Daily load curve clustering method, system and device and storage medium

Abstract

The invention relates to a daily load curve clustering method, system and device, and a storage medium. The method comprises the following steps: collecting a daily load curve of each user in a power system, and decomposing the daily load curve into subsequences of a plurality of time scales; calculating time features in each sub-sequence, constructing an adaptive clustering tree, and mapping the time features of the sub-sequence of each time scale to the adaptive clustering tree; clustering the daily load curve based on an adaptive clustering tree to obtain a preliminary clustering result; constructing a clustering multi-objective optimization function, and optimizing the preliminary clustering result to obtain a preliminary optimized clustering result; and introducing an external influence factor to correct the preliminary optimization clustering result to obtain a final clustering result.

Description

Technical Field

[0001] The present invention relates to a daily load curve clustering method, system, device and storage medium, belonging to the field of power technology. Background Art

[0002] The power load curve is a curve describing the change of electricity demand in the power system at different time points. The typical daily load curve reflects the supply-demand balance state of the power system within 24 hours of a day, and has significant time regularity, user category characteristics and seasonal characteristics.

[0003] The load curve can help the power dispatching center reasonably allocate power resources and reduce unnecessary power generation costs. By clustering the daily load curve, the electricity consumption patterns of different types of users (residential, commercial, industrial) can be identified, and power dispatching strategies can be customized for different user groups. Most traditional clustering methods pay more attention to the similarity of the overall load curve and ignore local details. The load curve has dynamic evolution characteristics over time, but static clustering methods are difficult to capture this change. At the same time, external factors such as weather, holidays, and price policies will significantly affect the load curve, and traditional clustering methods have not considered these external factors, resulting in inaccurate clustering results. Summary of the Invention

[0004] In order to solve the problems existing in the above-mentioned prior art, the present invention proposes a daily load curve clustering method, system, device and storage medium.

[0005] The technical solution of the present invention is as follows: On the one hand, the present invention provides a daily load curve clustering method, including the following steps: Collect the daily load curves of each user in the power system and decompose the daily load curves into subsequences of multiple time scales; Calculate the time features in each subsequence, construct an adaptive clustering tree, and map the time features of the subsequences of each time scale to the adaptive clustering tree; Cluster the daily load curve based on the adaptive clustering tree to obtain a preliminary clustering result; Construct a clustering multi-objective optimization function to optimize the preliminary clustering result to obtain a preliminary optimized clustering result; Introduce an external influence factor to correct the preliminary optimized clustering result to obtain a final clustering result.

[0006] As a preferred embodiment of the present invention, the specific steps of decomposing the daily load curve into subsequences of multiple time scales are: Preset multiple time-scale windows, segment the daily load curves of all users according to different time-scale windows, construct a subsequence for the segmentation result of each time-scale window, and each subsequence includes the statistical features of each segment under the current time-scale window.

[0007] As a preferred embodiment of the present invention, the specific steps for calculating the time features in each subsequence are as follows: Calculate the time entropy of each daily load curve in each subsequence, as shown in the following formula: ; Where: represents the time entropy of each segment of the current daily load curve; represents the number of time points of the segment; represents the statistical feature of the th time point in the segment; The sum of the time entropy of each segment of the daily load curve is used to obtain the time feature of the daily load curve under the current subsequence.

[0008] As a preferred embodiment of the present invention, construct an adaptive clustering tree, which consists of a root node, intermediate nodes, and leaf nodes. The root node is composed of several intermediate nodes, and the intermediate nodes are composed of several leaf nodes; Map the daily load curves of each user to the root node, map the subsequences of each time scale to the intermediate nodes, and set the time features of each daily load curve under the subsequence to the leaf nodes corresponding to the intermediate nodes.

[0009] As a preferred embodiment of the present invention, the specific steps for clustering the daily load curves based on the adaptive clustering tree are as follows: Calculate the similarity between each leaf node, as shown in the following formula: ; Where: represents the similarity between the th leaf node and the th leaf node; represents the time feature of the daily load curve corresponding to the th leaf node; represents the time feature of the daily load curve corresponding to the th leaf node; represents the adjustment parameter; Preliminarily classify the leaf nodes with similarity greater than the preset threshold into the same cluster class to obtain a preliminary clustering result.

[0010] As a preferred embodiment of the present invention, construct a clustering multi-objective optimization function as shown in the following formula: ; Among them: represents the clustering compactness; represents the clustering separation; represents the clustering temporal stability; The specific calculation formula d of the clustering compactness is shown as follows: ; Among them: represents the number of preliminary clusters; represents the th number of leaf nodes of the th preliminary cluster; represents the th daily load curve time feature corresponding to the th leaf node in the th preliminary cluster; The specific calculation formula of the clustering separation is shown as follows: ; Among them: represents the th clustering center of the th preliminary cluster; represents all the time features in the th preliminary cluster; represents all the time features in the th preliminary cluster; ; Among them: represents the th time feature at the th moment in the daily load curve corresponding to the th leaf node in the th preliminary cluster; represents the th time feature at the th moment in the daily load curve corresponding to the th leaf node in the The non - dominated sorting genetic algorithm is used to solve the clustering multi - objective optimization function to form the Pareto optimal solution, and the Pareto optimal solution is used as the preliminary optimized clustering result.

[0011] As a preferred embodiment of the present invention, the specific steps for correcting the preliminary optimized clustering result by introducing external influence factors are as follows: The external influence factors include time influence factors, environmental influence factors and social influence factors, and an external influence factor weight matrix is constructed; Calculate the correction deviation of each cluster in the preliminary optimized clustering result based on the external influence factor weight matrix, as shown in the following formula: ; Where: represents the correction deviation of the -th preliminary optimized clustering center; represents the number of external influence factors; represents the weight of the -th external influence factor; represents the -th external influence factor and the -th preliminary optimized clustering center of the influence deviation distance; Based on the correction deviation of the preliminary optimized clustering center, correct the preliminary optimized clustering center, as shown in the following formula: ; Where: represents the corrected preliminary optimized clustering center; represents the correction step size; Re-cluster according to the corrected preliminary optimized clustering center to obtain the final clustering result.

[0012] On the other hand, the present invention also provides a daily load curve clustering system, including a data acquisition module, a preliminary clustering module, a clustering optimization module, and a clustering correction module; The data acquisition module is used to collect the daily load curves of each user in the power system and decompose the daily load curves into subsequences of multiple time scales; The preliminary clustering module is used to calculate the time features in each subsequence, construct an adaptive clustering tree, map the time features of the subsequences of each time scale to the adaptive clustering tree; cluster the daily load curves based on the adaptive clustering tree to obtain a preliminary clustering result; The clustering optimization module is used to construct a clustering multi-objective optimization function, optimize the preliminary clustering result to obtain a preliminary optimized clustering result; The clustering correction module is used to introduce external influence factors to correct the preliminary optimized clustering result to obtain the final clustering result.

[0013] On yet another aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the program, it implements the method described in any embodiment of the present invention.

[0014] In another aspect, the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the method described in any embodiment of the present invention is implemented.

[0015] The present invention has the following beneficial effects: 1. The present invention decomposes the daily load curve at different time scales, can capture short-term, medium-term, and long-term characteristics, and avoids information loss; 2. The present invention combines clustering compactness, separability, and time stability for multi-objective optimization, improving the interpretability and practicality of the clustering results; 3. The present invention introduces external influencing factors to correct the clustering results of the daily load curve, making the clustering results more accurate. Description of the Drawings

[0016] Figure 1 It is a flowchart of the method of the present invention. Detailed Embodiments

[0017] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0018] It should be understood that the step numbers used in the text are only for convenient description and do not limit the execution order of the steps.

[0019] It should be understood that the terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless otherwise clearly specified in the context, the singular forms "a", "an", and "the" are intended to include the plural forms.

[0020] The terms "including" and "comprising" indicate the presence of the described features, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0021] The term "and / or" refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0022] Embodiment 1: Referring to Figure 1 , a method for clustering daily load curves includes the following steps: Collect the daily load curves of each user in the power system and decompose the daily load curves into subsequences of multiple time scales; Calculate the time features in each subsequence, construct an adaptive clustering tree, and map the time features of the subsequences of each time scale to the adaptive clustering tree; Cluster the daily load curves based on the adaptive clustering tree to obtain a preliminary clustering result; Construct a clustering multi-objective optimization function to optimize the preliminary clustering result to obtain a preliminary optimized clustering result; Introduce external influence factors to correct the preliminary optimized clustering result to obtain the final clustering result.

[0023] As a preferred implementation manner of this embodiment, the specific steps for decomposing the daily load curves into subsequences of multiple time scales are as follows: Preset multiple time scale windows (for example: hourly, four-hourly, daily), segment the daily load curves of all users according to different time scale windows, construct a subsequence for the segmentation result of each time scale window, and each subsequence includes the statistical features of each segment under the current time scale window. The statistical features include the average load, load standard deviation, load skewness, load kurtosis, and change rate of adjacent time points of each segment; As a preferred implementation manner of this embodiment, the specific steps for calculating the time features in each subsequence are as follows: Calculate the time entropy of each daily load curve in each subsequence, as shown in the following formula: ; Where: represents the time entropy of each segment of the current daily load curve; represents the number of time points of the segment; represents the statistical feature of the th time point in the segment; The sum of the time entropies of each segment of the daily load curve is used to obtain the time feature of the daily load curve under the current subsequence.

[0024] As a preferred implementation manner of this embodiment, construct an adaptive clustering tree. The adaptive clustering tree consists of a root node, intermediate nodes, and leaf nodes. The root node consists of several intermediate nodes, and the intermediate nodes consist of several leaf nodes; Map the daily load curves of each user to the root node, map the subsequences of each time scale to the intermediate nodes, and set the time features of each daily load curve under the subsequence to the leaf nodes corresponding to the intermediate nodes.

[0025] As a preferred implementation manner of this embodiment, the specific steps for clustering the daily load curves based on the adaptive clustering tree are as follows: Calculate the similarity between each leaf node, as shown in the following formula: ; Where: represents the similarity between the -th leaf node and the -th leaf node; represents the time characteristics of the daily load curve corresponding to the -th leaf node; represents the time characteristics of the daily load curve corresponding to the -th leaf node; represents the adjustment parameter; Preliminarily classify the leaf nodes with similarity greater than the preset threshold into the same cluster category to obtain the preliminary clustering result.

[0026] As a preferred implementation manner of this embodiment, construct the clustering multi-objective optimization function as shown in the following formula: ; Where: represents the clustering compactness; represents the clustering separation; represents the clustering time stability; The specific calculation formula d of the clustering compactness is as shown in the following formula: ; Where: represents the number of preliminary clusters; represents the number of leaf nodes in the -th preliminary cluster; represents the time characteristics of the -th leaf node in the -th preliminary cluster; represents the clustering center of the -th preliminary cluster; The specific calculation formula of the clustering separation is as shown in the following formula: ; Where: represents the clustering center of the -th preliminary cluster; represents all the time characteristics in the -th preliminary cluster; represents all the time characteristics in the -th preliminary cluster; The specific calculation formula of the clustering time stability is as shown in the following formula: ; Where: represents the The time characteristics at the th leaf node in the th initial clustering; Denote the th leaf node in the th initial clustering; The time characteristics at the Denote the time step; Solve the clustering multi-objective optimization function through the non-dominated sorting genetic algorithm to form the Pareto optimal solution, and take the Pareto optimal solution as the preliminary optimized clustering result.

[0027] As the preferred implementation manner of this embodiment, the specific steps for correcting the preliminary optimized clustering result by introducing external influence factors are as follows: The external influence factors include time influence factors, environmental influence factors, and social influence factors, and an external influence factor weight matrix is constructed; Calculate the correction deviation of each clustering in the preliminary optimized clustering result based on the external influence factor weight matrix, as shown in the following formula: ; Where: Denote the correction deviation of the th preliminary optimized clustering center; Denote the number of external influence factors; Denote the weight of the th external influence factor; Denote the th external influence factor and the th preliminary optimized clustering center The influence deviation distance (Euclidean distance or other metrics can be used); The calculation formula of the time influence factor is: ; Where: Denote the time influence factor; Denote the time step (hour level: 1, 2,..., 24); Denote the cycle parameter (24 hours, 7 days, etc.); Denote the date (1 - 365, representing the number of days in a year); Denote the key days (such as Spring Festival, National Day and other holidays); Denote the hour (0 - 23 hours); Denote the key hour (such as peak time: 18:00); 、 , respectively represent weight coefficients; , respectively represent time decay coefficients; represents a random time noise term, which satisfies the standard normal distribution; The calculation formula for the environmental impact factor is: ; Where: represents the environmental impact factor; represents the average temperature of the day; represents the optimal temperature (e.g., 23 °C); represents the average humidity of the day; represents the average wind speed of the day; represents the weather weight (e.g., 1.0 for sunny days, 1.2 for rainy days, 1.5 for snowy days); represents the critical temperature point (e.g., the high-temperature warning line of 35 °C); , , respectively represent weight coefficients; , represents the environmental regulation coefficient; The calculation formula for the social impact factor is: ; Where: represents the social impact factor; represents the number of social events; represents the th social event's impact intensity on the power load; represents the th social event's electricity price when it occurs; represents the electricity price threshold; represents the electricity price change rate parameter; represents the power demand change rate.

[0028] Based on the correction deviation of the preliminary optimized clustering center, the preliminary optimized clustering center is corrected as shown in the following formula: ; Where: represents the corrected preliminary optimized clustering center; represents the correction step size; Re-clustering is performed according to the corrected preliminary optimized clustering center to obtain the final clustering result.

[0029] The final clustering result includes: Residential user load curve: Characteristic: Obvious morning and evening peaks, relatively low electricity consumption during the day, and certain electricity demand at night.

[0030] Typical load curve: Double-peak type (morning peak from 7:00 to 9:00, evening peak from 18:00 to 22:00).

[0031] Industrial user load curve: Characteristic: Higher electricity load during the day, lower at night, and significant differences in load between weekdays and holidays.

[0032] Typical load curve: Single-peak type (high load during the day, low load at night).

[0033] Commercial user load curve: Characteristic: High load during the day, lower at night, and the peak usually occurs during the daytime working hours.

[0034] Typical load curve: Daytime peak type (from 9:00 to 18:00).

[0035] Agricultural user load curve: Characteristic: Strong correlation between electricity load and seasons (significantly high load during the irrigation season).

[0036] Typical load curve: Significant seasonal fluctuations, relatively stable daily load.

[0037] Mixed user load curve: Characteristic: Comprising the composite characteristics of residential, industrial, and commercial users.

[0038] Typical load curve: Multi-peak type or irregular type.

[0039] Example 2: A daily load curve clustering system, comprising a data acquisition module, a preliminary clustering module, a clustering optimization module, and a clustering correction module; The data acquisition module is used to collect the daily load curves of each user in the power system and decompose the daily load curves into subsequences of multiple time scales; The preliminary clustering module is used to calculate the time characteristics in each subsequence, construct an adaptive clustering tree, map the time characteristics of the subsequences of each time scale to the adaptive clustering tree; cluster the daily load curves based on the adaptive clustering tree to obtain a preliminary clustering result; The clustering optimization module is used to construct a clustering multi-objective optimization function, optimize the preliminary clustering result to obtain a preliminary optimized clustering result; The clustering correction module is used to introduce external influence factors to correct the preliminary optimized clustering result to obtain a final clustering result.

[0040] This system is used to implement the method in Example 1, which will not be elaborated here.

[0041] Example 3: This example provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the method described in any embodiment of the present invention is implemented.

[0042] Example 4: This example provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the method described in any embodiment of the present invention is implemented.

[0043] In the embodiments of the present application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent the cases of A existing alone, A and B existing simultaneously, and B existing alone. Where A and B may be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" and its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, and c may represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c may be single or multiple.

[0044] Those of ordinary skill in the art can realize that the units and algorithm steps described in the embodiments disclosed herein can be implemented by a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0045] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0046] In several embodiments provided by the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or a part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM for short), random access memories (RAM for short), magnetic disks, or optical discs that can store program codes.

[0047] The above are only the embodiments of the present invention, and thus do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A daily load curve clustering method, characterized in that: The following steps are involved: Collect the daily load curve of each user in the power system and decompose the daily load curve into subsequences of multiple time scales; Calculate the time features in each subsequence, construct an adaptive clustering tree, and map the time features of the subsequences at each time scale to the adaptive clustering tree; The daily load curves are clustered based on the adaptive clustering tree and the preliminary clustering results are obtained; Construct a clustering multi-objective optimization function, optimize the preliminary clustering results, and obtain preliminary optimized clustering results; External influencing factors are introduced to correct the preliminary optimized clustering results and obtain the final clustering results.

2. A daily load curve clustering method according to claim 1, characterized in that: The specific steps of decomposing the daily load curve into subsequences of multiple time scales are as follows: Multiple time scale windows are preset, and the daily load curves of all users are segmented according to different time scale windows. A subsequence is constructed for the segmentation result of each time scale window, and each subsequence includes the statistical features of each segment under the current time scale window.

3. A daily load curve clustering method according to claim 2, characterized in that: The specific steps for calculating the time features in each subsequence are: Calculate the time entropy of each daily load curve in each subsequence, as shown in the following formula: ； in: Represents the time entropy of each segment of the current daily load curve; Indicates the number of time points for segmentation; Indicates the first The statistical characteristics of each time point are obtained by summing the time entropy of each segment of the daily load curve to obtain the time characteristics of the daily load curve under the current subsequence.

4. A daily load curve clustering method according to claim 3, characterized in that: Constructing an adaptive clustering tree, wherein the adaptive clustering tree is composed of a root node, an intermediate node and a leaf node, wherein the root node is composed of a plurality of intermediate nodes, and the intermediate node is composed of a plurality of leaf nodes; Map the daily load curve of each user to the root node, map the subsequence of each time scale to the intermediate node, and set the time characteristics of each daily load curve under the subsequence to the leaf node corresponding to the intermediate node.

5. A daily load curve clustering method according to claim 4, characterized in that: The specific steps of clustering daily load curves based on the adaptive clustering tree are as follows: Calculate the similarity between leaf nodes, as shown in the following formula: ； in: Indicates The leaf node and The similarity of leaf nodes; Indicates Time characteristics of daily load curve corresponding to each leaf node; Indicates Time characteristics of daily load curve corresponding to each leaf node; represents the adjustment parameter; The leaf nodes with similarity greater than the preset threshold are initially divided into the same cluster to obtain the preliminary clustering results.

6. A daily load curve clustering method according to claim 5, characterized in that: The construction of clustering multi-objective optimization function is shown in the following formula: ； in: Indicates cluster compactness; Indicates cluster separation; Indicates the temporal stability of clustering; The calculation formula d of the cluster compactness is specifically shown as follows: ； in: Indicates the number of preliminary clusters; Indicates The number of leaf nodes of the preliminary clustering; Indicates The first of the preliminary clusters Time characteristics of daily load curve corresponding to each leaf node; Indicates The cluster centers of the initial clustering; The calculation formula of the cluster separation is specifically shown as follows: ； in: Indicates The cluster centers of the initial clustering; Indicates All temporal features in the first preliminary clusters; Indicates All temporal features in the first preliminary clusters; The calculation formula of the clustering time stability is specifically shown as follows: ； in: Indicates The first of the preliminary clusters The daily load curve corresponding to the leaf node The temporal characteristics of the moment; Indicates The first of the preliminary clusters The daily load curve corresponding to the leaf node The temporal characteristics of the moment; represents the time step; The clustering multi-objective optimization function is solved by non-dominated sorting genetic algorithm to form the Pareto optimal solution, which is used as the preliminary optimization clustering result.

7. A daily load curve clustering method according to claim 6, characterized in that: The specific steps of introducing external influencing factors to correct the preliminary optimization clustering results are: The external influencing factors include time influencing factors, environmental influencing factors and social influencing factors, and a weight matrix of external influencing factors is constructed; The correction deviation of each cluster in the preliminary optimization clustering result is calculated based on the external influencing factor weight matrix, as shown in the following formula: ； in: Indicates Correction deviation of the initial optimized cluster centers; Represents the number of external influencing factors; Indicates The weight of external influencing factors; Indicates External factors With Initially optimized cluster centers The influence deviation distance; The preliminary optimized cluster center is corrected based on the correction deviation of the preliminary optimized cluster center, as shown in the following formula: ； in: Represents the modified preliminary optimized cluster center; represents the correction step length; Re-clustering is performed based on the revised preliminary optimized cluster centers to obtain the final clustering results.

8. A daily load curve clustering system, characterized in that: It includes a data collection module, a preliminary clustering module, a clustering optimization module and a clustering correction module; The data acquisition module is used to collect the daily load curve of each user in the power system and decompose the daily load curve into subsequences of multiple time scales; The preliminary clustering module is used to calculate the time features in each subsequence, construct an adaptive clustering tree, and map the time features of the subsequences at each time scale to the adaptive clustering tree; The daily load curves are clustered based on the adaptive clustering tree and the preliminary clustering results are obtained; The clustering optimization module is used to construct a clustering multi-objective optimization function, optimize the preliminary clustering results, and obtain preliminary optimized clustering results; The clustering correction module is used to introduce external influencing factors to correct the preliminary optimized clustering results to obtain the final clustering results.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images