Distributed microgrid data processing method and system
By building distributed computing nodes in the microgrid and using deep learning models for data preprocessing and confidence verification, the problems of complexity and insufficient computing power utilization of microgrid data analysis are solved, and efficient data analysis and coordinated control are achieved.
Patent Information
- Application Number
- CN202411173010.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-26
- Publication Date
- 2024-12-17
AI Technical Summary
When the data volume of microgrid is large, data analysis is complex, artificial operation steps are many, and the computing power of edge equipment is not effectively utilized.
The distributed computing nodes and deep learning models are used to preprocess and confidence verification of the data at the edge of microgrid equipment. The built-in distributed computing nodes are used to verify the data, and the deep learning model is used to reduce human intervention and improve data analysis efficiency.
It reduces human intervention, improves data analysis accuracy, effectively utilizes the computing power of edge devices, reduces waste of computing resources, and realizes coordinated control of the microgrid.
Smart Images
Figure CN119149226A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of microgrids, and relates to a distributed microgrid data processing method and system. Background Art
[0002] A smart microgrid is a smaller, decentralized, independent system that integrates distributed power sources, energy storage devices, energy conversion devices, related loads, and monitoring and protection devices into a small power generation and distribution system. It is an autonomous system capable of self-control, protection, and management, and can operate with an external power grid or in isolation.
[0003] However, the amount of data that microgrids obtain from edge devices is very large. When the data volume is large, the data analysis is more restricted and the steps required for manual operation are more complicated. At the same time, the computing power in edge devices cannot be well utilized. Summary of the invention
[0004] In order to solve the problems existing in the background technology, the present invention proposes a distributed microgrid data processing method and system.
[0005] In order to achieve the above object, the technical solution adopted by the present invention is as follows:
[0006] On the one hand, the present invention provides a distributed microgrid data processing method, comprising:
[0007] Responding to a data acquisition instruction issued by the data acquisition interface, acquiring relevant operating data of the microgrid edge device;
[0008] The microgrid data processing center pre-processes the microgrid edge device data and cleans the microgrid edge device data;
[0009] The microgrid data processing center uses a deep learning model to verify the confidence of the microgrid edge device data after preprocessing;
[0010] The microgrid data processing center sends the microgrid operation mode data to the microgrid edge device through the data transmission link;
[0011] The microgrid edge device verifies its own data based on the built-in distributed computing nodes to determine whether it complies with the microgrid operation mode sent by the microgrid data processing center.
[0012] Furthermore, the method for the microgrid data processing center to preprocess the microgrid edge device data is:
[0013] S1. Traverse the original microgrid edge device data, find blank values and missing values, and record the location of blank data;
[0014] S2. Based on other data of a single microgrid edge device, predict blank values and missing values and complete the data;
[0015] S3. Classify the completed microgrid edge device data, and classify each attribute value of the microgrid edge device data according to the attribute name of the microgrid edge device data.
[0016] Furthermore, the method by which the microgrid data processing center verifies the confidence of the pre-processed microgrid edge device data through a deep learning model is as follows:
[0017] Submit historical microgrid edge device data to the deep learning model, and the deep learning model obtains the deep learning model judgment logic through training;
[0018] The existing microgrid edge device data is judged by deep learning model judgment logic to determine the data confidence.
[0019] Furthermore, the method for training the deep learning model is:
[0020] The deep learning model reads data from multiple microgrid edge devices and extracts multiple credibility indicators based on the distribution intervals of multiple attribute data of the microgrid edge devices;
[0021] The credibility index corresponds to the attributes of the edge device data acquired by the data acquisition module, and the attributes include attribute data such as time and spatial distance;
[0022] According to the credibility index, the credibility is judged by combining the edge device data attributes with the corresponding data of the credibility index. The deep learning model is trained based on the existing data to obtain the initial threshold, which is used to exclude low-credibility data. The credibility of each data is calculated by combining the credibility of the edge device and the system threshold.
[0023] The deep learning model analyzes the span range of multiple attribute data of the microgrid edge device data, extracts the stability of each attribute data based on the span range of each attribute data, and calculates the weight of multiple attribute data of the microgrid edge device data based on the stability;
[0024] The deep learning model integrates the credibility of data obtained from multiple data sources according to weights to form a comprehensive confidence assessment.
[0025] Furthermore, the cost function of the deep learning model is:
[0026]
[0027] Where m is the number of training samples, is the value predicted by the deep learning model for parameter i, y iis the actual value of parameter i. By controlling the value range of E, the prediction effect of the deep learning model can be controlled.
[0028] Furthermore, the distributed computing node is a built-in component of each microgrid edge device.
[0029] On the other hand, the present invention provides a distributed microgrid data processing system, comprising:
[0030] The microgrid edge device has a built-in distributed computing node to match and verify its own data with the microgrid operation mode sent by the microgrid data processing center;
[0031] A data acquisition module, in response to a data acquisition instruction issued by a data acquisition interface, acquires relevant operating data of a microgrid edge device;
[0032] A data transmission module, responsible for data transmission between the microgrid edge device, the first data processing module and the second data processing module;
[0033] The first data processing module pre-processes the microgrid edge device data and cleans the microgrid edge device data;
[0034] The second data processing module verifies the confidence of the microgrid edge device data after preprocessing through a deep learning model.
[0035] Compared with the prior art, the present invention has the following beneficial effects:
[0036] The present invention uses a deep learning model to perform credibility analysis on data of different attributes in the microgrid edge device data, with less human intervention. The accuracy of the deep learning model's confidence analysis of the microgrid edge device data can be controlled by manually adjusting the cost function, thereby reducing human errors. The present invention also has built-in computing nodes in the microgrid edge device, which can effectively utilize the computing power of the microgrid edge device, reduce the waste of computing resources, and realize coordinated control of the entire microgrid. BRIEF DESCRIPTION OF THE DRAWINGS
[0037] Figure 1 It is a flow chart of the operation of the method of the present invention;
[0038] Figure 2 It is a system module architecture diagram of the present invention. DETAILED DESCRIPTION
[0039] 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.
[0040] like Figure 1-Figure 2 As shown, the technical solution adopted by the present invention is as follows:
[0041] On the one hand, if Figure 1 As shown, the present invention provides a distributed microgrid data processing method, comprising:
[0042] The microgrid data processing system issues a data collection instruction in response to the data acquisition instruction issued by the data acquisition interface. The data acquisition module in the microgrid data processing system acquires the relevant operation data of the microgrid edge device.
[0043] After obtaining the relevant operating data of the microgrid edge device, the microgrid data processing center preprocesses the microgrid edge device data and cleans the microgrid edge device data.
[0044] The method by which the microgrid data processing center preprocesses the microgrid edge device data is as follows:
[0045] S1. Traverse the original microgrid edge device data, find blank values and missing values, and record the location of the blank data. Only record the blank values and missing values to avoid distortion of the original microgrid edge device data when modifying abnormal values, resulting in failure of control of the original microgrid edge device.
[0046] S2. Based on other data of a single microgrid edge device, predict blank values and missing values and complete the data. Mark the missing data as the first data, combine the other data of other microgrid edge devices and the correlation between the first data of other microgrid edge devices to predict the data corresponding to the missing values and blank values.
[0047] S3. Classify the completed microgrid edge device data. Classify each attribute value of the microgrid edge device data according to the attribute name of the microgrid edge device data. This ensures that multiple attributes in the microgrid edge device data do not interfere with each other, ensures the orderliness of the data, and provides clear data for the later deep learning model training.
[0048] The microgrid data processing center receives the pre-processed microgrid edge device data and verifies the confidence of the pre-processed microgrid edge device data through a deep learning model.
[0049] The method by which the microgrid data processing center verifies the confidence of the pre-processed microgrid edge device data through a deep learning model is as follows:
[0050] Submit historical microgrid edge device data to the deep learning model, and the deep learning model obtains the deep learning model judgment logic through training.
[0051] The method for training the deep learning model is:
[0052] The deep learning model reads data from multiple microgrid edge devices and extracts multiple credibility indicators based on the distribution intervals of multiple attribute data of the microgrid edge device data. The deep learning model reads data from multiple attribute data distribution intervals of the microgrid edge device data by itself, reducing human intervention, effectively improving the accuracy of the deep learning model judgment logic, and also reducing the use of human resources.
[0053] The credibility index corresponds to the attributes of the edge device data acquired by the data acquisition module, and the attributes include time attributes, spatial distance attributes and other data.
[0054] According to the credibility indicators extracted by the deep learning model, the credibility is judged by combining the edge device data attributes with the corresponding data of the credibility indicators. The deep learning model is trained based on the historical microgrid edge device data to obtain the initial threshold, which is used to exclude low-credibility data. The credibility of each data is calculated based on the credibility of the edge device and the system threshold.
[0055] The deep learning model analyzes the span range of multiple attribute data of microgrid edge device data, and extracts the stability of each attribute data based on the span range of each attribute data. Data with a large span has a small impact on microgrid edge devices, while data with a small span has a large impact on microgrid edge devices. The weights of multiple attribute data of microgrid edge device data are calculated based on stability.
[0056] The deep learning model records the weights extracted in the above steps, and fuses the credibility of data obtained from multiple data sources based on the weights to form a comprehensive confidence assessment.
[0057] The deep learning model has no human intervention during the training process and can decide the judgment logic on its own. It has a good data summary effect on the microgrid edge device and can reasonably conduct effective confidence assessment on the microgrid edge device data, effectively reducing the impact of human intervention and other problems on data confidence.
[0058] The cost function of the deep learning model is:
[0059]
[0060] Where m is the number of training samples, is the value predicted by the deep learning model for parameter i, y i is the actual value of parameter i. By controlling the value range of E, when the value of E is closer to 0, the judgment result of the deep learning model is more accurate. When the value of E deviates from 0, the judgment result of the deep learning model is more ambiguous. The prediction effect of the deep learning model can be controlled by monitoring and adjusting E. This cost function can effectively control the judgment accuracy of the deep learning model.
[0061] The existing microgrid edge device data is judged through the trained deep learning model judgment logic to determine the data confidence.
[0062] The microgrid adjusts the operation mode based on the obtained trusted data. The microgrid data processing center sends the microgrid operation mode data to the microgrid edge device through the data transmission link.
[0063] The microgrid edge device verifies its own data based on the built-in distributed computing nodes to determine whether it complies with the microgrid operation mode sent by the microgrid data processing center.
[0064] The distributed computing nodes mentioned above are built-in components of each microgrid edge device. They can effectively utilize the computing power of microgrid edge devices, reduce the waste of computing resources, and achieve coordinated control of the entire microgrid.
[0065] On the other hand, Figure 2 As shown, the present invention provides a distributed microgrid data processing system, comprising:
[0066] The microgrid edge device has a built-in distributed computing node to match and verify its own data with the microgrid operation mode sent by the microgrid data processing center.
[0067] The data acquisition module acquires relevant operating data of the microgrid edge device in response to the data acquisition instruction issued by the data acquisition interface.
[0068] The data transmission module is responsible for data transmission between the microgrid edge device, the first data processing module and the second data processing module.
[0069] The first data processing module pre-processes the microgrid edge device data and cleans the microgrid edge device data.
[0070] The second data processing module verifies the confidence of the microgrid edge device data after preprocessing through a deep learning model.
[0071] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A distributed microgrid data processing method, characterized in that: Included are: Responding to a data acquisition instruction issued by the data acquisition interface, acquiring relevant operating data of the microgrid edge device; The microgrid data processing center pre-processes the microgrid edge device data and cleans the microgrid edge device data; The microgrid data processing center uses a deep learning model to verify the confidence of the microgrid edge device data after preprocessing; The microgrid data processing center sends the microgrid operation mode data to the microgrid edge device through the data transmission link; The microgrid edge device verifies its own data based on the built-in distributed computing nodes to determine whether it complies with the microgrid operation mode sent by the microgrid data processing center.
2. A distributed microgrid data processing method according to claim 1, characterized in that ,The method for the microgrid data processing center to pre-process the microgrid edge device data is: S1. Traverse the original microgrid edge device data, find blank values and missing values, and record the location of blank data; S2. Based on other data of a single microgrid edge device, predict blank values and missing values and complete the data; S3. Classify the completed microgrid edge device data, and classify each attribute value of the microgrid edge device data according to the attribute name of the microgrid edge device data.
3. A distributed microgrid data processing method according to claim 1, characterized in that ,The method by which the microgrid data processing center verifies the confidence of the ,microgrid edge device data after preprocessing through the deep learning model is: Submit historical microgrid edge device data to the deep learning model, and the deep learning model obtains the deep learning model judgment logic through training; The existing microgrid edge device data is judged by deep learning model judgment logic to determine the data confidence.
4. A distributed microgrid data processing method according to claim 3, characterized in that ,The method of deep learning model training is: The deep learning model reads data from multiple microgrid edge devices and extracts multiple credibility indicators based on the distribution intervals of multiple attribute data of the microgrid edge devices; The credibility index corresponds to the attributes of the edge device data acquired by the data acquisition module, and the attributes include time attribute data and space distance attribute data; According to the credibility index, the credibility is judged by combining the edge device data attributes with the corresponding data of the credibility index. The deep learning model is trained based on the existing data to obtain the initial threshold, which is used to exclude low-credibility data. The credibility of each data is calculated by combining the credibility of the edge device and the system threshold. The deep learning model analyzes the span range of multiple attribute data of the microgrid edge device data, extracts the stability of each attribute data based on the span range of each attribute data, and calculates the weight of multiple attribute data of the microgrid edge device data based on the stability; The deep learning model integrates the credibility of data obtained from multiple data sources according to weights to form a comprehensive confidence assessment.
5. A distributed microgrid data processing method according to claim 3, characterized in that ,The cost function of the deep learning model is: Where m is the number of training samples, is the value predicted by the deep learning model for parameter i, y i is the actual value of parameter i. By controlling the value range of E, the prediction effect of the deep learning model can be controlled.
6. A distributed microgrid data processing method according to claim 1, characterized in that ,The distributed computing node is a built-in component of each microgrid edge device.
7. A distributed microgrid data processing system, characterized in that: Included are: The microgrid edge device has a built-in distributed computing node to match and verify its own data with the microgrid operation mode sent by the microgrid data processing center; A data acquisition module, in response to a data acquisition instruction issued by a data acquisition interface, acquires relevant operating data of a microgrid edge device; A data transmission module, responsible for data transmission between the microgrid edge device, the first data processing module and the second data processing module; The first data processing module pre-processes the microgrid edge device data and cleans the microgrid edge device data; The second data processing module verifies the confidence of the microgrid edge device data after preprocessing through a deep learning model.