Credit payment processing method and system for transaction platform

Abstract

The invention discloses a credit payment processing method and system for a transaction platform, and relates to the technical field of credit payment, and the method comprises the steps: obtaining basic data of all tenants in the transaction platform, and calculating a dynamic weight for each tenant according to the basic data; creating an independent policy sandbox environment for each tenant according to the dynamic weight; inputting the basic data of the tenant into a preset mixed credit assessment model, and executing credit assessment of the tenant based on a policy sandbox environment; dynamically adjusting the credit line of the tenant according to the mixed evaluation result; and executing compliance verification and calculating a risk score according to the real-time credit line of the tenant and the basic data, if the risk score exceeds a preset threshold, triggering risk interception, and executing a credit payment isolation instruction. Through dynamic weight calculation, independent strategy sandbox isolation, a mixed credit evaluation model and block chain compliance verification, the problems of credit strategy coupling, data leakage, evaluation lag and compliance dependence on manpower in a multi-tenant environment are solved.

Description

Technical Field

[0001] The present invention relates to the field of credit payment technology, and in particular to a credit payment processing method and system for a transaction platform. Background Art

[0002] The credit payment processing systems currently available in the prior art have the following defects in a multi-tenant environment: Traditional systems rely on static policy configurations and are unable to dynamically adjust credit policies based on the tenants' real-time behavior, risk level, or external environment, resulting in rigid resource allocation. For example, high-risk tenants occupy too many resources, while low-risk tenants have insufficient resources. In addition, the data and policies between tenants are not physically isolated, resulting in policy conflicts or data leakage risks. Credit assessment models often rely on historical data (such as the FICO model) and lack integrated analysis of real-time behavior and external risks, resulting in delayed assessment results. As a result, credit limit adjustments are usually based on fixed rules and are unable to respond to risk changes in real time, leading to increased risk exposure. Furthermore, compliance verification relies on manual rule-making, with long update cycles and a lack of multi-party collaborative verification. Risk scoring relies on a single threshold judgment and is not combined with the tenant's real-time risk index and credit limit, resulting in a high false interception rate or missed risk detection. Therefore, there is an urgent need for a solution that can achieve more intelligent credit payment processing in a multi-tenant environment. Summary of the Invention

[0003] In view of this, the present invention proposes a credit payment processing method and system for trading platforms, which can solve the problems of rigid resource allocation, delayed evaluation, manual compliance and lack of closed loop. The present invention provides the following technical solutions:

[0004] A credit payment processing method for a transaction platform, the method comprising:

[0005] Obtain basic data of all tenants in the trading platform and calculate dynamic weight for each tenant based on the basic data;

[0006] Creating an independent policy sandbox environment for each tenant according to the dynamic weight;

[0007] Inputting the basic data of the tenant into a preset hybrid credit assessment model, and performing a credit assessment of the tenant based on the policy sandbox environment to obtain a hybrid assessment result;

[0008] dynamically adjusting the tenant's credit limit based on the hybrid assessment results;

[0009] Compliance verification and risk score calculation are performed based on the tenant's real-time credit limit and basic data. If the risk score exceeds the preset threshold, risk interception is triggered and a credit payment isolation instruction is executed.

[0010] Optionally, the method further includes:

[0011] Dynamically update the tenant's dynamic weight based on the compliance check result;

[0012] Update the tenant's policy sandbox environment based on the updated dynamic weights;

[0013] If risk interception is triggered, the risk event will be recorded in the blockchain rule base as a compliance verification rule, and the hybrid credit assessment model and strategy sandbox environment will be updated at the same time.

[0014] Optionally, obtaining basic data of all tenants in the trading platform and calculating a dynamic weight for each tenant based on the basic data includes:

[0015] The basic data of the tenant includes historical behavior characteristics, real-time risk index, and external risk data;

[0016] The historical behavior characteristics, real-time risk index and external risk data are weighted and summed to calculate the tenant's dynamic weight W i , the calculation formula is:

[0017] Among them, F ik is the kth historical behavior feature value of tenant i, N is the total number of historical behavior features, ω k is the preset feature weight, R i is the real-time risk index, η(t) is the time decay factor, defined as: η(t) = e -λt Where t is the time interval between the occurrence of risk events;

[0018] At every preset time interval, the basic data of the tenant is collected again and the dynamic weight W is recalculated. i ;

[0019] If the tenant triggers risk interception, the dynamic weight W is updated immediately i The default maximum risk weight.

[0020] Optionally, creating an independent policy sandbox environment for each tenant according to the dynamic weight includes:

[0021] According to the dynamic weight W of tenant i i , calculate the resource credit of tenant i, the resource credit includes computing resource credit and storage resource credit, and the calculation formula of the computing resource credit and storage resource credit is:

[0022] C i =C base ×min(W i , W max ), S i =Sbase ×min(W i , W max ), where C i is the computing resource credit, S i is the storage resource credit, C base and S base is the preset base credit limit, W max is the preset maximum weight value;

[0023] Create an independent sandbox container for tenant i based on containerization technology, and configure computing resource limit parameters, storage isolation and network isolation. The computing resource limit parameters include computing resource credit C i The CPU and memory credits, the storage isolation includes independent storage resource credits S i ;

[0024] Identify the dynamic weight W i When the change exceeds the preset threshold, the resource credit C is recalculated and updated. i and storage resource credit S i ;

[0025] If the dynamic weight W is identified i If the value drops below a preset threshold, the sandbox container is triggered to shrink.

[0026] Optionally, the hybrid credit assessment model includes a time series prediction sub-model and a reinforcement learning sub-model; inputting the basic data of the tenant into a preset hybrid credit assessment model and performing a credit assessment of the tenant based on the policy sandbox environment to obtain a hybrid assessment result includes:

[0027] Obtain tenants' historical behavior data, real-time behavior data, and external risk data based on their basic data;

[0028] Input the historical behavior data into the time series prediction sub-model to calculate the historical credit score S h , the calculation formula is: S h =LSTM(H(t)), where H(t) is the behavioral data of day t and LSTM is the time series prediction sub-model;

[0029] Based on real-time behavior data, the behavior score S is calculated through the reinforcement learning sub-model b , the calculation formula is: Among them, r t is the behavior reward at time t, D(t) is the time discount factor, γ rl is the behavior weight coefficient;

[0030] Calculate the external score S by integrating external risk data e, the calculation formula is: S e =α ext ·E i +β ext ·M i , where E i is the external risk index, M i is the market volatility index, α ext and β ext is the weight coefficient of external risk;

[0031] Based on historical credit score S h , behavior score S b and external rating S e Calculate comprehensive credit score S total , the calculation formula is: S total =S h +S b +S e ;

[0032] Based on the comprehensive credit score S total Outputs mixed evaluation results.

[0033] Optionally, dynamically adjusting the tenant's credit limit according to the hybrid evaluation result includes:

[0034] Based on the comprehensive credit score S total Calculate the base credit limit C base , the calculation formula is: Among them, C0 is the preset initial credit limit, S max is the maximum credit score threshold;

[0035] Based on the dynamic weight W of the tenant i and real-time risk index R i , calculate the adjustment factor A, the calculation formula is:

[0036] Where θ is the preset sensitivity threshold, η(t) is the time decay factor, and ΔS is the score change: ΔS = S total -S prev , S prev is the previous credit score, R max is the maximum risk threshold, A max is the maximum adjustment coefficient;

[0037] Based on adjustment factor A and base credit limit C base Dynamically adjust credit limit C new , the formula is: C new =C base A;

[0038] The tenant's latest credit limit Cnew Write the corresponding policy sandbox environment and trigger reloading;

[0039] Each time the credit limit is updated, its change is calculated. When the credit limit change drops below the preset compliance warning, a high-risk warning for compliance verification is triggered.

[0040] Optionally, performing compliance verification and calculating a risk score based on the tenant's real-time credit limit and basic data, triggering risk interception if the risk score exceeds a preset threshold, and executing a credit payment isolation instruction includes:

[0041] Obtain tenants' current credit limits, historical behavioral characteristics, and real-time risk indexes to build tenant data;

[0042] Obtain the latest compliance rule set from the blockchain compliance rule library and match the tenant data with the compliance rule set to mark the number of violations N v , and calculate the risk score, the formula is:

[0043] Among them, α com , β com and γ com is the weight coefficient of compliance verification, C th is the preset credit limit threshold, R i is the real-time risk index, C c is the current credit limit;

[0044] If the risk score R s If the threshold is greater than or equal to the preset interception threshold, risk interception is triggered and the tenant is marked as high-risk.

[0045] Record current risk events to the blockchain compliance rule base and update the compliance rule set;

[0046] Execute the following credit payment isolation instructions:

[0047] Freeze the tenant’s credit line;

[0048] Isolate network access to the policy sandbox environment;

[0049] Mark the policy sandbox environment as high-risk and perform resource quota reduction;

[0050] The present invention further discloses a credit payment processing system for a transaction platform, comprising:

[0051] A dynamic weight calculation module is used to obtain basic data of all tenants in the trading platform and calculate a dynamic weight for each tenant based on the basic data;

[0052] A sandbox environment creation module, configured to create an independent policy sandbox environment for each tenant according to the dynamic weight;

[0053] A hybrid evaluation module, configured to input the basic data of the tenant into a preset hybrid credit evaluation model and perform a credit evaluation of the tenant based on the policy sandbox environment to obtain a hybrid evaluation result;

[0054] a credit limit adjustment module, configured to dynamically adjust the tenant's credit limit based on the hybrid assessment result;

[0055] The compliance verification and risk interception module is used to perform compliance verification and calculate risk scores based on the tenant's real-time credit limit and basic data. If the risk score exceeds the preset threshold, risk interception is triggered and a credit payment isolation instruction is executed.

[0056] The present invention further discloses a computer-readable storage medium, wherein the storage medium stores a computer program, and the computer program implements the above method when executed by a processor.

[0057] The present invention further discloses an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the program.

[0058] According to the technical solution of the present invention, dynamic weights corresponding to tenants are calculated based on their historical behavioral characteristics, real-time risk index, and external risk data. Time decay factors are then used to reduce the long-term impact of historical risk. Dynamic weights are then assigned to drive the allocation of resource quotas for subsequent sandbox containers, enabling high-weight tenants to obtain greater computing resources while low-weight tenants have limited resources. This effectively improves resource utilization and simultaneously isolates tenants' networks and storage, preventing data leaks or policy conflicts and effectively ensuring the security of credit payments. Furthermore, by integrating a time-series prediction sub-model, a reinforcement learning sub-model, and external risk data to calculate a comprehensive credit score, the response delay to risk changes can be significantly reduced. A multi-party verified rule set is then obtained from the blockchain compliance rule library. The risk score is then calculated using a formula based on the tenant's real-time credit limit and risk index. Risk events trigger network isolation, credit freezes, and resource reductions in the sandbox container, ensuring resource utilization while improving the security of credit payments. When a tenant's credit payment process triggers risk interception, dynamic weights are recalculated, combined with sandbox resource updates and hybrid model parameter optimization to form an "event-rule-execution" closed loop. This effectively improves the system's risk tolerance. Automatic resource quota reduction ensures that high-risk tenants cannot continue to occupy resources, significantly enhancing the system's adaptability. This ensures high security, high real-time performance, adaptive optimization, and multi-party collaborative governance for the credit payment system, resolving existing issues such as rigid resource allocation, lagging assessments, manual compliance, and the lack of a closed loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] For purposes of illustration and not limitation, the present invention will now be described with reference to embodiments thereof and the accompanying drawings, in which:

[0060] Figure 1 1 is a flow chart of a credit payment processing method for a trading platform in an embodiment of the present invention;

[0061] Figure 2 Schematic diagram of the structure of a credit payment processing system for a transaction platform in an embodiment of the present invention;

[0062] Figure 3 Schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to enable those skilled in the art to better understand the solution of this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of this application.

[0064] It should be noted that, in the absence of conflict, the embodiments of the present application and the features thereof can be combined with each other. The embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0065] refer to Figure 1 This embodiment discloses a credit payment processing method for a transaction platform, which includes the following steps:

[0066] S100: Obtain the basic data of all tenants on the trading platform and calculate a dynamic weight for each tenant based on this basic data. In a multi-tenant trading platform, data fusion and real-time adjustment of multi-dimensional basic data can achieve resource isolation, differentiated risk management, and adaptive system optimization among tenants. By calculating the dynamic weights of tenants, tenant risks can be accurately identified and strategies can be dynamically adjusted, ultimately achieving the core goals of efficient resource utilization, real-time risk control, and adaptive system optimization.

[0067] Tenants' basic data includes historical behavioral characteristics, real-time risk index, and external risk data. Historical behavioral characteristics refer to tenants' historical transaction data obtained from the trading platform database, including, for example, transaction frequency (e.g., number of transactions in the past 30 days), default rate (e.g., percentage of overdue payments), average transaction amount, and user repurchase rate. The real-time risk index is a quantitative measure of the current tenant's credit payment risk, calculated from external data sources obtained in real time through an API and the tenant's current behavioral data. External risk data includes publicly available industry risk ratings, market volatility indices, and public opinion data.

[0068] Based on this, the method for calculating the tenant's dynamic weight includes the following steps:

[0069] S110: Perform weighted summation on the historical behavior characteristics, real-time risk index and external risk data to calculate the tenant's dynamic weight W i , the calculation formula is:

[0070] Among them, F ik is the k-th historical behavior feature value of tenant i. For example, if the historical behavior is transaction frequency, which is 10 times per day, then F ik=10; N is the total number of historical behavior features. In this calculation method, a specific number of historical behavior features can be selected to perform the calculation, that is, n historical behavior features are selected from N; ω k is the preset feature weight, R i is the real-time risk index, η(t) is the time decay factor, defined as: η(t) = e -λt , where t is the time interval between risk events. The dynamic weight calculation method in this implementation avoids the limitations caused by a single indicator by integrating real-time and external data, which can effectively reduce the misjudgment of credit payment behavior. For example, tenants with high transaction frequency but low default rate may be misjudged as high risk. In this solution, the weight coefficient ω is used to calculate the risk of a tenant. k Balance multi-dimensional data and improve accuracy. The time decay factor η(t) is also introduced to reduce the long-term impact of historical risk on weights. This embodiment gives an example explanation. If t is 24 hours (λ=0.05), then η(t)=e -0.05×24 ≈0.3, which means that the contribution weight of historical risk is reduced by 70%, while the contribution weight of recent risk, such as t is 1 hour, is close to 1, and the weight at this time will increase significantly. Compared with the static weight of the existing technology, this solution can quickly respond to changes in tenant behavior.

[0071] S120: Recollect tenant basic data and recalculate dynamic weight W at every preset time interval i . Because the behavioral characteristics of tenants (such as transaction frequency, default rate) and the external environment (such as market risk index) are changing dynamically. For example, a tenant may experience a surge in transaction volume due to business expansion, or an increase in default rate due to a sudden risk event. If the system relies solely on the initial weights, it will not be able to respond to these changes in a timely manner, resulting in irrational resource allocation or delayed risk assessment. At the same time, if the weights are not updated for a long time, their calculation basis will deviate seriously from the current tenant status. For example, a tenant has performed well historically but has frequently defaulted recently, and the old weights still overestimate its credit, resulting in excessive credit. Therefore, regular recalculation of weights ensures that the system quickly incorporates external risk changes, such as reducing tenants' credit limits when the market is down, avoiding systemic risks, while ensuring that the weights are always based on the latest data to reduce decision-making bias.

[0072] S130: If the tenant triggers risk interception, the dynamic weight W is updated immediately i The default maximum risk weight is 1. Risk interception is triggered when high-risk tenant behavior is detected. It can make extreme adjustments to the tenant's weight to ensure that risk events are quickly isolated and their impact on the system is limited. Its core goal is to force the system to adopt the highest level of resource restrictions and policy isolation for high-risk tenants through sudden changes in weights, preventing the spread of risk or system crashes.

[0073] S200: Creating an independent policy sandbox environment for each tenant according to the dynamic weight.

[0074] In a multi-tenant trading platform, resources need to be allocated to tenants to complete credit transaction processing. However, traditional trading systems lack a dynamic resource allocation mechanism. High-risk or high-demand tenants may occupy excessive resources, such as CPU or storage resources, resulting in insufficient resources for other tenants. For example, a tenant's sudden high-frequency trading may deplete system resources and affect the normal service of other tenants. At the same time, if tenants' credit policies (such as risk thresholds and credit limit rules) are not isolated, they may interfere with each other. An unisolated environment cannot quickly isolate high-risk tenants, which may cause risk events to spread to the entire trading system. Therefore, by constructing a policy sandbox environment based on dynamic weights calculated based on tenants' historical behavior, real-time risks, and external data, physical isolation between tenants can be achieved. That is, each sandbox container has independent storage and network access rights to prevent data leakage or unauthorized access; each sandbox's policy configuration is stored independently to avoid policy conflicts. Furthermore, changes in dynamic weights trigger real-time updates to the sandbox environment, forming a closed-loop control. The construction of a policy sandbox environment specifically includes:

[0075] S210: Based on the dynamic weight W of tenant i i , calculate the resource credit of tenant i, the resource credit includes computing resource credit and storage resource credit, and the calculation formula of the computing resource credit and storage resource credit is:

[0076] C i =C base ×min(W i , W max ), S i =S base ×min(W i , W max ), where C i is the computing resource credit, S i is the storage resource credit, C base and S base is the preset base credit limit, W max The present embodiment takes CPU processing core as an example for computing resources and storage space as an example for storage resources. base For 2 CPU cores, S base For 10GB storage, W max is 2. If tenant i’s dynamic weight W i is 1.8, then C i =2×min(1.8,2)=3.6,S i=10×min(1.8, 2)=18, then 3.6 CPU cores and 18 GB of storage are allocated to tenant i.

[0077] S220: Create an independent sandbox container for tenant i based on containerization technology, and configure computing resource limit parameters, storage isolation, and network isolation. Each tenant's sandbox container data is stored in an independent shard, which is inaccessible to other tenants. Access to the core database port is prohibited through NetworkPolicy, and only access to the pre-authorized interface is allowed. The sandbox container loads a dedicated configuration file containing the tenant's credit threshold and risk rules. The computing resource limit parameters include computing resource credit limit C i The CPU credit and memory credit of the storage isolation include independent storage resource credit S i For example, Kubernetes is used to create an independent sandbox container for tenant i, with configuration parameters including 3.6 CPU cores, 4GB of memory, and 18GB of storage. Kubernetes NetworkPolicy is used to restrict the sandbox container to access only pre-authorized API interfaces. By completely isolating storage and networks to prevent data leakage or unauthorized access, the risk of system crash can be effectively reduced. The configured sandbox container access interface restrictions can effectively prevent unauthorized operations.

[0078] S230: Identify dynamic weight W i When the change exceeds the preset threshold, the resource credit C is recalculated and updated. i and storage resource credit S i The change in dynamic weight is recorded as ΔW. For example, when ΔW ≥ 0.5, C is recalculated. i and S i , and update the container resource credit, for example, when W i When the value increases from 1.5 to 2, the CPU core quota increases from 3 to 4.

[0079] S240: If the dynamic weight W is identified i If the W drops below a preset threshold, the shrinking of the sandbox container is triggered. i When it is less than 0.5, scaling down is triggered, and the CPU core quota is reduced from 1 to 0.5, and the storage is reduced from 5GB to 3GB.

[0080] By building a strategic sandbox environment, high-weight tenants can obtain sufficient resources, while low-weight tenants have limited resources, effectively improving resource utilization. After a risk intercept occurs, the sandbox environment immediately freezes the credit limit and isolates network access, effectively preventing the spread of risks. iDirectly driving the calculation of resource quotas can ensure that high-weight tenants receive higher resource priority. Sandbox environment isolation ensures that the credit assessment model only accesses tenant-exclusive data to avoid data contamination. After risk interception, sandbox network isolation takes effect immediately, which can quickly prevent high-risk transactions.

[0081] S300: Inputting the basic data of the tenant into a preset hybrid credit evaluation model, and performing credit evaluation of the tenant based on the policy sandbox environment to obtain a hybrid evaluation result.

[0082] The hybrid credit assessment model includes a time series prediction sub-model, a reinforcement learning sub-model, and an external risk fusion module. The time series prediction sub-model uses an LSTM neural network to analyze historical behavioral data and capture long-term transaction trends.

[0083] S310: The input data is a historical transaction sequence, that is, the historical behavior data in the basic data, denoted as H(t) = [H1, H2, ..., H 30 ], where H 30 The transaction data on the 30th day is shown here for example, and the specific number of days can be modified adaptively. The historical behavior data is input into the time series prediction sub-model to calculate the historical credit score S h , the calculation formula is: S h =LSTM(H(t)), where H(t) is the behavioral data of day t and LSTM is the time series prediction sub-model;

[0084] S320: Based on real-time behavior data, the behavior score S is calculated through the reinforcement learning sub-model. b , the calculation formula is: Among them, r t is the reward for the behavior at time t, for example, when the payment is successful r t = +1, when a refund occurs, r t =-0.5, when high-frequency abnormal transactions occur, r t =-2. D(t) is the time discount factor, γ rl is the behavior weight coefficient, and T is the behavior data in the past counting time. For example, a tenant has 10 successful payments and 2 refunds within 24 hours, then T = 24, let γ r1 = 0.3, define D(t) = 0.9 t , then the tenant's behavior score S b =0.3·(10×0.9 1 +(-0.5)×0.9 24 )≈2.7.

[0085] S330: Calculate external score S by integrating external risk data e , the calculation formula is: Se =α ext ·E i +β ext ·M i , where E i is the external risk index, M i is the market volatility index, α ext and β ext is the weight coefficient of external risk. Let α ext =0.6, β ext =0.4, external risk index E i For low risk, the value is 70; the market volatility index M i For low market volatility, the value is 30, then S e =0.6·70+0.4·30=42+12=54.

[0086] S340: Based on historical credit score S h , behavior score S b and external rating S e Calculate comprehensive credit score S total , where the historical credit score S h is obtained from the outside. The calculation formula is: S total =S h +S b +S e ; Based on comprehensive credit score S total Output hybrid assessment results, which are based on the comprehensive credit score S total Credit rating classification.

[0087] S400: Dynamically adjust the tenant's credit limit based on the hybrid evaluation result. Specifically including:

[0088] S410: Based on the comprehensive credit score S total Calculate the base credit limit C base , the calculation formula is: Among them, C0 is the preset initial credit limit, S max is the maximum credit score threshold;

[0089] S420: Based on the dynamic weight W of the tenant i and real-time risk index R i , calculate the adjustment factor A, the calculation formula is:

[0090] Where θ is the preset sensitivity threshold, η(t) is the time decay factor, and ΔS is the score change: ΔS = S total -S prev , S prev is the previous credit score, Rmax is the maximum risk threshold, A max is the maximum adjustment coefficient;

[0091] S430: Based on adjustment factor A and base credit limit C base Dynamically adjust credit limit C new , the formula is: C new =C base A;

[0092] S440: The tenant's latest credit limit C new Write the corresponding policy sandbox environment and trigger reloading;

[0093] S450: Each time the credit limit is updated, its change is calculated. When the credit limit change drops below the preset compliance warning, a high-risk warning for compliance verification is triggered.

[0094] By introducing the dynamic weight W i , we can determine the upper limit of the adjustment factor A and construct the calculation formula of the adjustment factor. When the initial adjustment space of high-weight tenants is large, we can suppress it through the risk index. After risk interception, the weight W i Update to the highest value, can achieve further compression of credit limit. Compared with the traditional single factor driven calculation method, can build a more comprehensive decision. At the same time, the comprehensive score S total Direct drive benchmark amount C base , which ensures a strong correlation between score improvement and credit limit adjustment. new After writing the sandbox configuration, resource quotas (CPU / storage) can be automatically adjusted. This means that by combining dynamic weights, risk indices, and score changes, we can avoid misjudgments based on a single factor and effectively improve the sensitivity of adjustment factor A. Furthermore, hot updates of sandbox containers ensure that new quotas take effect within seconds, significantly reducing transaction latency. Furthermore, by compressing the penalty items for high-risk tenants' quotas, we can free up idle resources and improve computing resource utilization.

[0095] S500: Compliance verification and risk score calculation are performed based on the tenant's real-time credit limit and basic data. If the risk score exceeds the preset threshold, risk interception is triggered and credit payment isolation instructions are executed. Based on the results of the compliance verification, the tenant's dynamic weight is dynamically updated; based on the updated dynamic weight, the tenant's policy sandbox environment is updated; if risk interception is triggered, the risk event is recorded in the blockchain rule library as a compliance verification rule, and the hybrid credit assessment model and policy sandbox environment are updated. Specifically including:

[0096] S510: Obtain the tenant's current credit limit, historical behavior characteristics and real-time risk index to construct tenant data. The current credit limit is recorded as C calculated in step S430. new .

[0097] S520: Obtain the latest compliance rule set from the blockchain compliance rule library and match the tenant data with the compliance rule set to mark the number of violations N v , and calculate the risk score, the formula is:

[0098] Among them, α com , β com and γ com is the weight coefficient of compliance verification, C th is the preset credit limit threshold, R i is the real-time risk index, C c The current credit limit. The rule set is stored in the consortium chain and verified by regulators, trading platforms, and tenants. It consists of multiple conditional statements, and the number of violations can be determined through the compliance rule set.

[0099] S530: If the risk score R s If the threshold is greater than or equal to the preset interception threshold, risk interception is triggered and the tenant is marked as high-risk.

[0100] S540: Record the current risk event to the blockchain compliance rule base and update the compliance rule set;

[0101] S550: Execute the following credit payment isolation instruction:

[0102] Freeze the tenant's credit line, c Set to 0;

[0103] Isolate network access in the policy sandbox environment. Use Kubernetes NetworkPolicy to restrict sandbox containers to only access compliance verification interfaces and prohibit payment ports.

[0104] Marks the policy sandbox environment as high-risk and performs resource quota reduction.

[0105] In summary, this implementation solves the problems of rigid multi-tenant resource allocation, delayed risk assessment, manual compliance rule updates, and insufficient risk isolation in traditional credit payment systems through multi-dimensional dynamic weight calculation, independent policy sandbox environment creation, hybrid credit assessment model, and blockchain-driven compliance verification technology: First, by introducing a weighted calculation formula that combines historical behavior characteristics, real-time risk index, and external risk data, the dynamic weight not only reflects the tenant's risk level, but also reduces the long-term impact of historical risks through an attenuation mechanism, thereby achieving on-demand allocation of resource quotas and avoiding high-risk tenants occupying too many resources or low-risk tenants from having insufficient resources; second, the sandbox environment created based on dynamic weights uses computing / storage resource quotas and network isolation parameters, such as Kubernetes NetworkPolicy restricts access interfaces to ensure complete isolation of data, policies, and computing resources between tenants, preventing policy conflicts or unauthorized access. At the same time, the dynamic update mechanism and scaling-down strategy of resource quotas triggered by weight changes effectively optimize resource utilization. The hybrid credit assessment model integrates a time series prediction sub-model, a reinforcement learning sub-model, and external risk data, and calculates a comprehensive credit score through weighted multi-modal data, solving the problems of assessment lag and insufficient accuracy caused by the traditional model's reliance on a single data source. The adjustment factor introduced when dynamically adjusting the credit limit balances risk sensitivity and credit stability, avoiding system fluctuations caused by sudden changes in the credit limit. The blockchain-based compliance verification rule library and real-time risk scoring system realize multi-party consensus updates of compliance rules and interception of risk events, forming a closed-loop control through freezing quotas, network isolation, and resource reduction instructions. Ultimately, the overall technical solution significantly improves system security, resource efficiency, real-time risk response, and adaptability, forming a high-precision, highly isolated, and self-optimizing technical closed loop for credit payment processing in a multi-tenant environment.

[0106] refer to Figure 2 This embodiment further discloses a credit payment processing system for a transaction platform, including:

[0107] The dynamic weight calculation module 21 is used to obtain the basic data of all tenants in the trading platform and calculate the dynamic weight for each tenant based on the basic data, including: the basic data of the tenant includes historical behavior characteristics, real-time risk index, and external risk data; the historical behavior characteristics, real-time risk index and external risk data are weighted and summed to calculate the tenant's dynamic weight W i , the calculation formula is: Among them, F ik is the kth historical behavior feature value of tenant i, N is the total number of historical behavior features, ω k is the preset feature weight, R i is the real-time risk index, η(t) is the time decay factor, defined as: η(t) = e -λtWhere t is the time interval between risk events; at every preset time interval, the basic data of the tenant is recollected and the dynamic weight W is recalculated. i ; If the tenant triggers risk interception, the dynamic weight W is updated immediately i The default maximum risk weight.

[0108] The sandbox environment creation module 22 is used to create an independent policy sandbox environment for each tenant according to the dynamic weight, including: i , calculate the resource credit of tenant i, the resource credit includes computing resource credit and storage resource credit, the calculation formula of computing resource credit and storage resource credit is: C i =C base ×min(W i , W max ), S i =S base ×min(W i , W max ), where C i is the computing resource credit, S i is the storage resource credit, C base and S base is the preset base credit limit, W max is the preset maximum weight value; based on containerization technology, an independent sandbox container is created for tenant i, and computing resource limitation parameters, storage isolation and network isolation are configured. The computing resource limitation parameters include computing resource credit C i The CPU and memory credits, the storage isolation includes independent storage resource credits S i ; Identify dynamic weight W i When the change exceeds the preset threshold, the resource credit C is recalculated and updated. i and storage resource credit S i ; If the dynamic weight W is identified i If the value drops below a preset threshold, the sandbox container is triggered to shrink.

[0109] The hybrid assessment module 23 is used to input the basic data of the tenant into a preset hybrid credit assessment model and perform the credit assessment of the tenant based on the policy sandbox environment to obtain a hybrid assessment result, including: obtaining the tenant's historical behavior data, real-time behavior data and external risk data based on the tenant's basic data; inputting the historical behavior data into the time series prediction sub-model to calculate the historical credit score S h , the calculation formula is: S h=LSTM(H(t)), where H(t) is the behavioral data of day t, and LSTM is the time series prediction sub-model; based on the real-time behavioral data, the behavior score S is calculated by the reinforcement learning sub-model. b , the calculation formula is: Among them, r t is the behavior reward at time t, D(t) is the time discount factor, γ rl is the behavior weight coefficient; the external score S is calculated by integrating external risk data e , the calculation formula is: S e =α ext ·E i +β ext ·M i , where E i is the external risk index, M i is the market volatility index, α ext and β ext is the weight coefficient of external risk; based on the historical credit score S h , behavior score S b and external rating S e Calculate comprehensive credit score S total , the calculation formula is: S total =S h +S b +S e ; Based on comprehensive credit score S total Outputs mixed evaluation results.

[0110] The credit limit adjustment module 24 is used to dynamically adjust the tenant's credit limit based on the mixed evaluation result, including: total Calculate the base credit limit C base , the calculation formula is: Among them, C0 is the preset initial credit limit, S max is the highest credit score threshold; based on the tenant's dynamic weight W i and real-time risk index R i , calculate the adjustment factor A, the calculation formula is: Where θ is the preset sensitivity threshold, η(t) is the time decay factor, and ΔS is the score change: ΔS = S total -S prev , S prev is the previous credit score, R max is the maximum risk threshold, A max is the maximum adjustment factor; based on the adjustment factor A and the base credit limit C base Dynamically adjust credit limit C new , the formula is: C new =C baseA; Set the tenant's latest credit limit C new Write the corresponding policy sandbox environment and trigger a reload; each time the credit limit is updated, calculate its change. When the credit limit change drops below the preset compliance warning, a high-risk warning for compliance verification is triggered.

[0111] The compliance verification and risk interception module 25 is used to perform compliance verification and calculate risk scores based on the tenant's real-time credit limit and basic data. If the risk score exceeds the preset threshold, risk interception is triggered and credit payment isolation instructions are executed, including obtaining the tenant's current credit limit, historical behavior characteristics and real-time risk index to construct tenant data; obtaining the latest compliance rule set from the blockchain compliance rule library, and matching the tenant data with the compliance rule set to mark the number of violations N v , and calculate the risk score, the formula is: Among them, α com , β com and γ com is the weight coefficient of compliance verification, C th is the preset credit limit threshold, R i is the real-time risk index, C c is the current credit limit; if the risk score R s If the value is greater than or equal to the preset interception threshold, risk interception is triggered and the tenant is marked as high-risk; the current risk event is recorded in the blockchain compliance rule library and the compliance rule set is updated; the following credit payment isolation instructions are executed: freeze the tenant's credit limit; isolate the network access of the policy sandbox environment; mark the policy sandbox environment as high-risk and execute resource quota reduction.

[0112] Figure 3 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as Figure 3 As shown, the electronic device 50 includes: a processor 501 (processor), a memory 502 (memory) and a bus 503;

[0113] The processor 501 and the memory 502 communicate with each other via the bus 503 ; the processor 501 is used to call program instructions in the memory 502 to execute the methods provided by the above-mentioned method implementation methods.

[0114] This embodiment provides a non-transitory computer-readable storage medium, which stores computer instructions. The computer instructions enable a computer to execute the methods provided by the above-mentioned method embodiments.

[0115] Those skilled in the art will understand that all or part of the steps for implementing the above-mentioned method implementation method can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above-mentioned method implementation method; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk, etc. Various storage media that can store program codes.

[0116] The device embodiments described above are merely illustrative. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of these modules may be selected based on actual needs to achieve the objectives of this embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0117] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, or of course, by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the existing technology can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or certain parts of the embodiment.

[0118] The above specific embodiments do not limit the scope of protection of the present invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may occur depending on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A credit payment processing method for a trading platform, characterized in that: The method comprises: Obtain basic data of all tenants in the trading platform and calculate dynamic weights for each tenant based on the basic data; Creating an independent policy sandbox environment for each tenant according to the dynamic weight; Inputting the basic data of the tenant into a preset hybrid credit assessment model, and performing a credit assessment of the tenant based on the policy sandbox environment to obtain a hybrid assessment result; dynamically adjusting the tenant's credit limit based on the hybrid assessment results; Compliance verification and risk score calculation are performed based on the tenant's real-time credit limit and basic data. If the risk score exceeds the preset threshold, risk interception is triggered and a credit payment isolation instruction is executed.

2. The credit payment processing method for a trading platform according to claim 1, characterized in that: The method further comprises: Dynamically update the tenant's dynamic weight based on the compliance check result; Update the tenant's policy sandbox environment based on the updated dynamic weights; If risk interception is triggered, the risk event will be recorded in the blockchain rule base as a compliance verification rule, and the hybrid credit assessment model and strategy sandbox environment will be updated at the same time.

3. The credit payment processing method for a trading platform according to claim 1, characterized in that: The obtaining of basic data of all tenants in the trading platform and calculating a dynamic weight for each tenant based on the basic data includes: The basic data of the tenant includes historical behavior characteristics, real-time risk index, and external risk data; The historical behavior characteristics, real-time risk index and external risk data are weighted and summed to calculate the tenant's dynamic weight W i , the calculation formula is: Among them, F ik is the kth historical behavior feature value of tenant i, N is the total number of historical behavior features, ω k is the preset feature weight, R i is the real-time risk index, η(t) is the time decay factor, defined as: η(t) = e -λt , where t is the time interval between risk events; At every preset time interval, the basic data of the tenant is collected again and the dynamic weight W is recalculated. i ; If the tenant triggers risk interception, the dynamic weight W is updated immediately i The default maximum risk weight.

4. The credit payment processing method for a trading platform according to claim 3, characterized in that: The creating an independent policy sandbox environment for each tenant according to the dynamic weight includes: According to the dynamic weight W of tenant i i , calculate the resource credit of tenant i, the resource credit includes computing resource credit and storage resource credit, and the calculation formula of the computing resource credit and storage resource credit is: C i =C base ×min(W i , W max ), S i =S base ×min(W i , W max ), where C i is the computing resource credit, S i is the storage resource credit, C base and S base is the preset base credit limit, W max is the preset maximum weight value; Create an independent sandbox container for tenant i based on containerization technology, and configure computing resource limit parameters, storage isolation and network isolation. The computing resource limit parameters include computing resource credit C i The CPU and memory credits, the storage isolation includes independent storage resource credits S i ; Identify the dynamic weight W i When the change exceeds the preset threshold, the resource credit C is recalculated and updated. i and storage resource credit S i ; If the dynamic weight W is identified i If the value drops below a preset threshold, the sandbox container is triggered to shrink.

5. The credit payment processing method for a trading platform according to claim 3, characterized in that: The hybrid credit assessment model includes a time series prediction sub-model and a reinforcement learning sub-model; inputting the basic data of the tenant into the preset hybrid credit assessment model and performing the credit assessment of the tenant based on the policy sandbox environment to obtain a hybrid assessment result includes: Obtain tenants' historical behavior data, real-time behavior data, and external risk data based on their basic data; Input the historical behavior data into the time series prediction sub-model to calculate the historical credit score S h , the calculation formula is: S h =LSTM(H(t)), where H(t) is the behavioral data of day t and LSTM is the time series prediction sub-model; Based on real-time behavior data, the behavior score S is calculated through the reinforcement learning sub-model b , the calculation formula is: Among them, r t is the behavior reward at time t, D(t) is the time discount factor, γ rl is the behavior weight coefficient; Calculate the external score S by integrating external risk data e , the calculation formula is: S e =α ext ·E i +β ext ·M i , where E i is the external risk index, M i is the market volatility index, α ext and β ext is the weight coefficient of external risk; Based on historical credit score S h , behavior score S b and external rating S e Calculate comprehensive credit score S total , the calculation formula is: S total =S h +S b +S e ; Based on the comprehensive credit score S total Outputs mixed evaluation results.

6. The credit payment processing method for a transaction platform according to claim 5, characterized in that: The dynamically adjusting the tenant's credit limit according to the hybrid evaluation result includes: Based on the comprehensive credit score S total Calculate the base credit limit C base , the calculation formula is: Among them, C0 is the preset initial credit limit, S max is the maximum credit score threshold; Based on the dynamic weight W of the tenant i and real-time risk index R i , calculate the adjustment factor A, the calculation formula is: Where θ is the preset sensitivity threshold, η(t) is the time decay factor, and ΔS is the score change: ΔS = S total -S prev , S prev is the previous credit score, R max is the maximum risk threshold, A max is the maximum adjustment coefficient; Based on adjustment factor A and base credit limit C base Dynamically adjust credit limit C new , the formula is: C new =C base A; The tenant's latest credit limit C new Write the corresponding policy sandbox environment and trigger reloading; Each time the credit limit is updated, its change is calculated. When the credit limit change drops below the preset compliance warning, a high-risk warning for compliance verification is triggered.

7. The credit payment processing method for a trading platform according to claim 3, characterized in that: The compliance check and risk score calculation based on the tenant's real-time credit limit and basic data are performed, and if the risk score exceeds a preset threshold, risk interception is triggered and a credit payment isolation instruction is executed, including: Obtain tenants' current credit limits, historical behavioral characteristics, and real-time risk indexes to build tenant data; Obtain the latest compliance rule set from the blockchain compliance rule library and match the tenant data with the compliance rule set to mark the number of violations N v , and calculate the risk score, the formula is: Among them, α com , β com and γ com is the weight coefficient of compliance verification, C th is the preset credit limit threshold, R i is the real-time risk index, C c is the current credit limit; If the risk score R s If the threshold is greater than or equal to the preset interception threshold, risk interception is triggered and the tenant is marked as high-risk. Record current risk events to the blockchain compliance rule base and update the compliance rule set; Execute the following credit payment isolation instructions: Freeze the tenant’s credit line; Isolate network access to a policy sandbox environment; Marks the policy sandbox environment as high-risk and performs resource quota reduction.

8. A credit payment processing system for a trading platform, characterized in that: include: A dynamic weight calculation module is used to obtain basic data of all tenants in the trading platform and calculate a dynamic weight for each tenant based on the basic data; A sandbox environment creation module, configured to create an independent policy sandbox environment for each tenant according to the dynamic weight; A hybrid evaluation module, configured to input the basic data of the tenant into a preset hybrid credit evaluation model and perform a credit evaluation of the tenant based on the policy sandbox environment to obtain a hybrid evaluation result; a credit limit adjustment module, configured to dynamically adjust the tenant's credit limit based on the hybrid assessment result; The compliance verification and risk interception module is used to perform compliance verification and calculate risk scores based on the tenant's real-time credit limit and basic data. If the risk score exceeds the preset threshold, risk interception is triggered and a credit payment isolation instruction is executed.

9. A computer-readable storage medium, characterized in that The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

Images