Detection method and recovery method for multi-service abnormal data of intelligent power distribution network

Abstract

The invention discloses a method for detecting multi-service abnormal data of an intelligent power distribution network, and the method comprises the following steps: setting a monitoring point, and collecting and monitoring various service data of the intelligent power distribution network through the monitoring point; respectively extracting data features of the collected service data according to the types of the service data; and according to the classified and extracted data features, adopting a distributed state estimation algorithm to respectively detect various data features, and screening abnormal data in the data features. According to the method, on one hand, the abnormal data detection work of multiple pieces of business data can be efficiently completed for mass business data, and the detection efficiency is greatly improved; and on the other hand, after the abnormal data in the intelligent power distribution network system business are detected, the abnormal data are recovered to normal data for the intelligent power distribution network to use, and the working effects of related detection and recovery are evaluated, so that the integration of detection, recovery and evaluation of the multi-business abnormal data of the intelligent power distribution network is realized.

Description

Technical Field

[0001] The present invention belongs to the field of control technology of smart distribution networks, and specifically relates to multi-service abnormal data detection and abnormal data recovery of smart distribution networks. Background Art

[0002] With the development of human society, the growing demand for energy has made the distribution network one of the important infrastructures in the process of urban construction. Traditional distribution network upgrade solutions require a lot of capital investment, and it is difficult to adapt to the access of new energy and the rapidly changing electricity demand. The application of smart distribution network technology can realize distributed energy management and seamless access to new energy, and improve the energy efficiency and power supply reliability of the distribution network. Smart distribution network technology is applied to distribution network planning, and the distribution network can be planned by establishing a system model. The system model can take into account a variety of factors, such as load distribution, energy reserves, energy access methods, etc., and finally come up with an efficient, reliable and sustainable distribution network construction plan. In addition, the system model can also evaluate the power supply reliability, response speed and other indicators of the distribution network to ensure the safe and stable operation of the distribution network. Through the system model, the distribution network can be built into an intelligent and digital energy management platform to maximize the efficiency of the distribution network.

[0003] The safe operation of smart distribution networks requires the collection of various data from the power grid and efficient analysis. In the past, in terms of power grid data collection, sensors and smart meters collected the overall power grid status operation data, and then packaged the power grid status operation data and sent it to the data and status operation analysis center of the power grid, and used the least squares state estimation algorithm to analyze the operation status. The main disadvantages of this method are that the algorithm calculation is too large, the requirements for computing hardware are too high, the computing hardware cost is high, and the calculation accuracy is not high.

[0004] Especially with the large-scale construction of smart grids in my country, more and more sensors, smart meters and distributed power supply equipment are connected to the smart grid system, which leads to the increasingly extensive system services of smart grids, the increasing and complex business data of smart grids and the increasingly complex grid structure of smart grids. The traditional centralized abnormal data detection method is not suitable for processing massive business data.

[0005] In addition, traditional power grid systems rarely consider abnormal data recovery, and basically focus on abnormal data detection, which makes it impossible to integrate abnormal data detection, recovery, and evaluation in smart grid systems. Summary of the invention

[0006] Purpose of the invention: The present invention aims to at least partially solve the above-mentioned technical problems. To this end, the first aspect of the present invention provides a method for detecting abnormal data of multiple services in a smart distribution network, which can efficiently complete the abnormal data detection work of multiple services for massive business data.

[0007] The second aspect of the present invention provides a method for recovering abnormal data of multiple services in a smart distribution network. After detecting abnormal data in the services of the smart distribution network system, the abnormal data is restored to normal data for use in the smart distribution network, and the working effects of related detection and recovery are evaluated, thereby realizing the integration of abnormal data detection, recovery and evaluation of multiple services in the smart distribution network.

[0008] The method for detecting abnormal data of multiple services in a smart distribution network provided by the present invention comprises the following steps:

[0009] S1. Set up monitoring points and use them to collect and monitor various business data of the smart distribution network;

[0010] S2. Extracting data features of the collected business data according to the types of business data;

[0011] S3. Based on the data features extracted by classification, a distributed state estimation algorithm is used to detect various data features respectively and filter out abnormal data.

[0012] In the present invention, further preferably, the step S1 of collecting and monitoring various business data of the smart distribution network using monitoring points includes:

[0013] Use monitoring points to collect and monitor the power detection data of the smart distribution network;

[0014] Use monitoring points to collect and monitor equipment status data of smart distribution networks;

[0015] The communication delay data of the smart distribution network is collected and monitored using monitoring points.

[0016] Preferably, the electric energy detection data includes the voltage, frequency and harmonic content of the smart distribution network;

[0017] The equipment status data includes operation status data of transmission lines, smart substations, smart switchgear, smart meters and communication equipment;

[0018] The communication delay data includes transmission delay data, processing delay data and queuing delay data.

[0019] Preferably, the step S2 extracts data features of the collected business data according to the types of business data, including:

[0020] The statistical characteristic value extraction method is used to extract the data features of the electric energy detection data of the smart distribution network;

[0021] The equipment status data of the smart distribution network uses the statistical eigenvalue extraction method to extract its data features;

[0022] The communication delay data of the smart distribution network uses the time domain eigenvalue extraction method to extract its data features.

[0023] Preferably, in step S3, according to the data features extracted by classification, a distributed state estimation algorithm is used to detect various data features respectively and filter out abnormal data therein, including:

[0024] S31. According to the data features extracted by classification, different distributed abnormal data detection mechanisms are established respectively;

[0025] The mathematical model of the collected data is as follows:

[0026] (1);

[0027] (2);

[0028] (3);

[0029] in, is the actual observation data model; h(x) is the observation function of the system, which transforms the extracted business data feature vector Mapped to the observation space to obtain the business observation value h; v is the abnormal data model of the system; is the average value of the data set, which is used to represent The average over n observations, is the standard deviation of the data set, which is used to measure the degree of dispersion between the data points and the mean;

[0030] The anomaly detection mathematical model established by the distributed anomaly data detection mechanism is:

[0031] (4);

[0032] If a data point With its average The absolute value of the difference is greater than the threshold k multiplied by the standard deviation , then this data point is considered abnormal; where k is the threshold, which is determined by the business history data and business experience of the smart distribution network;

[0033] S32, putting the classified data into corresponding anomaly detection mathematical models for detection, and screening out the abnormal data.

[0034] Preferably, the step S1 of setting the monitoring points includes:

[0035] Appropriate monitoring equipment is set up as monitoring points at energy storage stations, transmission lines, distribution rooms, substations and user load ends.

[0036] Preferably, the adapted monitoring device comprises:

[0037] Electric energy sensors, smart meters, and power grid data acquisition and monitoring control systems.

[0038] The present invention provides a method for recovering abnormal data of multiple services in a smart distribution network, comprising the following steps:

[0039] S4, obtaining abnormal data detected by the above detection method;

[0040] S5. Use random matrix theory to recover abnormal data.

[0041] Preferably, the recovery of abnormal data using random matrix theory in step S5 includes:

[0042] S51, dividing the abnormal data into business categories and obtaining data features of the corresponding business data;

[0043] S52. Establish a random matrix mathematical model:

[0044] (5);

[0045] (6);

[0046] (7);

[0047] Wherein, formula (5) is the matrix representation of formula (1), and Z is p n measurement data observation matrix, which contains p observation values, each observation value is an n-dimensional vector; H is p The measurement matrix of m is used to describe the mapping of data from the m-dimensional original data space to the p-dimensional observation space; X is m n is the original data matrix, containing the data to be restored; V is p The abnormal data expression matrix of n represents the abnormal values ​​in the observed data;

[0048] C in formula (6) is the covariance matrix of the observation matrix, which is a p The matrix of p describes the linear relationship between the observed data;

[0049] Formula (7) is m n data recovery matrix, which uses the least squares method to recover data, that is, by minimizing Z and The error between , is a p The matrix of p, is the covariance matrix With the observation matrix The product of is used to project the observed data back to the original data space;

[0050] S53, recovering the abnormal data through the established random matrix mathematical model.

[0051] As an advantage, the method further comprises the following steps:

[0052] S6. After the abnormal detection data is repaired, the above detection method is used again for detection;

[0053] S7. If abnormal data is detected, restore it again.

[0054] Beneficial effects: The present invention takes into account the current task requirements, and in order to ensure that the new requirements and new changes of smart grid business data are met, proposes an abnormal data detection method for smart distribution networks based on distributed state estimation, and an abnormal data recovery method for smart distribution networks based on random matrix theory. On the one hand, for massive business data, it can efficiently complete the abnormal data detection of multiple business data, greatly improving the detection efficiency. On the other hand, after detecting the abnormal data in the smart distribution network system business, the abnormal data is restored to normal data for use in the smart distribution network, and the working effects of related detection and recovery are evaluated, realizing the integration of multi-business abnormal data detection, recovery and evaluation of smart distribution networks. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a flow chart of a method for detecting and recovering abnormal data of multiple services in a smart distribution network according to the present invention;

[0056] Figure 2 The present invention is a flowchart of a method for recovering abnormal data of multiple services in a smart distribution network. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings in the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0058] Combine the following Figure 1-Figure 2 The present invention describes a method for detecting and recovering abnormal data of multiple services in a smart distribution network.

[0059] Embodiment 1: This embodiment provides a method for detecting abnormal data of multiple services in a smart distribution network, such as Figure 1 As shown, the following steps are included:

[0060] S1. Set up monitoring points and use them to collect and monitor various business data of the smart distribution network;

[0061] The setting of monitoring points includes:

[0062] Appropriate monitoring equipment is set up as monitoring points at energy storage stations, transmission lines, distribution rooms, substations and user load ends.

[0063] The adapted monitoring equipment includes:

[0064] Electric energy sensors, smart meters, and power grid data acquisition and monitoring control systems.

[0065] The monitoring points are used to collect and monitor various business data of the smart distribution network, including:

[0066] The monitoring points are used to collect and monitor the power detection data of the smart distribution network; the power detection data includes the voltage, frequency and harmonic content of the smart distribution network;

[0067] Use monitoring points to collect and monitor equipment status data of the smart distribution network; equipment status data includes operating status data of transmission lines, smart substations, smart switchgear, smart meters and communication equipment;

[0068] The monitoring points are used to collect and monitor the communication delay data of the smart distribution network; the communication delay data includes transmission delay data, processing delay data and queuing delay data.

[0069] S2. Extracting data features of the collected business data according to the types of business data; including:

[0070] The statistical characteristic value extraction method is used to extract the data features of the electric energy detection data of the smart distribution network;

[0071] The equipment status data of the smart distribution network uses the statistical eigenvalue extraction method to extract its data features;

[0072] The communication delay data of the smart distribution network uses the time domain eigenvalue extraction method to extract its data features.

[0073] S3. Based on the data features extracted by classification, a distributed state estimation algorithm is used to detect various data features and filter out abnormal data, including:

[0074] S31. According to the data features extracted by classification, different distributed abnormal data detection mechanisms are established respectively;

[0075] The mathematical model of the collected data is as follows:

[0076] (1);

[0077] (2);

[0078] (3);

[0079] in, is the actual observation data model; h(x) is the observation function of the system, which transforms the extracted business data feature vector Mapped to the observation space to obtain the business observation value h; v is the abnormal data model of the system; is the average value of the data set, which is used to represent The average over n observations, is the standard deviation of the data set, which is used to measure the degree of dispersion between the data points and the mean;

[0080] The anomaly detection mathematical model established by the distributed anomaly data detection mechanism is:

[0081] (4);

[0082] If a data point With its average The absolute value of the difference is greater than the threshold k multiplied by the standard deviation , then this data point is considered abnormal; where k is the threshold, which is determined by the business history data and business experience of the smart distribution network;

[0083] S32, putting the classified data into corresponding anomaly detection mathematical models for detection, and screening out the abnormal data.

[0084] If it is detected as abnormal data, the abnormal data will be sent to the data recovery module for abnormal data recovery;

[0085] If it is detected as not abnormal data, it will enter the evaluation module and send the collected business data of the distribution network to the situation assessment and analysis center of the smart distribution network to conduct situation assessment on the collected data.

[0086] Embodiment 2: This embodiment provides a method for recovering abnormal data of multiple services in a smart distribution network, such as Figure 1 and Figure 2 As shown, firstly, the detection method of Example 1 is used to detect abnormal data, and then the following steps are performed:

[0087] S4, obtaining detected abnormal data;

[0088] S5. Use random matrix theory to recover abnormal data, including:

[0089] S51, dividing the abnormal data into business categories and obtaining data features of the corresponding business data;

[0090] S52. Establish a random matrix mathematical model:

[0091] (5);

[0092] (6);

[0093] (7);

[0094] Wherein, formula (5) is the matrix representation of formula (1), and Z is p n measurement data observation matrix, which contains p observation values, each observation value is an n-dimensional vector; H is p The measurement matrix of m is used to describe the mapping of data from the m-dimensional original data space to the p-dimensional observation space; X is m n is the original data matrix, containing the data to be restored; V is p The abnormal data expression matrix of n represents the abnormal values ​​in the observed data;

[0095] C in formula (6) is the covariance matrix of the observation matrix, which is a p The matrix of p describes the linear relationship between the observed data;

[0096] Formula (7) is m n data recovery matrix, which uses the least squares method to recover data, that is, by minimizing Z and The error between , is a p The matrix of p, is the covariance matrix With the observation matrix The product of is used to project the observed data back to the original data space;

[0097] S53, recovering the abnormal data through the established random matrix mathematical model.

[0098] S6. After the abnormal detection data is repaired, use S2 and S3 of Example 1 to detect again; detect the accuracy of the restored data and which data are still abnormal data. If the detection result is still abnormal data, restore it again.

[0099] S7. If the detection result is not abnormal data, the recovery work is evaluated to evaluate the accuracy of the data recovery and the speed of the data recovery algorithm.

[0100] The present invention is mainly based on the distributed state estimation algorithm and random matrix theory to construct a smart grid multi-task abnormal data detection and recovery evaluation method to achieve the integrated construction goal of abnormal data detection, recovery and evaluation.

[0101] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention.

Claims

1. A method for detecting abnormal data of multiple services in a smart distribution network, characterized in that: The steps include: S1. Set up monitoring points and use them to collect and monitor various business data of the smart distribution network; S2. Extracting data features of the collected business data according to the types of business data; S3. Based on the data features extracted by classification, a distributed state estimation algorithm is used to detect various data features respectively and filter out abnormal data.

2. The method for detecting abnormal data of multiple services in a smart distribution network according to claim 1, characterized in that: Step S1 uses monitoring points to collect and monitor various business data of the smart distribution network, including: Use monitoring points to collect and monitor the power detection data of the smart distribution network; Use monitoring points to collect and monitor equipment status data of smart distribution networks; The monitoring points are used to collect and monitor the communication delay data of the smart distribution network.

3. The method for detecting abnormal data of multiple services in a smart distribution network according to claim 2, characterized in that: The electric energy detection data includes the voltage, frequency and harmonic content of the smart distribution network; The equipment status data includes operation status data of transmission lines, smart substations, smart switchgear, smart meters and communication equipment; The communication delay data includes transmission delay data, processing delay data and queuing delay data.

4. The method for detecting abnormal data of multiple services in a smart distribution network according to claim 2, characterized in that: Step S2 extracts data features of the collected business data according to the types of business data, including: The statistical characteristic value extraction method is used to extract the data features of the electric energy detection data of the smart distribution network; The equipment status data of the smart distribution network uses the statistical eigenvalue extraction method to extract its data features; The communication delay data of the smart distribution network uses the time domain eigenvalue extraction method to extract its data features.

5. The method for detecting abnormal data of multiple services in a smart distribution network according to claim 4, characterized in that: According to the data features extracted by classification in step S3, a distributed state estimation algorithm is used to detect various data features respectively and filter out abnormal data therein, including: S31. According to the data features extracted by classification, different distributed abnormal data detection mechanisms are established respectively; The mathematical model of the collected data is as follows: （1）； （2）； （3）； in, is the actual observation data model; h(x) is the observation function of the system, which transforms the extracted business data feature vector Mapped to the observation space to obtain the business observation value h; v is the abnormal data model of the system; is the average value of the data set, which is used to represent The average over n observations, is the standard deviation of the data set, which is used to measure the degree of dispersion between the data points and the mean; The anomaly detection mathematical model established by the distributed anomaly data detection mechanism is: （4）； If a data point With its average The absolute value of the difference is greater than the threshold k multiplied by the standard deviation , then this data point is considered abnormal; where k is the threshold, which is determined by the business history data and business experience of the smart distribution network; S32, putting the classified data into corresponding anomaly detection mathematical models for detection, and screening out the abnormal data.

6. The method for detecting abnormal data of multiple services in a smart distribution network according to claim 1, characterized in that: The step S1 of setting the monitoring points includes: Appropriate monitoring equipment is set up as monitoring points at energy storage stations, transmission lines, distribution rooms, substations and user load ends.

7. The method for detecting abnormal data of multiple services in a smart distribution network according to claim 6, characterized in that: The adapted monitoring equipment includes: Electric energy sensors, smart meters, and power grid data acquisition and monitoring control systems.

8. A method for recovering abnormal data of multiple services in a smart distribution network, characterized in that: The steps include: S4. Obtaining abnormal data detected by any detection method of claims 1 to 7; S5. Use random matrix theory to recover abnormal data.

9. The method for recovering abnormal data of multiple services in a smart distribution network according to claim 8, characterized in that: The step S5 uses random matrix theory to recover abnormal data, including: S51, dividing the abnormal data into business categories and obtaining data features of corresponding business data; S52. Establish a random matrix mathematical model: （5）； （6）； （7）； Wherein, formula (5) is the matrix representation of formula (1), and Z is p n measurement data observation matrix, which contains p observation values, each observation value is an n-dimensional vector; H is p The measurement matrix of m is used to describe the mapping of data from the m-dimensional original data space to the p-dimensional observation space; X is m n is the original data matrix, containing the data to be restored; V is p The abnormal data expression matrix of n represents the abnormal values ​​in the observed data; C in formula (6) is the covariance matrix of the observation matrix, which is a p The matrix of p describes the linear relationship between the observed data; Formula (7) is m n data recovery matrix, which uses the least squares method to recover data, that is, by minimizing Z and The error between , is a p The matrix of p, is the covariance matrix With the observation matrix The product of is used to project the observed data back to the original data space; S53, recovering the abnormal data through the established random matrix mathematical model.

10. The method for recovering abnormal data of multiple services in a smart distribution network according to claim 8, characterized in that: The following steps are also included: S6. After the abnormal detection data is repaired, use any one of the detection methods of claims 1 to 7 to perform abnormal detection again; S7. If abnormal data is detected, restore it again.

Images