and subsequently, the size of the target slots can be adjusted accordingly.

[0155] Among them, the slot pitch refers to the spacing distance between adjacent spacer slots, the reference gap spacing refers to the pre-configured standard slot pitch, and under the reference gap spacing, there will be no overlapping phenomenon between adjacent spacer slots. The target slot refers to the spacer slot whose slot pitch is less than the reference gap spacing.

[0156] B2. Calculate the spacing difference between the slot pitch and the reference gap spacing, and obtain the size offset coefficient according to the ratio of the spacing difference to the reference spacing difference.

[0157] Specifically, the difference between the slot pitch and the reference gap spacing, that is, the spacing difference, can be calculated. By calculating the ratio of the spacing difference to the reference spacing difference, the size offset coefficient for adjusting the size of the spacer slots can be obtained.

[0158] Among them, the spacing difference refers to the difference between the slot pitch and the reference gap spacing, and the size offset coefficient refers to the coefficient for adjusting the size of the spacer slots.

[0159] B3. Obtain the size adjustment value according to the product of the size offset coefficient and the size adjustment weight of the spacer slot, and obtain the adjusted size based on the difference between the standard size of the target slot and the size adjustment value.

[0160] Specifically, the pre-configured size adjustment weight corresponding to the spacer slot can be obtained. By multiplying the size adjustment weight by the corresponding size offset coefficient, the size corresponding to the adjusted size of the target slot, that is, the size adjustment value, can be obtained. The standard size corresponding to the target slot can be obtained. By calculating the difference between the standard size and the size adjustment value, the adjusted size when adjusting the size of the target slot can be obtained.

[0161] Among them, the size adjustment weight refers to the weight for adjusting the size offset coefficient, the size adjustment value refers to the size of the target slot calculated according to the size offset coefficient and the corresponding size adjustment weight, the standard size refers to the initial size of the target slot before any size adjustment, and the adjustment size refers to the difference between the standard size and the size adjustment value.

[0162] B4, Based on the adjustment size, reduce and adjust the slot size of the target slot to obtain an updated adjusted data column.

[0163] Specifically, after obtaining the adjustment size, the slot size of the target slot can be reduced and adjusted according to the adjustment size to obtain an updated adjusted data column.

[0164] A3, Or, based on the spacing display mode, obtain the similarity difference between adjacent data slots in the fusion data column, and adjust the spacing between adjacent data slots according to the similarity difference to obtain an adjusted data column.

[0165] Specifically, when the gap display mode selected by the push management end is the spacing display mode, the similarity difference between adjacent data slots can be obtained according to the spacing display mode, and the interval distance between adjacent data slots can be adjusted accordingly according to the similarity difference. For example, if two data slots are very similar, the spacing between them may be reduced, and if the difference is large, the spacing may be enlarged, so as to obtain an adjusted data column. Among them, the similarity difference refers to the difference between the data similarity coefficients corresponding to adjacent data slots.

[0166] In some embodiments, the specific implementation manner of step A3 may be:

[0167] A31, Obtain the similarity difference between the data similarity coefficients corresponding to adjacent data slots in the fusion data column.

[0168] Specifically, since each data slot in the fusion data column has a corresponding data similarity coefficient, the difference between the data similarity coefficients corresponding to adjacent data slots, that is, the similarity difference, can be obtained.

[0169] A32, Obtain a spacing adjustment coefficient according to the ratio of the similarity difference to the reference coefficient difference, and obtain the adjusted spacing corresponding to the adjacent data slots based on the product of the spacing adjustment coefficient and the standard spacing.

[0170] Specifically, by calculating the ratio of the similarity difference to the pre-configured reference coefficient difference, a spacing adjustment coefficient for offsetting the standard spacing can be obtained. By multiplying the spacing adjustment coefficient by the standard spacing, an adjusted spacing for adjusting the interval distance between adjacent data slots can be obtained. Here, the spacing adjustment coefficient refers to the coefficient for adjusting the interval distance between adjacent data slots, the standard spacing refers to the interval distance between adjacent data slots in the fused data column, and the adjusted spacing refers to the interval distance obtained after correspondingly adjusting the standard spacing.

[0171] A33, adjust the spacing between adjacent data slots based on the adjusted spacing to obtain an adjusted data column.

[0172] Specifically, after obtaining the adjusted spacing, the spacing between adjacent data slots can be correspondingly adjusted according to the adjusted spacing. When the adjusted spacing is less than the initial spacing, it indicates that the spacing between adjacent data slots needs to be reduced, and the corresponding data axis can be shortened. When the adjusted spacing is greater than the initial spacing, it can be considered that the spacing between adjacent data slots needs to be enlarged, and the corresponding data axis can be extended. Moreover, for the two different cases of spacing reduction and spacing enlargement, different pixel values can be used to correspondingly update the data axis.

[0173] For example, for the case of spacing reduction, gray pixel values can be used to update the corresponding data axis. For the case of spacing enlargement, red pixel values can be used to update the corresponding data axis, so as to display different adjustment situations. Refer to Figure 3 , which is a schematic diagram for determining an adjusted data column in a spacing display mode provided by an embodiment of the present invention. As shown in Figure 3 , the spacing between each data slot in the initial data column is equal. Compared with the initial data column, in the adjusted data column, the spacing between data slot 1 and data slot 2 is reduced. Therefore, the data axis between data slot 1 and data slot 2 can be shortened and can be displayed by gray pixel values. The spacing between data slot 2 and data slot 3 is enlarged. Therefore, the data axis between data slot 2 and data slot 3 can be extended and can be displayed by red pixel values.

[0174] Through the above implementation manner, the similarity difference between adjacent data slots in the fused data column can be more intuitively displayed, and the similarity and difference between data can be effectively reflected, providing a more intuitive basis for data push.

[0175] S3, obtain the fusion index of multiple dimension features corresponding to the user end in the fused data column, and calculate the user portrait data corresponding to each user end according to the fusion index.

[0176] After obtaining the fused data column, the data of the fusion metrics of multiple dimensional features in the fused data column can be obtained. By calculating the fusion metrics, the portrait data corresponding to each client can be obtained. Based on the portrait data corresponding to the client, the energy consumption situation of each client can be judged. Subsequently, relevant data can be pushed according to the energy consumption situation, and devices that are more suitable for the client can be determined according to the energy consumption situation. For example, when the required energy consumption of the client is large, a storage device with a larger capacity can be determined when determining the device.

[0177] Among them, the fusion metric refers to the data corresponding to each dimensional feature in the fused data column. For example, the electricity consumption corresponding to the resident feature, and the user portrait data refers to the data that can reflect the electricity consumption situation of the client.

[0178] Based on the above embodiments, the specific implementation manner of step S3 can be:

[0179] S31. Respond to the portrait prediction request of the push management terminal, and screen multiple clients corresponding to the fused data column to obtain reference clients.

[0180] In practical applications, there may be many clients. If the portrait data corresponding to each one is calculated separately, the data processing volume is large. This solution can screen from multiple clients, and predict the portrait data of the remaining clients through the portrait data corresponding to the screened clients, which can reduce the data processing volume and improve the data processing efficiency.

[0181] Specifically, it can respond to the portrait prediction request sent by the push management personnel, and then screen multiple clients in the fused data column, and determine the screened clients as reference clients.

[0182] Among them, the portrait prediction request refers to the request for predicting the portrait data of the remaining clients based on the portrait data of the screened clients, and the reference client refers to the client screened from multiple clients corresponding to the fused data column.

[0183] In some embodiments, the specific implementation manner of step S31 can be:

[0184] S311. Respond to the portrait prediction request, and obtain the spacing calculation amount according to the sum of the preset reference quantity and the spacing offset value.

[0185] Among them, the preset reference quantity refers to the number of reference clients configured in advance, the spacing offset value is 1, and the spacing calculation amount refers to the quantity after evenly dividing the fused data column at the same spacing.

[0186] Specifically, after receiving the portrait prediction request from the push management terminal, the corresponding number of reference users, i.e., the preset reference number, can be pre-configured. By adding the preset reference number to the spacing offset value, the number for equally dividing the fused data column, i.e., the spacing calculation amount, can be obtained.

[0187] S312. Obtain the data column length of the fused data column, and obtain the unit spacing according to the ratio of the data column length to the spacing calculation amount.

[0188] Among them, the data column length refers to the length of the fused data column, and the unit spacing refers to the value obtained by calculating the ratio of the data column length to the spacing calculation amount.

[0189] S313. Determine the reference points of the preset reference number in the fused data column according to the unit spacing, obtain the data slot closest to the reference point as its corresponding reference slot, and determine the user terminal corresponding to the reference slot as the reference user.

[0190] Specifically, after obtaining the unit spacing, the reference points can be determined in the fused data column according to the unit spacing. One reference point can be determined every unit spacing, and the reference points of the preset reference number can be determined in sequence. Since in the fused user data column, each reference point has a corresponding data slot closest to it, the data slot closest to the reference point can be determined as the reference slot, and the user terminal corresponding to the reference slot can be determined as the reference user.

[0191] Among them, the reference point refers to the point in the fused data column that can be used to determine the reference user, the reference slot refers to the data slot closest to the reference point, and the reference user refers to the user terminal corresponding to the reference slot.

[0192] S32. Obtain the fusion indicators of the multiple dimension features corresponding to the reference users, and the fusion indicators at least include the electricity consumption of the corresponding dimension features.

[0193] Specifically, each dimension feature has a corresponding fusion indicator. For example, for the residential dimension, the corresponding residential electricity consumption can be obtained. Therefore, the fusion indicators of the multiple dimension features corresponding to the reference users can be obtained. For example, the electricity consumption of the residential dimension, the electricity consumption of the commercial dimension, and the electricity consumption of the industrial dimension corresponding to the reference users can be obtained.

[0194] S33. Calculate the reference portrait data corresponding to each reference user according to the fusion indicators corresponding to each reference user.

[0195] Among them, the reference portrait data refers to the portrait data corresponding to the reference user obtained by calculating the fusion indicators corresponding to the reference user.

[0196] In some embodiments, the specific implementation of step S33 may be as follows:

[0197] S331. Determine the benchmark index corresponding to each dimension feature according to the mean value of the fusion indexes of each dimension feature corresponding to each reference user.

[0198] Specifically, the mean value calculation can be sequentially performed on the fusion indexes of the same dimension feature corresponding to multiple reference users to obtain the benchmark index of the corresponding dimension feature. For example, when there are 100 reference users, the electricity consumption of the resident dimension corresponding to 100 reference users can be obtained, and the mean value of the fusion indexes of the 100 resident dimensions can be calculated by taking the mean value of the electricity consumption of the 100 resident dimensions, and this mean value can be determined as the benchmark index corresponding to the resident dimension. Therefore, the benchmark indexes corresponding to each dimension feature can be obtained. Among them, the benchmark index refers to the average value of the fusion indexes of multiple reference users corresponding to the same dimension feature.

[0199] S332. Compare the fusion indexes of each dimension feature corresponding to each reference user with the benchmark index, determine that the dimension feature with the fusion index greater than the benchmark index corresponds to a high energy consumption attribute, and determine that the dimension feature with the fusion index less than the benchmark index corresponds to a low energy consumption attribute.

[0200] After obtaining the benchmark index, by comparing the fusion index of each dimension feature with the benchmark index, the energy consumption attribute corresponding to each dimension feature can be determined. If the fusion index of the dimension feature is greater than the benchmark index, it can be considered that the electricity consumption of this dimension feature is relatively high, and its corresponding energy consumption level is also relatively high, and it can be determined that the energy consumption attribute corresponding to this dimension feature is a high energy consumption attribute. If the fusion index of the dimension feature is less than the benchmark index, it can be considered that the electricity consumption of this dimension feature is relatively low, and its corresponding energy consumption level is also relatively low, and it can be determined that the energy consumption attribute corresponding to this dimension feature is a low energy consumption attribute. Subsequently, the portrait data corresponding to the user side can be determined according to the corresponding energy consumption attribute.

[0201] Among them, the high energy consumption attribute refers to the attribute corresponding to the dimension feature with a relatively high energy consumption level, and the low energy consumption attribute refers to the attribute corresponding to the dimension feature with a relatively low energy consumption level.

[0202] S333. Determine the energy consumption demand corresponding to the electricity consumption of the fusion index according to the high energy consumption attribute or the low energy consumption attribute, and obtain the reference portrait data of each reference user based on the energy consumption demands corresponding to each dimension feature.

[0203] Specifically, according to the energy consumption attributes corresponding to each dimensional feature, a model for calculating the energy consumption demand can be constructed. By inputting the power consumption data into the energy consumption demand calculation model, the energy consumption demand corresponding to each dimensional feature can be further determined through calculation. For example, for the residential dimension with high energy consumption attributes, it can be determined that the power consumption demand of this user may be relatively high. Combining the energy consumption attributes of each dimensional feature and their corresponding energy consumption demands, the reference portrait data corresponding to each reference user can be constructed. The reference portrait data can reflect the energy consumption demands of the reference user in each dimension and can provide a reference for predicting the portrait data of other users in the future.

[0204] Among them, the energy consumption demand refers to the energy consumption that a user may need determined according to the dimensional features of the user and their corresponding energy consumption attributes.

[0205] S34. Obtain the data difference coefficients between each of the remaining user terminals and the reference user, and calculate the predicted portrait data corresponding to each of the remaining user terminals according to the reference portrait data and the data difference coefficients. The user portrait data includes the reference portrait data and the predicted portrait data.

[0206] After obtaining the reference portrait data of the reference user, the portrait data corresponding to each of the remaining user terminals can be calculated through the differences between each of the remaining user terminals and the reference user. Specifically, according to the differences between each of the remaining user terminals and the reference user, the data difference coefficients between each of the remaining user terminals and the reference user can be obtained. Combining the data difference coefficients and the reference portrait data of the reference user, the portrait data corresponding to the remaining users can be calculated to obtain the predicted portrait data corresponding to each of the remaining user terminals. Among them, the data difference coefficient is a coefficient that can be used to measure the degree of difference between the corresponding data between the remaining user terminals and the reference user, and the predicted portrait data refers to the portrait data corresponding to each of the remaining user terminals.

[0207] Through the above implementation methods, the data processing volume can be reduced and the data processing efficiency can be improved.

[0208] In some embodiments, the specific implementation manner of step S34 may be:

[0209] S341. Determine the fusion data columns with interval slots as one type of data column, and obtain the reference user with the closest distance as the corresponding type of user for each of the remaining user terminals.

[0210] Specifically, there are two different adjustment display modes when adjusting the fusion data columns. The fusion data columns with interval slots can be determined as one type of data column. Since there may be multiple reference users, in order to reduce the data processing volume, the reference user with the closest distance to the remaining user terminals can be used as the corresponding type of user for the corresponding user terminals.

[0211] Among them, a first - type data column refers to a fused data column with interval slots, and a first - type user refers to a reference user in the first - type data column that is closest to the remaining client terminals.

[0212] S342. Obtain the number of slot positions of the interval slots between the client terminal and the first - type user, obtain a first - type offset coefficient according to the ratio of the number of slot positions to the reference number of slot positions, and obtain a data difference coefficient of the client terminal based on the product of the first - type offset coefficient and the first - type weight.

[0213] After determining the first - type user, there are corresponding numbers of interval slots between each of the remaining client terminals in the first - type data column and the first - type user. Since the interval slots can represent the similarity gap between data slots, the number of interval slots between the client terminal and the first - type user can represent the similarity gap between the client terminal and the first - type user. The more the number of interval slots, the greater the similarity gap, and the corresponding data difference coefficient of the client terminal will also be greater.

[0214] Specifically, the number of interval slots between the client terminal and the first - type user, that is, the number of slot positions, can be obtained. By calculating the ratio of the number of slot positions to the pre - configured reference number of slot positions, the corresponding first - type offset coefficient can be obtained. Then, the corresponding first - type weight can be retrieved. By multiplying the first - type weight by the first - type offset coefficient, the first - type offset coefficient can be adjusted to obtain the data difference coefficient corresponding to the client terminal on the first - type data column.

[0215] Among them, the number of slot positions refers to the number of interval slots between the client terminal on the first - type data column and the corresponding first - type user. The reference number of slot positions refers to the pre - configured standard number of slot positions. The first - type offset coefficient refers to a coefficient that can reflect the deviation degree of the similarity gap between the client terminal and the first - type user relative to the standard gap. The larger the first - type offset coefficient, the greater the similarity gap between the client terminal and the first - type user. The first - type weight refers to a value that can be used to adjust the first - type offset coefficient.

[0216] S343. Determine the fused data column without interval slots as the second - type data column, and obtain the reference user closest to the remaining corresponding client terminals as the second - type user corresponding to the client terminals.

[0217] Specifically, the fused data column without interval slots can be determined as the second - type data column. Similarly, the reference user closest to the remaining client terminals can be used as the second - type user corresponding to the client terminals.

[0218] Among them, the second - type data column refers to the fused data column obtained by combining the spacing display mode, and the second - type user refers to the reference user in the second - type data column that is closest to the remaining client terminals.

[0219] S344. Obtain the interval distance between the client and the secondary user, obtain the secondary offset coefficient according to the ratio of the interval distance to the reference interval distance, and obtain the data discrimination coefficient of the client based on the product of the secondary offset coefficient and the secondary weight.

[0220] After determining the secondary user, there is a certain interval distance between each of the remaining clients in the secondary data column and the secondary user. The greater the interval distance, the greater the similarity gap between the client and the secondary user, and the correspondingly greater the data discrimination coefficient of the client. Therefore, the interval distance between the client and the corresponding secondary user can be obtained. By calculating the ratio of the interval distance to the pre-configured reference interval distance, the secondary offset coefficient can be obtained. By multiplying the secondary offset coefficient by the corresponding secondary weight, the secondary offset coefficient can be adjusted to obtain the data discrimination coefficient corresponding to the client on the secondary data column.

[0221] Among them, the interval distance refers to the distance between the client on the secondary data column and the corresponding secondary user. The reference interval distance refers to the pre-configured standard interval distance. The secondary offset coefficient refers to the coefficient that can reflect the deviation degree of the similarity gap between the client and the secondary user relative to the standard gap. The greater the secondary offset coefficient, the greater the similarity gap between the client and the secondary user. The secondary weight refers to the value that can be used to adjust the secondary offset coefficient.

[0222] S345. Obtain the adjusted demand corresponding to each dimension feature of the client according to the product of the data discrimination coefficient and the energy consumption demand of each dimension feature corresponding to the primary user or the secondary user.

[0223] After obtaining the data discrimination coefficient corresponding to the client, for the client on the primary data column, by multiplying the data discrimination coefficient by the energy consumption demand of each dimension feature corresponding to the primary user, the adjusted demand of the dimension feature corresponding to the client can be obtained. For example, by multiplying the data discrimination coefficient by the energy consumption demand of the resident dimension corresponding to the primary user, the adjusted demand of the resident dimension corresponding to the client can be obtained. For the client on the secondary data column, by multiplying the data discrimination coefficient by the energy consumption demand of each dimension feature corresponding to the secondary user, the adjusted demand of the dimension feature corresponding to the client can be obtained.

[0224] Among them, the adjusted demand refers to the demand obtained after correspondingly adjusting the energy consumption demand corresponding to the dimension feature.

[0225] S346. Determine the adjustment trend of each client, determine the predicted demand of each client according to the adjustment trend and the adjustment demand, and obtain the predicted portrait data corresponding to the client based on the predicted demand corresponding to each dimension feature.

[0226] Specifically, the adjustment trend corresponding to the client can be determined by analyzing the position of the data slot of the client, and the predicted demand corresponding to the client can be calculated according to the corresponding adjustment trend and the adjustment demand. The predicted portrait data corresponding to the client can be determined through the predicted demand corresponding to each dimension feature.

[0227] Among them, the adjustment trend refers to the trend when adjusting the adjustment demand, and the predicted demand refers to the demand obtained by correspondingly adjusting the adjustment demand according to the adjustment trend.

[0228] In some embodiments, "determine the adjustment trend of each client, and determine the predicted demand of each client according to the adjustment trend and the adjustment demand" in step S346 includes the following steps:

[0229] S3461. Obtain the slot side of the data slot corresponding to the first-type user or the second-type user in the data slot of the client, and the slot side includes a decreasing side or an increasing side.

[0230] In practical applications, the data slot corresponding to the first-type user or the second-type user may be located on the decreasing side or the increasing side of the data slot corresponding to the client. For example, the decreasing side may be the left side of the data slot of the client, and the increasing side may be the right side of the data slot of the client. When the data slot corresponding to the first-type user or the second-type user is located on the left side of the data slot corresponding to the client, the demand corresponding to the client may decrease. When the data slot corresponding to the first-type user or the second-type user is located on the right side of the data slot corresponding to the client, the demand corresponding to the client may increase. Therefore, when determining the predicted demand of the client according to the adjustment demand, it is first necessary to determine the slot side of the data slot corresponding to the first-type user or the second-type user relative to the data slot of the client. Among them, the slot side refers to the relative position side of the data slot corresponding to the first-type user or the second-type user relative to the data slot corresponding to the client. Different position sides can represent different adjustment trends. The decreasing side refers to the slot side where the demand decreases in sequence, and the increasing side refers to the slot side where the demand increases in sequence.

[0231] S3462. Determine that the dimension feature corresponding to the decreasing side is a decreasing trend, and determine that the dimension feature corresponding to the increasing side is an increasing trend. The adjustment trend includes a decreasing trend and an increasing trend.

[0232] For example, when the decreasing side is the slot side on the left, the dimensional feature of the data slot to the left of the data slot corresponding to a first type of user or a second type of user can be determined as a decreasing trend. When the increasing side is the slot side on the right, the dimensional feature of the data slot to the right of the data slot corresponding to a first type of user or a second type of user can be determined as an increasing trend. Herein, the decreasing trend refers to the trend of decreasing demand, and the increasing trend refers to the trend of increasing demand.

[0233] S3463. Obtain the energy consumption demand with the decreasing trend corresponding to the reference user as the first type of demand, and obtain the predicted demand corresponding to the decreasing trend according to the difference between the first type of demand and the adjusted demand.

[0234] Specifically, the energy consumption demand with the decreasing trend corresponding to the reference user can be determined as the first type of demand. By calculating the difference between the first type of demand and the adjusted demand, the predicted demand corresponding to the decreasing trend can be obtained. Herein, the first type of demand refers to the energy consumption demand with the decreasing trend.

[0235] S3464. Determine the energy consumption demand with the increasing trend corresponding to the reference user as the second type of demand, and obtain the predicted demand corresponding to the increasing trend according to the sum of the second type of demand and the adjusted demand.

[0236] Specifically, the energy consumption demand with the increasing trend corresponding to the reference user can be determined as the second type of demand. By calculating the sum of the second type of demand and the adjusted demand, the predicted demand corresponding to the increasing trend can be obtained. Herein, the second type of demand refers to the energy consumption demand with the increasing trend.

[0237] S4. Generate the push data corresponding to each user terminal according to the user profile data, and send the push data to the corresponding user terminal.

[0238] This solution can recommend products to each user terminal according to the data in the fusion data column, specifically the recommendation of some energy-consuming devices, such as photovoltaic power generation devices, industrial cleaning devices, etc. Since the power consumption data corresponding to each user terminal is different, in order to achieve customized recommendations for user terminals, corresponding data push can be combined with the user profile data corresponding to the user terminal.

[0239] Specifically, according to the user profile data corresponding to each user terminal, the energy consumption demand required by each user terminal can be determined. According to the corresponding energy consumption demand, relevant data is pushed, the push data corresponding to the user terminal is determined, and the push data can be sent to the corresponding user terminal. The push data corresponding to each user terminal is different.

[0240] Herein, the push data refers to the data displayed on the user terminal when recommending devices to the user.

[0241] Based on the above embodiments, the specific implementation manner of step S4 may be as follows:

[0242] S41. Determine the high-demand devices corresponding to the dimensional features of the high energy consumption attributes, and determine the low-demand devices corresponding to the dimensional features of the low energy consumption attributes.

[0243] Specifically, it may be determined that the high-demand devices are the devices corresponding to the dimensional features of the high energy consumption attributes, and the low-demand devices are the devices corresponding to the dimensional features of the low energy consumption attributes. Among them, high-demand devices refer to devices with relatively high energy consumption requirements, such as high-power industrial cleaning devices, and low-demand devices refer to devices with relatively low energy consumption requirements, such as energy-saving compressors.

[0244] S42. Determine the device capacity corresponding to the high-demand devices or low-demand devices according to the energy consumption demand, and generate sub-push data corresponding to the dimensional features based on the high-demand devices or low-demand devices and the device capacity.

[0245] Specifically, according to the energy consumption demand corresponding to the dimensional features of the high energy consumption attributes, the device capacity corresponding to the high-demand devices can be determined. According to the energy consumption demand corresponding to the dimensional features of the low energy consumption attributes, the device capacity corresponding to the low-demand devices can be determined. The greater the energy consumption demand, the greater the corresponding device capacity. According to the high-demand devices and the corresponding device capacity, sub-push data corresponding to the dimensional features of the high energy consumption attributes can be generated. According to the low-demand devices and the corresponding device capacity, sub-push data corresponding to the dimensional features of the low energy consumption attributes can be generated.

[0246] Among them, the device capacity refers to the maximum power that the device can provide under normal operating conditions, and the sub-push data refers to the push data corresponding to each dimensional feature.

[0247] S43. Generate push data corresponding to each user terminal based on the sub-push data, and send the push data to the corresponding user terminal.

[0248] Specifically, according to the sub-push data corresponding to each dimensional feature, push data corresponding to the user terminal can be generated, and the generated push data can be sent to the corresponding user terminal to recommend corresponding data to the user.

[0249] Through the above implementation manner, customized data recommendation can be performed for the user terminal in combination with the requirements of the user terminal.

[0250] See Figure 4 , which is a schematic structural diagram of a multi-channel data fusion system provided by an embodiment of the present invention. The data processing system based on the multi-channel data fusion system includes:

[0251] A scraping module, configured to scrape data from data sources corresponding to each client based on a data collection unit, and obtain collected data of multiple dimension features corresponding to each client;

[0252] A construction module, configured to construct an initial data column corresponding to each client, and perform arrangement and update on the initial data column according to the collected data corresponding to each dimension feature to obtain a fused data column;

[0253] An acquisition module, configured to acquire fusion metrics of multiple dimension features corresponding to the client in the fused data column, and calculate user portrait data corresponding to each client according to the fusion metrics;

[0254] A generation module, configured to generate push data corresponding to each client according to the user portrait data, and send the push data to the corresponding client.

[0255] Figure 4 The device in the illustrated embodiment can correspondingly be used to execute Figure 1 the steps in the illustrated method embodiment, and its implementation principle and technical effects are similar, which will not be elaborated here.

[0256] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-channel data fusion method, characterized in that: include: Based on the data collection unit, data is captured from the data source corresponding to each user terminal to obtain the collected data of multiple dimensional features corresponding to each user terminal; Constructing an initial data column corresponding to each of the user terminals, and arranging and updating the initial data column according to the collected data corresponding to each of the dimensional features to obtain a fused data column; Obtaining fusion indicators of multiple dimensional features corresponding to the user terminals in the fusion data column, and calculating user portrait data corresponding to each of the user terminals according to the fusion indicators; Generate push data corresponding to each of the user terminals according to the user portrait data, and send the push data to the corresponding user terminal; Obtaining fusion indicators of multiple dimensional features corresponding to the user terminals in the fusion data column, and calculating user portrait data corresponding to each user terminal according to the fusion indicators, including: In response to the portrait prediction request of the push management end, multiple user ends corresponding to the fused data column are screened to obtain reference users; Acquire a fusion index of the plurality of dimensional features corresponding to the reference user, wherein the fusion index at least includes the power consumption of the corresponding dimensional features; The reference portrait data corresponding to each of the reference users is obtained by calculation according to the fusion index corresponding to each of the reference users; Obtaining a data distinction coefficient between each of the remaining user terminals and the reference user, and calculating predicted portrait data corresponding to each of the remaining user terminals according to the reference portrait data and the data distinction coefficient, wherein the user portrait data includes reference portrait data and predicted portrait data; The reference portrait data corresponding to each reference user is calculated according to the fusion index corresponding to each reference user, including: Determine the benchmark indicator corresponding to each dimensional feature according to the average value of the fusion indicator of each dimensional feature corresponding to each reference user; Compare the fusion index of each dimensional feature corresponding to each reference user with the benchmark index, determine that the dimensional feature whose fusion index is greater than the benchmark index corresponds to a high energy consumption attribute, and determine that the dimensional feature whose fusion index is less than the benchmark index corresponds to a low energy consumption attribute; Determine the energy consumption requirement corresponding to the power consumption of the fusion indicator according to the high energy consumption attribute or the low energy consumption attribute, and obtain reference portrait data of each reference user based on the energy consumption requirement corresponding to each dimensional feature; Constructing an initial data column corresponding to each of the user terminals, and arranging and updating the initial data column according to the collected data corresponding to each of the dimensional features to obtain a fused data column, including: Counting the number of users of the user terminal, determining the length of the data column corresponding to the number of users, and constructing an initial data axis according to the length of the data column; Constructing a corresponding number of data slots on the initial data axis according to the number of users, and binding each data slot with the collected data corresponding to the corresponding user terminal to obtain an initial data column; Generate a dimension display table based on the dimension features corresponding to each of the user terminals and send it to the push management terminal, and obtain the target dimension selected by the push management terminal based on the dimension display table; Obtaining a data similarity coefficient of each of the user terminals according to the collected data corresponding to the target dimension, and arranging and updating each of the data slots in the initial data column based on the data similarity coefficient to obtain a fused data column; Obtaining the data similarity coefficient of each user terminal according to the collected data corresponding to the target dimension, and arranging and updating each data slot in the initial data column based on the data similarity coefficient to obtain a fused data column, including: The collected data is parsed to obtain the power consumption corresponding to each of the target dimensions, and the power consumption corresponding to multiple target dimensions of each of the user terminals is counted to obtain the total power consumption of each of the user terminals; Determine the user terminal with the least total power consumption as the target user, and obtain the data similarity coefficient between the remaining user terminals and the target user according to the power consumption corresponding to each target dimension; Obtaining the user terminal with the largest data similarity coefficient as the next target user, and repeating the steps of obtaining the data similarity coefficient and the target user until the next target user is the last user terminal; Arrange and update each of the data slots according to the determined order of each of the target users to obtain the fused data column; Obtaining data similarity coefficients between the remaining user terminals and the target user according to the power consumption corresponding to each of the target dimensions, including: Obtaining the power consumption difference between the target user and the remaining user terminals in each target dimension; Retrieving the similarity adjustment weight corresponding to each of the target dimensions, and obtaining the sub-similarity coefficient corresponding to each of the target dimensions according to the product of the inverse of the power consumption difference and the similarity adjustment weight; Counting the sub-similarity coefficients corresponding to each of the target dimensions to obtain the data similarity coefficients between the target user and the remaining user terminals; After arranging and updating the data slots according to the determined order of the target users to obtain the fused data column, the method further includes: Receiving a gap display mode selected by the push management end, the gap display mode including a slot display mode and a spacing display mode; According to the slot display mode, the number of interval slots between adjacent data slots in the fused data column is obtained, and according to the number of interval slots, a corresponding number of interval slots are inserted between adjacent data slots to obtain an adjusted data column; or, Acquire similarity differences between adjacent data slots in the fused data column based on the spacing display mode, and adjust the spacing between adjacent data slots according to the similarity differences to obtain an adjusted data column; Acquiring the number of interval slots between adjacent data slots in the fused data sequence according to the slot display mode, and inserting a corresponding number of interval slots between adjacent data slots according to the number of interval slots to obtain an adjusted data sequence, including: Determine that the user terminals corresponding to the adjacent data slots in the fused data column are the same user group, and obtain similarity differences between the data similarity coefficients corresponding to the user groups; Compare the similarity difference value and the benchmark coefficient difference value, and determine the user group whose similarity difference value is greater than the benchmark coefficient difference value as the adjustment group; Determining the number of spacing slots of the adjustment group pair according to the ratio of the similarity difference to the reference coefficient difference; Inserting the number of interval slots between the data slots corresponding to the adjustment groups to obtain an adjustment data column; Acquiring similarity differences between adjacent data slots in the fused data column based on the spacing display mode, and adjusting the spacing between adjacent data slots according to the similarity differences to obtain an adjusted data column, including: Obtaining a similarity difference between data similarity coefficients corresponding to adjacent data slots in the fused data column; Obtaining a spacing adjustment coefficient according to a ratio of the similarity difference value to the reference coefficient difference value, and obtaining an adjustment spacing corresponding to the corresponding adjacent data slots based on a product of the spacing adjustment coefficient and the standard spacing; The spacing between adjacent data slots is adjusted based on the adjustment spacing to obtain an adjusted data column.

2. The method according to claim 1, characterized in that After inserting the number of interval slots between the data slots corresponding to the adjustment group to obtain the adjustment data column, the method further includes: Obtaining the slot spacing between adjacent spacing slots, and determining the spacing slot whose slot spacing is smaller than the reference gap spacing as the target slot; Wherein, the slot spacings between the plurality of adjacent spacing slots corresponding to the same adjustment group are equal; Calculating the spacing difference between the slot spacing and the reference spacing, and obtaining a dimension shift coefficient according to the ratio of the spacing difference to the reference spacing difference; A size adjustment value is obtained according to the product of the size deviation coefficient and the size adjustment weight of the spacing groove, and an adjustment size is obtained based on the difference between the standard size of the target groove and the size adjustment value; The slot size of the target slot is reduced and adjusted based on the adjustment size to obtain an updated adjustment data column.

3. The method according to claim 1, characterized in that In response to the portrait prediction request of the push management end, multiple user ends corresponding to the fused data column are screened to obtain reference users, including: In response to the image prediction request, obtaining a spacing calculation amount according to a sum of a preset reference quantity and a spacing offset value; Acquire the data column length of the fused data column, and obtain the unit spacing according to the ratio of the data column length to the spacing calculation amount; Determine the preset reference number of reference points in the fused data column according to the unit spacing, obtain the data slot closest to the reference point as the reference slot corresponding to it, and determine the user end corresponding to the reference slot as the reference user.

4. The method according to claim 1, characterized in that: Acquiring a data distinction coefficient between each of the remaining user terminals and the reference user, and calculating predicted portrait data corresponding to each of the remaining user terminals according to the reference portrait data and the data distinction coefficient, including: Determine that the fused data column with the gap slot is a type of data column, and obtain the reference user with the closest distance as the type of user corresponding to the other corresponding user terminals; Obtaining the number of slots of the interval slot between the user terminal and the class of users, obtaining a type of offset coefficient according to a ratio of the number of slots to the number of reference slots, and obtaining a data distinction coefficient of the user terminal based on a product of the type of offset coefficient and a type of weight; Determine that the fused data column without the interval slot is a second-category data column, and obtain the reference user with the closest distance as the second-category user corresponding to the other corresponding user terminals; Acquire the interval distance between the user terminal and the two types of users, obtain the second type of offset coefficient according to the ratio of the interval distance to the reference interval distance, and obtain the data distinction coefficient of the user terminal based on the product of the second type of offset coefficient and the second type of weight; According to the product of the data distinction coefficient and the energy consumption demand of each dimensional feature corresponding to the first type of user or the second type of user, the adjustment demand of each dimensional feature corresponding to the corresponding user terminal is obtained; Determine the adjustment trend of each of the user terminals, determine the predicted demand of each of the user terminals according to the adjustment trend and the adjustment demand, and obtain the predicted portrait data corresponding to the user terminal based on the predicted demand corresponding to each of the dimensional features.

5. The method according to claim 4, characterized in that Determining the adjustment trend of each of the user terminals, and determining the predicted demand of each of the user terminals according to the adjustment trend and the adjustment demand, including: Acquire the data slot corresponding to the first type of user or the second type of user on the slot side of the data slot of the user terminal, wherein the slot side includes a decreasing side or an increasing side; Determine that the dimensional feature corresponding to the decreasing side is a decreasing trend, and determine that the dimensional feature corresponding to the increasing side is an increasing trend, wherein the adjustment trend includes a decreasing trend and an increasing trend; The energy consumption demand of the decreasing trend corresponding to the reference user is obtained as a type of demand, and the predicted demand corresponding to the decreasing trend is obtained according to the difference between the type of demand and the adjusted demand; The energy consumption demand corresponding to the increasing trend of the reference user is determined to be a second-class demand, and a predicted demand corresponding to the increasing trend is obtained according to the sum of the second-class demand and the adjusted demand.

6. The method according to claim 1, characterized in that Generating push data corresponding to each of the user terminals according to the user portrait data, and sending the push data to the corresponding user terminals, including: Determine that the dimension feature of the high energy consumption attribute corresponds to a high-demand device, and determine that the dimension feature of the low energy consumption attribute corresponds to a low-demand device; Determine the device capacity corresponding to the high-demand device or the low-demand device according to the energy consumption demand, and generate sub-push data corresponding to the dimensional feature based on the high-demand device or the low-demand device and the device capacity; Generate push data corresponding to each of the user terminals based on the sub-push data, and send the push data to the corresponding user terminals.

7. A data fusion system for the multi-channel data fusion method according to claim 1, characterized in that: include: A capture module, used to capture data from a data source corresponding to each user terminal based on a data acquisition unit, and obtain collected data of multiple dimensional features corresponding to each user terminal; A construction module, used to construct an initial data column corresponding to each of the user terminals, and to arrange and update the initial data column according to the collected data corresponding to each of the dimensional features to obtain a fused data column; An acquisition module, used to acquire fusion indicators of multiple dimensional features corresponding to the user terminals in the fusion data column, and calculate user portrait data corresponding to each of the user terminals according to the fusion indicators; A generation module is used to generate push data corresponding to each of the user terminals based on the user portrait data, and send the push data to the corresponding user terminal.

Images