A security control method for financial accounts
By registering financial account application scenarios in the blockchain network and monitoring node behavior in real time, and dynamically assessing risks, the problem of flexibility and security limitation in financial account security control in the existing technology is solved, and more efficient risk response is achieved and accuracy.
Patent Information
- Application Number
- CN202410725372.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-06
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2044-06-06
AI Technical Summary
The existing technology lacks real-time monitoring and dynamic assessment capabilities for node behavior and associated risks in financial account security control, resulting in limited system flexibility and security, and the risk assessment mechanism cannot be adjusted in time.
Register a financial account application scenario in the blockchain network as a network node, evaluate the importance of nodes, monitor node behavior in real time, build a risk assessment model, dynamically set evaluation thresholds, and implement processing measures based on the risk assessment results, and record all data storage.
By dynamically assigning node roles and real-time risk assessment, the flexibility and security of node control are improved, and the response speed and accuracy of financial risks are improved.
Smart Images

Figure CN118469577B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of information security control, and in particular to a security control method for financial accounts. Background Art
[0002] In the era of digital finance, the security of financial accounts has become the focus of the industry. Traditional financial account security control methods rely on static authentication mechanisms, such as passwords and two-step verification. These methods are incapable of coping with increasingly complex network attacks and internal risk management. Especially in the context of the rapid development of financial technology, how to ensure the security and transparency of financial transactions while improving the flexibility and response speed of the system has become an important direction of technological development. In recent years, blockchain technology has been widely regarded as one of the key technologies for the innovation of the financial industry due to its unique advantages in security, transparency and immutability. The decentralized nature of blockchain reduces the risk of single point failures, and the public verifiability of all transaction records enhances the trust of the system. However, existing technologies often treat financial application scenarios as static nodes, lacking the ability to monitor and dynamically evaluate node behavior and associated risks in real time, which limits the flexibility and security of the system. In addition, the risk assessment mechanism is mostly preset and fixed, and cannot be adjusted in time according to actual environmental changes. Summary of the invention
[0003] In view of the above problems existing in the existing security control methods for financial accounts, the present invention is proposed.
[0004] Therefore, the problem to be solved by the present invention is that the existing technology often treats financial application scenarios as static nodes, lacks the ability to monitor and dynamically evaluate node behaviors and associated risks in real time, which limits the flexibility and security of the system. In addition, the risk assessment mechanism is mostly preset and fixed, and cannot be adjusted in time according to actual environmental changes.
[0005] To solve the above technical problems, the present invention provides the following technical solutions: a security control method for financial accounts, which includes registering financial account application scenarios as network nodes in the blockchain network, evaluating node importance, and setting node roles; real-time monitoring of node behavior, building a risk assessment model to perform risk assessment on the nodes; dynamically setting assessment thresholds, and implementing processing measures on the nodes based on the risk assessment results and the assessment thresholds; and recording and storing all data.
[0006] As a preferred solution of the security control method for financial accounts described in the present invention, wherein: the registration of financial account application scenarios as network nodes in the blockchain network refers to generating a unique node identity for each financial account application scenario, including a public key and a private key. The financial account application scenario uses the private key to sign the registration request and the public key and sends them to the blockchain network simultaneously. The blockchain network smart contract receives the registration request to verify the validity. After the verification is passed, the financial account application scenario information generates a network node and records it in the blockchain network.
[0007] As a preferred solution of the security control method for financial accounts of the present invention, wherein: the node importance evaluation and node role setting refer to:
[0008] Extracting node features includes:
[0009] Compute node efficiency affects NEIS: , where N is the number of transactions processed by the node within the considered time. is the processing time of the ith transaction, is the request time of the i-th transaction;
[0010] Calculate the interactive node trust propagation TPS: Where J is the total number of nodes interacting with the node, is the interaction weight of node j, is the historical trust propagation score of node j;
[0011] Compute node network stability NSCS: , where H is the number of hours the node is online, and F is the number of hours the node is down;
[0012] Combine the extracted node features to form a feature vector ;
[0013] Design the multi-layer perceptron structure as follows: , where x is the input node feature vector, , as well as is the weight of each layer of the multi-layer perceptron, , as well as is the bias of each layer of the multilayer perceptron, , is the output of the hidden layer in the multi-layer perceptron, It is the final output of the multi-layer perceptron, representing the importance of the node;
[0014] Use the training set to train the multilayer perceptron, annotate the training set data, define the mean square error as the loss function, and use the Adam optimizer to optimize the multilayer perceptron parameters, with the optimization goal of minimizing the loss.
[0015] After the trained multilayer perceptron is obtained, the extracted node feature vector is input into the multilayer perceptron to obtain the node importance, and the most important node is set as the central node. The central node generates node commands and comprehensively manages the remaining nodes.
[0016] As a preferred solution of the security control method for financial accounts described in the present invention, the real-time monitoring of node behavior refers to the central node generating a node behavior monitoring command and sending it to the remaining nodes, all nodes monitor their own behavior in real time and transmit the monitoring data to the central node through an encrypted transmission channel, and the central node cleans and filters the data after receiving the monitoring data and stores it in the central database.
[0017] As a preferred solution of the security control method for financial accounts of the present invention, wherein: the risk assessment model is constructed to conduct risk assessment on the node, including:
[0018] Design a risk assessment model based on the CNN model, including input layer, convolution layer, pooling layer, fully connected layer and output layer;
[0019] The convolution layer consists of two layers. The first convolution layer is a 1D convolution layer using 16 filters, and the second convolution layer is a 1D convolution layer using 32 filters. The size of each filter is set to 3. Both the first and second convolution layers use the ReLU function as the activation function.
[0020] The pooling layer is connected after the convolution layer, and the pooling size is set to 2 to perform maximum pooling on the convolution layer output;
[0021] The fully connected layer is set after the pooling layer, and 64 neurons are set to process the output of the pooling layer;
[0022] After the output layer is connected to the fully connected layer, sigmoid is used as the activation function to process the output of the last neuron in the fully connected layer and finally output it;
[0023] After building the risk assessment model, use the training set to train the model, annotate the training set data simultaneously, and define the loss function as:
[0024] ,
[0025] in For loss, is the total number of training set data, is the actual label of the i-th training data, is the output of the risk assessment model for the i-th training data;
[0026] Use the Adam optimizer to iteratively train the model, update the model parameters and calculate the loss after each iteration, and stop the iteration until the model loss does not decrease significantly to obtain the trained risk assessment model;
[0027] The risk assessment model is deployed to the central database to perform risk assessment on the collected behavior monitoring data of each node to obtain the risk assessment result R.
[0028] As a preferred solution of the security control method for financial accounts of the present invention, the dynamically set evaluation threshold refers to setting the initial threshold , after obtaining the new node risk assessment result R, calculate the mean of all node risk assessment results and standard deviation , using the logistic loss function to calculate the threshold update gradient:
[0029] ,
[0030] ,in is the logistic loss, Set to , where +1 represents risk and -1 represents no risk. To evaluate the threshold, is the updated gradient of k;
[0031] Dynamic update of evaluation threshold based on threshold update gradient: in is the learning rate.
[0032] As a preferred solution of the security control method for financial accounts of the present invention, wherein: the implementation of processing measures on nodes according to risk assessment results and assessment thresholds refers to obtaining the risk assessment result R through the risk assessment model and comparing it with the dynamically updated assessment threshold k to determine the node risk:
[0033] If R < k, the node is judged to be risk-free, and the central node assigns a detected mark to the node, regularly conducts risk assessment and judgment on the node, and generates records of risk assessment results and risk judgment results;
[0034] If R ≥ k, the node is judged to be at risk. The central node generates an unbinding command and sends it to the risk node. The smart contract automatically executes the unbinding operation to prevent all actions of the risk node, isolates the risk node from the blockchain network, and transfers it to the security sandbox for security risk analysis and processing.
[0035] As a preferred solution of the security control method for financial accounts described in the present invention, wherein: the recording and storage of all data refers to the central database synchronously generating records for storage of risk assessment results, risk assessment thresholds, risk judgment results and processing measures and setting security access measures, and backing up all stored data in the cloud, regularly performing integrity and consistency checks on stored data and backup data, generating and storing test records after the test, adding timestamp data, and adding them.
[0036] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of the above-mentioned security control method for financial accounts when executing the computer program.
[0037] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned security control method for a financial account.
[0038] The beneficial effects of the present invention are as follows: the present invention registers financial account application scenarios as nodes and dynamically allocates node roles, which greatly improves the flexibility and security of node control, collects node behavior data for node risk assessment, and dynamically sets assessment thresholds for node risk judgment, effectively improving the response speed and accuracy to financial risks. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.
[0040] Figure 1 The figure is a flowchart of a security control method for a financial account.
[0041] Figure 2 A schematic diagram of an implementation of a security control method for a financial account. DETAILED DESCRIPTION
[0042] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.
[0043] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0044] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments. Example 1
[0045] Reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and which provides a security control method for a financial account. The security control method for a financial account includes:
[0046] S1. Register the financial account application scenario as a network node in the blockchain network, evaluate the importance of the node, and set the node role;
[0047] Specifically, registering a financial account application scenario as a network node in a blockchain network means generating a unique node identity for each financial account application scenario, including a public key and a private key. The financial account application scenario uses the private key to sign a registration request and sends the public key to the blockchain network simultaneously. The blockchain network smart contract receives the registration request to verify its validity. After verification, the financial account application scenario information generates a network node and records it in the blockchain network.
[0048] By generating an independent public-private key pair for each financial account, the uniqueness of the transaction identity and the security of the transaction data are ensured. The public key can be made public to identify the node identity, while the private key remains private and is used to sign the transaction. Such a mechanism significantly enhances the immutability of the transaction and the reliability of identity authentication, and effectively prevents identity forgery and replay attacks, which are common in traditional financial systems. Through unique identity identification, the authenticity of each transaction can be independently verified through the blockchain network. The private key signature ensures that the sent data has not been tampered with during transmission. Any illegal modification of the registration data will cause the signature verification to fail, thereby ensuring the integrity and credibility of the data. Smart contracts automatically handle the verification process, reduce manual intervention, improve processing efficiency and reduce the possibility of errors or fraud. Once information is recorded on the blockchain, it cannot be changed or deleted, providing a reliable audit trail for the historical activities of financial accounts.
[0049] Furthermore, the importance of nodes is evaluated and the role of nodes is set.
[0050] Extracting node features includes:
[0051] Compute node efficiency affects NEIS: , where N is the number of transactions processed by the node within the considered time. is the processing time of the ith transaction, is the request time of the i-th transaction;
[0052] Calculate the interactive node trust propagation TPS: , where J is the total number of nodes interacting with the node, is the interaction weight of node j, is the historical trust propagation score of node j;
[0053] Compute node network stability NSCS: Where H is the number of online hours of the node, and F is the number of node failure hours;
[0054] Combine the extracted node features to form a feature vector ;
[0055] Design the multi-layer perceptron structure as follows:
[0056] , where x is the input node feature vector, , as well as is the weight of each layer of the multi-layer perceptron, , as well as is the bias of each layer of the multilayer perceptron, , is the output of the hidden layer in the multi-layer perceptron, It is the final output of the multi-layer perceptron, representing the importance of the node;
[0057] Use the training set to train the multilayer perceptron, annotate the training set data, define the mean square error as the loss function, and use the Adam optimizer to optimize the multilayer perceptron parameters, with the optimization goal of minimizing the loss.
[0058] After the trained multilayer perceptron is obtained, the extracted node feature vector is input into the multilayer perceptron to obtain the node importance, and the most important node is set as the central node. The central node generates node commands and comprehensively manages the remaining nodes. The node importance will be updated in real time according to the collected data, and the central node role will be reallocated according to the updated node importance.
[0059] By evaluating the efficiency of nodes in processing transactions, nodes with strong processing capabilities can be dynamically identified. This quantitative method directly reflects the actual contribution of each node to network efficiency, thereby promoting the rational allocation of resources. By quantifying the trust relationship and historical interaction quality between nodes, this step provides an objective trust evaluation mechanism. This mechanism evaluates the trust of nodes based on historical data and the interaction between nodes. By dynamically monitoring and evaluating node stability, it ensures that the network can respond quickly and adjust resource allocation when facing node failures. Taking into account multiple influencing factors, (multi-layer perceptron) is used for automatic learning and evaluation of node importance. This makes the allocation of node roles not only based on a single indicator, but a comprehensive feature vector, which increases the accuracy of the evaluation and the adaptability of the network. Traditional node evaluation methods may ignore the complex relationships between nodes and the impact of multi-dimensional performance indicators. Through machine learning methods, the present invention can learn and simulate these complex relationships, providing a more efficient and intelligent solution for node management.
[0060] S2, monitor node behavior in real time and build a risk assessment model to conduct risk assessment on nodes;
[0061] Specifically, real-time monitoring of node behavior means that the central node generates a node behavior monitoring command and sends it to other nodes. All nodes monitor their own behavior in real time and transmit the monitoring data to the central node through an encrypted transmission channel. After receiving the monitoring data, the central node cleans and filters the data and stores it in the central database.
[0062] By centrally managing and formulating monitoring commands, it can be ensured that all nodes collect and report data according to unified standards, enhancing the consistency and comparability of monitoring data, solving the data inconsistency and monitoring blind spots that may be caused by each node determining its own monitoring standards in a decentralized system, and improving the comprehensiveness and accuracy of network monitoring. Encrypted transmission ensures the security and privacy of data during transmission, prevents data from being intercepted or tampered with during transmission, effectively prevents the risk of network spies or malicious attackers intercepting and using node data, and enhances the security of data transmission. Data cleaning and filtering can reduce the waste of storage space and improve data processing efficiency, while also enhancing data availability and ease of use, effectively solving the problem of data redundancy in large-scale data collection, improving the efficiency of data processing and analysis, and ensuring the quality and reliability of data stored in the database.
[0063] Furthermore, a risk assessment model is constructed to conduct risk assessment on nodes, including:
[0064] Design a risk assessment model based on the CNN model, including input layer, convolution layer, pooling layer, fully connected layer and output layer;
[0065] The convolution layer consists of two layers. The first convolution layer is a 1D convolution layer using 16 filters, and the second convolution layer is a 1D convolution layer using 32 filters. The size of each filter is set to 3. Both the first and second convolution layers use the ReLU function as the activation function.
[0066] The pooling layer is connected after the convolution layer, and the pooling size is set to 2 to perform maximum pooling on the convolution layer output;
[0067] The fully connected layer is set after the pooling layer, and 64 neurons are set to process the output of the pooling layer;
[0068] After the output layer is connected to the fully connected layer, sigmoid is used as the activation function to process the output of the last neuron in the fully connected layer and finally output it;
[0069] After building the risk assessment model, use the training set to train the model, annotate the training set data simultaneously, and define the loss function as:
[0070] ,in For loss, is the total number of training set data, is the actual label of the i-th training data, is the output of the risk assessment model for the i-th training data; the Adam optimizer is used to iteratively train the model, and the model parameters are updated after each iteration to calculate the loss. The iteration is stopped until the model loss does not decrease significantly to obtain the trained risk assessment model;
[0071] The risk assessment model is deployed to the central database to perform risk assessment on the collected behavior monitoring data of each node to obtain the risk assessment result R.
[0072] By using two convolutional layers with increasing numbers of filters, the model can extract features from node behavior data more deeply, and gradually abstract the risk features of node behavior from simple to complex, solving the problem of insufficient recognition of complex behavior patterns in traditional risk assessment, and providing a method that can dynamically learn and adapt to new risk patterns. It is particularly suitable for processing large-scale and dynamically changing node behavior data. The cross entropy loss and Adam optimizer are used to optimize the efficiency and effect of model training. Especially when dealing with classification problems, this method can converge quickly and reduce the error rate. In a dynamic and changing network environment, it can effectively adapt to new risk scenarios and reduce risk judgment errors caused by insufficient model training, thereby improving the accuracy and reliability of risk assessment. By deploying the trained model in a central database, centralized management and real-time update of risk assessment logic can be achieved, ensuring that all node behavior data are processed in accordance with the latest risk assessment standards, effectively combining the risk assessment model with the actual operating status of the network, so that risk assessment can be based on real-time data, greatly improving the response speed and timeliness. At the same time, the centralized model update and deployment strategy also ensures the consistency of the assessment standards and the timeliness of the update.
[0073] S3, dynamically set the assessment threshold, and implement processing measures on the node according to the risk assessment results and the assessment threshold;
[0074] Specifically, dynamically setting the evaluation threshold means setting the initial threshold , after obtaining the new node risk assessment result R, calculate the mean of all node risk assessment results and standard deviation , using the logistic loss function to calculate the threshold update gradient:
[0075] ,in is the logistic loss, Set to , where +1 represents risk and -1 represents no risk. To evaluate the threshold, is the updated gradient of k;
[0076] Dynamic update of evaluation threshold based on threshold update gradient:
[0077] ,in is the learning rate.
[0078] By calculating the threshold update gradient in real time and making dynamic adjustments, the system's response speed to emerging risks is greatly improved. In financial networks such as blockchain, transactions and interactions are frequent and changeable, and fixed assessment thresholds are difficult to adapt to such a rapidly changing environment. By updating the threshold in real time, the system can respond quickly at the early stage of risk occurrence, thereby reducing potential economic losses. Dynamic threshold adjustment provides a method to adjust the risk assessment criteria in a personalized manner according to the specific situation and historical performance of each node, which not only increases the flexibility of the assessment model, but also improves its accuracy, because the threshold can be optimized according to real-time data and the specific behavior of the node, which is closer to the actual operating environment and risk conditions. By dynamically adjusting the threshold using a logical loss function, the present invention effectively balances the risks of false positives and false negatives. In the security-sensitive financial field, an excessively high false positive rate may lead to a waste of system resources, while a false negative may lead to real risks that are not handled in a timely manner, both of which will cause economic losses or reputation damage. The logical loss function helps the system to more accurately locate and adjust the risk threshold, making risk judgment more reasonable and accurate.
[0079] Further, implementing treatment measures on the node according to the risk assessment result and the assessment threshold refers to obtaining the risk assessment result R through the risk assessment model and comparing it with the dynamically updated assessment threshold k to determine the node risk:
[0080] If R < k, the node is judged to be risk-free, and the central node assigns a detected mark to the node, regularly conducts risk assessment and judgment on the node, and generates records of risk assessment results and risk judgment results;
[0081] If R ≥ k, the node is judged to be at risk. The central node generates an unbinding command and sends it to the risk node. The smart contract automatically executes the unbinding operation to prevent all actions of the risk node, isolates the risk node from the blockchain network, and transfers it to the security sandbox for security risk analysis and processing.
[0082] The introduction of the dynamic threshold k allows the system to flexibly adjust the risk assessment criteria according to the current network status and historical risk data, so as to more accurately match the actual security needs of the network, solve the problem of over-response or under-response that may be caused by static thresholds, ensure that risk assessment is synchronized with network status, and improve the effectiveness and timeliness of defense measures. Regular risk reassessment provides a continuous security mechanism for the network, ensuring that even nodes that are initially judged to be risk-free can be identified and handled in a timely manner if they show risk characteristics in subsequent operations, effectively preventing long-term or latent risks that may be ignored in one-time risk assessments, and enhancing the overall security and stability of the network. The immediate unbinding and isolation of high-risk nodes avoids the spread of potential security threats, and further analysis in the security sandbox helps to identify the root causes of risks, provide data support for preventing similar risks in the future, and solve the immediate threats that risk nodes may pose to the overall security and stability of the network. Through immediate isolation and in-depth analysis, not only the spread of risks is prevented, but also the understanding of the source of risks and future prevention capabilities are improved.
[0083] S4. Record and store all data;
[0084] Specifically, recording and storing all data means that the central database will synchronously generate records for storage of risk assessment results, risk assessment thresholds, risk judgment results and treatment measures and set up security access measures, and back up all stored data in the cloud, regularly perform integrity and consistency checks on stored data and backup data, and generate and store detection records after detection, add timestamp data, and store them.
[0085] By centrally recording key risk management data, risk management becomes more systematic and standardized, which helps to quickly identify and respond to risk changes in the network, solves the problem of data inconsistency that may occur when managing data in a decentralized manner, and improves the efficiency and accuracy of data management. The use of cloud backup and secure access strategies can ensure that important information can be quickly restored in the event of local data loss or damage, ensuring the continuous operation of the network, providing a solution to combat the risk of data loss caused by physical damage, network attacks or system failures, and ensuring the persistence and integrity of data. Regular testing provides a self-audit mechanism to ensure that stored data has not been modified or damaged without authorization, enhancing data reliability, solving the problem of tampering or misoperation that may occur during long-term storage of data, and ensuring the accuracy of data and the effectiveness of risk handling measures. Example 2
[0086] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.
[0087] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.
[0088] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.
[0089] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit with a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit with a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.
[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A security control method for a financial account, characterized in that: include, Register financial account application scenarios as network nodes in the blockchain network, evaluate node importance, and set node roles; Monitor node behavior in real time and build a risk assessment model to conduct risk assessment on nodes; Dynamically set assessment thresholds and implement processing measures on nodes based on risk assessment results and assessment thresholds; Record and store all data; Registering the financial account application scenario as a network node in the blockchain network refers to generating a unique node identity for each financial account application scenario, including a public key and a private key. The financial account application scenario uses the private key to sign a registration request and sends the public key to the blockchain network simultaneously. The blockchain network smart contract receives the registration request to verify the validity. After the verification is passed, the financial account application scenario information generates a network node and records it in the blockchain network; The evaluating node importance and setting the node role refers to: Extracting node features includes: Compute node efficiency affects NEIS: ; Where N is the number of transactions processed by the node within the considered time. is the processing time of the ith transaction, is the request time of the i-th transaction; Calculate the interactive node trust propagation TPS: ; Where J is the total number of nodes interacting with the node, is the interaction weight of node j, is the historical trust propagation score of node j; Compute node network stability NSCS: ; Where H is the number of online hours of the node, and F is the number of node failure hours; Combine the extracted node features to form a feature vector ; Design the multi-layer perceptron structure as follows: ; ; ; Where x is the input node feature vector, , as well as is the weight of each layer of the multi-layer perceptron, , as well as is the bias of each layer of the multilayer perceptron, , is the output of the hidden layer in the multi-layer perceptron, It is the final output of the multi-layer perceptron, representing the importance of the node; Use the training set to train the multilayer perceptron, annotate the training set data, define the mean square error as the loss function, and use the Adam optimizer to optimize the multilayer perceptron parameters, with the optimization goal of minimizing the loss. After the trained multi-layer perceptron is obtained, the extracted node feature vector is input into the multi-layer perceptron to obtain the node importance, and the most important node is set as the central node, which generates node commands and comprehensively manages the remaining nodes; Real-time monitoring of node behavior refers to the central node generating a node behavior monitoring command and sending it to other nodes. All nodes monitor their own behavior in real time and transmit the monitoring data to the central node through an encrypted transmission channel. After receiving the monitoring data, the central node cleans and filters the data and stores it in the central database. The risk assessment model is constructed to perform risk assessment on the node, including: Design a risk assessment model based on the CNN model, including input layer, convolution layer, pooling layer, fully connected layer and output layer; The convolution layer consists of two layers. The first convolution layer is a 1D convolution layer using 16 filters, and the second convolution layer is a 1D convolution layer using 32 filters. The size of each filter is set to 3. Both the first and second convolution layers use the ReLU function as the activation function. The pooling layer is connected after the convolution layer, and the pooling size is set to 2 to perform maximum pooling on the convolution layer output; The fully connected layer is set after the pooling layer, and 64 neurons are set to process the output of the pooling layer; After the output layer is connected to the fully connected layer, sigmoid is used as the activation function to process the output of the last neuron in the fully connected layer and finally output it; After building the risk assessment model, use the training set to train the model, annotate the training set data simultaneously, and define the loss function as: ; in For loss, is the total number of training set data, is the actual label of the i-th training data, is the output of the risk assessment model for the i-th training data; Use the Adam optimizer to iteratively train the model, update the model parameters and calculate the loss after each iteration, and stop the iteration until the model loss does not decrease significantly to obtain the trained risk assessment model; The risk assessment model is deployed to the central database to perform risk assessment on each node behavior monitoring data collected to obtain the risk assessment result R; The dynamically setting of the evaluation threshold refers to setting the initial threshold , after obtaining the new node risk assessment result R, calculate the mean of all node risk assessment results and standard deviation , using the logistic loss function to calculate the threshold update gradient: ; ; in is the logistic loss, Set to , where +1 represents risk and -1 represents no risk. To evaluate the threshold, is the updated gradient of k; Dynamic update of evaluation threshold based on threshold update gradient: ; in is the learning rate.
2. The security control method for a financial account according to claim 1, characterized in that: The implementation of treatment measures on the node according to the risk assessment result and the assessment threshold refers to obtaining the risk assessment result R through the risk assessment model and comparing it with the dynamically updated assessment threshold k to determine the node risk: If R < k, the node is judged to be risk-free, and the central node assigns a detected mark to the node, regularly conducts risk assessment and judgment on the node, and generates records of risk assessment results and risk judgment results; If R ≥ k, the node is judged to be at risk. The central node generates an unbinding command and sends it to the risk node. The smart contract automatically executes the unbinding operation to prevent all actions of the risk node, isolates the risk node from the blockchain network, and transfers it to the security sandbox for security risk analysis and processing.
3. The security control method for a financial account according to claim 2, characterized in that: The recording and storage of all data refers to the central database synchronously generating records for storage of risk assessment results, risk assessment thresholds, risk judgment results and treatment measures and setting security access measures, and backing up all stored data in the cloud, regularly checking the integrity and consistency of stored data and backup data, generating detection records after detection, adding timestamp data and storing them.
4. A computer device comprising: Memory and processor; The memory stores a computer program, wherein the processor implements the steps of the security control method for a financial account according to any one of claims 1 to 3 when executing the computer program.
5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the security control method for a financial account described in any one of claims 1 to 3 are implemented.
