Distributed new energy communication access method and system

Abstract

The invention discloses a distributed new energy communication access method and system, and relates to the technical field of communication methods. The communication access method of the distributed new energy comprises four core steps of equipment registration, identity authentication, protocol adaptive conversion and data security transmission, and a complete flow of accessing the distributed new energy equipment to an energy management platform is formed. According to the communication access method of the distributed new energy, by introducing protocol adaptive conversion based on the AI engine, various standard protocols and unknown private protocols can be dynamically analyzed, and automatic conversion from messages to standardized data is realized. Therefore, the dependence of the traditional scheme on the preset protocol is thoroughly solved, the access barrier caused by protocol fragmentation is remarkably reduced, the compatibility and access efficiency of the system on massive heterogeneous equipment are improved, and the deployment and maintenance cost is greatly reduced.

Description

Technical Field

[0001] This invention relates to the field of communication method technology, specifically to a communication access method and system for distributed new energy sources. Background Technology

[0002] With the rapid development of distributed new energy sources, the number of access devices has surged and they are highly heterogeneous. Existing communication access technologies have the following shortcomings:

[0003] First, there is serious protocol fragmentation: different manufacturers' devices use multiple standard protocols such as Modbus, IEC61850, and DNP3, as well as a large number of proprietary protocols. Traditional gateways need to pre-configure fixed resolution rules, resulting in poor compatibility and high deployment and maintenance costs.

[0004] Second, data security risks: device identity is easily impersonated, and there is a risk of data transmission being eavesdropped on or tampered with.

[0005] Third, communication efficiency is low: uploading massive amounts of raw data directly to the cloud platform consumes a large amount of bandwidth resources and cannot meet the low-latency requirements of real-time control of local devices (such as MPPT adjustment and charging / discharging strategies). These problems severely restrict the development of large-scale and intelligent management of distributed renewable energy. To address the shortcomings of existing technologies, this invention provides a communication access method and system for distributed renewable energy to solve the above problems. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a communication access method and system for distributed new energy sources. It solves protocol compatibility issues through an AI-driven protocol adaptive conversion engine, utilizes an edge computing optimization layer to achieve local decision-making and bandwidth saving, and integrates protocol fingerprint authentication, reinforcement learning dynamic updates, and active detection mechanisms to construct multi-layered security protection. Ultimately, it achieves efficient, secure, and intelligent access and local real-time optimized control of distributed new energy devices.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a communication access method for distributed new energy sources, comprising the following steps:

[0008] S1. Device Registration: Connect the new energy equipment to the gateway and register the device identifier and metadata;

[0009] S2. Identity Authentication: Two-way security authentication is completed based on device identification;

[0010] S3. Adaptive Protocol Conversion: Utilizes an AI engine to dynamically parse and standardize device communication messages, including:

[0011] S3.1 Passive listening device raw message stream;

[0012] S3.2 Extract message structure features and match them with the protocol knowledge graph;

[0013] S3.3 Infer the semantics of data points and generate initial mapping relationships;

[0014] S3.4 Correct the mapping relationship and update the knowledge graph based on feedback data;

[0015] S4. Secure Data Transmission: Upload standardized data to the energy management platform after encryption.

[0016] Preferably, the method for constructing the protocol knowledge graph includes:

[0017] Stores a multi-protocol syntax rule base: containing frame structure templates for Modbus, IEC61850, and DNP3;

[0018] Establish a semantic network: associate data point addresses, physical quantity types, units, and device models;

[0019] Dynamic update mechanism: New private protocol feature vectors are added through reinforcement learning, and the update formula is as follows:

[0020] V new =αV old +(1-α)(βF str +γF sem )

[0021] Where V is the protocol feature vector, Fstr is the structural feature matrix, Fsem is the semantic similarity, and α, β, and γ are weight coefficients.

[0022] Preferably, the message structure feature extraction in S3.2 employs a dual-channel neural network model:

[0023] Channel 1: CNN convolutional layers extract byte-level local features;

[0024] Channel 2: Extracting message temporal dependency features using the LSTM layer;

[0025] The output fused feature vector is input into the protocol classifier.

[0026] Preferably, the semantic inference in S3.3 includes:

[0027] Semantic similarity between computing device messages and knowledge graph protocol templates:

[0028] Sim=max(σ(W·[E p ;Ed]+b))Ep is the protocol template embedding vector, Ed is the device data embedding vector, and σ is the activation function;

[0029] If Sim < θ, then the private protocol analogy reasoning module is started, and the protocol template of the same manufacturer is associated with the device model.

[0030] Preferably, the feedback mechanism in S3.4 employs a reinforcement learning model:

[0031] Define the action as a mapping relationship adjustment strategy;

[0032] The reward function is designed as follows:

[0033] R = ω1·Acc + ω2·DC - ω3·AC

[0034] Where Acc is accuracy, DC is data consistency, AC is the number of adjustments, and ω1, ω2, and ω3 are weighting coefficients.

[0035] The policy network parameters are updated using Q-learning, prioritizing the mapping adjustment actions that result in the largest reward increase.

[0036] Preferably, an active detection mechanism is introduced during the feedback phase:

[0037] Send a preset set of instructions to the device to trigger a response at key data points;

[0038] By comparing the response message with the knowledge graph prediction, address mapping errors can be identified.

[0039] Generate adversarial examples and inject them into the training set.

[0040] Preferably, an edge computing optimization layer is added after S3:

[0041] At the gateway, local decisions are made based on the transformed data.

[0042] Dynamic adjustment of MPPT parameters for photovoltaic arrays;

[0043] Real-time calculation of battery charging and discharging strategies;

[0044] Only the decision results are uploaded to the cloud platform.

[0045] Preferably, the edge computing layer works in conjunction with the protocol conversion engine:

[0046] When abnormal fluctuations in device data are detected, the protocol sampling frequency is automatically increased;

[0047] If the data conversion exceeds the reasonable range N times consecutively, the device re-registration process will be triggered.

[0048] Preferably, the identity authentication phase binds the protocol fingerprint:

[0049] Extract the protocol characteristics of the device's first communication as a device fingerprint;

[0050] Subsequent authentication verifies the consistency of the protocol fingerprint to block protocol spoofing attacks.

[0051] The second aspect of this invention discloses a communication access system for distributed new energy sources, applied to the communication access method for said distributed new energy sources, comprising:

[0052] The device registration module is used to connect new energy devices to the gateway and register device identifiers and metadata;

[0053] The identity authentication module is used to complete two-way security authentication based on the device identifier;

[0054] The protocol adaptive conversion module is used to dynamically parse and standardize device communication messages using an AI engine, including:

[0055] Passive monitoring of the device's raw message stream;

[0056] Extract message structure features and match them with protocol knowledge graphs;

[0057] Infer the semantics of data points and generate initial mapping relationships;

[0058] The mapping relationship is corrected and the knowledge graph is updated based on feedback data;

[0059] The data security transmission module is used to encrypt standardized data before uploading it to the energy management platform.

[0060] This invention discloses a communication access method and system for distributed new energy sources, which has the following beneficial effects:

[0061] 1. This distributed new energy communication access method introduces AI engine-based adaptive protocol conversion, which can dynamically parse multiple standard protocols and unknown private protocols, and realize the automatic conversion of messages into standardized data. This completely solves the dependence of traditional solutions on preset protocols, significantly reduces the access barriers caused by protocol fragmentation, improves the system's compatibility and access efficiency for massive heterogeneous devices, and greatly reduces deployment and maintenance costs.

[0062] 2. This distributed renewable energy communication access method significantly reduces the amount of data that needs to be uploaded to the cloud platform by making local decisions based on standardized data at the gateway and only uploading the decision results, effectively alleviating network bandwidth pressure. Simultaneously, locally executed decisions significantly reduce control command latency, improving the real-time performance and autonomy of the distributed renewable energy system. The linkage between the edge computing layer and the protocol engine can also dynamically adjust the sampling frequency or trigger re-registration in case of data anomalies, ensuring data reliability and system stability.

[0063] 3. The communication access method of this distributed new energy source comprehensively utilizes two-way authentication of binding protocol fingerprints, encrypted data transmission, dynamic updating of knowledge graphs based on reinforcement learning, and feedback correction mechanism. In particular, the reinforcement learning model and active detection mechanism enable the system to continuously optimize protocol parsing accuracy, automatically adapt to new private protocols, and effectively identify and defend against address mapping errors and potential attacks. This constructs a multi-layered, adaptive security protection system and endows the system with the ability to continuously learn and evolve. Attached Figure Description

[0064] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0065] Figure 1 This is a flowchart of the overall method of the present invention;

[0066] Figure 2 This is a flowchart illustrating the adaptive conversion process of the protocol in this invention.

[0067] Figure 3 This is a flowchart illustrating the construction of the protocol knowledge graph for this invention.

[0068] Figure 4 This is a diagram of the dual-channel neural network model of the present invention;

[0069] Figure 5 This is a flowchart of the edge computing optimization layer of the present invention;

[0070] Figure 6 This is a flowchart of the fingerprint authentication protocol of the present invention;

[0071] Figure 7 This is a system architecture diagram of the present invention. Detailed Implementation

[0072] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only some embodiments of the present invention, 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.

[0073] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0074] This invention discloses a communication access method for distributed new energy sources, according to the appendix. Figure 1 To be continued Figure 7 As shown, it includes the following steps:

[0075] S1. Device Registration: Connect the new energy equipment to the gateway and register the device identifier and metadata. The device identifier is used to uniquely identify the device, while the metadata contains basic information about the device, such as device type, model, and production date. This provides basic information for subsequent device management and communication, facilitating unified management and monitoring of the equipment.

[0076] S2. Identity Authentication: Two-way security authentication is performed based on device identification; ensuring the legitimacy and security of accessing devices. Two-way authentication means that not only does the device need to verify the gateway's identity, but the gateway also needs to verify the device's identity. This prevents unauthorized devices from accessing the network, ensures communication security, and avoids data leaks and malicious attacks.

[0077] S3. Protocol Adaptive Conversion: Utilizing an AI engine, dynamic protocol parsing and standardized conversion of device communication messages are performed, based on the appendix... Figure 2 As shown, it includes:

[0078] S3.1 Passive listening device raw message stream;

[0079] S3.2 Extract message structure features and match them with the protocol knowledge graph;

[0080] S3.3 Infer the semantics of data points and generate initial mapping relationships;

[0081] S3.4 Correct the mapping relationship and update the knowledge graph based on feedback data;

[0082] The protocol adaptive conversion step can adapt to various communication protocols used by different devices, realize automatic protocol parsing and conversion, and improve system compatibility and flexibility.

[0083] S4. Secure Data Transmission: Standardized data is encrypted before being uploaded to the energy management platform. Encryption technology protects data security during transmission, preventing theft or tampering. It ensures the confidentiality and integrity of data during transmission, guaranteeing the energy management platform receives accurate and reliable data.

[0084] This method includes four core steps: device registration, identity authentication, protocol adaptive conversion, and secure data transmission, forming a complete process for distributed new energy devices to access the energy management platform.

[0085] Furthermore, according to the appendix Figure 3 As shown, the method for constructing the protocol knowledge graph includes:

[0086] The system stores a multi-protocol syntax rule base, including frame structure templates for Modbus, IEC61850, and DNP3, providing a fundamental reference for protocol parsing. Therefore, this method can quickly identify and process common communication protocols, improving the efficiency and accuracy of protocol parsing.

[0087] Establish a semantic network: Associate data point addresses, physical quantity types, units, and equipment models to form a semantic network. This allows for the rapid determination of the type and unit of a physical quantity based on information such as equipment model and data point address. This step enhances the understanding of data semantics and improves the accuracy and consistency of data conversion.

[0088] Dynamic update mechanism: New private protocol feature vectors are added through reinforcement learning, and the update formula is as follows:

[0089] V new =αV old +(1-α)(βF str +γF sem )

[0090] Where V is the protocol feature vector, Fstr is the structural feature matrix, Fsem is the semantic similarity, and α, β, and γ are weighting coefficients. The protocol feature vector is continuously adjusted based on feedback data to adapt to new proprietary protocols. This enables the protocol knowledge graph to continuously learn and update, adapting to the ever-changing communication protocol environment and improving the system's scalability and adaptability.

[0091] According to the appendix Figure 4 As shown, furthermore, the message structure feature extraction in S3.2 adopts a dual-channel neural network model:

[0092] Channel 1: CNN convolutional layers extract byte-level local features; by sliding the convolutional kernel across the byte sequence of the message, local features such as frame headers and checksums are extracted. These local features are crucial for identifying the type and structure of protocols. Channel 1 effectively captures these local features, improving the accuracy of protocol identification.

[0093] Channel Two: Extracting Temporal Dependency Features from LSTM Layers; LSTM (Long Short-Term Memory) networks can handle long-term dependencies in sequential data. In message structure feature extraction, it can extract temporal dependency features, such as the order and time intervals of various fields in the message. Considering the temporal information of the message allows for a more comprehensive understanding of its structure and meaning, improving the accuracy of protocol parsing.

[0094] The output fused feature vector is input into the protocol classifier. The feature vectors extracted from the CNN convolutional layers and LSTM layers are fused to form a comprehensive feature vector. This fused feature vector contains both local and temporal features of the message, enabling a more accurate description of the message structure. By integrating features from different levels and types, the accuracy and robustness of protocol classification are improved.

[0095] The semantic inference in S3.3 includes:

[0096] Semantic similarity between computing device messages and knowledge graph protocol templates:

[0097] Sim=max(σ(W·[E p ;Ed]+b))Ep is the protocol template embedding vector, Ed is the device data embedding vector, and σ is the activation function;

[0098] If Sim < θ, then the private protocol analogy reasoning module is started, and the protocol template of the same manufacturer is associated with the device model.

[0099] Device messages and protocol templates from the knowledge graph are converted into embedding vectors Ed and Ep, respectively. A neural network is then used to calculate their semantic similarity Sim, with the activation function σ mapping the results to a suitable range. This process quantifies the semantic similarity between device messages and protocol templates, providing an objective basis for semantic inference.

[0100] When the semantic similarity Sim is less than the threshold θ, the device message is considered to potentially use a proprietary protocol. In this case, by associating the device model with protocol templates from the same manufacturer, and using the similarity of protocols from the same manufacturer for analogical reasoning, the semantics of the device message are inferred. This improves the method's ability to identify and process proprietary protocols, and expands the system's protocol adaptability.

[0101] The feedback mechanism in S3.4 employs a reinforcement learning model:

[0102] The action is defined as a mapping adjustment strategy; by defining the mapping adjustment strategy as an action in reinforcement learning, the system can select different adjustment actions to correct the mapping relationship based on different feedback data. This enables the system to flexibly adjust the mapping relationship according to the actual situation, improving the accuracy of data transformation.

[0103] The reward function is designed as follows:

[0104] R = ω1·Acc + ω2·DC - ω3·AC

[0105] Where Acc represents accuracy, DC represents data consistency, AC represents the number of adjustments, and ω1, ω2, and ω3 are weighting coefficients; the reward function comprehensively considers three factors: accuracy (Acc), data consistency (DC), and the number of adjustments (AC). Higher accuracy and data consistency result in a larger reward; more adjustments result in a smaller reward. The weighting coefficients ω1, ω2, and ω3 can be used to adjust the importance of each factor. This guides the system to adjust towards improving accuracy and data consistency while avoiding over-adjustment.

[0106] The policy network parameters are updated using Q-learning, prioritizing the mapping adjustment action that maximizes the reward increase. Q-learning is a reinforcement learning algorithm that learns the optimal policy by continuously trying different actions and updating the Q-value according to the reward function. The system selects the mapping adjustment action with the largest reward increase based on the Q-value, continuously optimizing the mapping relationship. Therefore, this method can automatically learn the optimal mapping adjustment policy, improving the system's adaptability and learning efficiency.

[0107] Introduce an active detection mechanism during the feedback phase:

[0108] The system sends preset command sets to the device to trigger responses at key data points. These commands can trigger responses at key data points on the device. By observing the device's responses, the actual data from the device can be obtained. This proactive acquisition of key device data provides a basis for identifying address mapping errors.

[0109] By comparing the response message with the knowledge graph prediction, address mapping errors can be identified. The device's response message is compared with the predicted message in the knowledge graph; if discrepancies exist, an address mapping error may exist. This allows for timely detection of address mapping errors, improving the accuracy of data conversion.

[0110] Adversarial examples are generated and injected into the training set based on identified address mapping errors. These adversarial examples are then injected into the training set to retrain the model. This allows the model to learn more anomalies, improving its robustness and enhancing its ability to withstand interference, enabling it to better cope with various complex situations.

[0111] According to the appendix Figure 5 As shown, furthermore, an edge computing optimization layer is added after S3:

[0112] At the gateway, local decisions are made based on the transformed data.

[0113] Dynamic adjustment of MPPT parameters for photovoltaic arrays;

[0114] Real-time calculation of battery charging and discharging strategies;

[0115] Only the decision results are uploaded to the cloud platform.

[0116] The converted data is processed and analyzed in real time at the gateway, making local decisions based on preset algorithms and rules. For example, for photovoltaic arrays, the maximum power point tracking (MPPT) parameters are dynamically adjusted based on parameters such as light intensity and temperature; for batteries, charging and discharging strategies are calculated in real time based on factors such as battery capacity and load demand. Therefore, this method can quickly respond to changes in equipment, improving the system's real-time performance and response speed.

[0117] Instead of uploading all the raw data, only the decision-making results are uploaded to the cloud platform. This significantly reduces data transmission volume and bandwidth consumption, saving network resources and lowering communication costs, making it particularly suitable for distributed renewable energy systems with limited bandwidth.

[0118] The edge computing layer works in conjunction with the protocol conversion engine:

[0119] When abnormal fluctuations in device data are detected, the protocol sampling frequency is automatically increased;

[0120] If the data conversion exceeds the reasonable range N times consecutively, the device re-registration process will be triggered.

[0121] The edge computing layer monitors device data changes in real time. When abnormal fluctuations are detected, it indicates that more detailed data may be needed to analyze the device status. At this point, it automatically sends a command to the protocol conversion engine to increase the protocol sampling frequency and acquire more device data. Therefore, this method can promptly detect device anomalies and obtain more detailed information by increasing the sampling frequency, facilitating accurate determination of the device status.

[0122] If the data conversion exceeds a reasonable range N times consecutively, it indicates a potential serious problem with the device or an anomaly in the communication protocol. In this case, the device re-registration process is triggered, re-registering the device, re-authenticating, and re-adapting the protocol conversion to ensure normal communication. This improves system reliability and stability, promptly handles device anomalies, and guarantees the normal operation of the system.

[0123] According to the appendix Figure 6 As shown, further, the identity authentication phase binds a protocol fingerprint:

[0124] Extract the protocol characteristics of the device's first communication as a device fingerprint;

[0125] Subsequent authentication verifies the consistency of the protocol fingerprint to block protocol spoofing attacks.

[0126] During the initial communication between devices, their protocol characteristics, such as message format, field order, and data encoding method, are extracted and used as the device's unique fingerprint. A unique identifier is then created for each device to facilitate subsequent authentication and security protection.

[0127] In subsequent identity authentication processes, the device's protocol characteristics are extracted again and compared with the initially extracted device fingerprint. If they match, the device is considered legitimate; if they do not match, a protocol spoofing attack may have occurred, blocking the device's communication. This effectively prevents protocol spoofing attacks and improves the security and reliability of identity authentication.

[0128] According to the appendix Figure 7 As shown, in particular, the second aspect of this invention discloses a communication access system for distributed new energy sources, applied to the communication access method for said distributed new energy sources, comprising:

[0129] The device registration module is used to connect new energy devices to the gateway and register device identifiers and metadata;

[0130] The identity authentication module is used to complete two-way security authentication based on the device identifier;

[0131] The protocol adaptive conversion module is used to dynamically parse and standardize device communication messages using an AI engine, including:

[0132] The raw message stream of the passive monitoring device;

[0133] Extract message structure features and match them with protocol knowledge graphs;

[0134] Infer the semantics of data points and generate initial mapping relationships;

[0135] The mapping relationship is corrected and the knowledge graph is updated based on feedback data;

[0136] The data security transmission module is used to encrypt standardized data before uploading it to the energy management platform.

[0137] The system comprises a device registration module, an identity authentication module, a protocol adaptive conversion module, and a data security transmission module. The entire communication access process is divided into different modules, each responsible for a specific function. The device registration module handles device access and registration, the identity authentication module handles device identity verification, the protocol adaptive conversion module handles protocol parsing and conversion, and the data security transmission module handles data encryption and uploading. This improves the system's maintainability and scalability, facilitating independent development and optimization of each module.

[0138] Example 1: Photovoltaic power station equipment connection;

[0139] A large photovoltaic power plant has multiple photovoltaic inverters and sensors manufactured by different companies. These devices use different communication protocols, such as Modbus and IEC61850. These devices need to be connected to an energy management platform for unified monitoring and management.

[0140] Detailed steps

[0141] Equipment registration: Connect the photovoltaic inverter and sensor to the gateway and register the equipment identification (such as equipment number) and metadata (such as equipment model, production date, rated power, etc.).

[0142] Identity authentication: Two-way security authentication based on device identifiers to ensure legitimate device access.

[0143] Protocol adaptive conversion:

[0144] The raw message stream from the passively monitored device is used to extract message structure features using a dual-channel neural network model. CNN convolutional layers extract local features such as frame headers and checksums, while LSTM layers extract temporal dependency features of the message.

[0145] The extracted feature vectors are matched with the protocol knowledge graph to calculate semantic similarity. For example, if the semantic similarity between a message from a photovoltaic inverter and a Modbus protocol template in the knowledge graph is 0.85, which is greater than the threshold of 0.8, then that template is directly used for semantic inference.

[0146] An initial mapping relationship is generated and corrected based on feedback data. For example, if a significant deviation is found between the actual and predicted values ​​of a data point during data transmission, the mapping relationship is adjusted using a reinforcement learning model to update the protocol knowledge graph.

[0147] Edge computing optimization:

[0148] At the gateway, local decisions are made based on the converted data, dynamically adjusting the MPPT parameters of the photovoltaic array according to parameters such as illuminance and temperature. Assume the current illuminance is 800 W / m². 2 The temperature is 25℃. The optimal MPPT parameters calculated by the preset algorithm are 380V voltage and 10A current.

[0149] Only the decision results (such as MPPT parameters, battery charging and discharging strategies, etc.) are uploaded to the energy management platform to reduce bandwidth consumption.

[0150] Secure data transmission: Standardized data is encrypted before being uploaded to the energy management platform to ensure data security during transmission.

[0151] Assuming a photovoltaic power station has 100 devices, each generating 10 data points per second, with each data point occupying 4 bytes, without edge computing optimization, directly uploading all raw data would require transmitting 100 × 10 × 4 = 4000 bytes per second. With edge computing optimization, only decision results are uploaded; assuming each decision result occupies 10 bytes, the data transmission per second is 100 × 10 = 1000 bytes, reducing bandwidth consumption by 75%.

[0152] Example 2: Wind power plant equipment connection;

[0153] A wind farm has multiple wind turbine generators and related monitoring equipment. These devices use communication protocols including DNP3 and proprietary protocols. It is necessary to connect these devices to an energy management platform to achieve real-time monitoring and optimized control of the wind farm.

[0154] Detailed steps

[0155] Device registration and identity authentication: Similar to Example 1, the wind turbine generator set and monitoring equipment are connected to the gateway and their information is registered to complete two-way security authentication.

[0156] Protocol adaptive conversion:

[0157] For devices using the DNP3 protocol, protocol parsing and conversion are performed according to standard procedures.

[0158] For devices using proprietary protocols, the semantic similarity between the device message and the knowledge graph protocol template is first calculated. Assuming that the semantic similarity between a monitoring device's message and all templates in the knowledge graph is less than the threshold of 0.8, the proprietary protocol analogy reasoning module is activated. Based on the device model and its association with other device protocol templates from the same manufacturer, analogy reasoning is performed to infer the message's semantics.

[0159] An active detection mechanism is employed to send a pre-defined set of instructions to the device, triggering responses from key data points. The response messages are compared with the predicted values ​​from the knowledge graph to identify address mapping errors. For example, if the actual address of a data point is found to be inconsistent with its predicted address, adversarial examples are generated and injected into the training set to retrain the model and improve its robustness.

[0160] Edge computing optimization:

[0161] At the gateway, the optimal pitch angle and rotational speed of the wind turbine are calculated in real time based on parameters such as wind speed and wind direction. Assuming the current wind speed is 12 m / s, the optimal pitch angle is calculated to be 5° and the optimal rotational speed is 15 r / min using a preset algorithm.

[0162] When abnormal wind speed fluctuations are detected, the protocol sampling frequency is automatically increased from once per second to five times per second to obtain more detailed wind speed data so as to more accurately adjust the operating parameters of the generator set.

[0163] Secure data transmission: Standardized data is encrypted and then uploaded to the energy management platform.

[0164] Assume a wind farm has 20 wind turbines, each with 5 monitoring devices. Each monitoring device generates 5 data points per second, and each data point occupies 4 bytes. Without edge computing optimization, the amount of data to be transmitted per second is: 20 × 5 × 5 × 4 = 2000 bytes. After edge computing optimization, only decision results are uploaded. Assuming each decision result occupies 15 bytes, the amount of data to be transmitted per second is: 20 × 15 = 300 bytes, reducing bandwidth consumption by 85%.

[0165] Example 3: Access to small-scale distributed new energy system equipment;

[0166] A small-scale distributed renewable energy system includes solar water heaters, small wind turbines, and batteries. While the communication protocols of these devices are relatively simple, they also present certain compatibility issues. These devices need to be connected to a home energy management platform to achieve intelligent management of home energy resources.

[0167] Detailed steps

[0168] Device registration and identity authentication: Connect solar water heaters, small wind turbines and batteries to the gateway, register device identification and metadata, and complete two-way security authentication.

[0169] Protocol adaptive conversion:

[0170] A dual-channel neural network model is used to extract the structural features of device messages and match them with a protocol knowledge graph. Since device protocols are relatively simple, the semantic similarity calculation results are generally high, enabling rapid protocol parsing and conversion.

[0171] During the authentication phase, a protocol fingerprint is bound, extracting the protocol characteristics from the device's first communication as its fingerprint. Subsequent authentication verifies the consistency of the protocol fingerprint to prevent protocol spoofing attacks.

[0172] Edge computing optimization:

[0173] At the gateway, the charging and discharging strategy for the battery is calculated in real time based on information such as the water temperature of the solar water heater, the output power of the small wind turbine, and the battery charge. For example, when the water temperature of the solar water heater reaches the set value and the output power of the small wind turbine is relatively high, the battery is charged first.

[0174] If the data conversion exceeds the reasonable range three times in a row, the device re-registration process will be triggered to ensure normal device communication.

[0175] Secure data transmission: Standardized data is encrypted and then uploaded to the home energy management platform.

[0176] Assume a small-scale distributed renewable energy system has 3 devices, each generating 3 data points per second, with each data point occupying 4 bytes. Without edge computing optimization, the amount of data to be transmitted per second is 3 × 3 × 4 = 36 bytes. After edge computing optimization, only decision results are uploaded. Assuming each decision result occupies 8 bytes, the amount of data to be transmitted per second is 3 × 8 = 24 bytes, reducing bandwidth consumption by 33.3%.

[0177] Example 4: Access to multi-energy complementary system equipment;

[0178] A multi-energy complementary system includes various energy devices such as photovoltaic, wind power, and hydropower. These devices come from different manufacturers, use complex and diverse communication protocols, and include some proprietary protocols. It is necessary to connect these devices to an integrated energy management platform to achieve coordinated and optimized scheduling of multiple energy sources.

[0179] Detailed steps

[0180] Device registration and identity authentication: Connect various energy devices to the gateway, register detailed information, and complete strict two-way security authentication.

[0181] Protocol adaptive conversion:

[0182] For common protocols (such as Modbus, IEC61850, DNP3), parsing and conversion are performed according to standard procedures.

[0183] For proprietary protocols, a combination of semantic inference and proactive probing mechanisms is employed. First, semantic similarity is calculated; if the similarity is low, an analogy reasoning module is activated. Then, a preset instruction set is sent via the proactive probing mechanism to obtain the device's actual response data, further refining the protocol parsing results.

[0184] By continuously optimizing the mapping relationship using a reinforcement learning model and adjusting the policy network parameters according to the reward function, the accuracy and stability of protocol conversion can be improved.

[0185] Edge computing optimization:

[0186] At the gateway, the energy allocation strategy is dynamically adjusted based on the real-time output and load demand of various energy sources. For example, when photovoltaic power generation is high, local load demand is prioritized, and excess electricity is stored in batteries or transmitted to the grid; when wind power generation is insufficient, hydroelectric generators are started to supplement it.

[0187] The system monitors equipment data in real time, automatically increases the protocol sampling frequency when abnormal fluctuations are detected, and takes timely measures to ensure stable system operation.

[0188] Secure data transmission: Standardized data is encrypted and then uploaded to the integrated energy management platform.

[0189] Assuming a multi-energy complementary system has 50 devices, each generating 8 data points per second, with each data point occupying 4 bytes, without edge computing optimization, the amount of data to be transmitted per second is: 50 × 8 × 4 = 1600 bytes. After edge computing optimization, only decision results are uploaded, assuming each decision result occupies 12 bytes, the amount of data to be transmitted per second is: 50 × 12 = 600 bytes, reducing bandwidth consumption by 62.5%.

[0190] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0191] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A communication access method for distributed new energy sources, characterized in that, Includes the following steps: S1. Device Registration: Connect the new energy equipment to the gateway and register the device identifier and metadata; S2. Identity Authentication: Two-way security authentication is completed based on device identification; S3. Adaptive Protocol Conversion: Utilizes an AI engine to dynamically parse and standardize device communication messages, including: S3.1 Passive listening device raw message stream; S3.2 Extract message structure features and match them with the protocol knowledge graph; S3.3 Infer the semantics of data points and generate initial mapping relationships; S3.4 Correct the mapping relationship and update the knowledge graph based on feedback data; S4. Secure Data Transmission: Upload standardized data to the energy management platform after encryption.

2. The communication access method for distributed new energy sources according to claim 1, characterized in that, The method for constructing the protocol knowledge graph includes: Stores a multi-protocol syntax rule base: containing frame structure templates for Modbus, IEC61850, and DNP3; Establish a semantic network: associate data point addresses, physical quantity types, units, and device models; Dynamic update mechanism: New private protocol feature vectors are added through reinforcement learning, and the update formula is as follows: V new =αV old +(1-α)(βF str +γF sem ) Where V is the protocol feature vector, and Fs t r is the structural feature matrix, Fsem is the semantic similarity, and α, β, and γ are weight coefficients.

3. The communication access method for distributed new energy sources according to claim 2, characterized in that, The message structure feature extraction in S3.2 adopts a dual-channel neural network model: Channel 1: CNN convolutional layers extract byte-level local features; Channel 2: Extracting message temporal dependency features using the LSTM layer; The output fused feature vector is input into the protocol classifier.

4. The communication access method for distributed new energy sources according to claim 1, characterized in that, The semantic inference in S3.3 includes: Semantic similarity between computing device messages and knowledge graph protocol templates: Sim=max(σ(W·[E p [Ed]+b)) E p E is the protocol template embedding vector. d Let σ be the device data embedding vector, and σ be the activation function. If Sim < θ, then the private protocol analogy reasoning module is started, and the protocol template of the same manufacturer is associated with the device model.

5. The communication access method for distributed new energy sources according to claim 4, characterized in that, The feedback mechanism in S3.4 employs a reinforcement learning model: Define the action as a mapping relationship adjustment strategy; The reward function is designed as follows: R = ω1·Acc + ω2·DC - ω3·AC Where Acc is accuracy, DC is data consistency, AC is the number of adjustments, and ω1, ω2, and ω3 are weighting coefficients. The policy network parameters are updated using Q-learning, prioritizing the mapping adjustment actions that result in the largest reward increase.

6. The communication access method for distributed new energy sources according to claim 5, characterized in that, Introduce an active detection mechanism during the feedback phase: Send a preset set of instructions to the device to trigger a response at key data points; By comparing the response message with the knowledge graph prediction, address mapping errors can be identified. Generate adversarial examples and inject them into the training set.

7. The communication access method for distributed new energy sources according to claim 1, characterized in that, An edge computing optimization layer is added after S3: At the gateway, local decisions are made based on the transformed data. Dynamic adjustment of MPPT parameters for photovoltaic arrays; Real-time calculation of battery charging and discharging strategies; Only the decision results are uploaded to the cloud platform.

8. The communication access method for distributed new energy sources according to claim 7, characterized in that, The edge computing layer works in conjunction with the protocol conversion engine: When abnormal fluctuations in device data are detected, the protocol sampling frequency is automatically increased; If the data conversion exceeds the reasonable range N times consecutively, the device re-registration process will be triggered.

9. The communication access method for distributed new energy sources according to claim 1, characterized in that, Binding protocol fingerprint during identity authentication phase: Extract the protocol characteristics of the device's first communication as a device fingerprint; Subsequent authentication verifies the consistency of the protocol fingerprint to block protocol spoofing attacks.

10. A communication access system for distributed new energy sources, applied to the communication access method for distributed new energy sources as described in any one of claims 1-9, characterized in that, include: The device registration module is used to connect new energy devices to the gateway and register device identifiers and metadata; The identity authentication module is used to complete two-way security authentication based on the device identifier; The protocol adaptive conversion module is used to dynamically parse and standardize device communication messages using an AI engine, including: The raw message stream of the passive monitoring device; Extract message structure features and match them with protocol knowledge graphs; Infer the semantics of data points and generate initial mapping relationships; The mapping relationship is corrected and the knowledge graph is updated based on feedback data; The data security transmission module is used to encrypt standardized data before uploading it to the energy management platform.

Images