Wind power safety data sharing and protection method based on federated learning and blockchain

Abstract

The present invention discloses a method for sharing and protecting wind power safety data based on federated learning and blockchain, which relates to the field of new generation information technology for data processing, and includes the following steps: Step 1, establishing a confidential data model based on federated learning; Step 2, establishing proxy sharing integration based on blockchain; Step 3, establishing a data leakage risk assessment model based on a federated optimization algorithm. The technical solution provided by the present invention enhances security through federated learning and blockchain technology; utilizes differential privacy and encryption algorithms to ensure data privacy protection during data sharing; reduces data transmission requirements through distributed data processing, reduces system performance overhead, and improves overall efficiency; introduces a data leakage risk assessment model, which can monitor and warn potential security threats in real time, and enhances the system's warning capability.

Description

Technical Field

[0001] The present invention relates to the field of a new generation of information technology for data processing, and in particular to a method for sharing and protecting wind power safety data based on federated learning and blockchain. Background Art

[0002] With the increasing attention paid to renewable energy and the continuous advancement of technology, wind power, as an important component of clean energy, is developing and applying at an unprecedented rate around the world. The number of wind farms is increasing, the scale is expanding, and the amount of data generated is also growing explosively. These data cover wind resource assessment, wind turbine operating status, power generation efficiency, fault warning and maintenance records, etc., which are of great significance for optimizing wind farm operations, improving power generation efficiency and reducing operation and maintenance costs.

[0003] However, with the rapid development of the wind power industry, customers are increasingly concerned about the privacy of wind power data. Wind power data not only contains technical information, but also involves sensitive areas such as corporate trade secrets, user privacy, and national security. Once the data is leaked or abused, it will bring serious losses and risks to wind power companies, customers, and even the entire society.

[0004] Therefore, how to ensure the privacy and security of wind power data while ensuring efficient sharing of wind power data has become an important issue that the wind power industry needs to solve. Traditional data sharing methods often rely on centralized data platforms or cloud services. Although this method can achieve centralized management and processing of data, it has the risk of data leakage, tampering or illegal access. In addition, centralized data platforms may also cause the entire system to fail due to single point failures, affecting the normal operation of wind farms.

[0005] In the traditional wind power industry, methods to ensure the privacy and security of wind power data mainly rely on a series of technical means and management measures. These methods are designed to protect the confidentiality, integrity and availability of wind power data during collection, transmission, storage and processing. The following are some of the main traditional methods:

[0006] 1. Data encryption technology: Data encryption is the basic means of protecting the privacy and security of wind power data. By using encryption algorithms to convert wind power data into ciphertext that can only be decrypted by authorized users, it can effectively prevent data from being illegally obtained or tampered with during transmission and storage. Common encryption technologies include: symmetric encryption (such as AES) and asymmetric encryption (such as RSA), each with different application scenarios and advantages;

[0007] 2. Network security protection: Network security protection is an important means to protect wind power data from network attacks. Wind farms should deploy network security equipment such as firewalls, intrusion detection systems and security audit systems to monitor and defend against external network attacks. At the same time, the security management of the internal network of the wind farm should be strengthened to prevent internal personnel from leaking data or performing illegal operations;

[0008] 3. Access control mechanism: Access control is an important measure to limit access rights to wind power data. By establishing a strict access control policy, it can be ensured that only authorized personnel can access wind power data. This is usually achieved through identity authentication, role division, and authority allocation. For example, wind farm operation and maintenance personnel can only access data related to their work, while sensitive data such as customer privacy information is protected at a higher level.

[0009] The disadvantages of the above solutions are:

[0010] 1. Data encryption technology has a large performance overhead. The encryption algorithm involves a lot of calculations and conversions, which consumes computing resources and time. When processing large amounts of data or real-time communications, the performance cost of the encryption algorithm may affect the system's response speed and overall performance. The encryption process may cause data transmission and processing speeds to decrease, affecting the operational efficiency of wind farms. Moreover, key management is complex. Key management is an important part of data encryption, but it is also a tedious and sensitive task. The generation, storage, distribution, update, and destruction of keys require strict control and management. Once the key is leaked or lost, it may lead to illegal access to the data. Improper key management may also increase the complexity and cost of the system;

[0011] 2. The configuration of access control and permission management is highly complex. The access control and permission management system needs to be configured in detail according to the actual situation of the wind farm, including role definition, permission allocation, etc. The configuration process is complex and error-prone, and requires professional technicians to operate and maintain. Secondly, it lacks flexibility. Traditional access control and permission management systems often lack sufficient flexibility and are difficult to adapt to the rapid changes in wind farm business. When the organizational structure or business processes of a wind farm change, the permission system may need to be reconfigured, increasing management costs. Moreover, auditing and tracing are difficult. Although the access control and permission management system can record user operations, auditing and tracing become complicated and time-consuming when the amount of data is large and operations are frequent;

[0012] 3. The network security protection system needs to rely on specific hardware and software equipment to implement. These devices are expensive and need to be updated and maintained regularly. If the hardware equipment fails or the software has a vulnerability, the entire protection system may fail. Moreover, it is difficult to deal with new attacks. With the continuous development of network attack technology, new attack methods emerge in an endless stream. Traditional network security protection systems may find it difficult to effectively deal with these new attack methods, resulting in an increased risk of data leakage or system intrusion.

[0013] In summary, although data encryption, access control, and network security protection systems play an important role in wind power data privacy and security protection, they also have some shortcomings and limitations. Therefore, technicians in this field are committed to developing a new wind power security data sharing and protection method to solve the above defects in the prior art. Summary of the invention

[0014] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is how to solve the problem of data security sharing and protection in the wind power field.

[0015] To achieve the above objectives, in terms of data security, the present invention proactively establishes a series of protection measures to ensure data security; in terms of data protection, the present invention effectively defends against external intrusions to ensure data integrity and confidentiality. Specifically, the present invention provides a wind power safety data sharing and protection method based on federated learning and blockchain, comprising the following steps:

[0016] Step 1: Establish a confidential data model based on federated learning;

[0017] Step 2: Establishing blockchain-based proxy sharing and integration;

[0018] Step 3: Establish a data leakage risk assessment model based on the federated optimization algorithm.

[0019] Furthermore, the step 1 includes the following sub-steps:

[0020] Step 1.1: Perform data preprocessing;

[0021] Step 1.2: Train the local model to achieve distributed processing;

[0022] Step 1.3, encrypting the parameters of the local model and transmitting the encrypted parameters to the central server;

[0023] Step 1.4: Aggregate and update model parameters, generate new global model parameters, and distribute them;

[0024] Step 1.5: The local model receives the new global model parameters, performs model update and evaluation locally, and establishes the confidential data model.

[0025] Furthermore, the step 1.3 includes the following sub-steps:

[0026] Step 1.3.1, adding noise to the parameters of the local model to achieve differential privacy;

[0027] Step 1.3.2: Select an encryption algorithm to encrypt the parameters of the local model.

[0028] Furthermore, the step 2 includes the following sub-steps:

[0029] Step 2.1, generate a natural password;

[0030] Step 2.2, encrypting the data using the generated natural password;

[0031] Step 2.3: Use blockchain technology for data transmission;

[0032] Step 2.4: Decrypt data to achieve secure data sharing.

[0033] Furthermore, the step 2.1 includes the following sub-steps:

[0034] Step 2.1.1, extract dynamic features, use time-varying wind speed-direction-temperature data, take dynamic meteorological features as the information base, and ensure that the acquired feature data has uncertainty in the time dimension;

[0035] Step 2.1.2: Extract static features, obtain the device’s geographic location, device ID, MAC address, and IP address in real time through sensors and network interfaces to ensure the uniqueness of feature data;

[0036] Step 2.1.3: Combine the dynamic features with the static features and generate a unique natural password through a hash function.

[0037] Furthermore, the step 2.2 includes the following sub-steps:

[0038] Step 2.2.1, select encryption algorithm;

[0039] Step 2.2.2, Generate encryption keys through natural passwords, and dynamically generate new keys during each encryption process;

[0040] Step 2.2.3: Use the encryption key to encrypt the data to generate encrypted data.

[0041] Furthermore, the step 2.3 includes the following sub-steps:

[0042] Step 2.3.1. Design the blockchain structure. Each block contains the hash value, timestamp, transaction record and hash value of the previous block. Connect each block through a chain structure to ensure that the data cannot be tampered with.

[0043] Step 2.3.2, storing transaction records, storing the encrypted data as transaction records in the block, each transaction record contains the hash value, timestamp and signature information of the data;

[0044] Step 2.3.3. Select a consensus algorithm and verify and transmit transaction records through the consensus mechanism.

[0045] Furthermore, the step 2.4 includes the following sub-steps:

[0046] Step 2.4.1, extract dynamic and static features, using the same dynamic meteorological features and static environmental features extraction technology as the data sender to ensure the consistency and high accuracy of the extracted feature data;

[0047] Step 2.4.2: Use the same hash function to process the dynamic and static characteristics of the receiver to generate the same natural password as the sender;

[0048] Step 2.4.3: Use a symmetric decryption algorithm to decrypt the encrypted data using the encryption key generated by the natural password to restore the original data.

[0049] Furthermore, the step 3 includes the following sub-steps:

[0050] Step 3.1, designing the data leakage risk assessment model;

[0051] Step 3.2: Use a distributed collaborative learning framework to train the model and establish the data leakage risk assessment model based on the federated optimization algorithm;

[0052] The data leakage risk assessment model in step 3.1 selects information on personnel, hardware, and software as nodes, takes the relationship between the three as edges, calculates the weight of each node and the weight of the edge, and constructs a risk assessment algorithm based on complex network modeling, including the following sub-steps:

[0053] Step 3.1.1, calculate node weight;

[0054] Define a collection of people , Hardware Collection , Software Collection , setting personnel Have experienced data breaches Two hardware problems, then define their hardware feature set as , and its similarity with the hardware information in historical data is ; Setting personnel Have experienced data breaches Two software problems, then define their software feature set as , and its similarity with the software information in historical data is ; Through the above definition, we can get personnel The node weight , is the risk coefficient of the individual after normalization;

[0055] Step 3.1.2, calculate the edge weight; the similarity between personal characteristics of different people is defined as ;personnel With hardware The similarity between them is defined as ,personnel With software The similarity between them is defined as ;

[0056] Step 3.1.3, calculate the similarity between each node; use the Pearson correlation coefficient method to calculate the similarity index between different nodes, and then determine the degree of association between each node according to the size of the similarity index, and use the degree of association as the weight of the edge;

[0057] The overall similarity of the system is defined as ,Right now ;

[0058] Step 3.1.4: Assess data leakage risk.

[0059] Furthermore, the step 3.1.4 is specifically as follows: The systematic risk is expressed as , then the overall system risk is defined as , expressed as .

[0060] The wind power safety data sharing and protection method based on federated learning and blockchain provided by the present invention has at least the following technical effects:

[0061] Compared with the existing technology, this invention has the following significant advantages in terms of wind power data security sharing and protection:

[0062] 1. The technical solution provided by the present invention enhances the security of data during transmission and storage through federated learning and blockchain technology, and effectively prevents data leakage and illegal access;

[0063] 2. The technical solution provided by the present invention utilizes differential privacy and encryption algorithms to ensure data privacy protection during data sharing and avoid leakage of sensitive information;

[0064] 3. The technical solution provided by the present invention reduces the data transmission requirements through distributed data processing, reduces the performance overhead of the system, and improves the overall efficiency;

[0065] 4. The technical solution provided by the present invention introduces a data leakage risk assessment model, which can monitor and warn potential security threats in real time, thereby enhancing the early warning capability of the system.

[0066] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flow chart of a wind power confidentiality data model based on federated learning in a preferred embodiment of the present invention;

[0068] Figure 2 It is a flowchart of proxy sharing integration based on blockchain in a preferred embodiment of the present invention;

[0069] Figure 3 It is a schematic diagram of a wind power data leakage risk network according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0070] The following describes several preferred embodiments of the present invention with reference to the drawings in the specification, so that the technical content is clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0071] Example 1

[0072] The embodiment of the present invention provides a wind power safety data sharing and protection method based on federated learning and blockchain, which belongs to the field of artificial intelligence applications and involves federated learning technology and blockchain technology in artificial intelligence to solve the problem of data security sharing and protection in the wind power field.

[0073] The method comprises the following steps:

[0074] Step 1: Establish a confidential data model based on federated learning;

[0075] Step 2: Establishing blockchain-based proxy sharing and integration;

[0076] Step 3: Establish a data leakage risk assessment model based on the federated optimization algorithm.

[0077] In particular, in step 1, the rapid development of the wind power industry is accompanied by the generation and accumulation of a large amount of data, including wind speed, power generation, equipment status, etc. Effective use of this data for analysis and prediction can improve the operational efficiency and power generation of wind farms. However, traditional data processing methods have the risk of data leakage and privacy infringement, especially when sharing data among multiple wind farms. In order to solve these problems, a wind power confidentiality data model based on federated learning is proposed, which realizes the secure sharing and collaborative learning of wind power data through distributed data processing and privacy protection technology.

[0078] like Figure 1 As shown in the figure, it is a schematic diagram of the confidential data model based on federated learning, which includes the following sub-steps:

[0079] Step 1.1: Data preprocessing: Each wind farm preprocesses local data, including data cleaning, feature extraction, etc.

[0080] Step 1.2: Train the local model, that is, each wind farm performs data processing and model training locally to achieve distributed processing; each wind farm selects a suitable machine learning algorithm (such as neural network, support vector machine, etc.), trains the model on local data, and generates local model parameters;

[0081] Step 1.3: Encrypt the parameters of the local model and transmit the encrypted parameters to the central server; that is, before uploading the model parameters, each wind farm encrypts the parameters to ensure the security of the parameter transmission process;

[0082] Step 1.4: The central server aggregates and updates model parameters, generates new global model parameters, and distributes them;

[0083] Step 1.5: The local model receives the new global model parameters, updates and evaluates the model locally, and establishes a confidential data model.

[0084] Example 2

[0085] Based on Example 1, step 1.3 includes the following sub-steps:

[0086] Step 1.3.1: Add noise to the parameters of the local model to achieve differential privacy and ensure that it is difficult for attackers to infer the original data through the parameters.

[0087] Step 1.3.2: Select an encryption algorithm to encrypt the parameters of the local model to ensure confidentiality during parameter transmission. The encryption algorithm includes homomorphic encryption or Paillier encryption.

[0088] In particular, step 1.4 includes the following sub-steps:

[0089] Step 1.4.1, parameter verification, that is, the central server verifies the received encryption model parameters to ensure the validity and integrity of the parameters;

[0090] Step 1.4.2, parameter aggregation, that is, aggregating and updating the encrypted model parameters through secure multi-party computing technology to generate new global model parameters;

[0091] Step 1.4.3: Parameter distribution, that is, distributing the updated global model parameters to each wind farm to complete the iterative update of the model.

[0092] In particular, step 1.5 includes the following sub-steps:

[0093] Step 1.5.1, local model update, that is, each wind farm uses the updated global model parameters to update the local model to generate a new local model;

[0094] Step 1.5.2: Model evaluation, that is, each wind farm evaluates the updated local model, including indicators such as model accuracy and robustness, to ensure the effectiveness and reliability of the model.

[0095] Example 3

[0096] With the rapid development of the wind power industry, the issue of secure sharing and protection of data has become increasingly prominent. Traditional data storage and sharing methods face the risks brought by centralized management and are vulnerable to external attacks and data leakage. Therefore, a secure sharing model based on natural cryptography and blockchain underlying design is proposed. By integrating advanced cryptography and blockchain technology, distributed processing, secure transmission and shared protection of data can be achieved.

[0097] like Figure 2 As shown, based on Example 1 or 2, step 2 includes the following sub-steps:

[0098] Step 2.1, generate a natural password; in the process of data encryption and decryption, first extract the dynamic and static environmental features to generate a unique natural password;

[0099] Step 2.2: Use the generated natural password to encrypt the data to ensure the security of the data during transmission;

[0100] Step 2.3: Use blockchain technology for data transmission to achieve secure data transmission and storage;

[0101] Step 2.4: Decrypt the data to achieve secure data sharing. At the data receiver, extract the receiver's biometric and environmental features to generate a natural password that is the same as the sender's, and use the password to decrypt the data.

[0102] Example 4

[0103] Based on Example 1, 2 or 3, step 2.1 includes the following sub-steps:

[0104] Step 2.1.1, extract dynamic features, use time-varying wind speed-direction-temperature data, take dynamic meteorological features as the information base, and ensure that the acquired feature data has uncertainty in the time dimension;

[0105] Step 2.1.2: Extract static features, and obtain the device’s geographic location, device ID, MAC address, IP address and other information in real time through the device’s sensors and network interfaces to ensure the uniqueness of the feature data;

[0106] Step 2.1.3: Combine dynamic features with static features to generate a unique natural password through a hash function or other cryptographic algorithm.

[0107] In particular, step 2.2 includes the following sub-steps:

[0108] Step 2.2.1. Select encryption algorithm. Use AES (Advanced Encryption Standard) algorithm to encrypt data. AES has high security and processing speed, and is suitable for encryption of large-scale data.

[0109] Step 2.2.2, Generate encryption keys through natural passwords, and dynamically generate new keys during each encryption process to ensure the security and unpredictability of the keys;

[0110] Step 2.2.3: Use the encryption key to encrypt the data to generate encrypted data.

[0111] In particular, step 2.3 includes the following sub-steps:

[0112] Step 2.3.1. Design the blockchain structure. Each block contains the hash value, timestamp, transaction record and hash value of the previous block. Connect each block through a chain structure to ensure that the data cannot be tampered with.

[0113] Step 2.3.2: Store transaction records. Store encrypted data as transaction records in the block. Each transaction record contains the hash value, timestamp and signature information of the data to ensure the integrity and traceability of the data.

[0114] Step 2.3.3. Select a consensus algorithm and verify and transmit transaction records through the consensus mechanism to ensure the integrity and authenticity of the data. The consensus algorithm includes PoW, PoS or other consensus algorithms.

[0115] In particular, step 2.4 includes the following sub-steps:

[0116] Step 2.4.1, extract dynamic and static features, using the same dynamic meteorological features and static environmental features extraction technology as the data sender to ensure the consistency and high accuracy of the extracted feature data;

[0117] Step 2.4.2: Use the same hash function to process the dynamic and static characteristics of the receiver to generate the same natural password as the sender;

[0118] Step 2.4.3: Use the symmetric decryption algorithm to decrypt the encrypted data using the encryption key generated by the natural password to restore the original data.

[0119] Among them, the symmetric decryption algorithm is the same AES algorithm as the data encryption algorithm.

[0120] Example 5

[0121] Due to the large amount of data in the wind power industry, its wide distribution and high sensitivity, a data leakage risk assessment model based on a federated optimization algorithm is proposed to reduce the risk of data leakage. The model is based on data and starts from the three attributes of data: user, device and logic of use. It connects the construction of people, hardware and software from a network perspective, assumes that external intruders are the greatest leakage risk, and trains the data leakage risk assessment model to ensure the minimum leakage risk under the optimal logical structure.

[0122] Based on Embodiment 1, 2, 3 or 4, step 3 comprises the following sub-steps:

[0123] Step 3.1: Design a data leakage risk assessment model;

[0124] Step 3.2: Use a distributed collaborative learning framework to train the model and establish a data leakage risk assessment model based on a federated optimization algorithm.

[0125] Among them, in order to solve the dynamics and uncertainty of data leakage in the wind power industry and dynamically capture abnormal situations in the network, the data leakage risk assessment model in step 3.1 selects hardware, software, and personnel information as nodes in the leakage process, and takes the relationship between the three as edges. The weight of each node and the weight of the edge are calculated to construct a risk assessment algorithm based on complex network modeling, such as Figure 3 As shown in the figure, by measuring the correlation between multivariate data, calculating the weights of each node and edge, and building a risk assessment algorithm based on complex network modeling, the assessment and early warning of data leakage risks are completed. Compared with the traditional spatiotemporal assessment model, the complex network assessment model can build connections between data leakage from a network perspective and provide accurate early warnings for new data leakage.

[0126] by Figure 3Take the network as an example, the personnel information is represented as , which includes internal operators, external collaborators and external intruders; hardware information is represented by ; Software information is represented as .

[0127] Specifically, step 3.1 includes the following sub-steps:

[0128] Step 3.1.1, calculate node weight;

[0129] Define a collection of people , Hardware Collection , Software Collection , setting personnel Have experienced data breaches Two hardware problems, then define their hardware feature set as , and its similarity with the hardware information in historical data is ; Setting personnel Have experienced data breaches Two software problems, then define their software feature set as , and its similarity with the software information in historical data is ; Through the above definition, we can get personnel The node weight , is the risk coefficient of the individual after normalization;

[0130] Step 3.1.2, calculate the edge weight; the similarity between personal characteristics of different people is defined as ;personnel With hardware The similarity between them is defined as ,personnel With software The similarity between them is defined as ;

[0131] Step 3.1.3, calculate the similarity between each node; use the Pearson correlation coefficient method to calculate the similarity index between different nodes, and then determine the degree of association between each node according to the size of the similarity index, and use the degree of association as the weight of the edge;

[0132] The overall similarity of the system is defined as ,Right now ;

[0133] Step 3.1.4: Assess data breach risk; individual personnel The systematic risk is expressed as , then the overall system risk is defined as , expressed as .

[0134] In step 3.2, in order to protect the ownership and privacy of wind power data and reduce the risk of data leakage, a distributed collaborative learning framework can be used to train the risk assessment model. This technology ensures user privacy and completes collaborative training through parameter interaction rather than original data sharing to generate a global model, thereby effectively protecting data privacy. In this distributed collaborative learning framework, training data can be aggregated without exchanging original privacy information, thereby realizing the training of the global model.

[0135] The preferred specific embodiments of the present invention are described in detail above. It should be understood that ordinary technicians in the field can make many modifications and changes based on the concept of the present invention without creative work. Therefore, all technical solutions that can be obtained by technicians in the technical field based on the concept of the present invention through logical analysis, reasoning or limited experiments on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A wind power safety data sharing and protection method based on federated learning and blockchain, characterized in that: The method comprises the following steps: Step 1: Establish a confidential data model based on federated learning; Step 2: Establishing blockchain-based proxy sharing and integration; Step 3: Establish a data leakage risk assessment model based on the federated optimization algorithm; The step 2 includes the following sub-steps: Step 2.1, generate a natural password; Step 2.2, encrypting the data using the generated natural password; Step 2.3: Use blockchain technology for data transmission; Step 2.4: Decrypt data to achieve secure data sharing; The step 3 includes the following sub-steps: Step 3.1, designing the data leakage risk assessment model; Step 3.2: Use a distributed collaborative learning framework to train the model and establish the data leakage risk assessment model based on the federated optimization algorithm; The data leakage risk assessment model in step 3.1 selects information on personnel, hardware, and software as nodes, takes the relationship between the three as edges, calculates the weight of each node and the weight of the edge, and constructs a risk assessment algorithm based on complex network modeling, including the following sub-steps: Step 3.1.1, calculate node weight; Define a collection of people , Hardware Collection , Software Collection , setting personnel Have experienced data breaches Two hardware problems, then define their hardware feature set as , and its similarity with the hardware information in historical data is ; Setting personnel Have experienced data breaches Two software problems, then define their software feature set as , and its similarity with the software information in historical data is ; Through the above definition, we can get personnel The node weight , is the risk coefficient of the individual after normalization; Step 3.1.2, calculate the weight of the edge; the similarity between personal characteristics of different people is defined as ;personnel With hardware The similarity between them is defined as ,personnel With software The similarity between them is defined as ; Step 3.1.3, calculate the similarity between each node; use the Pearson correlation coefficient method to calculate the similarity index between different nodes, and then determine the degree of association between each node according to the size of the similarity index, and use the degree of association as the weight of the edge; The overall similarity of the system is defined as ,Right now ;in, is the similarity set between personnel and software information in historical data, It is a similarity set between personnel and hardware information in historical data; Step 3.1.4: Assess data leakage risk.

2. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 1 is characterized in that: The step 1 includes the following sub-steps: Step 1.1: Perform data preprocessing; Step 1.2: Train the local model to achieve distributed processing; Step 1.3, encrypting the parameters of the local model and transmitting the encrypted parameters to the central server; Step 1.4: Aggregate and update model parameters, generate new global model parameters, and distribute them; Step 1.5: The local model receives the new global model parameters, performs model update and evaluation locally, and establishes the confidential data model.

3. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 2 is characterized in that: The step 1.3 includes the following sub-steps: Step 1.3.1, adding noise to the parameters of the local model to achieve differential privacy; Step 1.3.2: Select an encryption algorithm to encrypt the parameters of the local model.

4. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 1, characterized in that: The step 2.1 includes the following sub-steps: Step 2.1.1, extract dynamic features, use time-varying wind speed-wind direction-temperature data, take dynamic meteorological features as the information base, and ensure that the acquired feature data has uncertainty in the time dimension; Step 2.1.2, extract static features, obtain the device's geographic location, device ID, MAC address, and IP address in real time through sensors and network interfaces to ensure the uniqueness of feature data; Step 2.1.3: Combine the dynamic features with the static features and generate a unique natural password through a hash function.

5. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 4 is characterized in that: The step 2.2 includes the following sub-steps: Step 2.2.

1. Select encryption algorithm; Step 2.2.2, Generate encryption keys through natural passwords, and dynamically generate new keys during each encryption process; Step 2.2.3: Use the encryption key to encrypt the data to generate encrypted data.

6. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 5 is characterized in that: The step 2.3 includes the following sub-steps: Step 2.3.

1. Design the blockchain structure. Each block contains the hash value, timestamp, transaction record and hash value of the previous block. Connect each block through a chain structure to ensure that the data cannot be tampered with. Step 2.3.2, storing transaction records, storing the encrypted data as transaction records in the block, each transaction record contains the hash value, timestamp and signature information of the data; Step 2.3.

3. Select a consensus algorithm and verify and transmit transaction records through the consensus mechanism.

7. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 6 is characterized in that: The step 2.4 includes the following sub-steps: Step 2.4.1, extract dynamic and static features, using the same dynamic meteorological features and static environmental features extraction technology as the data sender to ensure the consistency and high accuracy of the extracted feature data; Step 2.4.2: Use the same hash function to process the dynamic and static characteristics of the receiver to generate the same natural password as the sender; Step 2.4.3: Use a symmetric decryption algorithm to decrypt the encrypted data using the encryption key generated by the natural password to restore the original data.

8. The wind power safety data sharing and protection method based on federated learning and blockchain as claimed in claim 1, characterized in that: The step 3.1.4 is specifically as follows: The systematic risk is expressed as , then the overall system risk is defined as , expressed as ,in, For personnel The node weights, is the overall similarity of the system.

Images