A method for identifying the type of power quality disturbance event association

Abstract

The application provides a power quality disturbance event correlation type identification method, and relates to the technical field of power system analysis, and comprises the following steps: first, obtaining feature information of any two power quality disturbance events at a same target monitoring point; then, processing the feature information of the any two power quality disturbance events to obtain correlation feature information between the any two power quality disturbance events; finally, inputting the correlation feature information into a correlation type classification identification model for identification to obtain a correlation type identification result between the any two power quality disturbance events. The correlation feature information obtained through processing and a multi-classifier based on a least square support vector machine (LS-SVM) can effectively identify the correlation type between the any two power quality disturbance events, and improve the identification accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system analysis, and in particular to a power quality disturbance event correlation type identification method. BACKGROUND

[0002] In the existing technology, a complex power quality disturbance mode concept is proposed in the comprehensive diagnosis of complex power quality disturbance event space-time correlation, a method for mining possible complex disturbance modes in the power grid is established, and basic definitions of basic event correlation types are provided according to the disturbance formation mechanism. However, the existing technology does not study the identification of basic event correlation types. Since the basic event correlation type plays an important role in understanding the development and evolution process of complex power quality disturbance events, tracing the source of disturbance events, responsibility division, complex disturbance mode mining and disturbance governance, how to effectively identify the basic event correlation type is a problem to be solved. SUMMARY

[0003] The purpose of the present application is to provide a power quality disturbance event correlation type identification method to alleviate the technical problem that the existing technology cannot effectively identify the power quality disturbance event correlation type.

[0004] In a first aspect, the present application provides a power quality disturbance event correlation type identification method, which comprises: obtaining characteristic information of any two power quality disturbance events at the same target monitoring point; processing the characteristic information of the any two power quality disturbance events to obtain correlation characteristic information between the any two power quality disturbance events; inputting the correlation characteristic information into a correlation type classification identification model for identification to obtain a correlation type identification result between the any two power quality disturbance events; wherein the correlation type classification identification model is a multi-classifier based on least squares support vector machine (LS-SVM).

[0005] Further, the method further comprises: self-defining the correlation type between the power quality disturbance events; wherein the correlation type comprises at least one of the following: conduction type, chain type, concurrent type and development type; characterizing the correlation type to obtain correlation characteristic information; and constructing a matter element model for representing the correlation type according to the correlation characteristic information.

[0006] Further, the correlation type classification identification model comprises: a first correlation type classifier for identifying the chain type, a second correlation type classifier for identifying the concurrent type and a third correlation type classifier for identifying the development type; wherein the types of the first correlation type classifier, the second correlation type classifier and the third correlation type classifier are all LS-SVM classifiers.

[0007] Further, the association feature information is input into an association type classification recognition model for recognition to obtain an association type recognition result between the two arbitrary power quality disturbance events, including: inputting the association feature information into the first association type classifier for classification to obtain a first classification result; inputting the association feature information into the second association type classifier for classification to obtain a second classification result; inputting the association feature information into the third association type classifier for classification to obtain a third classification result; and determining the association type recognition result between the two arbitrary power quality disturbance events according to the first classification result, the second classification result and the third classification result.

[0008] Further, the feature information of the power quality disturbance event includes time features and type features; the feature information of the two arbitrary power quality disturbance events is processed to obtain association feature information between the two arbitrary power quality disturbance events, including: performing difference processing on the time features of the two arbitrary power quality disturbance events to obtain time association features between the two arbitrary power quality disturbance events; performing comparison processing on the type features of the two arbitrary power quality disturbance events to obtain type association features between the two arbitrary power quality disturbance events; and determining the time association features and the type association features as the association feature information between the two arbitrary power quality disturbance events.

[0009] Further, the method further includes: obtaining training association feature samples; inputting the training association feature samples into an initial LS-SVM classifier to obtain training association type recognition results corresponding to the training association feature samples; calculating a function value of a target function of the initial LS-SVM classifier based on the training association type recognition results; wherein the target function is a function constructed based on a Lagrange multiplier; adjusting parameters of the initial LS-SVM classifier through the function value of the target function to obtain the first association type classifier, the second association type classifier or the third association type classifier.

[0010] Further, the feature information of the two arbitrary power quality disturbance events at the same target monitoring point is obtained by using a Mallat algorithm to extract time features and type features of the two arbitrary power quality disturbance events at the same target monitoring point.

[0011] In a second aspect, the present application provides a device for identifying the correlation type of power quality disturbance events, comprising: a first acquisition unit configured to acquire feature information of any two power quality disturbance events at a same target monitoring point; a processing unit configured to process the feature information of the any two power quality disturbance events to obtain correlation feature information between the any two power quality disturbance events; and an identification unit configured to input the correlation feature information into a correlation type classification identification model to identify the correlation type between the any two power quality disturbance events, and obtain an identification result of the correlation type between the any two power quality disturbance events; wherein the correlation type classification identification model is a multi-classifier based on least squares support vector machine (LS-SVM).

[0012] In a third aspect, the present application provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor implements the steps of the method for identifying the correlation type of power quality disturbance events when executing the computer program.

[0013] In a fourth aspect, the present application provides a computer readable medium having non-volatile program codes executable by a processor, wherein the program codes enable the processor to execute the method for identifying the correlation type of power quality disturbance events.

[0014] The method for identifying the correlation type of power quality disturbance events provided by the present application comprises the following steps: first, acquiring feature information of any two power quality disturbance events at a same target monitoring point; then, processing the feature information of the any two power quality disturbance events to obtain correlation feature information between the any two power quality disturbance events; finally, inputting the correlation feature information into a correlation type classification identification model to identify the correlation type between the any two power quality disturbance events, and obtain an identification result of the correlation type between the any two power quality disturbance events; wherein the correlation type classification identification model is a multi-classifier based on least squares support vector machine (LS-SVM). The correlation feature information obtained by processing can provide a data basis for identifying the correlation type, and based on this, the multi-classifier based on least squares support vector machine (LS-SVM) can effectively identify the correlation type between the any two power quality disturbance events. Since the multi-classifiers can check the identification results of each other, the accuracy of identification is improved.

[0015] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art upon examination of the following or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and attained by the structure particularly pointed out in the description and the appended drawings.

[0016] In order to make the above objectives, features and advantages of the present application more apparent, the following will describe a preferred embodiment in detail, and the accompanying drawings will be referred to, as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 A flowchart of a method for identifying associated types of power quality disturbance events provided by an embodiment of the present invention;

[0019] Figure 2 A schematic diagram of the relationship between the four types of associations;

[0020] Figure 3 A schematic diagram of the structure of the association type classification identification model;

[0021] Figure 4 A schematic structural diagram of a device for identifying associated types of power quality disturbance events provided by an embodiment of the present invention.

[0022] icon:

[0023] 11-first acquisition unit; 12-processing unit; 13-identification unit. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] With the rapid development of power companies, power quality issues are becoming increasingly complex. The power system is a dynamic response process. When certain stimuli occur, such as motor starting, capacitor bank switching, nonlinear load integration, renewable energy generation system integration, and various faults, corresponding responses are generated within the power grid. These responses typically manifest as basic power quality disturbances such as voltage sags, voltage interruptions, transient oscillations, and transient pulses, as well as complex disturbances derived from these basic disturbances. Due to the electrical connectivity characteristics of the power grid, disturbances of varying severity are detected at different monitoring points. Furthermore, disturbance-induced grid events can also lead to the occurrence of new disturbances. These disturbances, arising in time and space based on the topology of the power grid, are not completely independent of each other. From the nature of their occurrence, they may have direct or indirect causal relationships. Describing the correlations between these disturbances is crucial for understanding the evolution of complex power quality disturbances, tracing their root causes, assigning responsibility, identifying complex disturbance patterns, and addressing them. The concept of correlation types has emerged as a response to this need.

[0026] At present, there has been a basic definition of the correlation types of power quality disturbance events, and useful exploration has been carried out in the research on power quality disturbance event correlation analysis. However, these theoretical foundations rely heavily on the definition and identification of correlation types, and the existing basic event correlation types only involve subsequent applications, not the identification process. Therefore, how to further improve the definition of correlation types and realize the classification and identification of correlation types is an urgent problem to be solved.

[0027] In addition, currently mature disturbance detection and identification technologies provide technical support for related research. The detection and identification of power quality disturbance events is the basis for the classification and identification of association types. In essence, the classification and identification process of power quality disturbance event association types is the process of mining the association characteristics between basic power quality disturbance events in time and space. In power quality disturbance detection, commonly used methods include Fourier transform, wavelet transform, S transform, Hilbert-Huang transform, variational mode decomposition (VMD), etc. The main disturbance classification and identification methods include fuzzy logic, decision tree, support vector machine, artificial neural network, etc. These disturbance detection and classification and identification algorithms create the basic conditions for the classification and identification of power quality disturbance event association types.

[0028] Based on this, the purpose of the present invention is to provide a method for identifying the correlation type of power quality disturbance events, which can effectively identify the correlation type between any two power quality disturbance events and improve the accuracy of identification.

[0029] For the convenience of understanding the present embodiment, firstly, a power quality disturbance event correlation type identification method disclosed by the present embodiment is described in detail.

[0030] Embodiment 1

[0031] According to the present embodiment, an embodiment of a power quality disturbance event correlation type identification method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described herein can be executed in a different order.

[0032] Figure 1 A flowchart of a power quality disturbance event correlation type identification method provided by the present embodiment is shown in FIG. 1, which includes the following steps: Figure 1

[0033] In step S101, the characteristic information of any two power quality disturbance events at the same target monitoring point is obtained. The present embodiment does not specifically limit the location of the target monitoring point, which can be determined according to actual needs. Since the power quality disturbance events are obtained at the same target monitoring point, the characteristic information of the above-mentioned power quality disturbance events includes time characteristics and type characteristics, wherein the time characteristics include the start time and the end time. When the power quality disturbance events at different target monitoring points are obtained, the above-mentioned characteristic information can also include the location characteristics. Since the present embodiment mainly analyzes transient power quality disturbance events, the type characteristics mainly include five types of voltage sag, voltage swell, voltage interruption, transient oscillation and transient pulse.

[0034] When there are multiple power quality disturbance events at the same target monitoring point, the present embodiment can first obtain the characteristic information of all power quality disturbance events at the same target monitoring point, and then determine two power quality disturbance events to be correlated from all power quality disturbance events. That is, the correlation type between the power quality disturbance events is identified two by two. For example: when there are N power quality disturbance events, first sort the N power quality disturbance events according to the start time, and after sorting, in order to determine the correlation type identification result between power quality disturbance event 1 and power quality disturbance event 2, the characteristic information of power quality disturbance event 1 and the characteristic information of power quality disturbance event 2 can be obtained. Similarly, if the correlation type identification result between power quality disturbance event 2 and power quality disturbance event 3 is to be determined, the characteristic information of power quality disturbance event 2 and the characteristic information of power quality disturbance event 3 can be obtained.

[0035] ​The application extends to: in the case that multiple power quality disturbance events exist on multiple monitoring points, the multiple power quality disturbance events also have a correlation relationship, for example: if three monitoring points all have a voltage sag event caused by the propagation of the disturbance, then the correlation type between the two disturbance events in the three disturbance events is all conduction type. The premise of the execution of the application is the detection and identification of power quality disturbance events. However, event detection and identification is a relatively mature research field in the prior art, so the application can realize the detection and identification of power quality disturbance events by using the prior art.

[0036] In step S102, the feature information of any two power quality disturbance events is processed to obtain the correlation feature information between any two power quality disturbance events. Since the feature information includes time characteristics and type characteristics, step S102 can process the time characteristics and the type characteristics respectively when performing the processing operation, and finally combine the two processing results to determine the correlation feature information between any two power quality disturbance events.

[0037] In step S103, the correlation feature information is input to the correlation type classification and identification model for identification to obtain the correlation type identification result between any two power quality disturbance events. The correlation type classification and identification model is a multi-classifier based on least squares support vector machine (LS-SVM), which can be referred to as LS-SVM multi-classifier. The correlation type identification result can be understood as the identification result of the type of the correlation relationship.

[0038] The power quality disturbance event correlation type identification method provided by the embodiment of the application can provide a data basis for the identification of the correlation type by processing the obtained correlation feature information. On this basis, the embodiment of the application can realize effective identification of the correlation type between any two power quality disturbance events by the multi-classifier based on LS-SVM. Since the multi-classifiers can perform mutual inspection of the identification results, the accuracy of the identification is improved.

[0039] In an optional embodiment, since the feature information of the power quality disturbance event includes time characteristics and type characteristics, in step S102, the feature information of any two power quality disturbance events is processed to obtain the correlation feature information between any two power quality disturbance events, including the following steps S201-S203:

[0040] Step S201, difference processing is performed on the time characteristics of any two power quality disturbance events to obtain time correlation characteristics between any two power quality disturbance events; step S202, comparison processing is performed on the type characteristics of any two power quality disturbance events to obtain type correlation characteristics between any two power quality disturbance events; and step S203, the time correlation characteristics and the type correlation characteristics are determined as the correlation characteristic information between any two power quality disturbance events.

[0041] Since the time characteristics include the start time and the end time, when step S201 is performed, the start time and the end time of any two power quality disturbance events can be respectively subjected to difference processing, and two time difference results are obtained by respectively taking absolute values, and the two time difference results are determined as the above-mentioned time correlation characteristics. Since the type characteristics include five types of voltage sag, voltage swell, voltage interruption transient oscillation, and transient pulse, the type correlation characteristics in step S202 can represent the same type or different types. The above-mentioned correlation characteristic information can further include waveform form characteristics, which are generally divided into three cases of covering type, overlapping type, and connecting type. The waveform form characteristics can be reflected in time, and therefore the above-mentioned time correlation characteristics further include a third time difference result obtained by taking an absolute value of a difference between the start time of the power quality disturbance event 1 and the end time of the power quality disturbance event 2, which can also be referred to as waveform form characteristics. The correlation characteristic information in step S203 is used to determine the corresponding correlation type. Through the above-mentioned steps S201 to S203, an effective data basis can be provided for the identification of the correlation type. The embodiment of the present application is extended as follows: when analyzing two power quality disturbance events at different monitoring points, the above-mentioned correlation characteristic information can further include location correlation characteristics.

[0042] In an optional embodiment, before step S101 is implemented, the method further includes steps S104 to S106, wherein:

[0043] Step S104, the correlation type between power quality disturbance events is self-defined. The correlation type includes at least one of the following: conduction type, chain type, concurrent type, and development type.

[0044] The above-mentioned four correlation types are defined as follows:

[0045] (1)Conductive, after a power quality disturbance event occurs in a branch, due to the electrical propagation characteristics of the power grid, the same type (i.e. transmission line propagation) or different type (i.e. transformer propagation) of power quality disturbance events of different degrees can be detected at the same time in other branches of the power grid. For example: when a single-phase short-circuit fault occurs in a branch, the monitoring point on the branch and the multiple monitoring points on the upstream branch of the branch detect the same type of basic disturbance event of different degrees at the same time, and the multiple monitoring points on the downstream branch of the branch detect the same type of basic disturbance event of different degrees at the same time. The association type between these basic disturbance events is conductive.

[0046] (2) Cascading, if a basic disturbance event occurs, another disturbance event occurs after the recovery stage of the event, and the complex disturbance waveform occurs in succession, and there is a clear boundary time between the basic disturbance events. The types of the two basic disturbance events (i.e. the abbreviation of the type characteristics described above) can be the same or different. For example: a three-phase short-circuit fault occurs in a branch, causing a basic disturbance event of the voltage sag type, and then the relay protection device acts. After the fault is removed, the reclosing successfully closes to restore power supply. If the downstream branch has a motor, the motor startup will cause another basic disturbance event of the voltage sag type. The association type between the two basic disturbance events of the voltage sag type occurring in succession in this process is cascading.

[0047] (3) Development, if a basic disturbance event occurs, another basic disturbance event occurs due to the continuous change of the parameters of the basic disturbance event, and there is no clear boundary time between the basic disturbance events, and there is no recovery stage in the middle of the event. The types of the basic disturbance events can be the same or different. For example: a lightning strike on a transmission line produces a basic disturbance event of the transient pulse type, and the basic disturbance event of the transient pulse type can be excited at the natural frequency point of the power grid to produce a basic disturbance event of the transient oscillation type. At this time, the association type between the basic disturbance event of the transient pulse type and the basic disturbance event of the transient oscillation type is development.

[0048] (4) Concurrent, different types of basic disturbance events detected by the same disturbance source at the same time and at the same monitoring point are mixed together. The types of the basic disturbance events must be different. For example: during the process of connecting a power capacitor to the grid, a basic disturbance event of the transient oscillation type and a basic disturbance event of the voltage sag type occur at the same time. At this time, the association type between the basic disturbance event of the transient oscillation type and the basic disturbance event of the voltage sag type is concurrent.

[0049] For example: Figure 2As shown, the schematic diagram of the relationship between the four association types is not intended to convey the idea that the individual association types can be converted into each other, but is intended to illustrate the differences between them. Taking the conduction type and the chain type as examples, the position and time between the two association types can mean that the two association types can be distinguished by the position association feature and the time association feature, wherein: the occurrence time (i.e., the time feature) of the plurality of basic disturbance events (i.e., the power quality disturbance event) with the conduction type is the same, and for the chain type, due to the change of the operating state of the power equipment, the occurrence time of the plurality of basic disturbance events of the chain type is not the same. Since the waveform morphology feature can be mapped in time, the waveform morphology feature of the two association types is different. Figure 2 The time between the chain type and the development type in the middle can also be used to reflect the different waveform morphology features of the two types. In addition, the power quality disturbance event with the conduction type association relationship must be at different monitoring points, and the chain type is analyzed at the same monitoring point.

[0050] In step S105, the association type is characterized to obtain the association feature information.

[0051] From the above definitions of the association types: conduction type, chain type, concurrent type, and development type, each association type has the association feature information as shown in Table 1:

[0052] Table 1 Association feature information of each association type

[0053]

[0054] From the above Table 1, (1) the conduction type has the following corresponding association feature information, wherein the position association feature is that the occurrence position (i.e., the position feature) is definitely different, the time association feature is that the start time and the end time of the two basic disturbance events are the same, the type association feature is that the type can be the same or different, and the waveform morphology feature is that there is no waveform morphology feature; (2) the chain type has the following corresponding association feature information, wherein the position association feature is that the occurrence position is the same, the time association feature is that the start time and the end time of the two basic disturbance events are different, the type association feature is that the type can be the same or different, and the waveform morphology feature is that the connection type; (3) the concurrent type has the following corresponding association feature information, wherein the position association feature is that the occurrence position is the same, the time association feature is that the start time and the end time of the two basic disturbance events are basically the same, the type association feature is that the type is different, and the waveform morphology feature is that the coverage type; (4) the development type has the following corresponding association feature information, wherein the position association feature is that the occurrence position is the same, the time association feature is that the start time and the end time of the two basic disturbance events are different, the type association feature is that the type is different, and the waveform morphology feature is that the overlap type.

[0055] Step S106, constructing a matter-element model for representing the association type according to the association feature information.

[0056] Although the association feature information can include four types of association features, i.e., position association feature, time association feature, type association feature and waveform form feature, the type association features of the conduction type and the linkage type both have uncertain factors, and therefore the embodiment of the present application can construct the matter-element model of each association type according to the position association feature, the time association feature and the waveform form feature except the type association feature. It should be noted that the embodiment of the present application does not limit the number of types of association features on which the construction of the matter-element model is based.

[0057] Taking the construction of the matter-element model of each association type according to the position association feature, the time association feature and the waveform form feature as an example, the following is described:

[0058] In an ideal case, a multi-dimensional form matter-element model is established for the association type, as shown in the following formula (1):

[0059]

[0060] wherein, g i represents the ith association type, when i=1, 2, 3, 4, g1, g2, g3, g4 represent the conduction type, the linkage type, the concurrent type and the development type respectively. The to p , t end , p in the matrix respectively represent the characteristic variables of the start time (i.e., the start time described above), the end time (i.e., the end time described above) and the occurrence position between two basic disturbance events with the ith association type. The waveform feature w is used to describe the waveform form feature of the complex disturbance event constituted by the two basic disturbance events, which is generally one of the following three cases: the covering type, the overlapping type and the connecting type. The last column of the matrix is used to describe the values of the above four characteristics. For the first three characteristics, the same characteristic is represented by 1, different characteristics are represented by -1, and not necessarily is represented by 0. If the characteristic does not exist, it is represented by null. For the last characteristic, if the characteristic does not exist, it is represented by null, the connecting type is represented by 1, the covering type is represented by -1, and the overlapping type is represented by 0. The following describes the characteristic analysis and quantification of the matter-element model of each association type:

[0061] (1) Conduction type, for two basic disturbance events with conduction type of associated type, since the disturbance propagation number is very fast, similar to the speed of light, so the occurrence time is basically equal, therefore the starting time of two basic disturbance events is the same, the occurrence position is different, and there is no waveform form feature. The disturbance type (and the above type feature) mainly depends on the propagation path of the basic disturbance event, if it propagates along the conductor, the disturbance type is the same, if it passes through the transformer, it may cause the disturbance type to change. Combined with formula (1), the matter element model of conduction type can be obtained as shown in formula (2):

[0062]

[0063] (2) Chain type, for two basic disturbance events with chain type of associated type, the event start and end time (including: starting time and ending time) is obviously different, the occurrence position is the same, the disturbance type may be the same or different, and the waveform form feature is connection type. Combined with formula (1), the matter element model of chain type can be obtained as shown in formula (3):

[0064]

[0065] (3) Concurrent type, for two basic disturbance events with concurrent type of associated type, the start and end time is basically the same, the disturbance type is different, the occurrence position is the same, and the waveforms are mixed together, so the waveform form feature is coverage type. Combined with formula (1), the matter element model of concurrent type can be obtained as shown in formula (4):

[0066]

[0067] (4) Development type, for two basic disturbance events with development type of associated type, the event start and end time is obviously different, the occurrence position is the same, the disturbance type is different, and the waveform form feature is overlap type. Combined with formula (1), the matter element model of development type can be obtained as shown in formula (5).

[0068]

[0069] Since the disturbance has propagation, whether it is a simple disturbance event (i.e. basic disturbance event) or a complex disturbance event (i.e. a set containing multiple basic disturbance events) will contain a conduction type relationship. In addition, the complex power quality disturbance event type identification and association analysis method research is the association type identification of a series of basic disturbance events monitored at the same monitoring point, that is, the number N of basic power quality disturbance events at the same monitoring point in the complex power quality disturbance event is known, and then the association relationship type between the N basic power quality disturbance events is identified. Therefore, this application only contains the classification and identification of three association types of chain type, concurrent type and development type.

[0070] In an optional embodiment, since the target of identification is the association type between any two power quality disturbance events on the same target monitoring point, the conductive type out of the same target monitoring point is not within the identification range. Therefore, the association type classification identification model applied in step S103 only needs to identify 3 association types, as shown in Figure 3 The association type classification identification model includes: a first association type classifier for identifying the cascading type, a second association type classifier for identifying the concurrent type, and a third association type classifier for identifying the developing type; wherein the types of the first association type classifier, the second association type classifier and the third association type classifier are all LS-SVM classifiers.

[0071] In an optional embodiment, since the association type classification identification model applied in step S103 includes: the first association type classifier, the second association type classifier and the third association type classifier, the above step S103 inputs the association feature information into the association type classification identification model for identification to obtain the association type identification result between any two power quality disturbance events, including the following steps S301-S304:

[0072] Step S301: input the association feature information into the first association type classifier for classification to obtain a first classification result; step S302: input the association feature information into the second association type classifier for classification to obtain a second classification result; step S303: input the association feature information into the third association type classifier for classification to obtain a third classification result; step S304: determine the association type identification result between any two power quality disturbance events according to the first classification result, the second classification result and the third classification result.

[0073] For example: if the first classification result is 1, the second classification result is 0, and the third classification result is 0, then the association type identification result is the cascading type according to Figure 3 If the first classification result is 0, the second classification result is 1, and the third classification result is 0, then the association type identification result is the concurrent type according to Figure 3 If the first classification result is 0, the second classification result is 0, and the third classification result is 1, then the association type identification result is the developing type according to Figure 3It can be known that the association type identification result is the development type; if two or more than two 1s appear in the first classification result, the second classification result and the third classification result, it indicates that the identification is wrong, and if all the first classification result, the second classification result and the third classification result are 0, it also indicates that the identification is wrong. As can be seen, the classification results of the three association type classifiers are also mutual inspection, so as to ensure the correctness of the association type identification result. In addition, in the case of identification error, the embodiment of the application can retrain each association type classifier of the association type classification identification model, so as to ensure the effectiveness of the identification.

[0074] In an optional embodiment, before step S101, the method further comprises steps S107-S110, wherein:

[0075] Step S107, obtaining a training association feature sample; step S108, inputting the training association feature sample into the initial LS-SVM classifier to obtain a training association type identification result corresponding to the training association feature sample; step S109, calculating a function value of a target function of the initial LS-SVM classifier based on the training association type identification result; wherein the target function is a function constructed based on a Lagrange multiplier; step S110, adjusting the parameters of the initial LS-SVM classifier through the function value of the target function to obtain the first association type classifier, the second association type classifier or the third association type classifier. The training association feature sample includes an input sample and an output sample, and step S107 realizes the analysis of the association feature information in the input sample, and steps S108-S110 realize the design of each association type classifier.

[0076] For step S107, the calculation of F1-F4 can be realized. First, define the indexes t ij in the time association feature, where i can take 1, 2, …, N, representing the i th basic disturbance event, N represents the total number of basic disturbance events, j can only take 1 or 2, when j = 1, t i1 represents the starting time of the i th basic disturbance event, and when j = 2, t i2 represents the ending time of the i th basic disturbance event. In combination with the waveform feature and the type association feature, it can be known that the input sample includes three time association features and one type feature, which are represented by F1, F2, F3 and F4 respectively, as shown in Table 2:

[0077] Table 2 Quantization of association feature information

[0078]

[0079] In Table 2, if F4 is 1, it means that the type characteristics of the i-1th basic disturbance event and the i-th basic disturbance event are the same. If F4 is -1, it means that the type characteristics of the two basic disturbance events are different. If F4 is 1, it means that the type characteristics of the two basic disturbance events may be the same.

[0080] The embodiment of the present invention introduces the initial LS-SVM classifier as follows: Since LS-SVM introduces the least squares method and the square loss function in terms of the calculation of the square of the error and the selection of the hyperplane, it can realize the transformation of the inequality constraint to the equality constraint in the objective function, so that it can solve the original quadratic programming problem by using a linear solution method. The initial objective function when using LS-SVM for classification optimization processing is formula (6), where:

[0081]

[0082] Where x i is the input sample in the i-th training associated feature sample, where x i ∈R m , is the m-dimensional input row vector, since x i Contains four features: F1, F2, F3, and F4, so m = 4. i is the output sample in the i-th training associated feature sample, y i is the output value of - or 1, which is used to represent the sample label, is the weight vector, b is the bias term, C is the penalty factor, ξ is the relaxation factor, which represents the degree of penalty for misclassified samples, Ф(x i ) is the implicit mapping function, * represents the optimal, and the initial objective function here is to find the optimal weight vector ω* and the optimal bias term b* when the right side of the equation reaches the minimum value under the constraints.

[0083] On the basis of the initial objective function, the Lagrange multiplier is introduced to construct the corresponding Lagrange function, as shown in the following formula (7):

[0084]

[0085] By taking the partial derivative of Equation (7) to eliminate ω and ξ, the optimal classification function (i.e., the objective function of the initial LS-SVM classifier) ​​is obtained:

[0086]

[0087] Where k(x,x i ) is the kernel function, α i is the Lagrange multiplier.

[0088] The extracted features F1, F2, F3, and F4 are input into the initial LS-SVM classifier as sample features in the form of a vector to perform association type classification. During training, the process of training the second association type classifier or the third association type classifier is consistent with the process of training the first association type classifier, and both adopt the above-mentioned steps S107 to S110, so the training process is not described in detail here. The classification categories in the embodiment of the present invention are relatively small, and a one-to-many method can be directly used. The main idea is to classify samples of a certain category into one category, and all other categories into another category. In this way, three LS-SVM classifiers (i.e., association type classifiers) are constructed for samples of three categories. The number of classifiers is small, so the classification speed is fast.

[0089] In an optional embodiment, step S101, obtaining characteristic information of any two power quality disturbance events at the same target monitoring point, includes: using the Mallat algorithm to extract time characteristics and type characteristics of any two power quality disturbance events at the same target monitoring point.

[0090] The extraction of the start and end times and types of basic disturbance events can be achieved by using the Mallat algorithm, and its operation and inverse operation formulas are shown in Equations (9) and (10):

[0091]

[0092]

[0093] In formula (9), f(t) is the original signal of the basic disturbance event, t is the time series, A l (k) is the discrete approximate component of the basic disturbance event, where l is the number of custom decomposition layers, A l (k) is a sequence, not a point, so k is the number of all points, that is, k is the number of collections, d l (k) is the discrete detail component of the basic disturbance event, φ l，k (t) is the scale parameter, which is a known quantity, ψ l,k (t) is the wavelet function, which is also a known quantity. Equation (10) is the reconstruction of the basic disturbance event, where C l-1,k is the reconstructed signal of the basic disturbance event after reconstruction. The type characteristics of the reconstructed signal are consistent with the original signal. If the original signal is a voltage sag signal, the reconstructed signal is also a voltage sag signal. h0 and h1 are two filter coefficients, k-2n represents 2x downsampling, and Z is an integer set.

[0094] The start and end time t of the basic power quality disturbance event in Table 2 i1 , t i2The extraction of the disturbance type can be obtained using a typical power quality disturbance event detection and identification algorithm, which will not be described in detail here. Formulas (9) and (10) are not direct mathematical expressions for extracting time features and type features in the embodiment of the present invention, but rather mathematical expressions for wavelet decomposition and reconstruction. Considering that the feature extraction and classification of basic disturbance events are not the research focus of the embodiment of the present invention, the implementation plan of the embodiment of the present invention is based on the premise that the time features and type features are known.

[0095] In summary, on the one hand, the embodiments of the present invention propose a basic concept for describing the association relationship between basic power quality disturbance events, namely association type, in view of the possible association relationship between basic power quality disturbance events that constitute complex power quality disturbance events. The possible association types of basic power quality disturbance events in time, space, and causal relationships are divided into conductive type, concurrent type, chain type, and development type. By studying the evolution mechanism of complex power quality disturbance events, a matter-element model of four association types under ideal conditions is established, and the corresponding association feature information is analyzed. In other words, the embodiments of the present invention improve the existing definition of association types between basic power quality disturbance events, characterize the association types to extract the classification features of the association types (i.e., the above-mentioned association feature information), and provide a data basis for the classification and identification of the association types of power quality disturbance events. On the other hand, the embodiments of the present invention design a multi-classifier based on the least squares support vector machine (LS-SVM) based on the classification features, and use the association feature information constructed by the matter-element model of the association type to realize the classification and identification of the three association types of development type, concurrent type, and chain type.

[0096] The embodiment of the present invention is aimed at the correlation analysis and identification of complex power quality disturbance events. It can establish a corresponding matter-element model according to the proposed correlation types of basic power quality disturbance events, analyze the correlation feature information of each correlation type, and construct a three-classifier based on the LS-SVM classifier that can accurately classify and identify chain type, concurrent type and development type correlation types, ultimately realizing the classification and identification of the correlation types of basic disturbance events.

[0097] Example 2:

[0098] An embodiment of the present invention provides a device for identifying the association type of power quality disturbance events. The device for identifying the association type of power quality disturbance events is mainly used to execute the method for identifying the association type of power quality disturbance events provided in the above content of Example 1. The following is a detailed introduction to the device for identifying the association type of power quality disturbance events provided in an embodiment of the present invention.

[0099] Figure 4 This is a schematic diagram of a device for identifying the type of power quality disturbance events provided by an embodiment of the present invention. Figure 4As shown, the power quality disturbance event correlation type identification device mainly includes: a first acquisition unit 11, a processing unit 12 and an identification unit 13, wherein:

[0100] The first acquisition unit 11 is used to acquire characteristic information of any two power quality disturbance events at the same target monitoring point;

[0101] The processing unit 12 is configured to process characteristic information of any two power quality disturbance events to obtain correlation characteristic information between the any two power quality disturbance events;

[0102] The identification unit 13 is used to input the associated feature information into the associated type classification identification model for identification, and obtain the associated type identification result between any two power quality disturbance events; wherein the associated type classification identification model is a multi-classifier based on the least squares support vector machine LS-SVM.

[0103] The power quality disturbance event correlation type identification device provided by the embodiment of the present invention can provide a data basis for the identification of correlation types by processing the correlation feature information obtained. On this basis, the embodiment of the present invention can realize the effective identification of the correlation type between any two power quality disturbance events through a multi-classifier based on the least squares support vector machine LS-SVM. Since the identification results can be mutually checked between the multiple classifiers, the accuracy of the identification is improved.

[0104] Optionally, the device further comprises a custom setting unit, a feature characterization unit and a construction unit, wherein:

[0105] A custom setting unit, used for custom setting the association type between power quality disturbance events; wherein the association type includes at least one of the following: conductive type, chain type, concurrent type, and development type;

[0106] A feature characterization unit is used to characterize the association type and obtain association feature information;

[0107] The construction unit is used to construct a matter-element model for characterizing the association type according to the association feature information.

[0108] Optionally, the association type classification recognition model includes: a first association type classifier for identifying the chain type, a second association type classifier for identifying the concurrent type, and a third association type classifier for identifying the developmental type; wherein, the first association type classifier, the second association type classifier, and the third association type classifier are all LS-SVM classifiers.

[0109] Optionally, the identification unit 13 includes a first classification module, a second classification module, a third classification module and a first determination module, wherein:

[0110] The first classification module is configured to input the association feature information into a first association type classifier to perform classification, and obtain a first classification result.

[0111] The second classification module is configured to input the association feature information into a second association type classifier to perform classification, and obtain a second classification result.

[0112] The third classification module is configured to input the association feature information into a third association type classifier to perform classification, and obtain a third classification result.

[0113] The first determination module is configured to determine an association type recognition result between any two power quality disturbance events according to the first classification result, the second classification result and the third classification result.

[0114] Optionally, the feature information of the power quality disturbance event includes time feature and type feature; the processing unit 12 includes a difference processing module, a comparison processing module and a second determination module, wherein:

[0115] The difference processing module is configured to perform difference processing on the time feature of any two power quality disturbance events, and obtain time association feature between the any two power quality disturbance events.

[0116] The comparison processing module is configured to perform comparison processing on the type feature of any two power quality disturbance events, and obtain type association feature between the any two power quality disturbance events.

[0117] The second determination module is configured to determine the time association feature and the type association feature as the association feature information between the any two power quality disturbance events.

[0118] Optionally, the device further includes a second acquisition unit, an input unit, a calculation unit and an adjustment unit, wherein:

[0119] The second acquisition unit is configured to acquire a training association feature sample.

[0120] The input unit is configured to input the training association feature sample into an initial LS-SVM classifier to obtain a training association type recognition result corresponding to the training association feature sample.

[0121] The calculation unit is configured to calculate a function value of a target function of the initial LS-SVM classifier based on the training association type recognition result; wherein the target function is a function constructed based on a Lagrange multiplier.

[0122] The adjustment unit is configured to adjust parameters of the initial LS-SVM classifier through the function value of the target function, and obtain the first association type classifier, the second association type classifier or the third association type classifier.

[0123] Optionally, the first acquisition unit is further configured to extract the time feature and the type feature of any two power quality disturbance events at the same target monitoring point by using a Mallat algorithm.

[0124] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the foregoing method embodiments, which will not be described here.

[0125] In an optional embodiment, the embodiment further provides an electronic device, including a memory and a processor, the memory stores a computer program executable on the processor, and the processor executes the computer program to implement the steps of the method embodiments.

[0126] In an optional embodiment, the embodiment further provides a computer readable medium having non-volatile program codes executable by a processor, wherein the program codes cause the processor to execute the method of the method embodiments.

[0127] In addition, the terms "first", "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0128] In several embodiments provided in the embodiment, it should be understood that the disclosed method and device can be implemented by other ways. The device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some communication interface, device or unit, and can be electrical, mechanical or other forms.

[0129] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, that is, they can be located in one place, or can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment.

[0130] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0131] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a nonvolatile computer readable storage medium executable by a processor. Based on this understanding, the technical solutions of the embodiments can be embodied in the form of a software product, the computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0132] Finally, it should be noted that: the above-described embodiments are merely specific embodiments of the present application, used to illustrate the technical solutions of the present application, and not to limit them. The protection scope of the present application is not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can make modifications or easily think of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed by the present application, or make equivalent replacements to some of the technical features; and these modifications, changes or replacements do not cause the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be covered within the protection scope of the present application.

Claims

1. A method for identifying the correlation type of power quality disturbance events, characterized in that: include: Obtain characteristic information of any two power quality disturbance events at the same target monitoring point; Processing the characteristic information of the arbitrary two power quality disturbance events to obtain correlation characteristic information between the arbitrary two power quality disturbance events; Inputting the associated feature information into an associated type classification and recognition model for identification, thereby obtaining an associated type identification result between any two power quality disturbance events; wherein the associated type classification and recognition model is a multi-classifier based on a least squares support vector machine (LS-SVM); The method further comprises: Customize the association type between power quality disturbance events; wherein the association type includes at least one of the following: chain type, concurrent type, and development type; Characterizing the association type to obtain association feature information; Constructing a matter-element model for characterizing the association type according to the association feature information; The characteristic information of the power quality disturbance event includes time characteristics and type characteristics, and the type characteristics include: voltage sag, voltage swell, voltage interruption, transient oscillation, and transient pulse; Processing the characteristic information of the arbitrary two power quality disturbance events to obtain correlation characteristic information between the arbitrary two power quality disturbance events includes: The first time difference result is obtained by subtracting the start time of any two power quality disturbance events and taking the absolute value; The second time difference result is obtained by subtracting the end time of any two power quality disturbance events and taking the absolute value; determining the first time difference result and the second time difference result as time correlation features; The time correlation feature further includes: a third time difference result obtained by taking the absolute value of the difference between the start time of the first power quality disturbance event and the end time of the second power quality disturbance event; Comparing the type features of the arbitrary two power quality disturbance events to obtain a type correlation feature between the arbitrary two power quality disturbance events, wherein the type correlation feature indicates whether the two power quality disturbance events are of the same type or different types; Determining the time correlation feature and the type correlation feature as correlation feature information between the arbitrary two power quality disturbance events; The associated feature information also includes: waveform morphology features, which are divided into covering type, overlapping type and connection type; The association type classification and recognition model includes: a first association type classifier for identifying the chain type, a second association type classifier for identifying the concurrent type, and a third association type classifier for identifying the developmental type; wherein the first association type classifier, the second association type classifier, and the third association type classifier are all LS-SVM classifiers; Inputting the correlation feature information into the correlation type classification recognition model for recognition, and obtaining the correlation type recognition result between the arbitrary two power quality disturbance events, including: Inputting the association feature information into the first association type classifier for classification to obtain a first classification result; Inputting the association feature information into the second association type classifier for classification to obtain a second classification result; Inputting the association feature information into the third association type classifier for classification to obtain a third classification result; According to the first classification result, the second classification result and the third classification result, an association type identification result between the arbitrary two power quality disturbance events is determined.

2. The method according to claim 1, characterized in that The method also includes: Obtain training associated feature samples; Inputting the training association feature sample into an initial LS-SVM classifier to obtain a training association type recognition result corresponding to the training association feature sample; Calculating a function value of an objective function of the initial LS-SVM classifier based on the training association type recognition result; wherein the objective function is a function constructed based on Lagrange multipliers; The parameters of the initial LS-SVM classifier are adjusted according to the function value of the objective function to obtain the first association type classifier, the second association type classifier or the third association type classifier.

3. The method according to claim 1, characterized in that Obtain characteristic information of any two power quality disturbance events at the same target monitoring point, including: The Mallat algorithm is used to extract the time characteristics and type characteristics of any two power quality disturbance events at the same target monitoring point.

4. A device for identifying the correlation type of power quality disturbance events, characterized in that: include: The first acquisition unit is used to acquire characteristic information of any two power quality disturbance events at the same target monitoring point; a processing unit, configured to process the characteristic information of the arbitrary two power quality disturbance events to obtain correlation characteristic information between the arbitrary two power quality disturbance events; an identification unit, configured to input the association feature information into an association type classification identification model for identification, and obtain an association type identification result between any two power quality disturbance events; wherein the association type classification identification model is a multi-classifier based on a least squares support vector machine (LS-SVM); The power quality disturbance event association type identification device is further used to: customize the association type between power quality disturbance events; wherein the association type includes at least one of the following: chain type, concurrent type, and development type; characterize the association type to obtain association feature information; and construct a matter-element model for characterizing the association type based on the association feature information; the feature information of the power quality disturbance event includes time features and type features, and the type features include: voltage sag, voltage swell, voltage interruption, transient oscillation, and transient pulse; The processing unit is also used for: performing a difference process on the start time of any two power quality disturbance events and taking an absolute value to obtain a first time difference result; performing a difference process on the end time of any two power quality disturbance events and taking an absolute value to obtain a second time difference result; determining the first time difference result and the second time difference result as a time correlation feature; the time correlation feature also includes: a third time difference result obtained by taking an absolute value of the difference between the start time of the first power quality disturbance event and the end time of the second power quality disturbance event; performing a comparison process on the type features of the any two power quality disturbance events to obtain a type correlation feature between the any two power quality disturbance events, the type correlation feature characterizing whether the types of the two power quality disturbance events are the same or different; determining the time correlation feature and the type correlation feature as correlation feature information between the any two power quality disturbance events; the correlation feature information also includes: waveform morphology feature, the waveform morphology feature is divided into covering type, overlapping type and connection type; The association type classification and recognition model includes: a first association type classifier for identifying the chain type, a second association type classifier for identifying the concurrent type, and a third association type classifier for identifying the developmental type; wherein the first association type classifier, the second association type classifier, and the third association type classifier are all LS-SVM classifiers; The identification unit is also used to: input the association feature information into the first association type classifier for classification to obtain a first classification result; input the association feature information into the second association type classifier for classification to obtain a second classification result; input the association feature information into the third association type classifier for classification to obtain a third classification result; and determine the association type identification result between any two power quality disturbance events based on the first classification result, the second classification result and the third classification result.

5. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, characterized in that: When a processor executes the computer program, the method according to any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium, wherein: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.

Images