Electric power customer management method based on curve similarity and dynamic clustering

Abstract

The invention relates to an electric power customer management method based on curve similarity and dynamic clustering. The method comprises the following steps: acquiring user side daily load data and power generation side power generation data; preprocessing the user side daily load data and the power generation side power generation data to obtain a user side daily load curve and a power generation side power generation curve; performing similarity calculation on the user-side daily load curve to obtain a comprehensive similarity matrix between users; performing multi-dimensional comprehensive matching on the user side daily load curve and the power generation side power generation curve to obtain a customer packet-power generation side curve comprehensive matching matrix; according to the inter-user comprehensive similarity matrix and the customer package-power generation side curve comprehensive matching matrix, constructing a business oriented customer package; and managing the service oriented customer package according to the service target. The method supports the design of a combined transaction product, achieves the purposes of reducing the purchase cost fluctuation, reducing the deviation electric quantity and improving the contract performance stability, and improves the bargaining capability of a power selling company and a power generation company in the power transaction process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power customer information management of power selling company, and particularly relates to a power customer management method based on curve similarity and dynamic clustering. BACKGROUND

[0002] With the promotion of power marketization reform in China, as one of the main bodies of the power market, the power selling company undertakes multiple functions such as proxying customers to purchase power, formulating power utilization schemes, participating in market transactions, and providing value-added services. In order to form an advantage in competition, the power selling company generally needs to carry out tag management and portrait construction on the proxyed customers to support customer stratified management, price negotiation, contract design, and personalized marketing services.

[0003] Most of the existing power selling customer management methods establish tags based on individual attributes and power utilization behavior characteristics of customers, such as customer power utilization scale, contract compliance rate, price sensitivity, and industry category. The overall load characteristics formed by multiple customers in combination cannot be identified, and moreover, once the tags of the existing customers are generated, they are long-term effective and cannot be dynamically adjusted with the changes in customer behavior, which easily leads to tag aging or even distortion, resulting in the inability to meet the actual needs of the power selling company in combination transactions, cost optimization, price negotiation, and risk control. SUMMARY

[0004] The present application provides a power customer management method based on curve similarity and dynamic clustering to overcome at least one of the above technical problems in the prior art.

[0005] To achieve the above purpose, the embodiments of the present application adopt the following technical solutions: In a first aspect, the present application provides a power customer management method based on curve similarity and dynamic clustering, comprising: obtaining user-side daily load data and power generation data on the power generation side, wherein the user-side daily load data comprises daily load data of multiple users; preprocessing the user-side daily load data and the power generation data on the power generation side to obtain user-side daily load curves and power generation curves on the power generation side; calculating the similarity of the user-side daily load curves to obtain an inter-user comprehensive similarity matrix; performing multi-dimensional comprehensive matching on the user-side daily load curves and the power generation curves on the power generation side to obtain a customer package-power generation side curve comprehensive matching matrix; constructing a business-oriented customer package according to the inter-user comprehensive similarity matrix and the customer package-power generation side curve comprehensive matching matrix; managing the business-oriented customer package according to a business target.

[0006] In one possible implementation of the first aspect, the preprocessing of the user-side daily load data and the generator-side power generation data to obtain the user-side daily load curve and the generator-side power generation curve includes: Data cleaning, missing value imputation, time alignment, and normalization are performed on the user-side daily load data and the generator-side power generation data respectively to obtain the user-side daily load curve and the generator-side power generation curve.

[0007] In one possible implementation of the first aspect, the step of calculating the similarity of the user-side daily load curves to obtain a comprehensive similarity matrix between users includes: Extract the curve features of the user-side daily load curve, and construct a multi-user load curve matrix based on the time point; The morphological features of the multi-user load curve matrix are extracted to construct a comprehensive morphological matrix; Based on the multi-user load curve matrix and the comprehensive morphology matrix, the similarity between users is analyzed to obtain the comprehensive similarity matrix between users.

[0008] In one possible implementation of the first aspect, the step of analyzing the similarity between users based on the multi-user load curve matrix and the comprehensive morphology matrix to obtain the comprehensive similarity matrix between users includes: Use the following formula to analyze trend similarity between users: ; in, For the first Load curve matrix for each customer For the first Load curve matrix for each customer The covariance of the two load curve matrices For the first Standard deviation of the load curve matrix for each customer For the first Standard deviation of the load curve matrix for each customer For customers and customers Trend similarity index; The following formula can be used to analyze the similarity of local change rhythms among users: ; ; ; in, For the first The change in load for a customer between two adjacent moments. and the load amount at the time point of and the time point of , the Euclidean distance of the fluctuation rhythm difference of the first and the second customer, the difference degree of the local change structure of the first and the second customer, the local change rhythm similarity index; the structure form similarity between users is analyzed by using the following formula: ; ; wherein, and are the comprehensive form matrices of the first and the second customer, the Euclidean distance of the overall structure characteristics of the first and the second customer, the structure form difference degree, the structure form similarity index; the trend similarity index, the local change rhythm similarity index and the structure form similarity index are weighted and fused to obtain the comprehensive similarity matrix between users.

[0009] In a possible implementation manner of the first aspect, the multi-dimensional comprehensive matching of the user-side daily load curve and the power generation curve of the power generation side to obtain a customer package-power generation side curve comprehensive matching matrix comprises: constructing a customer package total load according to the user-side daily load curve; analyzing the trend matching degree, the shape matching degree and the dynamic time warping distance matching degree of the customer package total load and the power generation curve of the power generation side to obtain the customer package-power generation side curve comprehensive matching matrix.

[0010] In a possible implementation manner of the first aspect, the constructing a customer package total load according to the user-side daily load curve comprises: the customer package total load is constructed by using the following formula: ; wherein, is the load value of the first customer at the time point of , is all user sets in the customer package, is the customer package total load, The time instant belongs to The total load value of all users in the set.

[0011] In a possible implementation manner of the first aspect, the analysis of the trend matching degree, the shape matching degree and the dynamic time warping distance matching degree of the customer package total load and the power generation side power generation curve, and the obtaining of the customer package-power generation side curve comprehensive matching matrix, comprises: The trend matching degree is analyzed by using the following formula: ; Wherein, is the gth curve of the power generation side, is the Pearson correlation coefficient, is the trend matching degree; According to the customer package total load, the customer package total load fluctuation amplitude and the customer package total load fluctuation strength are calculated, and according to the power generation side power generation curve, the power generation curve fluctuation amplitude and the power generation curve fluctuation strength are calculated; The shape matching degree is analyzed by using the following formula: ; ; Wherein, is the customer package total load fluctuation amplitude, is the customer package total load fluctuation strength, is the power generation curve fluctuation amplitude, is the power generation curve fluctuation strength, and is a weighting coefficient, is a small constant to prevent the denominator from being zero, is the shape difference degree of the customer package total load and the power generation side power generation curve, is the shape matching degree of the customer package total load and the power generation side power generation curve; The dynamic time warping distance matching degree is analyzed by using the following formula: ; Wherein, is the dynamic time warping distance, is the attenuation coefficient, is the dynamic time warping distance matching degree; The trend matching degree, the shape matching degree and the dynamic time warping distance matching degree are weighted and fused to obtain the customer package-power generation side curve comprehensive matching matrix.

[0012] In a possible implementation manner of the first aspect, the construction of the business-oriented customer package according to the inter-user comprehensive similarity matrix and the customer package-power generation side curve comprehensive matching matrix comprises: construct a similarity graph according to the inter-user comprehensive similarity matrix, and obtain a customer initial grouping structure; perform clustering processing on the customer initial grouping structure, and obtain an initial customer package; perform matching degree analysis on the initial customer package and the customer package-generating-side curve comprehensive matching matrix, and obtain a generating-side category matched with the initial customer package; based on the generating-side category, perform iterative optimization on the initial customer package by using a target function, eliminate unsuitable customers, introduce complementary customers, and obtain the business-oriented customer package.

[0013] In a possible implementation manner of the first aspect, the management of the business-oriented customer package according to the business target comprises: identifying a current life cycle in which the business-oriented customer package is located, and assigning a life cycle label representing the current life cycle; judging whether the current life cycle of the business-oriented customer package meets a transfer condition according to a preset life cycle transfer rule, and if so, adjusting the current life cycle of the business-oriented customer package and reassigning the life cycle label.

[0014] In a possible implementation manner of the first aspect, the management of the business-oriented customer package according to the business target further comprises: assigning a comprehensive label to each customer based on at least one life cycle management result, the comprehensive label comprising a basic label and a scenario label; periodically updating and / or event-triggered updating the comprehensive label.

[0015] Compared with the prior art, the present application has at least the following beneficial effects: The power customer management method based on curve similarity and dynamic clustering provided by the present application introduces inter-user similarity matching, so that the customer electricity consumption behavior can be re-described from the perspective of the group, thereby being able to identify the overall load characteristics generated after customer combination, and making the customer label dynamically expand from static attributes.

[0016] Further, the present application matches the customer package with the generating-side output curve to generate different types of load packages, supports the design of combination transaction products, realizes the goal of reducing procurement cost fluctuation, reducing deviation power, and improving contract compliance stability, and thus improves the bargaining power of the power selling company and the power generating company in the power transaction process.

[0017] Further, the present application introduces a life cycle management method into the power selling customer packaging system, provides dynamic maintenance, so that the customer package can automatically maintain optimal load smoothness and scale, and improves the tradability and continuous stability of the customer package.

[0018] In a second aspect, the present application provides a power customer management system based on curve similarity and dynamic clustering, comprising: a data acquisition module configured to acquire user-side daily load data and power generation data on the power generation side, wherein the user-side daily load data comprises daily load data of a plurality of users; a data preprocessing module configured to preprocess the user-side daily load data and the power generation data on the power generation side to obtain user-side daily load curves and power generation curves on the power generation side; a user-side similarity analysis module configured to calculate the similarity of the user-side daily load curves to obtain an inter-user comprehensive similarity matrix; a customer package-power generation side matching module configured to perform multi-dimensional comprehensive matching of the user-side daily load curves and the power generation curves on the power generation side to obtain a customer package-power generation side curve comprehensive matching matrix; a business-oriented customer package construction module configured to construct a business-oriented customer package according to the inter-user comprehensive similarity matrix and the customer package-power generation side curve comprehensive matching matrix; a guided customer package management module configured to manage the business-oriented customer package according to a business target.

[0019] In a third aspect, the present application provides an electronic device, comprising at least one processor and at least one memory, wherein the memory has computer readable instructions stored thereon; the computer readable instructions are executed by one or more processors, so that the electronic device implements the power customer management method based on curve similarity and dynamic clustering as in any one of the implementations of the first aspect.

[0020] In a fourth aspect, the present application provides a storage medium having a computer executable program stored thereon, wherein the computer executable program is configured to cause a computer to execute the power customer management method based on curve similarity and dynamic clustering as in any one of the implementations of the first aspect.

[0021] It can be understood that the system of the second aspect, the electronic device of the third aspect, and the storage medium of the fourth aspect provided above can achieve the beneficial effects as in the first aspect and any possible design thereof, which will not be described here again. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0023] Figure 1 A structural schematic diagram of an electronic device provided for an embodiment of the present application is shown in the figure. Figure 2 A method flowchart of a power customer management method based on curve similarity and dynamic clustering provided for an embodiment of the present application is shown in the figure. Figure 3 A structural block diagram of a power customer management system based on curve similarity and dynamic clustering provided for an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the present application will be described below with reference to the accompanying drawings. In the description of the present application, unless otherwise specified, “ / ” represents an “or” relationship between the objects before and after the “ / ”, for example, A / B can represent A or B; in the present application, “or” is only a description of the relationship between the objects, which means that there can be three relationships, for example, A or B, which means that A exists alone, A and B exist together, and B exists alone, where A and B can be singular or plural. In addition, in the description of the present application, unless otherwise specified, “multiple” means two or more than two. “At least one of the following” or the like means any combination of these items, including any combination of single item or multiple items.

[0025] In addition, in the embodiments of the present application, the words “exemplary” or “for example” are used to mean serving as an example, instance, or illustration. Any embodiment or design presented as “exemplary” or “for example” in the embodiments of the present application should not be interpreted as superior or superior to other embodiments or design solutions. Rather, the use of “exemplary” or “for example” is intended to present the relevant concept in a specific manner for understanding.

[0026] With the advancement of China's power market reform, as one of the main bodies of the power market, the power selling company undertakes multiple functions such as proxying customers to purchase electricity, developing electricity use plans, participating in market transactions, and providing value-added services. In order to form an advantage in competition, the power selling company generally needs to carry out tag management and portrait construction for the proxy customers to support customer stratified management, price negotiation, contract design, and personalized marketing services.

[0027] Most of the existing electricity sales customer management methods are based on customer individual attributes and electricity consumption behavior characteristics to establish labels, such as customer electricity consumption scale, contract compliance rate, price sensitivity and industry category. It is unable to identify the overall load characteristics formed by multiple customers in combination, and moreover, once the label of the existing customer is generated, it is long-term effective and unable to be dynamically adjusted with the change of customer behavior, which is easy to cause label aging or even distortion, resulting in the inability to meet the actual needs of the electricity sales company in combination transaction, cost optimization, price negotiation and risk control.

[0028] In view of this, in one aspect, the embodiment of the present application provides a power customer management method based on curve similarity and dynamic clustering, comprising: acquiring user side daily load data and power generation side power generation data, the user side daily load data comprising daily load data of multiple users; pre-processing the user side daily load data and the power generation side power generation data to obtain user side daily load curve and power generation side power generation curve; performing similarity calculation on the user side daily load curve to obtain inter-user comprehensive similarity matrix; performing multi-dimensional comprehensive matching on the user side daily load curve and the power generation side power generation curve to obtain customer package-power generation side curve comprehensive matching matrix; constructing business-oriented customer package according to the inter-user comprehensive similarity matrix and the customer package-power generation side curve comprehensive matching matrix; and managing the business-oriented customer package according to business targets.

[0029] The power customer management method based on curve similarity and dynamic clustering provided by the embodiment of the present application introduces inter-user similarity matching, so that the customer electricity consumption behavior can be re-described from the perspective of the group, thereby being able to identify the overall load characteristics generated after the combination of customers, and making the customer label dynamically expand from static attributes; by matching the customer package with the power generation side output curve, different types of load packages are generated, supporting the design of combination transaction products, achieving the goals of reducing procurement cost fluctuation, reducing deviation power, and improving contract compliance stability, thereby improving the bargaining power of the electricity sales company in the power transaction process with the power generation company; introduced into the electricity sales customer packaging system, providing dynamic maintenance, so that the customer package can automatically maintain optimal load smoothness and scale, improving the tradability and continuous stability of the customer package.

[0030] In some embodiments, the power customer management method based on curve similarity and dynamic clustering provided by the embodiment of the present application can be executed by any electronic device 20 with data processing capability, such as general-purpose computer, personal computer, notebook computer, switch or tablet computer, etc., and the specific implementation mode of the electronic device 20 is not limited here.

[0031] Figure 1 The hardware structure schematic diagram of the electronic device provided by the embodiment of the present application is shown. The electronic device 20 comprises a processor 210, a memory 220 and a communication interface 230.

[0032] The processor 210 can include one or more processing cores. The processor 210 connects various parts within the electronic device 20 with various interfaces and lines, performs various functions of the electronic device 20 and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 220, and calling data stored in the memory 220. Alternatively, the processor 210 can be implemented in at least one of a hardware form of a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processing (DSP), a field-programmable gate array (FPGA), a programmable logic array (PLA).

[0033] The memory 220 can include a random access memory (RAM) and can also include a read-only memory (ROM). Alternatively, the memory 220 includes a non-transitory computer-readable storage medium. The memory 220 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 220 can include a program storage area. The program storage area can store instructions for implementing an operating system, instructions for implementing at least one function (such as a data processing function, a data storage function and a display push function, etc.), instructions for implementing each of the above-mentioned method embodiments, etc.

[0034] The communication interface 230 is configured to communicate with other devices, equipment or communication networks, such as data storage devices, image processing equipment or Ethernet, a radio access network (RAN), a wireless local area network (WLAN), etc.

[0035] In physical implementation, each of the above-mentioned devices (such as the processor 210, the memory 220 and the communication interface 230) can be a device in the same device (such as a notebook computer). Alternatively, at least two of them can be arranged in the same device, i.e. as different devices in a device, such as the deployment of devices or components in a distributed system.

[0036] It is to be understood that the structure illustrated in the embodiments does not constitute a specific limitation on the electronic device 20. In other embodiments of the application, the electronic device 20 can include more or fewer components than shown, or combine certain components, or split certain components, or different arrangement of components. The components shown can be implemented in hardware, software, or a combination of software and hardware.

[0037] An electricity customer management method based on curve similarity and dynamic clustering is provided in the embodiments of the present application, which is described below in combination with the drawings of the specification.

[0038] As shown in the Figure 2 An electricity customer management method based on curve similarity and dynamic clustering is provided in the embodiments of the present application, which can but not limited to include: Step S1: Obtain user side daily load data and power generation side power generation data, wherein the user side daily load data includes daily load data of multiple users.

[0039] In the specific implementation process, the manner of obtaining the user side daily load data and the power generation side power generation data in the embodiments of the present application can but not limited to include obtaining from a power selling company internal data system, a dispatching platform interface, a data service platform or a power trading platform, which is not limited here.

[0040] In the specific implementation process, the user side daily load data in the embodiments of the present application can but not limited to include day-ahead load curve, voltage level, industry type, capacity scale, production operation cycle and historical power consumption and the like; the power generation side power generation data in the embodiments of the present application can but not limited to include photovoltaic output data, wind power output processing, thermal power unit typical output characteristics and maintenance plan and the like, which is not limited here.

[0041] In addition, it should be noted that in addition to obtaining the user side daily load data and the power generation side power generation data, the electricity trading side data can also be obtained in the embodiments of the present application, which can but not limited to include annual / monthly medium and long term contract information, deviation assessment rules, spot market clearing power, clearing price and multi-period quotation data and the like, which is not limited here.

[0042] Step S2: Preprocess the user side daily load data and the power generation side power generation data to obtain user side daily load curve and power generation side power generation curve.

[0043] In a feasible implementation scheme, the preprocessing of the user side daily load data and the power generation side power generation data to obtain the user side daily load curve and the power generation side power generation curve in the embodiments of the present application can but not limited to include: The user side daily load data and the power generation side power generation data are respectively subjected to data cleaning, missing value interpolation, time alignment and normalization processing to obtain the user side daily load curve and the power generation side power generation curve.

[0044] In the specific implementation process, the data cleaning in the embodiment of the application is used for identifying and removing abnormal data, so that the cleaned data meets the curve continuity and reasonable trend, and the method of data cleaning can include but is not limited to noise point removal, offset correction and metering anomaly detection, and the like, without limitation.

[0045] The missing value interpolation in the embodiment of the application can be but not limited to processed by using time series interpolation, similar day completion and the like. The interpolation rule can be but not limited to processed according to the length of missing period, in which: For isolated single point missing, direct interpolation is used for completion; for continuous missing and missing section with duration less than 1 hour, time series interpolation is used and supplemented by a smoothing algorithm for repair; for continuous missing section with duration greater than or equal to 1 hour, a reference day is selected according to the load curve shape similarity, and the missing period data is reconstructed by using the similar day completion process to ensure the stability of the overall structure of the load curve.

[0046] The time alignment in the embodiment of the application is used for mapping different granularity data to a unified time sequence, which can be but not limited to using a segmented average method, a linear interpolation method and the like to ensure the consistency of the input data structure.

[0047] The normalization processing in the embodiment of the application can be but not limited to using the following formula: ; The normalization processing is used for eliminating the dimension difference of different customers, and the normalized data is mapped to the [0, 1] interval, which is convenient for subsequent curve shape comparison.

[0048] Step S3: Similarity calculation is performed on the user side daily load curve to obtain a user comprehensive similarity matrix.

[0049] In a feasible implementation scheme, the similarity calculation on the user side daily load curve to obtain a user comprehensive similarity matrix in the embodiment of the application can include but not limited to: Curve features of the user side daily load curve are extracted, time points are taken as a reference, and a multi-user load curve matrix is constructed.

[0050] In the specific implementation process, the multi-user load curve matrix constructed in the embodiment of the application can be composed of a plurality of customers and a daily load curve of each customer, as shown in the following formula: ; Wherein, n is a customer, and m is a time.

[0051] In the specific implementation process, m in the embodiment of the application can be 96, that is, one day is divided into 96 time points according to 15 min as a time period.

[0052] The morphological feature of the multi-user load curve matrix is extracted, and a comprehensive morphological matrix is constructed.

[0053] In the specific implementation process, the morphological feature of the multi-user load curve matrix in the embodiment of the application is extracted, mainly extracting indexes reflecting the overall shape and local change characteristics of the curve, which can include but are not limited to peak power, valley power, peak-valley difference, load rate, volatility and change rate.

[0054] The peak power is , defined as the maximum value of the load curve in a day, .

[0055] The valley power is , defined as the minimum value of the load curve in a day, .

[0056] The peak-valley difference is , defined as the difference between the peak power and the valley power, used to reflect the load fluctuation intensity, .

[0057] The load rate is , used to reflect the smoothness of the load operation, , wherein is the average load.

[0058] The volatility is , used to reflect the overall volatility of the load curve around the average value, . Wherein, , indicating that there are m time periods in total, and the time length of each time period is equal, such as 96 time periods, each time period being 15 min.

[0059] The change rate is , used to reflect the intensity of the load change, . Wherein, , indicating that there are m time periods in total, and the time length of each time period is equal.

[0060] In the specific implementation process, after the morphological features are extracted, the multi-user load curve matrix can also be compressed by principal component according to the morphological features, that is, dimension reduction processing is performed, through the following formula: ; Wherein, is the load curve matrix after dimension reduction. compressed principal component matrix (usually ), a principal component analysis algorithm for feature decomposition of m-dimensional time series, a dimensional multi-user load curve matrix.

[0061] The principal component analysis algorithm specifically includes: zero-mean processing on each column of the matrix , calculating a zero-mean covariance matrix, performing eigenvalue decomposition on the covariance matrix, and selecting the first characteristic vectors arranged in descending order of eigenvalues to form a projection matrix , so as to map the original matrix to a low-dimensional space.

[0062] The embodiment of the application can reduce the computational complexity, remove noise components, and strengthen the main morphological trend by performing principal component compression on the multi-user load curve matrix.

[0063] In the specific implementation process, after the principal component compression on the multi-user load curve matrix, the embodiment of the application fuses the features and morphological features to construct a comprehensive morphological matrix, as shown in the following formula: wherein, is the th principal component feature vector corresponding to the th customer; is the peak power of the th customer; is the valley power of the th customer, is the peak-valley difference of the th customer; is the volatility index of the th customer.

[0064] In the specific implementation process, after the comprehensive morphological matrix is constructed, the embodiment of the application can also perform normalization processing on the comprehensive morphological matrix through the following formula: wherein, is the original value of the th customer in the th feature dimension in the comprehensive morphological matrix , and are respectively the original values of all customers in the ​​​The minimum and maximum values ​​in each feature dimension. The normalized matrix is ​​denoted as... .

[0065] This invention embodiment uses a comprehensive morphological matrix. Performing column-by-column normalization can eliminate the impact of inconsistencies in the dimensions and ranges of various features on distance metrics, thereby improving similarity calculation and clustering results.

[0066] Based on the multi-user load curve matrix and the comprehensive morphology matrix, the similarity between users is analyzed to obtain the comprehensive similarity matrix between users.

[0067] In one feasible implementation, the step of analyzing the similarity between users based on the multi-user load curve matrix and the comprehensive morphology matrix to obtain the comprehensive similarity matrix between users may include, but is not limited to: The following formula can be used to analyze the trend similarity among users and to identify customers with similar overall electricity consumption trends: ; in, For the first Load curve matrix for each customer For the first A load curve matrix for each customer, wherein the load curve matrix is ​​a time series curve; The covariance of the two load curve matrices is used to measure the consistency of trends. For the first Standard deviation of the load curve matrix for each customer For the first The standard deviation of the load curve matrix for each customer is used to reflect the degree of volatility. For customers and customers The trend similarity index has a value range of [-1, 1], and the closer it is to 1, the more consistent the overall trend is.

[0068] The following formula can be used to analyze the similarity of local change rhythms among users, which can be used to identify customers with consistent fluctuation rhythms and help to form customer groups with consistent fluctuation characteristics: ; ; ; in, For the first The change in load for a customer between two adjacent moments; and The first and the The load amount at the moment; The Euclidean distance of fluctuation rhythm difference of the first customer and the second customer; The difference degree of local change structure of the first customer and the second customer, and the smaller the value is, the more similar the load change amount trend of the two customers is; The local change rhythm similarity index, and the closer the value is to 1, the more consistent the load change rhythm of the two customers is. The Euclidean distance of fluctuation rhythm difference of the first customer and the second customer; The difference degree of local change structure of the first customer and the second customer, and the smaller the value is, the more similar the load change amount trend of the two customers is; The local change rhythm similarity index, and the closer the value is to 1, the more consistent the load change rhythm of the two customers is.

[0069] The overall load pattern similarity between different customers is analyzed by using the following formula: ; ; Wherein, and are the comprehensive pattern matrices of the first customer and the second customer respectively; The Euclidean distance of overall structure characteristics of the first customer and the second customer; The structure pattern difference degree; The structure pattern similarity index, and the value range is (0, 1]. The overall load pattern similarity between different customers is analyzed by using the following formula: .

[0070] The trend similarity index, the local change rhythm similarity index and the structure pattern similarity index are weighted and fused to obtain the comprehensive similarity matrix between the users, as shown in the following formula: .

[0071] Wherein, , and are the weights of different similarity dimensions, and need to satisfy the condition ; The final comprehensive similarity of the first customer and the second customer is a dimensional real number matrix. Step S4: The user side daily load curve and the power generation side power generation curve are comprehensively matched in multiple dimensions to obtain a customer package-power generation side curve comprehensive matching matrix.

[0072] Step S4: The user side daily load curve and the power generation side power generation curve are comprehensively matched in multiple dimensions to obtain a customer package-power generation side curve comprehensive matching matrix.

[0073] ​​​​In a feasible implementation, the multi-dimensional comprehensive matching of the user-side daily load curve and the power generation curve in the embodiment of the present application can include, but is not limited to, the following: According to the user-side daily load curve, a customer package total load is constructed.

[0074] The customer package total load can be constructed by using the following formula: ; Wherein, is the load value of the i-th customer at the t-th moment, is the total load value of all customers in the customer package, is the total load value of all customers belonging to the set at the t-th moment.

[0075] The trend matching degree, shape matching degree and dynamic time warping distance matching degree of the customer package total load and the power generation curve are analyzed to obtain the customer package-power generation curve comprehensive matching matrix.

[0076] In a feasible implementation, the analysis of the trend matching degree, shape matching degree and dynamic time warping distance matching degree of the customer package total load and the power generation curve in the embodiment of the present application to obtain the customer package-power generation curve comprehensive matching matrix can include, but is not limited to, the following: The trend matching degree is analyzed by using the following formula: ; Wherein, is the g-th curve of the power generation side (such as a photovoltaic output curve, a wind turbine output curve or a thermal power output curve, etc.), is the Pearson correlation coefficient, is the trend matching degree; According to the customer package total load, the customer package total load fluctuation amplitude and the customer package total load fluctuation intensity are calculated by using the following formula: ; Wherein, is the average load of the customer package; is the load fluctuation amplitude of the customer package; is the root mean square value of the difference between consecutive moments of the customer package, which is used to reflect the total load fluctuation intensity of the customer package, , indicating that there are m time periods in total, and the length of each time period is equal, such as 96 time periods, each time period being 15 minutes. ​​​​

[0077] According to the power generation side power generation curve, the power generation curve fluctuation amplitude and the power generation curve fluctuation strength are calculated in the same way.

[0078] The shape matching degree is analyzed by the following formula: ; .

[0079] Wherein, is the total load fluctuation amplitude of the customer package; is the total load fluctuation strength of the customer package; is the power generation curve fluctuation amplitude; is the power generation curve fluctuation strength; And is the weighted coefficient, satisfying ; is a tiny constant to prevent the denominator from being zero; is the shape difference between the total load of the customer package and the power generation side power generation curve, and the smaller the difference is, the closer the shape is; is the shape matching degree between the total load of the customer package and the power generation side power generation curve, and the larger the value is, the closer the customer package is to the power generation curve in the overall fluctuation form, and the value range is (0, 1].

[0080] In the specific implementation process, when is higher, and the corresponding is smaller, is lower, it can be determined that the customer package is a smooth load, which is more suitable for matching with thermal power or other stable power supply; when on the corresponding of the photovoltaic or wind power output curve is higher, and , is relatively large, it can be determined that the customer package is a fluctuation load package, which is more suitable for participating in the matching and consumption of the photovoltaic or wind turbine output curve.

[0081] The dynamic time warping distance matching degree is analyzed by the following formula: ; Wherein, is the dynamic time warping distance, which is used to measure the curve shape similarity; is the attenuation coefficient, which is used to control the sensitivity of distance and similarity mapping; is the dynamic time warping distance matching degree.

[0082] The embodiments of the present invention analyze the dynamic time warping distance matching degree to identify whether the peak-valley fluctuations with time-series characteristics, such as the morning peak or the evening ramp-up, match the volatility of the photovoltaic / power generation curve.

[0083] In the specific implementation process, the embodiments of the present invention The calculation process includes the original distance matrix. Construct a dynamic cumulative cost matrix The recursive formula, finally calculate.

[0084] In specific implementation, embodiments of the present invention can construct the original distance matrix using the following formula. : .

[0085] in, Package for customers in the first The time-to-generation curve at the first The load difference at any time is not required. This enables similarity point matching analysis after time distortion.

[0086] In specific implementation, the dynamic cumulative cost matrix in this embodiment of the invention... The following recursive formula can be used: .

[0087] The following boundary requirements must be met: .

[0088] The For the destination The minimum cumulative difference.

[0089] In specific implementation, embodiments of the present invention can use the following formula to calculate the dynamic time warp distance: .

[0090] The trend matching degree, shape matching degree, and dynamic time warping distance matching degree are weighted and fused using the following formula to obtain the comprehensive matching matrix of the customer package-generator side curve: ; in, , and For the weighting coefficients, satisfying .

[0091] Step S5: constructing a business-oriented customer package according to the inter-user comprehensive similarity matrix and the customer package-generating-side curve comprehensive matching matrix.

[0092] In a feasible implementation, the constructing a business-oriented customer package according to the inter-user comprehensive similarity matrix and the customer package-generating-side curve comprehensive matching matrix in the embodiment of the present application may, but is not limited to, include the following steps: constructing a similarity graph according to the inter-user comprehensive similarity matrix to obtain an initial grouping structure of customers.

[0093] In a specific implementation process, the similarity graph may be constructed by the following formula in the embodiment of the present application: ; wherein, is the similarity graph; is a set of all customers; is a connection edge that may form a clustering relationship between customers; is an edge weight matrix, used to reflect the similarity of load curves between customers, and is defined as , The higher the value is, the more likely it is to be put into the same customer package.

[0094] performing clustering processing on the initial grouping structure of customers to obtain an initial customer package.

[0095] In a specific implementation process, the initial grouping of customers may be clustered by using spectral clustering, hierarchical clustering or similarity threshold clustering in the embodiment of the present application, so that highly similar customers are put into the same package, so that the initial package has high consistency, and an initial customer package is obtained, as shown in the following formula: .

[0096] performing matching degree analysis on the initial customer package and the customer package-generating-side curve comprehensive matching matrix to obtain a generating-side category matched with the initial customer package.

[0097] In a specific implementation process, the process of performing matching degree analysis in the embodiment of the present application is as follows: obtaining an initial customer package , an initial customer package total load time sequence , a generating-side typical curve set , and performing determination by using the following formula: ; wherein, indicates different generating-side output curve categories, if , it indicates that the customer package curve is closest to the photovoltaic curve; if If the customer package volatility is most consistent with the wind power, then If the customer package smoothness is most consistent with the thermal power, then The meaning is to find the variable corresponding to the maximum value of .

[0098] For example, as shown in Table 1: Table 1 shows the value of under different . If , it indicates that the customer package smoothness is most consistent with the thermal power.

[0099] Based on the power generation side category, the initial customer package is iteratively optimized by using a target function, unsuitable customers are eliminated, complementary customers are introduced, and the business-oriented customer package is obtained.

[0100] In the specific implementation process, the iteratively optimizing the initial customer package by using a target function, eliminating unsuitable customers, introducing complementary customers, and obtaining the business-oriented customer package in the embodiment of the application can but not limited to include: The weight coefficient is constructed in a comprehensive evaluation manner of the similarity of the internal power consumption behavior of the customer package, the matching degree of the customer package load curve and the power generation curve, the overall volatility and smoothness of the customer package, and the penalty term for abnormal volatility, as shown in the following formula: Wherein, , , and are weight coefficients related to the power generation curve type matched with the customer package, which need to meet the following conditions: The weight coefficients related to the power generation curve type are set to adaptively adjust the target function of the customer package with the power generation curve type, and better match the business requirements.

[0101] If Type=PV or WT, the weight reflecting the matching degree of the load curve and the power generation curve is maximum; if Type=Thermal, the weights reflecting the overall volatility and smoothness of the customer package and the weight penalizing abnormal volatility are maximum.

[0102] The is used to quantify the similarity of the customer load curve in the same customer package, and the calculation formula is as follows: ; wherein, is the current number of customer packages.

[0103] The The formula for ensuring the overall volatility and smoothness of the optimized customer package is shown in the following formula: .

[0104] The The formula for avoiding the volatility of the customer package exceeding the threshold value to control the deviation electricity price risk is shown in the following formula: wherein, is the set volatility threshold value.

[0105] The unsuitable customers are eliminated, and the specific process is as follows: For the customer , the marginal contribution of the customer to the customer package is defined, and the formula for calculating the marginal contribution is shown in the following formula: ; If , it means that the customer has a negative effect on the optimization objective function, and the customer can be eliminated. Through , the formula reflects the influence of on the elimination. For the photovoltaic following type package, customers with high midnight load will be preferentially proposed; for the wind and light consumption type package, customers with too smooth load will be preferentially eliminated; for the smooth steady state type package, customers with strong random fluctuations or too large peak-valley difference will be preferentially proposed.

[0106] The complementary type customer is introduced, and the specific process is as follows: The customers that can improve the quality of the customer package are selected from the customer set , and the selection mechanism is shown in the following formula: ; If , the customer is added.

[0107] In the specific implementation process, the size, consistency and type matching of the business-oriented customer package are also constrained. Wherein: The size constraint is shown in the following formula: ; The size constraint is used to ensure the size of the customer package, so as to avoid the occurrence of too small customer package being not tradable or too large customer package being unstable.

[0108] The consistency constraint is shown in the following formula: ; wherein, is a set empirical value.

[0109] The consistency constraint is used to ensure the mutual adaptation of users within the customer package.

[0110] The type matching constraint is shown in the following formula: ; wherein, is a set empirical value.

[0111] The type matching constraint is used to ensure that the customer package meets the definition of its business type.

[0112] The condition for the iteration of the objective function to stop needs to meet the requirement that the current function converges, that is, or reaches the set iteration upper limit.

[0113] Step S6: managing the business-oriented customer package according to the business target.

[0114] In a feasible implementation, the management of the business-oriented customer package according to the business target in the embodiments of the present application can include, but is not limited to, the following: Identifying the current life cycle in which the business-oriented customer package is located, and assigning a life cycle label representing the current life cycle.

[0115] In the specific implementation process, the life cycle of the business-oriented customer package can be divided into four stages: a creation stage, an active stage, a decay stage, and a retirement stage.

[0116] The creation stage refers to the first completion of the construction of the customer package by the dynamic clustering module, and the stable structure has not yet been formed, and the business matching degree evaluation and consistency verification need to be performed.

[0117] The active stage refers to the stable internal structure of the customer package, and the matching degree of the power generation side is at a high level, which can be directly used for power selling side combination transaction, contract matching and deviation management, and is the core operating state in the life cycle.

[0118] The decay stage refers to the obvious downward trend of the internal consistency, matching degree or smoothness of the customer package, indicating that the package structure is deteriorating, and the optimization or reconstruction mechanism needs to be triggered.

[0119] The retirement stage refers to the customer package that cannot meet the business requirements, and the matching degree and consistency are lower than the minimum requirement, and needs to be disassembled, reconstructed or marked as an invalid package.

[0120] The state judgment index system is established by the following formula, which is used for identifying the life cycle state: ; ; ; ; ; Wherein, is the current life cycle health degree of the customer package, , , and are index weight coefficients, which need to satisfy ; and are the comprehensive matching degrees of the customer package in the current period and the last period with the target power generation curve; is the matching degree change, reflecting the change trend of the customer package business adaptation ability; and are the internal customer electricity behavior similarities of the customer package in the current period and the last period ; is the consistency change, which is used to judge whether the package internal structure is stable; and are the smoothness indexes of the customer package in the current period and the last period ; is the quantitative index of whether the customer package overall fluctuation increases; and are the penalty items of the customer package in the current period and the last period whose fluctuation exceeds the threshold value; is the change of the customer package risk degree.

[0121] According to , the customer package state is divided: Creation state: ; It indicates that the customer package is initially formed or has not yet stabilized.

[0122] Active state: ; It indicates that the customer package structure is stable, well matched with the power generation curve, and can be used for market transaction.

[0123] Decay state: ; indicates that the customer package has a downward trend in various indicators, and the member structure needs to be adjusted.

[0124] Retired state: ; indicates that the customer package is invalid and no longer has transaction value, and needs to be disassembled and rebuilt.

[0125] wherein, , and are set empirical values and satisfy the condition .

[0126] In the specific implementation process, the embodiment of the application assigns a life cycle tag representing the current life cycle, which can include but is not limited to: The following formula is used to assign a life cycle tag to each customer package: ; wherein, is a life cycle mapping function, is the customer package state (creation, active, decay, and retirement). The state and life cycle tag correspondence is shown in Table 2: Table 2 Life cycle tag mapping relationship wherein, the package can be used to participate in transactions, the package triggers an automatic optimization mechanism, and the package automatically enters the disassembly and reconstruction process.

[0127] According to the preset life cycle transfer rule, it is judged whether the current life cycle of the business-oriented customer package meets the transfer condition, and if it meets, the current life cycle of the business-oriented customer package is adjusted and the life cycle tag is re-assigned.

[0128] In the specific implementation process, the embodiment of the application can include but is not limited to the following: According to determine the state transfer rule of the business-oriented customer package in the four life cycle stages, which is as follows: The condition for transferring from the creation stage to the active stage is: , that is, to ensure that the matching degree and consistency both reach the available threshold.

[0129] The condition for transferring from the active stage to the decay stage is: and , that is, the matching degree and life cycle health degree are both decreased.

[0130] Conditions for transitioning from the decay phase to the active phase: .

[0131] Conditions for transitioning from the decline phase to the retirement phase: .

[0132] In one feasible implementation, the management of the business-oriented customer package according to business objectives described in this embodiment of the invention may, but is not limited to, include: Based on at least one lifecycle management result, assign a comprehensive tag to each customer, which includes a basic tag and a scenario tag; The integrated tags are updated periodically and / or triggered by events.

[0133] In specific implementation, the basic tags in this embodiment of the invention may include, but are not limited to, tags with long-term stability, low update frequency, and no dependence on transaction results, such as industry type, voltage level, electricity consumption level, regional characteristics, and credit rating. The basic tags can be constructed using the following formula: ; in, Number the customer. For the first A basic tag, The number of basic tags.

[0134] The scenario tags in this embodiment of the invention may include, but are not limited to, medium- and long-term transaction scenario tags, spot transaction scenario tags, and customer package scenario tags.

[0135] The medium- and long-term transaction scenario tags are used to reflect the customer's behavioral characteristics in contracts, guaranteed electricity volume, performance, and deviation management. These tags may include, but are not limited to, annual performance rate, contract yield, deviation electricity volume ratio, and contract default risk, and can be constructed using the following formula: ; in, Number the customer. For the first Long-term transaction scenario tags, This refers to the number of tags for medium- to long-term transaction scenarios.

[0136] The spot trading scenario tags are used to reflect the characteristics of customers in time-of-use pricing, market clearing, and dispatch response. These may include, but are not limited to, load price sensitivity, clearing participation frequency, and peak load shifting capability. They can be constructed using the following formula: ; in, Number the customer. For the first A spot trading scenario tag, This refers to the number of tags for spot trading scenarios.

[0137] The customer package scenario tags are used to reflect the status and characteristics of the customer package, and may include, but are not limited to, customer package business type, matching degree level, smoothness level, package stability, and package lifecycle status, etc., and can be constructed using the following formula: ; in, Number the customer. For the first The customer package scenario tags. The number of scene tags for the customer package.

[0138] The comprehensive tag in this embodiment of the invention integrates basic tags, medium- and long-term transaction scenario tags, spot transaction scenario tags, and customer package scenario tags to form a key tag that can ultimately be used for business strategy, marketing, pricing, and risk management.

[0139] The comprehensive tags are categorized based on the customer's primary electricity trading methods: (1) If the client mainly participates in medium- to long-term transactions, the fusion tag layer is: ; (2) If the customer mainly participates in spot trading, the fusion tag layer is: ; In specific implementation, the periodic updates described in this embodiment of the invention may include, but are not limited to: updating medium- and long-term labels based on the latest contract execution status, updating spot labels based on recent price responses and fluctuations, and updating customer package labels based on their lifecycle status.

[0140] The event-triggered update described in this embodiment of the invention may include, but is not limited to, updating the fusion tag immediately when a customer joins / leaves a customer package, the customer package lifecycle changes, or the customer behavior type changes significantly.

[0141] The above-mentioned power customer management method based on curve similarity and dynamic clustering provided in the embodiments of the present invention introduces similarity matching between users, so that customer electricity consumption behavior can be re-characterized from the perspective of the group, thereby enabling the identification of the overall load characteristics generated after customer combination, and enabling customer tags to be extended from static attributes to dynamic ones.

[0142] Furthermore, embodiments of the present invention generate different types of load packages by matching customer packages with the power output curves of the generation side, supporting the design of combined trading products, and achieving the goals of reducing procurement cost fluctuations, reducing deviation power volume, and improving contract performance stability, thereby enhancing the bargaining power of the power sales company and the power generation company in the power trading process.

[0143] Furthermore, this embodiment of the invention introduces lifecycle management methods into the electricity sales customer package system, providing dynamic maintenance so that customer packages can automatically maintain optimal load smoothness and scale, thereby improving the tradability and continuous stability of customer packages.

[0144] Based on the above-mentioned method for electricity customer management based on curve similarity and dynamic clustering provided in the first aspect, this embodiment of the invention provides an electricity customer management system based on curve similarity and dynamic clustering, such as... Figure 3 As shown, the power customer management system based on curve similarity and dynamic clustering includes: Data acquisition module 110 is used to acquire user-side daily load data and generator-side power generation data, wherein the user-side daily load data includes daily load data of multiple users; The data preprocessing module 120 is used to preprocess the user-side daily load data and the generator-side power generation data to obtain the user-side daily load curve and the generator-side power generation curve. The user-side similarity analysis module 130 is used to calculate the similarity of the user-side daily load curves and obtain a comprehensive similarity matrix between users. The customer package-generation side matching module 140 is used to perform multi-dimensional comprehensive matching of the user-side daily load curve and the generation-side generation curve to obtain a comprehensive matching matrix of the customer package-generation side curve. The business-oriented customer package construction module 150 is used to construct a business-oriented customer package based on the comprehensive similarity matrix between users and the comprehensive matching matrix of customer package-generation side curve. The customer package management module 160 is used to manage the business-oriented customer packages according to business objectives.

[0145] Based on the above-described method for electricity customer management based on curve similarity and dynamic clustering provided in the first aspect, this embodiment of the invention also provides a storage medium storing a computer-executable program. The computer-executable program is used to cause a computer to execute the electricity customer management method based on curve similarity and dynamic clustering as described in any implementation of the first aspect. Explanations of the relevant content and descriptions of the beneficial effects of any of the computer-readable storage media provided above can be found in the corresponding embodiments described above, and will not be repeated here.

[0146] Those skilled in the art will understand that the program for implementing all or part of the steps of the above embodiments, which can be executed by a program instructing related hardware, can be stored in a computer-readable storage medium. The storage medium mentioned above can be a read-only memory, a random access memory, etc. The processing unit or processor mentioned above can be a central processing unit, a general-purpose processor, an application-specific integrated circuit (ASIC), a microprocessor (DSP), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof.

[0147] This invention also provides a computer program product containing instructions that, when executed on a computer, cause the computer to perform any of the methods described in the above embodiments. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., SSD), etc.

[0148] It should be noted that the devices for storing computer instructions or computer programs provided in the embodiments of the present invention, such as, but not limited to, the aforementioned memory, computer-readable storage medium, and communication chip, are all non-transitory. Those skilled in the art should recognize that the functions described in the embodiments of the present invention in one or more of the above examples can be implemented using hardware, software, firmware, or any combination thereof. When implemented using software, these functions can be stored in a computer-readable storage medium or transmitted as one or more instructions or code on a computer-readable storage medium. Computer-readable storage media include computer storage media and communication media, wherein communication media include any medium that facilitates the transmission of computer programs from one place to another. Storage media can be any available medium accessible to general-purpose or special-purpose computers.

[0149] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A power customer management method based on curve similarity and dynamic clustering, characterized in that, include: Acquire user-side daily load data and generator-side power generation data, wherein the user-side daily load data includes daily load data from multiple users; The user-side daily load data and the generator-side power generation data are preprocessed to obtain the user-side daily load curve and the generator-side power generation curve. The similarity of the user-side daily load curves is calculated to obtain a comprehensive similarity matrix between users; The user-side daily load curve and the generator-side power generation curve are matched in multiple dimensions to obtain a customer package-generator-side curve comprehensive matching matrix. Based on the comprehensive similarity matrix between users and the comprehensive matching matrix of customer package-generation side curves, a business-oriented customer package is constructed; The business-oriented customer packages are managed according to business objectives.

2. The power customer management method based on curve similarity and dynamic clustering according to claim 1, characterized in that, The preprocessing of the user-side daily load data and the generator-side power generation data to obtain the user-side daily load curve and the generator-side power generation curve includes: Data cleaning, missing value imputation, time alignment, and normalization are performed on the user-side daily load data and the generator-side power generation data respectively to obtain the user-side daily load curve and the generator-side power generation curve.

3. The electricity customer management method based on curve similarity and dynamic clustering according to claim 1, characterized in that, The step of calculating the similarity of the user-side daily load curves to obtain a comprehensive similarity matrix between users includes: Extract the curve features of the user-side daily load curve, and construct a multi-user load curve matrix based on the time point; The morphological features of the multi-user load curve matrix are extracted to construct a comprehensive morphological matrix; Based on the multi-user load curve matrix and the comprehensive morphology matrix, the similarity between users is analyzed to obtain the comprehensive similarity matrix between users.

4. The power customer management method based on curve similarity and dynamic clustering according to claim 3, characterized in that, The step of analyzing the similarity between users based on the multi-user load curve matrix and the comprehensive morphology matrix to obtain the comprehensive similarity matrix between users includes: Use the following formula to analyze trend similarity between users: ； in, For the first Load curve matrix for each customer For the first Load curve matrix for each customer The covariance of the two load curve matrices For the first Standard deviation of the load curve matrix for each customer For the first Standard deviation of the load curve matrix for each customer For customers and customers Trend similarity index; The following formula can be used to analyze the similarity of local change rhythms among users: ； ； ； in, For the first The change in load for a customer between two adjacent moments. and The first and the Load at any given time For the first The and the first The European distance between individual clients with varying rhythms of life. For the first The and the first The degree of difference in the local structural variations of individual customers This serves as an indicator of the similarity of local change rhythms. The following formula is used to analyze the structural similarity between users: ； ； in, and The first The and the first A comprehensive customer profile matrix For the first The and the first European distance of the overall structural characteristics of each customer For structural morphological differences. As an index of structural morphological similarity; The trend similarity index, the local change rhythm similarity index, and the structural morphology similarity index are weighted and fused to obtain the comprehensive similarity matrix between users.

5. The power customer management method based on curve similarity and dynamic clustering according to claim 1, characterized in that, The step of performing multi-dimensional comprehensive matching of the user-side daily load curve and the generator-side generation curve to obtain a customer package-generator-side curve comprehensive matching matrix includes: Based on the user-side daily load curve, construct the total load of the customer package; The trend matching degree, shape matching degree, and dynamic time warping distance matching degree of the total load of the customer package and the power generation curve of the power generation side are analyzed to obtain the comprehensive matching matrix of the customer package-power generation side curve.

6. The electricity customer management method based on curve similarity and dynamic clustering according to claim 5, characterized in that, The step of constructing the total load of the customer package based on the user-side daily load curve includes: The total load of the customer package is constructed using the following formula: ； in, For the first The customer in the first Load value at any given time For the set of all users within the customer package, For the first Within a certain time period The total load value of all users in the set.

7. The electricity customer management method based on curve similarity and dynamic clustering according to claim 5, characterized in that, The analysis of the trend matching degree, shape matching degree, and dynamic time warping distance matching degree between the total load of the customer package and the power generation curve on the generation side yields a comprehensive matching matrix of the customer package-power generation curve, including: The following formula is used to analyze the trend matching degree: ； in, This is the g-th type curve on the power generation side. The Pearson correlation coefficient is used. For trend matching degree; Based on the total load of the customer package, calculate the fluctuation range and intensity of the total load fluctuation of the customer package; based on the power generation curve of the power generation side, calculate the fluctuation range and intensity of the power generation curve fluctuation. The shape matching degree is analyzed using the following formula: ； ； in, For the customer's total load fluctuation range, For the customer's total load fluctuation intensity, The amplitude of the power generation curve. The intensity of the power generation curve fluctuations. and These are weighting coefficients. To prevent tiny constants with a denominator of zero, The difference in shape between the total load of the customer package and the power generation curve on the power generation side. The shape matching degree between the total load of the customer package and the power generation curve on the power generation side; The dynamic time-warped distance matching degree is analyzed using the following formula: ； in, For dynamic time-normalized distance, The attenuation coefficient is... For dynamic time-warped distance matching degree; The trend matching degree, the shape matching degree, and the dynamic time warping distance matching degree are weighted and fused to obtain the customer package-generator side curve comprehensive matching matrix.

8. The power customer management method based on curve similarity and dynamic clustering according to claim 1, characterized in that, The step of constructing a business-oriented customer package based on the user-to-user comprehensive similarity matrix and the customer package-generation side curve comprehensive matching matrix includes: A similarity graph is constructed based on the comprehensive similarity matrix among users to obtain the initial customer grouping structure; Clustering is performed on the initial customer grouping structure to obtain the initial customer package; A matching degree analysis is performed on the initial customer package and the comprehensive matching matrix of the customer package-generation side curve to obtain the generation side category that matches the initial customer package; Based on the power generation category, the initial customer package is iteratively optimized using an objective function, eliminating unsuitable customers and introducing complementary customers to obtain the business-oriented customer package.

9. A power customer management method based on curve similarity and dynamic clustering according to claim 1, characterized in that, The management of the business-oriented customer package based on business objectives includes: Identify the current lifecycle of the business-oriented customer package and assign a lifecycle label that represents the current lifecycle; According to the preset lifecycle transfer rules, it is determined whether the current lifecycle of the business-oriented customer package meets the transfer conditions. If it does, the current lifecycle of the business-oriented customer package is adjusted and the lifecycle label is reassigned.

10. A power customer management method based on curve similarity and dynamic clustering according to claim 9, characterized in that, The management of the business-oriented customer package based on business objectives also includes: Based on at least one lifecycle management result, assign a comprehensive tag to each customer, which includes a basic tag and a scenario tag; The integrated tags are updated periodically and / or triggered by events.

Images