Load identification method and device based on variable sliding window multi-resolution power matching

Abstract

The invention provides a load identification method and device based on variable sliding window multi-resolution power matching, which can be applied to the technical field of load monitoring. The method comprises the following steps: carrying out load event detection on total power data to obtain initial load data; determining a load event type according to the length of the initial load data; under the condition that the load event type is determined to be a long transient event type, determining a sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched; determining C target section starting points according to the section starting points of the initial load data and a preset section starting point number; for the starting point of each target section, according to the length range of the sliding window and the starting point of each target section, performing dynamic data interception on the total electric power data to obtain a candidate load matrix; and based on a multi-resolution waveform matching algorithm, determining a load identification result according to the candidate load matrix and the template power data to be matched.

Description

Technical Field

[0001] The present invention relates to the technical field of load monitoring, and in particular to a load identification method and device based on variable sliding window multi-resolution power matching. Background Art

[0002] Energy conservation is becoming an increasingly important issue. Feeding back detailed power consumption information of different power-consuming devices to users can help users adjust their power consumption plans to reduce unnecessary power consumption. Therefore, it is important to extract the load data of different total power-consuming devices from power consumption. Existing load monitoring methods usually formulate rules or threshold comparisons based on common sense experience to determine whether it is a load event, and then extract load feature samples.

[0003] However, since the transient power waveforms generated by different electrical equipment during the working state transition process are not the same, and the state transition process is easily affected by background noise, and the background noise levels in different load scenarios may be different, therefore, for long transient type load events, the existing methods are difficult to ensure that the start and end times of the same type of load events generated by the same electrical equipment are accurately located, that is, the accuracy of the complete detection of the load characteristic samples of the electrical equipment is low. Summary of the invention

[0004] In view of the above problems, the present invention provides a load identification method and device based on variable sliding window multi-resolution power matching.

[0005] According to a first aspect of the present invention, there is provided a load identification method based on variable sliding window multi-resolution power matching, comprising: performing load event detection on total power data to obtain initial load data; determining the load event type according to the length of the initial load data; in the case where the load event type is determined to be a long transient event type, determining the sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched, wherein the length of the initial load data corresponding to the long transient event type is greater than or equal to a preset length threshold; determining C target segment starting points according to the segment starting point of the initial load data and the number of preset segment starting points, where C is a positive integer; for each target segment starting point among the C target segment starting points, dynamically intercepting the total power data according to the sliding window length range and each target segment starting point to obtain a candidate load matrix; based on a multi-resolution waveform matching algorithm, determining a load identification result according to the candidate load matrix and the template power data to be matched, the load identification result including the target load data and the target electrical equipment corresponding to the target load data.

[0006] Optionally, based on a multi-resolution waveform matching algorithm, according to the candidate load matrix and the template power data to be matched, a load identification result is obtained, including: performing N frequency reduction processing on the candidate load matrix to obtain N frequency-reduced candidate load matrices; performing N frequency reduction processing on the template power data to be matched to obtain N frequency-reduced template power data to be matched; for the nth frequency reduction in the N frequency reductions, according to the template power data to be matched after the nth frequency reduction and at least one frequency-reduced candidate load data in the candidate load matrix after the nth frequency reduction, an nth matching matrix is ​​obtained to obtain N matching matrices, wherein each matching matrix includes at least one matching result, and the matching result represents the distance value between the frequency-reduced template power data to be matched and the frequency-reduced candidate load data, N≥1, 1≤n≤N; determining a target matching result from at least one matching result, and when the target matching result meets a preset distance threshold range, determining the frequency-reduced candidate load data corresponding to the target matching result as the target load data, and determining the electrical equipment corresponding to the template power data to be matched as the target electrical equipment corresponding to the target load data.

[0007] Optionally, determining the target matching result from the at least one matching result includes: determining the minimum matching result among the at least one matching result as the target matching result.

[0008] Optionally, the load identification method based on variable sliding window multi-resolution power matching also includes: determining a first power difference value according to the power values ​​corresponding to the segment start point and the segment end point of the template power data to be matched; determining a second power difference value according to the power values ​​corresponding to the segment start point and the segment end point of the target load data; determining the similarity between the target load data and the template power data to be matched according to the first power difference value and the second power difference; when the similarity is less than a preset similarity threshold, adding the target load data to the initial load data sample library to obtain the target load data sample library; determining the number of samples of the sample load data in the target load data sample library; when the number of samples is greater than the preset number threshold, updating the initial power data template library according to the target load data sample library, wherein the template power data to be matched is stored in the initial power data template library.

[0009] Optionally, the initial power data template library includes template power data to be matched corresponding to each of the I types of electrical equipment, the template power data to be matched corresponding to the i-th electrical equipment includes central template power data, average template power data, and cluster template power data, and the initial load data sample library includes multiple sample load data corresponding to the i-th electrical equipment, i I, i, I are positive integers; wherein the central template power data corresponding to the i-th electrical equipment is determined based on the following operations: using a dynamic time warping distance algorithm to calculate the distance between every two sample load data in a plurality of sample load data to obtain a distance matrix; summing the column elements in the distance matrix to obtain a plurality of column element sums; and determining the sample load data corresponding to the minimum column element sum in the plurality of column element sums as the central template power data.

[0010] Optionally, the average template power data corresponding to the i-th electrical equipment is determined based on the following operations: based on the dynamic time warping distance algorithm, respectively calculating the twisted paths between multiple sample load data and the central template power data to obtain multiple twisted paths; using the average value calculation function to process the multiple twisted paths to obtain the average template power data.

[0011] Optionally, the cluster template power data corresponding to the i-th electrical equipment is determined based on the following operations: determining a relationship judgment threshold based on the mean and standard deviation of the elements in the upper triangular area of ​​the distance matrix, wherein the upper triangular area represents the area composed of the diagonal from the upper left corner to the lower right corner of the distance matrix and all the elements above it; clustering multiple sample load data based on the distance matrix and the relationship judgment threshold to obtain A clusters, where A is a positive integer; for each of the A clusters, obtaining A cluster template power data based on multiple sample load data in each cluster.

[0012] Optionally, the load identification method based on variable sliding window multi-resolution power matching also includes: when the load event type is a step event type, determining the power tolerance and length of the initial load data, wherein the length of the initial load data corresponding to the step event type is less than a preset length threshold; based on the power tolerance and length, matching multiple template power data to be matched and the initial load data, determining the target template power data corresponding to the initial load data, and determining the electrical equipment corresponding to the target template power data as the target electrical equipment corresponding to the initial load data.

[0013] Optionally, load event detection is performed on the total power data to obtain initial load data including:

[0014] Based on the first preset window length, segment data of the total electric power data is intercepted to obtain first segment power data, wherein the total electric power data represents the total data of power consumption of different electrical equipment; according to the power values ​​corresponding to the segment start point and the segment end point of the first segment power data, a third power difference is determined; when the third power difference is greater than or equal to the preset power threshold, segment data of the first segment power data is intercepted based on the second preset window length to obtain second segment power data; according to the power values ​​corresponding to the segment start point and the segment end point of the second segment power data, a fourth power difference is determined; when the fourth power difference is greater than or equal to the preset power threshold and the length of the second segment power data is within the preset length range, the second segment power data is determined as the initial load data.

[0015] The second aspect of the present invention provides a load identification device based on variable sliding window multi-resolution power matching, comprising: a detection module, used to perform load event detection on total power data to obtain initial load data; a judgment module, used to determine the load event type according to the length of the initial load data; a first determination module, used to determine the sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched when the load event type is determined to be a long transient event type, wherein the length of the initial load data corresponding to the long transient event type is greater than or equal to a preset length threshold; a second determination module, used to determine C target segment starting points according to the segment starting point of the initial load data and the number of preset segment starting points; a capture module, used to dynamically capture the total power data according to the sliding window length range and each target segment starting point for each target segment starting point among the C target segment starting points, to obtain a candidate load matrix; an identification module, used to obtain a load identification result according to the candidate load matrix and the template power data to be matched based on a multi-resolution waveform matching algorithm, the load identification result including the target load data and the target electrical equipment corresponding to the target load data.

[0016] According to the load identification method and device based on variable sliding window multi-resolution power matching provided by the present invention, by pre-applying the template power data corresponding to each scene-personalized electrical equipment to the target load data extraction stage, it is possible to adjust the sliding window length range according to the length of the template power data to be matched during each matching process, dynamically intercept multiple candidate load data based on the sliding window length range on the basis of the initial load data, and then match the candidate load data after multiple resolution conversion with the template power data to be matched, select the target load data with high integrity of the template power data to be matched when the matching is successful, make full use of the template information to extract the target load data with high integrity and high consistency, thereby improving the accuracy of load identification and solving the problem of poor integrity and accuracy in extracting load feature samples of long transient event types. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and other objects, features and advantages of the present invention will become more apparent through the following description of the embodiments of the present invention with reference to the accompanying drawings.

[0018] Figure 1 A flow chart of a load identification method based on variable sliding window multi-resolution power matching according to an embodiment of the present invention is shown.

[0019] Figure 2 An example diagram of generating a candidate load matrix according to an embodiment of the present invention is shown.

[0020] Figure 3 A flow chart of a load identification process based on variable sliding window multi-resolution power matching according to an embodiment of the present invention is shown.

[0021] Figure 4 A structural block diagram of a load identification device based on variable sliding window multi-resolution power matching according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0022] Below, embodiments of the present invention will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of embodiments of the present invention. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of concepts of the present invention.

[0023] The terms used herein are only for describing specific embodiments and are not intended to limit the present invention. The terms "comprise", "include", etc. used herein indicate the existence of features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0024] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0025] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0026] In view of this, an embodiment of the present invention provides a load identification method and device based on variable sliding window multi-resolution power matching. The method includes: performing load event detection on the total power data to obtain initial load data; determining the load event type according to the length of the initial load data; in the case where the load event type is determined to be a long transient event type, determining the sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched; determining C target segment starting points according to the segment starting point of the initial load data and the number of preset segment starting points; for each target segment starting point among the C target segment starting points, dynamically intercepting the total power data according to the sliding window length range and each target segment starting point to obtain a candidate load matrix; based on a multi-resolution waveform matching algorithm, obtaining a load identification result according to the candidate load matrix and the template power data to be matched.

[0027] In the technical solution of the present invention, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0028] It should be noted that the sequence numbers of the operations in the following method are only used as representations of the operations for the purpose of description, and should not be regarded as representing the execution order of the operations. Unless explicitly stated, the method does not need to be executed completely in the order shown.

[0029] Figure 1 A flow chart of a load identification method based on variable sliding window multi-resolution power matching according to an embodiment of the present invention is shown.

[0030] like Figure 1 As shown, the method 100 includes operations S110 to S160.

[0031] In operation S110, load event detection is performed on the total power data to obtain initial load data.

[0032] Optionally, the total power data is the total power time series variation data collected directly from the power supply entrance of the power user, and the total power time series includes the time series power variation data consumed by different power-consuming devices. For example, the power-consuming devices may be air conditioners, refrigerators, etc.

[0033] Optionally, the total electric power data is preprocessed using a median filtering algorithm to obtain filtered total active power data, and load event detection is performed on the filtered total active power data to obtain multiple initial load data.

[0034] Optionally, the initial load data is a power data variation section during a process in which the electrical equipment switches from a steady state to another steady state (ie, a transient process), and the power variation range of the initial load data is greater than or equal to a preset power threshold.

[0035] In operation S120, a load event type is determined according to the length of the initial load data.

[0036] Optionally, the length of the initial load data may be calculated using a sequence length calculation function, and the length of the initial load data may be detected to determine the load event type.

[0037] Optionally, the length of the initial load data is the length of the time series.

[0038] In operation S130, when it is determined that the load event type is a long transient event type, a sliding window length range corresponding to the template power data to be matched is determined according to the length of the template power data to be matched.

[0039] Optionally, a sequence length calculation function is used to calculate the length of the initial load data. When the length of the initial load data is greater than or equal to a preset length threshold, the load event type corresponding to the initial load data is determined as a long transient event type. A load event is an event in which an increase, decrease or transfer of load data occurs during a working state transition of an electrical equipment, causing power consumption. The transient process of the load event corresponding to the long transient event type lasts for a long time, which may be several seconds or even minutes.

[0040] Optionally, the initial power data template library stores template power data corresponding to different power-consuming devices. Multiple template power data can be sorted based on the length of the template power data to obtain a matching order, and the current template power data to be matched is determined based on the matching order. For example, the number of template power data belonging to the first power-consuming device is b. If the template power data to be matched is one of the b template power data corresponding to the first power-consuming device, then according to the standard deviation of the length of all template power data belonging to the power-consuming device in the load data sample library, and the length of the template power data to be matched , determine the sliding window length range.

[0041] In operation S140 , C target section starting points are determined according to the section starting point of the initial load data and the number of preset section starting points.

[0042] Optionally, the initial load data is obtained based on a preset window length. The initial load data is a time segment data. The time segment data includes a segment start point and a segment end point. The sliding start point and sliding end point of the window are respectively the segment start point and segment end point of the initial load data. The segment difference between the segment start point and the segment end point is the preset window length.

[0043] Optionally, the starting point of the segment of the initial load data is used as the reference point, and a time neighborhood with a width of r is determined according to the preset number of segment starting points. Samples are taken within the time domain of the reference point to obtain C target segment starting points. The target segment starting point is the sliding starting point of the sliding window, and C is a positive integer.

[0044] In operation S150, for each of the C target section starting points, the total power data is dynamically intercepted according to the sliding window length range and each target section starting point to obtain a candidate load matrix.

[0045] Optionally, for a target segment start point, the total power data is dynamically intercepted according to the sliding window length range to obtain multiple candidate load data, where the segment start points of the multiple candidate load data are the same but the segment end points are different, thereby obtaining a one-dimensional candidate load submatrix corresponding to the target segment start point. For example, the range of the segment end point is .

[0046] Optionally, for each target segment starting point, a one-dimensional candidate load sub-matrix corresponding to each target segment starting point is obtained based on the sliding window length range; then the candidate load sub-matrices corresponding to each target segment starting point are sorted in ascending order according to the position of the target segment starting point to obtain a multi-dimensional candidate load matrix.

[0047] Optionally, the elements in the candidate load matrix represent the intercepted candidate load data, and the number of rows and columns of the candidate load matrix represent the number of segment start points and the number of segment end points of the intercepted candidate load data.

[0048] Figure 2 An example diagram of generating a candidate load matrix according to an embodiment of the present invention is shown.

[0049] like Figure 2 As shown, according to the length of the template power data to be matched , determine the sliding window length range ; Take the starting point of the initial load data as the reference point G0, determine the time neighborhood with a width of r according to the number of preset section starting points, sample in the time neighborhood of the reference point G0, and obtain C target section starting points. For the target section starting point G1, according to the sliding window length range Dynamically intercept the total power data to obtain multiple candidate load data, where the candidate load data Q1 is based on the target section starting point G1 and the sliding window length The data is intercepted; for the starting point G2 of the target segment, according to the sliding window length range Dynamically intercept the total power data to obtain multiple candidate load data, where the candidate load data Q2 is based on the target section starting point G2 and the sliding window length The data is intercepted; for each of the C target section starting points, a candidate load matrix is ​​obtained according to a plurality of candidate load data corresponding to each target section starting point.

[0050] In operation S160, a load identification result is determined based on a multi-resolution waveform matching algorithm according to the candidate load matrix and the template power data to be matched.

[0051] Optionally, the load identification result includes target load data and target electrical equipment corresponding to the target load data.

[0052] Optionally, the candidate load matrix and the template power data to be matched are down-converted to the same frequency to obtain the down-converted candidate load matrix and the template power data to be matched, and then the down-converted template power data to be matched are matched with the candidate load data in the candidate load matrix. If the match is successful, the target load data is determined from multiple candidate load data.

[0053] Optionally, the electric device corresponding to the template power data to be matched is determined as the target electric device.

[0054] Optionally, the target load data represents high-integrity power data corresponding to a state transition process of the target electrical equipment.

[0055] Optionally, in the event of a matching failure, the template power data to be matched in the next matching process is determined according to the matching order, the sliding window length range and the candidate load matrix are re-determined, and the template power data to be matched and the candidate load matrix are matched again until the target load data is matched or all template power data in the initial power data template library fail to match.

[0056] Optionally, since the template power data corresponding to each scene-personalized electrical equipment is pre-applied to the target load data extraction stage, the sliding window length range can be adjusted according to the length of the template power data to be matched during each matching process, and multiple candidate load data are dynamically intercepted based on the sliding window length range on the basis of the initial load data, and then the candidate load data after multiple resolution conversions are matched with the template power data to be matched, the template information is fully utilized to extract target load data with high completeness and consistency, thereby improving the accuracy of target load data identification.

[0057] Optionally, load event detection is performed on the total electric power data to obtain initial load data, including: segment data interception of the total electric power data based on a first preset window length to obtain first segment power data, wherein the total electric power data represents total data of power consumption of different electrical equipment; determining a third power difference value based on power values ​​corresponding to a segment start point and a segment end point of the first segment power data; when the third power difference value is greater than or equal to a preset power threshold, segment data interception of the first segment power data based on a second preset window length to obtain second segment power data; determining a fourth power difference value based on power values ​​corresponding to a segment start point and a segment end point of the second segment power data; when the fourth power difference value is greater than or equal to the preset power threshold and the length of the second segment power data is within a preset length range, determining the second segment power data as the initial load data.

[0058] Optionally, a third power difference value is obtained according to a difference between a power value at an end point of the first segment power data and a power value at a start point of the segment.

[0059] Optionally, if the third power difference is greater than or equal to a preset power threshold, it indicates that the power data of the first section is an event section, and the power data of the first section is retained; if the third power difference is less than the preset power threshold, it indicates that the power data of the first section is a steady-state section, and the power data of the first section is discarded. An event section is a section where a large power variation range occurs, and a steady-state section is a section where a small power variation range occurs.

[0060] In one embodiment, the segment type is determined as shown in formula (1):

[0061] (1); among which, The segment where the power data of the first segment is located is the event segment. The segment where the power data of the first segment is located is a steady-state segment. Characterizes the preset power threshold, a power value at the end point of the segment representing the power data of the first segment, A power value representing the start point of the first segment of power data.

[0062] Optionally, when the third power difference is greater than or equal to a preset power threshold, segment data interception is performed on the first segment power data based on a second preset window length to obtain second segment power data.

[0063] Optionally, a fourth power difference is obtained according to a difference between a power value at an end point of the second segment power data and a power value at a start point of the segment.

[0064] Optionally, when the fourth power difference is greater than or equal to a preset power threshold, the power data of the first segment is retained, and it is continued to be determined whether the event segment is a load event; when the length of the power data of the second segment is within a preset length range, the event segment is determined to be a load event, thereby determining the power data of the second segment as the initial load data.

[0065] Optionally, when the fourth power difference is less than a preset power threshold or the length of the second section power data is not within a preset length range, the second section power data is discarded.

[0066] In one embodiment, the load event determination is as shown in formula (2):

[0067] (2) Among them, The event segment where the power data of the second segment is located is characterized as a load event. Characterize the power data of the second section as a non-load event, Characterizes the preset power threshold, a power value at the end point of the segment representing the power data of the second segment, a power value representing the start point of the second segment power data, Characterizes the length of the second segment power data, the preset length range is .

[0068] Optionally, the initial power data template library includes template power data to be matched corresponding to each of the I types of electrical equipment, the template power data to be matched corresponding to the i-th electrical equipment includes central template power data, average template power data, and cluster template power data, and the initial load data sample library includes multiple sample load data corresponding to the i-th electrical equipment, i I, i, I are positive integers; wherein the central template power data corresponding to the i-th electrical equipment is determined based on the following operations: using a dynamic time warping distance algorithm to calculate the distance between every two sample load data in a plurality of sample load data to obtain a distance matrix; summing the column elements in the distance matrix to obtain a plurality of column element sums; and determining the sample load data corresponding to the minimum column element sum in the plurality of column element sums as the central template power data.

[0069] Optionally, the initial load data sample library includes multiple sample load data corresponding to I types of electrical equipment, and based on the multiple sample load data corresponding to the i-th electrical equipment, the central template power data, average template power data, and cluster template power data corresponding to the i-th electrical equipment in the initial power data template library are generated.

[0070] For example, the initial load data sample library includes S sample load data corresponding to the i-th electrical equipment, and the distance between each two sample load data in the multiple sample load data is calculated using the dynamic time warping distance algorithm (DTW) to obtain an S×S distance matrix. Each element in the distance matrix represents the distance between two sample load data.

[0071] Optionally, for each column element in the distance matrix, the column elements are summed to obtain multiple column element sums, and the sample load data corresponding to the minimum column element sum in the multiple column element sums is determined as the central template power data.

[0072] For example, the initial load data sample library includes 3 sample load data corresponding to the i-th electrical equipment. The distance between every two sample load data is calculated to obtain a distance matrix of 3 rows and 3 columns. The sum of the column elements in the second column is the minimum sum of the column elements, and the second sample load data is determined as the central template power data corresponding to the i-th electrical equipment.

[0073] Optionally, the average template power data corresponding to the i-th electrical equipment is determined based on the following operations: based on the dynamic time warping distance algorithm, respectively calculating the twisted paths between multiple sample load data and the central template power data to obtain multiple twisted paths; using the average value calculation function to process the multiple twisted paths to obtain the average template power data.

[0074] Optionally, a dynamic time warping distance algorithm is used to calculate the distortion path between each sample load data and the central template power data to obtain multiple distortion paths.

[0075] Optionally, the warped path is the optimal path found from the first point of the sample load data to the last point of the center template power data.

[0076] Optionally, the distance matrix between each sample load data and the central template power data is first calculated using the dynamic time warping distance algorithm, and then the minimum cumulative distance from the first point of the sample load data to the last point of the central template power data is calculated using the dynamic programming method to form a cumulative distance matrix. Starting from the last point of the central template power data in the cumulative distance matrix, backtracking to the first point of the sample load data, the path with the minimum cumulative distance is found to obtain a distorted path. The distorted path is used to measure the optimal alignment path found when measuring the similarity between the sample load data and the central template power data.

[0077] In one embodiment, the twist path is as shown in formula (3):

[0078] (3); among which, Characterizing the power data of the center template The point corresponds to the vth point of the sample load data. Characterize the twisted paths between the sample load data x and the central template power data u of the plurality of sample load data.

[0079] In one embodiment, the average value calculation function mean() is as shown in formula (4):

[0080] (4);

[0081] in, represents the average result of the point corresponding to the kth point of the central template power data u, Represents a set of points in a plurality of sample load data that respectively correspond to the k-th point of the central template power data u.

[0082] Optionally, based on multiple warping paths, the average value calculation function is used to average the points in multiple sample load data corresponding to the points of the central template power data to obtain the average template power data. The number of iterations IT is set, and the average template power data is calculated repeatedly each time. By continuously calculating the distance matrix between each sample load data and the average template power data, the average template power data is gradually updated so that it gradually tends to reflect the overall shape of all sample load data, and finally the optimal average template power data after the iteration IT times is obtained.

[0083] Optionally, the cluster template power data corresponding to the i-th electrical equipment is determined based on the following operations: determining a relationship judgment threshold based on the mean and standard deviation of the elements in the upper triangular area of ​​the distance matrix, wherein the upper triangular area represents the area composed of the diagonal from the upper left corner to the lower right corner of the distance matrix and all the elements above it; clustering multiple sample load data based on the distance matrix and the relationship judgment threshold to obtain A clusters, where A is a positive integer; for each of the A clusters, obtaining A cluster template power data based on multiple sample load data in each cluster.

[0084] In one embodiment, the clustering process is as shown in formula (5):

[0085] (5); among which, Representing sample load data in distance matrix D With sample load data The distance value between Characterize the relationship judgment threshold, Represents the average value of the elements in the upper triangular area of ​​the distance matrix, Characterizes the standard deviation of the elements in the upper triangle area of ​​the distance matrix, 0 represents sample load data Between the sample load data They are not located in the same cluster. 1 represents the sample load data Between the sample load data In between are located in a cluster.

[0086] Optionally, based on the size between each element in the distance matrix and the relationship judgment threshold, the similarity relationship between every two sample load data is determined, so that a plurality of sample load data are clustered to obtain A clusters.

[0087] Optionally, for each of the A cluster clusters, the central template power data corresponding to this cluster cluster can be obtained based on multiple sample load data in each cluster cluster, so that the central template power data corresponding to this cluster cluster can be determined as the cluster template power data corresponding to this cluster cluster, thereby obtaining A cluster template power data.

[0088] Optionally, the initial power data template library stores central template power data, average template power data, and A cluster template power data corresponding to the i-th electrical device.

[0089] Optionally, based on multiple sample load data, central template power data, average template power data, and cluster template power data are generated that are highly representative of the original waveform, thereby making full use of the template power information to extract more complete and consistent target load data, helping users to better complete load status identification.

[0090] Optionally, based on a multi-resolution waveform matching algorithm, a load identification result is obtained according to a candidate load matrix and template power data to be matched, including: performing N frequency reduction processing on the candidate load matrix to obtain N frequency-reduced candidate load matrices; performing N frequency reduction processing on the template power data to be matched to obtain N frequency-reduced template power data to be matched; for the nth frequency reduction in the N frequency reductions, according to the template power data to be matched after the nth frequency reduction and at least one frequency-reduced candidate load data in the candidate load matrix after the nth frequency reduction, an nth matching matrix is ​​obtained, and N matching matrices are obtained, wherein each matching matrix includes at least one matching result, and the matching result represents the distance value between the template power data to be matched after the frequency reduction and the candidate load data after the frequency reduction, N≥1, 1≤n≤N; determining a target matching result from at least one matching result, and when the target matching result meets a preset distance threshold range, determining the frequency-reduced candidate load data corresponding to the target matching result as the target load data, and determining the electrical equipment corresponding to the template power data to be matched as the target electrical equipment corresponding to the target load data.

[0091] Optionally, each candidate load data in the candidate load matrix is ​​resampled to generate multiple waveform features at N time resolutions to obtain N down-converted candidate load matrices. For example, the candidate load matrix H is obtained. 1 , H 2 , H 3 , H 4 , H 5 , where H 5 It is the candidate load matrix after the resolution is downgraded 5 times to the original candidate load data.

[0092] Optionally, the template power data to be matched is subjected to the same number of frequency reduction processing to obtain a plurality of frequency-reduced template power data Y to be matched. 1 , Y 2 , Y 3 , Y 4 , Y 5 .

[0093] Optionally, the template power data to be matched after the nth frequency reduction has the same frequency as the candidate load matrix after the nth frequency reduction.

[0094] Optionally, a dynamic time warping distance algorithm is used to calculate the distance between the template power data to be matched after the nth frequency reduction and each candidate load data after the frequency reduction in the candidate load matrix after the nth frequency reduction to obtain the nth matching matrix.

[0095] Optionally, each matching matrix includes at least one matching result, and a target matching result is determined from at least one matching result in the N matching matrices. When the target matching result satisfies a preset distance threshold range, the down-converted candidate load data corresponding to the target matching result is determined as the target load data, and the electrical equipment corresponding to the template power data to be matched is determined as the target electrical equipment corresponding to the target load data.

[0096] Optionally, determining the target matching result from the at least one matching result includes: determining the minimum matching result among the at least one matching result as the target matching result.

[0097] Optionally, the preset distance threshold range may be determined according to the average value and standard deviation of elements in the upper triangular region of the distance matrix.

[0098] In one embodiment, the matching process of the long transient event type is shown in formula (6):

[0099] (6);

[0100] in, Characterize the target matching results, Characterizes the preset distance threshold range, Represents the average value of the elements in the upper triangular area of ​​the distance matrix, Represents the standard deviation of the elements in the upper triangle area of ​​the distance matrix. Sample is True if the match is successful, and Sample is True if the match fails.

[0101] Optionally, the target matching result satisfies a preset distance threshold range, indicating that the down-converted candidate load data corresponding to the target matching result and the template power data to be matched belong to the power conversion process of the same electrical equipment.

[0102] Optionally, the candidate load data after frequency reduction corresponding to the target matching result is determined as the target load data; and the electrical equipment corresponding to the template power data to be matched is determined as the target electrical equipment corresponding to the target load data.

[0103] Optionally, in the event of a matching failure, the template power data to be matched in the next matching process is determined according to the matching order, the sliding window length range and the candidate load matrix are re-determined, and the template power data to be matched and the candidate load matrix are matched again until the target load data is matched or all template power data in the initial power data template library fail to match.

[0104] Optionally, the matching order may be to sort multiple template power data of the same electrical device in descending order according to length, and select the next template power data to be matched starting from the first position. If multiple template power data of the same electrical device all fail to match, multiple template power data of the next electrical device are selected for matching.

[0105] Optionally, the load identification method also includes: determining a first power difference value based on the power values ​​corresponding to the segment start point and the segment end point of the template power data to be matched; determining a second power difference value based on the power values ​​corresponding to the segment start point and the segment end point of the target load data; determining the similarity between the target load data and the template power data to be matched based on the first power difference value and the second power difference; when the similarity is less than a preset similarity threshold, adding the target load data to the initial load data sample library to obtain the target load data sample library; determining the number of samples of the sample load data in the target load data sample library; when the number of samples is greater than the preset number threshold, updating the initial power data template library according to the target load data sample library, wherein the template power data to be matched is stored in the initial power data template library.

[0106] Optionally, when the matching is successful, a second verification process is performed on the template power data and the target load data to be matched to determine the validity of the matching.

[0107] In one embodiment, the similarity As shown in formula (7):

[0108] (7); among which, The power value corresponding to the end point of the segment representing the template power data to be matched, The power value corresponding to the starting point of the segment representing the template power data to be matched, The power value corresponding to the end point of the segment representing the template power data to be matched, The power value corresponding to the starting point of the segment of the template power data to be matched.

[0109] Optionally, if the similarity is less than a preset similarity threshold, it indicates that the match is valid, and the target load data is added to the target load data sample library; otherwise, the match is valid, and the target load data is recorded as an unknown event and stored in the unknown event list. Unknown events refer to load event types that are not included in the initial load data sample library. They may be load events that have not occurred before for known electrical equipment, load events generated by newly added electrical equipment, or aliased load events. Aliased load events refer to transient processes involving multiple different electrical equipment during the unstable period of load power corresponding to the detected load event.

[0110] Optionally, a quantitative statistics is performed on the target load data sample library to determine the number of samples of the sample load data in the target load data sample library; if the number of samples is greater than a preset quantity threshold, indicating that the batch processing scale has been reached, the sample load data in the target load data sample library is updated with various types of template power data corresponding to different electrical equipment in the initial power data template library, thereby obtaining an updated power data template library for the next matching.

[0111] Optionally, since the latest template power data corresponding to each scene-personalized electrical equipment is pre-applied to the target load data extraction stage, the problem of low accuracy in matching the load data extracted based on inherent parameters with the template information is avoided.

[0112] Optionally, the load identification method also includes: when the load event type is a step event type, determining the power tolerance and length of the initial load data, wherein the length of the initial load data corresponding to the step event type is less than a preset length threshold; based on the power tolerance and length, matching multiple template power data to be matched and the initial load data, determining the target template power data corresponding to the initial load data, and determining the electrical equipment corresponding to the target template power data as the target electrical equipment corresponding to the initial load data.

[0113] Optionally, when the length of the initial load data is less than a preset length threshold, the type of the event segment where the initial load data is located is determined to be a step event type.

[0114] Optionally, the preset length threshold can be ,in, is the sampling frequency of the data set, The value is The preset parameters of is the length of the longest template power data in the real scene.

[0115] Optionally, the power tolerance of the initial load data represents a difference between a maximum power value and a minimum power value in the initial load data.

[0116] In one embodiment, the matching process of the step event type is shown in formula (8):

[0117] (8);

[0118] in, Characterize the tolerance factor, Characterizes the power tolerance of the template power data to be matched, Characterizes the length of the template power data to be matched, Characterize the power tolerance of the initial load data, Indicates the length of the initial load data. Sample is True if the match is successful, and Sample is True if the match fails.

[0119] Optionally, template power data is selected in sequence from the initial power data template library to match the initial load data. If the match is successful, the template power data to be matched is determined as the target template power data corresponding to the initial load data, and the electrical equipment corresponding to the target template power data is determined as the target electrical equipment corresponding to the initial load data.

[0120] Optionally, a non-intrusive load monitoring (NILM) framework is introduced, in which load events are divided into step events and long transient events, and different load data extraction and identification strategies are set according to the load event type, thereby overcoming the problems of poor consistency and low identification accuracy of load data extracted based on conventional event detection algorithms.

[0121] Figure 3 A flow chart of a load identification process based on variable sliding window multi-resolution power matching according to an embodiment of the present invention is shown.

[0122] like Figure 3 As shown, the load identification process includes steps S301 to S311.

[0123] In step S301, an initial power data template library is generated. Then, the process goes to step S302.

[0124] In step S302, the total power data is collected and preprocessed. Then, the process goes to step S303.

[0125] In step S303, load event detection is performed, and the process proceeds to step S304.

[0126] In step S304, it is determined whether the load event type is a step event type. If it is a step event type, the process goes to step S305, otherwise, the process goes to step S306.

[0127] In step S305, matching is performed based on the power tolerance to obtain a load identification result. Then, the process proceeds to step S307.

[0128] In step S306, a candidate load matrix is ​​obtained based on the sliding window length range and the starting point of the target section, and then a load identification result is obtained by matching based on a multi-resolution waveform matching algorithm. Then, the process goes to step S307.

[0129] In step S307, it is determined whether the load identification result is successfully verified. If the verification is successful, the process goes to step S309, otherwise, the process goes to step S308.

[0130] In step S308, the target load data is stored in the unknown event list.

[0131] In step S309, the target load data is added to the initial load data sample library. Go to step S310.

[0132] In step S310, it is determined whether the number of sample load data in the target load data sample library is greater than a preset number threshold. If yes, go to step S311, otherwise go to step S302.

[0133] In step S311, the initial power data template library is updated according to the target load data sample library. Go to step S301.

[0134] Optionally, the present invention selects a measured data set with a sampling frequency of 1 Hz, the time span is 7 days, and the data set includes the total power data of different electrical equipment. According to the load identification method of the present invention and the existing load detection method, the data set is parsed and identified, and the recognition accuracy, recall rate and precision are calculated based on the number of correctly identified load data samples or load events, the number of incorrectly identified load data samples or load events, and the number of samples or events that have not been successfully identified among all real load data samples or events.

[0135] Table 1 shows the comparative effects of load identification based on the method of the present invention and the existing method.

[0136] Table 1 Comparison results

[0137]

[0138] Optionally, it can be seen from Table 1 that, compared with the existing methods, the present invention has higher accuracy and precision in identifying the extracted target load data. It also means that the present invention has extracted relatively complete target load data, reflecting a higher load decomposition capability, helping users to better understand the power consumption of each load in the home, so as to achieve the purpose of energy saving.

[0139] Based on the above load identification method, the present invention also provides a load identification device based on variable sliding window multi-resolution power matching. Figure 4 The device is described in detail.

[0140] Figure 4 A structural block diagram of a load identification device based on variable sliding window multi-resolution power matching according to an embodiment of the present invention is shown.

[0141] like Figure 4 As shown, the load identification device 400 of this embodiment includes a detection module 410 , a judgment module 420 , a first determination module 430 , a second determination module 440 , an interception module 450 and an identification module 460 .

[0142] The detection module 410 is used to perform load event detection on the total power data to obtain initial load data. In one embodiment, the detection module 410 can be used to perform the operation S110 described above, which will not be described in detail here.

[0143] The judgment module 420 is used to determine the load event type according to the length of the initial load data. In one embodiment, the judgment module 420 can be used to perform the operation S120 described above, which will not be repeated here.

[0144] The first determination module 430 is used to determine the sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched when the load event type is determined to be a long transient event type, wherein the length of the initial load data corresponding to the long transient event type is greater than or equal to a preset length threshold. In one embodiment, the first determination module 430 can be used to perform the operation S130 described above, which will not be repeated here.

[0145] The second determination module 440 is used to determine C target segment starting points according to the segment starting point of the initial load data and the number of preset segment starting points, where C is a positive integer. In one embodiment, the second determination module 440 can be used to perform the operation S140 described above, which will not be described in detail here.

[0146] The interception module 450 is used to dynamically intercept the total power data for each of the C target section starting points according to the sliding window length range and each target section starting point to obtain a candidate load matrix. In one embodiment, the interception module 450 can be used to perform the operation S150 described above, which will not be repeated here.

[0147] The identification module 460 is used to determine the load identification result based on the multi-resolution waveform matching algorithm according to the candidate load matrix and the template power data to be matched, and the load identification result includes the target load data and the target electrical equipment corresponding to the target load data. In one embodiment, the identification module 460 can be used to perform the operation S160 described above, which will not be repeated here.

[0148] Optionally, the identification module 460 includes a first identification submodule, a second identification submodule, a third identification submodule and a fourth identification submodule.

[0149] The first identification submodule is used to perform N frequency reduction processes on the candidate load matrix to obtain N frequency reduced candidate load matrices.

[0150] The second identification submodule is used to perform N frequency reduction processing on the template power data to be matched, so as to obtain N frequency reduced template power data to be matched.

[0151] The third identification submodule is used to determine the nth matching matrix for the nth frequency reduction among N frequency reductions, based on the template power data to be matched after the nth frequency reduction and at least one candidate load data after the frequency reduction in the candidate load matrix after the nth frequency reduction, so as to obtain N matching matrices, wherein each matching matrix includes at least one matching result, and the matching result represents the distance value between the template power data to be matched after the frequency reduction and the candidate load data after the frequency reduction, N≥1, 1≤n≤N.

[0152] The fourth identification submodule is used to determine the target matching result from at least one matching result. When the target matching result meets the preset distance threshold range, the down-converted candidate load data corresponding to the target matching result is determined as the target load data, and the electrical equipment corresponding to the template power data to be matched is determined as the target electrical equipment corresponding to the target load data.

[0153] Optionally, the fourth identification submodule includes a first identification unit.

[0154] The first recognition unit is used to determine the minimum matching result among the at least one matching result as the target matching result.

[0155] Optionally, the load device 400 further includes a third determination module, a fourth determination module, a fifth determination module, a first updating module, a sixth determination module and a second updating module.

[0156] The third determination module is used to determine the first power difference according to the power values ​​corresponding to the segment start point and the segment end point of the template power data to be matched.

[0157] The fourth determination module is used to determine the second power difference according to the power values ​​corresponding to the section start point and the section end point of the target load data.

[0158] The fifth determination module is used to determine the similarity between the target load data and the template power data to be matched according to the first power difference and the second power difference.

[0159] The first updating module is used to add the target load data to the initial load data sample library to obtain the target load data sample library when the similarity is less than a preset similarity threshold.

[0160] The sixth determination module is used to determine the number of samples of the sample load data in the target load data sample library.

[0161] The second updating module is used to update the initial power data template library according to the target load data sample library when the number of samples is greater than a preset number threshold, wherein the template power data to be matched is stored in the initial power data template library.

[0162] Optionally, the load device 400 further includes a first generating module, a second generating module and a third generating module.

[0163] The first generation module is used to calculate the distance between every two sample load data in a plurality of sample load data by using a dynamic time warping distance algorithm to obtain a distance matrix.

[0164] The second generating module is used to sum the column elements in the distance matrix to obtain multiple column element sums.

[0165] The third generating module is used to determine the minimum column element among the multiple column elements and the corresponding sample load data as the central template power data.

[0166] Optionally, the load device 400 further includes a fourth generating module and a fifth generating module.

[0167] The fourth generating module is used to calculate the distortion paths between the multiple sample load data and the central template power data respectively based on the dynamic time warping distance algorithm to obtain multiple distortion paths.

[0168] The fifth generation module is used to process the multiple twisted paths using an average value calculation function to obtain average template power data.

[0169] Optionally, the load device 400 further includes a sixth generation module, a seventh generation module and an eighth generation module.

[0170] The sixth generation module is used to determine the relationship judgment threshold according to the average value and standard deviation of the elements in the upper triangular area of ​​the distance matrix, wherein the upper triangular area represents the area composed of the diagonal line from the upper left corner to the lower right corner of the distance matrix and all the elements above it.

[0171] The seventh generation module is used to perform clustering processing on multiple sample load data based on the distance matrix and the relationship judgment threshold to obtain A clusters, where A is a positive integer.

[0172] The eighth generating module is used to obtain A cluster template power data for each cluster in the A clusters according to a plurality of sample load data in each cluster.

[0173] Optionally, the load device 400 further includes a seventh determination module and a matching module.

[0174] The seventh determination module is used to determine the power tolerance and length of the initial load data when the load event type is a step event type, wherein the length of the initial load data corresponding to the step event type is less than a preset length threshold.

[0175] The matching module is used to match multiple template power data to be matched and initial load data based on power tolerance and length, determine the target template power data corresponding to the initial load data, and determine the electrical equipment corresponding to the target template power data as the target electrical equipment corresponding to the initial load data.

[0176] Optionally, the detection module 410 includes a first detection submodule, a second detection submodule, a third detection submodule, a fourth detection submodule and a fifth detection submodule.

[0177] The first detection submodule is used to perform segment data interception on the total power data based on a first preset window length to obtain first segment power data, wherein the total power data represents the total data of power consumption of different electrical devices.

[0178] The second detection submodule is used to determine a third power difference value according to power values ​​corresponding to a segment start point and a segment end point of the first segment power data.

[0179] The third detection submodule is used to perform segment data interception on the first segment power data based on the second preset window length to obtain second segment power data when the third power difference is greater than or equal to the preset power threshold.

[0180] The fourth detection submodule is used to determine a fourth power difference according to the power values ​​corresponding to the segment start point and the segment end point of the second segment power data.

[0181] The fifth detection submodule is used to determine the second section power data as initial load data when the fourth power difference is greater than or equal to a preset power threshold and the length of the second section power data is within a preset length range.

[0182] Optionally, any multiple modules among the detection module 410, the judgment module 420, the first determination module 430, the second determination module 440, the interception module 450 and the identification module 460 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. Optionally, at least one of the detection module 410, the judgment module 420, the first determination module 430, the second determination module 440, the interception module 450 and the identification module 460 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in an appropriate combination of any of them. Alternatively, at least one of the detection module 410, the judgment module 420, the first determination module 430, the second determination module 440, the interception module 450 and the identification module 460 can be at least partially implemented as a computer program module, and when the computer program module is executed, the corresponding function can be performed.

[0183] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present invention. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways, even if such a combination or combination is not explicitly recorded in the present invention. In particular, without departing from the spirit and teachings of the present invention, the features recorded in the various embodiments of the present invention can be combined and / or combined in various ways. All these combinations and / or combinations fall within the scope of the present invention.

[0184] The embodiments of the present invention are described above. However, these embodiments are only for the purpose of illustration, and are not intended to limit the scope of the present invention. Although each embodiment is described above, it does not mean that the measures in each embodiment cannot be used in combination advantageously. Without departing from the scope of the present invention, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present invention.

Claims

1. A load identification method based on variable sliding window multi-resolution power matching, characterized in that: The method comprises: Perform load event detection on the total power data to obtain initial load data; determining a load event type according to the length of the initial load data; In the case where it is determined that the load event type is a long transient event type, determining a sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched, wherein the length of the initial load data corresponding to the long transient event type is greater than or equal to a preset length threshold; Determine C target section starting points according to the section starting point of the initial load data and the number of preset section starting points, where C is a positive integer; For each of the C target section starting points, dynamically intercept the total power data according to the sliding window length range and each of the target section starting points to obtain a candidate load matrix; Based on a multi-resolution waveform matching algorithm, a load identification result is determined according to the candidate load matrix and the template power data to be matched, and the load identification result includes target load data and target electrical equipment corresponding to the target load data.

2. The method according to claim 1, characterized in that The method of determining the load identification result based on the multi-resolution waveform matching algorithm according to the candidate load matrix and the template power data to be matched includes: Performing N frequency reduction processing on the candidate load matrix to obtain N frequency reduced candidate load matrices; Performing N frequency reduction processing on the template power data to be matched to obtain N frequency reduced template power data to be matched; For the nth frequency reduction in N frequency reductions, determine the nth matching matrix according to the template power data to be matched after the nth frequency reduction and at least one candidate load data after the frequency reduction in the candidate load matrix after the nth frequency reduction, so as to obtain N matching matrices, wherein each of the matching matrices includes at least one matching result, and the matching result represents the distance value between the template power data to be matched after the frequency reduction and the candidate load data after the frequency reduction, N≥1, 1≤n≤N; A target matching result is determined from at least one of the matching results. When the target matching result satisfies a preset distance threshold range, the down-converted candidate load data corresponding to the target matching result is determined as the target load data, and the electrical equipment corresponding to the template power data to be matched is determined as the target electrical equipment corresponding to the target load data.

3. The method according to claim 2, characterized in that Determining a target matching result from at least one of the matching results includes: The minimum matching result among at least one of the matching results is determined as the target matching result.

4. The method according to claim 1, characterized in that: The method further comprises: Determine a first power difference value according to power values ​​corresponding to a segment start point and a segment end point of the template power data to be matched; Determine a second power difference value according to power values ​​corresponding to a section start point and a section end point of the target load data; Determining the similarity between the target load data and the template power data to be matched according to the first power difference and the second power difference; When the similarity is less than a preset similarity threshold, the target load data is added to an initial load data sample library to obtain a target load data sample library; Determining the number of samples of the sample load data in the target load data sample library; When the number of samples is greater than a preset number threshold, the initial power data template library is updated according to the target load data sample library, wherein the template power data to be matched is stored in the initial power data template library.

5. The method according to claim 4, characterized in that The initial power data template library includes template power data to be matched corresponding to each of the I types of electrical equipment, the template power data to be matched corresponding to the i-th electrical equipment includes central template power data, average template power data, and cluster template power data, and the initial load data sample library includes multiple sample load data corresponding to the i-th electrical equipment, i I, i, I are positive integers; The central template power data corresponding to the i-th electrical equipment is determined based on the following operations: Calculate the distance between every two of the sample load data in the plurality of sample load data using a dynamic time warping distance algorithm to obtain a distance matrix; Summing the column elements in the distance matrix to obtain a plurality of column element sums; The minimum column element and the corresponding sample load data among the plurality of the column elements and are determined as the central template power data.

6. The method according to claim 5, characterized in that The average template power data corresponding to the i-th electrical device is determined based on the following operations: Based on a dynamic time warping distance algorithm, respectively calculating the distortion paths between the plurality of sample load data and the central template power data to obtain a plurality of distortion paths; The plurality of twisted paths are processed using an average value calculation function to obtain the average template power data.

7. The method according to claim 5, characterized in that The cluster template power data corresponding to the i-th electrical device is determined based on the following operations: Determine the relationship judgment threshold according to the average value and standard deviation of the elements in the upper triangular area of ​​the distance matrix, wherein the upper triangular area represents the area consisting of the diagonal line from the upper left corner to the lower right corner of the distance matrix and all the elements above it; Based on the distance matrix and the relationship judgment threshold, clustering is performed on the plurality of sample load data to obtain A clusters, where A is a positive integer; For each of the A clusters, A cluster template power data are obtained according to a plurality of the sample load data in each cluster.

8. The method according to claim 1, characterized in that The method further comprises: In the case where the load event type is a step event type, determining a power tolerance and a length of the initial load data, wherein the length of the initial load data corresponding to the step event type is less than the preset length threshold; Based on the power tolerance and the length, the multiple template power data to be matched are matched with the initial load data, the target template power data corresponding to the initial load data is determined, and the electrical equipment corresponding to the target template power data is determined as the target electrical equipment corresponding to the initial load data.

9. The method according to claim 1, characterized in that: Perform load event detection on the total power data to obtain the initial load data including: Based on a first preset window length, segment data interception is performed on the total power data to obtain first segment power data, wherein the total power data represents total data of power consumption of different electrical devices; Determine a third power difference value according to power values ​​corresponding to a segment start point and a segment end point of the first segment power data; When the third power difference is greater than or equal to a preset power threshold, segment data interception is performed on the first segment power data based on a second preset window length to obtain second segment power data; Determine a fourth power difference value according to power values ​​corresponding to a segment start point and a segment end point of the second segment power data; When the fourth power difference is greater than or equal to the preset power threshold and the length of the second section power data is within a preset length range, the second section power data is determined as initial load data.

10. A load identification device based on variable sliding window multi-resolution power matching, characterized in that: The device comprises: A detection module is used to perform load event detection on the total power data to obtain initial load data; A judgment module, used for determining the load event type according to the length of the initial load data; A first determination module is used to determine, when determining that the load event type is a long transient event type, a sliding window length range corresponding to the template power data to be matched according to the length of the template power data to be matched, wherein the length of the initial load data corresponding to the long transient event type is greater than or equal to a preset length threshold; A second determination module is used to determine C target section starting points according to the section starting point of the initial load data and the number of preset section starting points, where C is a positive integer; An interception module is used for dynamically intercepting the total power data for each of the C target section starting points according to the sliding window length range and each of the target section starting points to obtain a candidate load matrix; The identification module is used to determine the load identification result based on the multi-resolution waveform matching algorithm according to the candidate load matrix and the template power data to be matched, and the load identification result includes the target load data and the target electrical equipment corresponding to the target load data.

Images