Full-scene integrated cross-system parking recovery service component docking method

Abstract

The invention relates to the technical field of intelligent traffic and parking management, and discloses a full-scene integrated cross-system parking recovery service component docking method, which comprises the following steps of: acquiring vehicle information, parking records and payment states in a parking lot through an edge computing node or Internet of Things equipment; cleaning and standardizing the acquired data to form a structured parking and payment record; parking and payment data are stored and synchronized in real time in a distributed data storage mode; based on the parking behavior, the payment behavior and the credit score of the user, constructing a user arrearage risk prediction model, and performing risk layering on the user; and calling a dynamic rule engine to generate a corresponding arrearage recovery strategy based on the user risk level. Real-time collection and processing of parking data are realized through edge calculation, and the data transmission efficiency and accuracy are improved; a recovery strategy is dynamically generated through an intelligent risk prediction model, and the arrearage conversion rate is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation and parking management, and specifically to a method for docking parking recovery service components in a full-scenario integrated cross-system manner. Background Art

[0002] At present, parking lot management systems play an important role in urban traffic and vehicle management, especially in the parking fee collection process, which realizes the automated management of parking fees through license plate recognition, parking timer and payment systems. However, with the rapid increase in the number of parking lots and the increase in the complexity of management systems, problems such as data fragmentation between different parking lot systems, low user willingness to pay and imperfect arrears management have gradually emerged, resulting in reduced parking lot operation efficiency and deteriorating user experience. The traditional parking arrears management method mainly relies on the arrears records within a single parking lot system, and adopts manual verification or delayed arrears notification methods, which is not only inefficient, but also leads to a low arrears collection rate.

[0003] The existing parking fee collection system has basic functions such as parking record collection, fee collection information generation and user reminders, but it has many shortcomings. First, most systems only support data processing for a single parking lot or a single area, and cannot integrate fee collection records across parking lots and regions; second, the risk assessment method for fee collection is simple, based only on historical fee collection records or manual rules, making it difficult to achieve dynamic and personalized strategy adjustments; finally, user notification and payment methods are single, and payment failures or users ignoring notifications are common, which seriously affects the collection effect. These problems directly limit the collection efficiency and user satisfaction of the parking fee management system, and urgently need to be solved through intelligent and full-scenario technology upgrades. Summary of the invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a full-scenario integrated cross-system parking recovery service component docking method, which solves the problems of data dispersion, inaccurate risk prediction, single recovery strategy and inconvenient payment method in parking arrears management in the existing technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a full-scenario integrated cross-system parking recovery service component docking method, comprising the following steps: Collect vehicle information, parking records, and payment status in the parking lot through edge computing nodes or IoT devices; Clean and standardize the collected data to form structured parking and payment records; The parking and payment data are stored and synchronized in real time through distributed data storage; Based on the user's parking behavior, payment behavior and credit score, a user arrears risk prediction model is built to stratify the users by risk; Based on the user's risk level, the dynamic rule engine is called to generate corresponding overdue payment collection strategies; Notify users in arrears through multiple channels, and collect overdue payments by combining automatic deductions with alternative payment methods; Credit information sharing, parking service restrictions or legal collection for users with long-term or high-amount delinquency records; Ensure the security and compliance of user information through data encryption and privacy protection mechanisms.

[0006] Preferably, the edge computing node is used to process vehicle information in real time, including license plate recognition, parking time calculation and preliminary parking fee calculation.

[0007] Preferably, the cleaning and standardization process comprises the following steps: Remove duplicate parking records; Filter the license plate recognition data whose confidence level is lower than the set threshold; Parking records and payment records are linked by timestamp.

[0008] Preferably, the distributed data storage adopts a cache storage and historical storage hierarchical design, wherein: Cache storage is used to access current parking and payment status in real time; Historical storage is used to save long-term arrears data and support subsequent analysis.

[0009] Preferably, the arrears risk prediction model is constructed using a machine learning algorithm, combined with a time series model to predict user payment behavior, and user risk levels are stratified through a classification model.

[0010] Preferably, the dynamic rule engine dynamically generates the following recovery strategies based on user risk stratification: Send a single reminder notification to low-risk users; Increase reminder frequency for medium-risk users and provide installment payment options; For high-risk users, the overdue payment information will be shared with the credit platform and subsequent parking services will be restricted.

[0011] Preferably, the multi-channel notification method includes: SMS reminders; Mobile application push; Voice assistant notifications; Smart car networking system reminder.

[0012] Preferably, the automatic deduction method includes a pre-authorized deduction mechanism, which freezes part of the parking fee when the vehicle enters the parking lot and completes the deduction when the user exits the parking lot.

[0013] Preferably, the processing of users with long-term delinquency includes the following steps: Synchronize the delinquency records to a third-party credit scoring platform; Blacklist your vehicle and restrict access to partner parking lots; Triggering legal recovery procedures.

[0014] Preferably, the privacy protection measures for the user information include: Data storage is encrypted using AES-256; Data transmission uses TLS1.3 protocol; User data is anonymized and only part of the license plate number and payment information is displayed.

[0015] The present invention provides a full-scenario integrated cross-system parking recovery service component docking method. It has the following beneficial effects: 1. Through edge computing and real-time data synchronization technology, the present invention can quickly collect information such as license plate number, parking time, payment status, etc. in the parking lot, and complete preliminary processing on the edge device. The data is uploaded to the central server in real time through the MQTT protocol, reducing data delays and transmission packet loss problems. Compared with traditional manual collection and centralized processing modes, the present invention significantly improves the real-time and accuracy of parking data, and provides an efficient data basis for subsequent collection of overdue fees.

[0016] 2. The present invention adopts an intelligent model based on LSTM and random forest to accurately predict the user's arrears risk, and dynamically generates personalized collection strategies according to the risk level. Low-risk users receive a single friendly reminder, medium-risk users enjoy installment payment options, and high-risk users are restricted from parking services and share credit information. Compared with traditional static rules and single processing methods, the present invention realizes more intelligent and flexible arrears collection, greatly improving the arrears conversion rate and user experience.

[0017] 3. The present invention provides a variety of user notification and payment methods, including SMS, APP push, voice assistant reminder and pre-authorization deduction mechanism, to meet the preferences and payment needs of different users. In particular, the pre-authorization mechanism can freeze the fees when the vehicle enters the venue, ensuring that the user can complete the payment in time. Compared with a single payment and notification mode, the multi-channel interaction method of the present invention improves the notification success rate and payment efficiency, while optimizing the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a diagram of the method steps of the present invention. DETAILED DESCRIPTION

[0019] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] Please refer to the attached Figure 1 The embodiment of the present invention provides a fully detailed description of the full-scenario integrated cross-system parking recovery service component docking method, comprehensively revealing its core principles, technical implementation details and implementation solutions of key modules.

[0021] 1. Overall system implementation plan The present invention provides a full-scenario integrated cross-system parking collection service component docking method, which aims to solve the problems of scattered parking data, difficulty in collecting arrears and poor user experience. By designing edge computing acquisition module, data fusion module, AI intelligent analysis module, dynamic rule engine module, multi-channel execution module and privacy and compliance module, the full process and intelligent collection of parking arrears can be realized.

[0022] The system includes the following main components and working principles: Edge computing collection module: Edge devices are deployed in the parking lot to collect and process parking data in real time, including license plate recognition, parking time calculation, preliminary cost estimation, etc.

[0023] Data fusion and storage module: The collected parking and payment data are cleaned and integrated and stored in a distributed database, supporting real-time query and historical analysis.

[0024] AI intelligent analysis module: Build a payment arrears prediction model based on machine learning to classify users' payment arrears risks.

[0025] Dynamic rule engine module: dynamically generates recovery strategies based on user risk levels and works in conjunction with the execution module.

[0026] Payment and notification execution module: sends overdue payment notifications through multiple channels and supports multiple payment methods to complete collection.

[0027] Privacy and Compliance Module: Ensure the security and compliance of user data.

[0028] 2. Edge computing acquisition module 2.1 Working Principle The edge computing module is deployed in the parking lot to collect vehicle information in real time and perform preliminary calculations, reducing the pressure on the central server and network transmission costs.

[0029] 2.2 Technical Implementation License Plate Recognition: Use edge devices such as NVIDIA Jetson Nano to run lightweight YOLO or OCR models to recognize license plates.

[0030] Low-precision recognition results are filtered using a confidence threshold, and the recognized image is stored in association with the license plate number.

[0031] Parking time calculation: Entry and exit times are recorded via IoT sensors, and parking duration is calculated based on the time difference.

[0032] Preliminary cost calculation: Map the parking duration to the predefined billing rule table to generate preliminary parking fees.

[0033] 2.3 Data Transmission After preprocessing, the data is uploaded to the central database using the MQTT protocol, which supports breakpoint resumption and packet loss retransmission to ensure data stability.

[0034] 3. Data fusion and storage module 3.1 Data Cleansing Duplicate record removal: De-duplicate based on license plate number and timestamp to avoid duplicate uploaded parking records.

[0035] Outlier filtering: Delete records with negative parking duration or unreasonable fees.

[0036] 3.2 Data Fusion Link parking records to payment records based on license plate number.

[0037] Supplement the user's credit score and vehicle information to form a complete user portrait.

[0038] 3.3 Data Storage Tiered storage solution: Hot data (current parking status): stored in Redis, supporting real-time query.

[0039] Cold data (historical records): stored in a distributed file system (HDFS) for analysis and backup.

[0040] Database structure design: The parking record table, payment record table, and user information table are linked to support multi-dimensional queries.

[0041] 4. AI intelligent analysis module 4.1 Debt arrears risk prediction model Model selection: The LSTM model is used to perform time series prediction on payment behavior, and the risk of arrears is determined based on the user's parking habits.

[0042] Use random forest to classify users into risk categories (low risk, medium risk, high risk).

[0043] Feature Engineering: Static features: user’s credit score and historical arrears record.

[0044] Dynamic features: parking duration distribution, payment delay behavior, average cost.

[0045] Model training: Dataset division: 80% for training and 20% for testing.

[0046] Loss function: Binary cross entropy is used for classification tasks, mean squared error is used for regression tasks.

[0047] Online prediction: User parking and payment data are input into the model in real time, and the probability of overdue payment and classification results are output.

[0048] 5. Dynamic rule engine module 5.1 Working Mechanism The rule engine generates corresponding recovery strategies based on the user classification results and works in conjunction with the execution module.

[0049] 5.2 Rule Design Low-risk users: Send friendly reminders to reduce user resistance.

[0050] Provide payment incentives (such as small discounts).

[0051] Medium risk users: Increase reminder frequency and provide multi-channel notifications via SMS and push notifications.

[0052] Installment payment plans available.

[0053] High-risk users: Synchronize the overdue payment information to the credit platform.

[0054] Restrict vehicle access to partner parking lots.

[0055] 5.3 Dynamic Adjustment Use reinforcement learning algorithms to simulate the effects of different rules (such as payment conversion rate) and optimize rule weights.

[0056] 6. Payment and Notification Execution Module 6.1 Multi-channel notification Notification method: SMS reminder: sent in real time via SMS gateway.

[0057] APP push: Use Firebase push service.

[0058] Voice assistant: Reminders through car networking voice devices.

[0059] Notification Personalization: Use NLP technology to generate personalized notification content and improve user acceptance.

[0060] 6.2 Payment Execution Automatic deduction: Supports pre-authorization deductions, freezing part of the fees when the vehicle enters the venue.

[0061] Alternative payment: Provide multiple payment methods (bank cards, e-wallets, ETC accounts).

[0062] 7. Privacy and Compliance Module 7.1 Data Encryption User data is encrypted using the AES-256 algorithm.

[0063] Data transmission uses TLS1.3 protocol to ensure transmission security.

[0064] 7.2 Data Desensitization Display part of the license plate number (such as ABC123→ABC***) to protect user privacy.

[0065] 7.3 Compliance assurance Comply with GDPR and CCPA regulations.

[0066] Credit information sharing requires user authorization.

[0067] 8. Handling of users with long-term arrears 8.1 Credit Union Synchronize the overdue payment records of high-risk users to Sesame Credit or the central bank’s credit reporting system.

[0068] Lower the user's credit score and encourage active payment.

[0069] 8.2 Parking restrictions Long-term overdue users will be placed on the parking lot’s blacklist and prohibited from entering the partner parking lot.

[0070] 8.3 Legal Recovery Take legal measures and initiate litigation to recover the high-amount of debts owed by users.

[0071] 9. Effect feedback and system optimization 9.1 Data Feedback Collect data such as payment conversion rate and user complaint rate.

[0072] Record the effect of rule execution for model and rule optimization.

[0073] 9.2 Model Update Perform incremental training based on the latest data to optimize the arrears prediction model.

[0074] 9.3 Rule Optimization Dynamically adjust rule engine parameters using reinforcement learning Example 1: City-level parking fee collection system based on real-time data synchronization Specific steps: Edge device deployment: Deploy NVIDIA Jetson Nano as an edge device to collect vehicle information in real time.

[0075] The OCR model recognizes the license plate, and the confidence threshold is set to 90%.

[0076] Data transmission: The MQTT protocol is used to achieve real-time transmission of parking records, and the data transmission delay is controlled within 300ms.

[0077] Risk prediction of arrears: The LSTM model is trained on 5,000 user parking records with a time series length of 30 days.

[0078] Output the probability of defaulting on payments and classify users into low, medium, and high risk based on a threshold value (0.5).

[0079] Strategy Execution: Low-risk users will receive a one-time SMS reminder.

[0080] Medium-risk users will be notified three times and offered the option of paying in installments.

[0081] Parking lot entrance is restricted for high-risk users, and overdue payment information is shared with the credit platform.

[0082] Payment Execution: Pre-authorization deduction is used, and the frozen amount is 110% of the estimated cost.

[0083] The payment response time does not exceed 500ms.

[0084] Comparative Example 1: Traditional parking arrears management without real-time synchronization Data collection is done manually by uploading once a day.

[0085] User arrears are manually classified based only on static features (such as the number of historical arrears).

[0086] Only unified notifications are sent to users who have been in arrears for a long time, without any tiered strategy.

[0087] Payments are made manually, with no pre-authorization mechanism.

[0088] Example 2: A system for collecting overdue fees in multiple parking lots across regions Specific steps: Multi-parking lot data integration: The data format is unified in JSON format, and the fields include license plate number, parking duration, fee, etc.

[0089] The distributed database (MongoDB) stores the merged data and supports cross-region queries.

[0090] Dynamic policy generation: Based on Drools rule engine: Send discount reminders to low-risk users.

[0091] Installment payment function is provided for medium-risk users.

[0092] High-risk users are blacklisted and service is restricted.

[0093] User Notice: Use Firebase to push daily overdue payment reminders.

[0094] High-risk users access voice assistant notifications, with a notification success rate of 90%.

[0095] Data Analysis: Analyze the arrears data daily and update the rule weights.

[0096] The target for overdue payment conversion rate is increased by 10%.

[0097] Comparative Example 2: Single Parking Lot Decentralized Management Mode Each parking lot manages data independently, making cross-regional integration impossible.

[0098] There is no tiered strategy, and all users in arrears receive the same reminder.

[0099] User data is updated offline with a delay of more than 1 day.

[0100] There is no voice notification function and the notification form is single.

[0101] Example 3: Intelligent risk prediction system for users in arrears Specific steps: Data collection and cleaning: The camera is used to collect license plate data, and the YOLO model is used for license plate recognition with an accuracy rate of 95%.

[0102] Parking records and payment records are associated through timestamps to remove duplicate data.

[0103] Risk Profile: The LSTM model was used to train 100,000 parking records, and the classification model (random forest) was used to classify users into low, medium, and high risk.

[0104] Model training parameters: learning rate 0.01, batch size 128.

[0105] Payment solution optimization: Users with high arrears are provided with the option of paying in installments, with the minimum payment amount for each installment being 20% ​​of the total arrears.

[0106] The system’s automatic deduction success rate reaches 85%.

[0107] Long-term arrears processing: The overdue payment records are synchronized to the credit platform, reducing the user's credit score.

[0108] Vehicles with long-term overdue payments will be added to the parking lot blacklist and prohibited from entering the cooperative parking lot.

[0109] Comparative Example 3: Manual Management of Static Rules There is no dynamic risk prediction, only manual classification based on historical arrears records.

[0110] The user payment method is single and does not support installment payments or automatic deductions.

[0111] For users who have been in arrears for a long time, only SMS notifications are used, and credit records are not shared.

[0112] Comparative Experimental Design Purpose Verify the performance advantages of the full-scenario integrated cross-system parking recovery service component docking method in terms of arrears conversion rate, notification success rate, payment response time, etc.

[0113] Experimental parameters Experimental scale: 10,000 parking records and 5,000 arrears records were tested.

[0114] Comparison system: The above embodiment is compared with the corresponding proportional system.

[0115] Evaluation indicators: Arrears conversion rate (number of users who successfully paid their bills / total number of users who owed their bills).

[0116] Notification success rate (number of notifications responded to by users / total number of notifications).

[0117] Payment response time (the time it takes for a user to complete a payment).

[0118] Experimental data table: Experimental methods Debt conversion rate (%) Notification success rate (%) Payment response time (ms) System delay (ms) Example 1 85 95 450 300 Comparative Example 1 60 75 1200 1500 Example 2 88 97 500 350 Comparative Example 2 65 80 1100 1600 Example 3 90 98 420 250 Comparative Example 3 70 85 1000 1400 According to the data in the table, the conversion rates of the embodiments are higher than those of the comparison groups (85%-90% vs. 60%-70%), which reflects the advantages of risk prediction and stratification strategies. The dynamic installment payment scheme of Example 3 further enhances the payment willingness of users with high arrears. The notification success rate of the embodiments is significantly higher than that of the comparison groups (95%-98% vs. 75%-85%), especially voice notification and personalized push notification improve the user response rate. In the embodiments, the payment response time is short (420ms-500ms), while the comparison group has a longer time due to manual operation or the absence of a pre-authorization mechanism. The embodiments use edge computing and real-time synchronization to significantly reduce system latency (250ms-350ms vs. 1400ms-1600ms).

[0119] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A full-scenario integrated cross-system parking recovery service component docking method, characterized in that: The following steps are involved: Collect vehicle information, parking records, and payment status in the parking lot through edge computing nodes or IoT devices; Clean and standardize the collected data to form structured parking and payment records; The parking and payment data are stored and synchronized in real time through distributed data storage; Based on the user's parking behavior, payment behavior and credit score, a user arrears risk prediction model is built to stratify the users by risk; Based on the user's risk level, the dynamic rule engine is called to generate corresponding overdue payment collection strategies; Notify users in arrears through multiple channels, and collect overdue payments by combining automatic deductions with alternative payment methods; Credit information sharing, parking service restrictions or legal collection for users with long-term or high-amount delinquency records; Ensure the security and compliance of user information through data encryption and privacy protection mechanisms.

2. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The edge computing node is used to process vehicle information in real time, including license plate recognition, parking time calculation and preliminary parking fee calculation.

3. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The cleaning and standardization process comprises the following steps: Remove duplicate parking records; Filter the license plate recognition data whose confidence level is lower than the set threshold; Parking records and payment records are linked by timestamp.

4. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The distributed data storage adopts a cache storage and historical storage layered design, where: Cache storage is used to access current parking and payment status in real time; Historical storage is used to save long-term arrears data and support subsequent analysis.

5. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The arrears risk prediction model is constructed using a machine learning algorithm, combined with a time series model to predict user payment behavior, and user risk levels are stratified through a classification model.

6. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The dynamic rule engine dynamically generates the following recovery strategies based on user risk stratification: Send a single reminder notification to low-risk users; Increase reminder frequency for medium-risk users and provide installment payment options; For high-risk users, the overdue payment information will be shared with the credit platform and subsequent parking services will be restricted.

7. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The multi-channel notification method includes: SMS reminders; Mobile application push; Voice assistant notifications; Smart car networking system reminder.

8. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The automatic deduction method includes a pre-authorization deduction mechanism, which freezes part of the parking fee when the vehicle enters the parking lot and completes the deduction when the user exits the parking lot.

9. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The processing of users in long-term arrears includes the following steps: Synchronize the delinquency records to a third-party credit scoring platform; Blacklist your vehicle and restrict access to partner parking lots; Triggering legal recovery procedures.

10. The full-scenario integrated cross-system parking recovery service component docking method according to claim 1 is characterized in that: The privacy protection measures for user information include: Data storage is encrypted using AES-256; Data transmission uses TLS1.3 protocol; User data is anonymized and only part of the license plate number and payment information is displayed.

Images