Collection system and method for data fusion of electric vehicle and charging pile

Abstract

The invention relates to the technical field of charging piles, in particular to an electric vehicle and charging pile data fusion acquisition system and method, and the system comprises an electric vehicle charging data and charging pile data multi-level aggregation architecture, the electric vehicle and charging pile data multi-level aggregation framework comprises an electric vehicle charging data layer, a charging pile layer and an encryption aggregation layer. The charging pile layer comprises all charging pile nodes in a distribution network area, and each charging pile node is used for collecting and encrypting electric vehicle charging data and sharing the encrypted charging data to the encryption aggregation layer; the encryption aggregation layer comprises a plurality of encryption aggregation nodes, and each encryption aggregation node independently manages one charging pile node and is used for collecting charging data. According to the invention, the charging data of the electric vehicle is encrypted on the charging pile node, and the encryption aggregation layer is utilized to carry out multi-level data aggregation management, so that the charging data privacy of a user is effectively protected, and unauthorized access and data leakage are prevented.

Description

Technical Field

[0001] The present invention relates to the technical field of charging piles, and in particular to a data collection system and method for integrating data of electric vehicles and charging piles. Background Art

[0002] With the popularization of electric vehicles and the extensive construction of charging infrastructure, the number of electric vehicles and charging load have shown a rapid upward trend in recent years. Under the new situation, aggregating the charging load data of electric vehicles in various regions of the distribution network is an important basis for formulating electric vehicle demand response strategies and guiding the orderly charging of electric vehicles.

[0003] Among the related technologies, data aggregation methods will face feasibility issues when massive electric vehicles are connected. If charging data is collected directly by power grid companies, it will impose a heavy workload on them and there will be scalability issues. If the data is aggregated by a third-party service provider and then submitted to the power grid company, there will be risks of data tampering and privacy exposure during the data flow and transmission process.

[0004] Therefore, how to collect and aggregate electric vehicle charging data in a reliable, scalable, and privacy-protected manner is of great significance in future scenarios where distribution networks are connected to massive electric vehicle charging piles. Summary of the invention

[0005] In view of at least one of the above technical problems, the present invention provides a data collection system and method for fusion of electric vehicle and charging pile data to solve the technical problem of poor security of existing charging pile data collection.

[0006] According to a first aspect of the present invention, a data collection system for electric vehicle and charging pile fusion is provided, comprising:

[0007] A multi-level aggregation architecture for electric vehicle and charging pile data, the multi-level aggregation architecture for electric vehicle and charging pile data comprising an electric vehicle charging data layer, a charging pile layer and an encryption aggregation layer;

[0008] The electric vehicle charging data layer includes the charging data of the electric vehicle and the charging interface. The charging data includes: the battery and motor status of the user's electric vehicle, charging requirements, and charging plan;

[0009] The charging pile layer includes all charging pile nodes in the distribution network area, each of which is used to collect and encrypt the charging data of electric vehicles, and share the encrypted charging data to the encryption aggregation layer;

[0010] The encryption aggregation layer includes multiple encryption aggregation nodes, and the multiple encryption aggregation nodes independently manage one charging pile node to collect the charging data.

[0011] In some embodiments of the present invention, the system further includes a client, and the aggregation and encrypted transmission of electric vehicles are achieved between the client and each encryption aggregation node of the encryption aggregation layer through a federated learning architecture.

[0012] In some embodiments of the present invention, the federated learning architecture adopts an adaptive aggregation algorithm and a consensus protocol to improve data security.

[0013] In some embodiments of the present invention, the encryption aggregation layer is configured to be a primary encryption representative of the encryption aggregation node set k. Receive data collected from all charging piles {U i} k When the encryption master represents The encrypted slave representatives based on the PBFT consensus and the encrypted aggregation node set will reach an agreement on the aggregation results of the ciphertext charging plans for each area.

[0014] In some embodiments of the present invention, the encryption aggregation layer uses the following calculation formula when determining the charging data aggregation result of each area:

[0015]

[0016] In the formula, Represents the encrypted master representative in the encrypted aggregation node set k The calculated aggregation result of the ciphertext charging data of region k; Represents a set of charging piles in area k. In some embodiments of the present invention, the encryption aggregation node is configured to perform model aggregation after collecting local models participating in the deep network training deployed by the charging pile nodes, and adjust the model aggregation weight according to the quality evaluation result of the adaptive aggregation algorithm and the reputation score of the charging pile nodes in the collaboration process, and the collaboration process is a process in which the models of the deep networks deployed by the charging pile nodes are aggregated under federated learning and uniformly transmitted to the encryption aggregation node for updating.

[0017] In some embodiments of the present invention, the quality evaluation of the adaptive aggregation algorithm is based on the node Q k In the i-th round of training, the model loss is determined, and the node Q k Participate in the deployment of deep networks in the charging pile nodes and use cross entropy To evaluate the local model loss, the quality assessment weight G of the adaptive aggregation algorithm i (Q k ) is calculated as:

[0018]

[0019] Where:

[0020] is f(x i ), y i The cross entropy between i ) and y i They are the label and observation data of the ciphertext charging data of the i-th round of the encrypted aggregation node set, and N represents the total number of rounds.

[0021] In some embodiments of the present invention, each time the model aggregates the G of each node i (Q k ) are all recorded on the corresponding encryption aggregation node, node Q k The reputation in the aggregation process is represented by a cumulative reputation score S(Q k ) is used to represent the reputation score S(Q k ) is calculated as:

[0022]

[0023] Where τ is the node Q k The number of times the model has participated in aggregation.

[0024] In some embodiments of the present invention, the global model ζ in the adaptive aggregation algorithm Ω The calculation formula is:

[0025]

[0026] In the formula, ζ Ω (t) represents the global model parameters at time t, ζ Ω (t-1) represents the global model parameters at the moment before time t, M is the number of nodes participating in the model aggregation, represents the K-th round broadcast model parameter ζ k According to a second aspect of the present invention, there is also provided a method for collecting electric vehicle and charging pile data fusion, which is applied to the electric vehicle and charging pile data fusion collection system in the first aspect and comprises the following steps:

[0027] Create and publish initial encrypted data containing initialization information;

[0028] The client with a reputation score greater than the set value is obtained to participate in the training. The client uses the charging data and the feature comparison loss function to complete the data standardization model update, and sends the updated local model weight together with the digital signature in the form of encrypted data to the associated packaged data;

[0029] Receive the local model and verify its legitimacy, and put the transaction with legal signature into the transaction pool;

[0030] Score and sign all weights in the transaction pool, update the reputation score of each client, calculate the weight of the global model and generate new encrypted data;

[0031] Verify the legitimacy of encrypted data, calculate the weight of the global model and generate new encrypted data;

[0032] Download new blocks to obtain global weights, and start training from clients whose reputation scores are greater than the set value, and start the next round of global training until the total number of global training rounds is reached;

[0033] Download new encrypted data.

[0034] The beneficial effects of the present invention are as follows: the present invention effectively protects the user's charging data privacy and prevents unauthorized access and data leakage by encrypting the electric vehicle charging data at the charging pile node and using the encryption aggregation layer to perform multi-level data aggregation management. Since the system uses encryption technology for data storage and processing, the integrity and security of the charging data during transmission and storage are ensured, and data tampering and loss are prevented. Multiple encryption aggregation nodes independently manage encryption nodes, so that the system has a high degree of fault tolerance and scalability, and can be dynamically adjusted and expanded according to needs to adapt to application scenarios of different scales. Furthermore, through distributed architecture design and multi-level encryption aggregation strategies, the stability and reliability of data collection and processing are enhanced, ensuring normal operation in various complex network environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0036] Figure 1 It is a schematic diagram of the framework of the multi-level aggregation architecture of charging data in an embodiment of the present invention;

[0037] Figure 2 It is a schematic diagram of the framework of the data collection system for electric vehicle and charging pile fusion in an embodiment of the present invention;

[0038] Figure 3 It is a flow chart of a data collection system for electric vehicle and charging pile fusion in an embodiment of the present invention;

[0039] Figure 4 The figure is a schematic diagram of the steps of the method for fusion collection of electric vehicle and charging pile data in an embodiment of the present invention. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0041] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only and do not represent the only implementation method.

[0042] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0043] like Figures 1 to 3 The electric vehicle and charging pile data fusion collection system shown includes: a multi-level aggregation architecture for electric vehicle and charging pile data, which includes an electric vehicle charging data layer, a charging pile layer 1 and an encryption aggregation layer 2; the electric vehicle charging data layer includes the electric vehicle charging data and charging interface, and the charging data includes: the user's electric vehicle battery and motor status, charging requirements, and charging plan; the charging pile layer 1 includes all charging pile nodes 11 in the distribution network area, each charging pile node 11 is used to collect and encrypt the electric vehicle charging data, and share the encrypted charging data to the encryption aggregation layer 2; the encryption aggregation layer 2 includes multiple encryption aggregation nodes, and the multiple encryption aggregation nodes each independently manage a charging pile node 11 to collect charging data. Figure 1As shown, the multi-level aggregation architecture of charging data is the basis of the entire system, which is divided into two main layers: charging pile layer 1 and encryption aggregation layer 2, which are used to realize the distributed collection, encryption and aggregation processing of charging data; charging pile layer 1 is composed of all charging pile nodes 11 in the distribution network area, and each charging pile node 11 has the function of collecting electric vehicle charging data, and encrypts the collected data at the same time. The encrypted data is then shared to the encryption layer aggregation to ensure that the data is not leaked during transmission and protect user privacy; encryption aggregation layer 2 is composed of multiple encryption aggregation nodes, each encryption aggregation node independently manages an encryption node for storing and processing charging data. The encryption aggregation node is responsible for collecting encrypted charging data from the charging pile layer 1, and storing, processing and further aggregating the data through encryption technology to ensure data integrity and privacy protection.

[0044] Specifically, the system also includes a client, and the client and each encryption aggregation node of the encryption aggregation layer 2 realize the aggregation and encryption transmission of electric vehicles based on the federated learning architecture. Figure 2 As shown in the figure, the client receives data from electric vehicles and performs preliminary processing and encryption on the data through the federated learning architecture. During the data encryption process, the client uses privacy protection technology to ensure that user data will not be intercepted or tampered with by a third party during transmission. The client connects to each encryption aggregation node of the encryption aggregation layer 2 through encryption technology. The decentralized feature enables the client to perform distributed data processing while ensuring data security, further enhancing data privacy protection and the system's anti-attack capabilities. The client transmits the encrypted data to the relevant nodes of the encryption aggregation layer 2 through encryption, and obtains feedback information from the aggregation nodes in real time to ensure the integrity and timeliness of data transmission. As a bridge between the user and the system, the client is not only responsible for data collection and encryption, but also ensures the security and efficiency of data transmission through an advanced federated learning architecture.

[0045] Furthermore, the federated learning architecture uses adaptive aggregation algorithms and consensus protocols to improve data security. Figure 2 , the adaptive aggregation algorithm dynamically adjusts the data aggregation strategy according to the data characteristics and network status of the charging pile node 11, ensuring that data can be aggregated efficiently and accurately in different environments. This algorithm not only improves the response speed of the system, but also better allocates computing resources among different nodes, thereby enhancing the overall data processing capability; the consensus protocol ensures that all encrypted aggregation nodes have a unified opinion on the data processing process, preventing data from being tampered with or forged during transmission and aggregation. Based on the federated learning architecture, the adaptive aggregation algorithm and consensus protocol effectively enhance the security and reliability of the system, ensuring the privacy protection and efficient transmission of electric vehicle charging data.

[0046] In some embodiments of the present invention, the encryption aggregation layer 2 is configured to be a primary representative of the encryption aggregation node set k. Receive data collected from all charging piles {U i} k When the encryption master represents The encrypted slave representatives based on PBFT consensus and the encrypted aggregation node set will reach an agreement on the aggregation results of the encrypted charging plan for each area. The encrypted aggregation layer 2 includes an encrypted aggregation node set, which includes an encrypted master representative and several encrypted slave representatives. Figure 2 and Figure 3 As shown in the figure, the encryption master representative is responsible for coordinating and managing the data collected from each charging pile node 11; when the encryption master representative receives data from all charging pile nodes 11, it will start the PBFT consensus protocol. PBFT is a consensus mechanism widely used in encryption systems. It can ensure that most nodes in the system reach a consensus on the processing results even in the presence of malicious nodes. In this process, the encryption master representative will negotiate with the encryption slave representative on the charging plan aggregation results of each area ciphertext, and through the PBFT consensus protocol, ensure that all nodes reach a consensus on the final aggregation results. This not only improves the reliability of the aggregation results, but also ensures that the entire system can maintain stable operation in the face of potential malicious behavior; through the PBFT consensus protocol, the encryption aggregation layer 2 ensures that all encryption nodes reach a consensus on the aggregation results of the charging data. This process greatly improves the accuracy and reliability of data processing and prevents the risk of data tampering or forgery. Ultimately, this mechanism ensures that the system can provide reliable and accurate results when processing large amounts of charging data.

[0047] Preferably, the encryption aggregation layer 2 uses the following calculation formula when determining the charging data aggregation result of each area:

[0048]

[0049] In the formula, Represents the encrypted master representative in the encrypted aggregation node set k The calculated aggregation result of the ciphertext charging data of region k; represents the set of charging piles in area k. Figure 2 and Figure 3 Through this calculation method, the encryption aggregation layer 2 can obtain the encrypted data aggregation results of the entire area without decrypting a single charging data. This method not only ensures the privacy of the data, but also improves the efficiency and security of data processing.

[0050] In some embodiments of the present invention, the encryption aggregation node is configured to perform model aggregation after collecting local models participating in the deep network training deployed by the charging pile node, and adjust the model aggregation weight according to the quality evaluation results of the adaptive aggregation algorithm and the reputation score of the charging pile node 11 in the collaboration process. The collaboration process is the process of aggregating the models of the deep network deployed by each charging pile node 11 under federated learning and uniformly transmitting them to the encryption aggregation node for updating. Please continue to refer to Figure 2 and Figure 3 The adaptive aggregation algorithm will conduct a quality assessment on each uploaded local model. The evaluation criteria may include the accuracy, robustness and other relevant performance indicators of the model. Based on these evaluation results, the algorithm will determine the weight of each model in the final aggregation process. In addition to the quality assessment of the model itself, the encryption node will also score the reputation of the participating nodes based on their performance in the collaboration process. The reputation reflects the contribution, honesty and stability of the node throughout the training process. The model of a node with a high reputation may be given a greater weight, while the contribution of the node with a low reputation may be affected in the aggregation.

[0051] Furthermore, the quality evaluation of the adaptive aggregation algorithm is based on the node Q k In the i-th round of training, the model loss is determined, and the node Q k Participate in the deployment of deep networks in the charging pile nodes and use cross entropy To evaluate the local model loss, the quality assessment weight G of the adaptive aggregation algorithm i (Q k ) is calculated as:

[0052]

[0053] Where:

[0054] is f(x i ), y i The cross entropy between i ) and y i are the labels and observation data of the ciphertext charging data of the i-th round of the encrypted aggregation node set, and N represents the total number of rounds. Please continue to refer to Figure 2 and Figure 3 The adaptive aggregation algorithm determines the quality by evaluating the model loss generated by the node in the i-th round of local training. The model loss reflects the error or deviation of the node when processing data. The smaller the loss, the better the performance of the model. Cross entropy is a commonly used loss function, especially widely used in classification problems. This algorithm uses cross entropy to evaluate the loss of the local model. Specifically, cross entropy measures the predicted value f(x i ) and the true label y i, to quantify the loss of the model. Based on the above formula, the algorithm evaluates the model loss of each round and calculates the cross entropy value. Through these cross entropy values, the algorithm can determine the weight of each node in the model aggregation process. Nodes with smaller losses have greater model contributions, so they will be given higher weights in the final aggregation. By using cross entropy to evaluate model loss and adjust weights, the adaptive aggregation algorithm can more accurately reflect the contribution of each node. This not only improves the accuracy of the final aggregation model, but also ensures that each node is reasonably weighted according to its performance during the aggregation process, thereby optimizing the performance of the overall system.

[0055] Better, each time the model aggregates the Gi( Qk ) are all recorded on the encryption aggregation node, node Q k The reputation performance in the aggregation process can be expressed as a cumulative reputation score S( Qk ) is used to represent the reputation score S( Qk ) is calculated as:

[0056]

[0057] Where τ is the node Q k The number of times the model is aggregated. Figure 2 and Figure 3 As shown in the figure, when each node participates in model aggregation, its behavior and performance will be recorded on the encrypted node. These records not only ensure the transparency and immutability of the data, but also provide a reliable data basis for subsequent reputation evaluation. The reputation score is a comprehensive evaluation of the performance of the node in multiple model aggregation processes. It not only affects the weight distribution of the node in the subsequent model aggregation process, but also serves as a reference standard for the node's future participation in system decision-making and other cooperative tasks; nodes with high reputation scores are more likely to obtain higher weights in future model aggregations, while nodes with low reputation scores may be restricted or have reduced opportunities to participate.

[0058] Based on the above embodiment, the global model ζ in the adaptive aggregation algorithm Ω The calculation formula is:

[0059]

[0060] In the formula, ζ Ω (t) represents the global model parameters at time t, ζ Ω (t-1) represents the global model parameters at the moment before time t, M is the number of nodes participating in the model aggregation, represents the K-th round broadcast model parameter ζ k Please continue to refer to Figure 2 and Figure 3, in each round of model aggregation, the global model parameter ζ k It will be broadcasted to all participating nodes. Each node performs local training based on these parameters and its own data, and generates new model parameter estimates. In each round of broadcasting and updating, the system will continuously optimize these estimates and gradually converge to a more accurate global model.

[0061] In a second aspect of the present invention, a data collection method based on electric vehicle and charging pile data fusion is also provided, which is applied to the data collection system based on electric vehicle and charging pile data fusion in the first aspect, such as Figure 4 As shown, the following steps are included:

[0062] S10: Create and publish initial encrypted data containing initialization information;

[0063] S20: Obtain clients whose credit scores are greater than a set value to participate in training. The clients use charging data and feature comparison loss function to complete the data standardization model update, and send the updated local model weights together with the digital signature in the form of encrypted data to the associated packaged data;

[0064] S30: Receive the local model and verify its legitimacy, and put the transaction with the legal signature into the transaction pool;

[0065] S40: Score and sign all weights in the transaction pool, update the reputation score of each client, calculate the weight of the global model and generate new encrypted data;

[0066] S50: verify the legitimacy of the encrypted data, calculate the weight of the global model and generate new encrypted data;

[0067] S60: Download the new block to obtain the global weight, and start the next round of global training from the clients whose reputation scores are greater than the set value until the total number of global training rounds is reached;

[0068] S70: Download new encrypted data.

[0069] This data fusion collection method based on electric vehicles and charging piles ensures the security and privacy protection of data during the collection and processing process, and realizes efficient aggregation and analysis of data.

[0070] Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A data collection system for electric vehicles and charging piles, characterized in that: include: A multi-level aggregation architecture for electric vehicle and charging pile data, the multi-level aggregation architecture for electric vehicle and charging pile data comprising an electric vehicle charging data layer, a charging pile layer and an encryption aggregation layer; The electric vehicle charging data layer includes the charging data of the electric vehicle and the charging interface. The charging data includes: the battery and motor status of the user's electric vehicle, charging requirements, and charging plan; The charging pile layer includes all charging pile nodes in the distribution network area, each of which is used to collect and encrypt the charging data of electric vehicles, and share the encrypted charging data to the encryption aggregation layer; The encryption aggregation layer includes multiple encryption aggregation nodes, and the multiple encryption aggregation nodes independently manage one charging pile node to collect the charging data.

2. The electric vehicle and charging pile data fusion collection system according to claim 1 is characterized in that: The system also includes a client, and the aggregation and encrypted transmission of electric vehicles are achieved between the client and each encryption aggregation node of the encryption aggregation layer through a federated learning architecture.

3. The electric vehicle and charging pile data fusion collection system according to claim 2 is characterized in that: The federated learning architecture adopts an adaptive aggregation algorithm and a consensus protocol to improve data security.

4. The electric vehicle and charging pile data fusion collection system according to claim 1 is characterized in that: The encryption aggregation layer is configured to be an encryption master representative of the encryption aggregation node set k. Receive data collected from all charging piles {U i } k When the encryption master represents The encrypted slave representatives based on the PBFT consensus and the encrypted aggregation node set will reach an agreement on the aggregation results of the ciphertext charging plans for each area.

5. The electric vehicle and charging pile data fusion collection system according to claim 4 is characterized in that: The encryption aggregation layer uses the following calculation formula when determining the charging data aggregation results of each area: In the formula, Represents the encrypted master representative in the encrypted aggregation node set k The calculated aggregation result of the ciphertext charging data of region k; Represents the set of charging piles in area k.

6. The data collection system based on electric vehicle and charging pile data fusion according to claim 3 or 4 is characterized in that: The encrypted aggregation node is configured to perform model aggregation after collecting local models participating in the deep network training deployed by the charging pile node, and adjust the model aggregation weight according to the quality evaluation result of the adaptive aggregation algorithm and the reputation score of the charging pile node in the collaboration process. The collaboration process is a process in which the models of the deep network deployed by each charging pile node are aggregated under federated learning and uniformly transmitted to the encrypted aggregation node for updating.

7. The electric vehicle and charging pile data fusion collection system according to claim 6 is characterized in that: The quality evaluation of the adaptive aggregation algorithm is based on the node Q k In the i-th round of training, the model loss is determined, and the node Q k Participate in the deployment of deep networks in the charging pile nodes and use cross entropy To evaluate the local model loss, the quality assessment weight G of the adaptive aggregation algorithm i (Q k ) is calculated as: Where: is f(x i ),y i The cross entropy between i ) and y i They are the label and observation data of the ciphertext charging data of the i-th round of the encrypted aggregation node set, and N represents the total number of rounds.

8. The electric vehicle and charging pile data fusion collection system according to claim 7 is characterized in that: Each time the model aggregates the G of each node i (Q k ) are all recorded on the corresponding encryption aggregation node, node Q k The reputation in the aggregation process is represented by a cumulative reputation score S(Q k ) is used to represent the reputation score S(Q k ) is calculated as: Where τ is the node Q k The number of times the model participated in the aggregation.

9. The electric vehicle and charging pile data fusion collection system according to claim 8, characterized in that: The global model ζ in the adaptive aggregation algorithm Ω The calculation formula is: In the formula, ζ Ω (t) represents the global model parameters at time t, ζ Ω (t-1) represents the global model parameters at the moment before time t, M is the number of nodes participating in the model aggregation, represents the K-th round broadcast model parameter ζ k Estimates.

10. A method for collecting electric vehicle and charging pile data fusion, characterized in that: The data collection system for electric vehicle and charging pile fusion as claimed in claim 9 comprises the following steps: Create and publish initial encrypted data containing initialization information; The client with a reputation score greater than the set value is obtained to participate in the training. The client uses the charging data and the feature comparison loss function to complete the data standardization model update, and sends the updated local model weight together with the digital signature in the form of encrypted data to the associated packaged data; Receive the local model and verify its legitimacy, and put the transaction with legal signature into the transaction pool; Score and sign all weights in the transaction pool, update the reputation score of each client, calculate the weight of the global model and generate new encrypted data; Verify the legitimacy of encrypted data, calculate the weight of the global model and generate new encrypted data; Download new blocks to obtain global weights, and start training from clients whose reputation scores are greater than the set value, and start the next round of global training until the total number of global training rounds is reached; Download new encrypted data.

Images