Post-payment guarantee method and payment system with capital bottom and trust as targets

Abstract

The invention discloses a post-payment guarantee method and a payment system with funds and trust as targets, and aims to provide real-time credit assessment and risk control for users in a post-payment financial system. The method comprises the following steps: firstly, constructing a post-payment financial system, collecting multi-source historical transaction data of a user, and constructing a user portrait through data analysis; thirdly, performing preliminary credit evaluation according to the user portrait, determining post-payment permission and formulating a fund bottom-out strategy; then, monitoring user transaction data in real time by using a deep learning algorithm, predicting a default probability, and dynamically adjusting fund bottom-out strategy parameters; and finally, optimizing the fund bottom-out strategy based on the default risk coverage degree to form a dynamically adjusted fund bottom-out optimization strategy. The method effectively improves the transaction security, reduces the default risk, improves the user trust, is suitable for a plurality of scenes such as finance and e-commerce, and has a wide application prospect.

Description

Technical Field

[0001] The present invention relates to the field of financial technology, and in particular to a post-payment guarantee method and payment system with fund guarantee and trust as the goals. Background Art

[0002] With the rapid development of Internet finance, the post-paid financial model has gradually attracted market attention and has become an emerging transaction payment method. Compared with the traditional prepaid model, the post-paid model can better optimize the user experience and improve transaction efficiency, especially in the e-commerce platform and consumer finance fields. However, although the post-paid model provides consumers with a convenient payment method, it also exposes some serious problems, especially in terms of fund security, merchant rights protection and performance risk control.

[0003] The traditional post-paid financial model has the following major problems: First, the financial rights and interests of merchants are not effectively protected, and consumers may be at risk of malicious default or delayed payment, causing financial losses to merchants. Second, due to the lack of accurate assessment and dynamic supervision of consumer credit, it is difficult for financial platforms to effectively identify high-risk users, resulting in a high frequency of default events. Third, the existing post-paid system mostly relies on a static credit scoring mechanism, fails to effectively combine users' real-time transaction behavior for dynamic assessment, and cannot adjust the fund protection strategy in a timely manner, resulting in insufficient fund security.

[0004] To address the above issues, existing solutions mostly focus on providing prepaid security, credit guarantees, or using a single third-party payment tool. However, these methods often fail to fundamentally solve the problems of fund security and performance risk, and lack flexibility and adaptability, and cannot effectively respond to the ever-changing trading environment.

[0005] Therefore, how to build a post-paid financial security method and payment system with both fund guarantee and trust functions, which can not only protect the rights and interests of merchants, but also dynamically adjust the degree of fund risk coverage to cope with the ever-changing user credit and transaction behavior, has become a technical problem that the industry needs to solve urgently. To this end, the present invention proposes a post-paid security method and payment system with fund guarantee and trust as the goal, which provides a safer, more flexible and efficient solution for the post-paid financial system through multi-dimensional credit evaluation and dynamic risk control strategy. Summary of the invention

[0006] In order to solve at least one of the above technical problems, the present invention proposes a post-payment guarantee method and payment system with the goal of fund guarantee and trust.

[0007] The first aspect of the present invention provides a post-payment guarantee method with the goal of fund guarantee and trust, comprising: Building a post-paid financial system, acquiring multi-source historical transaction-related data of users of the post-paid financial system, and building a user profile based on the multi-source historical transaction-related data; Performing an initial credit rating assessment on a user of the post-paid financial system according to the user portrait, determining the user's post-paid authority according to the user's initial credit rating, and constructing a fund guarantee strategy for users with post-paid authority; Acquire user transaction-related data of the postpaid financial system in real time, analyze the user transaction-related data based on a deep learning algorithm, predict the user's default probability, and adjust the capital guarantee strategy parameters according to the default probability; The default risk coverage degree of the fund guarantee strategy after parameter adjustment is obtained, and the parameters of the fund guarantee strategy are adjusted twice according to the default risk coverage degree to obtain the fund guarantee optimization strategy.

[0008] In this solution, the post-paid financial system is constructed, and multi-source historical transaction-related data of users of the post-paid financial system is obtained, and user portraits are constructed based on the multi-source historical transaction-related data, specifically: Building a post-paid financial system, obtaining multi-source historical transaction-related data of users of the post-paid financial system, wherein the multi-source historical transaction-related data includes basic user information, third-party payment platform credit score, historical consumption record data, and bank account information; Standardizing the multi-source historical transaction related data, calculating the covariance matrix of each data item of the standardized multi-source historical transaction related data, and calculating the eigenvalues ​​and eigenvectors of the covariance matrix; Arrange the eigenvalues ​​in descending order, select the eigenvectors corresponding to the first n eigenvalues ​​as principal components, construct a transformation matrix with the eigenvectors of the principal components, and perform a dimensionality reduction operation on the multi-source historical transaction-related data according to the transformation matrix to obtain dimensionality-reduced multi-source historical transaction-related data; Introducing the K-means clustering algorithm, analyzing the dimension-reduced multi-source historical transaction-related data based on the elbow rule, determining the clustering number k value of the K-means clustering algorithm, and randomly selecting k data points of each data item in the dimension-reduced multi-source historical transaction-related data as initial clustering centers; Calculate the Euclidean distance of each data point in each data item to the initial cluster center of the corresponding data item, assign each data point to the cluster center with the closest Euclidean distance to obtain the initial cluster, calculate the mean of the initial cluster, use the mean as the new cluster center, calculate the Euclidean distance of the data point to the new cluster center and assign the data point until the cluster center no longer changes, and obtain the clustering result of the dimensionality reduction multi-source historical transaction related data; Clustering features of each clustering result are extracted, and a user profile of the post-paid financial system is constructed based on the clustering features.

[0009] In this solution, the initial credit rating of the user of the post-paid financial system is evaluated according to the user portrait, the user's post-paid authority is determined according to the user's initial credit rating, and a fund guarantee strategy for the user with post-paid authority is constructed, specifically as follows: Performing an initial credit rating assessment on users of the post-paid financial system according to the user portrait, granting post-paid rights to users whose initial credit rating is greater than a preset level according to the initial credit rating, and obtaining post-paid rights information for each user; According to the post-payment authority information, if the post-payment authority user uses the post-payment function of the post-payment financial system during the purchase of goods, obtaining the transaction amount information, transaction subject credit evaluation information, and transaction commodity category information of the post-payment authority user; The security requirement level of the user's commodity purchase is evaluated based on the transaction amount information, the credit assessment information of the transaction subject, and the transaction commodity category information. If the security requirement level is lower than the preset requirement level, the post-payment agreed time node of the current commodity purchase order of the post-payment permission user is obtained. If the user still fails to complete the payment of the current commodity purchase order within the agreed time node, the user is marked as an overdue user, and the post-payment financial system provides a fund guarantee for the overdue user and advances funds to the target merchant; If the overdue user needs to purchase the product again, the purchase transaction operation will be restricted for the overdue user until the overdue user completes the payment process of the current product purchase order, and the paid funds will be transferred to the trust insurance account as a special fund reserve for accident recovery; If the security requirement level is not lower than the preset requirement level, the user freezes the preset amount of funds in his or her own bank account or digital RMB account or corporate public account through pre-authorization before the transaction, and authorizes the system to pay the designated merchant after the electronic contract is confirmed. When the user needs to purchase the product again, the system will identify the user's current product purchase order payment status, determine the next order purchase authority based on the payment status, and transfer the funds paid for the current product purchase order to the trust insurance account as a special fund reserve for claim recovery, thereby obtaining a fund guarantee strategy for users with post-payment permissions.

[0010] In this solution, the user transaction-related data of the post-paid financial system is obtained in real time, and the user transaction-related data is analyzed based on a deep learning algorithm to predict the user's default probability, and the capital guarantee strategy parameters are adjusted according to the default probability, specifically: Acquire transaction-related data of each user in the postpaid financial system in real time, wherein the user transaction-related data includes user information, transaction object information, transaction relationship information between the user and the transaction object, and historical default data; The user and transaction object are defined as nodes of the graph structure respectively, the user information and transaction object information are used as attributes of the nodes, and the transaction relationship between the user and the transaction object is defined as the edge of the graph structure. The graph structure of user transactions is constructed according to the nodes, attributes, and edges to obtain a graph structure dataset; Dividing the graph structure data set into a training set and a test set according to a preset ratio, building a user default prediction model based on the GCN graph neural network, and initializing the parameters of the user default prediction model, including a weight matrix and a bias term; Determine the dimensions of the input layer and output layer, and the dimensions and number of layers of the intermediate hidden layer of the user default prediction model according to the characteristic dimensions of the graph structure; Importing the training set into the user default prediction model for training, aggregating the features of the user default prediction model based on forward propagation, and calculating the error between the predicted default probability and the actual default probability of the user default prediction model based on a binary cross entropy loss function; Analyzing the error through back propagation, and updating the parameters of the user default prediction model, wherein the parameters include a weight matrix, a bias term, a learning rate, a hidden layer dimension and number of layers, and an activation function parameter; Importing the test set into the user default prediction model to perform prediction performance evaluation, and adjusting the hyperparameters of the user default prediction model according to the prediction performance evaluation result; Acquire real-time transaction-related data of users of the postpaid financial system within a preset time period and import it into the user default prediction model to predict the user default probability and obtain a default probability prediction result; The fund guarantee strategy parameters are adjusted according to the default probability prediction result, and the fund guarantee strategy parameters include fund advance payment ratio, payment period, and guarantee threshold.

[0011] In this solution, the default risk coverage of the fund guarantee strategy after the parameter adjustment is obtained, and the fund guarantee strategy is adjusted for the second time according to the default risk coverage to obtain the fund guarantee optimization strategy, which is specifically: Introduce the Monte Carlo simulation algorithm to obtain the user transaction data of the fund guarantee strategy after parameter adjustment, and set the simulation scenario of the Monte Carlo simulation algorithm according to the user transaction data, user portrait, and current fund guarantee strategy parameters. The simulation scenario includes the user's transaction amount, transaction frequency, transaction type, user's transaction level distribution, and user's default probability distribution; The simulation scenario is simulated based on the Monte Carlo simulation algorithm, a preset number of user transactions and default events are randomly generated according to the simulation scenario, and whether the fund guarantee condition is triggered during the simulation of the user transaction and default time is determined according to the current fund guarantee strategy parameters, and the proportion of the default transaction amount guaranteed in each simulation process to the total default transaction amount is calculated; The ability of the fund guarantee strategy to make up for the funding gap is evaluated based on the ratio of the default transaction amount to the total default transaction amount, and the degree of default risk coverage of the fund guarantee strategy is determined based on the compensation ability; If the default risk coverage is less than the preset value, a secondary parameter adjustment is performed on the fund guarantee strategy based on the particle swarm optimization algorithm to obtain a fund guarantee optimization strategy.

[0012] In this solution, the particle swarm optimization algorithm is used to adjust the parameters of the fund guarantee strategy twice to obtain the fund guarantee optimization strategy, which is specifically: Obtain the reserve level of the special fund for claim recovery of the trust insurance account of the post-paid financial system, determine the default risk tolerance of the post-paid financial system based on the reserve level, and determine the default risk coverage threshold of the fund guarantee strategy based on the default risk tolerance; A particle swarm optimization algorithm is introduced to convert the parameters of the fund guarantee strategy into the particle position vector of the particle swarm optimization algorithm, and the default risk coverage threshold is used as the optimization target of the particle swarm optimization algorithm; A particle swarm is randomly generated, a velocity vector is initialized for each particle in the particle swarm, the default risk coverage of each parameter combination is calculated according to the parameter combination of the fund hedging strategy represented by the particle position vector, and the initial fitness of each particle position vector is determined according to the default risk coverage of each parameter combination; The highest initial fitness is taken as the global optimal fitness, the position vector of the particle with the highest fitness is taken as the global optimal position, the fitness and the global optimal position are analyzed based on the initialization velocity vector, the global optimal position is updated, and the default risk coverage of the parameter combination represented by the global optimal position after the update operation is recalculated until the default risk coverage is greater than the default risk coverage threshold, and the particle at the global optimal position is output; Identify the parameter combination of the particle in the global optimal position, perform secondary parameter adjustment on the fund guarantee strategy according to the parameter combination, and obtain the fund guarantee optimization strategy. The second aspect of the present invention also provides a payment system with fund guarantee and trust as the goal, the system includes: a memory, a processor, the memory includes a post-payment guarantee method program with fund guarantee and trust as the goal, and when the post-payment guarantee method program with fund guarantee and trust as the goal is executed by the processor, the following steps are implemented: Building a post-paid financial system, acquiring multi-source historical transaction-related data of users of the post-paid financial system, and building a user profile based on the multi-source historical transaction-related data; Performing an initial credit rating assessment on a user of the post-paid financial system according to the user portrait, determining the user's post-paid authority according to the user's initial credit rating, and constructing a fund guarantee strategy for users with post-paid authority; Acquire user transaction-related data of the postpaid financial system in real time, analyze the user transaction-related data based on a deep learning algorithm, predict the user's default probability, and adjust the capital guarantee strategy parameters according to the default probability; The default risk coverage degree of the fund guarantee strategy after parameter adjustment is obtained, and the parameters of the fund guarantee strategy are adjusted twice according to the default risk coverage degree to obtain the fund guarantee optimization strategy.

[0013] The present invention discloses a post-paid security method and payment system with the goal of fund guarantee and trust, aiming to provide real-time credit assessment and risk control for users in the post-paid financial system. First, a post-paid financial system is constructed and multi-source historical transaction data of users is collected, and user portraits are constructed through data analysis; then, a preliminary credit assessment is performed based on the user portrait, the post-paid authority is determined, and a fund guarantee strategy is formulated; then, a deep learning algorithm is used to monitor user transaction data in real time, predict the probability of default, and dynamically adjust the parameters of the fund guarantee strategy; finally, the fund guarantee strategy is optimized based on the degree of default risk coverage to form a dynamically adjusted fund guarantee optimization strategy. This method effectively improves transaction security, reduces default risk, and enhances user trust. It is suitable for multiple scenarios such as finance and e-commerce, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flow chart of a post-payment guarantee method with fund guarantee and trust as the goal of the present invention is shown; Figure 2 A flow chart showing a fund guarantee strategy for post-payment authorized users constructed in the present invention; Figure 3 The flowchart of the present invention for obtaining the fund guarantee optimization strategy is shown; Figure 4A block diagram of a payment system with fund guarantee and trust as the goal of the present invention is shown. DETAILED DESCRIPTION

[0015] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0016] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0017] Figure 1 The flowchart of the post-payment guarantee method of the present invention with the goal of fund guarantee and trust is shown.

[0018] like Figure 1 As shown, the first aspect of the present invention provides a post-payment guarantee method with the goal of fund guarantee and trust, comprising: S102, constructing a post-paid financial system, obtaining multi-source historical transaction-related data of users of the post-paid financial system, and constructing a user profile based on the multi-source historical transaction-related data; S104, performing an initial credit rating assessment on the user of the post-payment financial system according to the user portrait, determining the user's post-payment authority according to the user's initial credit rating, and constructing a fund guarantee strategy for the user with post-payment authority; S106, acquiring user transaction-related data of the postpaid financial system in real time, analyzing the user transaction-related data based on a deep learning algorithm, predicting the user's default probability, and adjusting the capital guarantee strategy parameters according to the default probability; S108, obtaining the default risk coverage of the fund guarantee strategy after parameter adjustment, and performing secondary parameter adjustment on the fund guarantee strategy according to the default risk coverage to obtain a fund guarantee optimization strategy.

[0019] It should be noted that by building a post-paid financial system and obtaining multi-source historical transaction-related data (including basic user information, credit scores, consumption records, etc.), we can fully understand the user's credit status and historical behavior, and then build an accurate user portrait based on data analysis; by analyzing the user portrait, we can conduct an initial credit rating assessment and reasonably allocate post-paid permissions based on the user's credit status. For example, users with better credit are granted higher amounts of post-paid permissions, while users with poorer credit are granted lower amounts or no post-paid permissions. This step not only helps the post-paid financial system to effectively assess risks, but also dynamically formulates personalized funding strategies for different users to ensure the safety of funds for platforms and merchants and reduce default and credit risks; through real-time analysis of user transaction-related data by deep learning algorithms, potential risk behaviors and default patterns can be identified. This step uses the predictive ability of deep learning models to conduct risk assessments on each transaction and timely predict the user's probability of default. This dynamic and intelligent analysis method can accurately adjust the fund guarantee strategy based on real-time transaction data, thereby achieving more flexible and effective risk control, avoiding the limitations of traditional static credit scores, and improving the real-time and accuracy of fund security; by obtaining the default risk coverage of the adjusted fund guarantee strategy, further evaluate whether the fund guarantee strategy can fully cover the potential default risk. If the risk coverage is insufficient, the system will automatically adjust the optimization strategy through secondary parameters to ensure that the fund guarantee can cope with possible default risks under different circumstances. This secondary optimization capability makes the fund guarantee strategy highly adaptable and can be adjusted at any time according to market changes and user behavior changes, further improving fund security and ensuring the long-term sustainable operation of the platform.

[0020] According to an embodiment of the present invention, the post-paid financial system is constructed by obtaining multi-source historical transaction-related data of users of the post-paid financial system, and constructing a user profile based on the multi-source historical transaction-related data, specifically: Building a post-paid financial system, obtaining multi-source historical transaction-related data of users of the post-paid financial system, wherein the multi-source historical transaction-related data includes basic user information, third-party payment platform credit score, historical consumption record data, and bank account information; Standardizing the multi-source historical transaction related data, calculating the covariance matrix of each data item of the standardized multi-source historical transaction related data, and calculating the eigenvalues ​​and eigenvectors of the covariance matrix; Arrange the eigenvalues ​​in descending order, select the eigenvectors corresponding to the first n eigenvalues ​​as principal components, construct a transformation matrix with the eigenvectors of the principal components, and perform a dimensionality reduction operation on the multi-source historical transaction-related data according to the transformation matrix to obtain dimensionality-reduced multi-source historical transaction-related data; Introducing the K-means clustering algorithm, analyzing the dimension-reduced multi-source historical transaction-related data based on the elbow rule, determining the clustering number k value of the K-means clustering algorithm, and randomly selecting k data points of each data item in the dimension-reduced multi-source historical transaction-related data as initial clustering centers; Calculate the Euclidean distance of each data point in each data item to the initial cluster center of the corresponding data item, assign each data point to the cluster center with the closest Euclidean distance to obtain the initial cluster, calculate the mean of the initial cluster, use the mean as the new cluster center, calculate the Euclidean distance of the data point to the new cluster center and assign the data point until the cluster center no longer changes, and obtain the clustering result of the dimensionality reduction multi-source historical transaction related data; Clustering features of each clustering result are extracted, and a user profile of the post-paid financial system is constructed based on the clustering features.

[0021] It should be noted that the present invention innovatively integrates the advantages of mainstream third-party payment platforms such as WeChat Pay's payment score and Alipay's Sesame Credit score, and constructs a post-paid financial framework with trust service capabilities. Under this system, users only need to complete the authorization operation of the third-party payment platform to break through the limitations of advance payment in the traditional transaction model and enjoy the value brought by the goods or services in advance. The system accurately and automatically determines whether to grant the user the right to post-pay according to the user's credit score. Standardizing multi-source historical transaction-related data (such as user basic information, third-party payment platform credit score, historical consumption record data, bank account information, etc.) can eliminate the dimensional differences between different data sources, so that the data can be compared under a unified scale. By calculating the covariance matrix of the standardized data and further calculating the eigenvalues ​​and eigenvectors, the main variation direction in the data set can be identified. It can effectively extract important information from the data. The data is reduced in dimension using principal component analysis (PCA) to retain the most representative and important features. By selecting the eigenvectors corresponding to the first n eigenvalues ​​as the principal components, the dimension of the data can be significantly reduced while ensuring that the main information of the data is not lost. The K-means clustering algorithm can classify users with similar behavior patterns by clustering user transaction data. In the user group, the clustering algorithm can identify different user groups (such as high-credit users, low-credit users, active users, potential default users, etc.) based on the user's consumption behavior, credit score and other characteristics. After the K-means clustering is completed, the clustering features of each clustering result (such as average consumption amount, default probability, consumption frequency, etc.) are extracted. Finally, by extracting features from each clustering result and constructing a user portrait, each user's credit status, consumption behavior, risk level, etc. can be accurately described. The user portrait contains a wealth of personal information, credit behavior, historical consumption records, payment ability and other dimensions.

[0022] Figure 2 The flowchart of the present invention for constructing a fund guarantee strategy for users with post-payment rights is shown.

[0023] According to an embodiment of the present invention, the initial credit rating of the user of the post-paid financial system is evaluated according to the user portrait, the post-paid authority of the user is determined according to the initial credit rating of the user, and a fund guarantee strategy for the user with post-paid authority is constructed, specifically as follows: S202, performing an initial credit rating assessment on users of the post-paid financial system according to the user portrait, granting post-paid rights to users whose initial credit ratings are greater than a preset level according to the initial credit ratings, and obtaining post-paid rights information for each user; S204, based on the post-payment authority information, if the post-payment authority user uses the post-payment function of the post-payment financial system during the purchase of goods, obtaining the transaction amount information, transaction subject credit evaluation information, and transaction commodity category information of the post-payment authority user; S206, evaluating the security requirement level of the user's commodity purchase according to the transaction amount information, the credit assessment information of the transaction subject, and the transaction commodity category information; if the security requirement level is lower than the preset requirement level, obtaining the post-payment agreed time node of the current commodity purchase order of the post-payment authorized user; if the user still fails to complete the payment of the current commodity purchase order within the agreed time node, marking the user as an overdue user, and the post-payment financial system provides a fund guarantee for the overdue user and advances funds to the target merchant; S208, if the overdue user needs to purchase the product again, restrict the overdue user from purchasing the product until the overdue user completes the payment process of the current product purchase order, and transfer the paid funds to the trust insurance account as a special fund reserve for accident recovery; S210, if the security requirement level is not lower than the preset requirement level, the user freezes the preset amount of funds in his own bank account or digital RMB account or corporate public account through pre-authorization before the transaction, and authorizes the system to pay the designated merchant after the electronic contract is confirmed. When the user needs to purchase goods again, the system identifies the user's current payment status of the purchase order of the goods, determines the next order purchase authority based on the payment status, and transfers the funds paid for the current purchase order to the trust insurance account as a special fund reserve for accident recovery, thereby obtaining a fund guarantee strategy for users with post-payment rights.

[0024] It should be noted that by conducting an initial credit rating assessment on users of the post-paid financial system based on user portraits and granting users post-paid rights based on their initial credit ratings, a flexible post-paid rights user fund guarantee strategy has been established. For post-paid financial systems based on small payments on third-party payment platforms, it can effectively reduce the risk of users' instant payments and protect the legitimate rights and interests of consumers. When users are unable to pay due to emergencies, the platform can play a guarantee role to ensure that merchants receive full payment within the agreed time node. When users purchase again, they need to complete the payment of the previous order. The funds paid will be transferred to the insurance account as a special fund reserve for accident recovery, completing a true post-paid payment closed loop. For payment scenarios that require a higher level of security, such as post-paid financial payment systems based on pre-authorized frozen funds for personal and corporate bank accounts, which are suitable for large and small transactions, it has designed a post-paid model based on digital RMB, personal bank cards, corporate public accounts, etc. Before the transaction, the user needs to pre-authorize and freeze a certain amount of funds in his own account and authorize the system to pay the designated merchant after the electronic contract is confirmed to ensure smooth settlement. When purchasing again, the previous order payment must be completed first. The funds paid are transferred to the insurance account reserve. The system introduces insurance companies as a third-party guarantee mechanism to insure each order. When the user cannot complete the payment, the insurance company initiates the claim procedure to protect the merchant's collection rights and interests and avoid affecting the user's personal credit. At the same time, this system does not make a fund pool and is not subject to the supervision of existing financial regulations. It is a new payment method with trust capabilities. Through these technical means, the comprehensive effects of transaction guarantee, fund guarantee, trust fund reserve, user and merchant rights protection are achieved, providing a more complete, innovative and secure solution for post-paid financial transactions. This system does not make a fund pool and is not subject to the supervision of existing financial regulations. It is a new payment method with trust capabilities. The security requirement level is a standard or level used to measure the security requirements during a transaction, which takes into account multiple factors such as transaction amount information, transaction subject credit assessment information, transaction commodity category information, etc.

[0025] According to an embodiment of the present invention, the real-time acquisition of user transaction-related data of the postpaid financial system, the analysis of the user transaction-related data based on a deep learning algorithm, the prediction of the user's default probability, and the adjustment of the fund guarantee strategy parameters according to the default probability are specifically as follows: Acquire transaction-related data of each user in the postpaid financial system in real time, wherein the user transaction-related data includes user information, transaction object information, transaction relationship information between the user and the transaction object, and historical default data; The user and transaction object are defined as nodes of the graph structure respectively, the user information and transaction object information are used as attributes of the nodes, and the transaction relationship between the user and the transaction object is defined as the edge of the graph structure. The graph structure of user transactions is constructed according to the nodes, attributes, and edges to obtain a graph structure dataset; Dividing the graph structure data set into a training set and a test set according to a preset ratio, building a user default prediction model based on the GCN graph neural network, and initializing the parameters of the user default prediction model, including a weight matrix and a bias term; Determine the dimensions of the input layer and output layer, and the dimensions and number of layers of the intermediate hidden layer of the user default prediction model according to the characteristic dimensions of the graph structure; Importing the training set into the user default prediction model for training, aggregating the features of the user default prediction model based on forward propagation, and calculating the error between the predicted default probability and the actual default probability of the user default prediction model based on a binary cross entropy loss function; Analyzing the error through back propagation, and updating the parameters of the user default prediction model, wherein the parameters include a weight matrix, a bias term, a learning rate, a hidden layer dimension and number of layers, and an activation function parameter; Importing the test set into the user default prediction model to perform prediction performance evaluation, and adjusting the hyperparameters of the user default prediction model according to the prediction performance evaluation result; Acquire real-time transaction-related data of users of the postpaid financial system within a preset time period and import it into the user default prediction model to predict the user default probability and obtain a default probability prediction result; The fund guarantee strategy parameters are adjusted according to the default probability prediction result, and the fund guarantee strategy parameters include fund advance payment ratio, payment period, and guarantee threshold.

[0026] It should be noted that multi-dimensional user transaction-related data including user information, transaction object information, transaction relationship information and historical default data are obtained in real time. Integrating these data into a graph structure, taking users and transaction objects as nodes, their corresponding information as attributes, and transaction relationships as edges, can comprehensively and intuitively reflect the complex network structure of user transactions. Compared with traditional data processing methods, this integration method can better capture the potential connections between data, such as the transaction patterns and transaction frequency between different users and transaction objects. The user default prediction model is constructed based on the GCN graph neural network. GCN can effectively process graph structure data and learn the feature representation of nodes by aggregating node neighbor information. In the user default prediction scenario, the characteristics of each node (user or transaction object) depend not only on its own information, but also on the surrounding neighbor nodes (objects with which it has a transaction relationship). GCN can automatically learn these complex relationships, thereby more accurately capturing the patterns and laws of user defaults. For example, by analyzing the credit status, transaction activity and other information of multiple transaction objects with which a user has a transaction relationship, combined with the user's own historical default data, the accuracy of the prediction of the user's default probability can be improved; based on the accurate default probability prediction results, the parameters of the fund guarantee strategy (fund advance ratio, payment period, guarantee threshold) are adjusted. For example, for users with a high probability of default, the fund advance ratio is appropriately reduced, the payment period is extended, and the guarantee threshold is increased to reduce potential risk losses; for users with a low probability of default, a relatively loose fund guarantee strategy can be adopted to ensure that funds are reasonably allocated among users with different risks, avoiding excessive guarantee of high-risk users and resulting in resource waste or potential losses. The user information includes age, gender, income level, historical transaction records, and credit score; the transaction object information includes but is not limited to the type of transaction object, credit rating, and business scope; the transaction relationship information includes transaction time, transaction amount, and transaction type; the feature dimension is the number of features. The guarantee threshold includes amount threshold, credit threshold, transaction type threshold, transaction risk level threshold, default time threshold, and transaction duration threshold. The hyper parameters include the hidden layer dimension, the number of hidden layers, the learning rate, the optimizer parameters (such as the momentum coefficient of the Adam optimizer), the batch size, the number of training rounds (Epochs), and the Dropout ratio of the user default prediction model.

[0027] Figure 3 A flow chart of the present invention for obtaining a fund guarantee optimization strategy is shown.

[0028] According to an embodiment of the present invention, the default risk coverage of the fund guarantee strategy after parameter adjustment is obtained, and the fund guarantee strategy is adjusted for a second time according to the default risk coverage to obtain a fund guarantee optimization strategy, specifically: S302, introducing a Monte Carlo simulation algorithm, obtaining user transaction data of the fund guarantee strategy after parameter adjustment, and setting a simulation scenario of the Monte Carlo simulation algorithm according to the user transaction data, user profile, and current fund guarantee strategy parameters, wherein the simulation scenario includes the user's transaction amount, transaction frequency, transaction type, user's transaction level distribution, and user's default probability distribution; S304, simulating the simulation scenario based on a Monte Carlo simulation algorithm, randomly generating a preset number of user transactions and default events according to the simulation scenario, determining whether a fund guarantee condition is triggered during the simulation of the user transactions and default time according to the current fund guarantee strategy parameters, and calculating the proportion of the default transaction amount guaranteed in each simulation process to the total default transaction amount; S306, evaluating the funding gap compensation capability of the fund guarantee strategy according to the ratio of the default transaction amount to the total default transaction amount, and determining the default risk coverage degree of the fund guarantee strategy according to the compensation capability; S308: If the default risk coverage is less than a preset value, a secondary parameter adjustment is performed on the fund guarantee strategy based on a particle swarm optimization algorithm to obtain a fund guarantee optimization strategy.

[0029] It should be noted that the Monte Carlo simulation algorithm can set a variety of simulation scenarios based on user transaction data, user portraits and current fund guarantee strategy parameters. It can cover multiple dimensions such as user transaction amount, transaction frequency, transaction type, user transaction level distribution and user default probability distribution. Through random sampling, the algorithm can simulate many different combinations of transactions and default events, which represent various complex situations that may occur in reality. For example, for different transaction amounts, there may be different default risks, and Monte Carlo simulation can generate various possible transaction amounts in a large range, from small daily consumption amounts to large commercial purchase amounts, simulating their situations under different default probabilities; for transaction frequency, it can also simulate various situations from high-frequency frequent transaction users to low-frequency occasional transaction users, so that the impact of various user behaviors and characteristics on default risks can be fully considered. By performing multiple simulation operations on the simulation scenario, according to the current fund guarantee strategy parameters, it can be determined whether the fund guarantee conditions are triggered in each simulated user transaction and default event. By calculating the proportion of the default transaction amount guaranteed in each simulation process to the total default transaction amount, the effect of the fund guarantee strategy in different situations can be intuitively evaluated. This ratio reflects the ability of the fund guarantee strategy to respond to default events, that is, the proportion of the default amount that the strategy can cover when a default occurs. For example, assuming that in a simulation, the total default transaction amount is 1 million yuan, and the default transaction amount covered is 800,000 yuan, then the default risk coverage is 80%. Through multiple simulations, a series of such coverage results can be obtained, and the fund guarantee strategy's ability to make up for the funding gap in different scenarios can be comprehensively evaluated. When the assessed default risk coverage is less than the preset degree value, the fund guarantee strategy can be adjusted for a second time based on the particle swarm optimization algorithm to obtain a fund guarantee optimization strategy. This dynamic adjustment mechanism can automatically optimize the parameters of the fund guarantee strategy according to different risk conditions and market environments, such as adjusting the fund advance payment ratio, payment period or guarantee threshold. For example, when it is found that the default risk coverage of a certain type of user is low, a more suitable parameter combination can be found through the particle swarm optimization algorithm, making the fund guarantee strategy more targeted and more adaptable to the actual risk situation, thereby improving the adaptability and effectiveness of the fund guarantee strategy.

[0030] According to an embodiment of the present invention, the particle swarm optimization algorithm is used to perform secondary parameter adjustment on the fund guarantee strategy to obtain a fund guarantee optimization strategy, which is specifically: Obtain the reserve level of the special fund for claim recovery of the trust insurance account of the post-paid financial system, determine the default risk tolerance of the post-paid financial system based on the reserve level, and determine the default risk coverage threshold of the fund guarantee strategy based on the default risk tolerance; A particle swarm optimization algorithm is introduced to convert the parameters of the fund guarantee strategy into the particle position vector of the particle swarm optimization algorithm, and the default risk coverage threshold is used as the optimization target of the particle swarm optimization algorithm; A particle swarm is randomly generated, a velocity vector is initialized for each particle in the particle swarm, the default risk coverage of each parameter combination is calculated according to the parameter combination of the fund hedging strategy represented by the particle position vector, and the initial fitness of each particle position vector is determined according to the default risk coverage of each parameter combination; The highest initial fitness is taken as the global optimal fitness, the position vector of the particle with the highest fitness is taken as the global optimal position, the fitness and the global optimal position are analyzed based on the initialization velocity vector, the global optimal position is updated, and the default risk coverage of the parameter combination represented by the global optimal position after the update operation is recalculated until the default risk coverage is greater than the default risk coverage threshold, and the particle at the global optimal position is output; The parameter combination of the particle at the global optimal position is identified, and the secondary parameter adjustment of the fund guarantee strategy is performed according to the parameter combination to obtain the fund guarantee optimization strategy.

[0031] It should be noted that by obtaining the reserve level of special funds for insurance claims in the trust insurance account of the post-paid financial system, this method can take the resource status of the system into consideration. This makes the optimization of the fund guarantee strategy not isolated, but closely linked to the actual resources of the system. The default risk tolerance is determined according to the reserve level, and then the default risk coverage threshold is determined to ensure that the adjustment of the fund guarantee strategy will not exceed the tolerance of the system. For example, if the special fund reserve for insurance claims in the trust insurance account is sufficient, the system may have a high default risk tolerance, and the corresponding default risk coverage threshold can be set higher; conversely, if the reserve level is low, the system's risk tolerance is weak, and the corresponding threshold will be lowered. This can avoid the situation where system resources are tight or resource utilization is unreasonable due to excessive guarantee. The particle swarm optimization algorithm is introduced to convert the parameters of the fund guarantee strategy into particle position vectors, and the characteristics of the particle swarm are used to search in a wide parameter space. The particle swarm optimization algorithm can evaluate and adjust multiple parameter combinations at the same time, has a strong global search capability, and avoids falling into a local optimal solution. For example, for multiple fund guarantee strategy parameters such as fund advance ratio, payment period, guarantee threshold, etc., by converting them into particle position vectors, the particle swarm algorithm can quickly explore in different value ranges, find different parameter combinations, and evaluate the advantages and disadvantages of each combination based on the default risk coverage. A particle swarm is randomly generated and the velocity vector is initialized. The default risk coverage is calculated and the fitness is determined based on the parameter combination represented by the particle position vector, and the global optimal fitness and global optimal position are continuously updated. By continuously analyzing and updating the fitness and global optimal position, the algorithm can continuously optimize the parameter combination until the default risk coverage is greater than the set threshold. This means that it can automatically adjust the parameters to find the optimal parameter combination that meets or exceeds the target coverage, thereby achieving dynamic optimization of the fund guarantee strategy.

[0032] According to an embodiment of the present invention, it also includes: Monitoring the transaction behavior of users of the post-paid financial system, extracting the transaction behavior characteristics of the users within a preset time period based on the transaction behavior, and marking the transaction behavior characteristics as benchmark behavior characteristics; Comparing the user's real-time transaction behavior with the benchmark behavior characteristics, determining the Manhattan distance between the real-time transaction behavior and the benchmark behavior characteristics, and evaluating the difference between the real-time transaction behavior and the benchmark behavior characteristics according to the Manhattan distance to obtain transaction behavior difference data; Performing transaction anomaly assessment on the user's real-time transaction behavior according to the transaction behavior difference data to obtain transaction risk data; According to the transaction risk data, if the transaction risk is higher than the preset value, the user's transaction is temporarily frozen, and the user whose transaction risk is higher than the preset value is subject to multi-level verification, which includes SMS verification code, fingerprint recognition, and video verification. A risk report is generated and sent to the system staff, and the transaction of users whose transaction risk is higher than the preset value is monitored; If the user's abnormal transaction frequency is greater than the preset frequency value, adjust the user's credit rating.

[0033] It should be noted that in the post-paid financial system, the transaction behavior of users often has certain risks, especially when the transaction mode of users is abnormal, it may lead to default, fraud or other improper behaviors. Therefore, how to identify and prevent these potential risks in time has become an important issue to protect the interests of the platform and merchants. The present invention monitors the transaction behavior of users in the post-paid financial system in real time, extracts the transaction behavior characteristics of users in a preset time period, and marks them as benchmark behavior characteristics. Based on this, the real-time transaction behavior of users is compared and analyzed, and the Manhattan distance between them and the benchmark behavior characteristics is determined, so as to evaluate the difference in transaction behavior. If the difference is large, it can be determined as an abnormal transaction, and the transaction risk data is further evaluated. By setting a risk threshold, when the transaction risk exceeds the preset value, the system will temporarily freeze the user's transaction, and at the same time start a multi-level verification mechanism, such as SMS verification code, fingerprint recognition, video verification, etc., to ensure the security of the transaction. In addition, for high-risk users, the system will also generate a detailed risk report and send it to the staff for key monitoring. By tracking and analyzing the frequency of abnormal transactions of users, if it is found that the frequency of abnormal transactions exceeds the preset frequency value, the system will automatically adjust the credit rating of the user. This technical solution can effectively improve the system's ability to identify and respond to abnormal transactions, thereby reducing the platform's potential risks and ensuring transaction security, while also improving users' credit management level and optimizing the post-paid financial environment.

[0034] Figure 4 A block diagram of a payment system with fund guarantee and trust as the goal of the present invention is shown.

[0035] The second aspect of the present invention further provides a payment system 4 with the goal of fund guarantee and trust, the system comprising: a memory 41, a processor 42, the memory comprising a post-payment guarantee method program with the goal of fund guarantee and trust, and when the post-payment guarantee method program with the goal of fund guarantee and trust is executed by the processor, the following steps are implemented: Building a post-paid financial system, acquiring multi-source historical transaction-related data of users of the post-paid financial system, and building a user profile based on the multi-source historical transaction-related data; Performing an initial credit rating assessment on a user of the post-paid financial system according to the user portrait, determining the user's post-paid authority according to the user's initial credit rating, and constructing a fund guarantee strategy for users with post-paid authority; Acquire user transaction-related data of the postpaid financial system in real time, analyze the user transaction-related data based on a deep learning algorithm, predict the user's default probability, and adjust the capital guarantee strategy parameters according to the default probability; The default risk coverage degree of the fund guarantee strategy after parameter adjustment is obtained, and the parameters of the fund guarantee strategy are adjusted twice according to the default risk coverage degree to obtain the fund guarantee optimization strategy.

[0036] The present invention discloses a post-paid security method and payment system with the goal of fund guarantee and trust, aiming to provide real-time credit assessment and risk control for users in the post-paid financial system. First, a post-paid financial system is constructed and multi-source historical transaction data of users is collected, and user portraits are constructed through data analysis; then, a preliminary credit assessment is performed based on the user portrait, the post-paid authority is determined, and a fund guarantee strategy is formulated; then, a deep learning algorithm is used to monitor user transaction data in real time, predict the probability of default, and dynamically adjust the parameters of the fund guarantee strategy; finally, the fund guarantee strategy is optimized based on the degree of default risk coverage to form a dynamically adjusted fund guarantee optimization strategy. This method effectively improves transaction security, reduces default risk, and enhances user trust. It is suitable for multiple scenarios such as finance and e-commerce, and has broad application prospects.

[0037] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0038] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A post-payment guarantee method with the goal of fund guarantee and trust, characterized in that: The following steps are involved: Building a post-paid financial system, acquiring multi-source historical transaction-related data of users of the post-paid financial system, and building a user profile based on the multi-source historical transaction-related data; Performing an initial credit rating assessment on a user of the post-paid financial system according to the user portrait, determining the user's post-paid authority according to the user's initial credit rating, and constructing a fund guarantee strategy for users with post-paid authority; Acquire user transaction-related data of the postpaid financial system in real time, analyze the user transaction-related data based on a deep learning algorithm, predict the user's default probability, and adjust the capital guarantee strategy parameters according to the default probability; The default risk coverage degree of the fund guarantee strategy after parameter adjustment is obtained, and the parameters of the fund guarantee strategy are adjusted twice according to the default risk coverage degree to obtain the fund guarantee optimization strategy.

2. A post-payment guarantee method with fund guarantee and trust as the goal according to claim 1, characterized in that: The post-paid financial system is constructed, and multi-source historical transaction-related data of users of the post-paid financial system is obtained, and user portraits are constructed according to the multi-source historical transaction-related data, specifically: Building a post-paid financial system, obtaining multi-source historical transaction-related data of users of the post-paid financial system, wherein the multi-source historical transaction-related data includes basic user information, third-party payment platform credit score, historical consumption record data, and bank account information; Standardizing the multi-source historical transaction related data, calculating the covariance matrix of each data item of the standardized multi-source historical transaction related data, and calculating the eigenvalues ​​and eigenvectors of the covariance matrix; Arrange the eigenvalues ​​in descending order, select the eigenvectors corresponding to the first n eigenvalues ​​as principal components, construct a transformation matrix with the eigenvectors of the principal components, and perform a dimensionality reduction operation on the multi-source historical transaction-related data according to the transformation matrix to obtain dimensionality-reduced multi-source historical transaction-related data; Introducing the K-means clustering algorithm, analyzing the dimension-reduced multi-source historical transaction-related data based on the elbow rule, determining the clustering number k value of the K-means clustering algorithm, and randomly selecting k data points of each data item in the dimension-reduced multi-source historical transaction-related data as initial clustering centers; Calculate the Euclidean distance of each data point in each data item to the initial cluster center of the corresponding data item, assign each data point to the cluster center with the closest Euclidean distance to obtain the initial cluster, calculate the mean of the initial cluster, use the mean as the new cluster center, calculate the Euclidean distance of the data point to the new cluster center and assign the data point until the cluster center no longer changes, and obtain the clustering result of the dimensionality reduction multi-source historical transaction related data; Clustering features of each clustering result are extracted, and a user profile of the post-paid financial system is constructed based on the clustering features.

3. A post-payment guarantee method with fund guarantee and trust as the goal according to claim 1, characterized in that: The initial credit rating of the user of the post-payment financial system is evaluated according to the user portrait, the post-payment authority of the user is determined according to the initial credit rating of the user, and a fund guarantee strategy for the user with post-payment authority is constructed, specifically: Performing an initial credit rating assessment on users of the post-paid financial system according to the user portrait, granting post-paid rights to users whose initial credit rating is greater than a preset level according to the initial credit rating, and obtaining post-paid rights information for each user; According to the post-payment authority information, if the post-payment authority user uses the post-payment function of the post-payment financial system during the purchase of goods, obtaining the transaction amount information, transaction subject credit evaluation information, and transaction commodity category information of the post-payment authority user; The security requirement level of the user's commodity purchase is evaluated based on the transaction amount information, the credit assessment information of the transaction subject, and the transaction commodity category information. If the security requirement level is lower than the preset requirement level, the post-payment agreed time node of the current commodity purchase order of the post-payment permission user is obtained. If the user still fails to complete the payment of the current commodity purchase order within the agreed time node, the user is marked as an overdue user, and the post-payment financial system provides a fund guarantee for the overdue user and advances funds to the target merchant; If the overdue user needs to purchase the product again, the purchase transaction operation will be restricted for the overdue user until the overdue user completes the payment process of the current product purchase order, and the paid funds will be transferred to the trust insurance account as a special fund reserve for accident recovery; If the security requirement level is not lower than the preset requirement level, the user freezes the preset amount of funds in his or her own bank account or digital RMB account or corporate public account through pre-authorization before the transaction, and authorizes the system to pay the designated merchant after the electronic contract is confirmed. When the user needs to purchase the product again, the system will identify the user's current product purchase order payment status, determine the next order purchase authority based on the payment status, and transfer the funds paid for the current product purchase order to the trust insurance account as a special fund reserve for claim recovery, thereby obtaining a fund guarantee strategy for users with post-payment permissions.

4. A post-payment guarantee method with fund guarantee and trust as the goal according to claim 1, characterized in that: The real-time acquisition of user transaction-related data of the post-paid financial system, analysis of the user transaction-related data based on a deep learning algorithm, prediction of the user's default probability, and adjustment of the fund guarantee strategy parameters according to the default probability are specifically as follows: Acquire transaction-related data of each user in the postpaid financial system in real time, wherein the user transaction-related data includes user information, transaction object information, transaction relationship information between the user and the transaction object, and historical default data; Define users and transaction objects as nodes of a graph structure, user information and transaction object information as attributes of the nodes, and the transaction relationship between users and transaction objects as edges of the graph structure. Build a graph structure of user transactions based on the nodes, attributes, and edges to obtain a graph structure dataset. Dividing the graph structure data set into a training set and a test set according to a preset ratio, building a user default prediction model based on the GCN graph neural network, and initializing the parameters of the user default prediction model, including a weight matrix and a bias term; Determine the dimensions of the input layer and output layer, and the dimensions and number of layers of the intermediate hidden layer of the user default prediction model according to the characteristic dimensions of the graph structure; Importing the training set into the user default prediction model for training, aggregating the features of the user default prediction model based on forward propagation, and calculating the error between the predicted default probability and the actual default probability of the user default prediction model based on a binary cross entropy loss function; Analyzing the error through back propagation, and updating the parameters of the user default prediction model, wherein the parameters include a weight matrix, a bias term, a learning rate, a hidden layer dimension and number of layers, and an activation function parameter; Importing the test set into the user default prediction model to perform prediction performance evaluation, and adjusting the hyperparameters of the user default prediction model according to the prediction performance evaluation result; Acquire real-time transaction-related data of users of the postpaid financial system within a preset time period and import it into the user default prediction model to predict the user default probability and obtain a default probability prediction result; The fund guarantee strategy parameters are adjusted according to the default probability prediction result, and the fund guarantee strategy parameters include fund advance payment ratio, payment period, and guarantee threshold.

5. A post-payment guarantee method with fund guarantee and trust as the goal according to claim 1, characterized in that: The default risk coverage of the fund guarantee strategy after the parameter adjustment is obtained, and the fund guarantee strategy is adjusted for the second time according to the default risk coverage to obtain the fund guarantee optimization strategy, which is specifically: Introduce the Monte Carlo simulation algorithm to obtain the user transaction data of the fund guarantee strategy after parameter adjustment, and set the simulation scenario of the Monte Carlo simulation algorithm according to the user transaction data, user portrait, and current fund guarantee strategy parameters. The simulation scenario includes the user's transaction amount, transaction frequency, transaction type, user's transaction level distribution, and user's default probability distribution; The simulation scenario is simulated based on the Monte Carlo simulation algorithm, a preset number of user transactions and default events are randomly generated according to the simulation scenario, and whether the fund guarantee condition is triggered during the simulation of the user transaction and default time is determined according to the current fund guarantee strategy parameters, and the proportion of the default transaction amount guaranteed in each simulation process to the total default transaction amount is calculated; The ability of the fund guarantee strategy to make up for the funding gap is evaluated based on the ratio of the default transaction amount to the total default transaction amount, and the degree of default risk coverage of the fund guarantee strategy is determined based on the compensation ability; If the default risk coverage is less than the preset value, a secondary parameter adjustment is performed on the fund guarantee strategy based on the particle swarm optimization algorithm to obtain the fund guarantee optimization strategy.

6. A post-payment guarantee method with fund guarantee and trust as the goal according to claim 5, characterized in that: The particle swarm optimization algorithm is used to perform secondary parameter adjustment on the fund guarantee strategy to obtain the fund guarantee optimization strategy, which is specifically: Obtain the reserve level of the special fund for claim recovery of the trust insurance account of the post-paid financial system, determine the default risk tolerance of the post-paid financial system based on the reserve level, and determine the default risk coverage threshold of the fund guarantee strategy based on the default risk tolerance; A particle swarm optimization algorithm is introduced to convert the parameters of the fund guarantee strategy into the particle position vector of the particle swarm optimization algorithm, and the default risk coverage threshold is used as the optimization target of the particle swarm optimization algorithm; A particle swarm is randomly generated, a velocity vector is initialized for each particle in the particle swarm, the default risk coverage of each parameter combination is calculated according to the parameter combination of the fund hedging strategy represented by the particle position vector, and the initial fitness of each particle position vector is determined according to the default risk coverage of each parameter combination; The highest initial fitness is taken as the global optimal fitness, the position vector of the particle with the highest fitness is taken as the global optimal position, the fitness and the global optimal position are analyzed based on the initialization velocity vector, the global optimal position is updated, and the default risk coverage of the parameter combination represented by the global optimal position after the update operation is recalculated until the default risk coverage is greater than the default risk coverage threshold, and the particle at the global optimal position is output; The parameter combination of the particle in the global optimal position is identified, and the secondary parameter adjustment of the fund guarantee strategy is performed according to the parameter combination to obtain the fund guarantee optimization strategy.

7. A payment system with the goal of fund guarantee and trust, characterized in that: The payment system with the goal of fund guarantee and trust includes a storage device and a processor. The storage device includes a post-payment guarantee method program with the goal of fund guarantee and trust. When the post-payment guarantee method program with the goal of fund guarantee and trust is executed by the processor, the following steps are implemented: Building a post-paid financial system, acquiring multi-source historical transaction-related data of users of the post-paid financial system, and building a user profile based on the multi-source historical transaction-related data; Performing an initial credit rating assessment on a user of the post-paid financial system according to the user portrait, determining the user's post-paid authority according to the user's initial credit rating, and constructing a fund guarantee strategy for users with post-paid authority; Acquire user transaction-related data of the postpaid financial system in real time, analyze the user transaction-related data based on a deep learning algorithm, predict the user's default probability, and adjust the capital guarantee strategy parameters according to the default probability; The default risk coverage degree of the fund guarantee strategy after parameter adjustment is obtained, and the parameters of the fund guarantee strategy are adjusted twice according to the default risk coverage degree to obtain the fund guarantee optimization strategy.

Images