Power sale quantity prediction method and system based on power consumer group clustering

Abstract

The invention discloses an electricity sale quantity prediction method and system based on power consumer group clustering, and relates to the field of electricity sale quantity prediction, and the method comprises the steps: obtaining a monthly electricity sale quantity customer electricity consumption condition ranking table, and dividing large consumers and small consumers; performing enterprise group division on the selected large users, and establishing a large user enterprise directory library; clustering and grouping are carried out based on a large-user enterprise directory library, and large-user electricity sale quantity prediction is carried out; the selected micro users are classified, and the electricity sales quantity of the micro users is predicted; summarizing the electricity sale quantity prediction results of the large users and the micro users according to different weight proportions to obtain an overall electricity sale quantity prediction result; according to the method, the problem of prediction result error caused by different weight ratios due to excessive overall data is solved; the problem that the electric quantity of a single user is greatly fluctuated due to the influence of enterprise group area layout and market strategies is solved, and the fluctuation of single user observation is reduced.

Description

Technical Field

[0001] The present invention relates to the field of power sales volume forecasting, and in particular to a power sales volume forecasting method and system based on power user group clustering. Background Art

[0002] With the continuous development of the power market, the importance of electricity sales forecasting has become increasingly prominent. Both power companies and regulatory agencies need to accurately forecast electricity sales to optimize resource allocation, improve economic benefits, ensure the safety of power grid operation, and provide data support for the formulation of corresponding energy policies.

[0003] At present, the prediction of electricity sales is mostly based on the overall prediction of electricity sales. Even if it is down to the influencing factors of electricity sales, it is mostly a whole-body disassembly application without detailed focus and specific disassembly sorting method. At the same time, if the electricity sales are broken down and predicted by individual users, the electricity consumption of a single user will be affected by many factors, resulting in volatility in the observation. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by the present invention is: how to improve the accuracy and practicality of electricity sales forecasting through refined user data processing and systematized forecasting process.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for predicting power sales based on power user group clustering, comprising:

[0008] Obtain a monthly electricity sales customer electricity usage ranking table and divide it into large and small users;

[0009] Divide the selected major users into enterprise groups and establish a major user enterprise directory database;

[0010] Based on the large-user enterprise directory, clustering and grouping are carried out to predict the large-user electricity sales volume;

[0011] Classify the selected micro-users and predict their electricity sales;

[0012] The electricity sales forecast results of large and small users are summarized according to different weight ratios to obtain the overall electricity sales forecast result.

[0013] As a preferred solution of the electricity sales forecasting method based on power user group clustering, the following is the solution:

[0014] The clustering and grouping based on the large user enterprise directory database includes:

[0015] Use the dictionary word segmentation algorithm to extract the keywords of user names and count the number of useless words. After deletion, a large user list with useless words deleted is initially obtained.

[0016] Count the frequency of word occurrences and sort them from high to low;

[0017] Eliminate entries whose frequency reaches a preset threshold;

[0018] Manually confirm the uniqueness of user information.

[0019] As a preferred solution of the electricity sales forecasting method based on power user group clustering, the following is the solution:

[0020] The large user electricity sales forecasting includes:

[0021] The large user enterprise directory database after clustering is grouped by city and industry category, and the corresponding auxiliary table is generated;

[0022] Add primary key level fields to the generated auxiliary table;

[0023] Merge all auxiliary tables into one table and modify it;

[0024] For each merged and modified record, the sum of the electricity sales of users in the corresponding user number group is counted to generate monthly enterprise electricity sales data.

[0025] As a preferred solution of the electricity sales forecasting method based on power user group clustering, the following is the solution:

[0026] The classifying of the selected micro-users and prediction of the power sales of the micro-users includes:

[0027] The selected micro-users are divided into four categories: primary industry, secondary industry, tertiary industry, and residential electricity consumption by industry name, and their electricity sales are predicted using multivariate regression.

[0028] As a preferred solution of the electricity sales forecasting method based on power user group clustering, the following is the solution:

[0029] The division of large users and small users includes:

[0030] Obtain a monthly electricity sales customer electricity consumption ranking table, select the top 60% of users ranked by total electricity sales as large users, and select the remaining 40% of users ranked by total electricity sales as small users.

[0031] As a preferred solution of the electricity sales forecasting method based on power user group clustering, the following is the solution:

[0032] The division of the selected major users into enterprise groups and establishment of a major user enterprise directory database includes:

[0033] The user lists classified as large users are named and stored in a standardized manner, and the user list data of large users are stored in a database according to the indicator names of "customer name", "customer address", "industry major category code", "industry medium category code", "industry small category code", "industry major category name", "industry medium category name", "industry small category name", "city", "city code", "value type", "value" and "unit" to obtain a large user enterprise directory database.

[0034] As a preferred solution of the electricity sales forecasting method based on power user group clustering, the following is the solution:

[0035] The power sales forecast results of large users and small users are summarized according to different weight proportions to obtain the overall power sales forecast results including:

[0036] Different weights are set according to the proportion of large and small users in the overall electricity sales;

[0037] Multiply the forecast results of large users and small users by their corresponding weights respectively, and add them together to get the overall electricity sales forecast result;

[0038] The summarized results will be generated into a report, including the current month / cumulative value of overall electricity sales, the value for the same period last year, and the year-on-year growth rate.

[0039] In a second aspect, an embodiment of the present invention provides a power sales amount prediction system based on power user group clustering, comprising:

[0040] The user classification module is used to obtain the monthly electricity sales customer electricity usage ranking table and classify large users and small users;

[0041] The group division module is used to divide the selected major users into enterprise groups and establish a major user enterprise directory database;

[0042] The large user electricity sales forecasting module is used to perform clustering and grouping based on the large user enterprise directory to forecast the large user electricity sales;

[0043] A small user electricity sales prediction module is used to classify the selected small users and predict the electricity sales of the small users;

[0044] The overall electricity sales forecast module is used to summarize the electricity sales forecast results of large users and small users according to different weight proportions to obtain the overall electricity sales forecast result.

[0045] In a third aspect, an embodiment of the present invention provides a computing device, including:

[0046] Memory and processor;

[0047] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the one or more programs are executed by the one or more processors, the one or more processors implement the power sales forecasting method based on power user group clustering as described in any embodiment of the present invention.

[0048] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, can implement the electricity sales amount prediction method based on power user group clustering.

[0049] Beneficial effects of the present invention: The present invention solves the problem that a single user will be greatly affected by fluctuations; solves the problem that the weight ratio has different influences, and divides the influences of different proportions into different sizes. This method divides all users under it into two categories according to the capacity level, one category is large user data, accounting for more than 60%, and the other category is the 40% of users who are ranked at the bottom of the overall proportion, which has little impact on the overall power sales; solves the problem of dividing large user data into enterprise groups and establishing an enterprise directory. This method summarizes the enterprise groups of the top 60% of users, processes the correspondence between "enterprise code" and "user number", "power plant number", "industry classification", and "city", and cleans and summarizes the data; solves the manual processing method when it is impossible to completely rely on the big data matching algorithm for comparison due to complex situations such as "typos in enterprise (user) names", "changes in enterprise names", "irregular entry of user names", and "group-type enterprises"; solves the overall processing of the second category of users who are ranked at the bottom of the overall proportion. It is divided into primary industry, secondary industry, tertiary industry and residential electricity consumption, and an overall forecast is made using its industrial attributes and then summarized with the forecast of large users; this solves the problem of establishing a systematic forecast system for the overall electricity sales volume. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0051] Figure 1 It is an overall flow chart of the power sales forecasting method based on power user group clustering according to the first embodiment of the present invention;

[0052] Figure 2It is a schematic diagram of the back-end initial screening results after filtering out useless keywords in a simulation example of the power sales forecasting method based on power user group clustering according to the first embodiment of the present invention;

[0053] Figure 3 It is a schematic diagram of the result after adding user keywords in the simulation example of the power sales prediction method based on power user group clustering according to the first embodiment of the present invention;

[0054] Figure 4 It is an auxiliary representation diagram in a simulation example of the power sales forecasting method based on power user group clustering described in the first embodiment of the present invention. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0057] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0058] Example 1

[0059] Reference Figure 1 , which is the first embodiment of the present invention, and provides a method for predicting power sales based on power user group clustering, comprising:

[0060] S1: Obtain the monthly electricity sales customer electricity consumption ranking table and divide it into large users and small users;

[0061] S2: Divide the selected major users into enterprise groups and establish a major user enterprise directory database;

[0062] S3: Clustering and grouping based on the large-user enterprise directory to predict the power sales volume of large users;

[0063] S4: classify the selected micro-users and predict the electricity sales of the micro-users;

[0064] S5: Summarize the power sales forecast results of large users and small users according to different weight proportions to obtain the overall power sales forecast result.

[0065] In the embodiment of the present application, the division of large users and small users in the above step S1 includes:

[0066] Obtain a monthly electricity sales customer electricity consumption ranking table, select the top 60% of users ranked by total electricity sales as large users, and select the remaining 40% of users ranked by total electricity sales as small users.

[0067] In the embodiment of the present application, in the above step S2, the selected major users are divided into enterprise groups, and the establishment of a major user enterprise directory database includes:

[0068] The user lists classified as large users are named and stored in a standardized manner, and the user list data of large users are stored in a database according to the indicator names of "customer name", "customer address", "industry major category code", "industry medium category code", "industry small category code", "industry major category name", "industry medium category name", "industry small category name", "city", "city code", "value type", "value" and "unit" to obtain a large user enterprise directory database.

[0069] In the embodiment of the present application, the clustering and grouping based on the large user enterprise directory in the above step S3 includes:

[0070] Use the dictionary word segmentation algorithm to extract the keywords of user names and count the number of useless words. After deletion, a large user list with useless words deleted is initially obtained.

[0071] In the embodiment of the present application, word segmentation technology is used to pre-process the enterprise directory and user profile information table to extract keywords from the enterprise / user names.

[0072] For example, the text is segmented into List <word>words = WordSegmenter.segWithStopWords("Angang Group Mining Co., Ltd."); divide the company name into: [Angang, Group, Mining, Co., Ltd."

[0073] In another possible implementation, the HanLP word segmentation technology can be used to perform word segmentation processing on the company name. HanLP is a Java-based natural language processing toolkit that supports multiple word segmentation algorithms and has high accuracy and flexibility.

[0074] For example, HanLP is used to segment "Ansteel Group Mining Co., Ltd.":

[0075] List <term>terms = HanLP.segment("Ansteel Group Mining Co., Ltd.");

[0076] for (Term term : terms) {

[0077] System.out.println(term.word);

[0078] }

[0079] The word segmentation result is:

Ansteel, Group, Mining, Co., Ltd.

[0080] The advantage of HanLP's word segmentation technology lies in its powerful dictionary support and various word segmentation modes (such as standard word segmentation, NLP word segmentation, index word segmentation, etc.), which can adapt to the needs of different scenarios and improve the accuracy and efficiency of enterprise name word segmentation.

[0081] Statistically analyze the word segmentation frequency and sort it in descending order.

[0082] Exemplarily, the word segments

Co., Ltd.

Group

[0083] Delete the entries whose occurrence frequency reaches the preset threshold;

[0084] Exemplarily, after deleting the useless entries from "Ansteel Group Mining Co., Ltd.", the keyword "Ansteel Mining" is obtained.

[0085] Manually confirm the uniqueness of user information.

[0086] It should be noted that the big data screening technology applied in this method is mainly based on the dictionary-based word segmentation algorithm. However, when there are complex situations such as "misspelled words in the enterprise (user) name", "enterprise name change", "irregular user name entry", "group enterprise", etc., it is impossible to rely entirely on the big data algorithm for comparison. At this time, to improve the matching success rate of the enterprise directory, it is necessary to determine the true name, address and business situation of the enterprise through manual verification to increase the comparison surface between the enterprise directory and the user file information and determine the uniqueness of its enterprise group.

[0087] In the embodiment of this application, the powerquery function in Excel can be used to verify the keyword fields such as the user name, industry major category, industry middle category, affiliated city, voltage level, rated capacity, etc. corresponding to each user number to see if there are situations such as "one user corresponding to different user names / industry major categories / industry middle categories / affiliated cities / voltage levels / rated capacities" at different time points. When such a situation exists, it is necessary to verify and distinguish the possible reasons for this situation and clarify the handling method.

[0088] Add "User Keywords" - Clean "User Name": remove "branch", "company", "shares", "limited", "liability" (clean "branch" first), remove "provincial" and "municipal" names, and replace "Chinese and English brackets, -" with "*".

[0089] Generate "user keywords", segment the cleaned "user name", and perform fuzzy matching based on the segmentation to generate the initial screening value of "user keywords". Manually review the initial screening value to ensure the uniqueness of the keyword pointing and generate the review value of "user keywords".

[0090] If one "user keyword" corresponds to multiple "user numbers", then "enterprise keyword = user keyword".

[0091] If a "user keyword" corresponds to a "user number", then "enterprise keyword = user abbreviation".

[0092] In another possible implementation, natural language processing (NLP) technology may be used to further improve the uniqueness and matching accuracy of user information:

[0093] Combined with NLP technology, such as named entity recognition (NER), key entity information in enterprise names (such as company name, address, industry category, etc.) is automatically extracted and compared with standard databases. This method can handle complex user name formats and structures more finely.

[0094] In another possible implementation, graph database technology can be used to further improve the uniqueness and matching accuracy of user information:

[0095] Use graph databases (such as Neo4j) to build a corporate relationship network, and use nodes and edges to represent the relationships between companies (such as parent companies and subsidiaries, partnerships, etc.). This method can more intuitively identify and match corporate groups through graph query and analysis.

[0096] In another possible implementation, data consistency checking and automatic correction can be used to further improve the uniqueness and matching accuracy of user information:

[0097] Develop automated tools to regularly check and correct data consistency issues.

[0098] For example, name changes of the same user at different time points are automatically detected and corrected according to preset rules. At the same time, business logic and rule engines can be combined to achieve automated data quality improvement.

[0099] In the embodiment of the present application, the large user power sales forecasting in the above step S3 includes:

[0100] The large user enterprise directory database after clustering is grouped by city and industry category, and the corresponding auxiliary table is generated;

[0101] Specifically, generate the auxiliary tables of [Provincial / City] & [Major Category / Medium Category / Minor Category]. Each auxiliary table contains the following fields: Enterprise keyword┃Industry minor category code┃Industry minor category name┃Industry medium category code┃Industry medium category name┃Industry major category code┃Industry major category name┃City code┃City name┃User number group.

[0102] Tables 1 to 3 are auxiliary tables for [small / medium / large categories of cities and prefectures]. The generation process is: group [user directory] by [enterprise keywords, city name, small / medium / large categories of industries]:

[0103] The generation rule of the [Industry Subcategory] field is: if the number of [Industry Subcategory] after deleting duplicates in a group is 1 (i.e., the industry subcategory of each user in the group is the same), then the [Industry Subcategory] of the group = the subcategory after deleting duplicates; if the number of [Industry Subcategory] of the group after deleting duplicates is greater than 1 (i.e., the industry subcategory of each user in the group is not completely consistent), then the [Industry Subcategory] of the group is recorded as: -

[0104] The generation rule of the [Industry Middle Category] field is: if the number of [Industry Middle Category] of a group after grouping is 1, then the [Industry Middle Category] of the group = Middle Category after deleting duplicates; if the number of [Industry Middle Category] of the group after deleting duplicates is greater than 1, then the [Industry Middle Category] of the group is recorded as: -

[0105] The generation rule of the [Industry Category] field is: if the number of [Industry Category] of a group after deleting duplicates is 1, then the [Industry Category] of the group = the category after deleting duplicates; if the number of [Industry Category] of the group after deleting duplicates is greater than 1, then the [Industry Category] of the group is recorded as: -

[0106] The generation rule of the [User Number Group] field is: after grouping, delete the duplicate [User Number] of a group and connect them into a string with a separator (such as a semicolon). For example, under the group [a certain industry category of a certain city of a certain enterprise keyword], the corresponding user numbers are 001, 002, and 003 respectively, then the generated [User Number Group] value should be: 001; 002; 003

[0107] Tables 4 to 6 are auxiliary tables for [Provincial Small / Medium / Major Categories]. The generation process is: group [User Directory] by [Enterprise Keywords, Industry Small / Medium / Major Categories]:

[0108] The generation rules for the fields [Industry Sub-category], [Industry Medium-category], [Industry Major-category], and [User Number Group] are the same as above.

[0109] The generation rule of the [city name] field is: if the number of duplicate-deleted [city name] in a group is 1, then the [city name] of the group = the city name after duplicate removal; if the number of duplicate-deleted [city name] in the group is greater than 1, then the [city name] of the group is recorded as: -

[0110] Add primary key level fields to the generated auxiliary table;

[0111] Specifically, the [Primary Key Level] field is added to all six tables, and the values ​​of this field are: 1_DSXL, 2_DSZL, 3_DSDL, 4_QSXL, 5_QSZL, 6_QSDL;

[0112] Merge all auxiliary tables into one table and modify it;

[0113] Specifically, after the merger, delete duplicates according to the [Enterprise Keywords, Cities, Industry Subcategories, Industry Medium Categories, Industry Major Categories] fields and modify the [Primary Key Level] field. Modify the rules: If there are multiple different [Primary Key Level] values ​​in the same group, sort these values ​​in reverse order and merge them together, separated by commas.

[0114] For each merged and modified record, the sum of the electricity sales of users in the corresponding user number group is counted to generate monthly enterprise electricity sales data.

[0115] It should be noted that each record is equivalent to the corporate information of an enterprise. For each enterprise, the sum of the electricity sales of users corresponding to the corresponding [User Number Group] is counted to generate monthly [Enterprise Electricity Sales] data such as the current month / cumulative value, the current month / cumulative value of the previous year, and the current month / cumulative year-on-year growth rate.

[0116] In the embodiment of the present application, the selected micro-users are classified in the above step S4, and the power sales amount of the micro-users is predicted, which includes:

[0117] The selected micro-users are divided into four categories: primary industry, secondary industry, tertiary industry, and residential electricity consumption by industry name, and their electricity sales are predicted using multivariate regression.

[0118] Specifically, data preparation:

[0119] Extract the historical electricity sales data of micro-users and related explanatory variables (such as temperature, season, weekdays / weekends, etc.). Ensure data quality and handle missing values ​​and outliers.

[0120] Model building: Use the scikit-learn library in Python or other statistical analysis tools to build a multivariate regression model.

[0121] Use historical data for model training and train models for each industry category separately.

[0122] Model prediction: Use the trained model to predict the future electricity sales of small users. Generate prediction results, including the current month / cumulative value, the same period last year, and the year-on-year growth rate.

[0123] In another possible implementation, the prediction results may be evaluated to calculate indicators such as model accuracy, mean square error (MSE), etc. According to the evaluation results, the model parameters may be adjusted or a more appropriate prediction method may be selected.

[0124] In the embodiment of the present application, in the above step S5, the power sales forecast results of large users and small users are summarized according to different weight proportions, and the overall power sales forecast result is obtained, including:

[0125] Different weights are set according to the proportion of large and small users in the overall electricity sales;

[0126] For example, it is assumed that the electricity sales to large users account for 70% of the total electricity sales, and that to small users account for 30%.

[0127] Multiply the forecast results of large users and small users by their corresponding weights respectively, and add them together to get the overall electricity sales forecast result;

[0128] Generate a report with the summarized results, including the current month / cumulative value of overall electricity sales, the value of the same period last year, and the year-on-year growth rate;

[0129] It should be noted that the report should list in detail the prediction results and weights of large and small users to facilitate subsequent analysis and decision-making.

[0130] In another possible implementation, the forecast trend of the overall electricity sales volume is analyzed for the obtained forecast results to see whether it conforms to historical rules and market expectations. If the forecast results deviate greatly from the actual situation, it is necessary to further adjust the model parameters or weight settings. Based on the actual electricity sales volume data, the forecast model is continuously fed back and optimized to improve the accuracy of the forecast.

[0131] The above is a schematic scheme of the method for predicting the power sales volume based on the clustering of power user groups of this embodiment. It should be noted that the technical scheme of the power sales volume prediction system based on the clustering of power user groups and the technical scheme of the method for predicting the power sales volume based on the clustering of power user groups belong to the same concept. For the details not described in detail in the technical scheme of the power sales volume prediction system based on the clustering of power user groups in this embodiment, please refer to the description of the technical scheme of the method for predicting the power sales volume based on the clustering of power user groups.

[0132] The power sales forecasting system based on power user group clustering in this embodiment includes:

[0133] The user classification module is used to obtain the monthly electricity sales customer electricity usage ranking table and classify large users and small users;

[0134] The group division module is used to divide the selected major users into enterprise groups and establish a major user enterprise directory database;

[0135] The large user electricity sales forecasting module is used to perform clustering and grouping based on the large user enterprise directory to forecast the large user electricity sales;

[0136] A small user electricity sales prediction module is used to classify the selected small users and predict the electricity sales of the small users;

[0137] The overall electricity sales forecast module is used to summarize the electricity sales forecast results of large users and small users according to different weight proportions to obtain the overall electricity sales forecast result.

[0138] This embodiment further provides a computing device, which is applicable to the case of a method for predicting power sales based on power user group clustering, and includes:

[0139] A memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the power sales amount prediction method based on power user group clustering as proposed in the above embodiment.

[0140] This embodiment further provides a storage medium on which a computer program is stored. When the program is executed by a processor, the power sales forecasting method based on power user group clustering as proposed in the above embodiment is implemented.

[0141] The storage medium proposed in this embodiment and the power sales prediction method based on power user group clustering proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0142] Example 2

[0143] Reference Figure 2-Figure 4 , which is an embodiment of the present invention, provides a method for predicting electricity sales based on power user group clustering. In order to verify the beneficial effects of the present invention, a scientific demonstration is carried out through simulation experiments.

[0144] An example demonstration is given using the electricity sales data of Liaoning Province by user:

[0145] [User Electricity Sales] Original table storage and field cleaning:

[0146] The acquired original data of electricity sales per user is stored in the database for processing. The field names are standardized as follows:

[0147] Added [Province] and [Province Code] fields, [Province] is [Liaoning Province], [Province Code] is [210000], and added [City] and [City Code] fields. By [identifying which city name in Liaoning Province is included in the "Electricity Address", extract the [City] and generate the [City Code] (6 digits) according to the [Administrative Division Code]. If the address is empty, set it to null.

[0148] Added [Industry Category Code] and [Industry Category Code] fields. [Industry Category Code] and [Industry Category Code] are obtained through the standard industry codes issued by the state. Rename [Industry] in the table to [Industry Category Name].

[0149] Added the [Indicator Name] field. The indicator names are [Contract capacity, installed capacity of self-provided power plants, self-generated and self-used electricity, electricity sales, and average price of electricity sold to households] in the table. Change [Capacity, increase or decrease over the same period*] in [Contract capacity] in table [Indicator name] to [Capacity at the end of period, increase or decrease over the same period at the end of period]; change [Capacity, same period last year, increase or decrease over the same period*] in [Installed capacity of self-owned power plants] in table [Indicator name] to [Capacity at the end of period, capacity at the end of last year, increase or decrease over the same period at the end of period]; change [Cumulative this year, last year, increase or decrease over the same period*] in [Self-generated and self-used electricity] in table [Indicator name] to [Cumulative value, cumulative value of last year, cumulative increase or decrease over the same period]; change [Cumulative this year, increase or decrease over the same period*, proportion of the company's total electricity sales (%)] in [Electricity sales] in table [Indicator name] to [Cumulative value, cumulative increase or decrease over the same period, cumulative proportion]; change [Cumulative this year, increase or decrease over the same period*] in [Average price of electricity sold to households] in table [Indicator name] to [Cumulative value, cumulative increase or decrease over the same period].

[0150] Added [Value Type], [Value], [Unit] fields. [Value Type], [Value]: [End of Period Capacity, End of Previous Year Capacity, End of Period Increase / Decrease, Cumulative Value, Cumulative Value of Previous Year, Cumulative Increase / Decrease, Cumulative Proportion] are reversed into [Value Type], [Value] fields. Specific values ​​assigned under [Value Type] include [End of Period Capacity, End of Previous Year Capacity, End of Period Increase / Decrease, Cumulative Value, Cumulative Value of Previous Year, Cumulative Increase / Decrease, Cumulative Proportion]; [Unit]: [Power Supply Voltage, Contract Capacity, Installed Capacity of Self-provided Power Plant, Self-generated and Self-used Electricity, Electricity Sales, Average Price of Electricity Sold to Households] in [Indicator Name] are as follows: kilovolt, kilowatt, kilowatt, billion kWh, billion kWh, yuan / kilowatt-hour. Divide the [value] of [Value type = cumulative value, cumulative value of the previous year, cumulative increase or decrease over the same period] under [Self-generated and self-used electricity] and [Electricity sales] in [Indicator name] by 10^8 (that is, convert it into 100 million kWh), and retain all decimal places; the [Unit] field uses [Value type = cumulative proportion], and the unit = %.

[0151] The result table after the above processing is recorded as t1, and t1 needs to be stored.

[0152] Based on this data table, large user clustering and grouping processing is performed.

[0153] First, we use language to perform big data screening technology to perform preliminary screening of user keywords, and the result is the initial screening result of the backend after filtering out useless keywords, such as Figure 2 shown.

[0154] Based on the initial screening results, manually recheck the [User List]: Confirm the uniqueness of user information:

[0155] [User information] Uniqueness confirmation. Check the key fields such as user name, industry category, industry sub-category, city, voltage level, and rated capacity corresponding to each user number of t1 to see if [a user has different user names / industry categories / industry sub-category / city / voltage level / rated capacity at different time points].

[0156] Add [User Keywords]. Here we take [Angang] as an example to briefly describe the verification process:

[0157] Fuzzy search [Angang] on [User Keywords] to confirm whether [Angang] is the only keyword that can represent [Angang Mining]. In the figure, [Angang] represents the following four customer names at the same time. Check whether its representativeness is accurate. If it is not unique, the keyword can be expanded to review its pointing uniqueness and confirm that the keyword includes all relevant users, such as Figure 3 shown.

[0158] The result table after the above processing is recorded as t2, and t2 needs to be stored.

[0159] [User keyword electricity sales] calculation, generate 6 auxiliary tables [Provincial / City] & [Major / Medium / Minor Category], such as Figure 4 shown.

[0160] The above result is recorded as t3, and t3 needs to be put into storage.

[0161] Each record in t3 is equivalent to the enterprise information of an enterprise. For each enterprise in t3, the sum of the user electricity sales corresponding to the corresponding [User Number Group] is counted to generate the monthly [Enterprise Electricity Sales] data for the current month / cumulative value, the previous year's current month / cumulative value, and the current month / cumulative year-on-year growth rate. The above result table is recorded as t4.

[0162] The predicted value is calculated based on the t4 result, and the prediction processing is performed by different methods using Python statements using vscode.

[0163] By combining the forecast results of 40% of the data of small and micro enterprises with the forecast results of the data of large-user enterprises, the overall electricity sales forecast is obtained.

[0164] 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 preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.< / term> < / word>

Claims

1. A method for predicting electricity sales based on power user group clustering, characterized in that: include: Obtain a monthly electricity sales customer electricity usage ranking table and divide it into large and small users; Divide the selected major users into enterprise groups and establish a major user enterprise directory database; Based on the large-user enterprise directory, clustering and grouping are carried out to predict the large-user electricity sales volume; Classify the selected micro-users and predict their electricity sales; The electricity sales forecast results of large and small users are summarized according to different weight ratios to obtain the overall electricity sales forecast result.

2. The method for predicting power sales based on power user group clustering according to claim 1, characterized in that: The clustering and grouping based on the large user enterprise directory database includes: Use the dictionary word segmentation algorithm to extract the keywords of user names and count the number of useless words. After deletion, a large user list with useless words deleted is initially obtained. Count the frequency of word occurrences and sort them from high to low; Eliminate entries whose frequency reaches a preset threshold; Manually confirm the uniqueness of user information.

3. The method for predicting power sales based on power user group clustering according to claim 2, characterized in that: The large user electricity sales forecasting includes: The large user enterprise directory database after clustering is grouped by city and industry category, and the corresponding auxiliary table is generated; Add primary key level fields to the generated auxiliary table; Merge all auxiliary tables into one table and modify it; For each merged and modified record, the sum of the electricity sales of users in the corresponding user number group is counted to generate monthly enterprise electricity sales data.

4. The method for predicting power sales based on power user group clustering according to claim 3, characterized in that: The classifying of the selected micro-users and prediction of the power sales of the micro-users includes: The selected micro-users are divided into four categories: primary industry, secondary industry, tertiary industry, and residential electricity consumption by industry name, and their electricity sales are predicted using multivariate regression.

5. The method for predicting power sales based on power user group clustering according to claim 4, characterized in that: The division of large users and small users includes: Obtain a monthly electricity sales customer electricity consumption ranking table, select the top 60% of users ranked by total electricity sales as large users, and select the bottom 40% of users ranked by total electricity sales as small users.

6. The method for predicting power sales based on power user group clustering according to claim 5, characterized in that: The division of the selected major users into enterprise groups and establishment of a major user enterprise directory database includes: The user lists classified as large users are named and stored in a standardized manner. The user list data of large users are stored in the database according to the indicator names of "customer name", "customer address", "industry major category code", "industry medium category code", "industry minor category code", "industry major category name", "industry medium category name", "industry minor category name", "city", "city code", "value type", "value" and "unit" to obtain the large user enterprise directory database.

7. The method for predicting power sales based on power user group clustering according to claim 6, characterized in that: The power sales forecast results of large users and small users are summarized according to different weight proportions to obtain the overall power sales forecast results including: Different weights are set according to the proportion of large and small users in the overall electricity sales; Multiply the forecast results of large users and small users by their corresponding weights respectively, and add them together to get the overall electricity sales forecast result; The summarized results will be generated into a report, including the current month / cumulative value of overall electricity sales, the value for the same period last year, and the year-on-year growth rate.

8. A system using the method for predicting power sales volume based on power user group clustering as claimed in any one of claims 1 to 7, characterized in that: include: The user classification module is used to obtain the monthly electricity sales customer electricity usage ranking table and classify large users and small users; The group division module is used to divide the selected major users into enterprise groups and establish a major user enterprise directory database; The large user electricity sales forecasting module is used to perform clustering and grouping based on the large user enterprise directory to forecast the large user electricity sales; A small user electricity sales prediction module is used to classify the selected small users and predict the electricity sales of the small users; The overall electricity sales forecast module is used to summarize the electricity sales forecast results of large users and small users according to different weight proportions to obtain the overall electricity sales forecast result.

9. A computing device comprising: Memory and processor; The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the power sales forecasting method based on power user group clustering according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for predicting power sales amount based on power user group clustering as described in any one of claims 1 to 7.

Images