A communication method and module for a low-voltage transformer area power line communication device side

Abstract

The application discloses a communication method and module of a low-voltage area power line communication device side, which comprises the following steps: receiving power data issued by a power line communication terminal of a low-voltage area, performing data noise reduction processing on the power data to obtain noise reduction data; constructing a training data set according to a transverse federal learning strategy and the noise reduction data, performing federal training on a local data model based on the training data set to obtain a global data model; wherein the global data model can be adapted to lightweight communication between the device side and the power line communication terminal; through an AI compression algorithm, repeated items and calculable items of the noise reduction data are investigated and deleted to obtain key data; and the global data model and the key data are uploaded to the power line communication terminal. The embodiment realizes lightweight power line communication, reduces data transmission complexity, and guarantees the safety and reliability of data transmission.

Description

Technical Field

[0001] This invention relates to the field of power line communication in low-voltage distribution areas, and more particularly to a communication method and module for power line communication equipment in low-voltage distribution areas. Background Technology

[0002] With the construction of new power systems, a large proportion of new energy equipment is being connected to low-voltage distribution areas, making the communication environment in these areas complex and subject to significant noise interference. This necessitates intelligent monitoring of low-voltage distribution areas. Power line carrier communication (PLC) is an important communication technology for low-voltage distribution networks. It utilizes transmission lines as the information transmission medium, offering advantages such as wide coverage and low communication costs, and exhibits good compatibility with intelligent equipment detection in low-voltage distribution areas. The power industry has lightweight requirements for transmitted communication data, and these requirements vary significantly across different power scenarios. Simultaneously, power services demand high security and reliability for data transmission. As a key component for terminal access, PLC terminal integration typically integrates communication modules, power supply modules, control modules, and various interfaces related to data transmission. However, the PLC terminal itself is limited by processor computing resources and cannot integrate compression and training strategies. Therefore, the PLC equipment-side module will play a crucial role. However, current modules are limited in form, possessing only communication capabilities and lacking lightweight model training and data compression capabilities, hindering the technology's replicability and widespread adoption.

[0003] Existing power line communication modules mainly consist of a power supply end and a regulation module. They address the technical problem that traditional power lines cannot provide differentiated power supplies to meet varying load demands, leading to inconvenience. However, these modules are complex in design and consume significant power. Existing technologies also integrate high-speed power line carrier communication modules with edge modules via serial communication. The edge module receives and processes data collected by the communication module, saving module space and addressing the single-function limitation of traditional high-speed power line carrier communication modules. However, this is only suitable for data processing in noise-free environments, limiting its applicability and making it unsuitable for widespread use. Furthermore, existing technologies only transmit raw data to assist power line communication terminals in distribution areas, neglecting noise interference from complex communication environments. They cannot eliminate interference to improve communication quality, compromising security and compatibility with power services. Finally, existing technologies cannot increase the information carrying capacity of power line communication, making it difficult to support reliable data transmission in low-voltage distribution areas, thus falling short of the requirements of power services.

[0004] Currently, the existing power line communication (PLC) equipment-side modules, PLC modules, and methods suffer from the following problems: First, although PLC modules and terminal equipment exist to meet the needs of power scenarios, they mostly rely on transmitting raw data, making it difficult to guarantee security and reliability. Second, existing equipment lacks lightweight capabilities, is costly and complex, and cannot meet the needs of power business scenarios. The communication environment in low-voltage distribution areas with a high proportion of new energy and massive equipment access is complex. Current PLC modules are of a single form, do not consider the lightweight requirements of power businesses, and lack lightweight model training and data compression capabilities, resulting in high data transmission complexity. Existing PLC modules use raw data transmission, failing to consider the coordinated transmission of lightweight communication models and critical data, making it impossible to ensure data transmission security and reliability while performing lightweight communication, thus falling short of power business requirements. To enable power terminals to quickly acquire lightweight and secure data transmission capabilities, lightweight equipment-side modules for PLC terminals are needed to meet the requirements of lightweight model training, data compression, and noise reduction in PLC communication. Summary of the Invention

[0005] This invention provides a communication method and module for power line communication equipment in low-voltage distribution areas, enabling lightweight power line communication, reducing data transmission complexity, and ensuring the security and reliability of data transmission.

[0006] To address the aforementioned technical problems, embodiments of the present invention provide a communication method for the power line communication equipment side in a low-voltage distribution area, comprising:

[0007] Receive power data sent by the power line communication terminal of the low-voltage distribution area, perform data noise reduction processing on the power data, and obtain noise-reduced data;

[0008] Based on the horizontal federated learning strategy and denoised data, a training dataset is constructed, and a local data model is trained in a federated manner on the training dataset to obtain a global data model; the global data model can be adapted to lightweight communication between the device side and the power line communication terminal.

[0009] By using AI compression algorithms, duplicate and inferable items in the noise-reduced data are identified and removed to obtain key data.

[0010] Upload the global data model and key data to the power line communication terminal.

[0011] In implementing this embodiment of the invention, power data transmitted from a power line communication terminal in a low-voltage distribution area is received, and the power data is subjected to data denoising processing to obtain denoised data. A training dataset is constructed based on a horizontal federated learning strategy and the denoised data. A local data model is then federatedly trained based on the training dataset to obtain a global data model. The global data model is adaptable to lightweight communication between the device and the power line communication terminal. Through an AI compression algorithm, duplicate and inferable items in the denoised data are identified and removed to obtain key data. The global data model and key data are then uploaded to the power line communication terminal. After acquiring power data from the power line, the power line communication terminal transmits the power data to the power line communication equipment. It does not store the power data in the terminal itself. Instead, federated learning training and AI compression are implemented on the power line communication equipment side. Through horizontal federated training and normalized compression, a global data model and key data supporting lightweight communication are acquired and uploaded. Compared to traditional modules that upload all data, this effectively reduces data transmission complexity. Only the global data model and key data are retained in the power line communication terminal to ensure communication, reducing the terminal's buffer and achieving lightweight transmission. Furthermore, noise identification and elimination, data model training, and key data extraction effectively improve data quality. Compared to power line communication modules that only upload the data model, this provides a reference for key data, ensuring both lightweight communication and data transmission security and reliability.

[0012] As a preferred approach, a training dataset is constructed based on the horizontal federated learning strategy and the denoised data, specifically as follows:

[0013] Based on the horizontal federated learning strategy, data with similar features but different data sources are extracted from the denoised data to construct a training dataset.

[0014] By implementing embodiments of the present invention, a horizontal federated learning strategy is adopted to extract key data with similar features and construct a training dataset, thereby training a lightweight communication data model to support secure and reliable lightweight power line communication.

[0015] As a preferred approach, the local data model is federated and trained based on the training dataset to obtain the global data model, specifically as follows:

[0016] Several participating servers receive the model from the central server. Each participating server trains its local data model based on the training dataset. Each trained local data model is then uploaded to the central server via encrypted gradients.

[0017] The central server performs federated training on each trained local data model to obtain an updated global data model.

[0018] As a preferred approach, AI compression algorithms are used to identify and remove duplicate and inferable items from the noise-reduced data to obtain key data, specifically:

[0019] Determine whether the denoised data is continuous. If it is not continuous, complete the denoised data to obtain continuous data; if it is continuous, treat the denoised data as continuous data.

[0020] Through intelligent search, duplicate items in continuous data are retrieved and deleted to obtain deduplicated data;

[0021] Based on the characteristic points of the oscillation waveform, non-uniform sampling is performed on the deduplicated data to extract key data that can reflect the status of the transformer area, while other inferable data is deleted to obtain compressed data.

[0022] Data compression is achieved by supplementing the compressed data with cubic Hermitian interpolation, thereby obtaining key data.

[0023] By implementing the embodiments of the present invention, duplicate and calculable items in the data are checked and deleted, key data is extracted through normalization compression, and only key data and models are uploaded back to the power line communication terminal. This ensures the security and reliability of data transmission while meeting the lightweight communication needs of low-voltage distribution areas.

[0024] As a preferred approach, the power data is subjected to noise reduction processing to obtain noise-reduced data, specifically as follows:

[0025] Noise identification is performed on power data to obtain identification results; the identification results include the types of noise and the distribution of each type of noise.

[0026] Based on the recognition results, different filters are selected to eliminate noise in the data and obtain denoised data.

[0027] As a preferred approach, noise identification is performed on the power data to obtain the identification results, specifically:

[0028] Based on the information in the noise database, noise matching analysis is performed on the power data to obtain the identification results;

[0029] The noise database is designed based on historical data and prior knowledge of power line communication noise in low-voltage distribution areas, and includes characteristic information of various types of noise in low-voltage distribution areas.

[0030] To address the same technical problem, this invention also provides a communication module for the power line communication equipment side of a low-voltage distribution area, including: a communication module, a noise identification and cancellation module, a federated training module, and an AI compression module.

[0031] The communication module is used to receive power data sent by the power line communication terminal of the low-voltage distribution area and upload the global data model and key data to the power line communication terminal.

[0032] The noise identification and cancellation module is used to perform noise reduction processing on power data to obtain noise-reduced data;

[0033] The federated training module is used to construct a training dataset based on the horizontal federated learning strategy and denoised data, and to perform federated training of the local data model based on the training dataset to obtain the global data model; the global data model can be adapted to lightweight communication between the device side and the power line communication terminal.

[0034] The AI ​​compression module is used to use AI compression algorithms to identify and remove duplicate and inferable items from the noise-reduced data, thereby obtaining key data.

[0035] As a preferred option, a power module is included;

[0036] The power module is used to supply power to the various modules of the communication module.

[0037] As a preferred embodiment, the noise identification and cancellation module includes a noise identification submodule and a noise cancellation submodule;

[0038] The noise identification submodule is used to identify noise in power data and obtain identification results; the identification results include the types of noise and the distribution of each type of noise.

[0039] The noise cancellation submodule is used to select different filters based on the recognition results to eliminate noise in the data and obtain denoised data.

[0040] To address the same technical problem, embodiments of the present invention also provide a computer device, including a processor and a memory, wherein the memory is used to store a computer program, and when the computer program is executed by the processor, it implements a communication method on the power line communication equipment side of a low-voltage distribution area. Attached Figure Description

[0041] Figure 1 : A schematic flowchart of an embodiment of a communication method on the power line communication equipment side of a low-voltage distribution area provided by the present invention;

[0042] Figure 2 : A schematic diagram of the structure of an embodiment of a communication module on the power line communication equipment side of a low-voltage distribution area provided by the present invention;

[0043] Figure 3 This is a schematic diagram of the connection of a communication module according to an embodiment of a communication module on the power line communication equipment side of a low-voltage distribution area, provided by the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Example 1

[0046] Please refer to Figure 1 This is a flowchart illustrating a communication method for a low-voltage distribution area power line communication device side according to an embodiment of the present invention. The communication method of this embodiment is applicable to communication on the low-voltage distribution area power line communication device side. This embodiment reduces data transmission complexity and ensures data transmission security and reliability through a horizontal federated learning strategy and an AI compression algorithm. The communication method includes steps 101 to 104, each step as follows:

[0047] Step 101: Receive power data sent by the power line communication terminal of the low-voltage distribution area, perform data noise reduction processing on the power data, and obtain noise-reduced data.

[0048] In this embodiment, the low-voltage distribution area power line communication equipment module connects to the power line communication terminal via a USB interface, interacts with the power line communication terminal, and receives power data sent by the low-voltage distribution area power line communication terminal.

[0049] Optionally, the power data can be denoised to obtain denoised data. Specifically, the power data can be noise identified to obtain identification results. The identification results include the types of noise and the distribution of each type of noise. Based on the identification results, different filters can be selected to eliminate the noise in the data to obtain denoised data.

[0050] Optionally, noise identification is performed on the power data to obtain identification results. Specifically, noise matching analysis is performed on the power data based on the information in the noise database to obtain identification results. The noise database is designed based on historical data and prior knowledge of power line communication noise in low-voltage distribution areas and includes characteristic information of various types of noise in low-voltage distribution areas.

[0051] In this embodiment, the noise database is designed based on historical data and prior knowledge of power line communication noise in low-voltage distribution areas. It stores characteristic information of various types of noise in the distribution area. Based on the information in the noise database, noise matching analysis can be performed on the access data to analyze the types of noise present in the data stream and the distribution of each type of noise. To achieve noise cancellation, filters are provided. Based on the noise identification results, different filters are selected to eliminate noise in the data and obtain noise-reduced data.

[0052] Step 102: Based on the horizontal federated learning strategy and the denoised data, construct a training dataset, perform federated training of the local data model based on the training dataset, and obtain a global data model; wherein, the global data model can be adapted to lightweight communication between the device side and the power line communication terminal.

[0053] Optionally, a training dataset is constructed based on the horizontal federated learning strategy and the denoised data. Specifically, the training dataset is constructed by extracting data with similar features but different data sources from the denoised data according to the horizontal federated learning strategy.

[0054] Optionally, a global data model can be obtained by federated training of local data models based on the training dataset. Specifically, several participating servers receive the model from the central server, each participating server trains its local data model based on the training dataset, and each trained local data model is uploaded to the central server via encrypted gradients. The central server then performs federated training on each trained local data model to obtain the updated global data model.

[0055] In this embodiment, federated training includes multiple participating servers and a master server. A horizontal federated learning strategy is employed, extracting data with similar features but different sources as key effective data for training the model. A data model supporting lightweight communication is trained based on this key effective data. The training process is as follows: the master server distributes the model to the participating servers; the participating servers train their local data models based on the extracted key data; the trained local data models are uploaded to the master server via encrypted gradients; and the master server performs federated training on the uploaded local data models to obtain an updated global data model.

[0056] Step 103: Using AI compression algorithms, identify and remove duplicate and inferable items from the noise-reduced data to obtain key data.

[0057] Optionally, step 103 specifically involves: using an AI compression algorithm to identify and remove duplicate and inferable items from the denoised data to obtain key data. Specifically, this includes: determining whether the denoised data is continuous; if not, completing the denoised data to obtain continuous data; if continuous, treating the denoised data as continuous data; using intelligent search to retrieve duplicate items from the continuous data and deleting them to obtain deduplicated data; performing non-uniform sampling on the deduplicated data based on oscillation waveform feature points to extract key data reflecting the status of the transformer area, deleting other inferable data, and obtaining compressed data; and supplementing the compressed data with cubic Hermitian interpolation to achieve data compression and obtain key data.

[0058] In this embodiment, AI compression is based on AI algorithms. After the data stream is compressed, it is first determined whether the data is continuous, and non-continuous data is supplemented. For continuous data, duplicate items are retrieved and deleted through intelligent search. Then, non-uniform sampling is performed on the data based on the feature points of the oscillation waveform to extract a small amount of key data that can reflect the status of the transformer area. Other inferable data is deleted, and important data is supplemented through cubic Hermit interpolation to achieve data compression and form key upload data (key data).

[0059] Step 104: Upload the global data model and key data to the power line communication terminal.

[0060] In this embodiment, key data and the global data model are uploaded to the power line communication terminal. The power line communication terminal only needs to retain the key data and the global data model to communicate, reducing caching and achieving lightweight transmission.

[0061] In implementing this embodiment of the invention, power data transmitted from a power line communication terminal in a low-voltage distribution area is received, and the power data is subjected to data denoising processing to obtain denoised data. A training dataset is constructed based on a horizontal federated learning strategy and the denoised data. A local data model is then federatedly trained based on the training dataset to obtain a global data model. The global data model is adaptable to lightweight communication between the device and the power line communication terminal. Through an AI compression algorithm, duplicate and inferable items in the denoised data are identified and removed to obtain key data. The global data model and key data are then uploaded to the power line communication terminal. After acquiring power data from the power line, the power line communication terminal transmits the power data to the power line communication equipment. It does not store the power data in the terminal itself. Instead, federated learning training and AI compression are implemented on the power line communication equipment side. Through horizontal federated training and normalized compression, a global data model and key data supporting lightweight communication are acquired and uploaded. Compared to traditional modules that upload all data, this effectively reduces data transmission complexity. Only the global data model and key data are retained in the power line communication terminal to ensure communication, reducing the terminal's buffer and achieving lightweight transmission. Furthermore, noise identification and elimination, data model training, and key data extraction effectively improve data quality. Compared to power line communication modules that only upload the data model, this provides a reference for key data, ensuring both lightweight communication and data transmission security and reliability.

[0062] Example 2

[0063] Accordingly, see Figure 2 , Figure 2 This is a schematic diagram of a second embodiment of a communication module for a low-voltage power line communication device provided by the present invention. A connection diagram of the communication module is shown below. Figure 3 As shown. Figure 2 The communication module on the low-voltage power line communication equipment side includes a communication module 201, a noise identification and cancellation module 202, a federated training module 203, an AI compression module 204, and a power module 205.

[0064] The communication module 201 is used to receive power data sent by the power line communication terminal of the low-voltage distribution area and upload the global data model and key data to the power line communication terminal.

[0065] The noise identification and cancellation module 202 is used to perform noise reduction processing on power data to obtain noise-reduced data.

[0066] The noise identification and cancellation module 202 includes a noise identification submodule 2021 and a noise cancellation submodule 2022.

[0067] The noise identification submodule 2021 is used to identify noise in power data and obtain identification results; the identification results include noise types and the distribution of each type of noise.

[0068] The noise cancellation submodule 2022 is used to select different filters based on the recognition results to eliminate noise in the data and obtain denoised data.

[0069] The federated training module 203 is used to construct a training dataset based on the horizontal federated learning strategy and the denoised data, and to perform federated training of the local data model based on the training dataset to obtain the global data model; wherein, the global data model can be adapted to lightweight communication between the device side and the power line communication terminal.

[0070] The AI ​​compression module 204 is used to use AI compression algorithms to identify and remove duplicate and inferable items from the noise-reduced data to obtain key data.

[0071] The power supply module 205 is used to supply power to the various modules of the communication module.

[0072] In addition, this application also provides a computer device, which includes a processor and a memory. The memory is used to store a computer program, and when the computer program is executed by the processor, it implements the steps in any of the above method embodiments.

[0073] Implementing this invention, traditional power line communication requires uploading all data. This method, however, only requires uploading key data and the model, eliminating the need for uploading all data to achieve information transmission. Furthermore, the data is compressed, enabling lightweight power line communication. The lightweight module on the device side for low-voltage distribution area power line communication terminals integrates a federated training module and an AI compression module. Through horizontal federated learning and normalized compression, it achieves data model construction and key data extraction. Only key data and the model need to be uploaded to the power line communication terminal, meeting the needs of lightweight power line communication in low-voltage distribution areas. Noise reduction, data model extraction, and data compression of the original data improve the data quality of power line communication, ensuring the security and reliability of lightweight communication. Compared to traditional federated learning methods that only upload the model, it exhibits better robustness and verifiability.

[0074] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A communication method on the power line communication equipment side of a low-voltage distribution area, characterized in that, include: Receive power data sent by the power line communication terminal of the low-voltage distribution area, perform data noise reduction processing on the power data, and obtain noise-reduced data; Based on the horizontal federated learning strategy and the denoised data, a training dataset is constructed, and a local data model is federatedly trained on the training dataset to obtain a global data model; wherein, the global data model can be adapted to lightweight communication between the device side and the power line communication terminal. By using AI compression algorithms, duplicate and inferable items in the noise-reduced data are identified and removed to obtain key data. The global data model and the key data are uploaded to the power line communication terminal.

2. The communication method on the low-voltage distribution area power line communication equipment side as described in claim 1, characterized in that, The construction of the training dataset based on the horizontal federated learning strategy and the denoised data is specifically as follows: Based on the horizontal federated learning strategy, data with similar features but different data sources are extracted from the denoised data to construct a training dataset.

3. The communication method on the low-voltage distribution area power line communication equipment side as described in claim 1, characterized in that, The process of performing federated training of the local data model based on the training dataset to obtain the global data model specifically involves: Several participating servers receive the model from the central server, each participating server trains the local data model based on the training dataset, and each trained local data model is uploaded to the central server via encrypted gradients. The central server performs federated training on each of the trained local data models to obtain the updated global data model.

4. The communication method on the low-voltage distribution area power line communication equipment side as described in claim 1, characterized in that, The process involves using an AI compression algorithm to identify and remove duplicate and inferable items from the noise-reduced data to obtain key data, specifically: Determine whether the noise reduction data is continuous. If it is not continuous, complete the noise reduction data to obtain continuous data; if it is continuous, use the noise reduction data as the continuous data. By using intelligent search, duplicate items in the continuous data are retrieved and deleted to obtain deduplicated data; Based on the characteristic points of the oscillation waveform, non-uniform sampling is performed on the deduplicated data to extract key data that can reflect the status of the transformer area, delete other inferable data, and obtain compressed data. The compressed data is supplemented by cubic Hermitian interpolation to achieve data compression and obtain the key data.

5. The communication method on the low-voltage distribution area power line communication equipment side as described in claim 1, characterized in that, The process of performing data noise reduction on the power data to obtain noise-reduced data specifically involves: The power data is subjected to noise identification to obtain identification results; wherein, the identification results include noise types and the distribution of each type of noise; Based on the recognition results, different filters are selected to eliminate noise in the data, thereby obtaining the noise-reduced data.

6. The communication method on the low-voltage distribution area power line communication equipment side as described in claim 5, characterized in that, The noise identification process for the power data, to obtain the identification result, specifically involves: Based on the information in the noise database, noise matching analysis is performed on the power data to obtain the identification results; The noise database is designed based on historical data and prior knowledge of power line communication noise in low-voltage distribution areas, and includes characteristic information of various types of noise in low-voltage distribution areas.

7. A communication module on the power line communication equipment side of a low-voltage distribution area, characterized in that, include: Communication module, noise recognition and cancellation module, federated training module, and AI compression module; The communication module is used to receive power data sent by the power line communication terminal of the low-voltage distribution area and upload the global data model and key data to the power line communication terminal. The noise identification and cancellation module is used to perform data noise reduction processing on the power data to obtain noise-reduced data; The federated training module is used to construct a training dataset based on the horizontal federated learning strategy and noise reduction data, and to perform federated training of the local data model based on the training dataset to obtain the global data model; wherein, the global data model can be adapted to lightweight communication between the device side and the power line communication terminal. The AI ​​compression module is used to use AI compression algorithms to identify and remove duplicate and calculable items from the noise-reduced data to obtain the key data.

8. The communication module on the low-voltage distribution area power line communication equipment side as described in claim 7, characterized in that, Including the power module; The power module is used to supply power to each module of the communication module.

9. The communication module on the low-voltage distribution area power line communication equipment side as described in claim 7, characterized in that, The noise identification and cancellation module includes a noise identification submodule and a noise cancellation submodule; The noise identification submodule is used to identify noise in the power data and obtain identification results; wherein, the identification results include noise types and the distribution of each type of noise; The noise cancellation submodule is used to select different filters based on the recognition results to eliminate noise in the data and obtain the noise-reduced data.

10. A computer device, characterized in that, It includes a processor and a memory, the memory being used to store a computer program, which, when executed by the processor, implements the communication method on the low-voltage distribution area power line communication equipment side as described in any one of claims 1 to 6.

Images