Characterization transformer area identification and division method and system

Abstract

The invention relates to a characterization transformer area identification and division method, which comprises the following steps of: establishing a system dynamics model by discovering a relationship between features and an association mapping relationship between the features and anomalies, further, selecting and combining representative significant features of a characterization transformer area based on the model and through a group decision feature root method, and finally, identifying and dividing the characterization transformer area. And on the basis of the significant features, through a binary logistic regression method, a characterization transformer area identification model is constructed. According to the method, the model can be used for identifying and judging whether any transformer area data is a set characteristic transformer area or not, and the system dynamics model corresponding to the characteristic transformer area can be used for knowing the abnormal mode of the characteristic transformer area by analyzing the internal structure mode of the characteristic transformer area, so that the effective treatment of the transformer area line loss is facilitated. Meanwhile, for the same type of special transformer area, the scheme of the invention can cover different types of abnormal recognition diagnosis, and also helps to improve the coverage rate and accuracy of the algorithm.

Description

Technical Field

[0001] The present application relates to the field of power grid operation and maintenance, and in particular to a method, system, computer device and computer-readable storage medium for characterizing substation area identification and division. Background Art

[0002] As the amount of electricity sold by the power grid increases, the issue of line loss management in the substation area has received increasing attention. However, the identification and management of line loss anomalies face multiple challenges. First, the number of power grid substations is huge, and the monitoring terminals generate massive amounts of data every day. Although the existing line loss management system has certain data analysis capabilities, its functions in refined management are still imperfect. Secondly, the access of distributed power sources has changed the flow of the substation area and increased the probability of line anomalies.

[0003] In related technologies, in order to accurately analyze the causes of abnormal line loss, researchers have conducted extensive research in topological relationship identification, household transformer relationship identification, phase identification, and abnormal substation identification. However, these studies mostly analyze and model a single abnormality, while in reality there is often more than one substation abnormality, and the characteristics of the substation with the same abnormality may also be very different.

[0004] Currently, no effective solution has been proposed to address the problem of poor coverage and accuracy of platform area identification and division methods in related technologies. Summary of the invention

[0005] The embodiments of the present application provide a method, system, computer device and computer-readable storage medium for characterizing station area identification and division, so as to at least solve the problem of poor coverage and accuracy of station area identification and division methods in related technologies.

[0006] In a first aspect, an embodiment of the present application provides a method for identifying and dividing a characteristic station area, the method comprising:

[0007] Identify and obtain the initial features of any type of characterized station area, and construct a system dynamics model based on the initial features to reflect the causal relationship between the features;

[0008] By using the system dynamics model, the initial features are combined and selected to obtain a plurality of intermediate features, and a group decision feature root method is used to identify and screen elements based on the intermediate features to obtain significant features reflecting the characteristics of the characterized station area;

[0009] Sample data corresponding to the significant features are collected and preprocessed. Based on the binary logistic regression method, a substation identification model is constructed with the preprocessed sample data as the independent variable and whether it is a characterized substation as the dependent variable. The substation identification model and the system dynamics model are used to identify and divide any target substation.

[0010] In some embodiments, combining and selecting the initial features to obtain a plurality of intermediate features includes:

[0011] By using the system dynamics model, a plurality of the initial features are integrated or interactively analyzed to obtain a more representative comprehensive feature;

[0012] Based on preset rules, selection is made among the initial features and / or the comprehensive features to obtain intermediate features, wherein the preset rules are formulated by analyzing the system dynamics model, including: direct measurability, mutual independence, overall completeness, target consistency and quantitative rationality.

[0013] In some embodiments, using the group decision feature root method to identify and screen elements based on the intermediate features includes:

[0014] Scoring the corresponding elements in the intermediate features according to the decision groups constructed based on the knowledge of multiple experts to obtain a scoring matrix;

[0015] In the scoring matrix, a scoring vector with the smallest sum of angles with the scoring vectors of each expert is constructed as an ideal scoring vector;

[0016] Based on the scoring matrix and its transposed matrix, an intermediate matrix is ​​obtained, the maximum eigenvalue vector of the intermediate matrix is ​​solved, the ideal scoring vector is determined, and the significant features of the characterized platform area are obtained.

[0017] In some embodiments, collecting sample data corresponding to the significant features and preprocessing the sample data includes:

[0018] Collecting sample data corresponding to the significant features, wherein the sample data includes first sample data of the characterized area and second sample data of the non-characterized area;

[0019] The operation method of the significant features is defined, and labels are added to the sample data, wherein if the sample data belongs to the characterized area, the label is 1, and if it does not belong to the characterized area, the label is 0.

[0020] In some embodiments, based on the binary logistic regression method, with the sample data after the preprocessing as the independent variable and whether it is a characterized area as the dependent variable, constructing an area identification model includes:

[0021] Obtaining a first Pearson correlation coefficient between the significant feature and the characterized area, and verifying the validity of selecting the significant feature according to the first Pearson correlation coefficient;

[0022] Calculating a second Pearson correlation coefficient between the significant features, and judging whether there are highly correlated variables between the significant features according to the second Pearson correlation coefficient,

[0023] If not, dividing the sample data into a training set and a test set; if so, after removing the highly correlated variables from the sample data, dividing the sample data into a training set and a test set;

[0024] The station area identification model is constructed based on the training set, and the prediction performance of the station area identification model is verified through the test set.

[0025] In some embodiments, the method further comprises:

[0026] Calculating a performance evaluation index through a binary confusion matrix, and evaluating the prediction accuracy of the station area identification model through the performance evaluation index;

[0027] The ROC curve corresponding to the platform area identification model is obtained, and the generalization ability of the platform area identification model is evaluated according to the area under the ROC curve.

[0028] In some of the embodiments, the characterized substation is a mountainous substation, and the significant features include: the number of users, power supply radius, abnormal probability of line loss in thunderstorm weather, maximum voltage difference between users and terminals, standard deviation of voltage during peak power consumption, probability of missing collection, and probability of voltage loss.

[0029] In a second aspect, an embodiment of the present application provides a characterized area identification and division system, the system comprising: a feature identification module, a feature screening module and a characterized area identification module, wherein:

[0030] The feature recognition module is used to identify and obtain the initial features of any type of characterized station area, and to construct a system dynamics model for reflecting the causal relationship between the features based on the initial features;

[0031] The feature screening module is used to combine and select the initial features to obtain multiple intermediate features through the system dynamics model, and adopt the group decision feature root method to identify and screen elements based on the intermediate features to obtain significant features reflecting the characteristics of the characterized station area;

[0032] The characterized station area identification module is used to collect sample data corresponding to the significant features, and pre-process the sample data, and build a station area identification model based on binary logistic regression method, taking the pre-processed sample data as the independent variable and whether it is a characterized station area as the dependent variable;

[0033] Furthermore, the area identification model and the system dynamics model are used to identify and divide any target area.

[0034] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the first aspect above when executing the computer program.

[0035] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect above.

[0036] Compared with the related art, the embodiment of the present application provides a method for characterizing substation identification and division, which links single factors together and runs through the characterizing substation research from a holistic and systematic perspective. For the characterizing substations with special circumstances, a system dynamics model is constructed by discovering the relationship between features and the correlation mapping relationship between features and anomalies. Furthermore, based on the model and through the group decision feature root method, representative significant features of the characterizing substation are selected and combined. Finally, based on the significant features, a model for characterizing substation identification is constructed through the binary logistic regression method. The model can be used to identify and determine whether any substation data is a given characterizing substation, and the system dynamics model corresponding to the characterizing substation can be used to understand its abnormal pattern by analyzing the structural pattern inside the characterizing substation, which is beneficial to the effective management of substation line loss. At the same time, for the same type of special substations, the present application scheme can cover its different types of abnormal identification and diagnosis, which also helps to improve the coverage and accuracy of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0038] Figure 1 is a flow chart of a method for characterizing station area identification and division according to an embodiment of the present application;

[0039] Figure 2 is a schematic diagram of a causal relationship between various characteristics of a characterized station area according to an embodiment of the present application;

[0040] Figure 3 is a schematic diagram of the causal relationship between the characteristics of a mountainous area and anomalies according to an embodiment of the present application;

[0041] Figure 4 is a schematic diagram of evaluating the generalization ability of a characterization station area recognition model according to an embodiment of the present application;

[0042] Figure 5is a structural block diagram of a characterization station area identification and division system according to an embodiment of the present application;

[0043] Figure 6 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application, DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0045] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0046] Reference to "embodiments" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0047] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there may be three relationships, for example, "A and / or B" can represent: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0048] With the continuous increase in power sales, the line loss management of the substation has received great attention, but the abnormal line loss identification and management process is full of difficulties. On the one hand, the number of power grid substations is relatively large, and the monitoring terminals of the power grid substations can receive massive line loss data every day. Although the line loss management system has realized the intelligent analysis of data to a certain extent, the current software functions are still imperfect for the refined management of substation line losses. On the other hand, the access of distributed power sources has changed the flow of the substation and increased the probability of abnormal substation lines. In order to accurately analyze the causes of abnormal substation line losses, many scholars have done a lot of research on intelligent diagnosis of line losses, such as substation topology relationship identification, household transformer relationship identification, phase identification, and abnormal substation identification.

[0049] In the related technology, Liu Dangwu and others proposed a method for diagnosing the cause of low voltage at the outlet of the substation based on feature recognition. This method extracts relevant data such as basic information of the low-voltage substation and power consumption characteristics, compares the calculated index with the preset value, and realizes real-time analysis and judgment of the cause of low voltage at the outlet of the substation. Chang Hongshan and He Guang invented a method for substation identification of small-load users. By collecting continuous load data from the total meter in the substation, the small-load sub-meter to be identified, and all identified sub-meters, the household-to-household relationship of the small-load sub-meter is matched based on the principle that load jumps will bring obvious voltage fluctuations. Zhang Tao and Zhang Bo disclosed a substation identification method based on substation feature information, which uses communication equipment to exchange substation feature information and identifies substations based on the similarity of substation feature information.

[0050] The inventor of this case found through analyzing the related literature of the prior art that the above-mentioned substation analysis is all for a single anomaly, and the cause of the anomaly is determined by obtaining data of a single abnormal substation for data analysis and algorithm modeling. However, in reality, there is often more than one substation anomaly, and the substation characteristics of the same anomaly may also be very different. The prior art usually cannot meet the requirements of generalization.

[0051] In the field of power system operation and maintenance management, a substation refers to a relatively independent distribution unit in the power supply system, usually consisting of a distribution transformer and the users it supplies power to. In China, the power distribution network of the power system usually divides a larger area into multiple substations, each of which is powered by a distribution transformer (usually called a substation transformer or public transformer), and the users in the substation obtain power supply through this transformer. Users in the substation can be residents, commercial users, or small industrial users.

[0052] Among them, the substations in mountainous areas, coastal areas, newly built areas, and demolished areas have their own unique characteristics, which are mainly reflected in the geographical environment, power demand, power supply challenges and solutions. In this embodiment, the substations with significant particularity are called characteristic substations;

[0053] The following are several exemplary analyses of the characterization area:

[0054] 1. Mountainous areas: complex geographical environment: mountainous areas have rugged terrain, inconvenient transportation, long power supply lines and are easily affected by natural disasters; scattered electricity demand: residential areas and agricultural and industrial power points are widely distributed and scattered, making power supply difficult; poor power supply stability: due to terrain and climate factors, mountainous areas often face problems such as unstable voltage and power outages.

[0055] 2. Coastal areas: vulnerable to natural disasters such as typhoons: Coastal areas are often hit by natural disasters such as typhoons and tsunamis, which damage the power grid. Large electricity demand: Coastal areas have a developed economy and a dense population, and industrial and commercial electricity demand is large. Marine climate characteristics: High air humidity can easily lead to equipment corrosion and aging.

[0056] For the above-mentioned characteristic substations, in this field, most of the current research is to analyze and demonstrate single factors such as basic information of the substation, load data, line loss, and anomalies. This type of method is aimed at a single anomaly, and the cause of the anomaly is determined by obtaining data from a single abnormal substation for data analysis and algorithm modeling. However, in fact, there is often more than one abnormal substation, and the characteristics of the substation with the same abnormality may also be very different.

[0057] Taking the above problems into consideration, this application scheme introduces characterized substation labels to carry out abnormal re-diagnosis in a targeted manner, which helps to improve the coverage and accuracy of the algorithm. Combining system science theory and computer simulation, the abnormal pattern is understood by analyzing the structural pattern inside the characterized substation. Characterized substation identification can improve the accuracy of characterized substation identification by associating abnormal maps, and on the other hand, it helps to effectively manage line losses in characterized substations.

[0058] Specifically, Figure 1 is a flow chart of a method for characterizing station area identification and division according to an embodiment of the present application, such as Figure 1 As shown, the process includes the following steps:

[0059] S101, identifying and obtaining initial features of any type of characterized station area, and constructing a system dynamics model for reflecting the causal relationship between the features based on the initial features;

[0060] The characteristic area can be any one of the areas with special characteristics such as mountainous areas, rural areas, newly built areas, coastal areas, etc.

[0061] Furthermore, the identified initial features that affect the identification of substations include but are not limited to: (1) basic features: number of users, number of photovoltaic users, power supply radius, line length, line diameter, etc.; (2) electrical features: current, voltage, power consumption, power supply, load density, voltage difference, etc.; (3) abnormal features: voltage loss, voltage over-limit, data collection abnormality, low voltage at the end of the line, household transformer, etc.

[0062] In addition, system dynamics is a computer simulation method for studying the dynamic behavior of social systems. It can not only deal with the modeling problems of complex systems, but also quantitatively regulate the relationship between the various elements of the system. It is a dynamic feedback system with significant characteristics such as systematicity, nonlinearity and complexity.

[0063] In this embodiment, in order to clarify the significant characteristics of special substations and analyze the differences between special substations, based on the relationship between basic characteristics and data performance and anomalies, the characteristics of the characterized substations and the anomalies that may be caused by the characteristics are regarded as a system, and a system dynamics model for studying the relationship between substation characteristics is constructed. The dynamic process of the formation of the relationship between special substation characteristics and anomalies is depicted through a causal feedback loop diagram.

[0064] Figure 2 is a schematic diagram of the causal relationship between various characteristics of a characterized area according to an embodiment of the present application, such as Figure 2 As shown, the schematic diagram is a visual example of the system dynamics model, wherein, according to the model, the mutual connection and interaction mode between various characteristic variables in the characterized substation, the series connection relationship between the variables that can transmit causal relationships in each module, and the correlation relationship between characteristics and abnormal states can be obtained.

[0065] Specifically, taking the mountainous area as an example, according to the task of identifying the characteristics of the mountainous area, the significant characteristics of the mountainous area that are different from other areas, the load data performance and possible anomalies are identified.

[0066] Among them, the characteristic that clearly distinguishes mountainous substations from other substations is that they are easily affected by the weather. When encountering thunderstorms, the probability of power outages for users in mountainous areas is higher than in other substations, which will cause a series of anomalies such as voltage phase failure, data collection loss, and clock abnormalities.

[0067] In addition, the line length in mountainous areas is long, and the power supply radius of the substation is large, which will lead to a large voltage difference between users and terminals, resulting in a large theoretical line loss rate, voltage dropout, low voltage at the end of the line, etc. Therefore, by sorting out the significant features of specific substations, exploring the data performance brought by these features and the anomalies that may be caused, and fully analyzing the internal connections between these factors, it is helpful to identify characteristic substations.

[0068] In an exemplary embodiment, the identification of mountain areas in the process of establishing a system dynamics model includes a cause tree and a result tree, and the cause-effect relationship constructed is as follows: Figure 3 As shown, Figure 3 Schematic diagram of the causal relationship between the characteristics of the mountainous area and abnormalities according to an embodiment of the present application: Figure 3 As shown in the figure, this model can be used to obtain the correlation information between various features in mountainous areas, as well as the correlation between the features and the possible anomalies.

[0069] S102, selecting and combining the initial features of the characterized area through the system dynamics model to obtain a plurality of intermediate features, and using the group decision feature root method to identify and screen elements based on the intermediate features of the characterized area to obtain significant features reflecting the characteristics of the characterized area;

[0070] It should be noted that it is obviously unreasonable to use all the above characteristics as the criteria for screening substations. Therefore, when screening substations in mountainous areas as characterized substations, the above system dynamics model can be used to formulate a series of principles to ensure the effectiveness, representativeness and practicality of the selected characteristics.

[0071] In order to further find the significant features that can highlight the mountainous area, in this embodiment, it is necessary to first select and combine the features, where the combination of features refers to the integration or interactive analysis of multiple individual features to form new and more representative comprehensive features. This method can help describe the characteristics of the mountainous area more comprehensively, so as to more effectively identify and distinguish them, which can reduce redundancy and improve the accuracy and efficiency of the model.

[0072] The combination of features includes: combining two or more features to generate a new feature. For example, "line length" and "power supply radius" can be combined to generate a new feature "ratio of line length to power supply radius" to better reflect the power supply characteristics of mountainous areas.

[0073] Feature aggregation: Aggregate similar or related features. For example, multiple voltage-related features (such as voltage phase failure, low voltage, and voltage loss) can be aggregated into a comprehensive voltage anomaly feature.

[0074] Furthermore, the selection of features follows the principles of direct measurability, mutual independence, overall completeness, goal consistency, and reasonable quantity, and some redundant feature indicators are eliminated to reduce the redundancy between indicators; in this embodiment, the preset rules can also be formulated by analyzing the system dynamics model.

[0075] Through the above method, feature combination can help to more effectively identify the significant features and abnormal performance of mountainous areas, and improve the accuracy and practicality of the model. Further, based on the preset rules, feature selection is performed to obtain intermediate features, which can better capture the significant features and abnormal performance of mountainous areas, thus reducing redundancy and improving the accuracy and efficiency of the model.

[0076] After obtaining the intermediate features, the group decision feature root method is further used to identify and screen the substation features, providing a scientific basis for factor selection. The group decision feature root method is a multi-feature selection and screening method. By combining the analysis results of multiple features, the key features that best represent the decision-making goals are determined.

[0077] Among them, the group decision feature root method is to sort the scores of various factors and select more important key features according to the sorting results. The specific steps include the following:

[0078] Step 1, preliminary selection: preliminary feature selection based on experience or predefined criteria, removing obviously irrelevant or redundant features;

[0079] Step 2: Construct a decision group based on the knowledge of multiple experts, and score each element in the intermediate features of the characterized substations through the characterized substation decision group to obtain a scoring result matrix;

[0080] Step 3, in the characteristic area scoring result matrix, construct a scoring vector with the smallest sum of angles with the expert scoring vectors as the ideal scoring vector;

[0081] Step 4, based on the characterized area scoring result matrix and its transposed matrix, obtain the intermediate matrix, solve the maximum eigenvalue vector of the intermediate matrix, which is the ideal scoring vector, and obtain the significant features of the characterized area.

[0082] In an exemplary embodiment, taking a mountainous area as an example, selecting significant features by using a group decision feature root method includes the following steps:

[0083] First, experts S1, S2, ..., Sm jointly form a decision-making group to evaluate the factors B1, B2, ..., Bn to form a scoring result matrix X, where xij represents the score of expert Si on factor Bj, where i = 1, 2, ..., m, j = 1, 2, ..., n.

[0084]

[0085] It should be noted that since expert decisions are not always reliable and will be affected by the environment, it is assumed that there is an ideal expert S* with the most correct scoring result. The ideal expert has an ideal scoring vector with the smallest sum of angles with the scoring vectors of all experts in the decision-making group. Its scoring vector X*∈En, Generally, ‖b‖2=1, so X* is the vector where the function takes the maximum value. The specific calculation method is:

[0086]

[0087] Further, let the matrix F be the product of the score matrix X and its transposed matrix F = X T X, solve the maximum eigenvalue vector of matrix F, use the power method to solve the vector X*, the calculation steps are:

[0088] First, set the accuracy ε. When k = 0,

[0089] y1=Fy0

[0090]

[0091] When k=1,2,…,y k+1 =Fz k

[0092]

[0093] Repeat the above formula until |z k→k+1 | is less than the previously set precision ε, the z obtained at this time k+1 That is the vector X*, which reflects the significant features of the characteristic area.

[0094] In this embodiment, the mountain area identification index determined based on the group decision characteristic root method includes the number of users, power supply radius, abnormal line loss probability in thunderstorm weather, maximum voltage difference between users and terminals, standard deviation of voltage during peak power consumption, probability of missing data collection, and probability of voltage loss. The group decision characteristic root method can scientifically identify and screen significant features that can reflect the unique characteristics of the characteristic area.

[0095] S103, collect sample data corresponding to significant features, and pre-process the sample data, based on the binary logistic regression method, use the pre-processed sample data as the independent variable, and use whether it is a characterized substation as the dependent variable to build a substation identification model, and use the substation identification model and the system dynamics model to identify and divide any target substation.

[0096] In this embodiment, before constructing the substation identification model, data definition description and preprocessing are required first. The purpose of the data definition description is to define and clarify the reference factors for characterizing substation identification using various significant features. In this embodiment, taking the mountain substation as an example, it specifically includes: first clarifying the time range of the sample data, and statistically analyzing the values ​​of each indicator based on the time interval.

[0097] (1) Probability of abnormal line loss in thunderstorm weather = cumulative number of days with abnormal line loss / cumulative number of days with thunderstorm weather. The cumulative number of days with thunderstorm weather in a period of time is obtained based on meteorological data, and the cumulative number of days with abnormal line loss on the corresponding date is counted through the data collection system.

[0098] (2) Maximum voltage difference between user and terminal = max(max(daily voltage difference between user and terminal), all users);

[0099] (3) Standard deviation of voltage during peak power consumption = std (user voltage during peak power consumption), where the peak power consumption period is determined by the following method: the power consumption of all users in the substation area is counted on an hourly basis. When the power consumption growth rate exceeds the threshold, the initial value of the period is set to the initial time of the peak period, time1. When the power consumption reduction rate exceeds the threshold, the initial value of the period is set to the end time of the peak period, time2. The period [time1, time2] is the peak power consumption period.

[0100] (4) Probability of missing data = cumulative number of missing data anomalies / cumulative number of days with thunderstorms. The cumulative number of days with thunderstorms within a period of time is obtained based on meteorological data, and the cumulative number of days with missing data anomalies on the corresponding date is counted by the data collection system.

[0101] (5) Voltage undervoltage probability = cumulative number of voltage undervoltage anomalies / number of statistical days. The number of statistical days is based on the sample time, and the cumulative number of days with voltage undervoltage anomalies on the corresponding date is counted by the sampling system.

[0102] Among them, the indicator data and labels were obtained through the use of sampling systems and surveys. According to statistics, there are 139 substation samples, of which mountainous substation samples account for 36% and non-mountainous substation samples account for 67%. In addition to basic data such as the number of users and power supply radius, the values ​​of other indicators are determined by logical equations;

[0103] Based on the above samples, the calculated values ​​corresponding to the significant features obtained in the above step S102 are calculated, and the labels of the mountainous areas are introduced. If the area is in a mountainous area, the label is configured as 1, otherwise it is 0.

[0104] Finally, the logistic regression model is trained by adding labeled sample data. The logistic regression model is an important classification model in machine learning. It is widely used because of its simple, efficient and easy to promote algorithm. This study mainly studies whether it is a mountainous area (characterized area), so the binary logistic regression method of the logistic regression model is selected.

[0105] In this embodiment, the binary logistic regression model can be expressed as:

[0106]

[0107] Among them, y∈{0,1}, The decision function is:

[0108] In order to find the parameter θ, the maximum likelihood estimation can be used to maximize P(y|X;θ);

[0109] This study selected whether the substation is in a mountainous area as the dependent variable, and the independent variables were the number of users, power supply radius, abnormal line loss probability in thunderstorm weather, maximum voltage difference between users and terminals, standard deviation of voltage during peak hours, probability of missing data collection, and probability of voltage loss. First, the linear correlation between different characteristics and mountainous areas (1 represents mountainous areas, 0 represents non-mountainous areas) was studied. The Pearson correlation coefficients of the seven characteristic factors and mountainous areas are shown in the following table.

[0110] Table 1: Pearson product moment correlation coefficients between seven characteristics and mountainous areas

[0111] X1 X2 X3 X4 X5 X6 X7 -0.52 0.65 0.70 0.03 0.45 0.35 0.35

[0112] It can be seen from the table that, except for the number of users, the correlation coefficients of other features and mountainous areas are all positive, that is, the fewer the number of users, the higher the probability of being a mountainous area, and the longer the power supply radius, the higher the probability of being a mountainous area, which is consistent with the actual situation. Sorting by the absolute value of the correlation coefficient, the maximum value of the voltage difference between the user and the terminal has the smallest correlation with the mountainous area, which preliminarily shows that the voltage difference has little effect on the identification of mountainous areas.

[0113] Secondly, in order to avoid multicollinearity problems between variables, the Pearson correlation coefficients are calculated for each of the seven features. Variables with correlation coefficients greater than 0.8 are considered highly correlated variables, and one of the highly correlated variables needs to be eliminated. The Pearson correlation coefficients between the seven features are shown in the following table, from which it can be seen that there is no high correlation between the variables.

[0114] Table 2: Pearson product-moment correlation coefficients between seven features

[0115]

[0116]

[0117] In the above exemplary embodiment, a binary logistic regression model is used to predict whether it is a mountainous area. The binary logistic regression model established by selecting 7 indicators has a Hosmer-Lemeshow test P>0.10, indicating that it is suitable for logistic regression modeling.

[0118] Furthermore, 70% of the samples were randomly selected as the training set and 30% of the samples were selected as the test set. By fitting the full model and stepwise regression, it was found that the three indicators of user number, power supply radius, and abnormal probability of line loss in thunderstorm weather were significant. The logistic regression model constructed by these three indicators had the smallest AIC. The fitting results are shown in the following table.

[0119] Table 3: Logistic regression fitting results

[0120] Estimated value Standard error Z value P-value Intercept -3.18 1.73 -1.84 0.066 X1 -0.096 0.03 -3.01 0.003 X2 0.007 0.002 2.92 0.003 X3 6.12 3.01 2.03 0.042

[0121] In addition, in order to evaluate the prediction performance of the binary logistic regression model for mountainous areas, the evaluation indicators used in this scheme are accuracy, precision, recall, F1 and AUC. In order to calculate the evaluation index value, the binary classification confusion matrix is ​​introduced, and the confusion matrix of the classification result is shown in the following table.

[0122] Table 4: Confusion matrix of classification results

[0123]

[0124] The calculation results of accuracy, precision, recall and F1 are shown in Table 4.

[0125] Evaluation Metrics Metrics accuracy 95.23％ precision 92.86％ recall 92.86％ F1 92.86％

[0126] Finally, in order to evaluate the generalization ability of the prediction model, the ROC curve is introduced. The area under the ROC curve (AUC) determines the generalization ability of the model. Figure 4 is a schematic diagram of evaluating the generalization ability of the characterized station area recognition model according to an embodiment of the present application, such as Figure 4 As shown, the area under the curve (AUC) is 0.97, indicating that the model has a good fitting effect.

[0127] It should be noted that although the above-mentioned model constructed based on the binary logistic regression method is explained using the mountainous area as an example, it should be understood that the inventive concept of this application can also be applied to characteristic areas such as rural areas, newly built areas, demolition areas and coastal areas.

[0128] In addition, it should be noted that under the inventive concept of the present application, the feature screening method involved in the specific real-time means can be replaced by methods such as hierarchical analysis method, entropy weight method, fuzzy comprehensive evaluation, etc.; further, the characterized substation classification model can also be replaced by classification algorithms such as decision trees, random forests, SVM, and naive Bayes.

[0129] Furthermore, the characterized substation identification model can be used to identify and determine whether any substation is a characterized substation; further, since the system dynamics establishes a causal network among the variables within the substation by sorting out the feedback structural relationship between the basic characteristics of the substation, load performance, and line loss anomalies, based on the study of the overall situation and the relationship between the overall situation, the system dynamics model corresponding to the characterized substation is selected for analysis based on the data, which can specifically find the source of the anomaly from the internal structure of various types of substations and classify the anomaly types.

[0130] Through the above steps S101 to S104, compared with the related art, the present application scheme links single factors together to conduct a holistic and systematic study of the characterized substation. For the characterized substations with special circumstances, a system dynamics model is constructed by discovering the relationship between features and the correlation mapping relationship between features and anomalies. Furthermore, based on the model and through the group decision feature root method, representative significant features of the characterized substation are selected and combined. Finally, based on the significant features, a binary logistic regression method is used to construct a model for identifying the characterized substation. The model can be used to identify and determine whether any substation data is a given characterized substation, and the system dynamics model corresponding to the characterized substation can be used to understand its abnormal pattern by analyzing the structural pattern inside the characterized substation, which is conducive to the effective management of the substation line loss. At the same time, for the same type of special substation, the present application scheme can cover its different types of abnormal identification and diagnosis, which also helps to improve the coverage and accuracy of the algorithm.

[0131] This embodiment also provides a characterization station area identification and division system, which is used to implement the above-mentioned embodiments and preferred implementation modes, and will not be repeated hereafter. As used below, the terms "module", "unit", "subunit", etc. can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0132] Figure 5 is a structural block diagram of a characterization station area identification and division system according to an embodiment of the present application, such as Figure 5 As shown, the system includes a feature recognition module 50, a feature screening module 51 and a feature-based station identification module 52, wherein:

[0133] The feature recognition module 50 is used to identify and obtain the initial features of any type of characterized station area, and to construct a system dynamics model for reflecting the causal relationship between the features based on the initial features;

[0134] The feature screening module 51 is used to combine and select the initial features to obtain multiple intermediate features through the system dynamics model, and adopt the group decision feature root method to identify and screen the elements based on the intermediate features to obtain the significant features reflecting the characteristics of the characterized substation area;

[0135] The characterized substation identification module 52 is used to collect sample data corresponding to significant features and preprocess the sample data, and build a substation identification model based on the binary logistic regression method, with the preprocessed sample data as the independent variable and whether it is a characterized substation as the dependent variable; and use the substation identification model and the system dynamics model to identify and divide any target substation.

[0136] Through the above system, for the characterized substations with special circumstances, the system dynamics model is constructed by discovering the relationship between features and the correlation mapping relationship between features and anomalies. Furthermore, based on the model and through the group decision feature root method, the representative significant features of the characterized substation are selected and combined. Finally, based on the significant features, a binary logistic regression method is used to construct a model for the identification of characterized substations. The model can be used to identify and determine whether any substation data is a given characterized substation, and the system dynamics model corresponding to the characterized substation can be used to understand its abnormal mode by analyzing the internal structural mode of the characterized substation, which is conducive to the effective management of the line loss in the substation.

[0137] In one embodiment, Figure 6 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application, such as Figure 6 As shown, an electronic device is provided, which may be a server, and its internal structure diagram may be as shown in Figure 6 As shown. The electronic device includes a processor, a network interface, an internal memory and a non-volatile memory connected through an internal bus, wherein the non-volatile memory stores an operating system, a computer program and a database. The processor is used to provide computing and control capabilities, the network interface is used to communicate with an external terminal through a network connection, the internal memory is used to provide an environment for the operation of the operating system and the computer program, the computer program is executed by the processor to implement a method for characterizing the identification and division of the station area, and the database is used to store data.

[0138] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0139] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0140] The above characterization area embodiments only express several implementation methods of the present application, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present application, which all belong to the protection scope of the present application. Therefore, the protection scope of the patent of this application shall be based on the attached claims.

Claims

1. A method for characterizing station area identification and division, characterized in that: The method comprises: Identify and obtain the initial features of any type of characterized station area, and construct a system dynamics model based on the initial features to reflect the causal relationship between the features; By using the system dynamics model, the initial features are combined and selected to obtain a plurality of intermediate features, and a group decision feature root method is used to identify and screen elements based on the intermediate features to obtain significant features reflecting the characteristics of the characterized station area; Collecting sample data corresponding to the significant features, and preprocessing the sample data, and constructing a substation identification model based on binary logistic regression method, taking the preprocessed sample data as an independent variable and taking whether it is a characterized substation as a dependent variable; The station area identification model and the system dynamics model are used to identify and divide any target station area.

2. The method according to claim 1, characterized in that Combining and selecting the initial features to obtain multiple intermediate features include: By using the system dynamics model, a plurality of the initial features are integrated or interactively analyzed to obtain a more representative comprehensive feature; Based on preset rules, selection is made among the initial features and / or the comprehensive features to obtain intermediate features, wherein the preset rules are formulated by analyzing the system dynamics model, including: direct measurability, mutual independence, overall completeness, target consistency and quantitative rationality.

3. The method according to claim 2, characterized in that Using the group decision feature root method to identify and screen elements based on the intermediate features includes: Scoring the corresponding elements in the intermediate features according to the decision groups constructed based on the knowledge of multiple experts to obtain a scoring matrix; In the scoring matrix, a scoring vector with the smallest sum of angles with the scoring vectors of each expert is constructed as an ideal scoring vector; Based on the scoring matrix and its transposed matrix, an intermediate matrix is ​​obtained, the maximum eigenvalue vector of the intermediate matrix is ​​solved, the ideal scoring vector is determined, and the significant features of the characterized platform area are obtained.

4. The method according to claim 1, characterized in that: Collecting sample data corresponding to the significant features and preprocessing the sample data includes: Collecting sample data corresponding to the significant features, wherein the sample data includes first sample data of the characterized area and second sample data of the non-characterized area; The operation method of the significant features is defined, and labels are added to the sample data, wherein if the sample data belongs to the characterized area, the label is 1, and if it does not belong to the characterized area, the label is 0.

5. The method according to claim 1, characterized in that Based on the binary logistic regression method, the sample data after the preprocessing is used as the independent variable, and whether it is a characterized substation is used as the dependent variable, and the substation identification model is constructed, including: Obtaining a first Pearson correlation coefficient between the significant feature and the characterized area, and verifying the validity of selecting the significant feature according to the first Pearson correlation coefficient; Calculating a second Pearson correlation coefficient between the significant features, and judging whether there are highly correlated variables between the significant features according to the second Pearson correlation coefficient, If not, dividing the sample data into a training set and a test set; if so, after removing the highly correlated variables from the sample data, dividing the sample data into a training set and a test set; The station area identification model is constructed based on the training set, and the prediction performance of the station area identification model is verified through the test set.

6. The method according to claim 5, characterized in that The method further comprises: Calculating a performance evaluation index through a binary confusion matrix, and evaluating the prediction accuracy of the station area identification model through the performance evaluation index; The ROC curve corresponding to the station area identification model is obtained, and the generalization ability of the station area identification model is evaluated according to the area under the ROC curve.

7. The method according to claim 1, characterized in that The characterized substation is a mountainous substation, and the significant features include: the number of users, power supply radius, abnormal line loss probability in thunderstorm weather, maximum voltage difference between users and terminals, standard deviation of voltage during peak power consumption, probability of missing collection, and probability of voltage loss.

8. A characterization area identification and division system, characterized in that: The system comprises: a feature recognition module, a feature screening module and a feature-based station area recognition module, wherein: The feature recognition module is used to identify and obtain the initial features of any type of characterized station area, and use the system dynamics model to identify the causal relationship between the features based on the initial features; The feature screening module is used to combine and select the initial features to obtain multiple intermediate features through the system dynamics model, and adopt the group decision feature root method to identify and screen elements based on the intermediate features to obtain significant features reflecting the characteristics of the characterized station area; The characterized station area identification module is used to collect sample data corresponding to the significant features, and pre-process the sample data, and build a station area identification model based on binary logistic regression method, taking the pre-processed sample data as the independent variable and whether it is a characterized station area as the dependent variable; Furthermore, the area identification model and the system dynamics model are used to identify and divide any target area.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images