Substation auxiliary control system for pre-judging bus fault based on neural network

By designing an auxiliary control system based on neural network in the substation, using data collection and multiple prediction modules to detect bus failures in advance, the problem of failures in the existing technology cannot be detected in advance, and the risk warning and loss of the power system is achieved avoid.

CN119044835AInactive Publication Date: 2024-11-29FUJIAN YOUDI ELECTRIC POWER TECH
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411536937.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-31
Publication Date
2024-11-29
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art cannot detect substation bus failure in advance, resulting in considerable losses in the power system when the fault occurs.

Method used

A substation auxiliary control system based on neural network was designed, including a data collection module, a preliminary prediction module, a fault prediction module and a fault warning module. By collecting temperature, humidity, current, voltage and bus appearance image data, the trained CNN is used. The neural network model performs preliminary prediction and secondary prediction, and combines multiple data to predict faults to achieve early warning.

Benefits of technology

It realizes early detection and early warning of busbar failures, avoids power losses, and improves the operational safety of the substation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119044835A_ABST
    Figure CN119044835A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of substation bus fault prediction, and discloses a substation auxiliary control system for pre-judging a bus fault based on a neural network. The substation auxiliary control system comprises a data collection module, a preliminary prediction module, a fault prediction module and a fault early warning module, the data collection module comprises a temperature collection unit, a humidity collection unit, a current collection unit, a voltage collection unit, a bus appearance image collection unit and a maintenance data collection unit. The temperature collection unit is used for collecting temperature data at the wiring position. The system performs normalization processing on data in a preliminary prediction data set when receiving the data, then sends the data to a trained CNN neural network model for preliminary prediction, and then calls each component in a fault prediction module for secondary prediction. And the temperature data, the humidity data, the image data of the appearance of the bus and the maintenance data are combined to pre-judge the fault of the bus in multiple aspects, so that the risk is processed in advance, and the power loss is avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of transformer substation bus fault prediction, and in particular to a transformer substation auxiliary control system for predicting bus faults based on a neural network. Background Art

[0002] Substations are an indispensable part of the power system. The operational safety of substations is related to the stability of the power system. The auxiliary control system of the substation, including the fire protection system, video surveillance system, lighting control system, water supply and drainage system, and air conditioning system, is an important support system of the substation. It is responsible for the daily safety and reliable operation and maintenance of the substation, and is related to the operational safety of the substation. Therefore, there are high requirements for the auxiliary control system;

[0003] The busbar is an essential component used to distribute electrical energy from transformers, generators or other power sources to different circuits or loads. They connect multiple power sources and loads in the power system together to ensure that electrical energy can be efficiently distributed to where it is needed, and provide an electrical connection point so that various electrical equipment in the substation (such as circuit breakers, switches, disconnectors, etc.) can be connected to each other. This makes it easy to operate and control the circuit. At the same time, the busbar gathers current from different sources and then distributes it to various branches or loads. It can carry large currents, so it is usually made of materials with good conductivity (such as copper or aluminum);

[0004] However, since busbar faults at this stage are all checked during maintenance or discovered after problems occur, they will cause considerable power losses. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In view of the deficiencies in the prior art, the present invention provides a substation auxiliary control system for predicting busbar faults based on a neural network, which has the advantages of detecting busbar faults in advance and solves the above-mentioned technical problems.

[0007] (II) Technical solution

[0008] To achieve the above object, the present invention provides the following technical solutions: a substation auxiliary control system for predicting busbar faults based on a neural network, comprising a data collection module, a preliminary prediction module, a fault prediction module and a fault warning module;

[0009] The data collection module includes a temperature collection unit, a humidity collection unit, a current collection unit, a voltage collection unit, a bus appearance image collection unit, and a maintenance data collection unit. The humidity collection unit is used to collect current humidity data of the environment, the current collection unit is used to collect current and historical current data passing through the bus, the voltage collection unit is used to collect current and historical voltage data passing through the bus, the temperature collection unit is used to collect current and historical temperature data of the bus, the bus appearance image collection unit is used to collect image data of the bus appearance, and the maintenance data collection unit is used to collect maintenance data of the bus. The data collection module sends the current and historical bus temperature data, the current and historical bus current data, and the current and historical bus voltage data to the preliminary prediction module, waits for the feedback result of the preliminary prediction module, and then determines whether to send the current and historical bus temperature data, current humidity data, bus appearance image data, and maintenance data to the fault prediction module according to the feedback result;

[0010] The preliminary prediction module is used to perform preliminary prediction of bus faults based on temperature data, current and historical current data passing through the bus, and current and historical voltage data passing through the bus. If the prediction result is abnormal, the feedback result of the fault is sent to the data collection module. If the prediction result is no abnormality, the feedback result of no abnormality is sent to the data collection module.

[0011] The fault prediction module is used to combine the current temperature data, the current humidity data, the image data of the busbar appearance and the maintenance data to predict the fault. If the prediction result is that a fault will occur, the fault warning module is called to warn. If the result is that no fault will occur, the operation is terminated.

[0012] The fault warning module is used to warn the prediction result fed back by the fault prediction module.

[0013] As a preferred technical solution of the present invention, the expressions of the current and historical bus temperature data collected by the temperature collection unit, the current and historical bus current data collected by the current collection unit, and the current and historical bus voltage data collected by the voltage collection unit are as follows:

[0014] in, , , , , , Respectively Temperature data at historical moments and current moment Temperature data, , , , , , Respectively Current data at historical moments and current moment Current data, , , , , , Respectively Voltage data at historical moments and current moment Voltage data, Represents the preliminary prediction dataset.

[0015] As a preferred technical solution of the present invention, the preliminary prediction module is equipped with a trained CNN neural network model. When receiving the data in the preliminary prediction data set, it is normalized and sent to the trained CNN neural network model for prediction, and the prediction result is given. When the trained CNN neural network model predicts that the probability of abnormality is greater than or equal to 0.3, the abnormality is fed back and the feedback result is sent to the data collection module.

[0016] As a preferred technical solution of the present invention, the fault prediction module includes an image analysis unit, a temperature and humidity analysis unit, a maintenance data analysis unit and a comprehensive analysis unit. After the analysis is completed, the image analysis unit, the temperature and humidity analysis unit and the maintenance data analysis unit send the results of the calculation and analysis to the comprehensive analysis unit for comprehensive analysis.

[0017] As a preferred technical solution of the present invention, the specific analysis process of the temperature and humidity analysis unit is as follows:

[0018] A1. The temperature data in the bus exceeds the safety threshold of the temperature data The point at which the humidity data in the bus exceeds the safety threshold of the humidity data is marked. Mark the points;

[0019] A2. Select the coincident points , the overlapped points mean that the safety threshold of the temperature data is met at the same time and exceeds the safety threshold of humidity data Points

[0020] A3. Select the coincident points Sent to the image analysis unit for analysis.

[0021] As a preferred technical solution of the present invention, the specific steps of the image analysis unit analyzing the image data of the busbar appearance are as follows:

[0022] B1. Select the coincident points The image data at the location is intercepted;

[0023] B2. Analyze the image data at the corresponding coincidence points of the busbar and send them to the comprehensive analysis unit.

[0024] As a preferred technical solution of the present invention, the expression for analyzing the image data of the coincident points in step B2 is as follows:

[0025] in, represents the damage rate, Indicates The first of the coincident points The area of ​​the region where the chromaticity is inconsistent with the busbar chromaticity, Expressing the Coincident points Sum the areas of the regions, express The sum of the areas of all the coincident points among the coincident points is calculated. Indicates the total number of coincident points. Indicates The total area is intercepted at the overlapping points.

[0026] As a preferred technical solution of the present invention, the analysis expression of the temperature data and humidity data by the temperature and humidity analysis unit is as follows:

[0027] in, and Respectively represent the rate of change of temperature data and humidity data, Indicates the current temperature data. Indicates the safety threshold corresponding to the temperature data, Indicates the humidity data at the current moment. Indicates the safety threshold corresponding to the humidity data.

[0028] As a preferred technical solution of the present invention, the specific expression for the calculation and analysis performed by the maintenance data analysis unit is as follows:

[0029] in, Indicates the damage rate of maintenance data. Indicates the number of maintenance days. represents a natural constant, Represents the coefficient of variation.

[0030] As a preferred technical solution of the present invention, the expression for comprehensive analysis calculation performed by the comprehensive analysis unit is as follows:

[0031] in, represents the comprehensive abnormality rate, and Respectively represent the rate of change of temperature data and humidity data, Indicates the damage rate of maintenance data. Indicates the damage rate.

[0032] Compared with the prior art, the present invention provides a substation auxiliary control system for predicting bus faults based on a neural network, which has the following beneficial effects:

[0033] The present invention performs normalization processing on the data in the preliminary prediction data set when receiving it, and then sends it to the trained CNN neural network model for preliminary prediction, and then calls various components in the fault prediction module for secondary prediction, combining temperature data, humidity data, image data of the bus appearance and maintenance data to predict the bus fault in many aspects, thereby realizing early processing of risks and avoiding power loss. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the system framework of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] See also Figure 1 , a substation auxiliary control system for predicting busbar faults based on neural networks, including a data collection module, a preliminary prediction module, a fault prediction module and a fault warning module;

[0037] The data collection module includes a temperature collection unit, a humidity collection unit, a current collection unit, a voltage collection unit, a bus appearance image collection unit, and a maintenance data collection unit. The humidity collection unit is used to collect the current humidity data of the environment, the current collection unit is used to collect the current and historical current data passing through the bus, the voltage collection unit is used to collect the current and historical voltage data passing through the bus, the temperature collection unit is used to collect the current and historical temperature data of the bus, the bus appearance image collection unit is used to collect the image data of the bus appearance, and the maintenance data collection unit is used to collect the maintenance data of the bus;

[0038] in, , , , , , Respectively Temperature data at historical moments and current moment Temperature data, , , , , , Respectively Current data at historical moments and current moment Current data, , , , , , Respectively Voltage data at historical moments and current moment Voltage data, represents the preliminary prediction data set;

[0039] The data collection module sends the current and historical bus temperature data, the current and historical bus current data, and the current and historical bus voltage data to the preliminary prediction module, and waits for the feedback result of the preliminary prediction module;

[0040] The preliminary prediction module is used to make a preliminary prediction of bus faults based on temperature data, current and historical current data passing through the bus, and current and historical voltage data passing through the bus. If the prediction result is abnormal, the feedback result of the fault is sent to the data collection module. If the prediction result is no abnormality, the feedback result of no abnormality is sent to the data collection module. The preliminary prediction module is equipped with a trained CNN neural network model. When receiving the data in the preliminary prediction data set, it is normalized and sent to the trained CNN neural network model for prediction, and the prediction result is given. When the prediction result of the trained CNN neural network model: the probability of abnormality is greater than or equal to 0.3, the abnormality is fed back and the feedback result is sent to the data collection module;

[0041] The process of fault prediction by CNN neural network model is as follows:

[0042] (1) Data standardization

[0043] in, represents the standardized current, represents the standardized voltage, represents the standardized temperature, is the input window, and the window length is , ;

[0044] (2) Construct and train a CNN neural network model, which includes a convolutional layer, a pooling layer, a fully connected layer, and an output layer. Their expressions are as follows:

[0045] in, Indicates The convolutional feature map of the layer, represents the RELU activation function, Indicates The weights of the convolution kernels of the layers, is the bias, Represents the convolution operation;

[0046] in, Represents the output value of the pooling layer, Express Perform maximum pooling operation;

[0047] in, represents the output of the fully connected layer, represents the RELU activation function, represents the weight matrix of the fully connected layer, represents the bias vector;

[0048] in, Represents the output prediction result, represents the sigmoid activation function, represents the output of the fully connected layer, represents the bias of the output layer, represents the weight matrix of the output layer;

[0049] The data collection module determines whether to send the current and historical bus temperature data, current humidity data, bus appearance image data and maintenance data to the fault prediction module according to the feedback results;

[0050] The fault prediction module is used to combine the current temperature data, the current humidity data, the image data of the busbar appearance and the maintenance data to predict the fault. The fault prediction module includes an image analysis unit, a temperature and humidity analysis unit, a maintenance data analysis unit and a comprehensive analysis unit. After the analysis is completed, the image analysis unit, the temperature and humidity analysis unit and the maintenance data analysis unit send the results of the calculation and analysis to the comprehensive analysis unit for comprehensive analysis. If the prediction result is that a fault will occur, the fault warning module is called for warning. If the result is that no fault will occur, the module is terminated and waits for re-prediction.

[0051] Temperature and humidity analysis unit analyzes temperature and humidity , At the same time, the temperature data in the bus exceeds the safety threshold of the temperature data The point at which the humidity data in the bus exceeds the safety threshold of the humidity data is marked. Mark the points, that is, the values ​​greater than 0 in the calculation results, and select the values ​​that meet the safety threshold of exceeding the temperature data. and exceeds the safety threshold of humidity data The point of , a total of N, the selected coincident points The image analysis unit selects the coincident points. The image data at the corresponding coincidence point of the busbar is intercepted and analyzed to obtain , where it is obvious that there are discoloration, cracking and other problems, and these are sent to the comprehensive analysis unit;

[0052] in, and Respectively represent the rate of change of temperature data and humidity data, Indicates the current temperature data. Indicates the safety threshold corresponding to the temperature data, Indicates the humidity data at the current moment. Indicates the safety threshold corresponding to the humidity data, represents the damage rate, Indicates The total area of ​​the intercepted part is obtained by the overlapping points. Indicates The first of the coincident points The area of ​​the region where the chromaticity is inconsistent with the busbar chromaticity, Expressing the Coincident points Sum the areas of the regions, express The sum of the areas of all the coincident points among the coincident points is calculated. Indicates the total number of coincident points;

[0053] Maintenance data analysis unit combined with maintenance days (Here the days are calculated in months. =1, it means that one month has passed since the last maintenance. If it is less than one month, the maintenance data damage rate is calculated by dividing the number of days in the month by the total number of days in the month (retain two decimal places). ;

[0054] in, Indicates the damage rate of maintenance data. Indicates the number of maintenance days. represents a natural constant, represents the coefficient of variation;

[0055] The expression for comprehensive analysis calculation performed by the comprehensive analysis unit is as follows:

[0056] in, represents the comprehensive abnormality rate, and Respectively represent the rate of change of temperature data and humidity data, Indicates the damage rate of maintenance data. Indicates the damage rate;

[0057] The fault warning module is used to warn the prediction results fed back by the fault prediction module.

[0058] Embodiment 1:

[0059] For the preliminary prediction module, there are the following test data; Table 1 Preliminary prediction module experimental data table time Current I (A) Voltage V (V) Temperature T(℃) 1 15.2 230.0 45.0 2 15.8 231.2 45.5 3 16.1 230.5 46.0 4 15.9 229.8 45.7 5 16.4 230.1 46.3 6 16.7 231.0 46.5 7 16.5 231.3 46.7 8 16.9 232.1 47.0 9 17.2 232.0 47.2 10 17.5 233.0 47.5

[0060] Calculate the mean and standard deviation ;

[0061] (3) Data standardization

[0062] Finally, the following standardized data is obtained: Table 2 Standardized data table time Current I Voltage V Temperature T 1 -1.69 -1.00 -1.24 2 -0.86 0.00 -0.69 3 -0.44 -0.58 -0.15 4 -0.71 -1.17 -0.48 5 -0.03 -0.92 0.17 6 0.38 -0.17 0.39 7 0.11 0.08 0.61 8 0.68 0.75 0.93 9 1.07 0.67 1.15 10 1.47 1.50 1.48

[0063] Assume the window size is 3, and each set of data is input into the model as a window sample:

[0064] And so on, we get 8 sets of data , these 8 sets of data are sent to the trained CNN network for calculation; Table 3 CNN network output table Window Sample Failure probability <![CDATA[X1]]> 0.05 <![CDATA[X2]]> 0.10 <![CDATA[X3]]> 0.20 <![CDATA[X4]]> 0.15 <![CDATA[X5]]> 0.08 <![CDATA[X6]]> 0.25 <![CDATA[X7]]> 0.30 <![CDATA[X8]]> 0.35

[0065] When the output is greater than or equal to 0.3, the feedback is abnormal, and the fault prediction module is called for secondary prediction;

[0066] Embodiment 2:

[0067] The calculation results of the fault prediction module will be given here:

[0068] Image analysis computing unit: , the specific data are as follows: Table 4 Image analysis unit experimental data table

[0069]

[0070] For temperature and humidity analysis unit:

[0071] For the maintenance data analysis unit, take =1.365:

[0072] For the comprehensive analysis unit:

[0073] At this time, an early warning is issued to remind the staff to carry out maintenance;

[0074] In summary, the present invention normalizes the data in the preliminary prediction data set when receiving it, and then sends it to the trained CNN neural network model for preliminary prediction. It then calls each component in the fault prediction module for secondary prediction, and predicts the busbar fault from multiple aspects such as image, temperature and humidity, and maintenance status, thereby achieving early risk processing and avoiding power loss.

[0075] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A substation auxiliary control system for predicting busbar faults based on a neural network, characterized by: It includes data collection module, preliminary prediction module, fault prediction module and fault warning module; The data collection module includes a temperature collection unit, a humidity collection unit, a current collection unit, a voltage collection unit, a bus appearance image collection unit, and a maintenance data collection unit. The humidity collection unit is used to collect current humidity data of the environment, the current collection unit is used to collect current and historical current data passing through the bus, the voltage collection unit is used to collect current and historical voltage data passing through the bus, the temperature collection unit is used to collect current and historical temperature data of the bus, the bus appearance image collection unit is used to collect image data of the bus appearance, and the maintenance data collection unit is used to collect maintenance data of the bus. The data collection module sends the current and historical bus temperature data, the current and historical bus current data, and the current and historical bus voltage data to the preliminary prediction module, waits for the feedback result of the preliminary prediction module, and then determines whether to send the current and historical bus temperature data, current humidity data, bus appearance image data, and maintenance data to the fault prediction module according to the feedback result; The preliminary prediction module is used to perform preliminary prediction of bus faults based on temperature data, current and historical current data passing through the bus, and current and historical voltage data passing through the bus. If the prediction result is abnormal, the feedback result of the fault is sent to the data collection module. If the prediction result is no abnormality, the feedback result of no abnormality is sent to the data collection module. The fault prediction module is used to combine the current temperature data, the current humidity data, the image data of the busbar appearance and the maintenance data to predict the fault. If the prediction result is that a fault will occur, the fault warning module is called to warn. If the result is that no fault will occur, the operation is terminated. The fault warning module is used to warn the prediction result fed back by the fault prediction module.

2. According to claim 1, a substation auxiliary control system for predicting bus failure based on neural network is characterized by: The expressions of the current and historical bus temperature data collected by the temperature collection unit, the current and historical bus current data collected by the current collection unit, and the current and historical bus voltage data collected by the voltage collection unit are as follows: ; in, , , , , , Respectively Temperature data at historical moments and current moment Temperature data, , , , , , Respectively Current data at historical moments and current moment Current data, , , , , , Respectively Voltage data at historical moments and current moment Voltage data, Represents the preliminary prediction dataset.

3. According to claim 1, a substation auxiliary control system for predicting bus failure based on neural network is characterized by: The preliminary prediction module is equipped with a trained CNN neural network model. When receiving the data in the preliminary prediction data set, it is normalized and sent to the trained CNN neural network model for prediction, and a prediction result is given. When the prediction result of the trained CNN neural network model is: the probability of abnormality is greater than or equal to 0.3, the abnormality is fed back and the feedback result is sent to the data collection module.

4. According to claim 1, a substation auxiliary control system for predicting bus failure based on neural network is characterized by: The fault prediction module includes an image analysis unit, a temperature and humidity analysis unit, a maintenance data analysis unit and a comprehensive analysis unit. After the analysis is completed, the image analysis unit, the temperature and humidity analysis unit and the maintenance data analysis unit send the results of the calculation and analysis to the comprehensive analysis unit for comprehensive analysis.

5. The substation auxiliary control system for predicting bus failure based on neural network according to claim 4 is characterized by: The specific analysis process of the temperature and humidity analysis unit is as follows: A1. The temperature data in the bus exceeds the safety threshold of the temperature data The point at which the humidity data in the bus exceeds the safety threshold of the humidity data is marked. Mark the points; A2. Select the coincident points , the overlapped points mean that the safety threshold of the temperature data is met at the same time and exceeds the safety threshold of humidity data Points; A3. Select the coincident points Sent to the image analysis unit for analysis.

6. The substation auxiliary control system for predicting bus failure based on neural network according to claim 5 is characterized by: The specific steps of the image analysis unit analyzing the image data of the busbar appearance are as follows: B1. Select the coincident points The image data at the location is intercepted; B2. Analyze the image data at the corresponding coincidence points of the busbar and send them to the comprehensive analysis unit.

7. The substation auxiliary control system based on neural network for predicting bus fault according to claim 6 is characterized by: The expression for analyzing the image data of the coincident points in step B2 is as follows: ; in, Indicates the damage rate, Indicates The first of the coincident points The area of ​​the region where the chromaticity is inconsistent with the busbar chromaticity, Indicates Coincident points Sum the areas of the regions, express The sum of the areas of all the coincident points among the coincident points is calculated. Indicates the total number of coincident points. Indicates The total area is intercepted at the overlapping points.

8. The substation auxiliary control system based on neural network for predicting bus faults according to claim 5 is characterized by: The analysis expression of the temperature data and humidity data by the temperature and humidity analysis unit is as follows: ; ; in, and Respectively represent the rate of change of temperature data and humidity data, Indicates the current temperature data. Indicates the safety threshold corresponding to the temperature data, Indicates the humidity data at the current moment. Indicates the safety threshold corresponding to the humidity data.

9. The substation auxiliary control system for predicting bus failure based on neural network according to claim 4 is characterized by: The specific expression for the calculation and analysis performed by the maintenance data analysis unit is as follows: ; in, Indicates the damage rate of maintenance data. Indicates the number of maintenance days. represents a natural constant, Represents the coefficient of variation.

10. The substation auxiliary control system based on neural network for predicting bus fault according to claim 8, characterized in that: The expression for comprehensive analysis calculation performed by the comprehensive analysis unit is as follows: ; in, represents the comprehensive abnormality rate, and Respectively represent the rate of change of temperature data and humidity data, Indicates the damage rate of maintenance data. Indicates the damage rate.

Citation Information

Patent Citations

  • Intelligent bus on-line monitoring and condition evaluation system

    CN103163389A

  • Equipment fault prediction method and device, computer equipment and storage medium

    CN117370848A

  • Cable fault pre-judgment analysis method based on neural network algorithm

    CN117609841A

  • Collection line fault early warning system and method

    CN117686829A

  • Fault recognition method and system based on neural network self-learning

    WO2015158198A1