Method, device and equipment for analyzing electricity utilization characteristics of scattered users and medium

Abstract

The invention discloses a distributed user electricity utilization characteristic analysis method, device and equipment and a medium, and belongs to the technical field of electric power data analysis. The method comprises the following steps: acquiring historical power grid load data of a plurality of users; based on historical power grid load data of each user, typical load curve data and load characteristic data corresponding to the historical power grid load data are obtained; according to the typical load curve data and the load characteristic data of the plurality of users, dividing a power utilization mode of each user through a clustering algorithm; and obtaining a power utilization characteristic analysis result of the user based on the typical load curve data, the load characteristic data and the power utilization mode of the user. According to the method, the typical load curve of the user is extracted through the segmented dimensionality reduction reconstruction method, the influence of abnormal and extreme data is reduced, the characteristics of the user load data are extracted from multiple dimensions, the user load data without labels are divided into different modes through the improved k-means algorithm, and labeling analysis of the electricity utilization characteristics of the scattered users is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power data analysis, and relates to a method, device, equipment and medium for analyzing power consumption characteristics of dispersed users. Background Art

[0002] In the context of the vigorous development of smart grids, power demand side management, as one of the core technologies of smart grids, is the main way for users to participate in the coordinated interaction of power grids. Compared with the traditional method of relying on increasing the capacity of generators to cope with the imbalance of power supply and demand, power demand side management emphasizes the initiative of power users, incorporates user-side resources into the scope of available dispatchable resources, and maintains the interests of both supply and demand, thereby improving the overall benefits of society. User power consumption characteristics analysis usually studies the behavioral habits of users in the process of power consumption. However, with the development of social economy, the transformation and upgrading of enterprises, the high refinement of production division of labor, and the continuous improvement of living standards, the user side of the power grid has undergone significant changes. The power demand of users shows highly differentiated and hierarchical characteristics, and the load characteristics are more complex than before, which brings new challenges to the load characteristics analysis.

[0003] By processing and analyzing the load data of the power grid, the power grid company can fully understand the power consumption habits of power users, help provide personalized power supply services to power users, and improve the refined service level of the power grid. At present, the analysis of user power consumption characteristics is mostly focused on the discussion under the traditional load characteristic indicator system of indicators such as daily maximum load and load rate. With the deepening of power reform, the use of traditional methods to analyze user power consumption characteristics can no longer meet the needs of refined power grid management, and it is necessary to build a more comprehensive and complete power user load portrait. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and to provide a method, device, equipment and medium for analyzing the electricity consumption characteristics of decentralized users, so as to realize large-scale analysis of the electricity consumption characteristics of decentralized users.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a method for analyzing power consumption characteristics of decentralized users, comprising:

[0007] Obtain historical grid load data for multiple users;

[0008] Based on the historical grid load data of each user, obtain the corresponding typical load curve data and load characteristic data;

[0009] Based on the typical load curve data and load characteristic data of multiple users, the power consumption pattern of each user is divided through clustering algorithm;

[0010] Based on the user's typical load curve data, load characteristic data and power consumption pattern, the user's power consumption characteristic analysis results are obtained.

[0011] Furthermore, it also includes: data cleaning of the acquired historical power grid load data.

[0012] Furthermore, based on the user's historical grid load data, the user's typical load curve data is obtained, including:

[0013] The user's historical power grid load data is divided into multiple historical daily load data by day, and the historical daily load data is represented as X = [ x 1 , x 2 , ⋯ , x i ,..., x n ] ;

[0014] The historical daily load data is reduced in dimension, and the data after dimension reduction is X ¯ = [ x ¯ 1 , x ¯ 2 , ⋯ , x ¯ i ,..., x ¯ w ] ,in, The calculation formula is:

[0015] ,

[0016] By using the symbolic set approximation method, according to The size, yes The data is symbolized and assigned values ​​to obtain a discrete symbol sequence Y ^ = [ y ^ 1 , y ^ 2 , ⋯ , y ^ i ,..., y ^ w ] ;

[0017] statistics The frequency of occurrence of various symbol sequences in the , and filter out the most common ones A sequence of symbols;

[0018] Historical daily load data In and described The load sequences corresponding to the symbol sequences are superimposed and averaged to obtain the typical load curve data of the user.

[0019] Furthermore, based on the historical grid load data of the user, the load characteristic data of the user is obtained, including:

[0020] Divide the user's historical grid load data into multiple time series by year, month, and day;

[0021] The load characteristic data of the users are obtained by extracting the characteristics of the divided time series from three indicators: load level, typical peak-valley characteristics and long-term load development law.

[0022] Among them, load level indicators include power consumption, maximum load, minimum load and peak-to-valley difference; typical peak-to-valley characteristic indicators include load rate, minimum load rate and peak-to-valley difference rate; long-term load development law indicators include practical coefficient, stage coefficient and saturated load density.

[0023] Furthermore, users in the year The calculation formula of the practical coefficient is:

[0024] ,

[0025] in, Indicates the user's year The annual maximum load, Indicates the capacity reported by the user when applying for access to electricity;

[0026] The calculation formula of the user's terminal utility coefficient is:

[0027] ,

[0028] in, Indicates the final year, ; Indicates the year from when the user connects to electricity to the end of the term Annual maximum load;

[0029] User in the year The calculation formula of the stage coefficient is:

[0030] ;

[0031] The calculation formula for user load density is:

[0032] ,

[0033] in, Indicates the user's building area.

[0034] Further, the clustering algorithm includes a k-means++ algorithm;

[0035] The analysis method also includes: identifying the optimal number of clusters k by applying a heuristic algorithm to a clustering evaluation index of a clustering algorithm; the clustering evaluation index includes a silhouette coefficient, a Kalinsky-Harabas index, and a Davis-Boulting index.

[0036] Furthermore, obtaining the user's electricity consumption characteristic analysis results includes: inputting the user's typical load curve data and load characteristic data into a pre-trained daily load curve prediction model under the corresponding power consumption mode to obtain the user's daily load curve prediction results.

[0037] In a second aspect, the present invention further provides a device for analyzing power consumption characteristics of decentralized users, the device comprising:

[0038] A historical power grid load data acquisition module is used to acquire historical power grid load data of multiple users;

[0039] A typical load curve and load characteristic acquisition module is used to acquire the corresponding typical load curve data and load characteristic data based on the historical power grid load data of each user;

[0040] The power consumption mode classification module is used to classify the power consumption mode of each user through a clustering algorithm according to the typical load curve data and load characteristic data of multiple users;

[0041] The power consumption characteristic analysis module is used to obtain the user's power consumption characteristic analysis results based on the user's typical load curve data, load characteristic data and power consumption mode.

[0042] In a third aspect, the present invention further provides a computer device, comprising:

[0043] Memory for storing computer programs;

[0044] A processor is used to execute the computer program to implement the steps of the above-mentioned decentralized user power consumption characteristics analysis method.

[0045] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the steps of the above-mentioned method for analyzing the electricity consumption characteristics of decentralized users.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] The method for analyzing the power consumption characteristics of decentralized users provided by the present invention extracts the typical load curve of users through the segmented dimension reduction reconstruction method, reduces the impact of abnormal and extreme data, and extracts important features hidden in the user load data from multiple dimensions, providing effective information support for the analysis of the power consumption characteristics of decentralized users on a large scale. The improved k-means algorithm is used to divide the unlabeled power user load data into different modes to achieve the labeling of the power consumption mode; based on the one-dimensional convolutional neural network, a daily load curve prediction model for large-scale users is constructed in parallel for different power consumption modes, and the daily power consumption characteristics of users are accurately modeled to help users understand their own power consumption characteristics more accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A schematic diagram of a flow chart of a method for analyzing power consumption characteristics of decentralized users provided by an embodiment of the present invention;

[0049] Figure 2A schematic diagram of a process for cleaning historical power grid load data in an embodiment of the present invention;

[0050] Figure 3 Schematic diagram of the structure of a one-dimensional convolutional neural network in an embodiment of the present invention;

[0051] Figure 4 A schematic diagram of the structure of a distributed user power consumption characteristic analysis device provided by an embodiment of the present invention;

[0052] Figure 5 An internal structure diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. The same reference numerals in the accompanying drawings indicate the same or similar components or parts. It should be understood by those skilled in the art that these drawings are not necessarily drawn to scale. The embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0054] The term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0055] Embodiment 1:

[0056] like Figures 1 to 3 As shown, an embodiment of the present invention provides a method for analyzing electricity consumption characteristics of decentralized users. Figure 1 The flowchart is a schematic diagram of the method for analyzing the power consumption characteristics of decentralized users. This flowchart only shows the logical sequence of the method described in this embodiment. In other possible embodiments of the present invention, different methods may be used without conflict. Figure 1 The steps shown or described are accomplished in the order shown.

[0057] The decentralized user power consumption characteristic analysis method provided in this embodiment can be applied to a terminal, and can be executed by a decentralized user power consumption characteristic analysis device, which can be implemented by software and / or hardware, and can be integrated in a terminal.

[0058] See also Figure 1 The method of the embodiment of the present invention specifically includes the following steps 1 to 4. Among them:

[0059] Step 1: Obtain historical grid load data of multiple users.

[0060] Among them, the user's historical power grid load data is collected by time and divided by year, month, and day to form multiple time series.

[0061] Complete and effective historical grid load data is the basis for load characteristics research and analysis. With the widespread implementation of distribution production management systems, distribution automation systems and power consumption collection systems, the distribution and consumption system has generated a large amount of multi-source heterogeneous data. Due to objective factors such as the wide distribution of the distribution system, complex network structure, and harsh operating environment, the original grid load data has some missing and abnormalities, and the overall data quality is not ideal. Therefore, before analyzing the user's power consumption characteristics, it is necessary to clean the user's historical grid load data to obtain more complete and effective data.

[0062] like Figure 2 As shown, the specific cleaning work of the user's historical power grid load data in this embodiment can be mainly divided into two parts: (1) identifying missing values ​​and abnormal values ​​in the collected original power grid load data; (2) completing the missing values ​​in the original power grid load data and correcting the abnormal values ​​in the original power grid load data.

[0063] Missing values ​​in power grid load data are easy to determine, while abnormal data need to be identified using relevant methods. Abnormal data in power grid load data usually appear as negative values ​​or values ​​that are much greater or much less than the average load level. Negative abnormal data are directly selected and replaced with null values, while outliers are identified and corrected using the absolute median difference method. The absolute median difference method is statistically a robust measurement of sample deviation of univariate numerical data. It is a method of screening outliers by calculating the sum of the distances between the sample point value and the average value.

[0064] Take the grid load data of a certain user on a certain day For example, the specific steps of the absolute median difference method are as follows:

[0065] Calculate the sampling points at each moment in the power grid load data With median The absolute deviation ;

[0066] Calculate the median of the absolute value deviation of each sampling point ;

[0067] Define the scale factor To set the load fluctuation range to [ m e d i a n ( x ) − k M A D , m e d i a n ( x ) + k M A D ] The load data value beyond the safety range is corrected according to the following formula. In the embodiment of the present invention, the proportional factor is set to .

[0068] .

[0069] After identifying missing and abnormal data through the above steps, the Lagrange interpolation method is used to effectively fill and repair the load time series data with continuous characteristics. For the missing data point to be filled, select the adjacent normal data points as reference data points and calculate the Lagrange polynomial:

[0070] .

[0071] in, and Respectively Middle and The sampling time of the data points is For the The load value of the data point, Indicates the number of reference data points used to construct the polynomial; the missing moments Substitution The repair value corresponding to the missing or abnormal moment can be calculated.

[0072] Since the power consumption characteristics of different users vary greatly, the corresponding load levels vary greatly. This paper mainly studies the power consumption characteristics of users and the load change trend. The significant difference in load levels may interfere with the subsequent analysis of user power consumption characteristics. Therefore, the linear proportional normalization method is used to normalize the repaired power grid load data:

[0073] ,

[0074] in, After normalization, Grid load data points, is the peak load of the user, is the sampling number of power grid load data.

[0075] Step 2: Based on the historical grid load data of each user, obtain the corresponding typical load curve data and load characteristic data.

[0076] User load curve analysis usually describes the user's electricity consumption characteristics based on the user's historical load data, focusing on the user's electricity consumption characteristics in different time spans such as days, months, and years, and focusing on the method of extracting the user's typical daily load curve. Among the existing conventional methods, the maximum daily load selection method and the daily load rate mean selection method cannot guarantee that the selected typical daily load curve is "typical" enough due to the single load characteristics adopted. Although the improved non-parametric kernel density estimation method fully considers a variety of load characteristics, it still cannot overcome the impact of abnormal and extreme atypical daily load curves in user daily load data.

[0077] The present invention adopts the piecewise dimensionality reduction reconstruction method to perform average piecewise dimensionality reduction on the user's historical daily load curve, and then re-expresses the load data after dimensionality reduction through symbolization. The frequency of the re-expressed symbol sequence is counted to screen the load mode with the highest frequency, thereby minimizing the impact of abnormal and extreme atypical daily load curves in the user's daily load data, and obtaining the average to obtain the final user typical load curve.

[0078] The specific steps of the segmented dimensionality reduction reconstruction method are as follows:

[0079] (1) Load data dimensionality reduction: historical daily load data X = [ x 1 , x 2 , ⋯ , x i ,..., x n ] Perform dimensionality reduction and select is the number of segments after the daily load data is processed by the segmented dimension reduction reconstruction method, then the daily load time series after dimension reduction is X ¯ = [ x ¯ 1 , x ¯ 2 , ⋯ , x ¯ i ,..., x ¯ w ] , The load data is reduced in dimension. The mean of the time series segments is calculated as follows:

[0080] .

[0081] (2) Symbolic representation: The symbolic aggregation approximate representation method is used to symbolize and assign values ​​to each segment of the load time series after dimensionality reduction according to the size of the segment mean. is the symbol set size, and the symbols assigned to each segment are located at a set of vertical split points according to the segment average value The position within can be allocated and can be found in the Gaussian distribution table shown in Table 1 to obtain the discretized symbol sequence Y ^ = [ y ^ 1 , y ^ 2 , ⋯ , y ^ i ,..., y ^ w ] .

[0082] Table 1 Gaussian distribution table

[0083]

[0084] (3) Symbol sequence frequency statistics: After the load data is symbolized, statistics are The frequency of occurrence of various symbol sequences in the , and filter out the most common ones A symbol sequence is used to filter out the most common load curve forms and eliminate abnormal load conditions.

[0085] (4) Typical load curve extraction: extract historical daily load data In and described The load sequences corresponding to the symbol sequences are superimposed and averaged to obtain the typical load curve data of the user. L d = [ l 1 d , l 2 d , ⋯ , l i d ,..., l n d ] , No. The calculation formula for each data point is:

[0086] ,

[0087] in, Indicates the The first load sequence The first data points.

[0088] In order to fully explore the effective information in the load data of a single user, deepen the analysis of the user's electricity consumption characteristics, and support the development of personalized and differentiated services for users in the distribution network, it is necessary to extract load characteristics from multiple dimensions to provide sufficient effective information support for the analysis of large-scale decentralized user electricity consumption characteristics. The present invention comprehensively considers the practicality of the indicators, and selects characteristic indicators belonging to the load level description type under the three dimensions of load level, typical peak and valley characteristics, and long-term load development law, and organizes them from three time spans: day, month, and year.

[0089] User load characteristics under the load level dimension usually refer to characteristics that can intuitively describe the actual scale of electricity consumption of users in different time spans such as day, month and year. They are expressed as specific dimensional values, including daily electricity consumption, daily maximum load, daily minimum load, daily peak-to-valley difference, etc. They can accurately characterize the actual electricity consumption level of users. The specific definition and calculation method of characteristic indicators are shown in Table 2.

[0090] Table 2 Load level dimension characteristics

[0091]

[0092] The user load characteristics under the typical peak-valley characteristic dimension usually represent the characteristics of the ratio type, which can characterize the peak-valley characteristics of the user's electricity consumption in different time spans such as day, month and year, and are expressed as dimensionless values, including daily load rate, daily peak-valley difference rate, etc. The characteristic data of this dimension can avoid the interference of the actual scale of user electricity consumption and intuitively characterize the habitual characteristics of user electricity consumption behavior. The specific definition and calculation method of the characteristic indicators are shown in Table 3.

[0093] Table 3 Typical peak-valley characteristic dimension characteristics

[0094]

[0095] The characteristic indicators under the dimension of long-term load development law usually provide a description of the change trend from the user's power access stage to the saturation stage over a longer time span, including practical coefficient, stage coefficient and saturated load density, etc., as an important measurement basis for summarizing and judging the long-term change trend of user power characteristics. The practical coefficient and stage coefficient are commonly used indicators in business expansion and other distribution network planning work, while the load density reflects the relationship between the size of loads of different natures or categories and the user's land area or building area, and represents the density of power load distribution.

[0096] The utility coefficient is usually used to characterize the relationship between the capacity reported by the user when applying for access to electricity and the terminal load when the load develops to saturation after the actual access to electricity. It directly reflects the utilization rate of the user's actual load level compared to the estimated capacity when applying for access. The calculation formula of the practical coefficient is:

[0097] ,

[0098] in, Indicates the user's year The annual maximum load, Indicates the capacity reported by the user when applying for electricity access.

[0099] The stage coefficient represents the ratio of the user's actual load in each year to the actual load in the final year during the process of gradually developing to saturation after the user is connected to electricity. It reflects how fast the user's load develops after the user is connected to electricity.

[0100] The calculation formula of the user's terminal utility coefficient is:

[0101] ,

[0102] in, Indicates the final year, ; Indicates the year from when the user connects to electricity to the end of the term Annual maximum load;

[0103] User in the year The calculation formula of the stage coefficient is:

[0104] .

[0105] Load density is the average power consumption per unit area, which usually describes the relationship between the user's power load and the building area. The specific value directly reflects the density of the power load distribution. By counting the terminal power load and floor area of ​​the user group in the distribution network planning area, the reference value of the saturated load density of each industry can be comprehensively calculated to effectively improve the accuracy of customer load estimation in business expansion or distribution network planning. The calculation formula for the user's load density is:

[0106] ,

[0107] in, Indicates the user's building area.

[0108] Aggregating all user load characteristics extracted from the three dimensions of load level, typical peak-valley characteristics, and long-term load development law can obtain the final load characteristics of the user. L f = [ l 1 f , ⋯ , l q f ] ,in, Represents the total number of load characteristics.

[0109] Step 3: Based on the typical load curve data and load characteristic data of multiple users, the power consumption pattern of each user is divided through a clustering algorithm.

[0110] The first The typical load curve data and load characteristic data of each user can be combined to obtain the Electricity consumption characteristic data of each user L i = [ l 1 d , ⋯ , l n d , l 1 f , ⋯ , l q f ] , The electricity consumption characteristics of a user group can be expressed as L = [ L 1 , ⋯ , L N ] . In order to solve the problem that the clustering algorithm is easy to iterate to the local optimal solution due to defects in the initialization of the user load type center point, the present invention adopts the k-means++ algorithm to optimize the selection of the initial cluster center through probability prior. At the same time, in view of the unknown number of clusters, the hybrid clustering evaluation index is comprehensively utilized through the heuristic algorithm to guide the optimal cluster number k of the sample set. In the process of initializing the cluster center, the farther the distance between the sample and the cluster center, the more likely it is to become a new cluster center. The specific process of the k-means++ algorithm is as follows:

[0111] Step 301: From the dataset L = [ L 1 , ⋯ , L N ] Randomly select any sample point as the initial cluster center .

[0112] Step 302: Calculate each sample point The shortest distance to the current cluster center , and then define the distance ratio as The probability of being selected as the next cluster center, and the sample point with the largest probability value is selected as the next cluster center ,Right now:

[0113] .

[0114] Step 303: Repeat step 302 until a Cluster centers.

[0115] Due to the lack of real labels of user power load samples, the present invention uses cluster internal evaluation indicators to evaluate the clustering results, and measures the actual performance of clustering by measuring the compactness within the cluster and the distinction between clusters. Common cluster internal evaluation indicators include silhouette coefficient (SC), Karlinsky-Harabas index (CHI) and Davis-Boulting index (DBI). By applying heuristic algorithms to cluster evaluation indicators, the optimal number of clusters k can be automatically identified.

[0116] The SC index is usually expressed as the ratio of the distance from each sample point to the sample in its cluster to the distance between its nearest cluster structure. The smaller the ratio, the farther the distance between the current cluster structure of the sample and its nearest cluster structure, that is, the better the clustering result. This index integrates both cohesion and separation into the evaluation of clustering performance, and has a better clustering evaluation effect. For the data set Samples in , respectively define the distance and .in, Representation sample The average distance to other samples in the same cluster, used to characterize the sample Dissimilarity within a cluster. Representation sample The minimum average distance to all samples of any other cluster, sample The silhouette coefficient It can be expressed as:

[0117] .

[0118] Dataset The silhouette coefficient By calculating the set The arithmetic mean of the silhouette coefficients of all samples in is obtained, and the calculation formula is:

[0119] .

[0120] In the CHI index, separation (SSB) is usually defined as the sum of the squares of the distances between the center of each cluster and the center of the entire data set, and tightness (SSW) is defined as the sum of the squares of the distances between each point in the cluster and the center of the cluster. The CHI index can be calculated by the ratio of separation to tightness. The larger the CHI index value, the higher the tightness within the cluster and the greater the dispersion between clusters, which can be expressed as:

[0121] ,

[0122] ,

[0123] ,

[0124] in, Indicates The number of samples in a cluster, is the total number of samples, is the number of clusters, Representation sample The cluster center point of the cluster to which it belongs, Indicates The center point of the cluster, is the center point of the entire data set.

[0125] The DBI index is usually expressed as the sum of the average distances from the samples in any two clusters to the cluster centers divided by the maximum distance between the center points of the two clusters. It can be specifically expressed as:

[0126] ,

[0127] in, Indicates All samples in clusters to cluster Center Point The average value of the distance, and Respectively represent clusters and Cluster The larger the DBI index value, the more dispersed the distribution within the cluster, and the closer the distribution between clusters.

[0128] By extracting the electricity usage patterns of users, it is possible to label the electricity usage patterns of dispersed users.

[0129] Step 4: Based on the user's typical load curve data, load characteristic data and power consumption pattern, obtain the user's power consumption characteristic analysis results.

[0130] In an embodiment of the present invention, obtaining the user's electricity consumption characteristic analysis results specifically includes: inputting the user's typical load curve data and load characteristic data into a pre-trained daily load curve prediction model under the corresponding power consumption mode, and obtaining the user's daily load curve prediction results.

[0131] The daily load curve prediction model of the present invention is constructed in parallel based on convolutional neural networks to accurately model the user's daily electricity consumption characteristics, helping users to understand their own electricity consumption characteristics more accurately.

[0132] like Figure 3 As shown in the figure, the convolutional neural network architecture is usually composed of alternating convolutional layers and pooling layers to filter and extract useful features from the input data. The leftmost side of the figure is the multi-dimensional time series data of the user's power grid load. Each convolution layer has a corresponding convolution kernel (filter), and each box on the input data in the figure represents a filter. The filter slides from top to bottom on the entire input matrix, and produces the convolution features of the input data through the coefficient matrix contained. The dimension of the convolution feature extracted by the filter is ,in It is related to the input dimension, filter size and convolution stride. Assume that the number of convolution kernels applied to the input data is , then the extracted convolution feature dimension is The convolution layer is usually followed by a nonlinear activation function, and then followed by a pooling layer. The embodiment of the present invention uses the Sigmoid function as a nonlinear activation function responsible for mapping the input of the neuron to the output end, and its function expression is as follows:

[0133] .

[0134] The pooling layer is a subsampling technique that can convert each convolution feature matrix into a low-dimensional feature matrix and aggregate it based on specific rules. For example, under the max-pooling rule, the maximum value in the current sliding window will be output, which is the most critical feature in the window. The pooling operation can enhance the robustness of the system and reduce the sensitivity of the pooling output to input changes.

[0135] Embodiment 2:

[0136] Based on the same inventive concept as in Example 1, the embodiment of the present invention further provides a decentralized user power consumption characteristic analysis device for implementing the decentralized user power consumption characteristic analysis method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the decentralized user power consumption characteristic analysis device embodiment provided below can refer to the limitations of the decentralized user power consumption characteristic analysis method above, and will not be repeated here.

[0137] like Figure 4As shown, an embodiment of the present invention provides a decentralized user power consumption characteristic analysis device, comprising:

[0138] A historical power grid load data acquisition module is used to acquire historical power grid load data of multiple users;

[0139] A typical load curve and load characteristic acquisition module is used to acquire the corresponding typical load curve data and load characteristic data based on the historical power grid load data of each user;

[0140] The power consumption mode classification module is used to classify the power consumption mode of each user through a clustering algorithm according to the typical load curve data and load characteristic data of multiple users;

[0141] The power consumption characteristic analysis module is used to obtain the user's power consumption characteristic analysis results based on the user's typical load curve data, load characteristic data and power consumption mode.

[0142] Embodiment 3:

[0143] The embodiment of the present invention further provides a computer device, which may be a server, and its internal structure diagram may be as shown in FIG. Figure 5 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the decentralized user power consumption characteristic analysis method in the aforementioned embodiment is implemented.

[0144] Those skilled in the art will understand that Figure 5 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0145] Embodiment 4:

[0146] The embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the following method are implemented:

[0147] Obtain historical grid load data for multiple users;

[0148] Based on the historical grid load data of each user, obtain the corresponding typical load curve data and load characteristic data;

[0149] Based on the typical load curve data and load characteristic data of multiple users, the power consumption pattern of each user is divided through clustering algorithm;

[0150] Based on the user's typical load curve data, load characteristic data and power consumption pattern, the user's power consumption characteristic analysis results are obtained.

[0151] It should be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products, and therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0152] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0153] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0154] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0155] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the enlightenment of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the purpose of the present invention and the claims, which all fall within the protection of the present invention.

Claims

1. A method for analyzing the power consumption characteristics of dispersed users, characterized in that: include: Obtain historical grid load data for multiple users; Based on the historical grid load data of each user, obtain the corresponding typical load curve data and load characteristic data; Based on the typical load curve data and load characteristic data of multiple users, the power consumption pattern of each user is divided through clustering algorithm; Based on the user's typical load curve data, load characteristic data and power consumption pattern, the user's power consumption characteristic analysis results are obtained.

2. The method for analyzing the power consumption characteristics of dispersed users according to claim 1, characterized in that: Also includes: Perform data cleaning on the acquired historical power grid load data.

3. The method for analyzing the power consumption characteristics of dispersed users according to claim 1, characterized in that: Based on the user's historical grid load data, obtain the user's typical load curve data, including: The user's historical power grid load data is divided into multiple historical daily load data by day, and the historical daily load data is represented as ; The historical daily load data is reduced in dimension, and the data after dimension reduction is ,in, The calculation formula is: ， By using the symbolic set approximation method, according to The size, yes The data is symbolized and assigned values ​​to obtain a discrete symbol sequence ; statistics The frequency of occurrence of various symbol sequences in the , and filter out the most common ones A sequence of symbols; Historical daily load data In and described The load sequences corresponding to the symbol sequences are superimposed and averaged to obtain the typical load curve data of the user.

4. The method for analyzing the power consumption characteristics of dispersed users according to claim 1, characterized in that: Based on the user's historical grid load data, obtain the user's load characteristic data, including: Divide the user's historical grid load data into multiple time series by year, month, and day; The load characteristic data of the users are obtained by extracting the characteristics of the divided time series from three indicators: load level, typical peak-valley characteristics and long-term load development law. Among them, load level indicators include power consumption, maximum load, minimum load and peak-to-valley difference; typical peak-to-valley characteristic indicators include load rate, minimum load rate and peak-to-valley difference rate; long-term load development law indicators include practical coefficient, stage coefficient and saturated load density.

5. The method for analyzing the power consumption characteristics of dispersed users according to claim 4, characterized in that: User in year The calculation formula of the practical coefficient is: ， in, Indicates the user's year The annual maximum load, Indicates the capacity reported by the user when applying for access to electricity; The calculation formula of the user's terminal utility coefficient is: ， in, Indicates the final year, ; Indicates the year from when the user connects to electricity to the end of the term Annual maximum load; User in the year The calculation formula of the stage coefficient is: ； The calculation formula for user load density is: ， in, Indicates the user's building area.

6. The method for analyzing the power consumption characteristics of dispersed users according to claim 1, characterized in that: The clustering algorithm includes a k-means++ algorithm; The analysis method also includes: identifying the optimal number of clusters k by applying a heuristic algorithm to a clustering evaluation index of a clustering algorithm; the clustering evaluation index includes a silhouette coefficient, a Kalinsky-Harabas index, and a Davis-Boulting index.

7. The method for analyzing the power consumption characteristics of dispersed users according to claim 1, characterized in that: Obtaining the user's electricity consumption characteristic analysis results, including: inputting the user's typical load curve data and load characteristic data into a pre-trained daily load curve prediction model under the corresponding power consumption mode, and obtaining the user's daily load curve prediction results.

8. A device for analyzing the power consumption characteristics of decentralized users, characterized in that: include: A historical power grid load data acquisition module is used to acquire historical power grid load data of multiple users; A typical load curve and load characteristic acquisition module is used to acquire the corresponding typical load curve data and load characteristic data based on the historical power grid load data of each user; The power consumption mode classification module is used to classify the power consumption mode of each user through a clustering algorithm according to the typical load curve data and load characteristic data of multiple users; The power consumption characteristic analysis module is used to obtain the user's power consumption characteristic analysis results based on the user's typical load curve data, load characteristic data and power consumption mode.

9. A computer device, characterized in that: include: Memory for storing computer programs; A processor is used to execute the computer program to implement the steps of the decentralized user power consumption characteristics analysis method according to any one of claims 1 to 7.

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

Images