Transformer substation relay protection setting value intelligent calculation system based on big data analysis

Abstract

The invention relates to a transformer substation relay protection setting value intelligent calculation system based on big data analysis, and belongs to the technical field of power system protection and automation. The system comprises a data acquisition module, a constant value equipment classification module, a data preprocessing module, a feature extraction module, a feature construction module, a feature coding module, a model construction module and a constant value result calculation output module. According to the system, the efficiency and accuracy of relay protection setting value calculation can be improved, and the calculation period is shortened; the manual intervention degree is reduced, and human errors are reduced; and a powerful guarantee is provided for safe and stable operation of a power system.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system protection and automation, and specifically relates to an intelligent calculation system for relay protection settings of substations based on big data analysis. Background Art

[0002] In substations, relay protection devices are important equipment to ensure the safe operation of power systems. Traditional relay protection setting calculations usually rely on manual experience, which has the problems of low efficiency and low accuracy. With the development of smart grids, higher requirements are placed on the speed and accuracy of relay protection setting calculations.

[0003] Honghe Power Grid manages up to 190 35-220kV substations. With the full implementation of intensification and comprehensive transformation projects, the calculation of the fixed values ​​of each substation takes a lot of time. In addition, most of the main transformers of 110kV and above are three-winding transformers. The calculation of the main transformer is affected by the small power on each side. The protection in the station includes bus differential, line, capacitor, grounding transformer, etc., which are large and complicated. The traditional protection fixed value calculation is based on the relevant setting regulations and is manually calculated according to the formula. The protection fixed value is exported through the three links of setting, review and approval. Taking a 110kV substation as an example, the station has two main transformers, one 110kV bus differential protection, one 110kV, 35kV, and 10kV standby automatic transfer device, 20 35kV and 10kV outgoing lines, and 3 capacitors. The calculation formulas involved in the calculation of the entire station are about 242. The station needs to perform 242*3=726 formula calculations for the three levels of setting, review and approval. The workload of setting calculation is huge, and there is a problem that one calculation error leads to the problem of recalculation of the whole set, which not only wastes time but is also prone to errors. In addition, the traditional calculation method protects the calculation manuscript in paper form, which cannot realize the digital management of fixed value calculation.

[0004] With the rapid development of smart grids, the intelligent, information-based and interactive features of substation relay protection systems are becoming increasingly prominent. At present, some methods based on data analysis and machine learning have been applied to the field of power systems, such as predicting power system loads and faults by building prediction models. However, these methods are mainly focused on load prediction, fault diagnosis, etc., and their application in substation protection setting calculation is relatively small. Traditional relay protection setting calculation methods often rely on empirical formulas and manual calculations, and have the following disadvantages:

[0005] (1) Complex calculation process: Traditional fixed value calculation requires manual participation in multiple steps, and the calculation process is cumbersome;

[0006] (2) Time-consuming: Manual calculation takes a lot of time and cannot meet the needs of rapid response;

[0007] (3) Error-prone: Manual calculations are prone to errors due to negligence or lack of experience;

[0008] The introduction of big data analysis technology provides a new solution for realizing fast and accurate relay protection setting calculation. In the existing technology, although there are some relay protection setting analysis methods based on big data, these methods often only focus on simple data processing and analysis, lack of systematicness and comprehensiveness, and cannot give full play to the advantages of big data technology.

[0009] Therefore, how to overcome the shortcomings of existing technologies is an urgent problem to be solved in the field of power system protection and automation technology. Summary of the invention

[0010] The purpose of the present invention is to solve the deficiencies of the prior art and to provide a substation relay protection setting intelligent calculation system based on big data analysis.

[0011] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0012] The intelligent calculation system for substation relay protection setting based on big data analysis includes:

[0013] The data acquisition module is used to collect historical substation protection setting calculation and result data; it is also used to collect equipment type and parameter information of the substation to be calculated;

[0014] The fixed value equipment classification module is connected to the data acquisition module and is used to classify the data collected by the data acquisition module according to the equipment type;

[0015] A data preprocessing module, connected to the fixed value equipment classification module, is used to preprocess the data classified by the fixed value equipment classification module;

[0016] The feature encoding modules are connected to the data preprocessing modules respectively, and are used to encode the non-numerical features in the data preprocessing modules and convert them into numerical features;

[0017] A feature extraction module, connected to the feature encoding module, is used to extract features from the data encoded by different device types;

[0018] A feature construction module, connected to the feature extraction module, is used to combine the features extracted by the feature extraction module to form a new feature;

[0019] The model construction module is connected to the feature extraction module, the feature construction module, and the feature encoding module respectively, and is used to train and test the model according to the data corresponding to the features obtained by the feature extraction module and the feature construction module according to different voltage levels and different equipment types, so as to obtain the relay protection setting calculation model corresponding to each voltage level and equipment type;

[0020] The fixed value result calculation output module is connected to the data acquisition module and the model construction module respectively. It is used to perform fixed value calculation on the substation to be calculated according to the equipment type and parameter information of the substation to be calculated collected by the data acquisition module and the corresponding relay protection fixed value calculation model constructed by the model construction module, obtain the fixed value result, and output it for display.

[0021] Furthermore, preferably, the specific contents collected by the data collection module include equipment type, parameter information and constant value calculation results corresponding to the parameters.

[0022] Furthermore, preferably, in the fixed value equipment classification module, the equipment types include substations, capacitors, busbars, lines, station transformers, grounding transformers, and backup protection equipment.

[0023] Furthermore, preferably, in the data preprocessing module, outliers are removed and missing values ​​are filled with 0.

[0024] Further, preferably, in the feature extraction module, the correlation between each parameter data type and the fixed value is first calculated, and the parameter data type with a correlation coefficient greater than 0.5 is selected as the primary screening feature;

[0025] Then, the initial screening features are first subjected to principal component analysis or recursive feature elimination to reduce their dimensions, and the features after dimension reduction are the final extracted features.

[0026] Furthermore, preferably, in the feature construction module, an aggregation operation is performed on the features extracted by the feature extraction module to create new features; the aggregation method adopts feature mapping to perform aggregation, and maps the original extracted feature data to a new feature space.

[0027] Furthermore, preferably, in the feature encoding module, non-numerical features are converted into numerical features by one-hot encoding, thereby converting the features into a binary matrix.

[0028] Furthermore, preferably, in the model construction module, the model specifically adopts the Tensorflow model to perform fixed value calculation model training to complete the construction of the substation protection fixed value calculation model.

[0029] Furthermore, preferably, in the model building module, the ratio of the number of data samples used for training and testing is 2:1.

[0030] The parameter information collected by the data acquisition module of the present invention includes manufacturer, rated power, impedance voltage, high-voltage side voltage regulation range, medium-voltage side voltage regulation range, high-voltage side calculated impedance, medium-voltage side calculated impedance, low-voltage side calculated impedance, high-voltage side TA ratio, high backup TA ratio, ratio braking coefficient, zero-sequence protection, extreme low voltage, capacitor parameters, reactor parameters, rated power, and rated current.

[0031] The present invention preferably adopts the Tensorflow model for fixed value calculation model training, but is not limited to this. Other neural network models that can achieve accurate calculation can also be used for fixed value calculation model training.

[0032] The present invention collects the fixed values ​​of each type of equipment every month and one year, and regards the fixed value attributes as the same, to construct a corresponding data set of equipment type-parameter-fixed value for subsequent feature construction.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] (1) Improve the efficiency and accuracy of relay protection setting calculation and shorten the calculation cycle;

[0035] (2) Reduce the degree of manual intervention and reduce human errors;

[0036] (3) Provide strong guarantee for the safe and stable operation of the power system.

[0037] (4) The existing manual calculation method has unclear rules and risk points for constant value calculation due to personnel changes and employment reasons. This method can avoid human factors and quickly complete the constant value calculation by inputting fixed parameter information. Before using this method, it took 3 days to calculate the constant value of a 110KV substation, but it only takes 1 hour after using this method. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a structural schematic diagram of the substation relay protection setting intelligent calculation system based on big data analysis of the present invention. DETAILED DESCRIPTION

[0039] The present invention is further described in detail below in conjunction with embodiments.

[0040] Those skilled in the art will appreciate that the following examples are only used to illustrate the present invention and should not be considered to limit the scope of the present invention. If no specific techniques or conditions are specified in the examples, the techniques or conditions described in the literature in the art or the product specifications are used. If the manufacturer of the materials or equipment used is not specified, they are all conventional products that can be purchased.

[0041] Example 1

[0042] The intelligent calculation system for substation relay protection setting based on big data analysis includes:

[0043] The data acquisition module 101 is used to collect historical substation protection setting value calculation and result data; it is also used to collect equipment type and parameter information of the substation to be calculated;

[0044] The fixed value equipment classification module 102 is connected to the data acquisition module 101 and is used to classify the data collected by the data acquisition module 101 according to the equipment type;

[0045] The data preprocessing module 103 is connected to the fixed value device classification module 102 and is used to preprocess the data classified by the fixed value device classification module 102;

[0046] The feature encoding module 106 is connected to the data preprocessing module 103 and is used to encode the non-numerical features in the data preprocessing module 103 and convert them into numerical features;

[0047] The feature extraction module 104 is connected to the feature encoding module 106 and is used to extract features from the data encoded by different device types;

[0048] The feature construction module 105 is connected to the feature extraction module 104 and is used to combine the features extracted by the feature extraction module 104 to form a new feature;

[0049] The model building module 107 is connected to the feature extraction module 104, the feature construction module 105, and the feature encoding module 106 respectively, and is used to train and test the model according to the data corresponding to the features obtained by the feature extraction module 104 and the feature construction module 105 according to different voltage levels and different equipment types, so as to obtain the relay protection setting calculation model corresponding to each voltage level and equipment type;

[0050] The constant value result calculation output module 108 is connected to the data acquisition module 101 and the model construction module 107 respectively, and is used to perform constant value calculation on the substation to be calculated according to the equipment type and parameter information of the substation to be calculated collected by the data acquisition module 101 and the corresponding relay protection constant value calculation model constructed by the model construction module 107, obtain the constant value result, and output it for display.

[0051] Example 2

[0052] The intelligent calculation system for substation relay protection setting based on big data analysis includes:

[0053] The data acquisition module 101 is used to collect historical substation protection setting value calculation and result data; it is also used to collect equipment type and parameter information of the substation to be calculated;

[0054] The fixed value equipment classification module 102 is connected to the data acquisition module 101 and is used to classify the data collected by the data acquisition module 101 according to the equipment type;

[0055] The data preprocessing module 103 is connected to the fixed value device classification module 102 and is used to preprocess the data classified by the fixed value device classification module 102;

[0056] The feature encoding module 106 is connected to the data preprocessing module 103 and is used to encode the non-numerical features in the data preprocessing module 103 and convert them into numerical features;

[0057] The feature extraction module 104 is connected to the feature encoding module 106 and is used to extract features from the data encoded by different device types;

[0058] The feature construction module 105 is connected to the feature extraction module 104 and is used to combine the features extracted by the feature extraction module 104 to form a new feature;

[0059] The model building module 107 is connected to the feature extraction module 104, the feature construction module 105, and the feature encoding module 106 respectively, and is used to train and test the model according to the data corresponding to the features obtained by the feature extraction module 104 and the feature construction module 105 according to different voltage levels and different equipment types, so as to obtain the relay protection setting calculation model corresponding to each voltage level and equipment type;

[0060] The constant value result calculation output module 108 is connected to the data acquisition module 101 and the model construction module 107 respectively, and is used to perform constant value calculation on the substation to be calculated according to the equipment type and parameter information of the substation to be calculated collected by the data acquisition module 101 and the corresponding relay protection constant value calculation model constructed by the model construction module 107, obtain the constant value result, and output it for display.

[0061] The specific contents collected by the data collection module 101 include equipment type, parameter information and constant value calculation results corresponding to the parameters.

[0062] In the fixed value equipment classification module 102, the equipment types include substations, capacitors, busbars, lines, station transformers, grounding transformers, and backup protection equipment.

[0063] In the data preprocessing module 103, outliers are removed and missing values ​​are filled with 0.

[0064] In the feature extraction module 104, the correlation between each parameter data type and the fixed value is first calculated, and the parameter data type with a correlation coefficient greater than 0.5 is selected as the primary screening feature;

[0065] Then, the initial screening features are first subjected to principal component analysis or recursive feature elimination to reduce their dimensions, and the features after dimension reduction are the final extracted features.

[0066] In the feature construction module 105, the features extracted by the feature extraction module 104 are aggregated to create new features. The aggregation method uses feature mapping to perform aggregation, mapping the original data to a new feature space. The new features include device categories, calculation coefficients, main parameters, and constants to reduce feature dimensions and retain important information of the data.

[0067] In the feature encoding module 106, non-numerical features are converted into numerical features by one-hot encoding, thereby converting the features into a binary matrix.

[0068] In the model construction module 107, the model specifically uses the Tensorflow model to train the fixed value calculation model to complete the construction of the substation protection fixed value calculation model.

[0069] In the model building module 107, the ratio of the number of data samples used for training and testing is 2:1.

[0070] Application Examples

[0071] This example collects samples of historical calculation data of protection settings of substations, organizes and classifies them, and uses a large model to train parameter calculation methods for different types of equipment. By giving parameters, the model automatically identifies the parameters, automatically calculates the settings of each type of equipment, and finally outputs the substation setting data.

[0072] The intelligent calculation system for substation relay protection setting based on big data analysis includes:

[0073] 1. Data acquisition module

[0074] Collect historical substation protection setting calculation and result data, including equipment type, parameter information (manufacturer, rated power, impedance voltage, high-voltage side voltage regulation range, medium-voltage side voltage regulation range, high-voltage side calculated impedance, medium-voltage side calculated impedance, low-voltage side calculated impedance, high-voltage side TA ratio, high backup TA ratio, ratio braking coefficient, zero-sequence protection, extreme low voltage, capacitor parameters, reactor parameters, rated power, rated current), and setting information.

[0075] 2. Fixed value equipment classification module

[0076] According to the existing substation equipment, the collected data is classified, mainly into substations, capacitors, bus differentials, lines, station transformers, grounding transformers, and backup protection equipment.

[0077] 3. Data preprocessing module

[0078] Remove outliers, fill missing values ​​in sample data with 0, and summarize constant value calculation records of the same equipment to extract features of similar equipment.

[0079] 4. Feature Encoding Module

[0080] The non-numerical features preprocessed by the data preprocessing module are converted into numerical features using one-hot encoding, and then converted into a binary matrix for subsequent calculations.

[0081] 5. Feature extraction module

[0082] Extract features that can represent the essence of the data from the collected data, and these features will be used to train the machine learning model. In the substation relay protection setting intelligent calculation system based on big data analysis, the main task of the feature extraction module is to extract key features from the collected historical calculation data, such as equipment parameters, system status, etc., in order to prepare for subsequent model training.

[0083] (1) Equipment parameter characteristics

[0084] Voltage level: 110KV, 35KV, 220KV;

[0085] Equipment types: substation, capacitor, busbar, line, station transformer, grounding transformer, backup protection equipment.

[0086] Equipment parameters: manufacturer, rated power, impedance voltage, high-voltage side voltage regulation range, medium-voltage side voltage regulation range, high-voltage side calculated impedance, medium-voltage side calculated impedance, low-voltage side calculated impedance, high-voltage side TA ratio, high-backup TA ratio, ratio braking coefficient, zero-sequence protection, extreme low voltage, capacitor parameters, reactor parameters, rated power, rated current.

[0087] (2) Feature screening

[0088] The purpose of feature screening is to remove irrelevant or redundant features to reduce the complexity of the model and improve the performance of the model. Common methods include:

[0089] Correlation analysis: By calculating the correlation between features and target variables, features with high correlation with target variables are selected.

[0090] Principal Component Analysis (PCA): Extracts the most important features through dimensionality reduction techniques.

[0091] Recursive Feature Elimination (RFE): Select features by recursively reducing the number of features.

[0092] 6. Feature Construction Module

[0093] Generate new features from existing data to help improve the performance of the model. Create new features by combining two or more features, such as the combination of device type and voltage level. Aggregate a set of features using feature mapping to map the original extracted feature data to a new feature space to reduce feature dimensions and retain important information in the data.

[0094] 7. Model building module

[0095] According to different voltage levels and different equipment types, different calculation models are constructed, training sets and result sets are constructed, the training set data is used to calculate the model and compare the result set data, and the calculation model is adjusted and optimized according to the correctness of the test results. The model specifically uses the TensorFlow model for fixed value calculation model training.

[0096] In order to ensure the accuracy and reliability of relay protection setting calculation, a visual interface was developed to evaluate and optimize the setting calculation model. The interface allows users to pass the prepared result set data to the system, and the system calls the corresponding algorithm model for calculation. The calculated setting will be compared with the actual setting value to verify the accuracy of the calculation result. In this way, the deficiencies of the model can be continuously discovered, and the algorithm model can be corrected and optimized accordingly, so as to achieve accurate calculation of the setting.

[0097] Detailed steps:

[0098] Data input: Users upload or select pre-prepared test set data through a visual interface.

[0099] Model call: The system automatically calls the trained algorithm model and performs fixed value calculations on the training set data.

[0100] Result comparison: The calculated constant value is compared with the set value of the result set, and the system automatically calculates evaluation indicators such as error and accuracy.

[0101] Model optimization: Based on the comparison results, the model hyperparameters are tuned according to these recommendations.

[0102] Iterative improvement: Repeat the above process until the model's calculation results reach satisfactory accuracy and reliability.

[0103] 8. Fixed value result calculation output module

[0104] The calculated protection settings will be presented to the user in an intuitive and easy-to-understand way, ensuring that the user can quickly understand and use the information. The system provides a variety of output formats, including charts, lists, etc., to meet the needs of different users. In addition, the system also provides download or export functions, allowing users to save the calculation results locally or share them with other relevant personnel.

[0105] Detailed features:

[0106] Result display: The calculated protection settings are displayed in the form of charts and lists, including the setting details of each device and the historical setting calculation results of similar devices.

[0107] Result interpretation: Provides explanations and instructions for fixed value calculations to help users understand the calculation process and the meaning of the results.

[0108] Download function: allows users to download calculation results into common file formats, such as PDF, Excel, etc.

[0109] Export function: supports exporting calculation results to other systems or platforms for further analysis and use.

[0110] User-defined: Allows users to customize the display and output format of results to meet specific needs and preferences.

[0111] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. The intelligent calculation system for relay protection setting value of substation based on big data analysis is characterized by: include: The data collection module (101) is used to collect historical substation protection setting value calculation and result data; and is also used to collect equipment type and parameter information of the substation to be calculated; A fixed value equipment classification module (102), connected to the data acquisition module (101), is used to classify the data collected by the data acquisition module (101) according to equipment type; A data preprocessing module (103) is connected to the fixed value equipment classification module (102) and is used to preprocess the data classified by the fixed value equipment classification module (102); The feature encoding modules (106) are respectively connected to the data preprocessing modules (103) and are used to encode the non-numerical features in the data preprocessing modules (103) and convert them into numerical features; A feature extraction module (104), connected to the feature encoding module (106), is used to extract features from the data encoded by different device types; A feature construction module (105), connected to the feature extraction module (104), is used to combine the features extracted by the feature extraction module (104) to form a new feature; The model construction module (107) is connected to the feature extraction module (104), the feature construction module (105), and the feature encoding module (106) respectively, and is used to train and test the model according to different voltage levels and different equipment types, based on the data corresponding to the features obtained by the feature extraction module (104) and the feature construction module (105), so as to obtain a relay protection setting calculation model corresponding to each voltage level and equipment type; The fixed value result calculation output module (108) is connected to the data acquisition module (101) and the model construction module (107) respectively, and is used to perform fixed value calculation on the substation to be calculated based on the equipment type and parameter information of the substation to be calculated collected by the data acquisition module (101) and the corresponding relay protection fixed value calculation model constructed by the model construction module (107), obtain the fixed value result, and output it for display.

2. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: The specific contents collected by the data collection module (101) include equipment type, parameter information and constant value calculation results corresponding to the parameters.

3. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the fixed value equipment classification module (102), the equipment types include substation, capacitor, busbar, line, station transformer, grounding transformer, and backup protection equipment.

4. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the data preprocessing module (103), outliers are removed and missing values ​​are filled with 0.

5. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the feature extraction module (104), the correlation between each parameter data type and the fixed value is first calculated, and the parameter data type with a correlation coefficient greater than 0.5 is selected as the primary screening feature; Then, the initial screening features are first subjected to principal component analysis or recursive feature elimination to reduce their dimensions, and the features after dimension reduction are the final extracted features.

6. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the feature construction module (105), the features extracted by the feature extraction module (104) are aggregated to create new features.

7. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the feature encoding module (106), non-numerical features are converted into numerical features using one-hot encoding, thereby converting the features into a binary matrix.

8. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the model construction module (107), the model specifically uses the Tensorflow model to train the constant value calculation model to complete the construction of the substation protection constant value calculation model.

9. The substation relay protection setting intelligent calculation system based on big data analysis according to claim 1 is characterized in that: In the model building module (107), the ratio of the number of data samples used for training and testing is 2:1.

Images