A method for transaction security monitoring in vending machines

Abstract

A transaction security monitoring method for vending machines, relating to the field of transaction security monitoring technology, involves constructing a transaction security cloud monitoring center to collect encrypted transaction data uploaded by the vending machines; when a new blockchain node is added to the tail of the blockchain network, a secondary identity verification test or transaction data authenticity assessment is performed; a transaction data authenticity assessment model is constructed, obtaining the threshold assessment ranges for various types of features corresponding to the transaction data in the new blockchain node, obtaining various types of features in the new blockchain node, comparing each type of feature with its corresponding threshold assessment range, and performing a secondary identity verification test or a product release operation based on the comparison results; the secondary identity verification test is performed, and the transaction data authenticity assessment model is updated in real time or a new blockchain node is removed based on the test results, ensuring the security requirements of the vending machine's transaction data.

Description

Technical Field

[0001] This invention relates to the field of transaction security monitoring technology, specifically a transaction security monitoring method for vending machines. Background Technology

[0002] With the development of the Internet of Things, mobile payment, and business automation, intelligent unmanned sales have become a trend in the retail industry, and unmanned vending machines have attracted much attention. The transaction data of unmanned vending machines involves users' personal information and consumption behavior, posing a potential risk of privacy leaks. Monitoring methods may have omissions or false alarms, meaning they may fail to detect fraudulent behavior in a timely manner or mistakenly identify normal transactions as fraud. This can lead to ineffective security issues or cause unnecessary trouble for users. How to collect and analyze transaction data to detect any abnormal patterns or suspicious transactions, such as credit card fraud or the use of false payment information, is a problem that urgently needs to be solved. This paper presents a transaction security monitoring method for unmanned vending machines. Summary of the Invention

[0003] To address the aforementioned technical problems, the present invention aims to provide a transaction security monitoring method for vending machines, comprising the following steps:

[0004] Step S1: Build a transaction security cloud monitoring center, using vending machines as blockchain nodes to collect encrypted transaction data uploaded by vending machines;

[0005] Step S2: When a new blockchain node is added to the tail of the blockchain network, extract the identity information and transaction data from the new blockchain node, obtain the number of other blockchain nodes in the blockchain network whose identity information is consistent with that of the new blockchain node, and perform a secondary identity recognition test or transaction data authenticity assessment based on the number.

[0006] Step S3: Construct a transaction data authenticity assessment model based on graph convolutional long short-term memory network. Output the threshold assessment range of each type of feature corresponding to the transaction data in the new blockchain node through the transaction data authenticity assessment model. Extract features from the transaction data in the new blockchain node to generate each type of feature. Compare each type of feature with the corresponding threshold assessment range of each type of feature. Perform secondary identity recognition test or product outbound operation based on the comparison result.

[0007] Step S4: Conduct a secondary identity verification test, and based on the test results, update the transaction data authenticity assessment model in real time or remove new blockchain nodes.

[0008] Furthermore, the process of building a transaction security cloud monitoring center includes:

[0009] A transaction security cloud monitoring center is constructed using blockchain technology. The transaction security monitoring center is equipped with several blockchain nodes. The blockchain nodes are communicatively connected to unmanned vending machines within a preset range, and each blockchain node is interconnected to form a blockchain network. Each unmanned vending machine is equipped with a data uploading terminal, which is used to upload transaction data to the blockchain node.

[0010] Furthermore, the vending machine is equipped with an interactive interface, through which users can select orders, scan codes to pay, and perform biometric data recognition.

[0011] After a user performs a QR code payment operation through the interactive interface, the system obtains the user's identity information and transaction data based on the order QR code payment operation. The transaction data includes transaction amount, transaction frequency, transaction time period, and vending machine location data. The identity information and transaction data are then encrypted and uploaded to the blockchain node.

[0012] Furthermore, the process of encrypting and uploading identity information and transaction data to the blockchain node includes:

[0013] By pre-setting the public and private keys of each vending machine using an asymmetric encryption algorithm, the identity information and transaction data are pre-processed into fixed-length binary data. The binary data is then encrypted using the vending machine's private key to generate an encrypted data packet, which is then uploaded to the blockchain node. The blockchain node decrypts the received encrypted data packet using the vending machine's public key to generate the identity information and transaction data. Subsequently, a new blockchain node is added at the end of the blockchain network, and the identity information and transaction data are packaged into the new blockchain node.

[0014] Furthermore, when a new blockchain node is added to the tail of the blockchain network, the process of extracting the identity information and transaction data from the new blockchain node, obtaining the number of other blockchain nodes in the blockchain network whose identity information matches that of the new blockchain node, and performing a secondary identity verification test or transaction data authenticity assessment based on the number of nodes includes:

[0015] When a new blockchain node is added to the tail of the blockchain network, the identity information and transaction data of the new blockchain node are extracted. The identity information of the new blockchain node is matched with the identity information of other blockchain nodes in the blockchain network to obtain other blockchain nodes whose identity information matches that of the new blockchain node. The number of the other blockchain nodes is obtained, and a node number threshold is preset. If the number of other blockchain nodes is less than the node number threshold, a secondary identity recognition test is performed.

[0016] If the number of other blockchain nodes is greater than or equal to the node number threshold, the identity information of the new blockchain node is input into the transaction data authenticity assessment model. The transaction data authenticity assessment model outputs the threshold assessment range of each type of feature. The transaction data in the new blockchain node is feature extracted to generate each type of feature. Each type of feature is compared with the corresponding threshold assessment range of each type of feature.

[0017] If all types of features are within the corresponding threshold evaluation range for each type of feature, then proceed with the goods outbound operation;

[0018] If each type of feature is not within the corresponding threshold evaluation range, a secondary identity verification test is performed, and the transaction data authenticity evaluation model is updated in real time based on the test results of the secondary identity verification test.

[0019] Furthermore, the process of constructing a transaction data authenticity assessment model based on graph convolutional long short-term memory networks includes:

[0020] A transaction data authenticity assessment model is constructed based on graph convolutional long short-term memory network. The model retrieves the identity information and transaction data of each blockchain node in the blockchain network. Based on the identity information, the blockchain nodes are aggregated. The features of different transaction time periods in several blockchain nodes with the same identity information are spliced ​​together according to the blockchain nodes to construct a three-dimensional feature matrix. The features include transaction amount and transaction frequency. The three-dimensional feature matrix includes three dimensions: features, transaction time period, and unmanned vending machine location data.

[0021] An adaptive adjacency matrix is ​​constructed based on a three-dimensional feature matrix. The adaptive adjacency matrix is ​​determined by the similarity between each blockchain node in the three-dimensional feature matrix. Graph data is constructed based on the three-dimensional feature matrix and the adaptive adjacency matrix. Multi-layer convolutional operations are performed on the graph data using a graph convolutional neural network.

[0022] After multi-layer graph convolution operation, based on the transaction time period in the three-dimensional feature matrix, the collection order of spatial features of each blockchain node in the three-dimensional feature matrix is ​​obtained. According to the collection order, the spatial features of each blockchain node in the three-dimensional feature matrix are input into the long short-term memory network. The spatiotemporal features of each blockchain node are output through the long short-term memory network. In this way, the spatiotemporal features of each blockchain node corresponding to each different identity information in the blockchain network are obtained.

[0023] A transaction data authenticity assessment model is constructed based on the RBF neural network. The spatiotemporal characteristics of each blockchain node corresponding to different identity information in the blockchain network are used as the test set and training set of the transaction data authenticity assessment model.

[0024] The training set is input into the transaction data authenticity assessment model for training until the loss function is stable and the model parameters are saved. The transaction data authenticity assessment model is then tested using a test set until it meets the preset requirements, and the transaction data authenticity assessment model is output.

[0025] The model for assessing the authenticity of transaction data outputs spatiotemporal characteristic prediction data for different identities.

[0026] Furthermore, the process of extracting features from the transaction data in the new blockchain node by outputting the threshold evaluation range for each type of feature through the transaction data authenticity assessment model, generating various types of features, and comparing each type of feature with its corresponding threshold evaluation range includes:

[0027] Based on the spatiotemporal feature prediction data of different identities, the threshold evaluation ranges of various types of features for different identities are obtained. The threshold evaluation ranges of various types of features include the threshold ranges of transaction amount and transaction frequency corresponding to different vending machine location data in different time periods.

[0028] Based on the identity information in the new blockchain node, obtain the corresponding threshold evaluation ranges for various types of features. Based on the transaction data in the new blockchain node, obtain the location data of the vending machine, the time period of the transaction, the transaction amount, and the transaction frequency. Obtain the transaction amount threshold range and the transaction frequency threshold range corresponding to the location data of the vending machine and the time period of the transaction. Compare the transaction amount and transaction frequency with the transaction amount threshold range and the transaction frequency threshold range.

[0029] Furthermore, the process of conducting secondary identity verification tests and updating the transaction data authenticity assessment model or removing new blockchain nodes in real time based on the test results includes:

[0030] The transaction security cloud monitoring center is equipped with an identity recognition biometric database, which includes biometric data pre-entered by users. The vending machine sends a secondary identity recognition test command through the interactive interface. The user performs a biometric data recognition operation according to the secondary identity recognition test command, generates temporary biometric data of the user, and sends it to the identity recognition biometric database for biometric data matching to determine whether there is biometric data in the identity recognition biometric database that matches the temporary biometric data.

[0031] If it exists, the spatiotemporal characteristics of the new blockchain node are obtained. Based on the spatiotemporal characteristics of each blockchain node corresponding to different identity information in the blockchain network, as well as the test set and training set of the new blockchain node as the transaction data authenticity evaluation model, the spatiotemporal characteristics of the new blockchain node are added to the test set and training set to obtain a new test set and training set. The transaction data authenticity evaluation model is retrained and updated based on the new test set and training set to obtain the updated transaction data authenticity evaluation model.

[0032] If it does not exist, an authentication failure message is sent to the vending machine, and the new blockchain node is removed.

[0033] Compared with the prior art, the beneficial effects of the present invention are:

[0034] 1. When a new blockchain node is added to the tail of the blockchain network, the identity information and transaction data of the new blockchain node are extracted, and the number of other blockchain nodes in the blockchain network that have the same identity information as the new blockchain node is obtained. If the number of other blockchain nodes is less than the node number threshold, a secondary identity recognition test is performed to avoid the transaction data authenticity assessment model being unable to obtain a more accurate assessment of the authenticity of user transaction data due to the sparse or missing historical transaction behavior records of users in the blockchain network.

[0035] 2. A transaction data authenticity assessment model is constructed based on a graph convolutional long short-term memory network. This model outputs threshold assessment ranges for various feature types corresponding to transaction data in new blockchain nodes. It obtains threshold ranges for transaction amount and frequency for different vending machine locations within different time periods, enabling reasonable prediction of user transaction behavior. Features are extracted from the transaction data in the new blockchain nodes to generate various feature types. These features are compared with their corresponding threshold assessment ranges. If a user's transaction behavior does not match their reasonable threshold range, a secondary identity verification test is performed. If the user's biometric data verification passes, the transaction data authenticity assessment model is updated in real time to ensure it effectively identifies fraudulent behavior and guarantees the security of vending machine transaction data. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of a transaction security monitoring method for vending machines according to an embodiment of this application. Detailed Implementation

[0037] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0038] like Figure 1 As shown, a transaction security monitoring method for vending machines includes the following steps:

[0039] Step S1: Build a transaction security cloud monitoring center, using vending machines as blockchain nodes to collect encrypted transaction data uploaded by vending machines;

[0040] Step S2: When a new blockchain node is added to the tail of the blockchain network, extract the identity information and transaction data from the new blockchain node, obtain the number of other blockchain nodes in the blockchain network whose identity information is consistent with that of the new blockchain node, and perform a secondary identity recognition test or transaction data authenticity assessment based on the number.

[0041] Step S3: Construct a transaction data authenticity assessment model based on graph convolutional long short-term memory network. Output the threshold assessment range of each type of feature corresponding to the transaction data in the new blockchain node through the transaction data authenticity assessment model. Extract features from the transaction data in the new blockchain node to generate each type of feature. Compare each type of feature with the corresponding threshold assessment range of each type of feature. Perform secondary identity recognition test or product outbound operation based on the comparison result.

[0042] Step S4: Conduct a secondary identity verification test, and based on the test results, update the transaction data authenticity assessment model in real time or remove new blockchain nodes.

[0043] It should be further explained that, in the specific implementation process, the construction of the transaction security cloud monitoring center includes:

[0044] A transaction security cloud monitoring center is constructed using blockchain technology. The transaction security monitoring center is equipped with several blockchain nodes. The blockchain nodes are communicatively connected to unmanned vending machines within a preset range, and each blockchain node is interconnected to form a blockchain network. Each unmanned vending machine is equipped with a data uploading terminal, which is used to upload transaction data to the blockchain node.

[0045] It should be further explained that, in the specific implementation process, the vending machine is equipped with an interactive interface, through which users can perform order selection, QR code payment and biometric data identification operations;

[0046] After a user performs a QR code payment operation through the interactive interface, the user's identity information and transaction data are obtained based on the order QR code payment operation. The transaction data includes transaction amount, transaction frequency, transaction time period, and vending machine location data. It should be noted that the process of obtaining vending machine location data is as follows: the latitude and longitude information of the vending machine deployment location is obtained through GIS means, the latitude and longitude information of each vending machine is used as the vending machine location data, and the identity information and transaction data are encrypted and uploaded to the blockchain node.

[0047] It should be further explained that, in the specific implementation process, the process of encrypting and uploading identity information and transaction data to the blockchain node includes:

[0048] By pre-setting the public and private keys of each vending machine using an asymmetric encryption algorithm, the identity information and transaction data are pre-processed into fixed-length binary data. The binary data is then encrypted using the vending machine's private key to generate an encrypted data packet, which is then uploaded to the blockchain node. The blockchain node decrypts the received encrypted data packet using the vending machine's public key to generate the identity information and transaction data. Subsequently, a new blockchain node is added at the end of the blockchain network, and the identity information and transaction data are packaged into the new blockchain node.

[0049] It should be further explained that, in the specific implementation process, when a new blockchain node is added to the tail of the blockchain network, the process of extracting the identity information and transaction data from the new blockchain node, obtaining the number of other blockchain nodes in the blockchain network whose identity information matches that of the new blockchain node, and performing secondary identity verification testing or transaction data authenticity assessment based on the stated number includes:

[0050] When a new blockchain node is added to the tail of the blockchain network, the identity information and transaction data of the new blockchain node are extracted. The identity information of the new blockchain node is matched with the identity information of other blockchain nodes in the blockchain network to obtain other blockchain nodes whose identity information matches that of the new blockchain node. The number of the other blockchain nodes is obtained, and a node number threshold is preset. If the number of other blockchain nodes is less than the node number threshold, a secondary identity recognition test is performed.

[0051] If the number of other blockchain nodes is greater than or equal to the node number threshold, the identity information of the new blockchain node is input into the transaction data authenticity assessment model. The transaction data authenticity assessment model outputs the threshold assessment range of each type of feature. The transaction data in the new blockchain node is feature extracted to generate each type of feature. Each type of feature is compared with the corresponding threshold assessment range of each type of feature.

[0052] If all types of features are within the corresponding threshold evaluation range for each type of feature, then proceed with the goods outbound operation;

[0053] If each type of feature is not within the corresponding threshold evaluation range, a secondary identity verification test is performed, and the transaction data authenticity evaluation model is updated in real time based on the test results of the secondary identity verification test.

[0054] It should be further explained that, in the specific implementation process, the process of constructing a transaction data authenticity assessment model based on graph convolutional long short-term memory networks includes:

[0055] A transaction data authenticity assessment model is constructed based on graph convolutional long short-term memory network. The model retrieves the identity information and transaction data of each blockchain node in the blockchain network. Based on the identity information, the blockchain nodes are aggregated. The features of different transaction time periods in several blockchain nodes with the same identity information are spliced ​​together according to the blockchain nodes to construct a three-dimensional feature matrix. The features include transaction amount and transaction frequency. The three-dimensional feature matrix includes three dimensions: features, transaction time period, and unmanned vending machine location data.

[0056] It should be further explained that the formula for the three-dimensional feature matrix is:

[0057] Q = concact{W1,W2,...,Wj,...,WL}, j = 1,2,...,L, where Q represents the three-dimensional feature matrix. Let L represent the two-dimensional data matrix of the j-th blockchain node, where concact represents the concatenation of the two-dimensional data matrices of each blockchain node according to the blockchain node, and L represents the total number of blockchain nodes.

[0058] =( , ,... ,..., ), n = 1, 2, ..., k;

[0059] in, The features of the nth transaction time period of the j-th blockchain node are represented, including transaction amount and transaction frequency, where n represents the temporal relationship of each feature of the j-th blockchain node, the transaction time periods of each feature are converted into the corresponding collection order, and k represents the total number of the collection order.

[0060] An adaptive adjacency matrix is ​​constructed based on a three-dimensional feature matrix. The adaptive adjacency matrix is ​​determined by the similarity between each blockchain node in the three-dimensional feature matrix. Graph data is constructed based on the three-dimensional feature matrix and the adaptive adjacency matrix. Multi-layer convolutional operations are performed on the graph data using a graph convolutional neural network.

[0061] It should be further explained that an adaptive adjacency matrix is ​​constructed based on the three-dimensional feature matrix. This adaptive adjacency matrix is ​​determined by the similarity between the blockchain nodes in the three-dimensional feature matrix. The formula for constructing the similarity between the blockchain nodes in the three-dimensional feature matrix is ​​as follows:

[0062] ;

[0063] Where q(dp) represents the similarity between the d-th blockchain node and the p-th blockchain node; The transaction amount of the d-th blockchain node during the t-th transaction period; This represents the average transaction amount of the d-th blockchain node; The transaction amount of the p-th blockchain node during the t-th transaction time period; This represents the average transaction amount of the p-th blockchain node; The transaction frequency of the d-th blockchain node during the t-th transaction time period; This represents the average transaction frequency of the d-th blockchain node; The transaction frequency of the p-th blockchain node during the t-th transaction time period; The p-th blockchain node represents the average transaction frequency; s represents the total number of transaction time periods. , This represents the weighting factor.

[0064] The similarity between blockchain nodes is used as the attention weight between them. A graph convolutional neural network is used to perform multiple convolution operations on the graph data. The attention weight is then combined with the features of each blockchain node for feature aggregation, updating the self-representation of each blockchain node. The embedded representation of the blockchain node is obtained according to the neighbor aggregation mechanism. The formula for aggregating neighboring nodes is as follows:

[0065] ,in, An embedded representation of a blockchain node. ) represents the activation function. Represents the feature transformation parameter matrix. This represents the set of other blockchain nodes.

[0066] After multi-layer graph convolution operation, based on the transaction time period in the three-dimensional feature matrix, the collection order of spatial features of each blockchain node in the three-dimensional feature matrix is ​​obtained. According to the collection order, the spatial features of each blockchain node in the three-dimensional feature matrix are input into the long short-term memory network. The spatiotemporal features of each blockchain node are output through the long short-term memory network. In this way, the spatiotemporal features of each blockchain node corresponding to each different identity information in the blockchain network are obtained.

[0067] A transaction data authenticity assessment model is constructed based on the RBF neural network. The spatiotemporal characteristics of each blockchain node corresponding to different identity information in the blockchain network are used as the test set and training set of the transaction data authenticity assessment model.

[0068] The training set is input into the transaction data authenticity assessment model for training until the loss function is stable and the model parameters are saved. The transaction data authenticity assessment model is then tested using a test set until it meets the preset requirements, and the transaction data authenticity assessment model is output.

[0069] The model for assessing the authenticity of transaction data outputs spatiotemporal characteristic prediction data for different identities.

[0070] It should be further explained that, in the specific implementation process, the process of extracting features from the transaction data in the new blockchain node by outputting the threshold evaluation ranges for each type of feature through the transaction data authenticity evaluation model, generating various types of features, and comparing each type of feature with its corresponding threshold evaluation range includes:

[0071] Based on the spatiotemporal feature prediction data of different identities, the threshold evaluation ranges of various types of features for different identities are obtained. The threshold evaluation ranges of various types of features include the threshold ranges of transaction amount and transaction frequency corresponding to different vending machine location data in different time periods.

[0072] Based on the identity information in the new blockchain node, obtain the corresponding threshold evaluation ranges for various types of features. Based on the transaction data in the new blockchain node, obtain the location data of the vending machine, the time period of the transaction, the transaction amount, and the transaction frequency. Obtain the transaction amount threshold range and the transaction frequency threshold range corresponding to the location data of the vending machine and the time period of the transaction. Compare the transaction amount and transaction frequency with the transaction amount threshold range and the transaction frequency threshold range.

[0073] It should be further explained that, in the specific implementation process, the real-time updating of the transaction data authenticity assessment model or the new blockchain node removal operation based on the test results of the secondary identity verification test includes:

[0074] The transaction security cloud monitoring center is equipped with an identity recognition biometric database, which includes biometric data pre-entered by users, including facial data, iris data, etc. The vending machine sends a secondary identity recognition test command through the interactive interface. The user performs a biometric data recognition operation according to the secondary identity recognition test command, generates temporary biometric data of the user, and sends it to the identity recognition biometric database for biometric data matching to determine whether there is biometric data in the identity recognition biometric database that matches the temporary biometric data.

[0075] If it exists, the spatiotemporal characteristics of the new blockchain node are obtained. Based on the spatiotemporal characteristics of each blockchain node corresponding to different identity information in the blockchain network, as well as the test set and training set of the new blockchain node as the transaction data authenticity evaluation model, the spatiotemporal characteristics of the new blockchain node are added to the test set and training set to obtain a new test set and training set. The transaction data authenticity evaluation model is retrained and updated based on the new test set and training set to obtain the updated transaction data authenticity evaluation model.

[0076] If it does not exist, an authentication failure message is sent to the vending machine, and the new blockchain node is removed.

[0077] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for monitoring transaction security in vending machines, characterized in that, Includes the following steps: Step S1: Build a transaction security cloud monitoring center, using vending machines as blockchain nodes to collect encrypted transaction data uploaded by vending machines; Step S2: When a new blockchain node is added to the tail of the blockchain network, extract the identity information and transaction data from the new blockchain node, obtain the number of other blockchain nodes in the blockchain network whose identity information is consistent with that of the new blockchain node, and perform a secondary identity recognition test or transaction data authenticity assessment based on the number. Step S3: Construct a transaction data authenticity assessment model based on graph convolutional long short-term memory network. Output the threshold assessment range of each type of feature corresponding to the transaction data in the new blockchain node through the transaction data authenticity assessment model. Extract features from the transaction data in the new blockchain node to generate each type of feature. Compare each type of feature with the corresponding threshold assessment range of each type of feature. Perform secondary identity recognition test or product outbound operation based on the comparison result. Step S4: Conduct a secondary identity verification test, and based on the test results, update the transaction data authenticity assessment model in real time or remove new blockchain nodes. The process of building a transaction security cloud monitoring center includes: A transaction security cloud monitoring center is constructed using blockchain technology. The transaction security monitoring center is equipped with several blockchain nodes. The blockchain nodes are connected to unmanned vending machines within a preset range, and each blockchain node is interconnected to form a blockchain network. Each unmanned vending machine is equipped with a data uplink segment, which is used to upload transaction data to the blockchain node. The vending machine is equipped with an interactive interface, through which users can select orders, scan codes to pay, and perform biometric data recognition. When a user performs a QR code payment operation through the interactive interface, the user's identity information and transaction data are obtained based on the order QR code payment operation. The transaction data includes transaction amount, transaction frequency, transaction time period and unmanned vending machine location data. The identity information and transaction data are then encrypted and uploaded to the blockchain node. The process of encrypting and uploading identity information and transaction data to a blockchain node includes: By pre-setting the public and private keys of each vending machine using an asymmetric encryption algorithm, the identity information and transaction data are pre-processed into fixed-length binary data. The binary data is then encrypted using the vending machine's private key to generate an encrypted data packet, which is then uploaded to the blockchain node. The blockchain node decrypts the received encrypted data packet using the vending machine's public key to generate the identity information and transaction data. Subsequently, a new blockchain node is added at the end of the blockchain network, and the identity information and transaction data are packaged into the new blockchain node. When a new blockchain node is added to the tail of the blockchain network, the process of extracting the identity information and transaction data from the new blockchain node, obtaining the number of other blockchain nodes in the blockchain network whose identity information matches that of the new blockchain node, and performing a secondary identity verification test or transaction data authenticity assessment based on the number of nodes includes: When a new blockchain node is added to the tail of the blockchain network, the identity information and transaction data of the new blockchain node are extracted. The identity information of the new blockchain node is matched with the identity information of other blockchain nodes in the blockchain network to obtain other blockchain nodes whose identity information matches that of the new blockchain node. The number of the other blockchain nodes is obtained, and a node number threshold is preset. If the number of other blockchain nodes is less than the node number threshold, a secondary identity recognition test is performed. If the number of other blockchain nodes is greater than or equal to the node number threshold, the identity information of the new blockchain node is input into the transaction data authenticity assessment model. The transaction data authenticity assessment model outputs the threshold assessment range of each type of feature. The transaction data in the new blockchain node is feature extracted to generate each type of feature. Each type of feature is compared with the corresponding threshold assessment range of each type of feature. If all types of features are within the corresponding threshold evaluation range for each type of feature, then proceed with the goods outbound operation; If each type of feature is not within the corresponding threshold evaluation range, a secondary identity recognition test is conducted, and the transaction data authenticity evaluation model is updated in real time based on the test results of the secondary identity recognition test. The process of constructing a transaction data authenticity assessment model based on graph convolutional long short-term memory networks includes: A transaction data authenticity assessment model is constructed based on graph convolutional long short-term memory network. The model retrieves the identity information and transaction data of each blockchain node in the blockchain network. Based on the identity information, the blockchain nodes are aggregated. The features of different transaction time periods in several blockchain nodes with the same identity information are spliced ​​together according to the blockchain nodes to construct a three-dimensional feature matrix. The features include transaction amount and transaction frequency. The three-dimensional feature matrix includes three dimensions: features, transaction time period, and unmanned vending machine location data. An adaptive adjacency matrix is ​​constructed based on a three-dimensional feature matrix. The adaptive adjacency matrix is ​​determined by the similarity between each blockchain node in the three-dimensional feature matrix. Graph data is constructed based on the three-dimensional feature matrix and the adaptive adjacency matrix. Multi-layer convolutional operations are performed on the graph data using a graph convolutional neural network. After multi-layer graph convolution operation, based on the transaction time period in the three-dimensional feature matrix, the collection order of spatial features of each blockchain node in the three-dimensional feature matrix is ​​obtained. According to the collection order, the spatial features of each blockchain node in the three-dimensional feature matrix are input into the long short-term memory network. The spatiotemporal features of each blockchain node are output through the long short-term memory network. In this way, the spatiotemporal features of each blockchain node corresponding to each different identity information in the blockchain network are obtained. A transaction data authenticity assessment model is constructed based on the RBF neural network. The spatiotemporal characteristics of each blockchain node corresponding to different identity information in the blockchain network are used as the test set and training set of the transaction data authenticity assessment model. The training set is input into the transaction data authenticity assessment model for training until the loss function is stable and the model parameters are saved. The transaction data authenticity assessment model is then tested using a test set until it meets the preset requirements, and the transaction data authenticity assessment model is output. The model for assessing the authenticity of transaction data outputs spatiotemporal characteristic prediction data for different identities. The process of extracting features from transaction data in new blockchain nodes by outputting threshold evaluation ranges for various types of features through a transaction data authenticity assessment model, and comparing each type of feature with its corresponding threshold evaluation range includes: Based on the spatiotemporal feature prediction data of different identities, the threshold evaluation ranges of various types of features for different identities are obtained. The threshold evaluation ranges of various types of features include the threshold ranges of transaction amount and transaction frequency corresponding to different vending machine location data in different time periods. Based on the identity information in the new blockchain node, obtain the corresponding threshold evaluation ranges for various types of features. Based on the transaction data in the new blockchain node, obtain the location data of the vending machine, the time period of the transaction, the transaction amount, and the transaction frequency. Obtain the transaction amount threshold range and the transaction frequency threshold range corresponding to the location data of the vending machine and the time period of the transaction. Compare the transaction amount and transaction frequency with the transaction amount threshold range and the transaction frequency threshold range. The process of conducting secondary identity verification tests and updating the transaction data authenticity assessment model or removing new blockchain nodes in real time based on the test results includes: The transaction security cloud monitoring center is equipped with an identity recognition biometric database, which includes biometric data pre-entered by users. The vending machine sends a secondary identity recognition test command through the interactive interface. The user performs a biometric data recognition operation according to the secondary identity recognition test command, generates temporary biometric data of the user, and sends it to the identity recognition biometric database for biometric data matching to determine whether there is biometric data in the identity recognition biometric database that matches the temporary biometric data. If it exists, the spatiotemporal characteristics of the new blockchain node are obtained. Based on the spatiotemporal characteristics of each blockchain node corresponding to different identity information in the blockchain network, as well as the test set and training set of the new blockchain node as the transaction data authenticity evaluation model, the spatiotemporal characteristics of the new blockchain node are added to the test set and training set to obtain a new test set and training set. The transaction data authenticity evaluation model is retrained and updated based on the new test set and training set to obtain the updated transaction data authenticity evaluation model. If it does not exist, an authentication failure message is sent to the vending machine, and the new blockchain node is removed.

Images