Multi-dimensional credit evaluation methods for capacity trading platforms

Abstract

This invention relates to the field of data processing methods, specifically to a multi-dimensional credit evaluation method for a transportation capacity trading platform. The method includes: acquiring user-level transportation capacity datasets for each user connected to the platform; mapping the user-level transportation capacity datasets to a multi-dimensional sample space to obtain vehicle data points corresponding to each vehicle; clustering the vehicle data points in the multi-dimensional sample space to obtain at least one vehicle subclass and determining the integrity index of each vehicle subclass; determining the user's anomaly resistance index; monitoring newly connected transportation capacity data and determining the change in the user's credit score based on the user's anomaly resistance index and the newly connected transportation capacity data; and updating the user's credit score. This invention achieves real-time updates to user credit scores by continuously monitoring newly connected transportation capacity data and dynamically updating the change in credit scores using the integrity index before and after the data access and the user's anomaly resistance index.

Description

Technical Field

[0001] This invention relates to the field of data processing methods, specifically to a multi-dimensional credit evaluation method for a transportation capacity trading platform. Background Technology

[0002] With the increasing online presence of road freight, capacity trading platforms integrate resources from shippers, carriers, and individual drivers to form a service ecosystem that features centralized matching, online transactions, and end-to-end management. The platform needs to conduct real-time assessments of the performance quality of participating parties to provide decision-making support in areas such as bidding, dispatching, financing, and risk control. Traditionally, this involves aggregating multi-dimensional operational data such as vehicle mileage, average speed, and cargo damage rate, calculating a static credit score using a fixed-weighting method, and periodically updating the results to form capacity credit.

[0003] Existing problems: When faced with high-frequency, heterogeneous capacity data, the above static weighted model is unable to reflect the differences in timeliness and trends of data from different dimensions, nor can it capture the real-time fluctuations of user capacity behavior, resulting in delayed and inaccurate credit rating results, which in turn affects the platform's judgment efficiency in transaction matching and risk control. Summary of the Invention

[0004] This invention provides a multi-dimensional credit evaluation method for a capacity trading platform to solve existing problems.

[0005] The multi-dimensional credit evaluation method for a capacity trading platform of the present invention adopts the following technical solution:

[0006] One embodiment of the present invention provides a multi-dimensional credit evaluation method for a capacity trading platform, the method comprising:

[0007] Obtain user-level capacity datasets for each user connected to the capacity trading platform; wherein, the user-level capacity datasets include multi-dimensional capacity data for all vehicles belonging to the same user;

[0008] The user-level capacity dataset is mapped to a multi-dimensional sample space to obtain vehicle data points corresponding to the vehicles.

[0009] In the multidimensional sample space, the vehicle data points are clustered to obtain at least one vehicle subclass, and an integrity index for each vehicle subclass is determined; wherein, the integrity index is used to characterize the density of the vehicle data points within the vehicle subclass.

[0010] Based on the integrity index of each vehicle subclass, a user anomaly resistance index is determined; wherein, the user anomaly resistance index is used to characterize the degree of resistance to interference caused by changes in the integrity index due to newly accessed capacity data.

[0011] Monitor the newly accessed capacity data, and determine the change in the user's credit score based on the user's anomaly resistance index and the newly accessed capacity data;

[0012] The user's credit score is updated using the change in credit score.

[0013] Furthermore, the clustering of the vehicle data points includes:

[0014] Starting from any of the vehicle data points, sequentially retrieve the vehicle data points that have the smallest geometric distance from the starting point and have not been classified in the multidimensional sample space, and temporarily merge the retrieved vehicle data points into the same subclass to be formed.

[0015] For each new vehicle data point incorporated, the integrity index of the subclass to be formed is calculated;

[0016] When the integrity index after incorporating a new vehicle data point is less than the previous integrity index, the search stops, and the vehicle data points that have been incorporated into the subclass to be formed are identified as vehicle subclasses; wherein, the previous integrity index is the integrity index before incorporating a new vehicle data point.

[0017] Further, calculating the integrity index of the subclass to be formed includes:

[0018] Determine the subclass compactness gain component and the subclass data coverage component of the subclass to be formed; wherein, the subclass compactness gain component is used to characterize the degree to which the new vehicle data points make the spatial distribution within the subclass to be formed tend to be compact, and the subclass data coverage component is used to characterize the proportion of the number of data points already contained in the subclass to be formed to the total number of data points of the corresponding user.

[0019] The integrity index of the subclass to be formed is determined based on the subclass density gain component and the subclass data coverage component.

[0020] Further, determining the subclass compactness gain component of the subclass to be formed includes:

[0021] Determine the change in spatial density of the subclass to be formed before and after incorporating the new vehicle data point; wherein the change in spatial density is used to characterize the change in the density of the subclass to be formed caused by incorporating the new vehicle data point.

[0022] Determine the dispersion of the vehicle data points in the subclass to be formed before the new vehicle data points are incorporated.

[0023] Based on the spatial density change and the dispersion, the subclass density gain component of the subclass to be formed is determined.

[0024] Furthermore, determining the user's anomaly resistance index based on the integrity index of each of the vehicle subclasses includes:

[0025] Based on the integrity index of each vehicle subclass, a subclass stability component and a subclass size normalization component are determined for each vehicle subclass; wherein, the subclass stability component is used to characterize the relative fluctuation of the integrity index among the vehicle subclasses corresponding to the user; and the subclass size normalization component is used to characterize the proportion of the number of vehicle subclasses corresponding to the user in the total number of vehicle subclasses in the capacity trading platform.

[0026] Based on the subclass stability component and the subclass size normalization component, the user's anomaly resistance index is determined.

[0027] Further, determining the subclass stability component of the vehicle subclass includes:

[0028] For each vehicle subclass corresponding to the user, the minimum difference between the integrity index of the current subclass and the integrity index of other subclasses is calculated one by one to determine the minimum integrity deviation corresponding to each vehicle subclass.

[0029] The minimum integrity deviations are summed to determine the total user-level integrity deviations;

[0030] Determine the mean integrity index of the vehicle subclass containing only one vehicle data point, and determine a baseline stable value based on the mean integrity index;

[0031] Based on the sum of the user-level integrity deviations and the baseline stability value, the subclass stability component of the vehicle subclass is determined.

[0032] Furthermore, determining the change in the user's credit score based on the user's anomaly resistance index and the newly accessed capacity data includes:

[0033] Based on the integrity index of the vehicle subclass before and after accessing the newly accessed capacity data, an integrity change coefficient is determined; wherein, the integrity change coefficient is used to characterize the direction of fluctuation of the integrity index caused by the newly accessed capacity data, a positive integrity change coefficient indicates that the newly accessed capacity data leads to an improvement in the integrity index, and a negative integrity change coefficient indicates that the newly accessed capacity data leads to a decrease in the integrity index;

[0034] Based on the user anomaly resistance capability index before and after accessing the newly accessed capacity data, an anomaly resistance correction coefficient is determined.

[0035] Based on the integrity change coefficient, the anomaly resistance correction coefficient, and the user anomaly resistance capability index after accessing the newly accessed capacity data, the change in the user's credit score is determined.

[0036] Furthermore, the method for determining the anomaly resistance correction coefficient includes:

[0037] Calculate the ratio of the user anomaly resistance index after accessing the new access capacity data to the user anomaly resistance index before accessing the new access capacity data, and determine the anomaly resistance correction coefficient.

[0038] Further, obtaining the user-level capacity dataset includes:

[0039] The multi-dimensional transportation capacity data is collected in real time during the vehicle's operation using positioning and sensing devices installed on the vehicle; wherein the multi-dimensional transportation capacity data includes mileage, driving time, average speed, single trip time, and cargo damage rate.

[0040] The multi-dimensional transport capacity data is collected according to the vehicle number to form a single-vehicle transport capacity dataset;

[0041] The user-level capacity dataset is obtained by aggregating the single-vehicle capacity datasets of all vehicles belonging to the same user.

[0042] Furthermore, the method also includes:

[0043] When a user's credit score falls below a preset credit score threshold, the user's trading qualification on the transportation capacity trading platform is revoked.

[0044] The beneficial effects of the technical solution of the present invention are:

[0045] In this embodiment of the invention, user-level capacity datasets of each user connected to the capacity trading platform are obtained; the user-level capacity datasets are mapped to a multi-dimensional sample space to obtain vehicle data points corresponding to the vehicles; in the multi-dimensional sample space, the vehicle data points are clustered to obtain at least one vehicle subclass, and the integrity index of each vehicle subclass is determined; based on the integrity index of each vehicle subclass, the user's anomaly resistance capability index is determined; newly connected capacity data is monitored, and based on the user's anomaly resistance capability index and the newly connected capacity data, the change in the user's credit score is determined; and the change in the credit score is used to update the user's credit score. This invention achieves real-time updates to user credit scores by continuously monitoring newly integrated capacity data and dynamically updating credit score changes using integrity indicators before and after data integration, as well as user anomaly resistance indicators. Furthermore, it maps user-level capacity datasets to a multi-dimensional sample space and performs clustering, quantifying the similarity of vehicle behavior at a geometric level to form highly cohesive vehicle subclasses. This reduces the heterogeneous vehicle mixing rate caused by single-dimensional threshold screening, improving the sensitivity and accuracy of anomaly detection. Additionally, the user anomaly resistance indicator, calculated based on subclass integrity indicators, embeds the relative change in anomaly resistance into an anomaly resistance correction coefficient in a proportional form. This makes the credit score change non-linearly adaptive to external disturbances, weakening score cliffs caused by occasional anomalies and enhancing the robustness of the evaluation system. Finally, the credit score change is driven by both the integrity change coefficient and the anomaly resistance correction coefficient, enabling rapid updates to the credit score. This allows the capacity trading platform to dynamically adjust transaction qualifications, risk control levels, and scheduling priorities based on the updated credit score, forming a data-scoring-control closed loop, reducing manual intervention, and improving operational efficiency. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 A flowchart illustrating the multi-dimensional credit evaluation method for a capacity trading platform provided in this embodiment of the invention;

[0048] Figure 2 This is a flowchart illustrating the vehicle data point clustering method provided in an embodiment of the present invention. Detailed Implementation

[0049] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a multi-dimensional credit evaluation method for a capacity trading platform proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0051] It should be noted that traditional credit scoring models rely on static weights and periodic batch updates, making it difficult to reflect instantaneous changes in capacity behavior in real time. This allows even occasional anomalies to lower the overall score, while small, high-frequency anomalies are easily masked by averaging. Therefore, this invention focuses on real-time and adaptive approaches. By continuously monitoring newly accessed capacity data, it first uses historical operational data to assess the user's original resilience baseline. Then, based on the direction and magnitude of the latest data's impact on the sub-category integrity index, it dynamically calculates the change in the updated credit score. After a second weighting by the user's anomaly resistance index, the change only significantly corrects deviations that truly reflect behavioral trends, while suppressing random, occasional anomalies due to the amplified resistance coefficient. Thus, the platform can quickly refresh the credit score, highly coupling the score with real-time capacity behavior. This avoids a precipitous drop in score due to a single anomaly and ensures that the impact of small-scale abnormal transactions on the overall score is limited to a controllable range, achieving refined credit updates with strong trend sensitivity and weak noise response.

[0052] Furthermore, it should be noted that all data obtained in the embodiments of the present invention are accessed, collected, stored, and used for subsequent analysis and processing after the data owner has been clearly informed of the content of the data collection, the purpose of the data, the processing method, and other information, and with the consent and authorization of the data owner.

[0053] The following description, in conjunction with the accompanying drawings, details a specific scheme for a multi-dimensional credit evaluation method for a capacity trading platform provided by this invention.

[0054] Please see Figure 1 This illustrates a multi-dimensional credit evaluation method for a capacity trading platform provided by an embodiment of the present invention, comprising:

[0055] Step S110: Obtain the user-level capacity dataset of each user connected to the capacity trading platform; wherein, the user-level capacity dataset includes multi-dimensional capacity data of all vehicles belonging to the same user.

[0056] The aforementioned transportation capacity trading platform can refer to a digital trading platform for road freight built on internet technology. It connects multiple entities, including shippers, carriers, individual drivers, and third-party service providers, through cloud servers, data interfaces, and vehicle terminals. This centralized platform can integrate dispersed resources such as cargo, vehicles, and drivers, forming a dynamically dispatchable transportation capacity pool. Simultaneously, the trading platform supports online bidding, contract signing, and payment settlement, replacing the traditional offline matching model and enabling data sharing across the industry chain (such as integration with enterprise ERP systems), thus improving supply chain visibility. Furthermore, the transportation capacity trading platform possesses unified data standards and communication protocols, continuously receiving and storing multi-dimensional transportation capacity data transmitted from vehicles, such as location, mileage, speed, driving time, and cargo damage rate. This forms a transportation capacity data pool that can be accessed in real-time by the credit rating system, enabling the output of user-level transportation capacity datasets for subsequent multi-dimensional sample space mapping and credit score calculation.

[0057] Preferably, in one embodiment of the present invention, the above step S110 of obtaining the user-level transportation capacity dataset may include: collecting multi-dimensional transportation capacity data in real time during the vehicle's operation through positioning and sensing devices installed on the vehicle; wherein, the multi-dimensional transportation capacity data includes mileage, driving time, average speed, single trip driving time, and cargo damage rate; aggregating the multi-dimensional transportation capacity data according to the vehicle number to form a single-vehicle transportation capacity dataset; and summarizing the single-vehicle transportation capacity datasets of all vehicles belonging to the same user to obtain the user-level transportation capacity dataset.

[0058] It should be noted that the aforementioned user-level capacity dataset is a standardized data unit formed by the capacity trading platform after aggregating data from all vehicles belonging to the same user. Its data source is the raw signals continuously transmitted by vehicle positioning and multi-sensor devices during operation. The capacity trading platform can capture these signals in real time through a unified data access gateway and parse them into structured fields such as mileage, driving time, average speed, single-trip travel time, and cargo damage rate based on the built-in capacity data dictionary. The parsed data is first aggregated using the vehicle number as a unique identifier, completing timeliness verification and missing data completion at the vehicle level to form a single-vehicle capacity dataset. Subsequently, based on the user-vehicle binding relationship table, the platform aligns and merges all single-vehicle datasets under the same user ID in the time-series dimension, and outputs them as a user-level capacity dataset that can be directly accessed for subsequent multi-dimensional mapping and clustering calculations after formatting and encapsulation. In the storage layer, the user-level capacity dataset can use the user as the primary key, the vehicle as the shard, and time as the index, supporting incremental updates and second-level reads to ensure that the credit rating system can obtain the latest, complete, and structurally consistent capacity data in real time.

[0059] Step S120: Map the user-level capacity dataset to a multi-dimensional sample space to obtain vehicle data points corresponding to the vehicles.

[0060] It should be noted that the above-mentioned multidimensional sample space is a high-dimensional geometric space constructed with key capacity indicators as coordinate axes. Each dimension corresponds to a quantifiable capacity parameter (such as mileage, driving time, average speed, single trip time, cargo damage rate, etc.). The direction of increase of the capacity data monitoring value can be taken as the positive direction of the data axis, and the number of types of capacity data can be used as the dimension of the data axis to establish the multidimensional sample space.

[0061] It should be further explained that the scheme for mapping user-level capacity datasets to a multidimensional sample space can be as follows: First, the user-level capacity dataset is standardized to eliminate differences in the dimensions of various indicators. Then, taking each vehicle as a unit, the standardized multidimensional parameter values ​​are sequentially written into the corresponding coordinate axes to generate a unique coordinate point, i.e., the vehicle data point corresponding to that vehicle. Furthermore, in the multidimensional sample space, vehicle data points belonging to the current user can be labeled as belonging to the same class, thereby comparing changes in capacity data among different users based on user category labels. The relative distances of all vehicle data points in space directly reflect the similarity of their transportation behavior characteristics, providing a measurable geometric basis for subsequent clustering, subclassing, and credit scoring.

[0062] Step S130: In the multidimensional sample space, cluster the vehicle data points to obtain at least one vehicle subclass, and determine the integrity index of each vehicle subclass; wherein, the integrity index is used to characterize the tightness of vehicle data points within the vehicle subclass.

[0063] It should be noted that, to quantify the stability of user transport capacity behavior, the distribution of vehicle data points in a multi-dimensional sample space can be used as the evaluation object. When vehicles maintain compliant operating parameters over a long period (e.g., a cargo damage rate of 1%-5% is considered compliant), their corresponding data points will concentrate within the allowable ranges of each dimension, forming clusters with spatial proximity and small coverage areas. Conversely, if operating parameters fluctuate drastically or anomalies occur, the clusters will appear discrete or shifted. By splitting these clusters using clustering methods, a subset of vehicles with similar transport characteristics can be identified, forming vehicle subclasses. Then, by monitoring subsequent data points on integrity indicators at the subclass level, precise location of abnormal behavior and refined updates to credit scores can be achieved.

[0064] Please see Figure 2 Preferably, in one embodiment of the present invention, the above step S130 of clustering vehicle data points may include:

[0065] Step S131: Starting from any vehicle data point, sequentially retrieve the unclassified vehicle data points with the smallest geometric distance from the starting point in the multidimensional sample space, and temporarily merge the retrieved vehicle data points into the same subclass to be formed.

[0066] Step S132: For each new vehicle data point incorporated, calculate the integrity index of the subclass to be formed;

[0067] Step S133: When the integrity index after incorporating a new vehicle data point is less than the previous integrity index, stop the search and determine the vehicle data points that have been incorporated into the subclass to be formed as vehicle subclasses; wherein, the previous integrity index is the integrity index before incorporating a new vehicle data point.

[0068] It should be noted that the above scheme can employ an incremental nearest neighbor clustering strategy. A vehicle data point is randomly selected in the multidimensional sample space as a seed, and its nearest neighbor points that have not yet been classified are retrieved in ascending geometric distance order and temporarily incorporated into the current subclass to be formed. Each time a point is added, the integrity index of that subclass is recalculated. When the index value decreases compared to the previous state, the expansion is immediately terminated, and the accumulated point set is confirmed as a vehicle subclass. For each vehicle subclass obtained, the vehicle data points corresponding to that subclass are removed, and the above process is repeated among the remaining vehicle data points until all data points are classified. This achieves adaptive clustering constrained by the integrity index, ensuring high density within the resulting subclasses and providing a consistent and stable benchmark unit for subsequent anomaly detection and credit scoring.

[0069] It should be further explained that the aforementioned integrity index is designed so that the larger the index value, the more compact and representative the single scheduling quantity within the subclass. When the index decreases after a new data point is added, it indicates that the addition of this point has made the spatial distribution of the subclass tend to be discrete, exceeding the acceptable compactness threshold; if it continues to expand, it will introduce heterogeneous behavior, weakening the statistical significance of the subclass as a similar capacity unit. Therefore, using the decrease in the index as a stopping condition ensures that the data points at the truncation boundary maintain the highest internal consistency, thereby outputting the vehicle subclass with the optimal compactness and the most sensitive subsequent anomaly detection.

[0070] Preferably, in one embodiment of the present invention, the above-mentioned calculation of the integrity index of the subclass to be formed may include: determining the subclass density gain component and the subclass data coverage component of the subclass to be formed; wherein, the subclass density gain component is used to characterize the degree to which the new vehicle data points make the spatial distribution within the subclass to be formed tend to be dense, and the subclass data coverage component is used to characterize the proportion of the number of data points already contained in the subclass to be formed to the total number of data points of the corresponding user; and determining the integrity index of the subclass to be formed based on the subclass density gain component and the subclass data coverage component.

[0071] It should be noted that the above-mentioned subclass compactness gain component can quantify the degree to which newly incorporated vehicle data points improve the spatial consistency within a subclass: the larger this component, the closer the new points are to the existing core area of ​​the subclass, the more compact the overall distribution of the subclass, and the lower the risk of deviation; conversely, it indicates that the new points introduce a significant dispersion trend, which is not conducive to maintaining a highly cohesive subclass structure. The above-mentioned subclass data coverage component can measure the proportion of the number of vehicle data points already incorporated into the current subclass to be formed to the total number of data points of the corresponding user: the higher the proportion, the more representative the subclass is of the user's main capacity behavior pattern, and the greater the weight of its statistical characteristics in subsequent anomaly detection and credit assessment; a low proportion means that the subclass only covers a local sample, and its completeness and representativeness are insufficient.

[0072] Preferably, in one embodiment of the present invention, the determination of the subclass compactness gain component of the subclass to be formed may include: determining the change in spatial compactness of the subclass to be formed before and after incorporating new vehicle data points; wherein the change in spatial compactness is used to characterize the change in the compactness of the subclass to be formed caused by incorporating new vehicle data points; determining the dispersion of vehicle data points in the subclass to be formed before incorporating new vehicle data points; and determining the subclass compactness gain component of the subclass to be formed based on the change in spatial compactness and the dispersion.

[0073] It should be noted that the aforementioned changes in spatial density reflect the difference in compactness within a subclass before and after the addition of a new data point. A smaller change indicates that the new point is located near the core region of the subclass, making the overall distribution more cohesive; a larger change indicates that the new point is far from the existing center, leading to subclass boundary expansion and decreased compactness. The dispersion of vehicle data points in the subclass to be formed before the addition of new data points reflects the degree of dispersion of the existing members of the subclass. Higher dispersion indicates more significant differences between members and a looser subclass structure; lower dispersion indicates more concentrated member aggregation, providing a benchmark for judging whether subsequent points will disrupt compactness.

[0074] It should be further explained that: the vehicle subcategories obtained by clustering form a high-similarity capacity set, which can instantly compare the data of newly added single vehicles with the collective behavior of the subcategories, and quickly amplify abnormal deviation signals; avoid the mixing of heterogeneous vehicles caused by single-dimensional threshold screening, ensure that the transportation behavior within the subcategories is highly consistent, and improve the sensitivity and discrimination accuracy of subsequent credit scoring to the actual operational quality.

[0075] The following is a method for calculating the integrity index of a class to be subclassed:

[0076] In the current subclass Incorporate new vehicle data points Then, the new subclass to be formed The integrity index is:

[0077]

[0078] in, For vehicle data points To the vehicle subclass that does not contain that point The distance between them; The average of the minimum distances between all vehicle data points in the subclass to be formed can be used to measure the density of the original subclasses; For subclass The standard deviation of the distance between vehicle data points can reflect the subclass. Internal dispersion; This refers to the change in spatial density; That is, the dispersion of vehicle data points in the subclass to be formed; This represents the number of data points already contained in the current subclass. This represents the total number of vehicle data points for the current user.

[0079] In addition, the above This refers to the subclass density gain component. The larger the value of this component, the more vehicle data points the current subclass includes. Subsequently, the distribution of vehicle data points in the new subclass to be formed is more dense, that is, representing vehicle data points It has a smaller impact on deviations from subclasses and a higher degree of integrity for the current subclass;

[0080] This refers to the subclass data coverage component. The larger the value of this component, the greater the proportion of the current subclass in the total number of vehicle data points for the current user, and the more fully the integrity of the current subclass is reflected.

[0081] Step S140: Determine the user anomaly resistance index based on the integrity index of each vehicle subclass; wherein, the user anomaly resistance index is used to characterize the degree of resistance to interference caused by changes in the integrity index due to newly accessed capacity data.

[0082] It should be noted that, to identify significant changes in user capacity behavior, their resilience to abnormal operating parameters can be quantified. When vehicle anomalies cause a shift in sub-class distribution, the overall capacity scale and sub-class stability determine the impact of this shift on the credit score. Calculating a user's anomaly resilience index can assess their ability to maintain their credit level under external disturbances, preventing occasional anomalies from lowering the score and ensuring that subsequent updates match the behavioral risk. Therefore, this invention provides the following solution:

[0083] Preferably, in one embodiment of the present invention, step S140 may include: determining a subclass stability component and a subclass size normalization component of a vehicle subclass based on the integrity index of each vehicle subclass; wherein, the subclass stability component is used to characterize the relative fluctuation of the integrity index among the various vehicle subclasses corresponding to the user; the subclass size normalization component is used to characterize the proportion of the number of vehicle subclasses corresponding to the user in the total number of vehicle subclasses on the capacity trading platform; and determining the user's anomaly resistance index based on the subclass stability component and the subclass size normalization component.

[0084] It should be noted that the above-mentioned subclass stability component can characterize the relative dispersion of the integrity index across all vehicle subclasses belonging to the same user. A smaller component indicates that the integrity indices of each subclass are similar, the overall distribution is stable, and the user's operational behavior is highly consistent; conversely, a larger component means significant differences between subclasses, larger fluctuations in the user's operational status, and poor stability. The above-mentioned subclass size normalization component reflects the proportion of the number of vehicle subclasses owned by the user relative to the total number of all subclasses on the platform. A higher proportion indicates a larger user operational capacity, a stronger ability to dilute individual anomalies to the platform's statistical characteristics, and a correspondingly improved resistance to interference; a low proportion indicates a limited scale, and anomaly shocks are easily amplified.

[0085] Preferably, in one embodiment of the present invention, the determination of the subclass stability component of a vehicle subclass may include: for each vehicle subclass corresponding to a user, calculating the minimum difference between the integrity index of the current subclass and the integrity index of other subclasses to determine the minimum integrity deviation corresponding to each vehicle subclass; summing the minimum integrity deviations to determine the total user-level integrity deviation; determining the mean integrity index of the vehicle subclass containing only one vehicle data point, and determining a baseline stability value based on the mean integrity index; and determining the subclass stability component of the vehicle subclass based on the total user-level integrity deviation and the baseline stability value. This implementation example is as follows:

[0086] Calculate the current user All subclasses User abnormal resistance index between :

[0087]

[0088] in, This represents the number of vehicle subclasses currently owned by the user. For vehicle subclass Integrity indicators; For the vehicle subclass The vehicle subclass with the closest integrity index Integrity indicators; For a vehicle subclass containing only one vehicle data point Integrity indicators; This is for the operation of taking the average; This represents the total number of vehicle subclasses in the multidimensional sample space.

[0089] That is, the vehicle subclass The subclass stability component; the smaller the value of this component, the stronger the stability of the current user. The average stability of the vehicle subclass is higher, and the transportation behavior of each vehicle in the subclass is more controllable, thus demonstrating the current user's... When a vehicle transport anomaly occurs, the vehicle's subclass will be determined. The impact is smaller, thus reflecting the current user The stronger the abnormal resistance; This is the baseline stable value;

[0090] This refers to the normalized component of the subclass size. The larger the value of this component, the stronger the current user's size. The overall vehicle fleet is large, thus exhibiting a few anomalies for the current user. The overall operational stability is minimally affected.

[0091] Step S150: Monitor newly added capacity data and determine the change in the user's credit score based on the user's anomaly resistance index and the newly added capacity data.

[0092] Preferably, in one embodiment of the present invention, step S150, which determines the change in a user's credit score based on the user's anomaly resistance index and newly accessed capacity data, includes: determining an integrity change coefficient based on the integrity index of vehicle subcategories before and after accessing the newly accessed capacity data; wherein the integrity change coefficient is used to characterize the direction of fluctuation in the integrity index caused by the newly accessed capacity data, a positive integrity change coefficient indicates that the newly accessed capacity data leads to an improvement in the integrity index, and a negative integrity change coefficient indicates that the newly accessed capacity data leads to a decrease in the integrity index; determining an anomaly resistance correction coefficient based on the user's anomaly resistance index before and after accessing the newly accessed capacity data; and determining the change in a user's credit score based on the integrity change coefficient, the anomaly resistance correction coefficient, and the user's anomaly resistance index after accessing the newly accessed capacity data.

[0093] Preferably, in one embodiment of the present invention, the method for determining the anomaly resistance correction coefficient includes: calculating the ratio of the user anomaly resistance capability index after accessing the new access capacity data to the user anomaly resistance capability index before accessing the new access capacity data, and determining the anomaly resistance correction coefficient.

[0094] For example, the implementation of the above scheme can quantify the impact of abnormal behavior on credit scores based on user anomaly resistance indicators and stability fluctuations caused by newly added capacity data, thereby determining high-reliability user credit scores. Specifically:

[0095] In the current user The vehicles belonging to When an abnormal attribute occurs, it manifests as the completeness of the current subclass. The decline means that after the new round of capacity updates, the impact on individual subcategories will be affected. Calculate its updated completeness And then through Determine the direction of integrity fluctuations and assess the subclass attribute anomalies caused by user vehicle anomalies;

[0096] Calculate the current user Changes in credit score:

[0097]

[0098] in, and These are the user anomaly resistance indicators before and after the integration of new capacity data; For vehicle subclasses before accessing new transport capacity data Integrity indicators; For vehicle subclasses after integrating newly added transport capacity data Integrity indicators;

[0099] This is the integrity variation coefficient. This is the anomaly resistance correction coefficient.

[0100] It should be noted that the above-mentioned anomaly resistance correction coefficient has the following characteristics:

[0101] (1) When the user's abnormal resistance index increases positively, that is At that time, the abnormal resistance correction coefficient is amplified at a square rate, and the positive gain of the integrity index is exponentially enhanced, forming an acceleration excitation.

[0102] (2) When resistance decreases negatively, that is At that time, the abnormal resistance correction coefficient decays at a square rate, and the negative impact of the integrity index is weakened exponentially, generating buffer protection.

[0103] (3) When the resistance remains unchanged, that is At that time, the abnormal resistance correction coefficient degenerates into linearity, and the credit score adjustment range remains at the benchmark level.

[0104] These characteristics enable the sensitivity of credit scores to external disturbances to change dynamically with the user's resistance ability, achieving nonlinear adaptive adjustment of high resistance users' high scores to accumulate rapidly and low resistance users' abnormal shocks to be suppressed, reducing the drastic fluctuations in scores caused by occasional events, and improving the robustness and discrimination of the evaluation system.

[0105] It should be further noted that, in addition to square-weighted methods, the factor representing the change in resistance can also be introduced into a higher-order function mapping, so that the weights change proportionally with the increase or decrease in resistance. The higher-order curve changes. Higher-order characteristics cause the positive incentives to increase at a faster slope when resistance increases, significantly accelerating the accumulation of high scores; when resistance decreases, the steep drop in the curve rapidly weakens the deduction caused by anomalies, forming a stronger non-linear buffer. This mechanism maintains a high sensitivity to real trends while further compressing the propagation space of random disturbances, achieving a flexible adjustment that accelerates the growth of high-resistance individuals and provides immediate protection for low-resistance individuals, thus enhancing the dynamic adaptability and risk isolation capabilities of the credit rating system.

[0106] Step S160: Update the user's credit score using the change in credit score.

[0107] It should be noted that after obtaining the changes in credit score, the user's credit score can be updated. Specifically:

[0108] Record the user's initial credit score as It can be set according to real-time scoring requirements, such as... , etc., the change in credit score at each time point. The sum of the total score and the score is used as the real-time updated score. :

[0109]

[0110] in, For the initial credit score; For a moment A definite change in credit score; For a moment The updated user credit score has been confirmed.

[0111] Preferably, in one embodiment of the present invention, the multi-dimensional credit evaluation method of the above-mentioned capacity trading platform may further include: canceling the user's trading qualification in the capacity trading platform when the user's credit score is lower than a preset credit score threshold.

[0112] It should be noted that the capacity trading platform can continuously compare real-time credit scores with preset transaction qualification thresholds. When a user's credit score falls below the threshold, the platform automatically triggers a qualification circuit breaker mechanism, immediately suspending their bidding, order acceptance, and contract signing permissions, and sending a rectification notice. Only after credit repair is completed and the score rises back above the threshold will the transaction permissions be restored, thereby realizing dynamic access and exit management based on credit.

[0113] It should be further noted that the aforementioned user credit scores can also be used in application scenarios such as intelligent order dispatch, capacity scheduling, and risk warning within the capacity trading platform. The platform can weight the order matching algorithm based on user credit scores, prioritizing vehicles from high-credit users in the recommendation list to improve matching success rates and fulfillment reliability. In the dynamic scheduling phase, the score serves as a ranking factor in the capacity pool, enabling high-credit vehicles to obtain better route planning and loading / unloading time windows, reducing waiting and empty-running rates. Simultaneously, the platform automatically upgrades the monitoring level for users whose scores decline, triggering enhanced risk control strategies such as encrypted data transmission, trajectory verification, and abnormal parking alarms, achieving differentiated operations and precise resource allocation.

[0114] This invention is now complete.

[0115] In summary, in this embodiment of the invention, the following steps are taken: First, a user-level capacity dataset of each user connected to the capacity trading platform is obtained. This dataset is then mapped to a multi-dimensional sample space to obtain vehicle data points corresponding to each vehicle. In the multi-dimensional sample space, the vehicle data points are clustered to obtain at least one vehicle subclass, and the integrity index of each vehicle subclass is determined. Based on the integrity index of each vehicle subclass, a user anomaly resistance index is determined. Second, newly connected capacity data is monitored, and based on the user anomaly resistance index and the newly connected capacity data, the change in the user's credit score is determined. Finally, the user's credit score is updated using the change in credit score. This invention achieves real-time updates to user credit scores by continuously monitoring newly added capacity data and dynamically updating credit score changes using integrity indicators before and after data integration, as well as user anomaly resistance indicators. Furthermore, it maps user-level capacity datasets to a multi-dimensional sample space and performs clustering, quantifying the similarity of vehicle behavior at a geometric level to form highly cohesive vehicle subclasses. This reduces the heterogeneous vehicle mixing rate caused by single-dimensional threshold screening, improving the sensitivity and accuracy of anomaly detection. Additionally, based on the subclass integrity indicator, the user anomaly resistance indicator embeds the relative change in anomaly resistance into an anomaly resistance correction coefficient in a proportional form, making the credit score change non-linearly adaptive to external disturbances, weakening score cliffs caused by occasional anomalies, and enhancing the robustness of the evaluation system. Finally, the credit score change is driven by both the integrity change coefficient and the anomaly resistance correction coefficient, enabling rapid updates to the credit score. This allows the capacity trading platform to dynamically adjust transaction qualifications, risk control levels, and scheduling priorities based on the updated credit score, forming a data-scoring-control closed loop, reducing manual intervention, and improving operational efficiency.

[0116] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-dimensional credit evaluation method for a transportation capacity trading platform, characterized in that, The method includes: Obtain user-level capacity datasets for each user connected to the capacity trading platform; wherein, the user-level capacity datasets include multi-dimensional capacity data for all vehicles belonging to the same user; The user-level capacity dataset is mapped to a multi-dimensional sample space to obtain vehicle data points corresponding to the vehicles. In the multidimensional sample space, the vehicle data points are clustered to obtain at least one vehicle subclass, and an integrity index for each vehicle subclass is determined; wherein, the integrity index is used to characterize the density of the vehicle data points within the vehicle subclass. Based on the integrity index of each vehicle subclass, a user anomaly resistance index is determined; wherein, the user anomaly resistance index is used to characterize the degree of resistance to interference caused by changes in the integrity index due to newly accessed capacity data. Monitor the newly accessed capacity data, and determine the change in the user's credit score based on the user's anomaly resistance index and the newly accessed capacity data; The user's credit score is updated using the change in credit score. In the current subclass Incorporate new vehicle data points Then, the new subclass to be formed The integrity index is: in, For vehicle data points For vehicle subclasses that do not contain this point The distance between them; This is the average of the minimum distances between all vehicle data points in the subclass to be formed; For subclass Standard deviation of the distance between vehicle data points; This refers to the change in spatial density; That is, the dispersion of vehicle data points in the subclass to be formed; This represents the number of data points already contained in the current subclass. This represents the total number of vehicle data points for the current user. For subclass compactness gain components; Cover components for subclass data; Calculate the current user All subclasses User abnormal resistance index between : in, This represents the number of vehicle subclasses currently owned by the user. For vehicle subclass Integrity indicators; For the vehicle subclass The vehicle subclass with the closest integrity index Integrity indicators; For a vehicle subclass containing only one vehicle data point Integrity indicators; This is for the operation of taking the average; This represents the total number of vehicle subclasses in the multidimensional sample space. For vehicle subclass Subclass stability components; For subclass size normalization components; Calculate the current user Changes in credit score: in, and These are the user anomaly resistance indicators before and after the new access capacity data was introduced; Vehicle subclasses before accessing new transport capacity data Integrity indicators; For vehicle subclasses after integrating newly added transport capacity data Integrity indicators; The integrity variation coefficient, This is the abnormal resistance correction coefficient.

2. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 1, characterized in that, The clustering of the vehicle data points includes: Starting from any of the vehicle data points, sequentially retrieve the vehicle data points that have the smallest geometric distance from the starting point and have not been classified in the multidimensional sample space, and temporarily merge the retrieved vehicle data points into the same subclass to be formed. For each new vehicle data point incorporated, the integrity index of the subclass to be formed is calculated; When the integrity index after incorporating a new vehicle data point is less than the previous integrity index, the search stops, and the vehicle data points that have been incorporated into the subclass to be formed are identified as vehicle subclasses; wherein, the previous integrity index is the integrity index before incorporating a new vehicle data point.

3. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 2, characterized in that, The calculation of the integrity index of the subclass to be formed includes: Determine the subclass compactness gain component and the subclass data coverage component of the subclass to be formed; wherein, the subclass compactness gain component is used to characterize the degree to which the new vehicle data points make the spatial distribution within the subclass to be formed tend to be compact, and the subclass data coverage component is used to characterize the proportion of the number of data points already contained in the subclass to be formed to the total number of data points of the corresponding user. The integrity index of the subclass to be formed is determined based on the subclass density gain component and the subclass data coverage component.

4. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 3, characterized in that, Determining the subclass compactness gain component of the subclass to be formed includes: Determine the change in spatial density of the subclass to be formed before and after incorporating the new vehicle data point; wherein the change in spatial density is used to characterize the change in the density of the subclass to be formed caused by incorporating the new vehicle data point. Determine the dispersion of the vehicle data points in the subclass to be formed before the new vehicle data points are incorporated. Based on the spatial density change and the dispersion, the subclass density gain component of the subclass to be formed is determined.

5. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 1, characterized in that, The determination of user anomaly resistance indicators based on the integrity indicators of each of the vehicle subclasses includes: Based on the integrity index of each vehicle subclass, a subclass stability component and a subclass size normalization component are determined for each vehicle subclass; wherein, the subclass stability component is used to characterize the relative fluctuation of the integrity index among the vehicle subclasses corresponding to the user; and the subclass size normalization component is used to characterize the proportion of the number of vehicle subclasses corresponding to the user in the total number of vehicle subclasses in the capacity trading platform. Based on the subclass stability component and the subclass size normalization component, the user's anomaly resistance index is determined.

6. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 5, characterized in that, Determining the subclass stability component of the vehicle subclass includes: For each vehicle subclass corresponding to the user, the minimum difference between the integrity index of the current subclass and the integrity index of other subclasses is calculated one by one to determine the minimum integrity deviation corresponding to each vehicle subclass. The minimum integrity deviations are summed to determine the total user-level integrity deviations; Determine the mean integrity index of the vehicle subclass containing only one vehicle data point, and determine a baseline stable value based on the mean integrity index; Based on the sum of the user-level integrity deviations and the baseline stability value, the subclass stability component of the vehicle subclass is determined.

7. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 1, characterized in that, The step of determining the change in the user's credit score based on the user's anomaly resistance index and the newly accessed capacity data includes: Based on the integrity index of the vehicle subclass before and after accessing the newly accessed capacity data, an integrity change coefficient is determined; wherein, the integrity change coefficient is used to characterize the direction of fluctuation of the integrity index caused by the newly accessed capacity data, a positive integrity change coefficient indicates that the newly accessed capacity data leads to an improvement in the integrity index, and a negative integrity change coefficient indicates that the newly accessed capacity data leads to a decrease in the integrity index; Based on the user anomaly resistance capability index before and after accessing the newly accessed capacity data, an anomaly resistance correction coefficient is determined. Based on the integrity change coefficient, the anomaly resistance correction coefficient, and the user anomaly resistance capability index after accessing the newly accessed capacity data, the change in the user's credit score is determined.

8. The multi-dimensional credit evaluation method for a capacity trading platform according to claim 7, characterized in that, The methods for determining the anomaly resistance correction coefficient include: Calculate the ratio of the user anomaly resistance index after accessing the new access capacity data to the user anomaly resistance index before accessing the new access capacity data, and determine the anomaly resistance correction coefficient.

9. The multi-dimensional credit evaluation method for a capacity trading platform according to any one of claims 1-8, characterized in that, Obtaining the user-level capacity dataset includes: The multi-dimensional transportation capacity data is collected in real time during the vehicle's operation using positioning and sensing devices installed on the vehicle; wherein the multi-dimensional transportation capacity data includes mileage, driving time, average speed, single trip time, and cargo damage rate. The multi-dimensional transport capacity data is collected according to the vehicle number to form a single-vehicle transport capacity dataset; The user-level capacity dataset is obtained by aggregating the single-vehicle capacity datasets of all vehicles belonging to the same user.

10. The multi-dimensional credit evaluation method for a capacity trading platform according to any one of claims 1-8, characterized in that, The method further includes: When a user's credit score falls below a preset credit score threshold, the user's trading qualification on the transportation capacity trading platform is revoked.

Images