An operation data feature analysis method suitable for power distribution network operation state judgment

Abstract

The application discloses a kind of operation data feature analysis methods suitable for distribution network operating state judgment, it includes: obtaining multiple scene dimensions under the data to be analyzed and pre-processing;Characteristic value calculation is carried out based on the data after pre-processing, and the characteristic value data of each scene dimension in different time section is obtained;According to the preset characteristic dimension, the characteristic value data is grouped, and multiple characteristic value data slices under each time section are generated, and different characteristic value data slices are associated according to the correlation between data;Characteristic value data slice with associated relationship is combined to generate characteristic value data block;Characteristic value data block under different time section is used as the domain of two-dimensional table, and two-dimensional characteristic value snapshot table is generated;Operation data feature analysis is carried out based on two-dimensional characteristic value snapshot table.The present application can organically comb the disordered and massive data, and be normalized into feature snapshot, facilitate the operation abnormal analysis of distribution network and distribution equipment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution automation, and particularly relates to a running data feature analysis method suitable for judging abnormal operation states of a power distribution network and power distribution equipment. BACKGROUND

[0002] Due to the large number and wide distribution of power distribution network equipment, the first-line operation and maintenance human resources are not matched with the scale of the power distribution network and the operation and maintenance workload. At present, the operation and maintenance mode mainly based on manual work cannot meet the requirement of high reliability, but the deployment of power distribution, power utilization, and metering systems has achieved certain results in supporting the operation and maintenance of the power distribution network. The power distribution automation coverage rate of more than 90% has been realized in the whole country, and the visual controllability and fault perception of the power distribution network have been basically realized. In the process, a large amount of planned management, steady-state operation, and transient recording data have been accumulated. Meanwhile, the power utilization system supports the monitoring of the operation of distribution transformers and guides the operation and maintenance of the distribution transformers.

[0003] However, the overall operation and maintenance of the existing power distribution network is still mainly based on traditional periodic inspection and fault repair. The overhead (hybrid) line of the power distribution network is the focus of operation and maintenance control. More than 85% of tripping faults have short-time grounding and short-time discharge phenomena before the faults occur, and can be self-recovered without manual processing, which has great concealment. Therefore, it is urgent to change the operation and maintenance management mode, and change the operation and maintenance control mode from “after-event repair and inspection” to “pre-event early warning and active processing”, so as to improve the operation and maintenance control ability of the power distribution network and the power supply service level.

[0004] At present, the power distribution network has accumulated a large amount of operation and transient recording data, but has not formed sufficient pre-event control basis for the power distribution network. The power distribution network has many types of equipment, complex channel environment, and great difficulty in line inspection. The existing periodic inspection mode mainly based on human work has weak pertinence and low detection efficiency. Based on this, it is necessary to provide a method for analyzing the operation state of the power distribution network and the power distribution equipment, to excavate the line abnormal state information contained in the operation data, to locate the abnormal area and abnormal reason in advance, and to dispatch operation personnel, so as to greatly improve the line inspection efficiency. SUMMARY

[0005] The purpose of the present application is to provide a running data feature analysis method suitable for judging the operation state of a power distribution network, which can organically sort out disordered and massive data and standardize the data into feature snapshots, so as to facilitate the operation abnormality analysis of the power distribution network and power distribution equipment. The technical scheme adopted by the present application is as follows.

[0006] In one aspect, the present application provides a running data feature analysis method suitable for judging the operation state of a power distribution network, comprising:

[0007] obtaining power distribution network operation data to be analyzed;

[0008] The power distribution network operation data to be analyzed is preprocessed to obtain a plurality of slice data corresponding to different analysis dimensions;

[0009] Characteristic value calculation is performed based on the slice data to obtain characteristic value data of each analysis dimension at different time sections;

[0010] The characteristic value data is grouped according to a preset characteristic dimension to generate a plurality of characteristic value data slices corresponding to different characteristic dimensions at each time section, and the slice data in different characteristic value data slices are associated according to the association relationship between the data;

[0011] Characteristic value data slices of different characteristic dimensions having an association relationship are combined to generate a characteristic value data tile;

[0012] Two-dimensional table tuple information determined in advance according to an operation data characteristic analysis target is obtained, and the characteristic value data tile at different time sections is taken as the domain of a two-dimensional table to generate a two-dimensional characteristic value snapshot table;

[0013] Operation data characteristic analysis is performed based on the two-dimensional characteristic value snapshot table.

[0014] In the present application, by generating a two-dimensional characteristic value snapshot table of operation data, the advantages of data snapshots in data alignment, retrieval, etc. are utilized, and the acquisition and associated comparative analysis of data can be conveniently specified, which can greatly improve the efficiency of subsequent power distribution network operation state analysis.

[0015] Optionally, the data to be analyzed includes power distribution network operation data, PMS operation management data, OMS system power distribution network control operation data, and meteorological environment data;

[0016] The analysis dimension includes at least one or more dimensions of steady-state data, transient data, model data, management data, and environmental data;

[0017] The characteristic dimension includes at least one or more of basic characteristics, safety characteristics, fault characteristics, abnormal characteristics, and typical scenario characteristics;

[0018] The characteristic value calculation based on the preprocessed data includes calculating one or more of historical steady-state data, real-time section data, operation data, and transient recording wave data according to a preset operation index;

[0019] The preset operation index includes at least one or more of power distribution network operation index, primary equipment operation index, secondary equipment operation index, transient analysis index, and environmental index;

[0020] The power distribution network operation indexes include: main transformer overload rate, main transformer overload rate, main transformer light load rate, power distribution transformer overload rate, power distribution transformer overload rate, power distribution transformer light load rate, 10kV line overload rate, 10kV line overload rate, 10kV line light load rate, main transformer 10kV bus voltage unqualified rate, substation voltage unqualified rate, low-voltage user voltage unqualified rate, power distribution transformer three-phase imbalance ratio and power distribution transformer serious three-phase imbalance ratio;

[0021] The primary equipment operation indexes include: switch tripping statistics, protection action statistics, morning exercise execution failure statistics, transformer temperature out-of-limit statistics, cable head temperature out-of-limit statistics and equipment partial discharge statistics;

[0022] The secondary equipment operation indexes include: long-term offline statistics, frequent on-off statistics, remote control failure statistics, data quality defect statistics, remote signal jitter, alternating current power failure, power failure, flow exceeding standard and terminal battery defect indexes;

[0023] The transient analysis indexes include: parameter identification, phase current mutation, first half-wave polarity, negative sequence current calculation, zero sequence admittance calculation, zero sequence current active component, 5th harmonic component, power frequency zero sequence current amplitude and three-phase power frequency zero sequence current direction;

[0024] The environmental indexes include: high temperature early warning statistics, strong wind early warning statistics, low temperature early warning statistics, lightning early warning statistics and water inlet alarm statistics.

[0025] The calculation of the above-mentioned various operation indexes, and the further classification of the feature data under each feature dimension according to the operation index calculation results, can refer to the prior art for specific calculation methods.

[0026] Optionally, the pre-processing of the data to be analyzed includes: rejecting bad data, classifying data, and respectively performing time alignment processing on various types of data.

[0027] Optionally, the rejecting bad data includes: for data with data quality codes, filtering non-real-time values, non-refresh values, invalid data values, bad data values and working condition exit values; for data with association relationships, checking data integrity and legality, filtering remote signal and remote measurement mismatch values, total plus data mismatch values and PQI calculation mismatch values;

[0028] The data classification obtained by classifying the data includes at least: real-time data, non-time-tagged measurement section data, fault SOE data, waveform data, time-tagged steady-state data and operation data;

[0029] The time alignment processing of various types of data respectively includes: for real-time data and non-time-target measurement section data, ignoring the time difference of uploading, generating measurement time section alignment; for fault SOE data, taking the time index of key data as the section, and filling in the data values of other times; for waveform data, optionally finding a segment of the waveform with a mutation characteristic waveform, taking the mutation characteristic waveform as the reference time of time alignment, and performing waveform data time alignment on the start point of other waveforms where mutations occur; for steady-state data with time marks, taking minutes as the interval and the whole point time as the reference line for alignment; for operation data, taking the execution time as the reference for time alignment; for transient recording waveform data, taking the recording waveform start time as the reference for time alignment.

[0030] The above data preprocessing can guarantee the reliability of the results of subsequent operation state analysis according to the snapshot table through the elimination of bad data and time alignment processing.

[0031] Optionally, the pre-processing of the data to be analyzed further includes: performing data correlation analysis to obtain the correlation relationship of the data in each preset characteristic dimension and the correlation relationship between different characteristic dimensions.

[0032] The feature value calculation based on the pre-processed data further includes: performing normalization processing on the calculated feature value data.

[0033] The above technical solution, the normalization processing of the feature value can make it more convenient and intuitive to use the snapshot table to compare the data between the snapshot table, and reduce the complexity of subsequent analysis. Since the feature value obtained after the feature value extraction calculation is a dimensionless floating point number, the present application analyzes the data correlation relationship before data slicing and feature value extraction, thereby reducing the difficulty of data correlation relationship analysis.

[0034] Optionally, the correlation relationship information of the data includes at least one or more of the following: topological correlation relationship, influence range relationship, time causal relationship, voltage level relationship, power supply upstream and downstream relationship, operation locking relationship, and weighted relationship.

[0035] The data correlation analysis includes at least one or more of the following:

[0036] 1) generating a full network topology mapping correlation relationship, recording the correlation relationship between the root node and the device object, wherein the correlation relationship between the root node and the device object includes the relationship between the main network outgoing line switch and the distribution network root node feeder section, the relationship between the double-sided power supply and the tie switch, and the relationship between the low-voltage user and the distribution transformer;

[0037] 2) marking each voltage level power point to generate the senior-junior relationship between power points;

[0038] 3) According to the power supply points of each voltage level, a power supply range relationship is generated, a dyeing method is used to dye the power supply equipment, the power supply relationship is recorded, and the power supply relationship between the distribution network bus and the main network bus is marked;

[0039] 4) An alarm event relationship is generated, historical alarm records are event aggregated, and alarm signals associated with the comprehensive event are marked;

[0040] 5) The time relationship of the event is formed by marking the time of the comprehensive event aggregation result;

[0041] 6) The task execution relationship and event processing relationship are formed by marking the dispatching operation information and operation task;

[0042] 7) The weighted relationship is formed by marking the event priority, signal priority and device priority.

[0043] Optionally, the plurality of feature value data slices are a plurality of groups of mutually pointing vector queues. Vector is a sequential container encapsulating a dynamic size array, supporting fast and direct access to any element in the sequence, and the association relationship between data in different slices is realized by mutual pointing of Vector.

[0044] Optionally, the combination of feature value data slices of different feature dimensions with an existing association relationship to generate feature value data chunks includes: taking any feature value data slice in the data chunk as a key, and corresponding sorting the data in other feature value data slices to improve the convenience of subsequent data analysis. The snapshot table determines the grouping through dictionary mapping, establishes an index for each group, and accelerates the alignment, arrangement, filtering and retrieval between snapshot tables.

[0045] Optionally, the pre-determined two-dimensional table tuple is an associated device;

[0046] The running data feature analysis based on the two-dimensional feature value snapshot table includes horizontal trend analysis and vertical correlation comparison analysis, the horizontal trend analysis is used to compare the change trend of the feature value data of the same associated device at different time sections, and the vertical correlation comparison analysis is used to compare the feature value data relationship between the associated devices with an existing association relationship at the same time section.

[0047] Optionally, the vertical correlation comparison analysis includes: determining a device object set participating in comparison, and for all device objects participating in comparison, according to the feature value data chunk at the same time section, obtaining the device object with a feature value escape situation in the device object set as the device object with a state anomaly.

[0048] In a second aspect, the present application provides a running data feature analysis device suitable for power distribution network running state judgment, comprising:

[0049] A data to be analyzed acquisition module configured to acquire power distribution network operation data to be analyzed;

[0050] A preprocessing module configured to preprocess the power distribution network operation data to be analyzed to obtain a plurality of slice data corresponding to different analysis dimensions;

[0051] A feature value calculation module configured to calculate feature values based on the slice data to obtain feature value data of each analysis dimension at different time sections;

[0052] A feature value slice generation module configured to group the feature value data according to a preset feature dimension to generate a plurality of feature value data slices corresponding to different feature dimensions at each time section, and the slice data in different feature value data slices are associated according to the association relationship between the data;

[0053] A feature value chunk generation module configured to combine the feature value data slices of different feature dimensions with an association relationship to generate a feature value data chunk;

[0054] A snapshot table generation module configured to obtain two-dimensional table tuple information determined in advance according to an operation data feature analysis target, and generate a two-dimensional feature value snapshot table by taking the feature value data chunk at different time sections as the domain of the two-dimensional table;

[0055] And an analysis module configured to perform operation data feature analysis based on the two-dimensional feature value snapshot table.

[0056] In a third aspect, the present application provides a storage medium containing computer executable instructions, which, when executed by a computer processor, implement the steps of the operation data feature analysis method for power distribution network operation state judgment according to the first aspect.

[0057] Advantages

[0058] The operation data feature analysis method of the present application converts the chaotic and massive data into feature values of multiple dimensions through the conversion from the scene dimension to the feature dimension, and normalizes the feature values into feature snapshots, which facilitates the use of longitudinal trend analysis and horizontal correlation comparison analysis methods to obtain more accurate state analysis results. The method can be applied to power distribution network operation data analysis in various scenarios, and based on the historical data, steady-state data, transient waveform data and other data accumulated by the power distribution network, the method can analyze and obtain the analysis and judgment results of the safety state, abnormal state and fault state of the power grid and equipment, and form the basis for pre-fault management and control of the distribution network. The present application can support the operation management of the power distribution network and power distribution equipment, and can greatly improve the line patrol efficiency, improve the power supply reliability and fault handling capacity of the power distribution network. BRIEF DESCRIPTION OF DRAWINGS

[0059] Figure 1 Fig. 1 shows a flowchart of a power distribution network and equipment operation state analysis method according to an embodiment of the present application;

[0060] Figure 2 Fig. 2 shows a schematic diagram of a data collection process in the operation data feature analysis method according to the present application;

[0061] Figure 3 Fig. 3 shows a schematic diagram of a two-dimensional snapshot table generation flow according to an embodiment of the present application;

[0062] Figure 4 Fig. 4 shows a schematic diagram of real-time data alignment processing when data preprocessing is performed according to an embodiment of the present application;

[0063] Figure 5 Fig. 5 shows a schematic diagram of SOE data alignment processing when data preprocessing is performed according to an embodiment of the present application;

[0064] Figure 6 Fig. 6 shows a schematic diagram of waveform data alignment processing when data preprocessing is performed according to an embodiment of the present application;

[0065] Figure 7 Fig. 7 shows a schematic diagram of a horizontal trend analysis according to an embodiment of the present application;

[0066] Figure 8 Fig. 8 shows a schematic diagram of associated equipment division when longitudinal correlation comparison analysis is performed according to an embodiment of the present application;

[0067] Figure 9 Fig. 9 shows a schematic diagram of a feature escape phenomenon of associated equipment when longitudinal correlation comparison analysis is performed according to an embodiment of the present application;

[0068] Figure 10 Fig. 10 shows a schematic diagram of feature escape statistics of associated equipment when longitudinal correlation comparison analysis is performed according to an embodiment of the present application. DETAILED DESCRIPTION

[0069] The present application will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0070] The technical concept of the present application is to collect models, parameters, environments, and operation data of the entire power distribution network and entire service, form a continuous operation feature data snapshot of the entire power grid, and thus exhibit the full-dimensional data rules of associated equipment or power grid from multiple dimensions such as time, space, and object, so as to provide service support for monitoring, prevention, and decision-making of production control personnel, and realize deep analysis of power distribution network operation data and equipment features.

[0071] Embodiment 1

[0072] The embodiment introduces an operation data feature analysis method for power distribution network operation state judgment, comprising:

[0073] Obtain the operation data of the power distribution network to be analyzed;

[0074] Preprocess the operation data of the power distribution network to be analyzed to obtain a plurality of slice data corresponding to different analysis dimensions;

[0075] Calculate the eigenvalues based on the slice data obtained after preprocessing to obtain eigenvalue data of each analysis dimension at different time sections;

[0076] Group the eigenvalue data according to the preset feature dimension to generate a plurality of eigenvalue data slices corresponding to different feature dimensions at each time section, and the slice data in different eigenvalue data slices are associated according to the association relationship between the data;

[0077] Combine the eigenvalue data slices of different feature dimensions with the association relationship to generate eigenvalue data chunks;

[0078] Obtain two-dimensional table tuple information determined in advance according to the operation data feature analysis target, and use the eigenvalue data chunks at different time sections as the domain of the two-dimensional table to generate a two-dimensional eigenvalue snapshot table;

[0079] Perform operation data feature analysis based on the two-dimensional eigenvalue snapshot table.

[0080] The generation process of the snapshot table is shown in Figure 3 By generating the two-dimensional eigenvalue snapshot table of the operation data, the advantages of data alignment, retrieval, etc. can be utilized to facilitate the acquisition and associated comparative analysis of the data, and the efficiency of subsequent power distribution network operation state analysis can be improved.

[0081] Please refer to Figure 1 The operation data feature analysis method of the embodiment, when used for the power distribution network operation state judgment of the embodiment, involves the following aspects: data collection and storage, data preprocessing and generation of dimensionally sliced data, eigenvalue calculation, eigenvalue collection, snapshot generation, and operation state analysis based on the snapshot table. The operation state analysis includes snapshot trend analysis and snapshot comparative analysis to ultimately obtain power distribution network and equipment operation state information. The following will be introduced in detail.

[0082] I. Data collection and storage

[0083] The embodiment first obtains the operation data of the power distribution network and equipment. For the convenience of subsequent analysis and calculation, the full-service model data of the power distribution network can be collected at the same time.

[0084] Please refer to Figure 2As shown, the data to be analyzed is obtained through multi-service system data access. The accessed service system data includes: DMS power distribution network operation data, PMS operation management data, OMS system power distribution network control operation data, meteorological environment data, recording wave data, online monitoring data, etc., specifically including: OMS system scheduling log data, operation ticket data, etc.; PMS system power failure plan data, equipment defect inspection data, equipment fault inspection data, equipment maintenance data, etc.; state monitoring system equipment state evaluation data, fault diagnosis result data, transformer oil chromatographic monitoring data, cabinet partial discharge data, etc.; transient recording wave file of fault recording wave system; temperature, rainfall, icing, lightning, typhoon forecast data of meteorological system, etc.

[0085] When accessing data, the data of multiple systems is transmitted across systems through a distributed message queue (Kafka), ETL data extraction, a data bus, and a message bus to complete data collection. When storing the collected data, the data is supported in the loading link according to the business data classification configured by the data source, and is loaded through traditional ETL tools such as Kettle and Sqoop tools supporting distributed storage, and is output to distributed storage through a data service bus and a message bus. The distributed storage architecture includes a distributed relational database, a distributed time series database, a distributed real-time database, and a distributed file system.

[0086] II. Data preprocessing and generation of multi-dimensional slice data

[0087] Reference Figure 3 In this embodiment, the preprocessing of the collected and stored data includes: obtaining the correlation relationship of the data in each preset characteristic dimension and the correlation relationship between different characteristic dimensions through data correlation analysis, eliminating bad data, classifying the data, and performing time alignment processing on each type of data.

[0088] In this embodiment, the above-mentioned preset characteristic dimensions include: basic characteristics, safety characteristics, fault characteristics, abnormal characteristics, typical scene characteristics, and the like.

[0089] The content of data correlation analysis can include:

[0090] ①Generate the correlation relationship of the whole network topology mapping: record the correlation relationship between the root node and the device object, including the relationship between the main network outgoing switch and the distribution network root node feeder section, the relationship between the double-sided power supply and the tie switch, and the relationship between the low-voltage user and the distribution transformer, etc.

[0091] ②Mark each voltage level power point to generate the superior-inferior relationship between power points.

[0092] ③According to the power supply point of each voltage level, a power supply range relationship is generated: the power supply equipment is colored by using the dyeing method, the power supply relationship is recorded, and the power supply relationship between the distribution network bus and the main network bus is marked;

[0093] ④An alarm event relationship is generated: the historical alarm records are collected, and the alarm signals associated with the comprehensive event are marked;

[0094] ⑤The comprehensive event collection result is time marked, and the time relationship of the event is formed;

[0095] ⑥The dispatching operation information and the operation task are marked, and the task execution relationship and the event processing relationship are formed;

[0096] The event priority, signal priority and device priority are marked, and a weighted relationship is formed.

[0097] The above bad data rejection includes: for data with data quality code, filtering non-real-time value, non-refresh value, invalid data value, bad data value and working condition exit value; for data related to association, checking data integrity and legality, filtering remote signal and remote measurement mismatch value, total data mismatch value and PQI calculation mismatch value;

[0098] The above classification of data obtains these types of data: real-time data, non-time-marked measurement section data, fault SOE data, waveform data, time-marked steady-state data and operation data.

[0099] Based on the above classification, the time alignment processing of each type of data is carried out respectively, including: Figure 4 For real-time data and non-time-marked measurement section data, ignore the upload time difference, and generate measurement alignment time section; for example Figure 5 For fault SOE data, the time index of key data is used as the section, and the data values of other times are filled in; for example Figure 6 For waveform data, a characteristic waveform with mutation is selected from a section of waveform, and the characteristic waveform with mutation is used as the reference time for time alignment. For time-marked steady-state data, minute-level intervals and whole point time reference lines are used for alignment. For operation data, the execution time is used as the reference for time alignment. For transient recording wave data, the recording start time is used as the reference for time alignment.

[0100] The above data preprocessing, through bad data rejection and time alignment processing, can guarantee the reliability of the results of subsequent running state analysis according to the snapshot table.

[0101] After the above data preprocessing, a plurality of slice data corresponding to different analysis dimensions can be further obtained, and the analysis dimensions can be divided into: steady-state data, transient data, model data, management data and environmental data.

[0102] III. Eigenvalue calculation

[0103] The eigenvalue calculation based on the preprocessed data includes: calculating historical steady-state data, real-time section data, operation data, transient recording wave data, etc. according to preset operation indexes.

[0104] The above-mentioned preset operation indexes include: distribution network operation indexes, primary equipment operation indexes, transient analysis indexes, and environmental indexes. Among them:

[0105] The distribution network operation indexes can be calculated according to real-time data, and the distribution network operation indexes include: main transformer overload rate, main transformer overload rate, main transformer light load rate, distribution transformer overload rate, distribution transformer overload rate, distribution transformer light load rate, 10kV line overload rate, 10kV line overload rate, 10kV line light load rate, main transformer 10kV bus voltage unqualified rate, substation gate voltage unqualified rate, low-voltage user voltage unqualified rate, distribution transformer three-phase imbalance ratio and distribution transformer serious three-phase imbalance ratio, etc.

[0106] The primary equipment operation indexes can be calculated according to historical data and operation data, and the primary equipment operation indexes include: switch tripping statistics, protection action statistics, morning exercise execution failure statistics, transformer temperature out-of-limit statistics, cable head temperature out-of-limit statistics, and equipment partial discharge statistics, etc.

[0107] The secondary equipment operation indexes can be calculated according to historical data and operation data, and the secondary equipment operation indexes include: long-term offline statistics, frequent on-off statistics, remote control failure statistics, data quality defect statistics, remote signal jitter, alternating current power failure, power failure, flow exceeding standard, and terminal battery defect indexes, etc.

[0108] The transient analysis indexes can be calculated according to recording wave data, and the transient analysis indexes include: parameter identification, phase current mutation, first half-wave polarity, negative sequence current calculation, zero sequence admittance calculation, zero sequence current active component, 5th harmonic component, power frequency zero sequence current amplitude and three-phase power frequency zero sequence current direction, etc.

[0109] The environmental indexes can be calculated according to environmental data, and the environmental indexes include: high temperature early warning statistics, strong wind early warning statistics, low temperature early warning statistics, lightning early warning statistics, and water inlet alarm statistics, etc.

[0110] After the characteristic value is calculated, the calculated characteristic value data is normalized for subsequent comparison and analysis. Since the characteristic value obtained after the characteristic value extraction calculation is a dimensionless floating point number, the data correlation analysis is performed before data slicing and characteristic value extraction in this embodiment, thereby reducing the difficulty of data correlation analysis.

[0111] The calculation of each type of operation index and the further classification of characteristic data under each characteristic dimension according to the operation index calculation result can refer to the prior art for specific calculation methods.

[0112] IV. Characteristic value collection

[0113] Reference Figure 3 After the characteristic value data is calculated, the characteristic value is grouped according to the data correlation relationship and the preset characteristic dimension in this embodiment, the correlation characteristic value on the same characteristic dimension forms a characteristic value data slice, that is, a data queue. The correlation characteristic value on multiple dimensions forms a characteristic value data block, and any data column on the data block can be used as a key to sort other data slices.

[0114] According to the data collection needs, some row records can be filtered out in the sorting result, and the data slice is sampled. The calculation content includes: count, average value, median value, most probable value, standard deviation, and other custom functions.

[0115] V. Snapshot generation

[0116] After the characteristic value slices and blocks on different time sections are obtained, the characteristic value block can be used as the domain of a two-dimensional table to generate a two-dimensional characteristic value snapshot table, such as Figure 3 On the snapshot table of

[0117] When the snapshot table is generated, an index dictionary is defined to provide an entry for snapshot query for analysis and application. The calculation result of the data characteristic value is stored in the database in the form of a snapshot table, that is, a two-dimensional table. According to the index dictionary, an index is established for each two-dimensional table to accelerate the alignment, arrangement, filtering and retrieval between snapshot tables.

[0118] VI. Running state analysis based on snapshot

[0119] The running data characteristic analysis based on the two-dimensional characteristic value snapshot table in this embodiment includes horizontal trend analysis and vertical correlation comparison analysis, that is, snapshot trend analysis and snapshot comparison analysis. The horizontal trend analysis is used to compare the characteristic value data change trend of the same correlation device on different time sections, and the vertical correlation comparison analysis is used to compare the characteristic value data relationship between the correlation devices with correlation relationship on the same time section.

[0120] In the horizontal trend analysis, the embodiment can normalize the feature values, and induce the floating point results of the same type of data calculated by different algorithms into two values of obvious and not obvious features (1, 0). The normalized feature values are classified and weighted to obtain a real number in a state space based on grouping. Trend analysis is performed on the obtained state aggregation results. In the trend analysis, the trends are observed according to safety trend, abnormal trend and fault trend.

[0121] With reference to Figure 3 , in the operation state analysis of the power distribution network, the safety trend is mainly observed to determine whether the trend value is limited in a certain interval and presents a gentle fluctuation. When the trend presents a large amplitude change and exceeds the limit, it indicates that there is a large batch change between the feature values associated with the trend curve. The safety state fluctuation can be further observed by the safety sub-trend curve to determine which factors such as operation, load, maintenance and communication cause the fluctuation. The abnormal trend is mainly used to observe the equipment operation state. Since the abnormal trend mainly reflects the counting statistics of the feature values, it presents a slow asymptotic development trend under normal circumstances. Whether the abnormal trend presents a slow asymptotic development trend is observed. When the slope of the trend curve obviously increases, it indicates that the counting statistics of the feature values change dramatically, indicating that the equipment operation state is abnormal. The fault trend is mainly used to observe whether the power distribution network and the power distribution primary / secondary equipment have faults. Since the fault feature is strongly related to the escape of the feature value, the fault trend presents a horizontal line under normal circumstances. Whether the fault trend presents a sharp peak fluctuation is observed. Once the sharp peak fluctuation appears, it indicates that a fault feature appears, and further fault analysis and processing can be performed.

[0122] Compared with the fault trend analysis, the safety state and abnormal state trends of the power distribution network and the power distribution equipment are not obvious, and cannot be accurately determined by the trend change of the feature values. Snapshot comparison analysis provides a supplementary basis for the analysis of the safe operation state and abnormal operation state of the power distribution network and the power distribution equipment.

[0123] The longitudinal correlation comparison analysis includes: determining a set of device objects participating in comparison, and obtaining device objects with feature value escape from the set of device objects as device objects with state abnormalities according to feature value data blocks on the same time section.

[0124] With reference to Figure 8 , the longitudinal correlation comparison analysis focuses on dividing comparison reference objects. The safe operation state snapshot comparison is mainly divided according to the device type and voltage level. The abnormal operation state snapshot comparison is mainly divided according to the regional grid limitation based on the safe operation state correlation device division. The fault operation state snapshot comparison is mainly divided according to the power supply relationship and power supply range of the upper and lower power supplies. Figure 9 and Figure 10Observe the eigenvalue escape phenomenon on the same time slice, and count the equipment that has eigenvalue escape phenomenon; observe the aggregation result of the eigenvalue escape object, and the object that meets the linear development trend can be determined as the state abnormality.

[0125] The embodiment can be applied to power grid fault abnormality, DC blocking fault, load abnormal change, major celebration activities, severe weather and other scenarios for distribution network and equipment operation state analysis.

[0126] Embodiment 2

[0127] Based on the same inventive concept as embodiment 1, this embodiment introduces an operation data feature analysis device suitable for distribution network operation state judgment, which comprises:

[0128] The to-be-analyzed data acquisition module is configured to acquire the to-be-analyzed distribution network operation data;

[0129] The preprocessing module is configured to preprocess the to-be-analyzed distribution network operation data to obtain a plurality of slice data corresponding to different analysis dimensions;

[0130] The eigenvalue calculation module is configured to calculate eigenvalues based on the slice data to obtain eigenvalue data of each analysis dimension at different time sections;

[0131] The eigenvalue slice generation module is configured to group the eigenvalue data according to a preset feature dimension to generate a plurality of eigenvalue data slices corresponding to different feature dimensions at each time section, and the slice data in different eigenvalue data slices are associated according to the association relationship between the data;

[0132] The eigenvalue slice generation module is configured to group the eigenvalue data according to a preset feature dimension to generate a plurality of eigenvalue data slices corresponding to different feature dimensions at each time section, and the slice data in different eigenvalue data slices are associated according to the association relationship between the data;

[0133] The snapshot table generation module is configured to acquire two-dimensional table tuple information determined in advance according to the operation data feature analysis target, and take the eigenvalue data slice at different time sections as the domain of the two-dimensional table to generate a two-dimensional eigenvalue snapshot table;

[0134] And the analysis module is configured to perform operation data feature analysis based on the two-dimensional eigenvalue snapshot table.

[0135] The specific implementation of each functional module is referred to the corresponding content in the method of embodiment 1, which will not be repeated.

[0136] Embodiment 3

[0137] The embodiment introduces a storage medium containing computer executable instructions, which realize the steps of the operation data feature analysis method suitable for power distribution network operation state judgment in embodiment 1 when executed by a computer process.

[0138] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0139] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of the flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device implemented in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in the flow(s) or block(s).

[0140] These computer program instructions can also be stored in a computer-readable memory that can direct the computer or other programmable data processing apparatus to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction apparatus, which realizes the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in the flow(s) or block(s).

[0141] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus, so that a series of operation steps are performed on the computer or other programmable data processing apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable data processing apparatus provide a process for realizing the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 an apparatus that performs the functions specified in the flow(s) or block(s).

[0142] The embodiments of the present application are described above with reference to the accompanying drawings, but the present application is not limited to the above-described specific embodiments, and the above-described specific embodiments are merely illustrative, but not restrictive, and a person of ordinary skill in the art can make many forms under the inspiration of the present application without departing from the purpose of the present application and the scope protected by the claims, and these all belong to the protection of the present application.

Claims

1. A method for analyzing operational data features suitable for determining the operational status of a power distribution network, characterized in that, include: Obtain the power distribution network operation data to be analyzed; The power distribution network operation data to be analyzed is preprocessed to obtain multiple slice data corresponding to different analysis dimensions; Based on the sliced ​​data, feature values ​​are calculated to obtain feature value data of each analysis dimension at different time sections; The feature value data is grouped according to a preset feature dimension to generate multiple feature value data slices corresponding to different feature dimensions at each time segment. The slice data in different feature value data slices are associated according to the correlation between the data. Feature value data slices from different feature dimensions that have a relationship are combined to generate feature value data blocks; Obtain the tuple information of the two-dimensional table based on the target of the operation data feature analysis, and use the feature value data blocks under different time sections as the fields of the two-dimensional table to generate a two-dimensional feature value snapshot table; Perform runtime data feature analysis based on the aforementioned two-dimensional feature value snapshot table.

2. The method according to claim 1, characterized in that, The data to be analyzed includes distribution network operation data, PMS operation management data, OMS system distribution network control operation data, and meteorological environment data; The analysis dimensions include at least one or more of the following: steady-state data, transient data, model data, management data, and environmental data. The feature dimensions include at least one or more of the following: basic features, security features, fault features, anomaly features, and typical scenario features; The feature value calculation based on the preprocessed data includes: calculating one or more of the following according to preset operating indicators: historical steady-state data, real-time cross-sectional data, operational data, and transient waveform data. The preset operating indicators include at least one or more of the following: distribution network operating indicators, primary equipment operating indicators, secondary equipment operating indicators, transient analysis indicators, and environmental indicators; The power distribution network operation indicators include: main transformer heavy load rate, main transformer overload rate, main transformer light load rate, distribution transformer heavy load rate, distribution transformer overload rate, distribution transformer light load rate, 10kV line heavy load rate, 10kV line overload rate, 10kV line light load rate, main transformer 10kV bus voltage non-compliance rate, transformer area gate voltage non-compliance rate, low-voltage user voltage non-compliance rate, distribution transformer three-phase imbalance ratio, and distribution transformer severe three-phase imbalance ratio. The primary equipment operation indicators include: switch tripping statistics, protection action statistics, morning exercise execution failure statistics, transformer temperature exceeding limit statistics, cable head temperature exceeding limit statistics, and equipment partial discharge statistics; The secondary equipment operation indicators include: long-term offline statistics, frequent commissioning and decommissioning statistics, remote control failure statistics, data quality defect statistics, remote signal jitter, AC power failure, power supply failure, excessive traffic and terminal battery defect indicators. The transient analysis indicators include: parameter identification, phase current mutation, first half-wave polarity, negative sequence current calculation, zero sequence admittance calculation, active component of zero sequence current, fifth harmonic component, power frequency zero sequence current amplitude, and direction of three-phase power frequency zero sequence current. The environmental indicators include: high temperature warning statistics, strong wind warning statistics, low temperature warning statistics, lightning warning statistics, and water ingress alarm statistics.

3. The method according to claim 1, characterized in that, The preprocessing of the data to be analyzed includes: removing bad data, classifying the data, and performing time alignment processing on each type of data.

4. The method according to claim 3, characterized in that, The process of removing bad data includes: for data with a data quality code, filtering out non-real-time values, non-refreshed values, invalid data values, bad data values, and operating condition exit values; for data with correlation, checking data integrity and legality, and filtering out remote signaling and telemetry mismatch values, total data mismatch values, and PQI calculation mismatch values. The data classification obtained by classifying the data includes at least: real-time data, measurement section data without time scale, fault SOE data, waveform data, steady-state data with time scale, and operational data; The time alignment processing for various types of data includes: for real-time data and measurement section data without time scales, ignoring the upload time difference, generating a time section for measurement alignment; for fault SOE data, using the time index of key data as the section, supplementing the data values ​​of other times; for waveform data, randomly selecting a waveform segment to find the abrupt change characteristic waveform, using the abrupt change characteristic waveform as the reference time for time alignment, and performing waveform data time alignment on the starting point of abrupt changes in other waveforms; for steady-state data with time scales, aligning at minute intervals with the hour as the reference line; for operational data, aligning time based on the execution time; and for transient waveform recording data, aligning time based on the waveform recording start time.

5. The method according to claim 3, characterized in that, The preprocessing of the data to be analyzed also includes: performing data correlation analysis to obtain the correlation relationship of the data on each preset feature dimension and the correlation relationship between different feature dimensions; The eigenvalue calculation based on the preprocessed data also includes: normalizing the calculated eigenvalue data.

6. The method according to claim 5, characterized in that, The data correlation information includes at least one or more of the following correlation information: topological correlation, influence range correlation, time causal correlation, voltage level correlation, upstream and downstream power supply correlation, operation interlocking correlation, and weighted correlation; The data correlation analysis includes at least one or more of the following: 1) Generate the network topology mapping relationship and record the relationship between the root node and the device object. The relationship between the root node and the device object includes the relationship between the main network outgoing switch and the distribution network root node feeder segment, the relationship between the dual power supply and the tie switch, and the relationship between the low-voltage user and the distribution transformer. 2) Mark the power supply points at each voltage level and generate the hierarchical relationship between the power supply points; 3) Based on the power supply points of each voltage level, generate the power supply range relationship, use the coloring method to color the power supply equipment, record the power supply relationship, and mark the power supply relationship between the distribution network bus and the main network bus; 4) Generate alarm event relationships, collect historical alarm records, and mark alarm signals associated with comprehensive events; 5) Time-stamp the comprehensive event aggregation results to establish the temporal relationships between events; 6) Mark the scheduling operation information and operation tasks to form task execution relationships and event handling relationships; 7) Mark the event priority, signal priority, and device priority to form a weighted relationship.

7. The method according to claim 1, characterized in that, The multiple feature value data slices are multiple sets of mutually referencing vector queues.

8. The method according to claim 1, characterized in that, The step of combining feature value data slices of different feature dimensions that have a relationship to generate feature value data blocks includes: using any feature value data slice in the data block as a key to sort the data in other feature value data slices accordingly.

9. The method according to claim 1, characterized in that, The predetermined two-dimensional table group is the associated device; The operational data feature analysis based on the two-dimensional feature value snapshot table includes horizontal trend analysis and vertical correlation comparison analysis. The horizontal trend analysis is used to compare the trend of feature value data changes of the same related equipment at different time sections, and the vertical correlation comparison analysis is used to compare the feature value data relationship between related equipment that have a correlation relationship at the same time section.

10. The method according to claim 9, characterized in that, The longitudinal correlation comparison analysis includes: determining the set of device objects to be compared; for all device objects to be compared, according to the feature value data blocks on the same time section, obtaining the device objects in the set of device objects that have feature value escape, as the device objects with abnormal states.

11. An operational data feature analysis device suitable for determining the operational status of a power distribution network, characterized in that, include: The data acquisition module is configured to acquire the power distribution network operation data to be analyzed. The preprocessing module is configured to preprocess the power distribution network operation data to be analyzed to obtain multiple slice data corresponding to different analysis dimensions. The eigenvalue calculation module is configured to perform eigenvalue calculation based on the slice data to obtain eigenvalue data of each analysis dimension at different time sections. The feature value slice generation module is configured to group the feature value data according to a preset feature dimension, generate multiple feature value data slices corresponding to different feature dimensions at each time section, and associate the slice data in different feature value data slices according to the correlation between the data. The feature value chunk generation module is configured to combine feature value data slices of different feature dimensions that have a relationship to generate feature value data chunks. The snapshot table generation module is configured to obtain two-dimensional table tuple information predetermined according to the target of running data feature analysis, and use the feature value data blocks under different time sections as the fields of the two-dimensional table to generate a two-dimensional feature value snapshot table. Additionally, an analysis module is configured to perform runtime data feature analysis based on the two-dimensional feature value snapshot table.

12. A storage medium containing computer-executable instructions, characterized in that, When the computer-executable instructions are processed and executed by a computer, they implement the steps of the operation data feature analysis method for determining the operation status of a power distribution network as described in any one of claims 1-10.

Images