Power grid operation business volume abnormity monitoring and reason tracing method, system and device and medium

Abstract

The invention relates to the technical field of power grid operation business volume abnormity monitoring and reason tracing, and discloses a power grid operation business volume abnormity monitoring and reason tracing method, system and device and a medium, and the method comprises the steps: obtaining target power grid operation business volume statistical data, and carrying out the preprocessing; presetting a target statistical magnitude and solving a control limit, and calculating a real-time numerical value according to the preprocessed data; if the real-time numerical value exceeds the control limit, determining that the operation business volume is abnormal; calculating contribution values and control limits of different variables to the target statistics; when the operation business volume is abnormal, calculating contribution values of different variables to the overrun statistics and comparing the contribution values with the control limit; and finally tracing reasons according to a comparison result. The method improves the accuracy and efficiency of power grid operation business volume abnormity monitoring, realizes rapid tracing of abnormity reasons, and provides guarantee for safe and stable operation of a power grid. By presetting the target statistics and solving the control limit, whether the business volume is abnormal or not can be judged, and subjectivity and uncertainty of manual threshold setting in a traditional method are avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power grid operation business volume anomaly monitoring and cause tracing, and particularly relates to a power grid operation business volume anomaly monitoring and cause tracing method, system, device and medium. BACKGROUND

[0002] With the continuous expansion of the power grid scale and the continuous improvement of the intelligent level, power grid operations are becoming increasingly complex and variable. Power grid operations not only include daily maintenance, repair, patrol and other production operations, but also cover construction, marketing, planning and other management operations, and involve devices at multiple levels such as substations, transmission lines and distribution networks. The normal fluctuations of these operation data reflect the stability of the power grid operation, while abnormal fluctuations may indicate potential safety hazards, management loopholes or external interference. By monitoring the abnormal changes in business volume, abnormal events in power grid operation can be detected in a timely manner, such as device failure, management disorder or human error, so that measures can be taken to prevent accidents from expanding. In addition, identifying and tracing the causes of abnormal operation volume can help optimize resource scheduling and operation arrangement, improve operation efficiency, and enhance the intelligent level of power grid operation. At the same time, by analyzing the multi-dimensional characteristics of operation volume, scientific data support can be provided for power grid managers to improve the accuracy and foresight of management decisions. Therefore, how to build a multivariate anomaly detection and cause tracing method suitable for power grid operation business volume monitoring has become a key problem that needs to be solved in the development of smart grid.

[0003] Power grid operation data comes from multiple business systems such as production management system, device management system, construction management system, etc. When analyzing, not only the total amount of operation business needs to be analyzed, but also the changes in the number of different types of operation business such as production, construction, etc. and their internal relationship need to be considered. In addition, operation volume is affected by factors such as season, weather, load change, holidays, etc., showing obvious time-varying characteristics. This makes it difficult for traditional methods based on a single indicator or simple threshold judgment to effectively cope with it. Therefore, it is necessary to introduce advanced statistical modeling methods that can handle multivariate, non-Gaussian data to achieve accurate monitoring and anomaly identification of power grid operation business volume.

[0004] There has been some research foundation in multivariate anomaly detection at home and abroad, but in the field of power grid operation business volume monitoring, especially in multi-dimensional modeling and cause tracing, there is still a big research gap. The existing methods have the following main limitations in power grid operation business volume monitoring: (1) strong dependence on single indicators, only focusing on single indicators such as operation quantity or operation time, ignoring multi-dimensional factors such as operation type; (2) simple threshold judgment method is difficult to adapt to the dynamic changes of operation volume, and is prone to false positives or false negatives; (3) lack of in-depth analysis and tracing of the causes of abnormality, which cannot provide effective support for operation decision-making. SUMMARY

[0005] In view of the above existing problems, the present application is proposed.

[0006] Therefore, the present application provides a power grid operation business volume anomaly monitoring and reason tracing method, system, device and medium, which can solve the problems of multi-dimensional data processing and dynamic change adaptability in power grid operation business volume anomaly monitoring, and improve the accuracy of anomaly detection and the effectiveness of reason tracing.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, the present application provides a power grid operation business volume anomaly monitoring and reason tracing method, comprising:

[0009] Obtaining target power grid operation business volume statistical data, and performing first preprocessing on the statistical data;

[0010] The first preprocessing includes non-Gaussian feature extraction and Gaussian feature extraction;

[0011] A target statistical quantity is preset, a control limit for the target statistical quantity is solved, and a real-time value for the target statistical quantity is calculated according to the statistical data after the first preprocessing;

[0012] If the real-time value exceeds the control limit of the target statistical quantity, it is determined that the operation business volume is abnormal;

[0013] The contribution value of different variables to the target statistical quantity is calculated, and a contribution value control limit is calculated;

[0014] When the operation business volume is abnormal, the contribution value of different variables to the over-limit statistical quantity is calculated, and the contribution value of different variables to the over-limit statistical quantity is compared with the contribution value control limit;

[0015] Reason tracing is performed according to the comparison result.

[0016] As a preferred scheme of the power grid operation business volume anomaly monitoring and reason tracing method, wherein: the reason tracing according to the comparison result comprises:

[0017] If the contribution value of different variables to the over-limit statistical quantity is not less than the contribution value control limit, the corresponding variable is considered as an abnormal variable;

[0018] The over-limit rate of the corresponding variable is calculated, and the high and low of the abnormal level is determined according to the size of the over-limit rate;

[0019] The over-limit statistical quantity is the statistical quantity that produces over-limit.

[0020] As a preferred scheme of the power grid operation business volume abnormality monitoring and cause tracing method, the target statistics include independent principal component space statistics, principal component space statistics and residual space statistics.

[0021] If the real-time value of any target statistics exceeds the control limit of the target statistics, it is determined that the operation business volume is abnormal.

[0022] This preferred scheme can more comprehensively reflect the overall state of the power grid operation business volume, and improve the accuracy and sensitivity of abnormality monitoring. The independent principal component space statistics, the principal component space statistics and the residual space statistics respectively describe the characteristics of the power grid operation business volume from different angles. The independent principal component space statistics reflects the main trend of the data, the principal component space statistics reveals the main components of the data, and the residual space statistics captures the part of the data that cannot be explained by the principal components. When any of the three statistics is abnormal, it means that there may be a problem with the power grid operation business volume, thereby triggering the abnormality monitoring mechanism in time and providing strong support for subsequent cause tracing.

[0023] As a preferred scheme of the power grid operation business volume abnormality monitoring and cause tracing method, the calculation of the over-limit rate of the corresponding variable includes:

[0024] The contribution value of the corresponding variable to the over-limit statistics at the current time is calculated.

[0025] The contribution value control limit of the corresponding variable to the over-limit statistics is obtained.

[0026] The over-limit rate of the corresponding variable is calculated according to the contribution value of the corresponding variable to the over-limit statistics at the current time and the contribution value control limit of the corresponding variable to the over-limit statistics.

[0027] As a preferred scheme of the power grid operation business volume abnormality monitoring and cause tracing method, the control limit of the target statistics and the contribution value control limit are both calculated by the kernel density estimation method.

[0028] As a preferred scheme of the power grid operation business volume abnormality monitoring and cause tracing method, the non-higher order feature extraction and the higher order feature extraction include:

[0029] The non-higher order feature extraction is used to obtain a decomposition matrix, a mixing matrix, independent components and a residual space from the statistical data.

[0030] The higher order feature extraction is used to perform principal component decomposition on the residual space, and obtain a principal component matrix, a loading matrix and a residual matrix.

[0031] As a preferred solution of the method for abnormal monitoring and cause tracing of power grid operation traffic of the present invention, the first preprocessing of the statistical data includes:

[0032] The first preprocessing further includes performing data centering and dimension removal on the statistical data and then performing non-Gaussian feature extraction and Gaussian feature extraction in sequence;

[0033] The features extracted from the non-Gaussian features and the Gaussian features constitute the different variables.

[0034] In a second aspect, the present invention provides a system for monitoring abnormal power grid operation traffic and tracing its causes, comprising:

[0035] A preprocessing module, configured to obtain target power grid operation traffic statistics and perform a first preprocessing on the statistics;

[0036] The first preprocessing includes non-Gaussian feature extraction and Gaussian feature extraction;

[0037] a limit value acquisition module, configured to preset a target statistic, solve a control limit for the target statistic, and calculate a real-time value for the target statistic based on the statistical data after the first preprocessing;

[0038] A judgment module, configured to determine that an abnormality exists in the operation volume if the real-time value exceeds the control limit of the target statistic;

[0039] A calculation module, used to calculate the contribution values ​​of different variables to the target statistic and the contribution value control limits;

[0040] A judgment module is used to calculate the contribution values ​​of different variables to the out-of-limit statistics when there is an abnormality in the operation business volume, and compare the contribution values ​​of different variables to the out-of-limit statistics with the contribution value control limits;

[0041] The tracing module is used to trace the cause based on the comparison results.

[0042] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method described above when executing the computer program.

[0043] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method described above when the computer program is executed by a processor.

[0044] Compared with the prior art, the present application has the beneficial effects that: the present application provides a power grid operation business volume anomaly monitoring and reason tracing method, obtains target power grid operation business volume statistical data, and performs first preprocessing on the statistical data; a target statistical quantity is preset, a control limit for the target statistical quantity is solved, and a real-time value for the target statistical quantity is calculated according to the first preprocessed statistical data; if the real-time value exceeds the control limit of the target statistical quantity, it is determined that the operation business volume is abnormal; the contribution value of different variables to the target statistical quantity and the contribution value control limit are calculated; when the operation business volume is abnormal, the contribution value of different variables to the over-limit statistical quantity is calculated, and the contribution value of different variables to the over-limit statistical quantity is compared with the contribution value control limit; and the reason is traced according to the comparison result. The method not only improves the accuracy and efficiency of the power grid operation business volume anomaly monitoring, but also realizes the rapid tracing of the abnormal reason, thereby providing a strong guarantee for the safe and stable operation of the power grid. By presetting the target statistical quantity and solving the control limit, it can be more scientifically judged whether the operation business volume is abnormal, and the subjectivity and uncertainty of the threshold value set by the traditional method are avoided. At the same time, the contribution value of different variables to the target statistical quantity and the control limit are calculated, the reason for the abnormality is further analyzed, and targeted measures are taken for improvement and optimization. In addition, the method is also applicable to different types of power grid operation business volume statistical data, and has strong universality and practicality. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0046] Figure 1 The method flow chart of the power grid operation business volume anomaly monitoring and reason tracing method provided by an embodiment of the present application.

[0047] Figure 2 The modeling data independent component schematic diagram of the power grid operation business volume anomaly monitoring and reason tracing method provided by an embodiment of the present application.

[0048] Figure 3 The modeling data principal component schematic diagram of the power grid operation business volume anomaly monitoring and reason tracing method provided by an embodiment of the present application.

[0049] Figure 4 The statistical quantity offline monitoring diagram of the power grid operation business volume anomaly monitoring and reason tracing method provided by an embodiment of the present application.

[0050] Figure 5A control limit diagram of variable contribution to statistical quantity of an electric grid operation business volume abnormality monitoring and cause tracing method provided for an embodiment of the present application.

[0051] Figure 6 A test data monitoring statistical quantity diagram of an electric grid operation business volume abnormality monitoring and cause tracing method provided for an embodiment of the present application.

[0052] Figure 7 A diagram of a 79th test sample variable contribution to T2 exceeding limit rate of an electric grid operation business volume abnormality monitoring and cause tracing method provided for an embodiment of the present application.

[0053] Figure 8 An internal structure diagram of an electronic device of an electric grid operation business volume abnormality monitoring and cause tracing method provided for an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the above objectives, characteristics and advantages of the present application more apparent, obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should fall within the scope of protection of the present application.

[0055] Embodiment 1, with reference to Figures 1-8 For the first embodiment of the present application, the embodiment provides an electric grid operation business volume abnormality monitoring and cause tracing method, comprising:

[0056] In the prior related art, there are some problems, for example, for the abnormal monitoring of the electric grid operation business volume, the traditional method often relies on a single index or a simple threshold judgment, which is difficult to adapt to the multidimensionality and dynamic change characteristics of the operation volume.

[0057] The present application provides a method that can effectively solve the above-mentioned problems. Next, how to implement the electric grid operation business volume abnormality monitoring and cause tracing method will be described in detail in combination with multiple embodiments.

[0058] Figure 1 A method flowchart of an electric grid operation business volume abnormality monitoring and cause tracing method is shown, comprising:

[0059] S101, obtaining target electric grid operation business volume statistical data, and performing first preprocessing on the statistical data;

[0060] It should be noted that in order to realize the power grid operation business volume anomaly monitoring and reason tracing, the operation business volume statistical data of the target power grid in a certain time range needs to be obtained, which may come from production management system, equipment management system, capital construction management system and other business systems. These data contain various information of power grid operation, such as operation type, operation time, operation quantity, etc.

[0061] It should also be noted that the obtained statistical data is relatively raw and complex, and may contain noise, redundant information and data of different dimensions, so the data needs to be preprocessed for subsequent analysis and processing.

[0062] In an optional embodiment, the target power grid operation business volume statistical data can be obtained by interfacing with various business systems of the power grid enterprise, and the relevant data can be obtained in real time or at regular intervals. These data may include but are not limited to the type, quantity, execution time, execution personnel, related equipment information and environmental factors of the operation task.

[0063] In an optional embodiment, the target power grid operation business volume statistical data can also be obtained by manual input, file import and other methods. For example, the power grid enterprise can export the data of various business systems into Excel or CSV format files, and then import these data for processing and analysis by the method or system provided by the present application.

[0064] In the embodiment of the present application, 638 pieces of power grid business volume data of a power grid company in 202X are obtained, each piece of data containing 8 variables, including: total number of control plans, production items, capital construction items, marketing items, high-risk items, medium-risk items, low-risk items and acceptable risk items.

[0065] In an optional embodiment, the first preprocessing of the statistical data can include but is not limited to data cleaning, data conversion and data normalization. Data cleaning mainly removes duplicate data, handles missing values and outliers, etc., to ensure the accuracy and integrity of the data. Data conversion is to convert the original data into a form suitable for subsequent analysis and processing, such as converting categorical data into numerical data. Data normalization is to scale the data according to a certain proportion so that it falls within a small specific interval, so as to eliminate the influence of different dimensions on the data and improve the convergence speed and accuracy of the algorithm. In the embodiment of the present application, the first preprocessing of the obtained 638 pieces of power grid business volume data includes filling missing values, correcting outliers, and normalizing the data to 0-1, laying a foundation for subsequent analysis and processing.

[0066] However, in order to accurately realize the power grid operation business volume anomaly monitoring and reason tracing, more professional preprocessing steps are needed, so the present application designs the following first preprocessing method.

[0067] In the embodiments of the present application, the first preprocessing comprises non-Gaussian feature extraction and Gaussian feature extraction.

[0068] In the embodiments of the present application, the non-Gaussian feature extraction and the Gaussian feature extraction comprise:

[0069] The non-Gaussian feature extraction is used to obtain a decomposition matrix, a mixing matrix, independent components and a residual space from the statistical data.

[0070] The Gaussian feature extraction is used to perform a principal component decomposition on the residual space, and obtain a principal component matrix, a loading matrix and a residual matrix.

[0071] In the embodiments of the present application, the first preprocessing of the statistical data comprises:

[0072] The first preprocessing further comprises centering and de-dimensioning the statistical data, and then performing the non-Gaussian feature extraction and the Gaussian feature extraction in sequence.

[0073] The features after the non-Gaussian feature extraction and the Gaussian feature extraction constitute different variables.

[0074] For example, for the power grid operation business data obtained from a power grid company (510 is the number of data samples, and 8 is the number of variables) is standardized, that is, it is centered and the influence of different dimensions is removed:

[0075]

[0076] wherein, (i = 1, 2,..., 510; j = 1, 2,..., 8) represents the value of the jth variable at the ith sample; and respectively represent the mean and variance of the jth variable (j = 1, 2,..., 8); X i,j is the value of the jth variable at the ith sample after standardization, which is also the value of the jth variable at the ith sample in the standardized data matrix X (510 x 8).

[0077] Further, the non-Gaussian feature extraction is performed: the present application first performs whitening processing on the standardized data, and then extracts independent components by using a FASTICA algorithm.

[0078] The specific steps comprise: performing whitening processing on the standardized data X (510 x 8):

[0079] Z = Λ -1 / 2 U T X = QX (2)

[0080] wherein the obtaining is the whitened data, Q = Λ -1 / 2 U T is the whitening matrix, U is an orthogonal eigenvector matrix, Λ is a corresponding eigenvalue diagonal matrix, U and Λ are obtained by eigenvalue decomposition of the covariance matrix E(XX T ), and p is the number of eigenvalues greater than zero in the diagonal of Λ. In the embodiment, p is 8.

[0081] In an alternative embodiment, the whitening matrix Z is subjected to independent component extraction using the FASTICA algorithm proposed by Hyvarine, a Finnish scholar:

[0082]

[0083] The whitening matrix can be expressed as

[0084] Z T = AS + F T (4)

[0085] wherein S represents an independent principal component matrix; and 1 is the number of retained independent principal components, which is determined by checking the Gaussianity of the residual space F; is a decomposition matrix, which is solved iteratively according to the non-Gaussian maximization principle. is a mixing matrix, which is the inverse form of the decomposition matrix W; is a residual matrix and is subject to Gaussian distribution. In the embodiment, 2 independent principal components are retained, i.e. 1 = 2, and the independent principal components are as shown in Figure 2 .

[0086] Further, Gaussian feature extraction is performed: after the independent components S of the work load data are extracted, the residual information subject to Gaussian distribution in the data needs to be analyzed using the PCA model, i.e. principal component decomposition is performed on the residual matrix F:

[0087] F = TP T + E (5)

[0088] wherein and are the score matrix and the loading matrix, respectively; is the residual information after model explanation; and k is the number of retained principal components, which is generally determined by the cumulative percentage of variance (CPV), i.e. by calculating the cumulative percentage of variance of the first k principal components.

[0089] In the embodiment, the CPV value is 0.85, the number of retained principal components k is 2, and the principal components are as shown in Figure 3 .

[0090] It should be noted that the target power grid operation business volume statistical data is acquired, and the statistical data is first preprocessed, and the data after preprocessing can more accurately reflect the actual situation of the power grid operation business volume, and provides a reliable basis for subsequent anomaly monitoring and reason tracing. Meanwhile, through non-high dimensional feature extraction and high dimensional feature extraction, potential information in the data can be more effectively mined, and the accuracy of anomaly monitoring and the efficiency of reason tracing are improved.

[0091] In S102, a target statistical quantity is preset, a control limit for the target statistical quantity is solved, and a real-time value for the target statistical quantity is calculated according to the statistical data after the first preprocessing;

[0092] It should be noted that the target statistical quantity is a key indicator for measuring whether the power grid operation business volume is abnormal. The preset target statistical quantity is obtained based on in-depth understanding of the power grid operation business volume and statistical analysis of historical data. These statistical quantities can reflect some key characteristics of the power grid operation business volume, such as volatility, trend, periodicity, etc. When the target statistical quantity is preset, the actual situation of the power grid operation business volume and the monitoring requirements need to be considered to ensure that the selected statistical quantity can accurately reflect the abnormal state of the operation business volume.

[0093] In an optional embodiment, solving the control limit for the target statistical quantity is an important link of anomaly monitoring. The control limit is calculated according to historical data, and is a key threshold for judging whether the current statistical quantity is within a normal range. When the real-time value of the statistical quantity exceeds the control limit, it is considered that the power grid operation business volume is abnormal. There are many methods for solving the control limit, such as control limit calculation based on 3σ principle, control limit calculation based on kernel density estimation, etc.

[0094] In an optional embodiment, when the control limit calculation based on 3σ principle is selected, the mean and standard deviation of the target statistical quantity are first calculated, and then the control limit is determined according to the 3σ principle.

[0095] Specifically, the control limit can be set to the mean plus or minus 3 times the standard deviation, that is, the upper control limit is the mean plus 3 times the standard deviation, and the lower control limit is the mean minus 3 times the standard deviation. When the real-time value of the statistical quantity exceeds the upper control limit or is lower than the lower control limit, it is determined to be abnormal. This method is simple and intuitive, and is suitable for anomaly monitoring in most cases. However, it should be noted that the control limit calculation based on 3σ principle assumes that the data follows a normal distribution, and if the data distribution deviates from the normal distribution, it may affect the accuracy of the monitoring result. Therefore, in actual application, a suitable control limit calculation method needs to be selected according to the distribution of the data.

[0096] In an optional embodiment, when the control limit calculation based on kernel density estimation is selected, kernel density estimation needs to be performed on the historical data of the target statistical quantity first to obtain its probability density function. Kernel density estimation is a non-parametric estimation method, which does not need to assume that the data is subject to a certain distribution in advance, but estimates the probability density of the data through a kernel function. After obtaining the probability density function, the control limit can be determined according to the required confidence level.

[0097] Specifically, a suitable confidence level (such as 95% or 99%) can be selected, and then the corresponding quantile is found as the control limit. When the real-time value of the statistical quantity exceeds the control limit, it is determined to be abnormal. This method can more flexibly adapt to different data distribution conditions and improve the accuracy of abnormal monitoring.

[0098] In the embodiment of the present application, the control limit calculation method based on kernel density estimation is used to perform abnormal monitoring on the preprocessed power grid operation business volume data.

[0099] In the embodiment of the present application, if the real-time value exceeds the control limit of the target statistical quantity, it is determined that the operation business volume is abnormal.

[0100] In an optional embodiment, the selection of the target statistical quantity is a complex and key process. First, it needs to be clear what needs to be monitored, that is, which aspects of the power grid operation business volume need to be concerned. This usually includes key indicators such as the number, type, and time distribution of operations. Secondly, according to the historical data and business logic, statistical quantities that can accurately reflect these targets need to be selected. These statistical quantities should be representative, sensitive, and stable, and can timely discover abnormal changes in business volume and trace the causes of abnormalities.

[0101] In the embodiment of the present application, the target statistical quantity includes an independent principal component space statistical quantity, a principal component space statistical quantity, and a residual space statistical quantity.

[0102] If the real-time value of any target statistical quantity exceeds the control limit of the target statistical quantity, it is determined that the operation business volume is abnormal.

[0103] For example, after processing by the ICA and PCA models, ICA and PCA are two commonly used data analysis and dimensionality reduction techniques. ICA (Independent Component Analysis) is mainly used to extract non-Gaussian features in data, that is, independent components in data, which can reflect the essential structure of data. PCA (Principal Component Analysis) is used to extract Gaussian features in data, that is, the main trend of data, which can reveal the overall distribution and variation law of data.

[0104] The present application monitors the business volume data from three spaces, namely the independent principal component space, the principal component space, and the residual space, and the corresponding three statistical quantities are S2 statistic, T 2 statistic and SPE statistic.

[0105] S constructed by independent component space 2 The statistic is defined as:

[0106] S 2 = S T S (6)

[0107] T constructed by principal component space 2 The statistic is defined as:

[0108] T 2 = T T Ω -1 T (7)

[0109] In the formula, Ω = diag (λ1, λ2) is a diagonal matrix composed of the first two eigenvalues of the covariance matrix COV (F) of F.

[0110] The definition of the SPE statistic is as follows:

[0111] SPE = EE T (8)

[0112] After obtaining the three statistics, the present application obtains the density distribution of the statistics by using the kernel density estimation method. And given the quantile α, the upper α quantile corresponding to the control limit.

[0113] In the embodiment, the quantile is taken as 0.05, S 2 statistic, T 2 statistic and SPE statistic control limit is recorded as SPE lim = 0.69.

[0114] It should be noted that the calculation of the real-time value of the target statistic according to the first pre-processed statistical data is a continuous process, and the power grid operation business volume data needs to be tracked and analyzed in real time. In the embodiment of the present application, a fixed time interval (such as every hour, every day or every week) is set to calculate the real-time value of the target statistic regularly, and compared with the preset control limit. If the real-time value exceeds the control limit, an abnormal alarm is triggered to prompt that the power grid operation business volume is abnormal.

[0115] It also needs to be explained that the preset target statistical quantity, the control limit for the target statistical quantity is solved, and the real-time value for the target statistical quantity calculated according to the first preprocessed statistical data can realize real-time monitoring and abnormal early warning of the power grid operation business volume. By setting a reasonable target statistical quantity and control limit, abnormal changes in the operation business volume can be found in time, providing important clues and basis for subsequent cause tracing. At the same time, this real-time monitoring method can also improve the efficiency and accuracy of monitoring, avoiding misjudgment and missed judgment due to the lag and subjectivity of manual monitoring. On this basis, the application can further analyze and trace the causes of the abnormality, so as to find the root cause of the abnormal business volume and provide strong support for the management and optimization of power grid operation.

[0116] S103, calculating the contribution value of different variables to the target statistical quantity and the contribution value control limit;

[0117] In an optional embodiment, the contribution value of different variables to the target statistical quantity and the contribution value control limit are used to measure the degree of influence of each variable on the abnormality of the target statistical quantity. The calculation of the contribution value can help the application to identify which variables are the key factors leading to the abnormality of the business volume.

[0118] In an optional embodiment, the variable projection-based method can be used when calculating the contribution value. Specifically, each variable is projected onto the direction of the target statistical quantity, and the length of the projection represents the contribution value of the variable to the target statistical quantity. The greater the contribution value, the greater the influence of the variable on the target statistical quantity, and the more likely it is to be a key factor leading to the abnormality of the business volume. At the same time, the contribution value control limit needs to be set to determine whether the contribution value is abnormal. When the contribution value of a variable exceeds the contribution value control limit, it can be considered that the variable is one of the main reasons leading to the abnormality of the business volume.

[0119] In an optional embodiment, the contribution value calculation method based on reconstruction error can also be used. Specifically, first, the original data is reconstructed into data in the independent component space, the principal component space and the residual space according to the ICA and PCA models. Then, for each variable, the value of the variable in the reconstructed data is set to zero to obtain new reconstructed data. Next, the difference between the new reconstructed data and the original data, i.e. the reconstruction error, is calculated. The greater the reconstruction error, the greater the influence of the variable on the target statistical quantity. Finally, the contribution value of each variable to the target statistical quantity is determined according to the size of the reconstruction error.

[0120] In an optional embodiment, after determining the contribution value, the contribution value control limit needs to be set. The setting of the contribution value control limit can be adjusted according to historical data and business requirements. When the contribution value of a variable exceeds the contribution value control limit, it can be considered that the variable is one of the main reasons leading to the abnormality of the business volume, and needs to be further analyzed and traced.

[0121] The above method of calculating the contribution value and the contribution value control limit cannot directly reflect the specific influence path of each variable on the traffic anomaly. In order to more deeply understand the cause of the anomaly, the present application proposes the following way of calculating the contribution value and the contribution value control limit.

[0122] In the embodiment of the present application, when there is an anomaly in the job traffic, the contribution value of different variables to the over-limit statistics is calculated, and the contribution value of different variables to the over-limit statistics is compared with the contribution value control limit;

[0123] In the embodiment of the present application, the control limit of the target statistics and the contribution value control limit are both calculated by the kernel density estimation method.

[0124] Specifically, the contribution of each variable in each sample to the three statistics control limits

[0125] The contribution of each variable to S 2 :

[0126] C S2con,ij = s ij s ij T (9)

[0127] The contribution of each variable to T 2 :

[0128]

[0129] The contribution of each variable to SPE:

[0130]

[0131] In the formula, C SPE,ij respectively represent the contribution of the jth (j = 1, 2, …, 8) variable of the ith (i = 1, 2, …, 510) sample to the S 2 , T2, and SPE statistics; s ij , t ij , and e ij respectively represent the jth value of the ith independent vector s i , the jth value of the ith principal component vector t i , and the jth value of the residual vector e i .

[0132] The KDE estimation method is used to calculate the control limit of the contribution value of each variable, respectively denoted as LmtC SPEcon,j (j = 1, 2, 3, …, 8).

[0133] In the embodiment, the control limits of the three statistical contribution values are as shown in Figure 4

[0134] It should be noted that calculating the contribution values of different variables to the target statistical quantity and the contribution value control limits can more accurately locate the key factors causing the abnormality of the power grid operation business volume. By comparing the contribution values of each variable to the target statistical quantity with the control limits, variables that have a significant impact on the abnormality of the business volume can be screened out, thereby narrowing the scope of problem troubleshooting. This not only improves the efficiency of abnormal cause tracing, but also provides targeted suggestions for subsequent fault repair and management optimization. For example, if it is found that a certain specific variable frequently exceeds its contribution value control limit, then the power grid operation link related to the variable can be focused on, and whether there are problems such as equipment failure, improper operation, or process defects can be checked. In this way, the problem source can be more accurately located, and effective measures can be taken to solve the problem, thereby ensuring the stability and efficiency of the power grid operation.

[0135] S104, performing cause tracing according to the comparison result.

[0136] In the embodiment of the present application, performing cause tracing according to the comparison result comprises:

[0137] If the contribution value of the different variable to the out-of-limit statistical quantity is not less than the contribution value control limit, the corresponding variable is considered to be an abnormal variable;

[0138] The out-of-limit rate of the corresponding variable is calculated, and the high and low of the abnormal level is determined according to the size of the out-of-limit rate;

[0139] The out-of-limit statistical quantity is a statistical quantity that generates an out-of-limit.

[0140] In the embodiment of the present application, calculating the out-of-limit rate of the corresponding variable comprises:

[0141] The contribution value of the corresponding variable to the statistical quantity that generates an out-of-limit at the current time is calculated;

[0142] The contribution value control limit of the corresponding variable to the statistical quantity that generates an out-of-limit is obtained;

[0143] The out-of-limit rate of the corresponding variable is calculated according to the contribution value of the corresponding variable to the statistical quantity that generates an out-of-limit at the current time and the contribution value control limit of the corresponding variable to the statistical quantity that generates an out-of-limit.

[0144] ​In summary, the present application provides a power grid operation business volume anomaly monitoring and reason tracing method, obtains target power grid operation business volume statistical data, and performs first preprocessing on the statistical data; a target statistical quantity is preset, a control limit for the target statistical quantity is solved, a real-time value for the target statistical quantity is calculated according to the first preprocessed statistical data; if the real-time value exceeds the control limit of the target statistical quantity, it is determined that the operation business volume is abnormal; the contribution value of different variables to the target statistical quantity and the contribution value control limit are calculated; when the operation business volume is abnormal, the contribution value of different variables to the over-limit statistical quantity is calculated, and the contribution value of different variables to the over-limit statistical quantity is compared with the contribution value control limit; the reason is traced according to the comparison result. The method not only improves the accuracy and efficiency of the power grid operation business volume anomaly monitoring, but also realizes the rapid tracing of the abnormal reason, and provides a strong guarantee for the safe and stable operation of the power grid. By presetting the target statistical quantity and solving the control limit, it can be more scientifically judged whether the operation business volume is abnormal, and the subjectivity and uncertainty of the threshold value set by the traditional method are avoided. At the same time, the contribution value of different variables to the target statistical quantity and the control limit are calculated, which can further analyze the reason of the abnormality, so that targeted measures can be taken for improvement and optimization. In addition, the method is also applicable to different types of power grid operation business volume statistical data, and has strong universality and practicality.

[0145] In one preferred embodiment, 128 business volume samples are selected as the test set. According to the offline model, the principal component score matrix T, the load matrix P, T 2 The statistical quantity and the SPE statistical quantity and the control limit of the contribution are monitored.

[0146] The tth test data of the power grid operation business volume After that, the jth variable is standardized according to the obtained mean and variance:

[0147]

[0148] Wherein, and respectively represent the mean and variance of the jth variable (j=1, 2,..., m).

[0149] According to the obtained whitening transformation matrix Q, the whitening vector of the current power grid operation business volume is obtained:

[0150]

[0151] According to the obtained decomposition matrix W, the independent principal component representation of the current power grid operation business volume is obtained:

[0152]

[0153] According to the mixed matrix A in the embodiment, the data representation of the current grid operation service volume is obtained:

[0154]

[0155] Wherein, f t is the residual error vector after non-Gaussian information extraction.

[0156] According to the load matrix P, the principal component representation of the Gaussian information in the current grid operation service volume data is obtained:

[0157] t t = f t P (16)

[0158] The statistical quantity index of the current grid operation service volume data is calculated:

[0159] Real-time S 2 statistical quantity:

[0160]

[0161] Real-time T 2 statistical quantity:

[0162]

[0163] Real-time SPE statistical quantity:

[0164]

[0165] In the formula, T t 2 and SPE t are the S 2 , T 2 and SPE statistical quantity values calculated at the current time; s t is the independent component at the current time; t t is the principal component score at the current time; is the load matrix; e t = f t -f t PP T is the residual error information after Gaussian information extraction.

[0166] In the embodiment, the monitoring statistical quantity of the test data is as shown in Table 1. Figure 6 Through comparison with the label, it is found that the accuracy rate of this abnormality detection is 93%.

[0167] If there is a statistical quantity over limit, it can be considered that the power grid operation business volume has an abnormal situation, at this time the variable contribution chart is used to diagnose the reason. The contribution value of each variable to the over limit statistical quantity is calculated, compared with the control limit of the corresponding contribution, the variable exceeding the control limit is considered as the abnormal variable, and its over limit rate is calculated, and the abnormal level is determined according to the size of the over limit rate.

[0168] Determine whether the real-time statistical quantity exceeds the control limit, if it exceeds the control limit, it means that the business volume data is abnormal, and the contribution of each variable to the statistical quantity is calculated:

[0169] The contribution of each variable to S 2 Real-time contribution:

[0170]

[0171] The contribution of each variable to T 2 Real-time contribution:

[0172]

[0173] Real-time contribution of each variable to SPE

[0174]

[0175] In the formula, C SPE,tj respectively represent the contribution of the jth variable of the current sample to S 2 , T 2 , and SPE statistical quantity; s tj , t tj , and e tj respectively represent the jth value of the independent vector s t , the principal element vector t t , and the residual error vector e t .

[0176] The control limit of the contribution value of each variable to the statistical quantity is shown in Figure 6 When the real-time statistical quantity exceeds the control limit, the contribution of each process variable to the over limit statistical quantity at the current time is calculated, and by comparing with the control limit of the corresponding contribution, the process variable exceeding the control limit can be determined, which is considered as the fault variable, and its over limit rate is calculated.

[0177] In this embodiment, the 79th test sample is taken as an example, the S2 statistical quantity of the sample is 5.97, which is less than the control limit 21.79. The T2 statistical quantity is 7.70, which exceeds the control limit 7.45; the SPE statistical quantity is 1.37, which exceeds the control limit 0.69. Therefore, the over limit rate of each variable of the sample to T2 and SPE statistical quantity is calculated according to formula (23). Taking T2 statistical quantity as an example, as shown in Figure 7As shown, the 2nd and 4th variables have a greater contribution to T2 and exceed the control limit, indicating that the abnormal variables are the production item and the marketing item. However, the two variables have a small over-limit rate, indicating that the abnormal level is low.

[0178] In this embodiment, an electric power grid operation service volume anomaly monitoring and cause tracing system is also provided, comprising:

[0179] A preprocessing module is configured to acquire target electric power grid operation service volume statistical data and perform first preprocessing on the statistical data.

[0180] The first preprocessing includes non-Gaussian feature extraction and Gaussian feature extraction.

[0181] A limit value acquisition module is configured to preset a target statistical quantity, solve a control limit for the target statistical quantity, and calculate a real-time value for the target statistical quantity according to the first-processed statistical data.

[0182] A judgment module is configured to determine that the operation service volume is abnormal if the real-time value exceeds the control limit of the target statistical quantity.

[0183] A calculation module is configured to calculate a contribution value of different variables for the target statistical quantity and a contribution value control limit.

[0184] The judgment module is configured to, when the operation service volume is abnormal, calculate a contribution value of different variables for the over-limit statistical quantity, and compare the contribution value of different variables for the over-limit statistical quantity with the contribution value control limit.

[0185] A tracing module is configured to perform cause tracing according to a comparison result.

[0186] The above-mentioned unit modules can be embedded in or independent of a processor in an electronic device in a hardware form, or can be stored in a memory in the electronic device in a software form, so as to be called and executed by the processor to perform operations corresponding to the above-mentioned modules.

[0187] The electronic device can be a terminal, and an internal structure diagram of the electronic device can be as shown in FIG. 6. Figure 8As shown. The electronic device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner. The wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for monitoring abnormal power grid operation business volume and tracing the cause is implemented. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen. The input device of the electronic device can be a touch layer covering the display screen, or a key, trackball or touchpad provided on the electronic device housing, or an external keyboard, touchpad or mouse.

[0188] This embodiment further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the following steps are implemented:

[0189] Obtaining target power grid operation business volume statistical data and performing a first preprocessing on the statistical data;

[0190] The first preprocessing includes non-Gaussian feature extraction and Gaussian feature extraction;

[0191] Presetting a target statistic, solving a control limit for the target statistic, and calculating a real-time value for the target statistic based on the statistical data after the first preprocessing;

[0192] If the real-time value exceeds the control limit of the target statistic, it is determined that the operation volume is abnormal;

[0193] Calculate the contribution values ​​of different variables to the target statistic and the control limits of the contribution values;

[0194] When there is an abnormality in the operation business volume, the contribution values ​​of different variables to the out-of-limit statistics are calculated and compared with the contribution value control limits;

[0195] Trace the cause based on the comparison results.

[0196] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0197] Those skilled in the art will appreciate that embodiments of the present application can be readily used as software, hardware, or a combination of software and hardware. In one embodiment, the present application can be implemented in software and can be stored on a computer readable medium, which can include random access memory (RAM), read only memory (ROM), magnetic disk or optical disk, or the like. The software implementation of the present application files can further be transmitted or received over a modem or network connection.

[0198] The present application is described in reference to the flowchart illustrations and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations 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, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in one or more of the flowchart illustrations and / or block diagrams. Figure 1 means for performing each of the functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0199] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture including instructions which implement the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in one or more of the flowchart illustrations and / or block diagrams. Figure 1 means for performing each of the functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0200] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in one or more of the flowchart illustrations and / or block diagrams. Figure 1 Figure 1 means for performing each of the functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0201] While preferred embodiments of the application have been described, modifications and alterations thereto can occur to those skilled in the art upon reading the preceding description. It is intended to include all such modifications and alterations insofar as they come within the scope of the basic inventive concepts disclosed herein.

[0202] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method for monitoring abnormal power grid operation traffic and tracing its causes, characterized in that: include: Obtaining target power grid operation traffic statistical data, and performing a first preprocessing on the statistical data; The first preprocessing includes non-Gaussian feature extraction and Gaussian feature extraction; Presetting a target statistic, solving a control limit for the target statistic, and calculating a real-time value for the target statistic based on the statistical data after the first preprocessing; If the real-time value exceeds the control limit of the target statistic, it is determined that the operation volume is abnormal; Calculate the contribution values ​​of different variables to the target statistic and the control limits of the contribution values; When there is an abnormality in the operation business volume, the contribution values ​​of different variables to the out-of-limit statistics are calculated and compared with the contribution value control limits; Trace the cause based on the comparison results.

2. A method for monitoring abnormal power grid operation traffic and tracing its causes according to claim 1, characterized in that: The cause tracing according to the comparison result includes: If the contribution value of different variables to the out-of-limit statistics is not less than the contribution control limit, the corresponding variable is considered to be an abnormal variable; Calculate the limit-exceeding rate of the corresponding variable and determine the abnormality level based on the size of the limit-exceeding rate; The limit-exceeding statistic is a statistic that generates the limit-exceeding statistic.

3. A method for monitoring abnormal power grid operation traffic and tracing its causes according to claim 2, characterized in that: The target statistics include independent principal component space statistics, principal component space statistics and residual space statistics; If the real-time value of any target statistic exceeds the control limit of the target statistic, it is determined that there is an abnormality in the operation business volume.

4. A method for monitoring abnormal power grid operation traffic and tracing its causes according to claim 3, characterized in that: Calculating the limit-exceeding rate of the corresponding variable includes: Calculate the contribution of the corresponding variable to the statistic that produces the excess limit at the current moment; Obtain the control limit of the contribution value of the corresponding variable to the statistic that exceeds the limit; The exceeding rate of the corresponding variable is calculated according to the contribution value of the corresponding variable to the statistic that generates the exceeding limit at the current moment and the control limit of the contribution value of the corresponding variable to the statistic that generates the exceeding limit.

5. A method for monitoring abnormal power grid operation traffic and tracing its causes according to claim 4, characterized in that: The control limits of the target statistic and the control limits of the contribution value are both calculated by the kernel density estimation method.

6. A method for monitoring abnormal power grid operation traffic and tracing its causes according to claim 5, characterized in that: The non-Gaussian feature extraction and Gaussian feature extraction include: The non-Gaussian feature extraction is used to obtain a decomposition matrix, a mixing matrix, independent components and a residual space from statistical data; The Gaussian feature extraction is used to perform principal component decomposition on the residual space and obtain the principal component matrix, the load matrix and the residual matrix.

7. A method for monitoring abnormal power grid operation traffic and tracing its causes according to claim 6, characterized in that: The performing a first preprocessing on the statistical data includes: The first preprocessing further includes performing data centering and dimension removal on the statistical data and then performing non-Gaussian feature extraction and Gaussian feature extraction in sequence; The features extracted from the non-Gaussian features and the Gaussian features constitute the different variables.

8. A system for monitoring abnormal power grid operation traffic and tracing its causes, applying the method according to any one of claims 1 to 7, characterized in that: include: A preprocessing module, configured to obtain target power grid operation traffic statistics and perform a first preprocessing on the statistics; The first preprocessing includes non-Gaussian feature extraction and Gaussian feature extraction; a limit value acquisition module, configured to preset a target statistic, solve a control limit for the target statistic, and calculate a real-time value for the target statistic based on the statistical data after the first preprocessing; A judgment module, configured to determine that an abnormality exists in the operation volume if the real-time value exceeds the control limit of the target statistic; A calculation module, used to calculate the contribution values ​​of different variables to the target statistic and the contribution value control limits; A judgment module is used to calculate the contribution values ​​of different variables to the out-of-limit statistics when there is an abnormality in the operation business volume, and compare the contribution values ​​of different variables to the out-of-limit statistics with the contribution value control limits; The tracing module is used to trace the cause based on the comparison results.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of a method for abnormal monitoring and cause tracing of power grid operation traffic volume are implemented as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for abnormal monitoring and cause tracing of power grid operation traffic volume are implemented as described in any one of claims 1 to 7.

Images