Load curve similarity visualization method based on multi-distance measurement and weekly calendar mode

Abstract

The invention discloses a load curve similarity visualization method based on multi-distance measurement and a weekly calendar mode, discloses a system with the load curve similarity visualization method based on the multi-distance measurement and the weekly calendar mode, and provides a load curve similarity evaluation method based on the multi-distance measurement. Comprehensively representing load curve differences from a plurality of dimensions such as time alignment and form similarity; constructing a similarity matrix reconstruction method based on a weekly calendar mode, and converting an annual time sequence load similar distance sequence into a weekly calendar thermal matrix with form similarity; according to a thermodynamic diagram-based visual analysis method, the distribution rule and mode characteristics of load curve similarity under weekly calendar dimensions are intuitively revealed.

Description

Technical Field

[0001] This invention relates to the field of power big data analysis and visualization technology, and in particular to a method for visualizing load curve similarity based on multiple distance metrics and a weekly calendar pattern. Background Technology

[0002] In power system load analysis, traditional load curve analysis methods mainly focus on tasks such as load clustering or classification, load forecasting, anomaly detection, and user segmentation, lacking visual representation tools for the changing patterns of load curve shapes. Existing methods primarily have the following characteristics:

[0003] 1) Traditional load analysis methods often use a single distance metric, such as Euclidean distance, which cannot fully reflect the multidimensional characteristics of load curves in terms of time alignment and shape similarity.

[0004] 2) Visualization methods are difficult to effectively show the annual cyclical pattern of load curve similarity under the weekly calendar pattern, and cannot intuitively reveal the visual differences between weekday and weekend load patterns.

[0005] 3) Load curve data visualization tools lack a method to transform numerical results into a "one-map-for-the-year" visual mode, which limits the breadth and depth of power application scenarios such as load management. Summary of the Invention

[0006] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention proposes a load curve similarity visualization method based on multiple distance metrics and a weekly calendar pattern. This method provides a load curve similarity assessment method based on multiple distance metrics, comprehensively characterizing load curve differences from multiple dimensions such as time alignment and morphological similarity; and constructs a similarity matrix reconstruction method based on a weekly calendar pattern, transforming the annual time-series load similarity distance sequence into a weekly calendar heatmap matrix based on morphological similarity; simultaneously, a heatmap-based visualization analysis method intuitively reveals the distribution patterns and characteristics of load curve similarity under the weekly calendar dimension.

[0007] This invention also proposes a system with a method for visualizing load curve similarity based on multiple distance metrics and a weekly calendar pattern.

[0008] The method for visualizing load curve similarity based on multiple distance metrics and weekly calendar patterns according to a first aspect of the present invention is characterized by comprising the following steps:

[0009] Obtain the daily load curve data for the whole year, and normalize the daily load curve data;

[0010] Based on the normalized daily load curve data, the multi-dimensional distance between adjacent daily load curves is calculated to obtain the adjacent daily similarity distance sequence.

[0011] The adjacent day similarity distance sequence is reconstructed into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week.

[0012] A heatmap is generated based on the aforementioned cyclical distance matrix to visualize the distribution patterns of load curve similarity.

[0013] The load curve similarity visualization method based on multiple distance metrics and weekly calendar patterns according to embodiments of the present invention has at least the following beneficial effects:

[0014] The method provided in this invention is based on a multi-distance metric-based load curve neighboring day similarity assessment method, which comprehensively characterizes load curve differences from multiple dimensions such as time alignment and morphological similarity; it constructs a similarity matrix reconstruction method based on a weekly calendar pattern, which transforms the annual time-series load similarity distance sequence into a weekly calendar heat matrix of morphological similarity; and it uses a heatmap-based visualization analysis method to intuitively reveal the distribution law and pattern characteristics of load curve similarity in the weekly calendar dimension.

[0015] According to some embodiments of the present invention, the normalization process employs a min-max normalization method, specifically as follows:

[0016]

[0017] Where x is the original daily load curve sequence, x min and x max These are the minimum daily load value and the maximum daily load value, respectively.

[0018] According to some embodiments of the present invention, the multidimensional distance includes at least one of Euclidean distance, Euclidean distance and first-order difference weighted distance, dynamic time warping distance, and derivative dynamic time warping distance.

[0019] According to some embodiments of the present invention, the derivative dynamic time warping distance is obtained by calculating the first derivative sequence of the load curve and applying the dynamic time warping algorithm to the first derivative sequence.

[0020] According to some embodiments of the present invention, the weekly distance matrix is ​​a 52-row, 7-column matrix, where the element in the i-th row and j-th column represents the load curve similarity distance from week j to week j+1 in the i-th week of the year.

[0021] According to some embodiments of the present invention, the heat map displays the distribution pattern of load similarity through color gradients, and is used to identify load pattern characteristics.

[0022] According to some embodiments of the present invention, the load pattern characteristics include at least one of the following: weekday start-of-week pattern abrupt change, weekend load pattern stability characteristics, seasonal load pattern evolution, intra-week load pattern periodicity, and abnormal load pattern detection.

[0023] According to some embodiments of the present invention, the method further includes inputting the heatmap as a multi-channel image feature into a deep learning model for load clustering, classification, or user behavior analysis.

[0024] According to some embodiments of the present invention, the daily load curve data is 96-point daily load curve data, and the sampling frequency is 15 minutes.

[0025] According to some embodiments of the present invention, when calculating the multidimensional distance between adjacent daily load curves, 364 similar distances between adjacent days are calculated for 365 days of the year.

[0026] According to some embodiments of the present invention, the dynamic time warping distance d DTW The solution is obtained using dynamic programming, and the calculation formula is as follows:

[0027]

[0028] Where π is a curved path that satisfies monotonicity, continuity, and boundary conditions, and the DTW distance is solved recursively through dynamic programming.

[0029] According to some embodiments of the present invention, the reconstruction of the weekly distance matrix is ​​based on the similarity distance sequence of adjacent days throughout the year, and the 364 distance values ​​are arranged in a matrix form of 52 weeks and 7 days according to the weekly pattern.

[0030] According to some embodiments of the present invention, the heat map visualization is used to explore the seasonal evolution and periodic patterns of the load curve, and the magnitude of similar distances is intuitively represented by the intensity of color.

[0031] According to some embodiments of the present invention, the calculation of the multidimensional distance includes at least two different distance measurement methods to comprehensively characterize the differences in load curves from multiple dimensions of time alignment and morphological similarity.

[0032] A load curve similarity visualization system based on multiple distance metrics and a weekly calendar pattern according to a second aspect of the present invention is characterized in that it includes:

[0033] The data preprocessing module is capable of acquiring daily load curve data for the whole year and normalizing the daily load curve data.

[0034] The multi-dimensional distance measurement module can calculate the multi-dimensional distance between adjacent daily load curves based on normalized daily load curve data, and obtain the similarity distance sequence between adjacent days;

[0035] The annual adjacent day similarity matrix construction module can reconstruct the adjacent day similarity distance sequence into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week.

[0036] The heatmap visualization and similarity evolution module can generate heatmaps based on the cyclical distance matrix to visualize the distribution patterns of load curve similarity.

[0037] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0038] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0039] Figure 1 This is a schematic diagram illustrating the steps of the load curve similarity visualization method based on multiple distance metrics and weekly calendar patterns according to an embodiment of the present invention.

[0040] Figure 2 This invention provides an Euclidean distance (EU) heatmap in a load curve similarity visualization method based on multiple distance metrics and a weekly calendar pattern for embodiments of the present invention.

[0041] Figure 3 This invention provides a dynamic time warp distance (DTW) heatmap in a load curve similarity visualization method based on multiple distance metrics and a weekly calendar pattern for embodiments of the present invention.

[0042] Figure 4 This invention provides a derivative dynamic time warped distance (DDTW) heatmap in a load curve similarity visualization method based on multiple distance metrics and calendar patterns for embodiments of the present invention.

[0043] Figure 5 This invention provides a distance EU_Diff heatmap, weighted by Euclidean distance and differential distance, in a load curve similarity visualization method based on multiple distance metrics and a weekly calendar pattern, for embodiments of the present invention.

[0044] Figure 6 The diagram below shows the structure of a load curve similarity visualization system based on multiple distance metrics and a weekly calendar pattern, as provided in an embodiment of the present invention. Detailed Implementation

[0045] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0046] In the description of this invention, it should be understood that the orientation descriptions, such as up, down, front, back, left, right, etc., are based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting this invention.

[0047] In the description of this invention, "several" means one or more, "more than" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0048] In the description of this invention, unless otherwise explicitly defined, terms such as "set up," "install," and "connect" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0049] Example 1

[0050] Traditional load curve analysis methods mainly focus on tasks such as load clustering or classification, load forecasting, anomaly detection, and user segmentation, lacking visual representation tools for the changing patterns of load curve shapes. Based on existing technologies, this invention provides a load curve similarity visualization method based on multiple distance metrics and a weekly calendar pattern, such as... Figure 1 As shown, it specifically includes:

[0051] Step S100: Obtain the daily load curve data for the whole year and normalize the daily load curve data.

[0052] Obtain the daily load curve data for the entire year, and standardize the 96-point load curve using the min-max normalization method. The formula satisfies:

[0053]

[0054] Where x is the original daily load curve sequence, x min and x max These are the minimum daily load value and the maximum daily load value, respectively.

[0055] Step S200: Based on the normalized daily load curve data, calculate the multi-dimensional distance between adjacent daily load curves to obtain the adjacent daily similarity distance sequence.

[0056] Based on the normalized load curves, multi-dimensional distances between adjacent daily load curves are calculated. The metrics include classic Euclidean distance (EU), Euclidean distance plus first-order difference weighted distance (EU_Diff), dynamic time-warped distance (DTW), and derivative dynamic time-warped distance (DDTW).

[0057] Step S300: Reconstruct the adjacent day similarity distance sequence into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week.

[0058] Based on the similarity distances between adjacent days of the load curves calculated for 364 adjacent days throughout the year, a 52×7 annual weekly distance matrix D is reconstructed. Here, D is a 52-row, 7-column matrix, where the element in the i-th row and j-th column represents the similarity distance of the load curves from week j to week j+1 in the i-th week of the year, and the column dimensions correspond to the time transition pattern from Monday to Sunday.

[0059] Step S400: Generate a heat map based on the weekly distance matrix to visualize the distribution pattern of load curve similarity.

[0060] Heatmaps are generated based on the reconstructed weekly distance matrix, which intuitively display the distribution patterns of load similarity through color gradients, identify key load pattern characteristics, and provide the annual similarity evolution of daily load curves under multiple dimensions.

[0061] Example 2

[0062] To verify the effectiveness of Example 1 and provide a more detailed explanation, this embodiment of the invention uses a residential load scenario as an example, sampling a daily load curve dataset of 96 points with a sampling frequency of 15 minutes, spanning from January 1st to December 31st, 2012. The method is specifically as follows:

[0063] Step A100: Obtain the daily load curve data for the whole year and normalize the daily load curve data.

[0064] Using the aforementioned minimum-maximum normalized daily load curve, we only focus on and measure the morphological similarity of the daily load curve. The normalized values ​​of the 96-point load curves for each day fall between [0,1].

[0065] Step A200: Based on the normalized daily load curve data, calculate the multi-dimensional distance between adjacent daily load curves to obtain the adjacent daily similarity distance sequence.

[0066] Based on the normalized load curves, the multi-dimensional distance between adjacent daily load curves is calculated. The Euclidean distance (EU) is calculated as follows, measuring the absolute difference between curves aligned over time:

[0067]

[0068] Dynamic Time Warped (DTW) distance measures the similarity of load curve shapes and can eliminate the influence of time offset. The DTW distance is calculated as follows:

[0069]

[0070] Where π is a curved path that satisfies monotonicity, continuity, and boundary conditions, and the DTW distance is solved recursively through dynamic programming.

[0071] Derivative Dynamic Time Warping (DDTW) is a variant of Dynamic Time Warping (DTW). Compared to the commonly used EU and DTW distances, DDTW effectively overcomes the pathological alignment problems present in DTW by aligning the morphological context of points in the load sequence. This invention introduces it into the similarity distance metric of load curves. DDTW is calculated as follows:

[0072]

[0073] The distance EU_Diff, formed by Euclidean distance and first-order difference weighting, is a generalization of Euclidean distance. It is also a time-aligned load curve similarity measure. EU_Diff is calculated as follows (where α is the weight coefficient in [0,1]):

[0074]

[0075] Step A300: Reconstruct the adjacent day similarity distance sequence into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week.

[0076] The adjacent day distance sequence of length 364 is reconstructed into a 52×7 weekly distance matrix D. D is a 52-row, 7-column matrix, where the element in the i-th row and j-th column represents the similarity distance of the load curve from week j to week j+1 in the i-th week of the year, and the column dimension corresponds to the time transition pattern from Monday to Sunday.

[0077] Step A400: Generate a heat map based on the weekly distance matrix to visualize the distribution pattern of load curve similarity.

[0078] A heatmap is generated based on the reconstructed weekly distance matrix, visually displaying the distribution patterns of load similarity through color gradients. Based on a household's 365-day load curve, a weekly matrix of similarity values ​​for 364 adjacent daily load curves is calculated. The heatmap visualization of its Euclidean distance, weighted distance (α = 0.5), DTW distance, and DDTW distance is shown below. Figures 2-5 As shown.

[0079] by Figure 2 For example, based on Euclidean distance, it can be found that in terms of load similarity measures with strict time alignment, the load curve of this user throughout the year shows a significant difference in similarity between Monday and the previous Sunday (the larger the value, the brighter the color, indicating a greater difference in load curves between adjacent days). In terms of user load similarity between Thursday and Wednesday, there are significant differences and fluctuations in several weeks, while the similarity of the electricity load curves at the weekend is very high (the smaller the value, the darker the color, indicating a lower difference and higher similarity between adjacent days). This indicates that, based on the similarity measure of absolute time alignment, the user's electricity consumption behavior at the weekend is almost a very stable load pattern throughout the year.

[0080] Depend on Figure 4 Based on the derivative dynamic time warp distance (DDTW), after overcoming the pathological alignment present in DTW, this method disregards absolute time alignment and focuses only on the measurement of load patterns. It was found that the household's annual load curve showed relatively stable load patterns from Monday to Friday, while the load patterns on weekends exhibited more significant differences (mostly brighter colors).

[0081] Other characteristics of load curves, such as seasonal load pattern evolution, periodicity, and anomaly detection, can be intuitively explored and extracted from annual load calendar heatmaps with different distance metrics, which will not be elaborated here. Furthermore, it can also be used as multi-channel image feature input for training deep learning models such as load clustering or classification and load behavior analysis, expanding the channels for mining the value of power data.

[0082] Another embodiment of the present invention provides a load curve similarity visualization system based on multiple distance metrics and weekly calendar patterns, such as... Figure 6 As shown, the system includes:

[0083] The data preprocessing module is capable of acquiring daily load curve data for the whole year and normalizing the daily load curve data.

[0084] The multi-dimensional distance measurement module can calculate the multi-dimensional distance between adjacent daily load curves based on normalized daily load curve data, and obtain the similarity distance sequence between adjacent days;

[0085] The annual adjacent day similarity matrix construction module can reconstruct the adjacent day similarity distance sequence into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week.

[0086] The heatmap visualization and similarity evolution module can generate heatmaps based on the cyclical distance matrix to visualize the distribution patterns of load curve similarity.

[0087] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0088] It will be understood by those skilled in the art that all or some of the steps and systems in the methods disclosed above can be implemented as software, firmware, hardware, and suitable combinations thereof. Some or all of the physical components can be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit. Such software can be distributed on a computer-readable medium, which can include computer storage media (or non-transitory media) and communication media (or transient media). As is known to those skilled in the art, the term computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, as is known to those skilled in the art, communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0089] The above is a detailed description of the preferred embodiments of this application. However, this application is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of this application. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for visualizing the similarity of load curves based on multiple distance metrics and a weekly calendar pattern, characterized in that, Includes the following steps: Obtain the daily load curve data for the whole year, and normalize the daily load curve data; Based on the normalized daily load curve data, the multi-dimensional distance between adjacent daily load curves is calculated to obtain the adjacent daily similarity distance sequence. The adjacent day similarity distance sequence is reconstructed into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week. A heatmap is generated based on the aforementioned cyclical distance matrix to visualize the distribution patterns of load curve similarity.

2. The method according to claim 1, characterized in that, The normalization process uses the min-max normalization method, and the specific formula is as follows: Where x is the original daily load curve sequence, x min and x max These are the minimum daily load value and the maximum daily load value, respectively.

3. The method according to claim 1, characterized in that, The multidimensional distance includes at least one of Euclidean distance, Euclidean distance plus first-order difference weighted distance, dynamic time warping distance, and derivative dynamic time warping distance.

4. The method according to claim 3, characterized in that, The derivative dynamic time warping distance is obtained by calculating the first derivative sequence of the load curve and applying the dynamic time warping algorithm to the first derivative sequence.

5. The method according to claim 1, characterized in that, The weekly distance matrix is ​​a 52-row, 7-column matrix, where the element in the i-th row and j-th column represents the similarity distance of the load curve from week j to week j+1 in the i-th week of the year.

6. The method according to claim 1, characterized in that, The heatmap uses color gradients to display the distribution patterns of load similarity, which is used to identify load pattern characteristics.

7. The method according to claim 6, characterized in that, The load pattern characteristics include at least one of the following: weekday start pattern abrupt change pattern, weekend load pattern stability characteristics, seasonal load pattern evolution, weekday load pattern periodicity, and abnormal load pattern detection.

8. The method according to claim 1, characterized in that, The method also includes inputting the heatmap as a multi-channel image feature into a deep learning model for load clustering, classification, or user behavior analysis.

9. The method according to claim 1, characterized in that, The daily load curve data consists of 96 points of daily load curve data, with a sampling frequency of 15 minutes.

10. The method according to claim 1, characterized in that, When calculating the multidimensional distance between load curves of adjacent days, for a year of 365 days, the similarity distance between 364 adjacent days is calculated.

11. The method according to claim 3, characterized in that, The dynamic time warping distance d DTW The solution is obtained using dynamic programming, and the calculation formula is as follows: Where π is a curved path that satisfies monotonicity, continuity, and boundary conditions, and the DTW distance is solved recursively through dynamic programming.

12. The method according to claim 1, characterized in that, The reconstruction of the weekly distance matrix is ​​based on the similarity distance sequence of adjacent days throughout the year, and the 364 distance values ​​are arranged into a matrix form of 52 weeks and 7 days according to the weekly pattern.

13. The method according to claim 1, characterized in that, The heatmap visualization is used to explore the seasonal evolution and periodic patterns of the load curve, and the magnitude of similar distances is intuitively represented by the intensity of color.

14. The method according to claim 1, characterized in that, The calculation of the multidimensional distance includes at least two different distance measurement methods to comprehensively characterize the differences in load curves from multiple dimensions, including time alignment and morphological similarity.

15. A load curve similarity visualization system based on multiple distance metrics and weekly calendar patterns, characterized in that, include: The data preprocessing module is capable of acquiring daily load curve data for the whole year and normalizing the daily load curve data. The multi-dimensional distance measurement module can calculate the multi-dimensional distance between adjacent daily load curves based on normalized daily load curve data, and obtain the similarity distance sequence between adjacent days; The annual adjacent day similarity matrix construction module can reconstruct the adjacent day similarity distance sequence into a weekly calendar distance matrix, wherein the rows of the weekly calendar distance matrix represent the week number, the columns represent the day of the week, and the matrix elements represent the load curve similarity distances corresponding to the week number and day of the week. The heatmap visualization and similarity evolution module can generate heatmaps based on the cyclical distance matrix to visualize the distribution patterns of load curve similarity.

Images