A demand response adjustable user potential evaluation method and system based on WGCNA

Abstract

The present invention proposes a demand response adjustable user potential assessment method and system based on WGCNA, which relates to the field of distribution network reconstruction optimization technology. Compared with traditional methods that can only process point-to-point, linear or Euclidean distance relationships, the WGCNA of the present invention can capture more complex nonlinear structural associations between user load curves by constructing a network, realize higher-dimensional user load behavior pattern recognition, and can characterize the complex correlation structure between load curves. Traditional clustering methods often rely on cluster center vectors, resulting in poor aggregation effects on non-convex and asymmetric user curves, and are unable to retain the true peak and valley fluctuation information of user load curves. The present invention retains the dynamic morphological consistency characteristics between individuals, retains the overall dynamic morphology of the load curve, and does not rely on the "averaging loss" of traditional clustering centers.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network reconstruction optimization, and in particular to a demand response adjustable user potential evaluation method and system based on WGCNA. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] User load response assessment systematically evaluates the ability, extent, and effectiveness of users to adjust their electricity consumption in response to electricity price signals, incentive mechanisms, or grid dispatch instructions. User load response assessment plays multiple key roles in power system operation, energy management, and market mechanism design. User load clustering directly impacts the accuracy of user load response assessment.

[0004] Currently, user load clustering methods primarily rely on classic algorithms such as K-Means (an unsupervised learning clustering algorithm), DBSCAN (density-based spatial clustering with applications and noise, or density clustering), and hierarchical clustering. These algorithms typically construct a feature vector space by manually selecting statistical features (such as peak load and load factor) and classify users into several clusters. However, these methods have three shortcomings:

[0005] (1) It relies heavily on manual experience, and feature selection is subjective;

[0006] (2) The complete time series information of the 24-hour load curve cannot be retained, resulting in information loss;

[0007] (3) The collaborative electricity consumption relationship between users is ignored, and there is a lack of topological structure modeling, which makes it difficult to reveal potential behavioral modules and response collaborative groups.

[0008] Although some existing technologies have introduced techniques such as dynamic time warping (DTW), principal component analysis (PCA), or t-distributed stochastic neighbor embedding (t-SNE) to improve clustering performance, these methods are often computationally complex, sensitive to parameters, and difficult to achieve unified expression of structural modeling and potential evaluation of relationships between users. Summary of the Invention

[0009] The purpose of the invention is to provide a demand-responsive and adjustable user potential assessment method and system based on WGCNA. In response to the problems existing in the existing technology, by constructing a network through WGCNA, it is possible to capture more complex nonlinear structural correlations between user load curves and achieve higher-dimensional user load behavior pattern recognition.

[0010] To achieve the above purpose, the invention adopts the following technical solutions:

[0011] A first aspect of the present invention provides a demand response adjustable user potential assessment method based on WGCNA, comprising the following steps:

[0012] Acquiring power load data and preprocessing the power load data;

[0013] Construct a user correlation matrix based on the pre-processed electricity load data;

[0014] By introducing the exponential soft threshold function, the correlation matrix is ​​transformed into a weighted adjacency matrix;

[0015] Wherein, the soft threshold index is determined by the WGCNA standard process;

[0016] Based on the weighted adjacency matrix, the topological overlap matrix is ​​constructed;

[0017] A hierarchical clustering method is used to perform cluster analysis on the topological overlap matrix to identify the number of power consumption behavior modules; and the optimal number of clusters is determined using the DBI indicator.

[0018] Extract characteristic load curves for each type of electricity consumption behavior module;

[0019] The demand response adjustable user potential of each electricity behavior module is evaluated according to the characteristic load curve.

[0020] Furthermore, the obtaining of electricity load data specifically includes: obtaining electricity load data from an energy consumption monitoring system, a smart meter or a related energy management platform.

[0021] Furthermore, the Spearman rank correlation coefficient is used as a preliminary user similarity evaluation indicator to construct the user correlation matrix.

[0022] Furthermore, the weighted adjacency matrix is ​​constructed using the Pearson correlation coefficient.

[0023] Furthermore, methods for evaluating the demand response adjustable user potential of each electricity behavior module include:

[0024] Calculate the response potential of the central user in the electricity consumption behavior module, the weighted response potential of individual users in the electricity consumption behavior module, and the comprehensive response potential;

[0025] First, the response potential of the central user in the electricity consumption behavior module and the weighted response potential of individual users in the electricity consumption behavior module are calculated. Then, the comprehensive response potential index is defined by combining the central index of the electricity consumption behavior module and the individual weighted index.

[0026] Furthermore, the demand response adjustable user potential of each electricity behavior module is evaluated by time period, including:

[0027] First, the central response potential assessment of the electricity consumption behavior module uses time as the dimension to calculate the degree of matching between the central curve of each module and the unified load criterion in different time periods, and measure the overall adjustability of the module in each time period; secondly, the weighted response potential assessment of individual users within the module continues to be carried out within the time period, evaluating the weighted response capabilities of individual users in different time periods; finally, the comprehensive response potential assessment is based on the analysis results of the first two steps in the time dimension, integrating the time period response characteristics at the module level and the individual user level, and deriving the demand response adjustable user potential distribution for the entire power system in each key time period.

[0028] A second aspect of the present invention provides a WGCNA-based demand-response adjustable user potential assessment system, based on a WGCNA-based demand-response adjustable user potential assessment method provided in the first aspect of the present invention, comprising:

[0029] The data acquisition and processing module is configured to: acquire power load data and pre-process the power load data;

[0030] The topological overlap matrix construction module is configured to: construct a user correlation matrix based on preprocessed power load data; convert the correlation matrix into a weighted adjacency matrix by introducing an exponential soft threshold function; wherein the soft threshold index is determined according to the WGCNA standard process; and construct a topological overlap matrix based on the weighted adjacency matrix;

[0031] The clustering module is configured to: perform cluster analysis on the topological overlap matrix using a hierarchical clustering method to identify the number of types of power consumption behavior modules; and determine the optimal number of clusters using the DBI indicator;

[0032] The feature extraction module is configured to extract a characteristic load curve of each type of power consumption behavior module as a representative load behavior pattern of the power consumption behavior module;

[0033] The result output module is configured to compare the characteristic load curve of the electricity consumption behavior module with the standard load baseline, and evaluate the demand response adjustable user potential of each electricity consumption behavior module by combining the load fluctuation amplitude, response depth and duration indicators to generate an adjustable load resource portrait.

[0034] A third aspect of the present invention provides a computer program product, which, when executed by a processor, implements the steps of the WGCNA-based demand-response adjustable user potential assessment method as described in the first aspect of the present invention.

[0035] A fourth aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of the WGCNA-based demand-response adjustable user potential assessment method as described in the first aspect of the present invention.

[0036] A fifth aspect of the present invention provides an electronic device, comprising a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps of the WGCNA-based demand-responsive adjustable user potential assessment method as described in the first aspect of the present invention are implemented.

[0037] The technical solution of the present invention has the following beneficial effects:

[0038] 1. Traditional methods (such as direct similarity matching, KMeans clustering, and direct distance measurement) can only process point-to-point, linear, or Euclidean distance relationships. However, the WGCNA of the present invention, by constructing a network, can capture more complex nonlinear structural associations between user load curves, achieve higher-dimensional user load behavior pattern recognition, and characterize the complex correlation structure between load curves.

[0039] 2. Traditional clustering methods (such as K-Means) often rely on cluster center vectors, resulting in poor clustering performance for non-convex and asymmetric user curves and failing to preserve the true peak-to-valley fluctuations of user load curves. However, the WGCNA method of the present invention divides user groups into modules across the entire network, preserving the dynamic consistency of individual models. This method is particularly suitable for processing the complex fluctuations of 24-hour power load curves. This method preserves the overall dynamic shape of the load curve and does not rely on the "averaging loss" of traditional cluster centers.

[0040] 3. The TOM-based module partitioning in the WGCNA of the present invention is more stable in the presence of small amounts of noise or data fluctuations. Unlike traditional K-Means or hierarchical clustering, WGCNA's module partitioning does not rely on an initial random seed and is not sensitive to the choice of distance function. It is well-suited to the naturally "noisy" characteristics of real-world residential load curve data. This makes clustering results more interpretable. For example, in experiments, it was possible to clearly identify electricity users with distinct physiological behavioral characteristics, such as "households with high load throughout the day," "nighttime peak users," and "midday electricity users." The module (clustering) of the present invention is highly stable, and the identification results are consistent with physiological behavior.

[0041] The WGCNA of this invention calculates correlations between clustered modules and external variables (in this study, load criterion), further quantifying the degree of responsiveness of the module as a whole to the response standard. By calculating the Pearson correlation coefficient between the module characteristic curve and the criterion, it is possible to clearly assess which groups have the greatest response potential. Furthermore, representative users can be further identified within the module, enabling the formulation of precise demand response incentive strategies. This invention supports the association of modules with behavioral factors (such as load criterion), enhancing interpretability and operability.

[0042] 5. This invention is highly scalable and suitable for subsequent load forecasting, user classification, anomaly detection, and other tasks. The load-user network constructed can be directly used for other electricity behavior research tasks. This invention can predict load curves based on module center vectors; build user profiles by combining behavioral tags; and identify abnormal user behavior for power fraud detection or abnormal electricity usage diagnosis.

[0043] 6. The present invention does not require manual feature extraction, and the full curve input retains the original time series information of the load data; it has network modeling capabilities and can characterize the potential collaborative power consumption patterns among users; it identifies clearly structured response modules and supports module-level and user-level potential quantification; it is suitable for large-scale user data analysis and has good scalability and engineering adaptability.

[0044] 7. The present invention provides a highly robust, highly structured, demand-responsive and adjustable user identification method. In the process of constructing a new power system, it can provide a strong analytical basis and scheduling support for the establishment of a coordinated interaction mechanism of "source, grid, load and storage" and demand-side resource regulation, and has significant application prospects and promotion value.

[0045] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0047] Figure 1 This is a framework diagram of an adjustable user potential assessment method based on WGCNA demand response according to the present invention;

[0048] Figure 2 (a) shows the average connectivity curve corresponding to different β values;

[0049] Figure 2(b) shows the scale-free fitting index corresponding to different β values curve;

[0050] Figure 3 It is a clustering algorithm DBI indicator of an adjustable user potential evaluation method based on WGCNA demand response of the present invention;

[0051] Figure 4 This is a hierarchical clustering effect diagram of an adjustable user potential evaluation method based on WGCNA demand response in the present invention;

[0052] Figure 5 It is a cluster scatter plot of an adjustable user potential evaluation method based on WGCNA demand response of the present invention;

[0053] Figure 6 is the correlation coefficient of each power consumption mode of an adjustable user potential evaluation method based on WGCNA demand response of the present invention;

[0054] Figure 7 It is a cluster center curve of an adjustable user potential evaluation method based on WGCNA demand response of the present invention;

[0055] Figure 8(a) shows the load curve clustering result of the first type of electricity consumption mode. There are 179 samples in the first type. The blue curve in the figure represents the cluster center curve of this type.

[0056] Figure 8(b) shows the load curve clustering result of the second type of electricity consumption mode. There are 553 samples in the second type. The blue curve in the figure represents the cluster center curve of this type.

[0057] Figure 8(c) shows the load curve clustering result of the third type of electricity consumption mode. There are 354 samples in the third type and 553 samples in the second type. The blue curve in the figure represents the cluster center curve of this type.

[0058] Figure 8(d) shows the load curve clustering result of the fourth type of electricity consumption mode. There are 134 samples in the fourth type. The blue curve in the figure represents the cluster center curve of this type.

[0059] Figure 8(e) shows the load curve clustering result of the fifth type of electricity consumption mode. There are 180 samples in the fifth type. The blue curve in the figure represents the cluster center curve of this type.

[0060] Figure 9 is a comprehensive evaluation result of an adjustable user potential evaluation method based on WGCNA demand response of the present invention;

[0061] Figure 10 (a) shows the potential evaluation scores of central users with five electricity consumption patterns during peak, flat, and valley periods based on the comprehensive evaluation method of the WGCNA method.

[0062] Figure 10(b) shows the potential evaluation scores of individual users with five electricity consumption patterns during peak, flat, and off-peak periods based on the comprehensive evaluation method of the WGCNA method.

[0063] Figure 10(c) shows the potential evaluation scores of the five electricity consumption patterns under the comprehensive evaluation based on the WGCNA method in the peak, flat and valley periods.

[0064] Figure 11 The present invention is a flow chart of an adjustable user potential evaluation method based on WGCNA demand response. DETAILED DESCRIPTION

[0065] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0066] It should be noted that the terms used herein are for describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present invention.

[0067] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0068] Example 1

[0069] WGCNA stands for Weighted Gene Co-Expression Network Analysis. WGCNA is a bioinformatics analysis method based on gene expression data. The core idea is to construct a co-expression network between genes, mine gene modules with coordinated expression patterns, and explore the association between these modules and phenotypes or external variables, thereby screening out key genes (hub genes) that may be involved in specific biological processes.

[0070] This embodiment provides a demand response adjustable user potential evaluation method based on WGCNA, such as Figure 1 and Figure 11 As shown, the following steps are taken:

[0071] Step 1: Obtain power load data and preprocess the power load data.

[0072] Step 1.1: Obtain electricity load data. Specifically, obtain electricity load data from energy consumption monitoring systems, smart meters, or related energy management platforms. This data may include load data at different time scales (e.g., hourly, daily, and monthly). Ensure that the data covers a sufficiently long period of time to capture various patterns in energy consumption data.

[0073] Preferably, the power load data of multiple users within a period of time at a uniform time granularity (such as 24 hours per day) is collected.

[0074] Step 1.2: Preprocess the power load data.

[0075] Step 1.2.1 Data cleaning:

[0076] Remove noise and missing values. For missing values, you can use mean filling:

[0077] (1)

[0078] in, is data point a, n is the number of data points, For the i Electricity consumption data ( i =1).

[0079] Or using linear interpolation:

[0080] (2)

[0081] in, t is the time point corresponding to the missing value, and For adjacent time points, and is the data value at the corresponding time point.

[0082] Data standardization. Standardize or mean-center the power load data of each user to eliminate the impact of the difference in power consumption scale of different users on the correlation calculation. First, consider a power curve, record t The power load data at the moment is l t This power curve is in the period T The total amount of electricity in the battery is:

[0083] (3)

[0084] The power curve is normalized based on the total power w. The value at each moment after normalization is l * t ,Right now l * t =lt / w. l * t The constructed time series is recorded as the shape of the original power curve.

[0085] Step 2: Construct a user correlation matrix based on the preprocessed electricity load data.

[0086] Considering that user load curves may have large differences in numerical scale, but the trend (i.e. the relative position of peak and valley power consumption) can better reflect the power consumption behavior pattern, the present invention preferentially adopts the Spearman rank correlation coefficient as the preliminary user similarity evaluation index to construct the user correlation matrix :

[0087] (4)

[0088] in r i,t、 r j,t For users i 、 j For the moment t The Spearman coefficient can effectively suppress the interference of extreme values ​​on correlation and is suitable for situations where load data has abnormal fluctuations. Each row in the user correlation matrix represents a user, and the columns represent the power load values ​​at the corresponding time.

[0089] Step 3: Construct a weighted adjacency matrix based on the correlation matrix.

[0090] In determining the soft threshold index β In the process, the Spearman correlation matrix in step 2 is still used. S ( s ), find the network closest to the power law network through the WGCNA standard process (pickSoftThreshold) β However, when constructing the network adjacency matrix, in order to preserve the true numerical structure and linear coupling relationship of the load curve between users, the Pearson correlation coefficient is used. S ( p )Construct a weighted adjacency matrix :

[0091] (5)

[0092] in:

[0093] (6)

[0094] in, and After normalization i and j Power load data. is the soft threshold index, which controls the sparsity of the network. The smallest value that satisfies the scale-free topology is selected. β , that is, fitting Relationship reached >0.85. This approach combines the robustness of trend consistency assessment with the linear accuracy of network structure expression, which helps to improve the stability and topological rationality of power consumption behavior module identification.

[0095] like Figure 4 As shown, Figure 4This figure shows the hierarchical clustering results of user electricity usage behavior using the WGCNA method. This figure illustrates the similarity structure between users' behavioral characteristics, as well as the resulting multiple highly correlated user cluster modules. Different colors represent different electricity usage patterns: orange lines represent pattern 1, green lines represent pattern 2, red lines represent pattern 3, purple lines represent pattern 4, and brown lines represent pattern 1.

[0096] like Figure 6 As shown, Figure 6 The correlation coefficients between different electricity usage patterns are displayed, visually reflecting the degree of similarity between modules. This figure demonstrates that the modules in the constructed behavioral network have good differentiation and internal consistency, which serves as the basis for subsequent evaluation of module response potential.

[0097] As shown in Figure 2(a)-Figure 2(b), the soft threshold is selected β =9, the figure shows different β The scale-free fitting index corresponding to the value and average connectivity curve; specifically, Figure 2 (a) is different β The average connectivity curve corresponding to the value is shown in Figure 2(b). β The scale-free fitting index corresponding to the value curve.

[0098] As a rule of thumb, when β When the value is 9, The value exceeded 0.85 for the first time and tended to be stable while maintaining an appropriate level of average connectivity, so this value was selected as the final soft threshold for constructing subsequent network models.

[0099] Step 4: Construct a topological overlap matrix based on the weighted adjacency matrix;

[0100] According to the weighted adjacency matrix , calculate the topological overlap between each pair of nodes:

[0101] (7)

[0102] in, Represents the number of weighted neighbors that are commonly connected, and the formula is:

[0103] (8)

[0104] in, is the relationship between node i and its neighbors The weighted connection value of is the relationship between node j and its neighbors The weighted connection value of is the number of nodes, For the nodes.

[0105] represents the weighted number of commonly connected neighbors; For users i The weighted degree, For users j The weighted degree is calculated by the formula Equivalence;

[0106] A ij For users i and j The direct connection strength is obtained. Finally, the TOM matrix T∈R N×N The TOM matrix has the ability to remove noise and can reflect the “structural similarity” between users in the network.

[0107] By converting the correlation matrix into a topological overlap matrix (TOM), the algorithm not only considers the direct similarity between two users but also the similarity of "shared neighbors," thereby improving its robustness to group behavioral similarity. For example, if two users may not be directly correlated, but are highly correlated with the same set of typical electricity usage behaviors, WGCNA can identify this "indirect co-expression" association.

[0108] Step 5: Analyze the topological overlap matrix to obtain the label vector of the power consumption behavior module;

[0109] use As a distance matrix, perform hierarchical clustering on all users. Then use DBI (Davies-Bouldin) indicator, select the optimal number of clusters:

[0110] (9)

[0111] in: For class i The average distance between samples within the For class j Intra-sample mean distance; d ( c i , c j ) For class i 、 j The distance between the centers of K is the number of clusters.

[0112] like Figure 3 As shown, when the number of clusters is, for example KWhen DBI = 5, it reaches its minimum value, indicating that the electricity usage behavior module is most reasonable. Therefore, this number of clusters is selected as the final number of behavior module divisions in this paper. The DBI index calculated at different cluster numbers is used to evaluate the rationality of the clustering results. DBI reflects the balance between the degree of intra-cluster closeness and the degree of separation between clusters. The smaller the value, the better the clustering effect.

[0113] like Figure 5 As shown, Figure 5 This tool visualizes the distribution of users with different electricity usage behaviors in the clustering results. Each point represents a user, and their behavior characteristics are mapped into a two-dimensional space. Different colors or markers represent different module categories, thereby verifying the separability and representativeness of the clustering results.

[0114] Step 6: Based on the WGCNA user cluster analysis and electricity consumption behavior module, evaluate the demand response potential of various user groups.

[0115] Specifically, a three-tiered evaluation method is proposed to assess the demand response potential of various user groups, including the response potential of the central user in the electricity behavior module, the weighted response potential of individual users within the electricity behavior module, and finally the comprehensive response potential. (First, the response potential of the central user in the electricity behavior module and the weighted response potential of individual users within the electricity behavior module are calculated, and then the comprehensive response potential is calculated).

[0116] (I) Evaluation of the central response potential of the electricity consumption behavior module. For each clustered electricity consumption behavior module, the most representative central user (i.e., the user with the greatest correlation with the main component load curve of the electricity consumption behavior module) is selected, and the standardized load curve of a certain type of electricity consumption pattern is set as x c ( t ), the unified load line is B ( t ), in order to measure the similarity between the behavior module and the benchmark load curve, the behavior pattern response consistency index is defined R ME is the correlation coefficient between the two, and the calculation formula is as follows:

[0117] (10)

[0118] in, represents the central user curve, Indicates the load line, and They represent the mean of the central user curve and the load directrix respectively.

[0119] In order to characterize the response potential of the central user, the load difference with the guideline is introduced D norm :

[0120] (11)

[0121] Based on this, the response potential index of the electricity consumption behavior module center is defined as:

[0122] (12)

[0123] This indicator comprehensively considers the difference and directional consistency between the central user and the guideline. The greater the difference and the lower the correlation, the higher the response potential.

[0124] (2) Assessment of weighted response potential of individual users within the electricity consumption behavior module.

[0125] For each electricity behavior module, n users u i , respectively calculated R ij 、 D ij and kME i Three indicators, specifically:

[0126] R ij :user i With load alignment B ( t ) correlation coefficient; the calculation formula is:

[0127] (13)

[0128] x i ( t )and B ( t ) respectively represent t When the user i Normalized data of electricity load; Represents the mean of the user curve.

[0129] D ij :user i Average deviation in the area of ​​difference from the directrix:

[0130] (14)

[0131] kME i :user i The correlation coefficient with the principal component of this electricity consumption behavior module indicates its representativeness.

[0132] (15)

[0133] kME i Represents a user i The correlation coefficient with the load curve of the central user of the power consumption mode is in the range of [-1,1]; ME ( t ) represents the load curve of the central user in the power consumption mode.

[0134] Based on this, the weighted average response potential index is constructed:

[0135] (16)

[0136] This indicator takes into account the difference between each user and the load line ( D ij )、directional consistency(| R ij |) and representative weight ( kME i ), thereby improving the comprehensiveness and robustness of response assessment.

[0137] Figures 8(a)-8(e) show the clustering results of user load curves obtained based on the WGCNA method, reflecting the clustering and differences in the power consumption curves of user groups. Each group of curves represents a power consumption pattern, highlighting the ability of the present invention to cluster and classify complex load behavior data. Specifically:

[0138] Figure 8 (a) shows the load curve clustering results of the first type of electricity consumption mode. There are 179 samples in the first type. The blue curve in the figure represents the cluster center curve of this type.

[0139] Figure 8(b) shows the load curve clustering results of the second type of electricity consumption mode. There are 553 samples in the second type. The blue curve in the figure represents the cluster center curve of this type.

[0140] Figure 8 (c) shows the load curve clustering results of the third type of electricity consumption mode. There are 354 samples in the third type and 553 samples in the second type. The blue curve in the figure represents the cluster center curve of this type.

[0141] Figure 8 (d) shows the load curve clustering results of the fourth type of electricity consumption mode. There are 134 samples in the fourth type. The blue curve in the figure represents the cluster center curve of this type.

[0142] Figure 8(e) shows the load curve clustering results of the fifth type of electricity consumption mode. There are 180 samples in the fifth type. The blue curve in the figure represents the cluster center curve of this type.

[0143] The clustering result is the basic data structure for implementing subsequent response potential calculations and supports the overall evaluation framework of the present invention.

[0144] (3) Comprehensive response potential assessment

[0145] In order to uniformly characterize the overall response potential of each electricity consumption behavior module, the central indicator of the electricity consumption behavior module and the individual weighted indicator are combined to define the comprehensive response potential index as follows:

[0146] (17)

[0147] in, α ∈[0,1] is an adjustable parameter that controls the weight of the center perspective and the group average perspective; the recommended setting value is 0.3~0.5, which can be flexibly adjusted according to scene preferences.

[0148] Based on the above potential assessment results, users with higher response potential scores can be given priority in the demand response incentive mechanism to achieve precise load management and response strategy formulation based on behavioral characteristics.

[0149] (IV) Evaluation by time period

[0150] In this embodiment, "assessment by time period" runs through the three response potential assessment methods, and a multi-level assessment system is constructed from the whole to the individual, and from the local to the global. First, the central response potential assessment of the electricity consumption behavior module uses time as the dimension to calculate the degree of matching between the central curve of each module and the unified load criterion in different time periods, and measure the adjustability of the module as a whole in each time period; secondly, the weighted response potential assessment of individual users in the module continues to be carried out within the time period, and the weighted response capabilities of individual users in different time periods are assessed; finally, the comprehensive response potential assessment is based on the analysis results of the first two steps in the time dimension, integrating the time period response characteristics of the module level and the individual user level, and deriving the response potential distribution for the entire power system in each key time period. The three methods uniformly use the time period as the analysis basis to ensure that the assessment results not only reflect the group rules, but also accurately portray individual response behaviors.

[0151] like Figure 7 As shown, Figure 7 The cluster center curves for each electricity consumption pattern are displayed, representing the daily electricity load trends of the central user in each cluster. Different colors represent different electricity consumption patterns: red lines represent pattern 1, blue lines represent pattern 2, green lines represent pattern 3, orange lines represent pattern 4, and purple lines represent pattern 1. This graph reflects the differences in daily load characteristics for different electricity consumption patterns and is used to characterize the representative load characteristics of the user behavior module. By extracting the center curves of each cluster, it is possible to identify central users that are representative and controllable in demand response, providing a benchmark load behavior template for subsequent response potential assessment.

[0152] First, divide the day into three periods: peak period (08:00-11:00, 18:00-21:00), flat period (06:00-08:00, 11:00-18:00, 21:00-22:00), valley period (22:00-24:00, 00:00-06:00), and perform segmented evaluation for each period according to the above description.

[0153] 1. Single user response potential:

[0154] (18)

[0155] 2. Center user response potential:

[0156] (19)

[0157] 3. Comprehensive score:

[0158] (20)

[0159] in, is the mean absolute error of all users throughout the day; D u,T Represents a user u In the period T Total deviation from the guideline; R u,T Represents a user u In the period T Pearson correlation coefficient with the directrix; DR u,T Represents a user u In the period T Response potential score; DR center,T Indicates the cluster center user in the time period T responsiveness potential; DR user,T represents the average response potential of all users in the cluster in time period T; DR total,T is the comprehensive response potential of clustering in this period.

[0160] like Figure 9 As shown in the figure, the central user potential assessment scores, individual user potential assessment scores and comprehensive scores of the five electricity consumption modes are displayed under the comprehensive evaluation based on the WGCNA method.

[0161] As shown in Figure 10 (a) to Figure 10 (c), the potential evaluation scores of central users, individual users and comprehensive scores of five electricity consumption modes are displayed under the comprehensive evaluation based on the WGCNA method, during peak, flat and valley periods.

[0162] Figure 10 (a) shows the potential evaluation scores of central users with five electricity consumption patterns during peak, flat, and valley periods based on the comprehensive evaluation method of the WGCNA method.

[0163] Figure 10 (b) shows the potential evaluation scores of individual users with five electricity consumption patterns during peak, flat, and valley periods based on the comprehensive evaluation method of the WGCNA method.

[0164] Figure 10 (c) shows the potential assessment scores of the five electricity consumption patterns in the peak, flat, and valley periods under the comprehensive evaluation based on the WGCNA method.

[0165] Example 2

[0166] This embodiment discloses a demand-responsive and adjustable user potential assessment system based on WGCNA, which adopts the steps of the demand-responsive and adjustable user potential assessment method based on WGCNA described in Example 1, including:

[0167] The data acquisition and processing module is configured to: acquire power load data and pre-process the power load data;

[0168] The topological overlap matrix construction module is configured to: construct a user correlation matrix based on preprocessed power load data; convert the correlation matrix into a weighted adjacency matrix by introducing an exponential soft threshold function; wherein the soft threshold index is determined according to the WGCNA standard process; and construct a topological overlap matrix based on the weighted adjacency matrix;

[0169] The clustering module is configured to: perform cluster analysis on the topological overlap matrix using a hierarchical clustering method to identify the number of types of power consumption behavior modules; and determine the optimal number of clusters using the DBI indicator;

[0170] The feature extraction module is configured to extract a characteristic load curve of each type of power consumption behavior module as a representative load behavior pattern of the power consumption behavior module;

[0171] The result output module is configured to compare the characteristic load curve of the electricity consumption behavior module with the standard load baseline, and evaluate the demand response adjustable user potential of each electricity consumption behavior module by combining the load fluctuation amplitude, response depth and duration indicators to generate an adjustable load resource portrait.

[0172] Example 3

[0173] The purpose of this embodiment is to provide a computer program product. When executed by a processor, the computer program product implements the steps of the demand response adjustable user potential assessment method based on WGCNA as described in the first embodiment of the present disclosure.

[0174] Example 4

[0175] The purpose of this embodiment is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the WGCNA-based demand-response adjustable user potential assessment method as described in the first embodiment of the present disclosure.

[0176] Example 5

[0177] The purpose of this embodiment is to provide an electronic device. The electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the WGCNA-based demand-response adjustable user potential assessment method described in the first embodiment of the present disclosure are implemented.

[0178] The steps involved in the apparatuses of Examples 2, 3, 4, and 5 above correspond to those of Method Example 1. For detailed implementations, please refer to the relevant description of Example 1. The term "computer-readable storage medium" should be understood to mean a single medium or multiple media containing one or more instruction sets; it should also be understood to include any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and causing the processor to perform any method of the present invention.

[0179] Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using a general-purpose computer device. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. Alternatively, they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0180] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A demand-responsive and adjustable user potential assessment method based on WGCNA, characterized in that: The steps include: Acquiring power load data and preprocessing the power load data; Construct a user correlation matrix based on the pre-processed electricity load data; By introducing the exponential soft threshold function, the correlation matrix is ​​transformed into a weighted adjacency matrix; Among them, the soft threshold index is determined by the WGCNA standard process; Based on the weighted adjacency matrix, the topological overlap matrix is ​​constructed; A hierarchical clustering method is used to perform cluster analysis on the topological overlap matrix to identify the number of power consumption behavior modules; and the optimal number of clusters is determined using the DBI indicator. Extract characteristic load curves for each type of electricity consumption behavior module; The demand response adjustable user potential of each electricity behavior module is evaluated according to the characteristic load curve.

2. The WGCNA-based demand response adjustable user potential assessment method according to claim 1, characterized in that: The obtaining of electricity load data specifically includes: obtaining electricity load data from an energy consumption monitoring system, a smart meter or a related energy management platform.

3. The WGCNA-based demand response adjustable user potential assessment method according to claim 1, characterized in that: The Spearman rank correlation coefficient is used as a preliminary user similarity evaluation indicator to construct the user correlation matrix.

4. The WGCNA-based demand response adjustable user potential assessment method according to claim 1, characterized in that: The weighted adjacency matrix is ​​constructed using the Pearson correlation coefficient.

5. The WGCNA-based demand response adjustable user potential assessment method according to claim 1, characterized in that: Methods for evaluating the demand response adjustable user potential of each electricity behavior module include: Calculate the response potential of the central user in the electricity consumption behavior module, the weighted response potential of individual users in the electricity consumption behavior module, and the comprehensive response potential; The weighted response potential assessment of individual users in the electricity behavior module includes: n users u i , respectively calculated R ij 、 D ij and kME i Three indicators are used to construct a weighted average response potential index. The weighted average response potential index takes into account the difference between each user and the load line, the direction consistency and the representative weight; in, R ij Represents a user i Correlation coefficient with the load criterion; kME i Represents a user i Correlation coefficient with the principal components of this electricity consumption behavior module; D ij Represents a user i Average deviation in the area of ​​difference from the directrix: in, represents the central user curve, Indicates the load line; The comprehensive response potential index is: in, α ∈[0,1] is an adjustable parameter that controls the weight of the central perspective and the group average perspective; It is the response potential indicator of the electricity consumption behavior module center; is the weighted average response potential index; First, the response potential of the central user in the electricity consumption behavior module and the weighted response potential of individual users in the electricity consumption behavior module are calculated. Then, the comprehensive response potential index is defined by combining the central index of the electricity consumption behavior module and the individual weighted index.

6. The WGCNA-based demand response adjustable user potential assessment method according to claim 5, characterized in that: The demand response adjustable user potential of each electricity behavior module is evaluated by time period, including: First, the central response potential assessment of the electricity consumption behavior module uses time as the dimension to calculate the degree of matching between the central curve of each module and the unified load criterion in different time periods, and measure the overall adjustability of the module in each time period; secondly, the weighted response potential assessment of individual users within the module continues to be carried out within the time period, evaluating the weighted response capabilities of individual users in different time periods; finally, the comprehensive response potential assessment is based on the analysis results of the first two steps in the time dimension, integrating the time period response characteristics at the module level and the individual user level, and deriving the demand response adjustable user potential distribution for the entire power system in each key time period.

7. A demand-response adjustable user potential assessment system based on WGCNA, based on a demand-response adjustable user potential assessment method based on WGCNA according to any one of claims 1 to 6, characterized in that: include: The data acquisition and processing module is configured to: acquire power load data and pre-process the power load data; The topological overlap matrix construction module is configured to: construct a user correlation matrix based on preprocessed electricity load data; convert the correlation matrix into a weighted adjacency matrix by introducing an exponential soft threshold function; wherein the soft threshold exponent is determined according to the WGCNA standard process; and construct a topological overlap matrix based on the weighted adjacency matrix. The clustering module is configured to: perform cluster analysis on the topological overlap matrix using a hierarchical clustering method to identify the number of types of power consumption behavior modules; and determine the optimal number of clusters using the DBI indicator; The feature extraction module is configured to extract a characteristic load curve of each type of power consumption behavior module as a representative load behavior pattern of the power consumption behavior module; The result output module is configured to compare the characteristic load curve of the electricity consumption behavior module with the standard load baseline, and evaluate the demand response adjustable user potential of each electricity consumption behavior module by combining the load fluctuation amplitude, response depth and duration indicators to generate an adjustable load resource portrait.

8. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the WGCNA-based demand response adjustable user potential assessment method are implemented.

9. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps of the WGCNA-based demand response adjustable user potential assessment method are implemented.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the WGCNA-based demand response adjustable user potential assessment method are implemented.

Images