Reservation type vehicle and goods matching method and device applying AI model and storage medium

Abstract

The invention discloses a reservation type vehicle and cargo matching method and device applying an AI model and a storage medium, and the method comprises the steps: receiving a cargo owner transportation plan, and generating a vehicle finding demand; obtaining reserved driver information, and determining a transport capacity pool to which a reserved driver belongs and a corresponding transport capacity classification attribute; the AI model with the built-in vehicle and cargo matching rule is combined with the transport capacity pool and the corresponding transport capacity classification attribute to intelligently match a target driver, and a transport task is generated to notify the target driver to receive an order and carry; and the intelligent matching process improves the priority of the allocated transportation tasks of the reservation drivers with the goods-fixing certificates. And receiving an activation signal of the order certificate, distributing the AI digital person to follow up the performance condition of the target driver in real time, automatically identifying the risk, and sending out a guide prompt. According to the invention, on the basis of an AI large model technology, reservation type vehicle and goods matching based on a goods fixing certificate is realized, the space-time matching efficiency of a vehicle source and a goods source is optimized, the vehicle and goods matching efficiency is improved, a driver is assisted in conveniently and rapidly completing a performance transaction in a full-service process through an AI digital person, and the service quality of the driver is improved.

Description

Technical Field

[0001] The present invention relates to the field of logistics technology, and in particular, to a reservation-based vehicle-cargo matching method, device, and storage medium applying an AI model. Background Art

[0002] Vehicle-cargo matching is an important link in the logistics industry, and its purpose is to effectively connect the goods transportation demand with suitable transportation vehicles to achieve efficient utilization of resources and optimization of transportation costs. Traditional vehicle-cargo matching methods mainly rely on manual operations and offline communication to complete coordination.

[0003] Some logistics platforms use GPS positioning and basic algorithms for matching, and the scheduling does not have a strong association with the transport capacity, resulting in problems such as low matching efficiency, information asymmetry, and low intelligence level. Summary of the Invention

[0004] To solve the technical problems in the background art, the present invention proposes a reservation-based vehicle-cargo matching method, device, and storage medium applying an AI model.

[0005] The present invention proposes a reservation-based vehicle-cargo matching method applying an AI model, including:

[0006] Receiving the shipper's transportation plan and generating a vehicle-finding demand;

[0007] Obtaining the information of reserved drivers, and determining the transport capacity pool to which the reserved drivers belong and the corresponding transport capacity classification attributes;

[0008] An AI model with built-in vehicle-cargo matching rules combines the transport capacity pool to which it belongs and the corresponding transport capacity classification attributes, intelligently matches the target driver, and generates a transportation task to notify the target driver to take the order and carry out the transportation;

[0009] The intelligent matching process improves the priority of reserved drivers with order certificates being assigned transportation tasks.

[0010] Receiving the activation signal of the order certificate, allocating an AI digital human to track the performance of the target driver in real time, automatically identifying risks, and sending guiding reminders.

[0011] As a further improvement of the present invention, a transport capacity pool is pre-constructed before applying the AI model, specifically including:

[0012] Constructing a long-term cooperation potential model and setting driver credit score rules;

[0013] Introducing historical credit data on the number of driver cooperations and driver evaluations to determine the driver credit score;

[0014] Screening drivers with a credit score higher than the set value to construct a recommended transport capacity pool and issuing order certificates;

[0015] It also obtains the shipper's dynamic business requirements and driver behavior data to generate a dynamic transport capacity profile.

[0016] As a further improvement of the present invention, during the intelligent matching process,

[0017] Map the transportation task characteristics required by the shipper into a first vector, where the transportation task characteristics at least include vehicle type requirements, load limit, transportation route, and risk preference;

[0018] Map the driver and vehicle characteristics of the dynamic transport capacity profile into a second vector, where the driver and vehicle characteristics at least include vehicle type capabilities, historical performance fulfillment rate, heat map of frequently traveled routes, and real-time credit score;

[0019] Calculate the direction consistency between the first vector and the second vector through the cosine similarity algorithm to generate a matching degree quantification index, and sort the reserved drivers according to the matching degree index;

[0020] Recommend reserved drivers with a similarity higher than the preset threshold to complete the vehicle-cargo matching.

[0021] As a further improvement of the present invention, before applying the AI model, it also includes setting transport capacity classification and transport capacity grading - transaction rules;

[0022] Obtain the driver's behavioral time series data to determine spatio-temporal characteristics;

[0023] In the AI model, use the clustering algorithm to fuse spatio-temporal characteristics, divide the basic transport capacity groups, and set the transport capacity group levels;

[0024] Obtain the driver's real-time behavior data for dynamic update of transport capacity grouping and levels;

[0025] Analyze the GPS trajectory in the driver's time series data to identify abnormal behaviors;

[0026] Design dynamic credit score rules to deduct credit points for abnormal behaviors and add credit points for continuous compliance behaviors to evaluate the transport capacity credit score;

[0027] According to the mapping relationship between the credit score and the transport capacity risk level, design corresponding transaction strategies to be triggered.

[0028] As a further improvement of the present invention, during the intelligent matching process, it also includes:

[0029] Identify the transport level of the transport plan, determine the lowest transport capacity group level suitable for the transport level, and exclude reserved drivers whose transport capacity group level is lower than the lowest transport capacity group level;

[0030] Detect the transport capacity group level of the target driver and match the transaction strategy corresponding to the transport capacity group level.

[0031] As a further improvement of the present invention, the real-time follow-up of the AI digital human includes:

[0032] Analyze the deviation between the vehicle trajectory and the planned route in real time, dynamically adjust the warning threshold in combination with the estimated arrival time algorithm, and predict the impact of route deviation on the delivery time;

[0033] Dynamically predict the arrival time by combining multi-dimensional data such as real-time road conditions, distance measurement, and vehicle status;

[0034] Trigger a warning when it is detected that the driver spends more time than the preset threshold in the setting session;

[0035] Trigger a warning when the driver fails to complete the task as planned.

[0036] As a further improvement of the present invention, it further includes:

[0037] Obtain user demand information, preprocess the user demand information, and extract key information;

[0038] Input the key information into the intention classification model, and output the user intention, where the user intention includes providing goods sources, seeking transportation capacity, and making a reservation for carriage.

[0039] The present invention also proposes a reservation-based vehicle-cargo matching device applying an AI model, including: a memory and a processor, the memory is used to store information including program instructions, the processor is used to control the execution of the program instructions, and when the program instructions are loaded and executed by the processor, the above-mentioned reservation-based vehicle-cargo matching method applying an AI model is implemented.

[0040] The present invention also proposes a computer storage medium, the storage medium includes a stored program, and the processor executes the program to implement the above-mentioned reservation-based vehicle-cargo matching method applying an AI model.

[0041] Beneficial effects:

[0042] Receive the shipper's transportation plan, generate a vehicle-finding demand, obtain reservation driver information, determine the capacity pool to which the reservation driver belongs and the corresponding capacity classification attributes, and the AI model with built-in vehicle-cargo matching rules combines the capacity pool to which it belongs and the corresponding capacity classification attributes to intelligently match the target driver, and generate a transportation task to notify the target driver to accept the order for carriage; the intelligent matching process improves the priority of reservation drivers with order certificates to be assigned transportation tasks. Receive the activation signal of the order certificate, allocate an AI digital human to follow up the performance of the target driver in real time, automatically identify risks, and issue guidance reminders. Utilize the AI large model technology to realize reservation-based vehicle-cargo matching based on the order certificate, improve the vehicle-cargo matching efficiency, and assist the driver to complete the performance transaction conveniently and quickly through the AI digital human in the whole business process. Description of the Drawings

[0043] Figure 1Flow chart of a reservation-based vehicle-cargo matching method using an AI model proposed by the present invention;

[0044] Figure 2 Business flow chart of vehicle-cargo matching based on the AI large model proposed by the present invention. Detailed implementation manners

[0045] As Figure 1 shown, Figure 1 is the flow chart of a reservation-based vehicle-cargo matching method using an AI model according to an embodiment of the present invention;

[0046] Referring to Figures 1 to 2 , a reservation-based vehicle-cargo matching method using an AI model proposed by an embodiment of the present invention includes the following steps:

[0047] S10: Receive the shipper's transportation plan and generate a vehicle-finding requirement;

[0048] S20: Obtain the reservation driver information, and determine the capacity pool to which the reservation driver belongs and the corresponding capacity classification attributes; the capacity pool and capacity classification are pre-set before vehicle-cargo matching;

[0049] S30: The AI model with built-in vehicle-cargo matching rules combines the capacity pool to which it belongs and the corresponding capacity classification attributes, intelligently matches the target driver, and generates a transportation task to notify the target driver to accept the order and carry out the transportation;

[0050] S40: Improve the priority of reservation drivers with order confirmation certificates being assigned transportation tasks during the intelligent matching process;

[0051] S50: Upon receiving the activation signal of the order confirmation certificate, assign an AI digital human to track the performance of the target driver in real time, automatically identify risks, and issue guidance reminders.

[0052] Drivers make early reservations on the APP side. When the enterprise side releases the goods source, intelligent matching is performed on the reserved drivers (in dimensions such as vehicle type, vehicle length, vehicle location, route, risk level, capacity type, capacity level, etc.), simplifying the driver reservation and order acceptance process and improving the matching efficiency.

[0053] This method applies the principle of intelligent agent construction and combination: adopting a divide-and-conquer strategy, each intelligent agent is designed as a software entity with specific functions, such as a task guidance intelligent agent, an information collection intelligent agent, a matching intelligent agent, a decision-making intelligent agent, etc. The information collection intelligent agent is responsible for collecting vehicle source and goods source data from different channels; the matching intelligent agent analyzes and matches the collected vehicle source and goods source information based on the AI large model and specific matching algorithms; the decision-making intelligent agent makes the final matching decision according to the matching results and business rules. By combining these intelligent agents, an organic system is formed to achieve efficient vehicle-cargo matching.

[0054] It also applies the principle of the rule engine: Business rules (such as vehicle load limits, transportation distance requirements, cargo type matching, etc.) are defined in a structured form and stored in the rule library. When new vehicle source or cargo source information enters the system, or various situations occur during the matching process, the rule engine will automatically retrieve the rule library, evaluate and judge the data according to the preset rules, and monitor in real time whether the business meets the requirements. Once an abnormality or a situation that does not conform to the rules (such as overloading reservation, etc.) is found, an alarm will be issued in a timely manner and intervention will be carried out or an intelligent agent will be called in to intervene and guide to ensure the compliance and efficiency of the vehicle-cargo matching business.

[0055] In some embodiments, it further includes:

[0056] Obtain user demand information, preprocess the user demand information, and extract key information;

[0057] Input the key information into the intent classification model, and output the user intent, where the user intent includes providing cargo sources, seeking transportation capacity, and making reservation for carriage.

[0058] Principle of intent recognition: Using large model technology, when shippers and truck owners input demands or provide information on the platform, the AI large model performs preprocessing such as word segmentation, part-of-speech tagging, and syntactic analysis on the text, and extracts keywords and key phrases. Then, through the intent classification model trained by machine learning algorithms, the input text is matched with predefined intent categories (such as seeking transportation capacity, providing cargo sources, etc.) to determine the true intent of the user.

[0059] Based on the AI application development platform, using AI large model technology, through intent recognition and the construction and combination of intelligent agents, combined with the business real-time monitoring system based on the rule engine, realize the reservation-based vehicle-cargo matching based on order certificates, optimize the spatio-temporal matching efficiency of vehicle sources and cargo sources, reduce the high transaction costs in business matching, reduce the empty driving rate and transportation costs, and improve the vehicle-cargo matching efficiency.

[0060] In some embodiments, a transportation capacity pool is pre-constructed before the application of the AI model, which specifically includes:

[0061] Integrate the shipper's own fleet, long-term cooperative drivers, and transportation capacity of third-party platforms into a unified transportation capacity pool;

[0062] Construct a long-term cooperation potential model and set driver credit score rules;

[0063] Introduce historical credit data of driver cooperation times and driver evaluations to evaluate the driver credit score;

[0064] Screen drivers with a credit score higher than the set value to construct a recommended transportation capacity pool and issue order certificates;

[0065] It also obtains the shipper's dynamic business demands and driver behavior data to generate a dynamic transportation capacity profile.

[0066] During the process of building a transportation capacity pool, data collection can be carried out by using OCR to recognize driver's licenses / vehicle operation licenses, and structured data such as vehicle models, vehicle lengths, and load tonnages can be obtained. Driver behavior data includes heat maps of frequently traveled routes (based on GPS trajectory clustering analysis) and historical load utilization rates (load tonnage / actual transported cargo weight). Credit data includes the decay coefficient of the number of cooperations (the higher the weight of more recent cooperations) and driver transaction fulfillment data. Evaluate the credit data according to the driver credit score rules to determine the driver credit score.

[0067] In the implementation method, for the construction of a dynamic transportation capacity pool based on long-term cooperation relationships, introduce two-dimensional historical data of "number of cooperations + driver evaluation", build a long-term cooperation potential model, give priority to recommending drivers with high credit, form a stable transportation capacity pool, and reduce transportation risks. The transportation capacity portrait technology with multi-dimensional feature fusion combines the dynamic business needs of the enterprise (such as seasonal fluctuations) with driver behavior data (frequently traveled routes, load utilization rates) to generate a dynamic transportation capacity portrait and achieve precise supply-demand matching.

[0068] In some implementation methods, the intelligent matching process further includes:

[0069] Map the transportation task characteristics required by the shipper into a first vector. The transportation task characteristics at least include vehicle model requirements, load restrictions, transportation routes, and risk preferences, and the transportation task characteristics can be determined through the vehicle finding requirements;

[0070] Map the driver and vehicle characteristics of the dynamic transportation capacity portrait into a second vector. The driver and vehicle characteristics at least include vehicle model capabilities, historical fulfillment rates, heat maps of frequently traveled routes, and real-time credit scores;

[0071] Calculate the direction consistency between the first vector and the second vector through the cosine similarity algorithm to generate a matching degree quantization index, and sort the reserved drivers according to the matching degree index;

[0072] Recommend reserved drivers with a similarity higher than the preset threshold to complete the vehicle-cargo matching.

[0073] In the AI model, the collaborative filtering recommendation algorithm is adopted to map the enterprise requirements and the dynamic transportation capacity portrait into a vector space and calculate the similarity (such as cosine similarity).

[0074] Also, through reinforcement learning for dynamic weight allocation, realize the adaptive adjustment of the enterprise requirement priorities, and feedback and adjust the feature weights according to business results (such as transportation on-time rate, cargo damage rate) (for example, give priority to matching drivers on "frequently traveled routes" for urgent orders). Another example is to automatically increase the weight of "route compliance" during the epidemic.

[0075] Furthermore, it also includes the replenishment of the transportation capacity pool:

[0076] Through time series analysis of the shipper's historical business data, predict the shipper's future transportation capacity gap;

[0077] Initiate targeted recruitment of transportation capacity in advance.

[0078] Through time series analysis of historical business data, predict the future transportation capacity gap of the enterprise, initiate targeted recruitment in advance, and solve the problem of ensuring transportation timeliness. A demand prediction model can be constructed, using the Prophet time series model to analyze the periodicity of the enterprise's historical orders, and combining external data (such as macroeconomic indices, regional sales data) to predict the future transportation capacity gap through an LSTM neural network.

[0079] Targeted recruitment strategy: Through a graph neural network (GNN), construct a "enterprise - driver - route" relationship graph to identify potential high-quality transportation capacity (such as recommending drivers who have cooperated on similar routes with the target shipper). Increase the flexibility of transportation capacity expansion, and conduct dynamic auction matching for temporary gaps with drivers outside the transportation capacity pool. For example, design a two-objective optimization model: the lowest cost for the enterprise + the highest income for the driver.

[0080] In some implementation manners, before the application of the AI model, it also includes transportation capacity classification:

[0081] Obtain the behavioral time series data of the driver, and determine the spatio-temporal characteristics. Among them, the behavioral time series evidence includes GPS trajectories and loading / unloading timeliness, and the spatio-temporal characteristics include the regional density of frequently traveled routes.

[0082] In the AI model, use a clustering algorithm to fuse spatio-temporal characteristics, divide the basic transportation capacity groups, and set the levels of transportation capacity groups.

[0083] Obtain the real-time behavioral data of the driver, and perform dynamic updates of the transportation capacity groups and levels.

[0084] In addition, generate a feature importance report for each transportation capacity group (such as the "high-risk group" due to a route deviation rate > 15% and a credit score < 60).

[0085] Through real-time behavioral data + AI dynamic clustering, achieve automatic updates and level adjustments of transportation capacity groups (such as automatically downgrading "high-risk drivers" to low-priority groups). Through incremental learning, absorb new data in real time and update the groups every hour (such as adjusting the grouping of drivers who violate risk control to the risk group in a timely manner).

[0086] Also, through a graph neural network (GNN), construct a "transportation capacity - order - enterprise" relationship graph to identify hidden associations (such as a driver with a low score but long-term service to high-value customers, who is classified into a special protection group).

[0087] The enterprise groups and labels the transportation capacity under its own transportation capacity pool, supports the integration of transportation capacity information in the ways of enterprise grouping, custom grouping, and intelligent grouping, supports viewing, adding, and removing transportation capacity information under the transportation capacity groups, and realizes more refined and visual management of transportation capacity information.

[0088] Furthermore, during the intelligent matching process, it also includes:

[0089] Identify the transportation level of the transportation plan, determine the lowest transport capacity grouping level suitable for the transportation level, and exclude those with the transport capacity grouping level of the reserved driver less than the lowest transport capacity grouping level; Example, high-value orders are only open to high-quality transport capacity with order certificates, and ordinary orders are open to low-level transport capacity;

[0090] Detect the transport capacity group level of the target driver and match the trading strategy corresponding to the transport capacity group level.

[0091] Design a grouping-trading rule engine, where the transport capacity of different groups automatically triggers different trading strategies (such as high-level transport capacity participating in bidding first, and low-level transport capacity needing to pay a deposit).

[0092] Furthermore, before applying the AI model, it also includes setting transport capacity grading-trading rules:

[0093] Analyze the GPS trajectory to identify abnormal behaviors such as route deviation, long-term stay, and loading / unloading overtime;

[0094] Design a dynamic credit score rule to deduct credit points for abnormal behaviors and add credit points for continuous compliance behaviors, and evaluate the transport capacity credit score. Abnormal behaviors include route deviation, long-term stay, and loading / unloading overtime;

[0095] Design corresponding trading strategies to trigger according to the mapping relationship between the credit score and the risk level.

[0096] Through real-time monitoring of various abnormal behaviors and risk behaviors in the entire chain, conduct intelligent grouping of transport capacity, adjust the transport capacity level, assist in transport capacity trading decisions, and reduce the risk of transport capacity use.

[0097] The transport capacity grading driven by risks across the entire chain is based on a multi-dimensional anomaly detection model (such as route deviation, loading / unloading overtime, credit fluctuation), and the transport capacity level is adjusted in real time to achieve risk pre-interception. Conduct time-series anomaly detection on the driver's time-series data, use the Transformer model to analyze the GPS trajectory, and identify risks such as route deviation and long-term stay. Also add multi-modal fusion detection, such as combining the goods list recognized by OCR to detect problems such as inconsistent goods lists.

[0098] When evaluating the dynamic transport capacity level, introduce a fuzzy logic algorithm to handle uncertain risks (such as different deductions for "slight overtime" and "severe overtime"). Design a dynamic credit score algorithm to deduct points for abnormal behaviors (such as -5 points for each route deviation) and add points for continuous compliance behaviors (such as +20 points for 30 consecutive orders without anomalies). Example of the risk level mapping, credit score → four levels of A, B, C, D, triggering different trading strategies (such as D-level transport capacity requires manual review to accept orders).

[0099] In some embodiments, the real-time follow-up of the AI digital human includes:

[0100] Analyze the deviation between the vehicle trajectory and the planned route in real time, dynamically adjust the warning threshold in combination with the estimated arrival time algorithm, and predict the impact of the deviation on the delivery time;

[0101] Combine multi-dimensional data such as real-time road conditions, distance measurement, and vehicle status (such as staying) to dynamically predict the arrival time;

[0102] Trigger a warning when the driver spends more time than the preset threshold in a certain link (such as loading, unloading, or resting);

[0103] Trigger a warning when the driver fails to complete the task as planned (such as canceling the order or not arriving on time).

[0104] Based on information such as the route, category, tons / pieces, loading place, and unloading place in the transportation plan, after being identified by AI learning, screen and activate the drivers with order certificates among the reserved drivers, and quickly, accurately, and correctly complete the matching of the transportation plan and the drivers.

[0105] After the driver makes a reservation, the AI ​​constantly monitors the driver's arrival situation. By calculating the real-time position of the driver and the address of the loading place, it judges whether the driver may be late or default on arrival, and uses the AI digital human to remind the driver in the APP. In case of situations where the reservation cannot arrive in time or defaults on arrival, the AI digital human will comfort the driver emotionally in a high EQ manner in combination with the business situation, and promptly notify the consignor or logistics enterprise to minimize the possible differences and quarrels between the two parties.

[0106] In the form of an AI digital human, serve as the driver's business assistant throughout the entire business process, assist the driver in conveniently and quickly completing the performance transaction before, during, and after the transaction, and improve the driver's own service quality.

[0107] It should be noted that in the above embodiments, the program can be written in any combination of one or more programming languages ​​to write the program code for performing the operations of the embodiments of the present application. The programming languages ​​include object-oriented programming languages, such as Java, C++, etc., and also include conventional procedural programming languages, such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computing device, partially on the user's device, executed as an independent software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0108] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the division of the above functional modules is used as an example. In actual applications, the above functions can be allocated to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. For the specific working process of the device described above, reference can be made to the corresponding process in the foregoing method embodiments, which will not be elaborated here.

[0109] As mentioned above, the above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A reservation-based vehicle-cargo matching method using an AI model, characterized in that, Including: Receiving the shipper's transportation plan and generating a vehicle search requirement; Obtaining the reserved driver information and determining the capacity pool to which the reserved driver belongs and the corresponding capacity classification attributes; An AI model with built-in vehicle-cargo matching rules combines the capacity pool to which it belongs and the corresponding capacity classification attributes, intelligently matches the target driver, and generates a transportation task to notify the target driver to accept the order and carry out the transportation; The intelligent matching process improves the priority of reserved drivers with ordering certificates to be assigned transportation tasks; Upon receiving the activation signal of the ordering certificate, an AI digital human is assigned to track the performance of the target driver in real time, automatically identify risks, and issue guiding reminders.

2. The appointment-based vehicle-cargo matching method using an AI model according to claim 1, wherein Before applying the AI model, a capacity pool is pre-constructed, specifically including: Constructing a long-term cooperation potential model and setting driver credit score rules; Introducing historical credit data on the number of driver cooperations and driver evaluations to determine the driver credit score; Screening drivers with a credit score higher than the set value to construct a recommended capacity pool and issuing ordering certificates; It also obtains the shipper's dynamic business requirements and driver behavior data to generate a dynamic capacity profile.

3. The appointment-based vehicle-cargo matching method applying an AI model according to claim 2, characterized in that During the intelligent matching process, Mapping the transportation task characteristics required by the shipper into a first vector, where the transportation task characteristics at least include vehicle type requirements, load limits, transportation routes, and risk preferences; Mapping the driver and vehicle characteristics of the dynamic capacity profile into a second vector, where the driver and vehicle characteristics at least include vehicle type capabilities, historical performance rates, heat maps of frequently traveled routes, and real-time credit scores; Calculating the direction consistency between the first vector and the second vector through the cosine similarity algorithm to generate a matching degree quantization index, and sorting the reserved drivers according to the matching degree index; Recommending reserved drivers with a similarity higher than the preset threshold to complete the vehicle-cargo matching.

4. The appointment-based truck-load matching method using an AI model according to claim 2, wherein Before applying the AI model, it also includes setting capacity classification and capacity grading - transaction rules; Obtaining the driver's behavioral time series data to determine spatio-temporal characteristics; Using a clustering algorithm in the AI model to fuse spatio-temporal characteristics, dividing the basic capacity groups, and setting the capacity group levels; Obtaining the driver's real-time behavior data for dynamic updating of capacity grouping and levels; Analyzing the GPS trajectories in the driver's time series data to identify abnormal behaviors; Designing dynamic credit score rules to deduct credit points for abnormal behaviors and add credit points for continuous compliance behaviors to evaluate the capacity credit score; Designing corresponding trading strategies to be triggered according to the mapping relationship between the credit score and the capacity risk level.

5. The appointment-based vehicle-cargo matching method applying an AI model according to claim 4, characterized in that, During the intelligent matching process, it also includes: Identifying the transportation level of the transportation plan, determining the lowest capacity group level suitable for the transportation level, and excluding reserved drivers whose capacity group level is lower than the lowest capacity group level; Detecting the capacity group level of the target driver and matching the trading strategy corresponding to the capacity group level.

6. The appointment-based vehicle-cargo matching method applying an AI model according to claim 1, wherein The real-time follow-up by the AI digital human includes: Real-time analyzing the deviation between the vehicle trajectory and the planned route, dynamically adjusting the warning threshold in combination with the estimated arrival time algorithm, and predicting the impact of route deviation on the delivery time; Dynamically predicting the arrival time in combination with multi-dimensional data such as real-time road conditions, distance measurement, and vehicle status; Triggering a warning when it is detected that the driver spends more time than the preset threshold in the set link; Triggering a warning when the driver fails to complete the task as planned.

7. The appointment-based vehicle-cargo matching method using an AI model according to claim 1, characterized in that, It also includes: Obtaining user demand information, preprocessing the user demand information, and extracting key information; Input key information into the intent classification model to output the user intent, where the user intent includes providing goods sources, seeking transportation capacity, and booking carriage.

8. A reservation-based vehicle-cargo matching device applying an AI model, characterized in that, It includes: A memory and a processor, where the memory is used to store information including program instructions, and the processor is used to control the execution of the program instructions. It is characterized in that when the program instructions are loaded and executed by the processor, it implements an appointment-based vehicle-cargo matching method using an AI model according to any one of claims 1 to 5.

9. A computer storage medium, the storage medium comprising a stored program, characterized in that, The processor executes the program to implement an appointment-based vehicle-cargo matching method using an AI model according to any one of claims 1 to 5.

Images