A method for quantitatively verifying a declared address based on a GPS track of a vehicle in a loan process and a computer device

Abstract

The application provides a kind of based on the address quantification verification method and computer equipment of reporting of loaning vehicle GPS track, collect the GPS track coordinate sequence of target vehicle in the preset time window, and the address information reported by borrower;The address information reported by borrower is longitude and latitude analysis, and the longitude and latitude coordinates of the address reported are generated;The GPS track coordinate sequence is stopped point identification, and the list of events of stopping in loaning is generated;The address information reported by borrower is matched with the event of stopping, and the target event of stopping in the preset distance threshold is identified with the address reported;Based on the matching result of target event of stopping, the geographical distance score, the frequency score of stopping, the duration score of stopping and the time period characteristic score are calculated respectively, and the consistency score is obtained by the weighted sum through the preset weight calculation;According to the comparison result of consistency score and preset threshold, the corresponding risk level and suggestion are output.The application strengthens the verification strength of the authenticity of the information reported by borrower.

Description

Technical Field

[0001] This invention belongs to the field of vehicle location recognition technology. Specifically, this invention relates to a method and computer equipment for quantitative verification of declared addresses based on vehicle-mounted GPS trajectories. Background Technology

[0002] In auto finance, borrowers are required to declare their residential and work addresses when applying for a loan. Financial institutions use this information as a crucial basis for pre-loan approval, and also as a fundamental reference for loan management during and after the loan disbursement. However, verifying the authenticity of these declared addresses, especially in ongoing monitoring after loan disbursement to confirm whether the vehicle's actual parking location matches the declared address, remains a significant technical challenge in the field of auto finance risk control.

[0003] Chinese Patent 102236871A provides a loan management method integrating a mobile communication system, comprising the following steps: providing a mobile phone application financing module to transmit a mobile phone signature for identity verification; providing a loan processing module to access credit information; providing a mobile phone credit module to confirm company and residential locations using a Global Positioning System (GPS); providing a credit management module to determine a loan amount and interest rate; and providing an account management module for disbursement.

[0004] Existing technologies identify a borrower's frequent locations by collecting GPS / WiFi / cell tower location data from their mobile phone. However, in the context of auto finance, location tracking via the borrower's mobile phone cannot accurately reflect the security status of the collateral. Summary of the Invention

[0005] The present invention aims to provide a method and computer equipment for quantitative verification of the declared address based on the GPS trajectory of the vehicle during the loan process, so as to enhance the technical purpose of verifying the authenticity of the information declared by the borrower.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] This invention provides a method for quantitative verification of declared addresses based on vehicle GPS trajectories during loan processing, comprising the following steps:

[0008] Step 1: Collect the GPS trajectory coordinate sequence of the target vehicle within a preset time window, as well as the address information declared by the borrower corresponding to the target vehicle;

[0009] Step 2: Analyze the latitude and longitude of the address information declared by the borrower to generate the latitude and longitude coordinates of the declared address;

[0010] Step 3: Identify stop points in the GPS trajectory coordinate sequence and generate a list of stop events; each stop event in the list includes the center coordinates of the stop, the start time of the stop, the end time of the stop, and the duration of the stop;

[0011] Step 4: Spatial match the latitude and longitude coordinates of the address information declared by the borrower with the stay events in the loan stay event list to identify target stay events that are within a preset distance threshold from the declared address;

[0012] Step 5: Based on the matching results of the target stay events, calculate the geographical distance score, stay frequency score, stay duration score, and time period feature score respectively, and calculate the weighted sum through preset weights to obtain the consistency score;

[0013] Step 6: Based on the comparison results between the consistency score and the preset threshold, output the corresponding risk level and recommendations.

[0014] In step two, the declared address text is standardized by calling the geocoding API to convert the standardized address into latitude and longitude coordinates.

[0015] In step three, before identifying the stop points in the GPS trajectory coordinate sequence, the GPS trajectory coordinate sequence is preprocessed, including removing outliers and sorting the GPS trajectory coordinate sequence by timestamp.

[0016] In step three, the Havesing spherical distance between adjacent GPS track coordinates is calculated. When the Havesing spherical distance is greater than or equal to a preset distance threshold, it is determined whether the dwell time of the target vehicle is greater than a preset time threshold. If the dwell time of the target vehicle is greater than the preset time threshold, the center coordinates of the dwell are determined and the dwell event is recorded. All GPS track coordinates are traversed, and a list of dwell events is output.

[0017] In step five, the geographical distance score uses the Havesing formula to calculate the distance between the latitude and longitude coordinates of the address declared by the borrower and the center coordinates of the target stay event, and calculates the score according to the preset distance-score mapping function.

[0018] In step five, the dwell frequency scoring method is as follows: within the preset time window of the loan, the ratio of the number of target dwell events to the preset expected number is calculated; the preset expected number is set to the number of days in the preset time window of the loan.

[0019] In step five, the dwell time scoring method is as follows: calculate the ratio of the average dwell time of the target dwell event to the preset expected dwell time; the expected dwell time is set according to the type of address declared by the borrower.

[0020] In step five, the time period feature scoring method is as follows: calculate the ratio of the number of stay events that meet the time period matching conditions to the total number of target stay events.

[0021] In step six, when the consistency score is lower than the preset threshold, a list of risk factors is output. The risk factors include at least one of the following: excessive deviation of the declared address, low frequency of stay, insufficient stay duration, and mismatch of time period characteristics.

[0022] The present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0023] The technical effects of this invention are as follows:

[0024] (1) The data source of this invention is accurate and has strong anti-interference ability. The core positioning data comes from the vehicle's dedicated 4G Cat.1 vehicle GPS device and reports trajectory data in compliance with the JT808 protocol. It is completely different from the borrower's mobile terminal positioning, manually entered address, and third-party structured information such as industrial and commercial, judicial and tax information. From the data source level, it avoids problems such as human tampering, mobile device lending, virtual positioning, and lag and distortion of third-party public information.

[0025] (2) This invention provides accurate identification of dwell events. It employs a dwell point identification algorithm that uses both spatial and temporal thresholds, with a preferred spatial threshold of 50 meters and a temporal threshold of 30 minutes. This allows for precise identification of valid vehicle dwell behaviors and effectively filters out short pauses with no business significance, such as waiting at red lights, temporary parking, or brief loading / unloading during vehicle operation. Through complete trajectory traversal, dwell segment creation, and duration verification, only dwell events with risk control value, such as overnight parking and permanent parking at the workplace, are retained, thus avoiding invalid data from participating in address matching and scoring.

[0026] (3) The risk control assessment dimensions of this invention are comprehensive, and it constructs four core scoring dimensions: geographical distance, frequency of stay, duration of stay, and time period characteristics, which comprehensively cover the spatial matching degree, stability of stay, compliance of stay, and time period matching of vehicle stay. Different expected duration and matching time period standards are configured for the two core declaration addresses of residence and workplace, respectively, which are in line with the real behavioral characteristics of users' daily residence and work, and avoid misjudgment caused by uniform evaluation standards.

[0027] (4) This invention provides clear risk classification, precise loan management, and sets up a graded scoring range corresponding to low, medium and high risk levels. It also matches standardized and implementable loan management suggestions to achieve graded reuse of risk control resources, avoid a one-size-fits-all management model, and take into account both the effectiveness of risk control and the efficiency of business processing.

[0028] (5) This invention provides precise risk tracing. In high-risk scenarios, it can accurately output a list of risk factors, accurately pinpointing the core reasons for address verification anomalies, including specific issues such as excessive deviation in declared address distance, low frequency of stay, insufficient stay duration, and mismatch of time period characteristics. This solves the problem that traditional risk control can only determine the existence of risk but cannot pinpoint its root cause. Staff can conduct targeted verification, investigation, and risk disposal work based on clearly defined risk factors. At the same time, it can optimize pre-loan approval rules in reverse, improving the level of refinement in auto finance risk control.

[0029] (6) This invention reduces risk control costs by replacing the traditional manual work mode of manually verifying addresses and identifying vehicle parking behavior through automated GPS trajectory analysis, intelligent scoring and classification, and graded risk handling. This significantly reduces the time, labor, and verification errors of manual risk control. At the same time, through continuous dynamic address verification during the loan process, it enables full-cycle risk control monitoring after loan disbursement, and timely detection of potential risks such as inaccurate information declared by borrowers and changes in actual residential and work addresses. Attached Figure Description

[0030] This manual includes the following figures, which illustrate the following:

[0031] Figure 1 This is a flowchart of a method and computer device for quantitative verification of declared addresses based on vehicle-mounted GPS trajectories in loan applications, according to the present invention. Detailed Implementation

[0032] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, in order to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention, and to facilitate its implementation.

[0033] This invention provides a method for quantitative verification of declared addresses based on vehicle GPS trajectories during loan processing, comprising the following steps:

[0034] Step 1: Collect the GPS trajectory coordinate sequence of the target vehicle within a preset time window, as well as the address information declared by the borrower corresponding to the target vehicle;

[0035] Step 2: Analyze the latitude and longitude of the address information declared by the borrower to generate the latitude and longitude coordinates of the declared address;

[0036] Step 3: Identify stop points in the GPS trajectory coordinate sequence and generate a list of stop events; each stop event in the list includes the center coordinates of the stop, the start time of the stop, the end time of the stop, and the duration of the stop;

[0037] Step 4: Spatial match the latitude and longitude coordinates of the address information declared by the borrower with the stay events in the loan stay event list to identify target stay events that are within a preset distance threshold from the declared address;

[0038] Step 5: Based on the matching results of the target stay events, calculate the geographical distance score, stay frequency score, stay duration score, and time period feature score respectively, and calculate the weighted sum through preset weights to obtain the consistency score;

[0039] Step 6: Based on the comparison results between the consistency score and the preset threshold, output the corresponding risk level and recommendations.

[0040] In step two, the declared address text is standardized by calling the geocoding API to convert the standardized address into latitude and longitude coordinates.

[0041] In step three, before identifying the stop points in the GPS trajectory coordinate sequence, the GPS trajectory coordinate sequence is preprocessed, including removing outliers and sorting the GPS trajectory coordinate sequence by timestamp.

[0042] In step three, the Havesing spherical distance between adjacent GPS track coordinates is calculated. When the Havesing spherical distance is greater than or equal to a preset distance threshold, it is determined whether the dwell time of the target vehicle is greater than a preset time threshold. If the dwell time of the target vehicle is greater than the preset time threshold, the center coordinates of the dwell are determined and the dwell event is recorded. All GPS track coordinates are traversed, and a list of dwell events is output.

[0043] In step five, the geographical distance score uses the Havesing formula to calculate the distance between the latitude and longitude coordinates of the address declared by the borrower and the center coordinates of the target stay event, and calculates the score according to the preset distance-score mapping function.

[0044] In step five, the dwell frequency scoring method is as follows: within the preset time window of the loan, the ratio of the number of target dwell events to the preset expected number is calculated; the preset expected number is set to the number of days in the preset time window of the loan.

[0045] In step five, the dwell time scoring method is as follows: calculate the ratio of the average dwell time of the target dwell event to the preset expected dwell time; the expected dwell time is set according to the type of address declared by the borrower.

[0046] In step five, the time period feature scoring method is as follows: calculate the ratio of the number of stay events that meet the time period matching conditions to the total number of target stay events.

[0047] In step six, when the consistency score is lower than the preset threshold, a list of risk factors is output. The risk factors include at least one of the following: excessive deviation of the declared address, low frequency of stay, insufficient stay duration, and mismatch of time period characteristics.

[0048] The present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the above-described method.

[0049] The present invention will now be described in detail.

[0050] The system collects the GPS trajectory coordinate sequence of the target vehicle within a preset time window, which is the period between loan disbursement and the borrower's default. It also collects the address information declared by the borrower corresponding to the target vehicle. The GPS trajectory coordinate sequence originates from a GPS tracking device installed on the target vehicle (such as a 4G Cat.1 vehicle-mounted GPS device, reporting location data using the JT808 protocol), which is distinct from the location data of the borrower's mobile terminal device, text address input, and structured enterprise information such as business registration, judicial, and tax information.

[0051] The address information declared by the borrower is parsed for latitude and longitude to generate the latitude and longitude coordinates of the declared address. Specifically, the declared address text is standardized and geocoding APIs (such as Gaode Map, Baidu Map, and Tencent Map) are called to convert the standardized address into latitude and longitude coordinates.

[0052] The GPS trajectory coordinate sequence is preprocessed, including outlier removal and sorting by timestamp. Stop point identification is performed on the GPS trajectory coordinate sequence to generate a list of stop events; each stop event in the list includes the center coordinates of the stop, the start time of the stop, the end time of the stop, and the duration of the stop.

[0053] The stop point identification algorithm includes the following steps:

[0054] Step S1: Input the GPS track coordinate sequence Points = [p1, p2, ..., p n ].

[0055] Step S2: Initialize the stay event list stay_events to be empty, and the current stay segment current_stay to be empty.

[0056] Step S3: Traverse the trajectory points, setting the loop variable i from 1 to n-1.

[0057] Step S4: Calculate the Havesing spherical distance (dist) between two adjacent points p_i and p_{i+1}.

[0058] Step S5: Determine if dist is less than the spatial threshold (preferably 50 meters).

[0059] Step S6: If dist < 50 meters, determine if the current stop segment is empty; if it is empty, create a new stop segment and record the start time, then add p_{i+1} to the current stop segment.

[0060] Step S7: If dist ≥ 50 meters, determine whether the current stop segment is not empty; if not empty, calculate the stop duration.

[0061] Step S8: Determine whether the dwell time is greater than or equal to the time threshold (preferably 1800 seconds, i.e. 30 minutes).

[0062] Step S9: If the stay duration is ≥30 minutes, calculate the stay center point (take the average latitude and longitude of all points in the segment), record the stay event (including the coordinates of the stay center point, start time, end time, and duration), and add it to the stay event list.

[0063] Step S10: Reset the current pause segment to empty.

[0064] Step S11: Increment the loop variable i by 1 and continue iterating.

[0065] Step S12: After the traversal is complete, output the list of stay events: Stays = [s1, s2, ..., s m ].

[0066] The algorithm effectively eliminates short pauses during vehicle travel (such as waiting at red lights or temporary stops) by using both spatial and temporal thresholds, retaining only stop events with business significance (such as overnight parking or parking at the workplace).

[0067] The conditions for identifying stop points include:

[0068] Spatial conditions: The spatial distance between consecutive GPS track points is greater than or equal to the first threshold (preferably 50 meters).

[0069] Time condition: The time span of continuous trajectory points is greater than the second threshold (preferably 1800 seconds, i.e. 30 minutes).

[0070] The system spatially matches the latitude and longitude coordinates of the borrower's declared address with the stay events in the loan stay event list to identify target stay events within a preset distance threshold of the declared address. Spatial distance is calculated using the Haversine formula to calculate spherical distance:

[0071]

[0072] in The spherical distance between the address declared by the borrower and the center point of residence; The radius of the Earth is taken as an average of 6371 km. The latitude and longitude of the address declared by the borrower (which needs to be converted to radians); The latitude of the center point of the stop (must be converted to radians); The longitude of the address declared by the borrower (which needs to be converted to radians); The longitude of the center point (which needs to be converted to radians). The radian value is calculated as: (x*π) / 180, where x is the longitude or latitude value.

[0073] Based on the matching results of the target stay events, the geographical distance score, stay frequency score, stay duration score and time period feature score are calculated respectively, and the consistency score is obtained by weighted summation through preset weights.

[0074] The geographical distance score uses the Havesing formula to calculate the distance between the latitude and longitude coordinates of the address declared by the borrower and the center coordinates of the target stay event, and calculates the score according to the preset distance-score mapping function.

[0075] ≤300 meters 1.0 Completely trustworthy 300-3000 meters 1.0 - (distance - 300) / 100 × 0.01 linearly decreasing >3000 meters 0 Completely unreliable

[0076] The dwell frequency scoring method is as follows: within a preset time window in the loan (default 30 days), the ratio of the number of target dwell events to the preset expected number of times is calculated; the preset expected number of times is set to the number of days in the preset time window in the loan.

[0077] The dwell time scoring method is as follows: calculate the ratio of the average dwell time of the target dwell event to the preset expected dwell time; the expected dwell time is set according to the type of address declared by the borrower.

[0078] place of residence 8 hours Normal overnight stay Workplace 9 hours Normal daytime working hours

[0079] The time period feature scoring method is as follows: calculate the ratio of the number of stay events that meet the time period matching conditions to the total number of target stay events.

[0080] place of residence Nighttime (22:00-06:00) Normal overnight stay at home Workplace Daytime (09:00-17:00) Normal working hours

[0081] In this embodiment, the weights of the geographical distance score, the frequency of stay score, the duration of stay score, and the time period characteristic score are set to 0.3, 0.25, 0.25, and 0.2, respectively.

[0082] Based on the comparison between the consistency score and the preset threshold, the corresponding risk level and recommendations are output.

[0083] 0.8 ≤ S ≤ 1.0 Low risk Verification passed, stored as baseline. 0.5 ≤ S<0.8 Medium risk Included in the random inspection list, verification via SMS / App 0 ≤ S<0.5 High risk An alert has been triggered; an on-site inspection is recommended.

[0084] When the consistency score is lower than the preset threshold (preferably 0.5), a list of risk factors is output. The risk factors include at least one of the following: excessive deviation of the declared address, low frequency of stay, insufficient stay duration, and mismatch of time period characteristics.

[0085] The beneficial effects of the present invention are described in detail below.

[0086] The data source of this invention is accurate and has strong anti-interference capabilities. The core positioning data comes from the vehicle's dedicated 4G Cat.1 vehicle GPS device and reports trajectory data in compliance with the JT808 protocol. This is completely different from the borrower's mobile terminal positioning, manually entered address, and third-party structured information such as business, judicial, and tax information. From the data source level, it avoids problems such as human tampering, mobile device lending, virtual positioning, and the lag and distortion of third-party public information.

[0087] This invention provides accurate identification of dwell events, employing a dwell point identification algorithm that uses both spatial and temporal thresholds. It optimally selects a 50-meter spatial threshold and a 30-minute temporal threshold to accurately identify valid vehicle dwell behaviors, effectively filtering out short, meaningless pauses such as waiting at red lights, temporary roadside stops, and brief loading / unloading during vehicle movement. Through complete trajectory traversal, dwell segment creation, and duration verification, only dwell events with risk control value, such as overnight parking and permanent workplace parking, are retained, preventing invalid data from participating in address matching and scoring.

[0088] This invention offers comprehensive risk control assessment dimensions, constructing four core scoring dimensions: geographical distance, frequency of stay, duration of stay, and time period characteristics. These dimensions comprehensively cover the spatial matching degree, stability, compliance, and time period matching of vehicle stays. For the two core declaration addresses—residential and workplace—differentiated expected durations and matching time period standards are configured to align with users' actual daily living and working behaviors, avoiding misjudgments caused by standardized assessment criteria.

[0089] This invention provides clear risk classification and precise loan management. It sets up a tiered scoring range corresponding to low, medium, and high risk levels, and matches it with standardized and implementable loan management suggestions. This enables the tiered reuse of risk control resources, avoids a one-size-fits-all management model, and balances the effectiveness of risk control with the efficiency of business processing.

[0090] This invention offers precise risk tracing, accurately generating a list of risk factors in high-risk scenarios. It can pinpoint the core reasons for address verification anomalies, including excessive deviations in declared address distance, low frequency of stays, insufficient stay duration, and mismatched time period characteristics. This solves the problem of traditional risk control methods that can only determine the existence of risk but cannot pinpoint its root cause. Staff can conduct targeted verification, investigation, and risk management based on clearly defined risk factors. Simultaneously, it can optimize pre-loan approval rules, improving the precision of auto finance risk control.

[0091] This invention reduces risk control costs by replacing the traditional manual work mode of manually verifying addresses and identifying vehicle parking behavior through automated GPS trajectory analysis, intelligent scoring and classification, and tiered risk handling. This significantly reduces the time, labor, and verification errors associated with manual risk control. Simultaneously, through continuous dynamic address verification during the loan process, it enables full-cycle risk control monitoring after loan disbursement, promptly identifying potential risks such as inaccurate borrower information and changes in actual residential or work addresses.

[0092] The present invention has been described above by way of example with reference to the accompanying drawings. Obviously, the specific implementation of the present invention is not limited to the above-described manner. Any non-substantial improvements made using the inventive concept and technical solution; or the direct application of the inventive concept and technical solution to other situations without modification, are all within the protection scope of the present invention.

Claims

1. A method for quantitative verification of declared addresses based on vehicle-mounted GPS trajectories during loan processing, characterized in that, Includes the following steps: Step 1: Collect the GPS trajectory coordinate sequence of the target vehicle within a preset time window, as well as the address information declared by the borrower corresponding to the target vehicle; Step 2: Analyze the latitude and longitude of the address information declared by the borrower to generate the latitude and longitude coordinates of the declared address; Step 3: Identify stop points in the GPS trajectory coordinate sequence and generate a list of stop events; each stop event in the list includes the center coordinates of the stop, the start time of the stop, the end time of the stop, and the duration of the stop; Step 4: Spatial match the latitude and longitude coordinates of the address information declared by the borrower with the stay events in the loan stay event list to identify target stay events that are within a preset distance threshold from the declared address; Step 5: Based on the matching results of the target stay events, calculate the geographical distance score, stay frequency score, stay duration score, and time period feature score respectively, and calculate the weighted sum through preset weights to obtain the consistency score; Step 6: Based on the comparison results between the consistency score and the preset threshold, output the corresponding risk level and recommendations.

2. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step two, the declared address text is standardized by calling the geocoding API to convert the standardized address into latitude and longitude coordinates.

3. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step three, before identifying the stop points in the GPS trajectory coordinate sequence, the GPS trajectory coordinate sequence is preprocessed, including removing outliers and sorting the GPS trajectory coordinate sequence by timestamp.

4. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step three, the Havesing spherical distance between adjacent GPS track coordinates is calculated. When the Havesing spherical distance is greater than or equal to a preset distance threshold, it is determined whether the dwell time of the target vehicle is greater than a preset time threshold. If the dwell time of the target vehicle is greater than the preset time threshold, the center coordinates of the dwell are determined and the dwell event is recorded. All GPS track coordinates are traversed, and a list of dwell events is output.

5. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step five, the geographical distance score uses the Havesing formula to calculate the distance between the latitude and longitude coordinates of the address declared by the borrower and the center coordinates of the target stay event, and calculates the score according to the preset distance-score mapping function.

6. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step five, the dwell frequency scoring method is as follows: within the preset time window of the loan, the ratio of the number of target dwell events to the preset expected number is calculated; the preset expected number is set to the number of days in the preset time window of the loan.

7. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step five, the dwell time scoring method is as follows: calculate the ratio of the average dwell time of the target dwell event to the preset expected dwell time; the expected dwell time is set according to the type of address declared by the borrower.

8. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step five, the time period feature scoring method is as follows: calculate the ratio of the number of stay events that meet the time period matching conditions to the total number of target stay events.

9. The method for quantitative verification of declared addresses based on vehicle GPS trajectory during loan processing, as described in claim 1, is characterized in that: In step six, when the consistency score is lower than the preset threshold, a list of risk factors is output. The risk factors include at least one of the following: excessive deviation of the declared address, low frequency of stay, insufficient stay duration, and mismatch of time period characteristics.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the method of any one of claims 1 to 9.

Images