Method and system for identifying the same source of voltage sag based on association rule mining

Abstract

The present invention relates to the technical field of power quality management, and discloses a method and system for identifying the same source of voltage sag based on association rule mining. The candidate same-source sag events are screened out to form a transaction, and a transaction table is formed. The support size of each item set in the transaction table is used to screen out multiple frequent item sets to construct a standard sequence tree corresponding to the transaction table, thereby reducing the number of misleading transactions and improving the effectiveness of identifying the same source of voltage sag. At the same time, the association rules of the candidate same-source sag events corresponding to each two of the frequent item sets are determined through the standard sequence tree, the voltage waveforms corresponding to the candidate same-source sag events are compared for waveform similarity, and it is determined whether the two candidate same-source sag events are same-source sag events, thereby improving the accuracy of identifying the same source of voltage sag events.

Description

Technical Field

[0001] The present invention relates to the technical field of power quality management, and in particular to a method and system for identifying the same source of voltage sag based on association rule mining. Background Art

[0002] As the number of voltage sag monitoring devices installed increases, voltage sag monitoring data presents a massive feature. However, this does not mean that a corresponding number of voltage sag events have occurred in the power grid. Since voltage sags will propagate in the power grid after they occur, multiple monitoring devices may repeatedly record the sag data of the same voltage sag event within the propagation time slice. This not only causes a large amount of data redundancy, but may also lead to an over-evaluation of the severity of the voltage sag in the regional power grid. Therefore, it is very meaningful to perform homologous detection on voltage sag monitoring data.

[0003] By identifying repeatedly recorded voltage sag events, the source of the voltage sag event can be discovered, which also helps to analyze the scope of the voltage sag impact and improve the efficiency of voltage sag problem management.

[0004] The existing voltage sag homology identification algorithm mainly measures the similarity between multiple voltage sag waveform data through the Wasserstein distance, and performs cluster analysis on the measurement results based on the density-based spatial clustering of applications with noise (DBSAN) clustering algorithm.

[0005] However, the existing voltage sag homologous source identification algorithm has poor recognition effectiveness, which easily leads to low accuracy in homologous source identification of voltage sag events. Summary of the invention

[0006] The present invention provides a method and system for identifying the same source of voltage sag based on association rule mining, which solves the technical problem that the existing same source identification algorithm for voltage sag has poor recognition effectiveness and easily leads to low accuracy of same source identification of voltage sag events.

[0007] In view of this, a first aspect of the present invention provides a method for identifying the same source of voltage sag based on association rule mining, comprising:

[0008] According to a plurality of event groups, transactions corresponding to the event groups are sequentially organized to form a transaction table, wherein the event groups include two different candidate homologous temporary drop events;

[0009] Filter out multiple frequent item sets according to the support of each item set in the transaction table, wherein the item sets are candidate homologous temporary drop events;

[0010] Constructing a standard sequence tree corresponding to the transaction table according to the support of the plurality of frequent item sets, wherein the standard sequence tree includes the plurality of frequent item sets;

[0011] Determine, according to the standard sequence tree, an association rule of candidate homologous temporary drop events corresponding to each two of the frequent item sets, wherein the association rule is used to characterize that the two candidate homologous temporary drop events are in a mapping relationship;

[0012] A waveform similarity comparison is performed based on the voltage waveforms respectively corresponding to the two candidate homologous sag events that are in a mapping relationship, and whether the two candidate homologous sag events are homologous sag events is determined based on the waveform similarity comparison result.

[0013] Preferably, before the step of sequentially organizing transactions corresponding to the event groups according to the plurality of event groups and forming a transaction table, wherein the event group includes two different candidate homologous temporary drop events, the step further includes:

[0014] Sorting a plurality of voltage sag events according to occurrence time, and selecting a plurality of the voltage sag events as voltage sag reference events according to the order;

[0015] For each voltage sag reference event, calculating an occurrence time difference between the voltage sag reference event and each other voltage sag event except the voltage sag reference event;

[0016] The sag reference events and voltage sag events whose occurrence time differences are less than a preset occurrence time difference threshold are screened out as candidate homologous sag events, and the candidate homologous sag events are grouped into an event group.

[0017] Preferably, the step of selecting a plurality of frequent item sets according to the support sizes of the item sets in the transaction table, wherein the item sets are candidate homologous temporary drop events, comprises:

[0018] Taking the candidate homologous temporary drop events in each transaction in the transaction table as item sets, and determining the frequency of each item set appearing in the transaction table as the support of each item set;

[0019] For each item set, the support is compared with a preset support threshold, and the item sets whose support is greater than the preset support threshold are screened out as frequent item sets.

[0020] Preferably, the preset support threshold is determined according to the number of association rules mined under multiple different support thresholds; the step of determining the preset support threshold according to the number of association rules mined under multiple different support thresholds includes:

[0021] Initialize and set the initial value of the support threshold;

[0022] The initial value of the support threshold is used as the current support threshold, and the number of association rules mined under the current support threshold is determined;

[0023] The objective function is constructed by maximizing the product of the current support threshold and the number of association rules mined under the current support threshold as the optimization goal;

[0024] The objective function is iteratively optimized, and the current support threshold is iteratively updated until convergence, and the converged current support threshold is output as the preset support threshold.

[0025] Preferably, the step of constructing a standard sequence tree corresponding to the transaction table according to the support of the plurality of frequent item sets, wherein the standard sequence tree includes the plurality of frequent item sets, comprises:

[0026] Arrange the multiple frequent item sets in descending order according to the support of the multiple frequent item sets;

[0027] A multi-level standard order tree is constructed according to the frequent item sets arranged in descending order. The standard order tree includes multiple paths, each path includes multiple nodes, and the standard order tree arranges each node in descending order into a multi-level structure. Each path is a group of transactions sharing the same prefix node, each node corresponds to a frequent item set, and the prefix node is the frequent item set with the highest support.

[0028] Preferably, the step of determining the association rules of candidate homologous temporary drop events corresponding to each two frequent item sets according to the standard sequence tree comprises:

[0029] For each path in the standard sequence tree, the candidate homologous temporary drop events corresponding to every two frequent item sets in each path are combined to obtain an association rule.

[0030] Preferably, the step of performing waveform similarity comparison on the voltage waveforms respectively corresponding to the two candidate co-origin sag events in a mapping relationship, and judging whether the two candidate co-origin sag events are co-origin sag events according to the waveform similarity comparison result comprises:

[0031] Determine whether the temporary drop types corresponding to the two candidate homologous temporary drop events in a mapping relationship match each other;

[0032] If it is determined that the temporary drop types corresponding to the two candidate homologous temporary drop events in a mapping relationship do not match, then the transactions constituted by the two candidate homologous temporary drop events are filtered out from the transaction table, and the process proceeds to the step of selecting a plurality of frequent item sets according to the support sizes of the item sets in the transaction table, wherein the item sets are candidate homologous temporary drop events;

[0033] If it is determined that the sag types respectively corresponding to the two candidate co-source sag events in a mapping relationship match, the voltage waveforms respectively corresponding to the two candidate co-source sag events in a mapping relationship are transformed to the same side of the transformer;

[0034] The waveform similarity of the two transformed voltage waveforms is calculated based on the constrained differential dynamic time wrapping algorithm, and the waveform similarity is characterized by distance;

[0035] Comparing the waveform similarity of the two transformed voltage waveforms with a preset waveform similarity threshold;

[0036] If the waveform similarity of the two transformed voltage waveforms is less than or equal to the preset waveform similarity threshold, then determining that the two candidate common-origin sag events are common-origin sag events;

[0037] If the waveform similarity of the two transformed voltage waveforms is greater than the preset waveform similarity threshold, it is determined that the two candidate homologous sag events are not homologous sag events, and the transactions constituted by the two candidate homologous sag events are filtered out from the transaction table, and the process proceeds to the step of selecting multiple frequent item sets according to the support size of each item set in the transaction table, where the item sets are candidate homologous sag events.

[0038] Preferably, the step of transforming the voltage waveforms respectively corresponding to the two candidate homologous sag events in a mapping relationship to the same side of the transformer comprises:

[0039] Determine whether two candidate homologous sag events in a mapping relationship occur on the same side of the transformer;

[0040] If it is determined that the two candidate co-source sag events in a mapping relationship do not occur on the same side of the transformer, determining the transformer type involved in the sag propagation process according to the voltage sag types respectively corresponding to the two candidate co-source sag events;

[0041] Determining a voltage waveform transformation matrix according to the transformer type;

[0042] The voltage waveform of one of the candidate homologous sag events is transformed using the voltage waveform transformation matrix, wherein the transformed voltage waveform and the voltage waveform of another candidate homologous sag event are on the same side of the transformer.

[0043] Preferably, the step of calculating waveform similarity of two transformed voltage waveforms based on the constrained differential dynamic time wrapping algorithm, wherein the waveform similarity is characterized by distance, comprises:

[0044] Performing first-order derivatives on the two transformed voltage waveforms respectively to obtain first-order derivatives corresponding to the two voltage waveforms respectively;

[0045] Determine the Euclidean distance between points in the two voltage waveforms according to the first-order derivatives corresponding to the two voltage waveforms;

[0046] Construct a cost matrix based on the Euclidean distance between points in the two voltage waveforms;

[0047] Constructing a cumulative cost matrix within a preset constrained search area according to the cost matrix;

[0048] Determining the global cumulative cost of the two transformed voltage waveforms according to the cumulative cost matrix in the preset constrained search area;

[0049] The waveform similarity of the two voltage waveforms is determined according to the global cumulative cost and the maximum time series length of the two voltage waveforms.

[0050] In a second aspect, the present invention further provides a voltage sag homology identification system based on association rule mining, comprising:

[0051] A transaction construction module, used to sequentially construct transactions corresponding to each of the event groups according to a plurality of event groups, and form a transaction table, wherein the event group includes two different candidate homologous temporary drop events;

[0052] An item set screening module, used to screen out a plurality of frequent item sets according to the support of each item set in the transaction table, wherein the item sets are candidate homologous temporary drop events;

[0053] A tree construction module, configured to construct a standard sequence tree corresponding to the transaction table according to the support of the plurality of frequent item sets, wherein the standard sequence tree includes the plurality of frequent item sets;

[0054] An association mining module, used for determining, according to the standard sequence tree, an association rule of candidate homologous temporary drop events corresponding to each two of the frequent item sets, wherein the association rule is used for characterizing that the two candidate homologous temporary drop events are in a mapping relationship;

[0055] The homologous source identification module is used to compare the waveform similarity according to the voltage waveforms corresponding to the two candidate homologous source temporary sag events that are in a mapping relationship, and determine whether the two candidate homologous source temporary sag events are homologous source temporary sag events according to the waveform similarity comparison result.

[0056] It can be seen from the above technical solutions that the present invention has the following advantages:

[0057] The present invention forms a transaction by screening candidate homologous sag events and forms a transaction table, and screens out multiple frequent item sets by using the support size of each item set in the transaction table to construct a standard sequence tree corresponding to the transaction table, thereby reducing the number of misleading transactions and improving the effectiveness of identifying the homologous source of voltage sags. At the same time, the association rules of candidate homologous sag events corresponding to each two frequent item sets are determined by the standard sequence tree, the voltage waveforms corresponding to the candidate homologous sag events are compared for waveform similarity, and it is determined whether the two candidate homologous sag events are homologous sag events, thereby improving the accuracy of identifying the homologous source of voltage sag events. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A flowchart of a method for identifying the same source of voltage sag based on association rule mining provided by an embodiment of the present invention;

[0059] Figure 2 A flowchart for determining a candidate homologous temporary drop event provided by an embodiment of the present invention;

[0060] Figure 3 A flowchart for filtering multiple frequent item sets according to the support size of each item set in the transaction table provided by an embodiment of the present invention;

[0061] Figure 4 A schematic diagram of the structure of a standard sequence tree provided by an embodiment of the present invention;

[0062] Figure 5 A flowchart for determining whether two candidate homologous temporary drop events are homologous temporary drop events provided by an embodiment of the present invention;

[0063] Figure 6 A flowchart of converting voltage waveforms corresponding to two candidate homologous sag events in a mapping relationship to the same side of a transformer provided in an embodiment of the present invention;

[0064] Figure 7 A flowchart of calculating waveform similarity of two transformed voltage waveforms based on a constrained differential dynamic time wrapping algorithm provided in an embodiment of the present invention;

[0065] Figure 8 A schematic diagram of the structure of a voltage sag homology identification system based on association rule mining provided by an embodiment of the present invention;

[0066] Fig. 9 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0068] An embodiment of the present invention provides a method for identifying the same source of voltage sags based on association rule mining. The embodiment of the present invention can be applicable to the situation of identifying the same source of voltage sags. The method can be executed by a device for identifying the same source of voltage sags. The device for identifying the same source of voltage sags can be implemented in the form of hardware and / or software. The device for identifying the same source of voltage sags can be configured in a computer device.

[0069] like Figure 1 As shown, an embodiment of the present invention provides a method for identifying the same source of voltage sag based on association rule mining, comprising steps S1 to S5:

[0070] Step S1: sequentially organize transactions corresponding to each event group according to multiple event groups, and form a transaction table, wherein the event group includes two different candidate homologous temporary drop events.

[0071] Two different candidate same-source sag events refer to voltage sag events that are selected through preliminary screening and may be caused by the same sag source. The voltage sag events can be monitored by a power quality monitoring device.

[0072] In order to identify candidate homologous transient drop events and use two different candidate homologous transient drop events to form an event group, such as Figure 2 As shown, the embodiment of the present invention formulates a determination rule for candidate homologous temporary drop events, which specifically includes steps A1 to A3:

[0073] Step A1: sort multiple voltage sag events according to occurrence time, and select multiple voltage sag events as voltage sag reference events according to the sorting.

[0074] The sag reference event refers to a selected voltage sag event, and each voltage sag event can be a sag reference event.

[0075] Step A2: for each voltage sag reference event, calculate the occurrence time difference between the voltage sag reference event and each other voltage sag event except the voltage sag reference event.

[0076] Step A3: Screen out the voltage sag reference events and voltage sag events whose occurrence time difference is less than a preset occurrence time difference threshold as candidate same-source voltage sag events, and group the candidate same-source voltage sag events into an event group.

[0077] For example, N voltage sag events obtained from the power quality monitoring device are sorted by occurrence time, and the i-th voltage sag event is selected as the voltage sag benchmark event. Assume that the occurrence time of the voltage sag benchmark event i is t i , the occurrence time of the temporary drop event j is t j , according to the comparison function in formula (1), the candidate homologous sag events of the sag benchmark event i are selected. If the difference between the occurrence time of the sag event j and the occurrence time of the sag benchmark event i is less than the set occurrence time difference threshold, the sag benchmark event i and the sag event j may be caused by the same sag source.

[0078] (1)

[0079] Where δ is the occurrence time difference threshold.

[0080] If the voltage sag benchmark event i and voltage sag event j are selected as candidate sag events of the same source, and the candidate sag events of the same source are grouped into event groups according to formula (2), the corresponding transaction TR k In this way, different temporary drop benchmark events and corresponding candidate homologous temporary drop events are selected to form different transactions in turn.

[0081] (2)

[0082] In the formula, serial number (i) refers to the serial number of the power quality monitoring device that monitors the i-th sag event, sag type (i) refers to the type of voltage sag waveform monitored by the power quality monitoring device of the i-th sag event, serial number (j) refers to the serial number of the power quality monitoring device that monitors the j-th sag event, and sag type (j) refers to the type of voltage sag waveform monitored by the power quality monitoring device of the j-th sag event.

[0083] The power quality monitoring device refers to a device that monitors changes in grid voltage. The types of voltage sag waveforms include but are not limited to symmetrical voltage sag, asymmetrical voltage sag, sudden voltage sag, slow-falling voltage sag, oscillating voltage sag, pulse voltage sag and phase jump voltage sag.

[0084] All transactions generated by combining any basic event with its corresponding candidate homologous drop event can form a transaction table G, as shown in formula (3).

[0085] (3)

[0086] Where M is the number of transactions contained in the transaction table.

[0087] Step S2: Filter out multiple frequent item sets according to the support of each item set in the transaction table, and the item sets are candidate homologous temporary drop events.

[0088] Among them, the item set refers to the candidate homologous temporary drop events in each transaction in the transaction table. For each transaction, it generally contains two different item sets.

[0089] It should be noted that in the transaction table, some transactions are generated by two non-homologous events. In association rule mining, the transactions generated by these non-homologous events in the table and their corresponding association rules should be eliminated.

[0090] For an item set in the transaction table G, denoted as X, the support number of the item set X can be determined by the number of transactions containing the item set X in the transaction table G, denoted as count(X).

[0091] Then, according to the support number of item set X and the number of all transactions in transaction table G, the frequency of item set X appearing in the transaction table is determined as the support of item set X.

[0092] Specifically, if Figure 3 As shown, the process of selecting multiple frequent item sets according to the support size of each item set in the transaction table includes steps S201~S202:

[0093] Step S201: taking candidate homologous temporary drop events in each transaction in the transaction table as item sets, and determining the frequency of occurrence of each item set in the transaction table as the support of each item set.

[0094] Among them, the support is:

[0095] (4)

[0096] In the formula, is the support of item set X, The number of transactions in the transaction table.

[0097] Step S202: for each item set, compare the support with a preset support threshold, and select the item sets with support greater than the preset support threshold as frequent item sets.

[0098] Among them, when the support of item set X is greater than the preset support threshold, the item set X is determined to be a frequent item set, otherwise it is an infrequent item set.

[0099] Association rule mining mainly extracts rules with high credibility from frequent item sets. Therefore, it is very important to screen frequent item sets. If the support threshold is set too high, frequent items will be eliminated during the association rule mining process, which may result in too few matching rules being mined. If the support threshold is set too low, too many rules may be generated, which may conflict with each other and lead to inaccurate results. Therefore, it is necessary to determine an adaptive and optimized support threshold to improve the accuracy of the mined association rules.

[0100] In the embodiment of the present invention, the preset support threshold is determined according to the number of association rules mined under multiple different support thresholds; the step of determining the preset support threshold according to the number of association rules mined under multiple different support thresholds includes steps S21 to S24:

[0101] Step S21, initializing and setting the initial value of the support threshold.

[0102] In a general example, the initial value of the support threshold can be set according to an empirical value.

[0103] Step S22: taking the initial value of the support threshold as the current support threshold and determining the number of association rules mined under the current support threshold.

[0104] The number of association rules is determined by filtering out frequent itemsets according to the current support threshold, and then executing steps S3 and S4 using the frequent itemsets filtered by the current support threshold to obtain candidate association rules of homologous temporary drop events and their number.

[0105] Step S23, constructing an objective function by taking the maximization of the product of the current support threshold and the number of association rules mined under the current support threshold as the optimization goal.

[0106] Among them, the objective function is:

[0107] (5)

[0108] In the formula, is the updated support threshold, is the support threshold before updating, is the number of association rules mined under the support threshold before updating.

[0109] Step S24, iteratively optimize the objective function, and iteratively update the current support threshold until convergence, and output the converged current support threshold as the preset support threshold.

[0110] Among them, a mathematical solver is used to iteratively optimize the objective function, and iteratively update the current support threshold until convergence. The convergence condition can be that the maximum number of convergences is reached or the change difference of the updated support threshold is less than the error value, and the current support threshold after convergence is output as the preset support threshold.

[0111] Step S3: construct a standard sequence tree corresponding to the transaction table according to the support of multiple frequent item sets, where the standard sequence tree contains multiple frequent item sets.

[0112] In some specific embodiments, constructing a standard sequence tree corresponding to a transaction table includes steps S301-S302:

[0113] Step S301: Arrange multiple frequent item sets in descending order according to their support degrees.

[0114] Step S302: construct a multi-level standard order tree according to the frequent itemsets arranged in descending order. The standard order tree includes multiple paths, each path includes multiple nodes, and the standard order tree arranges each node in descending order into a multi-level structure. Each path is a group of transactions sharing the same prefix node, each node corresponds to a frequent itemset, and the prefix node is the frequent itemset with the highest support.

[0115] The canonical-order tree (CanTree) structure includes multiple paths, each path includes multiple nodes, each path represents a set of transactions sharing the same prefix, and each node corresponds to an item set.

[0116] Exemplarily, the transactions shown in Table 1 are used to construct a multi-level standard sequence tree, wherein the frequent item sets contained in the transaction table are item sets a~j.

[0117] Table 1

[0118]

[0119] The support of each frequent item set a~j is sorted in descending order as follows:

[0120] a: 5; b: 3; c: 3; d: 3; e: 2; f: 2; g: 2; h: 1; i: 1; j: 1.

[0121] Then, the frequent item sets in the transactions shown in Table 1 are sorted according to the support of each frequent item set, and the transaction sorting results shown in Table 2 are obtained.

[0122] Table 2

[0123]

[0124] According to the transaction sorting results, the item sets in each transaction are inserted to construct the corresponding multi-level standard sequence tree, such as Figure 4 shown. Figure 4 The structure of the standard sequence tree is shown in Figure 1. Figure 4 There are 5 paths in , the first one contains four nodes, corresponding to item sets a, c, f, h, respectively. Figure 4 It can be seen that all five paths contain item set a, so the item set a is used as the prefix of the standard order tree. All transactions in the transaction table are traversed one by one, and the item set of each transaction is compared with the child node in the standard order tree structure. If the same item set exists in both the transaction and the child node, the item set is merged with the node. Otherwise, the item set of the transaction will be added to the last merged node as a new branch.

[0125] Step S4: determining association rules of candidate homologous temporary drop events corresponding to each two frequent item sets according to the standard sequence tree, wherein the association rules are used to characterize that two candidate homologous temporary drop events are in a mapping relationship.

[0126] Among them, for each path in the standard sequential tree, the candidate homologous temporary drop events corresponding to every two frequent item sets in each path are combined to obtain association rules.

[0127] The association rule is used to characterize two candidate homologous transient drop events as a mapping relationship, as shown in the following formula (6).

[0128] (6)

[0129] In the formula, R k For the mapping relationship, , .

[0130] It can be understood that the waveform change of the waveform feature of a sag event between the two monitoring points i and j can be segmented through the mapping relationship.

[0131] Step S5, performing waveform similarity comparison based on the voltage waveforms corresponding to the two candidate co-origin sag events that are in a mapping relationship, and judging whether the two candidate co-origin sag events are co-origin sag events based on the waveform similarity comparison result.

[0132] Specifically, if Figure 5 As shown, waveform similarity comparison is performed based on the voltage waveforms corresponding to the two candidate co-origin sag events to determine whether the two candidate co-origin sag events are co-origin sag events, including steps S501-S507:

[0133] Step S501: determine whether the temporary drop types corresponding to two candidate homologous temporary drop events in a mapping relationship match each other.

[0134] It can be understood that if two candidate co-source sag events are caused by the same voltage sag source, the sag types of the two candidate co-source sag events are consistent. The embodiment of the present invention determines whether the sag types of the two candidate co-source sag events are consistent by judging whether the sag types corresponding to the two candidate co-source sag events in a mapping relationship match.

[0135] Step S502: If it is determined that the temporary drop types corresponding to the two candidate temporary drop events of the same source in a mapping relationship do not match, the transaction consisting of the two candidate temporary drop events of the same source is filtered out from the transaction table, and the process goes to step S2.

[0136] The transactions consisting of two candidate homologous temporary drop events in the transaction table are filtered out, thereby updating the support of each item set, and then the process goes to step S2 to re-determine the support of each item set.

[0137] Step S503: if it is determined that the sag types respectively corresponding to the two candidate co-source sag events in a mapping relationship match, the voltage waveforms respectively corresponding to the two candidate co-source sag events in a mapping relationship are transformed to the same side of the transformer.

[0138] The transformer includes a primary side and a secondary side. The voltage levels on different sides are not the same. Therefore, the voltage waveforms on the two sides are difficult to be comparable. Therefore, it is necessary to transform the voltage waveforms corresponding to two candidate homologous sag events with a mapping relationship to the same side of the transformer.

[0139] In some embodiments, Figure 6 As shown, the process of transforming the voltage waveforms corresponding to two candidate homologous sag events in a mapping relationship to the same side of the transformer includes steps S5031-S5034:

[0140] Step S5031: determine whether two candidate homologous sag events in a mapping relationship occur on the same side of the transformer.

[0141] Step S5032: if it is determined that the two candidate co-source sag events in a mapping relationship do not occur on the same side of the transformer, the transformer type involved in the sag propagation process is determined according to the voltage sag types respectively corresponding to the two candidate co-source sag events.

[0142] There is a mapping relationship between the candidate homologous sag events and the transformer type, or the transformer type can be determined through the power grid data system. For example, the transformer is divided into Y 0 / Y 0 –12, Y / Y 0 –12. Y 0 / Y-12, Y / Y-12, Y / -11.

[0143] Y 0 / Y 0 –12 means that both windings are connected in star (Y) fashion and have a neutral point (marked as 0), with no phase difference between the two windings (12 o'clock position, i.e. 0 degrees phase difference).

[0144] Y / Y 0 –12 means that the high voltage side adopts star connection, the low voltage side adopts star connection with neutral point lead-out, and there is no phase difference between the two windings.

[0145] Y 0 / Y-12 means that the high voltage side is star-connected with a neutral point lead, the low voltage side is star-connected but without a neutral point lead, and there is no phase difference between the two windings.

[0146] Y / Y-12 means that both sides are star-connected, there is no neutral point lead, and there is no phase difference between the two windings.

[0147] Y / -11 means the high voltage side is connected in star shape and the low voltage side is connected in triangle shape (△), with a phase difference of 30 degrees between the two windings (-11 o'clock position).

[0148] Step S5033: determine the voltage waveform transformation matrix according to the transformer type.

[0149] Among them, the connection form of the transformer winding is Y 0 / Y 0 The voltage sag type of -12 will not change after passing through the first type transformer, that is, the voltage on the high and low voltage sides of the transformer is the same, and its voltage waveform transformation matrix T 1 for:

[0150] (7)

[0151] The connection form of the transformer winding is Y / Y 0 –12. Y 0 / Y-12 or Y / Y-12, which does not allow the zero-sequence component to pass through, and its voltage waveform transformation matrix T 2 for:

[0152] (8)

[0153] For transformer windings with a connection form of Y / -11, the voltage will change phase by 30° after passing through this type of transformer. Its voltage waveform transformation matrix T 3 for:

[0154] (9)

[0155] Step S5034: transform the voltage waveform of one of the candidate co-source sag events using a voltage waveform transformation matrix, wherein the transformed voltage waveform and the voltage waveform of another candidate co-source sag event are on the same side of the transformer.

[0156] The transformation method is as follows:

[0157] (10)

[0158] Where U a , U b , Uc are the a, b, c phase voltages on the primary side of the transformer respectively, , , are the a, b, and c phase voltages on the secondary side of the transformer respectively, and T is the voltage waveform transformation matrix corresponding to the transformer type.

[0159] Step S504: calculating waveform similarity of the two transformed voltage waveforms based on the constrained differential dynamic time wrapping algorithm, wherein the waveform similarity is characterized by distance.

[0160] Considering the possibility of sampling frequency variation between power quality monitoring devices, which may lead to waveforms of different lengths, a Constraint Differential-Dynamic TimeWarping (CD-DTW) algorithm is proposed. Compared with the traditional DTW algorithm, this algorithm can handle singular points, that is, a single point on one waveform corresponds to multiple points on another waveform, solve the problem of waveform pathological alignment, and better perform waveform similarity evaluation.

[0161] Specifically, if Figure 7 As shown, the process of calculating the waveform similarity of two transformed voltage waveforms based on the constrained differential dynamic time wrapping algorithm includes steps S5041 to S5046:

[0162] Step S5041, performing first-order derivatives on the two transformed voltage waveforms respectively to obtain first-order derivatives corresponding to the two voltage waveforms respectively.

[0163] Among them, for the waveform with a recorded time series length of m , and the waveform with time series length n , their first-order derivatives are calculated using the following formula:

[0164] (11)

[0165] (12)

[0166] In the formula, is the first-order derivative of the i-th sampling point in waveform C, is the first-order derivative of the j-th sampling point in waveform D.

[0167] Step S5042: Determine the Euclidean distance between points in the two voltage waveforms according to the first-order derivatives corresponding to the two voltage waveforms.

[0168] Step S5043: construct a cost matrix according to the Euclidean distance between points in the two voltage waveforms.

[0169] Among them, the cost matrix A is:

[0170] (13)

[0171] In the formula, , refers to the Euclidean distance between two points.

[0172] Step S5044: construct a cumulative cost matrix within the preset constrained search area according to the cost matrix.

[0173] Among them, the dynamic programming method is used to calculate the cumulative cost. For each cell B(i,j) in the cumulative cost matrix B, the cumulative cost is determined by the following recursive relationship, which is obtained:

[0174] (14)

[0175] In formula (14), indicators i and j should satisfy the following constraints:

[0176] (15)

[0177] Where λ is the width of the constraint search area, and its value is generally set to half a period, that is, 10ms.

[0178] Step S5045: Determine the global cumulative cost of the two transformed voltage waveforms according to the cumulative cost matrix in the preset constraint search area.

[0179] The global cumulative cost is usually used to describe the cumulative cost from one point to another in the global domain. After calculating the cumulative cost matrix in the constrained search area, the cumulative cost of the cell (m, n) where the last row and the last column intersect can be easily recursively deduced through the time series length m of waveform C and the time series length n of waveform D as the global cumulative cost, recorded as .

[0180] Step S5046: Determine the waveform similarity of the two voltage waveforms according to the global cumulative cost and the maximum time series length of the two voltage waveforms.

[0181] Among them, the waveform similarity is calculated as:

[0182] (16)

[0183] In the formula, is the waveform similarity between waveforms C and D, is the global cumulative cost, It is the maximum time series length in waveforms C and D.

[0184] Step S505: compare the waveform similarity of the two transformed voltage waveforms with a preset waveform similarity threshold.

[0185] The embodiment of the present invention adopts the 3σ-criterion to determine whether waveform C and waveform D are homologous, that is, the determination condition for determining whether waveform C and waveform D are homologous is:

[0186] (17)

[0187] Where µ and σ are the mean and standard deviation of the similarity index of all pairs of waveforms of homologous events, respectively.

[0188] Step S506: if the waveform similarity of the two transformed voltage waveforms is less than or equal to a preset waveform similarity threshold, then the two candidate common-origin sag events are determined to be common-origin sag events.

[0189] Step S507: if the waveform similarity of the two transformed voltage waveforms is greater than the preset waveform similarity threshold, the two candidate common-source sag events are determined not to be common-source sag events, the transaction consisting of the two candidate common-source sag events is filtered out from the transaction table, and the process goes to step S2.

[0190] It should be noted that the embodiments of the present invention take into account the occurrence time of the sag event, the correlation between the sag types and their waveform characteristics, establish relevant standards from multiple angles to verify the accuracy of sag homologous detection, ensure matching accuracy, and use the association rule mining method to analyze the monitoring data. The voltage sag homologous detection method can be applied to complex scenarios where the grid parameters or topology are unknown.

[0191] Among them, if it is determined that the two candidate homologous temporary sag events are not homologous temporary sag events, it is necessary to select other candidate homologous temporary sag events and the benchmark temporary sag event in turn to perform waveform similarity evaluation to determine whether they are caused by the same temporary sag source as the benchmark event.

[0192] Considering that the occurrence of new sag events or the daily operation reconstruction of the distribution network will change the propagation rules of voltage sags, the association rules may change, causing the previously mined rules to become inaccurate. To this end, the transactions consisting of two candidate sag events of the same source are filtered out from the transaction table, and the support of each item set is updated, then the process goes to step S2 to redetermine the support of each item set, and then re-mine the association rules of sag propagation, update the association rule data, and then perform waveform similarity evaluation on the two candidate sag events of the same source to detect the same source of voltage sags.

[0193] It should be noted that the present invention forms transactions by screening candidate homologous sag events and forms a transaction table, and screens out multiple frequent item sets by using the support size of each item set in the transaction table to construct a standard sequence tree corresponding to the transaction table, thereby reducing the number of misleading transactions and improving the effectiveness of identifying the homologous source of voltage sags. At the same time, the association rules of candidate homologous sag events corresponding to each two frequent item sets are determined by the standard sequence tree, and the voltage waveforms corresponding to the candidate homologous sag events are compared for waveform similarity, and it is determined whether the two candidate homologous sag events are homologous sag events, thereby improving the accuracy of identifying the homologous source of voltage sag events.

[0194] Based on the same inventive concept, an embodiment of the present application further provides a voltage sag homologous source identification system based on association rule mining for implementing the above-mentioned voltage sag homologous source identification method based on association rule mining.

[0195] The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above method. Therefore, the specific limitations in one or more embodiments of the voltage sag homologous identification system based on association rule mining provided below can refer to the limitations of the voltage sag homologous identification method based on association rule mining above, and will not be repeated here.

[0196] like Figure 8 As shown, an embodiment of the present invention provides a voltage sag homology identification system based on association rule mining, comprising:

[0197] The transaction construction module 100 is used to sequentially construct transactions corresponding to each event group according to multiple event groups, and form a transaction table, wherein the event group includes two different candidate homologous temporary drop events;

[0198] An item set screening module 200 is used to screen out a plurality of frequent item sets according to the support of each item set in the transaction table, and the item sets are candidate homologous temporary drop events;

[0199] A tree construction module 300 is used to construct a standard order tree corresponding to the transaction table according to the support of multiple frequent item sets, and the standard order tree contains multiple frequent item sets;

[0200] An association mining module 400 is used to determine the association rules of candidate homologous temporary drop events corresponding to each two frequent item sets according to the standard sequence tree, and the association rules are used to characterize that two candidate homologous temporary drop events are in a mapping relationship;

[0201] The homologous source identification module 500 is used to compare the waveform similarity according to the voltage waveforms corresponding to two candidate homologous source temporary sag events in a mapping relationship, and determine whether the two candidate homologous source temporary sag events are homologous source temporary sag events according to the waveform similarity comparison result.

[0202] In some embodiments, the system includes:

[0203] A reference sorting module is used to sort multiple voltage sag events according to the occurrence time, and select multiple voltage sag events as voltage sag reference events according to the sorting;

[0204] A time difference calculation module, used for calculating, for each sag reference event, a time difference between the sag reference event and each other voltage sag event except the sag reference event;

[0205] The candidate homologous screening module is used to screen out the voltage sag benchmark events and voltage sag events whose occurrence time difference is less than a preset occurrence time difference threshold as candidate homologous sag events, and group the candidate homologous sag events into an event group.

[0206] In some embodiments, the item set screening module 200 is specifically used to treat candidate homologous temporary drop events in each transaction in the transaction table as item sets, and determine the frequency of occurrence of each item set in the transaction table as the support of each item set; it is also used to compare the support of each item set with a preset support threshold, and screen out item sets whose support is greater than the preset support threshold as frequent item sets.

[0207] In some embodiments, the preset support threshold is determined according to the number of association rules mined under multiple different support thresholds; determining the preset support threshold according to the number of association rules mined under multiple different support thresholds includes:

[0208] Initialize and set the initial value of the support threshold;

[0209] The initial value of the support threshold is used as the current support threshold, and the number of association rules mined under the current support threshold is determined;

[0210] The objective function is constructed by maximizing the product of the current support threshold and the number of association rules mined under the current support threshold as the optimization goal;

[0211] The objective function is iteratively optimized, and the current support threshold is iteratively updated until convergence, and the current support threshold after convergence is output as the preset support threshold.

[0212] In some embodiments, the tree construction module 300 is specifically used to arrange multiple frequent item sets in descending order according to the support of the multiple frequent item sets; construct a multi-level standard order tree based on the frequent item sets arranged in descending order, the standard order tree includes multiple paths, each path includes multiple nodes, the standard order tree arranges each node in descending order into a multi-level structure, each path is a group of transactions sharing the same prefix node, each node corresponds to a frequent item set, and the prefix node is the frequent item set with the highest support.

[0213] In some embodiments, the association mining module 400 is specifically used to combine candidate homologous temporary drop events corresponding to every two frequent item sets in each path in the standard sequence tree to obtain association rules.

[0214] In some embodiments, the homology identification module 500 is specifically used to determine whether the temporary drop types corresponding to two candidate homology temporary drop events in a mapping relationship match each other;

[0215] If it is determined that the temporary drop types corresponding to the two candidate homologous temporary drop events in a mapping relationship do not match, the transactions consisting of the two candidate homologous temporary drop events are filtered out from the transaction table, and the process proceeds to the step of selecting multiple frequent item sets according to the support size of each item set in the transaction table, where the item sets are candidate homologous temporary drop events;

[0216] If it is determined that the sag types respectively corresponding to the two candidate co-source sag events in a mapping relationship match, the voltage waveforms respectively corresponding to the two candidate co-source sag events in a mapping relationship are transformed to the same side of the transformer;

[0217] The waveform similarity of the two transformed voltage waveforms is calculated based on the constrained differential dynamic time wrapping algorithm, and the waveform similarity is characterized by distance.

[0218] Comparing the waveform similarity of the two transformed voltage waveforms with a preset waveform similarity threshold;

[0219] If the waveform similarity of the two transformed voltage waveforms is less than or equal to a preset waveform similarity threshold, the two candidate common-source sag events are determined to be common-source sag events;

[0220] If the waveform similarity of the two transformed voltage waveforms is greater than the preset waveform similarity threshold, the two candidate common-source sag events are determined not to be common-source sag events, and the transactions consisting of the two candidate common-source sag events are filtered out from the transaction table, and multiple frequent item sets are selected according to the support size of each item set in the transaction table, and the item sets are candidate common-source sag events.

[0221] The voltage waveforms corresponding to two candidate homologous sag events in a mapping relationship are transformed to the same side of the transformer, including:

[0222] Determine whether two candidate homologous sag events in a mapping relationship occur on the same side of the transformer;

[0223] If it is determined that the two candidate co-source sag events in a mapping relationship do not occur on the same side of the transformer, the transformer type involved in the sag propagation process is determined according to the voltage sag types respectively corresponding to the two candidate co-source sag events;

[0224] Determine the voltage waveform transformation matrix according to the transformer type;

[0225] The voltage waveform of one of the candidate co-source sag events is transformed using a voltage waveform transformation matrix, wherein the transformed voltage waveform and the voltage waveform of another candidate co-source sag event are on the same side of the transformer.

[0226] Among them, the waveform similarity of the two transformed voltage waveforms is calculated based on the constrained differential dynamic time wrapping algorithm, and the waveform similarity is characterized by distance, including:

[0227] Performing first-order derivatives on the two transformed voltage waveforms respectively to obtain first-order derivatives corresponding to the two voltage waveforms respectively;

[0228] Determine the Euclidean distance between points in the two voltage waveforms according to the first-order derivatives corresponding to the two voltage waveforms;

[0229] Construct a cost matrix based on the Euclidean distance between points in the two voltage waveforms;

[0230] Construct a cumulative cost matrix within the preset constrained search area based on the cost matrix;

[0231] Determining the global cumulative cost of the two transformed voltage waveforms according to the cumulative cost matrix in the preset constrained search area;

[0232] The waveform similarity of the two voltage waveforms is determined according to the global cumulative cost and the maximum time series length of the two voltage waveforms.

[0233] like Fig. 9 As shown, an embodiment of the present invention further provides an electronic device, the electronic device 10 includes a memory 20 and a processor 30;

[0234] The memory 20 is used to store programs;

[0235] The processor 30 executes the program to implement the steps of a method for identifying the same source of voltage sag based on association rule mining in any of the above embodiments.

[0236] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the steps of a method for identifying the same source of voltage sag based on association rule mining in any of the above embodiments are implemented.

[0237] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, electronic device and computer storage medium can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0238] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or apparatus.

[0239] In several embodiments provided by the present invention, it is understood that each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and a part of a module, a program segment or a code includes one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved.

[0240] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, electronic devices, computer storage media and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0241] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0242] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0243] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for executing all or part of the steps of the various embodiments of the method of the present invention through a computer device (which can be a personal computer, a server, or a network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk, etc. Various media that can store program codes.

[0244] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying the same source of voltage sag based on association rule mining, characterized in that: include: According to a plurality of event groups, transactions corresponding to the event groups are sequentially organized to form a transaction table, wherein the event groups include two different candidate homologous temporary drop events; Filter out multiple frequent item sets according to the support of each item set in the transaction table, wherein the item sets are candidate homologous temporary drop events; Constructing a standard sequence tree corresponding to the transaction table according to the support of the plurality of frequent item sets, wherein the standard sequence tree includes the plurality of frequent item sets; Determine, according to the standard sequence tree, an association rule of candidate homologous temporary drop events corresponding to each two of the frequent item sets, wherein the association rule is used to characterize that the two candidate homologous temporary drop events are in a mapping relationship; Comparing the voltage waveforms corresponding to the two candidate co-origin sag events that are in a mapping relationship, and determining whether the two candidate co-origin sag events are co-origin sag events according to the waveform similarity comparison result, includes: Determine whether the temporary drop types corresponding to the two candidate homologous temporary drop events in a mapping relationship match each other; If it is determined that the temporary drop types corresponding to the two candidate temporary drop events of the same source in a mapping relationship do not match, the transactions consisting of the two candidate temporary drop events of the same source are filtered out from the transaction table, and the process proceeds to the step of selecting a plurality of frequent item sets according to the support sizes of the item sets in the transaction table; If it is determined that the sag types respectively corresponding to the two candidate co-source sag events in a mapping relationship match, the voltage waveforms respectively corresponding to the two candidate co-source sag events in a mapping relationship are transformed to the same side of the transformer; The waveform similarity of the two transformed voltage waveforms is calculated based on the constrained differential dynamic time wrapping algorithm, and the waveform similarity is characterized by distance; Comparing the waveform similarity of the two transformed voltage waveforms with a preset waveform similarity threshold; If the waveform similarity of the two transformed voltage waveforms is less than or equal to the preset waveform similarity threshold, then determining that the two candidate common-origin sag events are common-origin sag events; If the waveform similarity of the two transformed voltage waveforms is greater than the preset waveform similarity threshold, it is determined that the two candidate homologous sag events are not homologous sag events, and the transactions constituted by the two candidate homologous sag events are filtered out from the transaction table, and the process proceeds to the step of screening out multiple frequent item sets according to the support size of each item set in the transaction table.

2. The method for identifying the same source of voltage sag based on association rule mining according to claim 1 is characterized in that: Before the step of sequentially organizing transactions corresponding to the event groups according to the plurality of event groups and forming a transaction table, wherein the event group includes two different candidate homologous temporary drop events, the step further includes: Sorting a plurality of voltage sag events according to occurrence time, and selecting a plurality of the voltage sag events as voltage sag reference events according to the order; For each voltage sag reference event, calculating an occurrence time difference between the voltage sag reference event and each other voltage sag event except the voltage sag reference event; The sag reference events and voltage sag events whose occurrence time differences are less than a preset occurrence time difference threshold are screened out as candidate homologous sag events, and the candidate homologous sag events are grouped into an event group.

3. The method for identifying the same source of voltage sag based on association rule mining according to claim 1 is characterized in that: The step of selecting a plurality of frequent item sets according to the support sizes of the item sets in the transaction table, wherein the item sets are candidate homologous temporary drop events, comprises: Taking the candidate homologous temporary drop events in each transaction in the transaction table as item sets, and determining the frequency of each item set appearing in the transaction table as the support of each item set; For each item set, the support is compared with a preset support threshold, and the item sets whose support is greater than the preset support threshold are screened out as frequent item sets.

4. The method for identifying the same source of voltage sag based on association rule mining according to claim 3 is characterized in that: The preset support threshold is determined according to the number of association rules mined under multiple different support thresholds; the step of determining the preset support threshold according to the number of association rules mined under multiple different support thresholds includes: Initialize and set the initial value of the support threshold; The initial value of the support threshold is used as the current support threshold, and the number of association rules mined under the current support threshold is determined; The objective function is constructed by maximizing the product of the current support threshold and the number of association rules mined under the current support threshold as the optimization goal; The objective function is iteratively optimized, and the current support threshold is iteratively updated until convergence, and the converged current support threshold is output as the preset support threshold.

5. The method for identifying the same source of voltage sag based on association rule mining according to claim 1, characterized in that: The step of constructing a standard sequence tree corresponding to the transaction table according to the support of the plurality of frequent item sets, wherein the standard sequence tree includes the plurality of frequent item sets, comprises: Arrange the multiple frequent item sets in descending order according to the support of the multiple frequent item sets; A multi-level standard order tree is constructed according to the frequent item sets arranged in descending order. The standard order tree includes multiple paths, each path includes multiple nodes, and the standard order tree arranges each node in descending order into a multi-level structure. Each path is a group of transactions sharing the same prefix node, each node corresponds to a frequent item set, and the prefix node is the frequent item set with the highest support.

6. The method for identifying the same source of voltage sag based on association rule mining according to claim 1, characterized in that: The step of determining the association rules of candidate homologous temporary drop events corresponding to each two frequent item sets according to the standard sequence tree comprises: For each path in the standard sequence tree, the candidate homologous temporary drop events corresponding to every two frequent item sets in each path are combined to obtain an association rule.

7. The method for identifying the same source of voltage sag based on association rule mining according to claim 1, characterized in that: The step of transforming the voltage waveforms corresponding to the two candidate homologous sag events in a mapping relationship to the same side of the transformer includes: Determine whether two candidate homologous sag events in a mapping relationship occur on the same side of the transformer; If it is determined that the two candidate co-source sag events in a mapping relationship do not occur on the same side of the transformer, determining the transformer type involved in the sag propagation process according to the voltage sag types respectively corresponding to the two candidate co-source sag events; Determining a voltage waveform transformation matrix according to the transformer type; The voltage waveform of one of the candidate homologous sag events is transformed using the voltage waveform transformation matrix, wherein the transformed voltage waveform and the voltage waveform of another candidate homologous sag event are on the same side of the transformer.

8. The method for identifying the same source of voltage sag based on association rule mining according to claim 1, characterized in that: The step of calculating waveform similarity of two transformed voltage waveforms based on the constrained differential dynamic time wrapping algorithm, wherein the waveform similarity is characterized by distance, comprises: Performing first-order derivatives on the two transformed voltage waveforms respectively to obtain first-order derivatives corresponding to the two voltage waveforms respectively; Determine the Euclidean distance between points in the two voltage waveforms according to the first-order derivatives corresponding to the two voltage waveforms; Construct a cost matrix based on the Euclidean distance between points in the two voltage waveforms; Constructing a cumulative cost matrix within a preset constrained search area according to the cost matrix; Determining the global cumulative cost of the two transformed voltage waveforms according to the cumulative cost matrix in the preset constrained search area; The waveform similarity of the two voltage waveforms is determined according to the global cumulative cost and the maximum time series length of the two voltage waveforms.

9. A voltage sag homology identification system based on association rule mining, characterized in that: include: A transaction construction module, used to sequentially construct transactions corresponding to each of the event groups according to a plurality of event groups, and form a transaction table, wherein the event group includes two different candidate homologous temporary drop events; An item set screening module, used to screen out a plurality of frequent item sets according to the support of each item set in the transaction table, wherein the item sets are candidate homologous temporary drop events; A tree construction module, configured to construct a standard sequence tree corresponding to the transaction table according to the support of the plurality of frequent item sets, wherein the standard sequence tree includes the plurality of frequent item sets; An association mining module, used for determining, according to the standard sequence tree, an association rule of candidate homologous temporary drop events corresponding to each two of the frequent item sets, wherein the association rule is used for characterizing that the two candidate homologous temporary drop events are in a mapping relationship; A homologous identification module is used to compare the waveform similarity according to the voltage waveforms corresponding to the two candidate homologous sag events that are in a mapping relationship, and to determine whether the two candidate homologous sag events are homologous sag events according to the waveform similarity comparison result; specifically, it is used to determine whether the sag types corresponding to the two candidate homologous sag events that are in a mapping relationship match; If it is determined that the temporary drop types corresponding to the two candidate temporary drop events of the same source in the mapping relationship do not match, the transactions consisting of the two candidate temporary drop events of the same source are filtered out from the transaction table, and multiple frequent item sets are selected according to the support size of each item set in the transaction table; If it is determined that the sag types respectively corresponding to the two candidate co-source sag events in a mapping relationship match, the voltage waveforms respectively corresponding to the two candidate co-source sag events in a mapping relationship are transformed to the same side of the transformer; The waveform similarity of the two transformed voltage waveforms is calculated based on the constrained differential dynamic time wrapping algorithm, and the waveform similarity is characterized by distance. Comparing the waveform similarity of the two transformed voltage waveforms with a preset waveform similarity threshold; If the waveform similarity of the two transformed voltage waveforms is less than or equal to a preset waveform similarity threshold, the two candidate common-source sag events are determined to be common-source sag events; If the waveform similarity of the two transformed voltage waveforms is greater than the preset waveform similarity threshold, the two candidate common-source sag events are determined not to be common-source sag events, and the transactions consisting of the two candidate common-source sag events are filtered out from the transaction table, and multiple frequent item sets are selected according to the support size of each item set in the transaction table.

Images