Abnormal Power Grid Data Identification Method, Device, Terminal Equipment, and Storage Medium

Abstract

The present invention discloses a method for identifying abnormal power grid data, including: dividing the power grid data to be identified by using a date division strategy to obtain a plurality of initial division data; dividing the plurality of initial division data by using an event division strategy to obtain a plurality of result division data; determining a plurality of selected classifiers corresponding to the plurality of result division data from a preset classifier set, where the preset classifier set includes classifiers corresponding to different division data; inputting each group of result division data into the corresponding selected classifier to obtain a classification result corresponding to each group of result division data; and obtaining the abnormal power grid data corresponding to the power grid data to be identified based on the plurality of classification results. The present invention also discloses an apparatus for identifying abnormal power grid data, a terminal device, and a storage medium. By using the method of the present invention, the accuracy rate of the classification result is relatively high, thus achieving the technical effect of improving the identification accuracy rate of abnormal power grid data.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid data management, and particularly to a method, device, terminal device and storage medium for identifying abnormal power grid data. Background Art

[0002] When resonance occurs in the power grid, overvoltage and surge current are induced, which will cause problems such as misoperation of power grid protection, threat to the insulation of power equipment, and burnout of power grid voltage transformers. In recent years, with the rapid development of new energy, the topology of the power grid has become more complex, resulting in an increased possibility of resonance in the power grid.

[0003] Existing resonance identification technologies usually extract features based on information such as power grid voltage and current to obtain effective power grid data. At present, methods for identifying abnormal power grid data have been proposed, such as identifying abnormal power grid data through distance or by means of normal distribution.

[0004] However, with the existing technical means, the identification accuracy of abnormal power grid data is relatively low. Summary of the Invention

[0005] The main purpose of the present invention is to provide a method, device, terminal device and storage medium for identifying abnormal power grid data, aiming to solve the technical problem that the identification accuracy of abnormal power grid data is relatively low with the existing technical means in the prior art.

[0006] To achieve the above purpose, the present invention proposes a method for identifying abnormal power grid data, the method comprising the following steps:

[0007] When the power grid data to be identified is obtained, using a date partitioning strategy, partitioning the power grid data to be identified to obtain a plurality of initial partition data;

[0008] Using an event partitioning strategy, partitioning the plurality of initial partition data to obtain a plurality of result partition data;

[0009] Determining a plurality of selected classifiers corresponding to the plurality of result partition data in a preset classifier set, the preset classifier set including classifiers corresponding to different partition data;

[0010] Inputting each group of the result partition data into the corresponding selected classifier to obtain a classification result corresponding to each group of the result partition data;

[0011] Based on the plurality of classification results, obtaining the abnormal power grid data corresponding to the power grid data to be identified.

[0012] Optionally, the date partitioning strategy includes holiday partitioning strategies corresponding to different holidays; the step of partitioning the to-be-identified grid data by using the date partitioning strategy to obtain a plurality of initial partition data includes:

[0013] Partition the to-be-identified grid data by using the time information of the to-be-identified grid data and the preset seasonal time information to obtain seasonal partition data;

[0014] If there is a holiday in the preset holiday set that matches the seasonal partition data, determine the holiday that matches the seasonal partition data as the selected holiday;

[0015] Partition the seasonal partition data by using the selected holiday partitioning strategy corresponding to the selected holiday in the preset holiday set to obtain a plurality of initial partition data.

[0016] Optionally, the date partitioning strategy further includes a workweek partitioning strategy; after the step of partitioning the to-be-identified grid data by using the time information of the to-be-identified grid data and the preset seasonal time information to obtain seasonal partition data, the method further includes:

[0017] If there is no holiday in the preset holiday set that matches the seasonal partition data, partition the seasonal partition data by using the workweek partitioning strategy to obtain a plurality of initial partition data.

[0018] Optionally, the event partitioning strategy includes a plurality of preset major events; the step of partitioning the plurality of initial partition data by using the event partitioning strategy to obtain a plurality of result partition data includes:

[0019] Extract first sub-result partition data that matches the plurality of preset major events from each of the initial partition data;

[0020] Determine the data that does not match the plurality of preset major events in each of the initial partition data as second sub-result partition data;

[0021] Summarize the first sub-result partition data corresponding to each of the initial partition data and the first sub-result partition data corresponding to each of the initial partition data to obtain a plurality of the result partition data.

[0022] Optionally, before the step of determining a plurality of selected classifiers corresponding to the plurality of result partition data in the preset classifier set, the method further includes:

[0023] Obtain a training data set, where the training data set includes a plurality of training data groups, and the plurality of training data are obtained by partitioning historical grid data by using the date partitioning strategy and the event partitioning strategy;

[0024] Input multiple said training data groups into an initial classifier for training respectively to obtain multiple preset classifiers corresponding to the multiple said training data groups;

[0025] Summarize multiple said preset classifiers to obtain the preset classifier set.

[0026] Optionally, the initial classifier includes an input layer, a first long short-term memory network layer, a second long short-term memory network layer, and an output layer;

[0027] The input layer is connected to the first long short-term memory network layer, the first long short-term memory network layer is connected to the second long short-term memory network layer, and the second long short-term memory network layer is connected to the output layer;

[0028] Two adjacent network units in the first long short-term memory network layer are connected, and the output of the previous network unit in the first long short-term memory network layer is the input of the next network unit;

[0029] Two adjacent network units in the first long short-term memory network layer are connected, and the output of the previous network unit in the second long short-term memory network layer is the input of the next network unit;

[0030] The output of the nth network unit in the first long short-term memory network layer is the input of the nth network unit in the second long short-term memory network layer, where n is a natural number other than 0.

[0031] Optionally, the output layer includes a softmax function, and the softmax function is as follows:

[0032]

[0033] where y i is the output of the softmax function, e is the natural constant, is the output of the last network unit in the second long short-term memory network layer.

[0034] In addition, to achieve the above object, the present invention also proposes an abnormal power grid data recognition device, and the device includes:

[0035] A first partitioning module, configured to, when obtaining the power grid data to be recognized, use a date partitioning strategy to partition the power grid data to be recognized to obtain multiple initial partition data;

[0036] A second partitioning module, configured to use an event partitioning strategy to partition multiple said initial partition data to obtain multiple result partition data;

[0037] A determination module, configured to determine a plurality of selected classifiers corresponding to the plurality of result division data from a preset classifier set, where the preset classifier set includes classifiers corresponding to different division data;

[0038] A classification module, configured to input each group of the result division data into the corresponding selected classifier to obtain a classification result corresponding to each group of the result division data;

[0039] An obtaining module, configured to obtain abnormal power grid data corresponding to the power grid data to be recognized based on the plurality of classification results.

[0040] In addition, to achieve the above object, the present invention further provides a terminal device, where the terminal device includes: a memory, a processor, and an abnormal power grid data recognition program stored on the memory and running on the processor. When the abnormal power grid data recognition program is executed by the processor, the steps of the abnormal power grid data recognition method described in any one of the above are implemented.

[0041] In addition, to achieve the above object, the present invention further provides a storage medium, where an abnormal power grid data recognition program is stored on the storage medium. When the abnormal power grid data recognition program is executed by a processor, the steps of the abnormal power grid data recognition method described in any one of the above are implemented.

[0042] The technical solution of the present invention provides a method for recognizing abnormal power grid data. When the power grid data to be recognized is obtained, the date division strategy is used to divide the power grid data to be recognized to obtain a plurality of initial division data; the event division strategy is used to divide the plurality of initial division data to obtain a plurality of result division data; a plurality of selected classifiers corresponding to the plurality of result division data are determined from a preset classifier set, where the preset classifier set includes classifiers corresponding to different division data; each group of the result division data is input into the corresponding selected classifier to obtain a classification result corresponding to each group of the result division data; and abnormal power grid data corresponding to the power grid data to be recognized is obtained based on the plurality of classification results.

[0043] In the existing method, when dividing the power grid data to be recognized, the time factor and different event factors are not considered, resulting in small differences among the multiple data after division. When using a classifier to classify the multiple data, the accuracy of the classification result is poor, and the recognition accuracy of the abnormal power grid data is low. Using the method of the present invention, the time division strategy and the event division strategy are used to divide the power grid data to be recognized, and the differences among the obtained plurality of result division data are large. When using a plurality of selected classifiers to classify the plurality of result division data, the accuracy of the obtained classification result is high, thereby achieving the technical effect of improving the recognition accuracy of the abnormal power grid data. Description of the Drawings

[0044] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on the structures shown in these drawings.

[0045] Figure 1 It is a schematic structural diagram of a terminal device for the hardware operating environment related to the solution of the embodiment of the present invention;

[0046] Figure 2 It is a schematic flowchart of the first embodiment of the method for identifying abnormal power grid data of the present invention;

[0047] Figure 3 It is a schematic structural diagram of the initial classifier of the present invention;

[0048] Figure 4 It is a structural block diagram of the first embodiment of the device for identifying abnormal power grid data of the present invention.

[0049] The realization, functional features and advantages of the object of the present invention will be further described in conjunction with the embodiments with reference to the accompanying drawings. Detailed implementation manners

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0051] Refer to Figure 1 , Figure 1 It is a schematic structural diagram of a terminal device for the hardware operating environment related to the solution of the embodiment of the present invention.

[0052] Generally, the terminal device includes: at least one processor 301, a memory 302, and an identification program for abnormal power grid data stored on the memory and operable on the processor. The identification program for abnormal power grid data is configured to implement the steps of the method for identifying abnormal power grid data as described above.

[0053] The processor 301 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. The processor 301 may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), or PLA (Programmable Logic Array). The processor 301 may also include a main processor and a coprocessor. The main processor is a processor used to process data in the wake state, also known as the CPU (Central Processing Unit); the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 301 may be integrated with a GPU (Graphics Processing Unit), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. The processor 301 may also include an AI (Artificial Intelligence) processor, which is used to process the operation of the abnormal power grid data recognition method, so that the abnormal power grid data recognition method model can be autonomously trained and learned to improve efficiency and accuracy.

[0054] The memory 302 may include one or more storage media, and the storage media may be non-transitory. The memory 302 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices and flash storage devices. In some embodiments, the non-transitory storage medium in the memory 302 is used to store at least one instruction, and the at least one instruction is used to be executed by the processor 301 to implement the abnormal power grid data recognition method provided in the method embodiments of the present application.

[0055] In some embodiments, the terminal may also optionally include: a communication interface 303 and at least one peripheral device. The processor 301, the memory 302, and the communication interface 303 may be connected through a bus or signal lines. Each peripheral device may be connected to the communication interface 303 through a bus, signal lines, or a circuit board. Specifically, the peripheral devices include at least one of a radio frequency circuit 304, a display screen 305, and a power supply 306.

[0056] The communication interface 303 can be used to connect at least one peripheral device related to I / O (Input / Output) to the processor 301 and the memory 302. In some embodiments, the processor 301, the memory 302, and the communication interface 303 are integrated on the same chip or circuit board; in some other embodiments, any one or two of the processor 301, the memory 302, and the communication interface 303 can be implemented on separate chips or circuit boards, and this embodiment does not limit this.

[0057] The radio frequency circuit 304 is used to receive and transmit RF (Radio Frequency) signals, also known as electromagnetic signals. The radio frequency circuit 304 communicates with the communication network and other communication devices through electromagnetic signals. The radio frequency circuit 304 converts an electrical signal into an electromagnetic signal for transmission, or converts the received electromagnetic signal into an electrical signal. Optionally, the radio frequency circuit 304 includes: an antenna system, an RF transceiver, one or more amplifiers, a tuner, an oscillator, a digital signal processor, a codec chipset, a subscriber identity module card, and so on. The radio frequency circuit 304 can communicate with other terminals through at least one wireless communication protocol. The wireless communication protocol includes but is not limited to: metropolitan area network, various generations of mobile communication networks (2G, 3G, 4G, and 5G), wireless local area network, and / or WiFi (Wireless Fidelity) network. In some embodiments, the radio frequency circuit 304 may further include a circuit related to NFC (Near Field Communication), and this application does not limit this.

[0058] The display screen 305 is used to display the UI (User Interface). The UI may include graphics, text, icons, videos, and any combination thereof. When the display screen 305 is a touch display screen, the display screen 305 also has the ability to collect touch signals on or above the surface of the display screen 305. The touch signals can be input as control signals to the processor 301 for processing. At this time, the display screen 305 can also be used to provide virtual buttons and / or a virtual keyboard, also known as soft buttons and / or a soft keyboard. In some embodiments, the display screen 305 can be one, the front panel of the electronic device; in other embodiments, the display screen 305 can be at least two, respectively arranged on different surfaces of the electronic device or in a foldable design; in still other embodiments, the display screen 305 can be a flexible display screen, arranged on the curved surface or the folding surface of the electronic device. Even, the display screen 305 can also be set to an irregular non-rectangular shape, that is, a special-shaped screen. The display screen 305 can be prepared using materials such as LCD (Liquid Crystal Display) and OLED (Organic Light-Emitting Diode).

[0059] The power supply 306 is used to supply power to each component in the electronic device. The power supply 306 can be alternating current, direct current, a disposable battery, or a rechargeable battery. When the power supply 306 includes a rechargeable battery, the rechargeable battery can support wired charging or wireless charging. The rechargeable battery can also be used to support fast charging technology.

[0060] Those skilled in the art can understand that Figure 1 the structure shown in does not constitute a limitation on the terminal device, and may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0061] In addition, an embodiment of the present invention also proposes a storage medium, on which an identification program for abnormal power grid data is stored. When the identification program for abnormal power grid data is executed by a processor, the steps of the method for identifying abnormal power grid data as described above are implemented. Therefore, it will not be elaborated here. In addition, the beneficial effects of using the same method will not be described again. For the technical details not disclosed in the embodiment of the storage medium involved in this application, please refer to the description of the method embodiment of this application. By way of example, the program instructions can be deployed to be executed on a terminal device, or on multiple terminal devices located at one location, or on multiple terminal devices distributed at multiple locations and interconnected through a communication network.

[0062] Those of ordinary skill in the art can understand that all or part of the processes of implementing the methods in the above embodiments can be completed by instructing relevant hardware through a computer program. The above program can be stored in a storage medium. When the program is executed, it can include the processes of the embodiments of the above methods. Among them, the above storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0063] Based on the above hardware structure, an embodiment of the method for identifying abnormal power grid data of the present invention is proposed.

[0064] Referring to Figure 2 , Figure 2 is a schematic flowchart of the first embodiment of the method for identifying abnormal power grid data of the present invention. The method is used for a terminal device, and the method includes the following steps:

[0065] Step S11: When the power grid data to be identified is obtained, use a date partitioning strategy to partition the power grid data to be identified to obtain a plurality of initial partitioned data.

[0066] It should be noted that the execution subject of the present invention is a terminal device. The terminal device is installed with an identification program for abnormal power grid data. When the terminal device executes the identification program for abnormal power grid data, the steps of the method for identifying abnormal power grid data of the present invention are implemented.

[0067] Generally, the power grid data to be identified is the data in the power grid to be identified for abnormal power grid data. There is abnormal power grid data in the power grid data to be identified, and the method of the present invention needs to be used to identify the abnormal power grid data. First, it is necessary to use a date partitioning strategy to partition the power grid data to be identified to obtain a plurality of initial partitioned data.

[0068] Specifically, the date partitioning strategy includes holiday partitioning strategies corresponding to different holidays; the step of using the date partitioning strategy to partition the power grid data to be identified to obtain a plurality of initial partitioned data includes: using the time information of the power grid data to be identified and the preset seasonal time information to partition the power grid data to be identified to obtain seasonal partitioned data; if there are holidays in the preset holiday set that match the seasonal partitioned data, then determine the holidays that match the seasonal partitioned data as the selected holidays; use the selected holiday partitioning strategy corresponding to the selected holidays in the preset holiday set to partition the seasonal partitioned data to obtain a plurality of initial partitioned data.

[0069] Alternatively, the date partitioning strategy further includes a workweek partitioning strategy; after the step of partitioning the to-be-identified grid data using the time information of the to-be-identified grid data and the preset seasonal time information to obtain seasonal partition data, the method further includes: if there is no holiday in the preset holiday set that matches the seasonal partition data, then use the workweek partitioning strategy to partition the seasonal partition data to obtain a plurality of initial partition data.

[0070] The time information of the grid data is the specific time of the grid data, and the preset seasonal time information is usually the time information of the four seasons recognized now: spring from March to May, summer from June to August, autumn from September to November, and winter from December to February. Using the preset seasonal time information and the time information of the grid data, the grid data is partitioned into seasonal partition data. One seasonal partition data is the grid data of one season, and there may be at most four seasonal partition data.

[0071] For each seasonal partition data, it needs to be partitioned again in the above two ways. Among them, the preset holiday set refers to Table 1, and Table 1 is as follows:

[0072] Table 1

[0073]

[0074]

[0075] Among them, the holiday partitioning strategy refers to partitioning in the manner corresponding to the left column in Table 1. When the time information of the seasonal partition data matches the date described for a certain holiday in Table 1, it indicates that there is a holiday in the preset holiday set that matches the seasonal partition data, then the seasonal partition data needs to be further partitioned, and the selected holiday partitioning strategy is used for partitioning.

[0076] For example, if the time information corresponding to the seasonal partition data is from September 30th to October 8th, then the seasonal partition data matches the 7-day vacation (National Day) in Table 1, the selected holiday is the National Day, and the three holiday partitioning strategies corresponding to the 7-day vacation (the last three holiday partitioning strategies in Table 1) are the selected holiday partitioning strategies. Then the seasonal partition data is partitioned into 3 groups of initial partition data: the data on September 30th is one group, the data from October 1st to October 7th is one group, and the data on October 8th is one group. In the present invention, usually the data of each day is used as input for recognition, then for the data from October 1st to October 7th, it needs to be divided into seven times and recognized separately.

[0077] If there is no holiday in the preset holiday concentration that matches the season division data, that is, the time information of the season division data does not match the dates corresponding to any of the holidays described in Table 1, then the working week division strategy is used for division. The working week division strategy is shown in Table 2, and Table 2 is as follows:

[0078] Table 2

[0079] Type Description Weekend Regular day off Monday The first working day of each week Friday The last working day of each week Tuesday, Wednesday, Thursday Regular working days

[0080] It can be understood that for the working week division strategy and the holiday division strategy, the present invention only exemplarily presents Table 1 and Table 2, and users can set new strategies based on requirements and the inspiration of the present invention.

[0081] Step S12: Use the event division strategy to divide the multiple initial division data to obtain multiple result division data.

[0082] It should be noted that the event division strategy includes multiple preset major events; the step of using the event division strategy to divide the multiple initial division data to obtain multiple result division data includes: extracting first sub-result division data that matches the multiple preset major events from each of the initial division data; determining the data that does not match the multiple preset major events in each of the initial division data as second sub-result division data; and summarizing the first sub-result division data corresponding to each of the initial division data and the first sub-result division data corresponding to each of the initial division data to obtain the multiple result division data.

[0083] Among them, the multiple preset major events may include political events (policy events), medical events (such as disease outbreak events), etc., which are not limited in the present invention. Among them, if the event information is similar to any one of the multiple preset major events, then the event information matches the multiple preset major events, otherwise, it does not match.

[0084] If the multiple initial division data correspond to 5 first sub-result division data and 3 second sub-result division data, then the total number of result division data after summarization is 8, that is, the first sub-result division data and the second sub-result division data are mixed together as the result division data.

[0085] It can be understood that for an initial division data, it may include data for a relatively long time period. For example, an initial division data includes all the regular working day data within a certain month, and at the same time, there are several days in this month when preset major events occur. Then, this initial division data needs to be divided into an initial division data with preset major events - the first sub-result division data, and an initial division data without preset major events - the second sub-result division data, that is, two result division data are obtained.

[0086] In addition, for an initial partition data, it may only include data within the time period involved in the preset major events, then the entire initial partition data is determined as a result partition data (the first sub-result partition data). For an initial partition data that does not include any data within the time period involved in the preset major events, the entire initial partition data is determined as a result partition data (the second sub-result partition data).

[0087] Step S13: Determine multiple selected classifiers corresponding to the multiple result partition data in the preset classifier set, where the preset classifier set includes classifiers corresponding to different partition data.

[0088] Step S14: Input each group of the result partition data into the corresponding selected classifier to obtain the classification result corresponding to each group of the result partition data.

[0089] Step S15: Based on the multiple classification results, obtain the abnormal power grid data corresponding to the power grid data to be recognized.

[0090] Refer to Figure 3 , Figure 3 , which is a schematic structural diagram of the initial classifier of the present invention; the initial classifier includes an input layer, a first long short-term memory network layer (the first LSTM layer), a second long short-term memory network layer (the second LSTM layer), and an output layer; the input layer is connected to the first long short-term memory network layer, the first long short-term memory network layer is connected to the second long short-term memory network layer, and the second long short-term memory network layer is connected to the output layer; two adjacent network units in the first long short-term memory network layer are connected, and the output of the previous network unit in the first long short-term memory network layer is the input of the next network unit; two adjacent network units in the first long short-term memory network layer are connected, and the output of the previous network unit in the second long short-term memory network layer is the input of the next network unit; the output of the nth network unit in the first long short-term memory network layer is the input of the nth network unit in the second long short-term memory network layer, where n is a natural number other than 0.

[0091] Among them, when the initial classifier is trained, the data input by the input layer is the data of one day, that is, Xt can refer to the data of the same day divided into different t time periods (24, one input data per hour). The output layer includes a softmax function, and the softmax function is as follows:

[0092]

[0093] Among them, y i is the output of the softmax function, e is the natural constant, is the output of the last network unit in the second long short-term memory network layer. Since this invention patent only considers abnormal power grid data and normal data, the value of i is 1 and 2 (where it can be that 1 corresponds to abnormal power grid data, or 2 corresponds to abnormal power grid data, which is set by the user based on requirements).

[0094] Before the step of determining multiple selected classifiers corresponding to the multiple result division data in the preset classifier set, the method further includes: obtaining a training data set, the training data set includes multiple training data groups, and the multiple training data are obtained by dividing historical power grid data using a date division strategy and the event division strategy; respectively inputting the multiple training data groups into an initial classifier for training to obtain multiple preset classifiers corresponding to the multiple training data groups; and summarizing the multiple preset classifiers to obtain the preset classifier set.

[0095] Divide the historical power grid data in the above manner of the present invention (using the date division strategy and the event division strategy for division) to obtain multiple training data groups. Each training data group is used to train the initial classifier once. The data in each training data group may include data for multiple days, and the data for each day is divided into data for multiple time periods as the input Xt, and the data between different days is input in batches for training.

[0096] Dividing the historical power grid data in the manner of the present invention may involve 20 training data groups, and then training the initial classifier more than 20 times to obtain more than 20 corresponding preset classifiers. To ensure better classification effects of the preset classifiers in the preset classifier set, when dividing the historical power grid data in the manner of the present invention, the higher the division fineness, the better.

[0097] Training of the first LSTM layer: After the first LSTM (the first network unit) on the left side of the first layer is trained and calculated, its result is input into the next LSTM in this layer and the corresponding connected LSTM in the second layer. The second LSTM on the left side of the first layer is trained based on the input of the first LSTM and the data of the input layer, and the results are respectively given to the next LSTM in this layer and the corresponding connected LSTM in the second layer, and the training of the first layer of LSTM is completed in sequence. Training of the second layer, similarly, the second LSTM layer relies on the LSTM connected to the first layer and this layer as the input to complete the training.

[0098] For the actual multiple result division data, determine the corresponding selected preset classifier based on its data type. For example, if the result division data is a regular working day, without a preset major event, and in summer, the preset classifier corresponding to the training data group of regular working days, without a preset major event, and in summer in the preset classifier set is determined as the selected preset classifier.

[0099] Divide the data using multiple actual results corresponding to the power grid data to be recognized, determine multiple corresponding selected classifiers, and classify each result-divided data with a corresponding selected classifier to obtain classification results. Synthesize the classification results corresponding to all the result-divided data, and summarize the abnormal power grid data corresponding to each classification result, which is the abnormal power grid data in step S15.

[0100] The technical solution of the present invention proposes a method for identifying abnormal power grid data. When the power grid data to be recognized is obtained, use the date division strategy to divide the power grid data to be recognized to obtain multiple initial divided data; use the event division strategy to divide the multiple initial divided data to obtain multiple result-divided data; determine multiple selected classifiers corresponding to the multiple result-divided data in a preset classifier set, and the preset classifier set includes classifiers corresponding to different divided data; input each group of the result-divided data into the corresponding selected classifier to obtain the classification result corresponding to each group of the result-divided data; based on the multiple classification results, obtain the abnormal power grid data corresponding to the power grid data to be recognized.

[0101] In the existing method, when dividing the power grid data to be recognized, the time factor and different event factors are not considered, resulting in a small difference in the multiple data after division. When using a classifier to classify the multiple data, the accuracy of the classification result is poor, and the recognition accuracy of abnormal power grid data is low. Using the method of the present invention, the power grid data to be recognized is divided using the time division strategy and the event division strategy, and the difference in the multiple result-divided data obtained is large. When using multiple selected classifiers to classify the multiple result-divided data, the accuracy of the obtained classification result is high, thus achieving the technical effect of improving the recognition accuracy of abnormal power grid data.

[0102] Refer to Figure 4 , Figure 4 As shown in the structural block diagram of the first embodiment of the device for identifying abnormal power grid data of the present invention, the device is used for a terminal device. Based on the same inventive concept as the foregoing embodiment, the device includes:

[0103] The first division module 10 is used to divide the power grid data to be recognized using the date division strategy to obtain multiple initial divided data when the power grid data to be recognized is obtained;

[0104] The second division module 20 is used to divide the multiple initial divided data using the event division strategy to obtain multiple result-divided data;

[0105] A determining module 30, configured to determine a plurality of selected classifiers corresponding to the plurality of result partitioning data from a preset classifier set, where the preset classifier set includes classifiers corresponding to different partitioning data;

[0106] A classifying module 30, configured to input each group of the result partitioning data into a corresponding selected classifier to obtain a classification result corresponding to each group of the result partitioning data;

[0107] An obtaining module 50, configured to obtain abnormal grid data corresponding to the grid data to be recognized based on the plurality of classification results.

[0108] It should be noted that since the steps executed by the device in this embodiment are the same as those in the foregoing method embodiment, the specific implementation manner and the achievable technical effects can refer to the foregoing embodiment, and will not be elaborated here.

[0109] The foregoing are only optional embodiments of the present invention, and do not limit the patent scope of the present invention. Any equivalent structural transformation made by using the content of the specification and drawings of the present invention under the inventive concept of the present invention, or direct / indirect application in other related technical fields is included in the patent protection scope of the present invention.

Claims

1. A method for identifying abnormal power grid data, characterized in that, the method comprises the following steps: When obtaining the power grid data to be identified, using a date partitioning strategy to partition the power grid data to be identified, and obtaining a plurality of initial partition data; Using an event partitioning strategy to partition the plurality of initial partition data, and obtaining a plurality of result partition data; Obtaining a training data set, the training data set includes a plurality of training data groups, and the plurality of training data are obtained by partitioning historical power grid data using a date partitioning strategy and the event partitioning strategy; Respectively inputting the plurality of training data groups into an initial classifier for training, and obtaining a plurality of preset classifiers corresponding to the plurality of training data groups; Aggregating the plurality of preset classifiers to obtain the preset classifier set; Determining a plurality of selected classifiers corresponding to the plurality of result partition data in the preset classifier set, and the preset classifier set includes classifiers corresponding to different partition data; Inputting each group of the result partition data into the corresponding selected classifier, and obtaining a classification result corresponding to each group of the result partition data; Based on the plurality of classification results, obtaining abnormal power grid data corresponding to the power grid data to be identified.

2. The method according to claim 1, characterized in that, the date partitioning strategy includes holiday partitioning strategies corresponding to different holidays; the step of using the date partitioning strategy to partition the power grid data to be identified and obtaining a plurality of initial partition data includes: Using the time information of the power grid data to be identified and preset seasonal time information to partition the power grid data to be identified, and obtaining seasonal partition data; If there is a holiday in the preset holiday set that matches the seasonal partition data, determining the holiday that matches the seasonal partition data as the selected holiday; Using the selected holiday partitioning strategy corresponding to the selected holiday in the preset holiday set to partition the seasonal partition data, and obtaining a plurality of initial partition data.

3. The method according to claim 2, characterized in that, the date partitioning strategy further includes a work week partitioning strategy; after the step of using the time information of the power grid data to be identified and preset seasonal time information to partition the power grid data to be identified and obtaining seasonal partition data, the method further includes: If there is no holiday in the preset holiday set that matches the seasonal partition data, using the work week partitioning strategy to partition the seasonal partition data, and obtaining a plurality of initial partition data.

4. The method according to claim 3, characterized in that, the event partitioning strategy includes a plurality of preset major events; the step of using the event partitioning strategy to partition the plurality of initial partition data and obtaining a plurality of result partition data includes: Extracting first sub-result partition data that matches the plurality of preset major events from each of the initial partition data; Determining the data in each of the initial partition data that does not match the plurality of preset major events as second sub-result partition data; Summarize the first sub-result partition data corresponding to each of the initial partition data and the second sub-result partition data corresponding to each of the initial partition data to obtain a plurality of the result partition data.

5. The method according to claim 1, wherein, the initial classifier includes an input layer, a first long short-term memory network layer, a second long short-term memory network layer, and an output layer; the input layer is connected to the first long short-term memory network layer, the first long short-term memory network layer is connected to the second long short-term memory network layer, and the second long short-term memory network layer is connected to the output layer; two adjacent network units in the first long short-term memory network layer are connected, and the output of the previous network unit in the first long short-term memory network layer is the input of the subsequent network unit; two adjacent network units in the first long short-term memory network layer are connected, and the output of the previous network unit in the second long short-term memory network layer is the input of the subsequent network unit; the output of the nth network unit in the first long short-term memory network layer is the input of the nth network unit in the second long short-term memory network layer, where n is a natural number other than 0.

6. The method according to claim 5, wherein, the output layer includes a softmax function, and the softmax function is as follows: Among them, is the output of the softmax function, is the natural constant, is the output of the last network unit in the second long short-term memory network layer.

7. An abnormal power grid data recognition device, wherein, the device includes: a first partitioning module, configured to, when obtaining power grid data to be recognized, use a date partitioning strategy to partition the power grid data to be recognized to obtain a plurality of initial partition data; a second partitioning module, configured to use an event partitioning strategy to partition the plurality of initial partition data to obtain a plurality of result partition data; a determination module, configured to obtain a training data set, the training data set including a plurality of training data groups, and the plurality of training data are obtained by partitioning historical power grid data using a date partitioning strategy and the event partitioning strategy; input the plurality of training data groups into an initial classifier for training respectively to obtain a plurality of preset classifiers corresponding to the plurality of training data groups; summarize the plurality of preset classifiers to obtain the preset classifier set; and further configured to determine a plurality of selected classifiers corresponding to the plurality of result partition data in the preset classifier set, where the preset classifier set includes classifiers corresponding to different partition data; a classification module, configured to input each group of the result partition data into the corresponding selected classifier to obtain a classification result corresponding to each group of the result partition data; an obtaining module, configured to obtain abnormal power grid data corresponding to the power grid data to be recognized based on the plurality of classification results.

8. A terminal device, wherein, the terminal device includes: a memory, a processor, and an abnormal power grid data recognition program stored on the memory and running on the processor, and when the abnormal power grid data recognition program is executed by the processor, the steps of the abnormal power grid data recognition method according to any one of claims 1 to 6 are implemented.

9. A storage medium, wherein, An identification program for abnormal power grid data is stored on the storage medium. When the identification program for abnormal power grid data is executed by a processor, the steps of the identification method for abnormal power grid data according to any one of claims 1 to 6 are implemented.

Images