Online car-hailing order matching method and system capable of dynamically adjusting pick-up range

Abstract

The invention provides an online car-hailing order matching method and system capable of dynamically adjusting a pick-up range, and the method comprises the steps: taking the pick-up range as a distance constraint parameter for limiting the formation of a matching relation between an order and a driver, and carrying out the screening of potential matching relations according to the pick-up range; in a driving range decision-making period of the coarse time granularity, based on a platform supply and demand state, historical operation information and a business constraint condition, decision-making input is constructed, reasoning is performed through a large language model, and the driving range is dynamically adjusted; and in an order dispatching period with relatively fine time granularity, under the constraint of the receiving range, completing a one-to-one matching decision between an order and a driver based on a matching optimization algorithm, and executing order dispatching. According to the invention, by introducing the pick-up range as a dynamically adjustable decision variable, adaptive control of the order matching space scale and structure is realized, so that the platform can flexibly balance the order completion rate, the driver pick-up cost and the passenger waiting experience under different supply and demand conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and travel service scheduling technology, specifically to a ride-hailing order matching method and system with dynamically adjustable pick-up range. Background Technology

[0002] With the development of mobile internet and shared mobility models, ride-hailing platforms have become an important part of the urban transportation system. These platforms need to make real-time matching decisions between orders and drivers in a dynamic environment where orders are constantly arriving and driver locations and statuses are constantly changing. Their dispatch efficiency directly impacts passenger travel experience, driver operating income, and the overall operational effectiveness of the platform. In practice, order demand and driver supply exhibit significant imbalances in both time and space, with marked differences in supply and demand structures across different time periods and regions. This places high demands on the adaptability and stability of order matching strategies.

[0003] Chinese patent document CN119887316A discloses an order processing method, apparatus, computer equipment, and storage medium based on ride-hailing capacity pre-matching, comprising: responding to a ride-hailing passenger's capacity pre-matching request and receiving the passenger's origin and destination information; obtaining the passenger's surrounding available capacity information based on the origin information; determining whether the pre-matching rules are met based on the surrounding available capacity information and the current time; if so, determining one or more target vehicles based on the origin and destination information, the current time, the surrounding available capacity information, the driver's willingness to accept orders in the surrounding available capacity information, and the driver's historical order rejection information; creating a virtual order based on the one or more target vehicles; and sending the virtual order to the ride-hailing passenger's user terminal.

[0004] Existing ride-hailing platforms typically use fixed pick-up ranges during order matching to limit the maximum distance a driver is allowed to travel to the passenger's pick-up point. While this method simplifies the scale of the matching problem, it has significant shortcomings in complex and dynamic scenarios: when the pick-up range is set too small, it can easily lead to orders not being matched in a timely manner during high-demand or supply-demand imbalance scenarios; when the pick-up range is set too large, it increases the driver's pick-up distance and the passenger's waiting time, reducing overall service efficiency. Some methods attempt to adjust the pick-up range through empirical rules or manual adjustments, but these are difficult to adaptively adjust according to real-time supply and demand conditions and lack a systematic modeling and optimization mechanism.

[0005] Therefore, there is an urgent need for a ride-hailing order matching method and system that can adjust the range of drivers in a dynamic supply and demand environment and integrate it with the order matching process, so as to improve the overall scheduling performance of the platform in complex operating scenarios. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the purpose of this invention is to provide a ride-hailing order matching method and system with dynamically adjustable pick-up range.

[0007] A ride-hailing order matching method with dynamically adjustable pick-up range according to the present invention includes the following steps: Step S1: Obtain the order status information and driver status information of the ride-hailing platform at the current operating time. The order status information includes the order origin location and the order arrival time. The driver status information includes the driver's current location and service status. Step S2: Based on the order status information and driver status information, construct a potential matching relationship between the order node and the driver node, wherein the pick-up range is used as a distance constraint to determine whether a matching relationship can be formed between the order node and the driver node; Step S3: Within a decision cycle with a coarser time granularity, based on the platform's current supply and demand status, historical operating information, and business constraints, construct decision inputs for large language model reasoning, and call the large language model to dynamically adjust the connection range, generating corresponding executable connection range parameters; Step S4: Within a finer time granularity dispatch cycle, potential matching relationships are filtered according to the pick-up range parameters to generate an executable matching structure that satisfies the pick-up range constraints. Based on the matching optimization algorithm, a one-to-one matching decision between the order and the driver is made on the executable matching structure.

[0008] Preferably, step S1 includes: Step S1.1: Collect order origin location, order arrival time, and order service attribute information from the ride-hailing platform's operating system; Step S1.2: Collect real-time location information, service status information, and historical service capacity parameters of idle drivers from the driver dispatch system; Step S1.3: Perform time alignment and spatial unification processing on the order information and driver information to generate basic state inputs for constructing potential matching relationships between orders and drivers.

[0009] Preferably, step S2 includes: Step S2.1: Based on order status information and driver status information, establish a set of order nodes and a set of driver nodes, and assign a unique identifier to each order node and driver node; Step S2.2: For each order node and each driver node, calculate the candidate pick-up distance based on their spatial location; Step S2.3: Generate a set of potential matching edges between order nodes and driver nodes based on the candidate pick-up distance, and organize the set of potential matching edges into a bipartite graph data structure.

[0010] Preferably, step S3 includes: Step S3.1: Within the preset pick-up range decision period, extract supply and demand statistical features for adjusting the pick-up range from the platform operation data. The supply and demand statistical features include the number of orders to be matched, the number of idle drivers, the regional supply and demand ratio, the average pick-up distance statistics, and the average waiting time statistics. Step S3.2: Normalize and summarize the supply and demand statistical characteristics and historical order dispatch information to form a status input for adjusting the pick-up range, and load platform business constraint parameters; based on preset prompt word engineering rules, organize the status input and business constraint parameters into prompt word input for large language model inference; Step S3.3: Based on the input of the prompt word, call the large language model to perform inference, generate the pick-up range adjustment result, and output the executable pick-up range parameters after constraint verification.

[0011] Preferably, step S4 includes: Step S4.1: Based on the executable pick-up range parameter, perform distance threshold filtering on the potential matching edge set, retain the potential matching edges whose candidate pick-up distance does not exceed the pick-up range, and obtain the executable matching edge set that satisfies the pick-up range constraint; Step S4.2: Construct an executable bipartite graph based on the executable matching edge set, and calculate the matching weights for the matching edges in the executable bipartite graph; Step S4.3: Perform matching optimization on the executable bipartite graph to obtain a one-to-one matching result between the order and the driver, and send the matching result to the dispatch execution module to complete the order assignment.

[0012] A ride-hailing order matching system with dynamically adjustable pick-up range according to the present invention includes: The status awareness and preprocessing module is used to collect order status information and driver status information during the operation of the ride-hailing platform. It performs time synchronization and spatial unified processing of order origin location, order arrival time, driver current location and driver service status, and generates basic status data for subsequent matching and decision-making. The matching structure construction module is used to calculate the candidate pick-up distance between the order and the driver based on the basic state data, construct the potential matching relationship between the order node and the driver node, and organize the potential matching relationship into a matching structure that can be filtered and solved. The dynamic adjustment decision module for pick-up range is used to construct a large language model inference input based on the platform's supply and demand status characteristics, historical order dispatching information and business constraints within a preset pick-up range decision period, and call the large language model to dynamically adjust the pick-up range, generating executable pick-up range parameters that take effect in subsequent order dispatching periods. The order matching and dispatching module is used to filter potential matching relationships based on the executable pick-up range parameters within the dispatching cycle, construct an executable matching structure that satisfies the pick-up range constraints, and solve the one-to-one matching result between the order and the driver based on the matching optimization algorithm on the executable matching structure, thereby completing the order assignment.

[0013] Preferably, the state perception and preprocessing module includes: The order status collection unit is used to collect the starting location information, order arrival time, and order service attribute information of the orders to be matched; The driver status acquisition unit is used to collect real-time location information, driver service status, and order acceptance capability information of idle drivers. The status alignment and verification unit is used to perform time synchronization and spatial unification processing of order status information and driver status information, and to filter and verify abnormal status data according to preset rules.

[0014] Preferably, the matching structure construction module includes: The node construction unit is used to create a set of order nodes and a set of driver nodes, and to assign a unique identifier to each order node and driver node. The distance calculation unit is used to calculate the candidate pick-up distance between the order node and the driver node; The matching relationship generation unit is used to generate potential matching relationships based on the candidate meeting distances and organize them into a matching relationship structure in the form of a bipartite graph.

[0015] Preferably, the dynamically adjustable pick-up range decision module includes: The supply and demand characteristic statistics unit is used to collect supply and demand characteristic information of the platform's operating status during the decision-making period of the receiving area; The adjustment input construction unit is used to normalize and combine the supply and demand characteristic information and historical order dispatch operation information, and to construct the large language model prompt word input based on the prompt word engineering rules; The pick-up range decision unit is used to call the large language model for reasoning, generate the pick-up range adjustment result, and output the executable pick-up range parameters after business constraint verification.

[0016] Preferably, the order matching and dispatching module includes: The matching relationship filtering unit is used to filter potential matching relationships based on the executable access range parameter using a distance threshold, and generate executable matching relationships that meet the constraints. An executable structural building unit is used to construct an executable bipartite graph based on executable matching relationships and to calculate matching weights for matching edges. The matching solution and order dispatching unit is used to perform matching optimization on the executable bipartite graph, obtain one-to-one matching results, and complete order dispatch.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention introduces the pick-up range as a dynamically adjustable decision variable and uses it as an important constraint to limit the matching relationship between orders and drivers. It can adaptively adjust the order matching structure under supply and demand scenarios in different times and regions, avoiding the problem of matching efficiency imbalance caused by a fixed pick-up range. 2. This invention dynamically adjusts the pick-up range based on the platform's supply and demand status, historical order dispatch information, and business constraints within a decision-making cycle with a relatively coarse time granularity. This enables the expansion of the candidate driver range in high-demand or supply-demand imbalance scenarios, and the effective control of the pick-up distance in supply-demand balanced scenarios, thereby balancing order completion rate and driver pick-up cost. 3. Within a dispatch cycle with finer time granularity, this invention constructs an executable matching structure and completes one-to-one matching decisions under the constraint of the pick-up range, reducing the introduction of invalid matching relationships, lowering the complexity of matching calculations, and improving the real-time response capability and operational stability of the system. 4. By organically combining the pick-up range adjustment process with the order matching process, this invention can operate stably in a dynamic supply and demand environment, thereby improving the overall scheduling efficiency and service quality of ride-hailing platforms in multi-time and multi-regional operation scenarios. Attached Figure Description

[0018] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of the ride-hailing order matching method with dynamically adjustable pick-up range in an embodiment of the present invention; Figure 2 This is a schematic diagram of the pick-up range decision module based on prompting engineering in an embodiment of the present invention; Figure 3 This is a schematic diagram of ride-hailing order matching in an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0020] like Figures 1-3As shown, this invention provides a method and system for matching ride-hailing orders with dynamically adjustable pick-up range. Specifically, the method includes the following steps: Step S1: Obtain the order status information and driver status information of the ride-hailing platform at the current operating time. The order status information includes the order origin location and the order arrival time, and the driver status information includes the driver's current location and service status. Step S2: Based on order status information and driver status information, construct the potential matching relationship between order nodes and driver nodes, and use the pick-up range as a distance constraint to limit whether a matching relationship can be formed between order nodes and driver nodes; Step S3: Within a decision cycle with a coarser time granularity, based on the platform's current supply and demand status, historical operating information, and business constraints, construct decision inputs for large language model reasoning, and call the large language model to dynamically adjust the connection range, generating corresponding executable connection range parameters; Step S4: Within a finer time granularity dispatch cycle, potential matching relationships are filtered based on the pick-up range parameter to generate an executable matching structure that satisfies the pick-up range constraint. Based on the executable matching structure, a one-to-one matching decision between the order and the driver is made using a matching optimization algorithm.

[0021] Step S1 includes the following:

[0022] Step S1.1: Collect the current running time from the ride-hailing platform's operating system. The order status information is collected and organized into an order set. Order collection Each order status element Represented as , Indicates the first The starting position of each order. Indicates the order arrival time. This indicates the service attribute information of the order; This indicates the number of orders to be matched. The collection results of step S1.1 correspond to the functional requirement of "collecting the order's starting location, order arrival time, and order service attribute information".

[0023] Step S1.2: Collect the current running time from the driver dispatching system. The driver status information is collected, and the driver status information that meets the dispatch conditions is organized into a driver set. Drivers gather Each driver state element Represented as , Indicates the first The current location of each driver. Indicates the driver's service status. This indicates the driver's historical service capacity or order-accepting capacity; the number of orders to be matched. Number of available drivers The data collection results of step S1.2 correspond to the functional requirement of "collecting real-time location information, service status information, and historical service capacity parameters of idle drivers".

[0024] Step S1.3: Process the order set obtained in step S1.1 The driver set obtained in step S1.2 Time alignment and spatial unification are performed. In the time dimension, the order dispatch execution time is used. As a unified time benchmark, order status information and driver status information are time-aligned to ensure that order information and driver information used for the same round of dispatch decisions have a consistent time reference; in the spatial dimension, the order's starting position is... Current location of the driver Mapped to a unified spatial reference frame Spatial reference frame It uses one of the urban road network coordinate systems or a predefined discrete grid coordinate system to eliminate spatial representation differences between different data sources; after completing time alignment and spatial unification processing, it performs validity verification and abnormal state filtering on order information and driver information, and finally generates basic state inputs for constructing potential matching relationships between orders and drivers.

[0025] Step S2 includes the following:

[0026] Step S2.1: Based on the basic state input generated in Step S1, establish a set of order nodes and a set of driver nodes, and assign unique identifiers to each order node and driver node. Specifically, the order set obtained in Step S1 is denoted as... ,in Indicates the current order dispatch time The number of valid orders participating in the matching process after status verification, and the status element of each order. Mapped to an order node The set of drivers obtained in step S1 is denoted as... ,in Indicates the current order dispatch time The number of valid drivers who are in an order-accepting state after status verification, and the status element of each driver. Mapped to a driver node To distinguish different nodes, a unique identifier is assigned to each order node and driver node when constructing the node set, where the order node set is denoted as . The set of driver nodes is denoted as .

[0027] Step S2.2: For each order node and each driver node established in step S2.1, calculate the candidate pick-up distance based on their corresponding spatial location. Specifically, let the order node... The corresponding order starting position is Driver node The corresponding driver's current location is The distance to the candidate pick-up point is recorded as follows. Candidate pick-up distance The distance can be calculated using the shortest path distance of the road network, or by using approximate road network distance calculation methods such as Euclidean distance and Manhattan distance. In engineering implementation, different distance calculation methods can be selected according to real-time requirements, and the distance calculation results can be cached or approximated to reduce the overall computational complexity.

[0028] Step S2.3: Generate a set of potential matching edges between order nodes and driver nodes based on the candidate pick-up distances obtained in Step S2.2. Specifically, for each order node... With driver node Construct a potential matching edge and the distance of the candidate pick-up This serves as the associated attribute of the potential matching edge; the resulting set of potential matching edges is denoted as... Furthermore, the order node set... Driver node set and the set of potential matching edges The organization is a bipartite diagram structure. The order node and driver node are located on opposite sides of the bipartite graph, forming a bipartite graph structure. This serves as the input data structure for subsequent pick-up range filtering and order matching.

[0029] Step S3 includes the following:

[0030] Step S3.1: Decision-making cycle based on the preset pick-up range according to the frequency of changes in platform supply and demand and the stability of order dispatch. Within this framework, supply and demand statistical features for adjusting pick-up ranges are extracted from ride-hailing platform operational data, and a structured state description required for large language model inference is constructed. Specifically, during the decision-making cycle... The covered multiple dispatching times The platform's operational status is statistically summarized to obtain a set of supply and demand statistical characteristics. The supply and demand statistical characteristics include at least the number of orders to be matched. Number of available drivers Regional supply and demand ratio Average pick-up distance statistics and average order wait time statistics The aforementioned supply and demand statistical features are used to characterize the overall supply and demand structure and operational pressure of the platform within the current decision-making cycle, and serve as core state information for subsequent prompt word construction and reception range inference.

[0031] Step S3.2: Normalize, summarize, and semantically organize the supply and demand statistical features and historical order dispatch information to construct prompt words for large language model inference and load platform business constraint parameters. Specifically, first, the supply and demand statistical features... The values ​​of the reception range and corresponding operational indicators in the historical decision-making cycle are normalized to form a structured state vector. Subsequently, the state vector Historical pick-up range adjustment records and platform business constraint parameters Joint organization for prompt word input Among them, business constraint parameters At least include the upper and lower limits of the pick-up and drop-off area. Single adjustment step size Update frequency constraints and service quality-related constraints. Prompt input. It is constructed using predefined templates, and guides the large language model to reason about the interface range adjustment strategy by combining natural language and structured fields, while meeting business constraints.

[0032] Step S3.3: Input based on prompt words The large language model is invoked for inference generation to obtain the candidate pick-up range adjustment results, which are then mapped to executable pick-up range parameters. Specifically, the large language model receives prompt word input. Subsequently, based on its inherent reasoning ability and historical pattern learning results, the system outputs a reasoning result that includes the direction and magnitude of the pick-up range adjustment. The system then parses and verifies the reasoning result to ensure that it meets the business constraint parameters. The upper and lower limits and adjustment rules in the parameters are used to ultimately generate the executable pick-up range parameters. Executable pick-up range parameters Decision-making cycle within the current pick-up area The value remains unchanged and applies to the matching and filtering process at multiple dispatch times within that cycle, until the next decision cycle begins. The pick-up range adjustment will be retried at that time.

[0033] Step S4 includes the following:

[0034] Step S4.1: Based on the executable pick-up range parameters output in step S3 The potential matching edge set constructed in step S2 is filtered by a distance threshold to generate an executable matching edge set that satisfies the pick-up range constraint. Specifically, assuming that at the dispatch execution time... The set of potential matching edges constructed is Each potential matching edge The corresponding candidate pick-up distance is Based on the pick-up range parameters Each potential matching edge is evaluated, and only those that satisfy the condition are retained. Potential matching edges are selected and grouped into an executable matching edge set. Through the above screening process, the pick-up range constraint is explicitly introduced into the matching structure construction process, thereby effectively controlling the size of the matching search space during the dispatching phase.

[0035] Step S4.2: Based on the set of executable matching edges Construct an executable bipartite graph and compute matching weights for matching edges in the executable bipartite graph. Specifically, using the set of order nodes... With driver node set Each is represented as a set of nodes on one side of a bipartite graph, and the other as a set of executable matching edges. As a set of edges, construct an executable bipartite graph. For each matching edge Calculate its corresponding matching weight Matching weight This is used to comprehensively represent factors such as order revenue, driver pick-up distance cost, and passenger waiting time cost, and can be expressed as a linear weighted or nonlinear combination of these factors. In one specific implementation, the matching weight... It can be represented as:

[0036] in, Indicates order The corresponding order revenue or service value, Indicates driver Pick-up orders The distance or time cost of picking up the driver Indicates order The cumulative waiting time , , These are non-negative weighting coefficients used to balance different optimization objectives.

[0037] By introducing matching weights, the matching relationships between different orders and drivers become comparable in the executable bipartite graph, providing a quantitative basis for subsequent matching optimization solutions.

[0038] Step S4.3: In the executable bipartite diagram The system performs matching optimization to obtain a one-to-one match between orders and drivers. Specifically, the matching optimization can be modeled as a matching optimization problem on a weighted bipartite graph. Under the constraints that each order node can match at most one driver node and each driver node can serve at most one order, the system seeks the matching scheme with the optimal overall matching weight. In one implementation, maximum weight matching or minimum cost matching algorithms can be used to solve the executable bipartite graph. In another implementation, to adapt to large-scale online order dispatch scenarios, greedy matching, partitioned parallel matching, or approximate matching algorithms can be combined to reduce computational complexity. After obtaining the one-to-one match between orders and drivers, the matching results are sent to the order dispatch execution module during the order dispatch process. Complete order assignment.

[0039] This embodiment primarily addresses the challenge of balancing matching efficiency and service costs faced by ride-hailing platforms in dynamic supply and demand environments when matching orders. It proposes a ride-hailing order matching method with dynamically adjustable pick-up ranges. On one hand, by uniformly modeling order and driver status information and introducing the pick-up range as a crucial constraint limiting the matchability between orders and drivers, effective control over the scale and form of potential matching structures is achieved. On the other hand, within a coarser-grained time-range decision-making cycle for the pick-up range, the method dynamically adjusts the pick-up range based on platform supply and demand characteristics, historical dispatch information, and business constraints. This allows the range to adapt to different operational scenarios and remain stable across multiple dispatch periods. In operational tests across various urban areas and time periods, this method significantly improves key indicators such as order completion rate, average pick-up distance, and passenger waiting time compared to dispatch strategies with fixed pick-up ranges. It exhibits particularly better stability and scheduling efficiency in scenarios with supply and demand imbalances or during peak hours. Although this embodiment is mainly designed for ride-hailing order matching scenarios, the matching modeling and decision-making framework based on dynamic adjustment of pick-up range is also applicable to other scheduling systems that need to make resource allocation and matching decisions in a dynamic supply and demand environment.

[0040] The present invention also provides a ride-hailing order matching system with dynamically adjustable pick-up range. The ride-hailing order matching system with dynamically adjustable pick-up range can be implemented by executing the process steps of the ride-hailing order matching method with dynamically adjustable pick-up range. That is, those skilled in the art can understand the ride-hailing order matching method with dynamically adjustable pick-up range as a preferred embodiment of the ride-hailing order matching system with dynamically adjustable pick-up range.

[0041] This invention provides a ride-hailing order matching system with dynamically adjustable pick-up range, comprising the following modules: The status awareness and preprocessing module is used to collect order status information and driver status information during the operation of the ride-hailing platform, obtain the order origin location, order arrival time, driver current location and driver service status, and perform time synchronization and spatial unified processing on the order status information and driver status information to generate basic status data for subsequent matching and decision-making. The matching structure construction module is used to calculate the candidate pick-up distance between orders and drivers based on basic state data, construct the potential matching relationship between order nodes and driver nodes, and organize the potential matching relationship into a matching structure that can be filtered and solved. The pick-up range dynamic adjustment decision module is used to construct a large language model inference input based on the platform's supply and demand status characteristics, historical order dispatching information and business constraints within a preset pick-up range decision period, and call the large language model to dynamically adjust the pick-up range, generating executable pick-up range parameters that take effect in subsequent order dispatching periods. The order matching and dispatching module is used to filter potential matching relationships based on the executable pick-up range parameter within the dispatching cycle, construct an executable matching structure that meets the pick-up range constraint, and solve the one-to-one matching result between the order and the driver based on the matching optimization algorithm on the executable matching structure, thereby completing the order assignment.

[0042] The state awareness and preprocessing module includes the following units, which are used to implement the functions corresponding to steps S1.1 to S1.3 at the system level.

[0043] The order status acquisition unit is used to obtain the order status information of orders to be matched from the ride-hailing platform's order system during the dispatching process, and organize the order status information into a data structure that the system can process. The order status information corresponds to the current dispatching time. order collection ,in Each order status element At least include the order's origin location information. Order arrival time and order service attribute information The order status acquisition unit is used to provide order status input in a unified format to subsequent modules to support the construction of order nodes and the generation of matching relationships.

[0044] The driver status acquisition unit is used to obtain the status information of idle drivers who meet the dispatch conditions from the driver dispatch system, and organize the driver status information into a unified data representation within the system. The driver status information corresponds to the current dispatch execution time. driver collection ,in Each driver state element At least include the driver's current location Driver service status And order acceptance capacity information The driver status acquisition unit is used to provide standardized driver status input for candidate pick-up distance calculation and matching structure construction.

[0045] The status alignment and verification unit performs system-level alignment and verification processing on the order status information and driver status information output by the order status acquisition unit and driver status acquisition unit to generate basic status data that meets subsequent decision-making requirements. Specifically, the status alignment and verification unit synchronizes the order status information and driver status information in the time dimension, ensuring they correspond to the same dispatch execution time. In terms of spatial dimension, the starting point of the order is... Current location of the driver Mapped to a unified spatial reference frame This eliminates the spatial representation differences between different data sources; simultaneously, the state alignment and verification unit filters abnormal or invalid state data according to preset validity discrimination rules, and outputs the filtered order set. Meet with the driver , which serves as the basic state input in the system for constructing potential matching relationships and dynamically adjusting the range of execution.

[0046] The matching structure building module includes the following units, which are used to build the potential matching structure between orders and drivers at the system level.

[0047] The node construction unit is used to establish order node sets and driver node sets within the system based on the basic state input output from the state awareness and preprocessing module. Specifically, the node construction unit is used to process the filtered order set... Each order status element in the process is mapped to an order node. And the filtered drivers are collected. Each driver state element in the algorithm is mapped to a driver node. This results in the formation of separate order node sets. With driver node set At the same time, a unique identifier is assigned to each order node and driver node to facilitate the differentiation and indexing of different nodes during the subsequent matching relationship construction and matching solution process.

[0048] The distance calculation unit is used to calculate the candidate pick-up distance based on the spatial location information of the order node and the driver node. Specifically, the distance calculation unit is used to calculate the distance based on the order node... Corresponding order starting position With driver node The corresponding driver's current location Calculate the candidate pick-up distance between the order node and the driver node. Among them, the distance to the candidate pick-up point It can be the shortest path distance based on the urban road network, or a spatial distance metric such as Euclidean distance or Manhattan distance used to approximate road network distances. The distance calculation results are stored inside the system as important attribute information for the subsequent generation of potential matching relationships.

[0049] The matching relationship generation unit is used to construct potential matching relationships between order nodes and driver nodes based on the candidate pick-up distances calculated by the distance calculation unit. Specifically, the matching relationship generation unit is used to generate matching relationships for each order node. With driver node Generate a potential matching relationship And organize potential matches into a set of matches. Furthermore, the matching relationship generation unit will combine the order node set... Driver node set and potential matching set The matching relationship structure is uniformly organized into a bipartite graph form. The bipartite graph matching relationship structure serves as the basic data structure in the system for filtering the pick-up range and solving order matching.

[0050] The dynamically adjustable pick-up range decision module includes the following units, which are used to realize the dynamically adjustable pick-up range decision at the system level.

[0051] Supply and demand characteristic statistical unit, corresponding to Figure 2 The scene parameter module is used for decision-making within the pick-up range. The system performs statistical analysis on the operational status of ride-hailing platforms to generate basic supply and demand characteristic information for adjusting pick-up and drop-off areas. Specifically, the supply and demand characteristic statistical unit is used in the decision-making cycle. The covered multiple dispatching times The platform's operational data is aggregated and statistically analyzed to form a set of supply and demand characteristics. Supply and demand characteristics set At least include the number of orders to be matched. Number of available drivers Regional supply and demand ratio Average pick-up distance statistics and average order wait time statistics At least one of these is used to characterize the platform's supply and demand environment and operational status within the current decision-making cycle.

[0052] Adjusting the input building blocks, corresponding to Figure 2The historical similarity case module, recent decision record module, and indicator definition and weight module are used to construct, at the system level, the range adjustment decision input for large language model inference. Specifically, the adjustment input construction unit is used to receive the supply and demand feature set. Based on this, and further combining historical case information similar to the current scenario from historical dispatch data, as well as recent pick-up range decision records and their corresponding operational effects, the above information is normalized, summarized, and processed with feature combination to form a state representation within the system. Simultaneously, the adjustment input construction unit is used to load and organize preset indicator definitions and weight configurations to describe the relative importance of different operational indicators during the adjustment process, thereby guiding subsequent decision-making. State Representation Historical case information, recent decision records, and indicator weight configurations are organized into structured fields and constructed into prompt input for large language model inference using predefined prompt word templates. .

[0053] The decision-making unit for the scope of receiving the emperor corresponds to... Figure 2 The optimization rule module and decision result output module are used to generate and output executable pick-up range parameters based on the prompt word input. Specifically, the pick-up range decision unit is used to call the large language model to process the prompt word input. The process involves reasoning to generate candidate pick-up range adjustments. Subsequently, the candidate results are constrained and mapped using pre-defined business optimization rules, which include at least upper and lower limits for the pick-up range. Single adjustment step size constraint Update frequency constraints and service quality-related constraints to generate a set of parameters that satisfy business constraints. Required executable pick-up range parameters ; Executable pick-up range parameters Decision-making cycle within the current pick-up area The data remains valid and serves as a unified parameter input for matching relationships in subsequent dispatch cycles until it is re-triggered and updated by the dynamically adjustable pick-up range decision module when entering the next decision cycle.

[0054] The order matching and dispatching module includes the following units, which are used to implement the order matching and dispatching functions at the system level.

[0055] The matching relationship filtering unit is used to dynamically adjust the executable pick-up range parameters output by the decision module based on the pick-up range. The system filters potential matching relationships constructed by the dynamically adjustable pick-up range decision module using distance thresholds, generating executable matching relationships that satisfy the pick-up range constraints. Specifically, the matching relationship filtering unit receives data at the dispatch execution time. Constructed potential matching set Each potential match The corresponding candidate pick-up distance is The matching relationship filtering unit is based on the pick-up range parameter. Each potential match is evaluated individually, and only those that satisfy the criteria are retained. Potential matching relationships are then grouped into a set of executable matching relationships. This explicitly introduces a range constraint at the system level, limiting the search space size for subsequent matching solutions.

[0056] An executable structure building unit is used to construct an executable bipartite graph structure based on executable matching relationships and to calculate matching weights for the matching relationships in the bipartite graph. Specifically, the executable structure building unit is used to construct an executable bipartite graph structure based on an order node set. With driver node set As the set of nodes on both sides of a bipartite graph, and as the set of executable matching relationships. As a set of edges, construct an executable bipartite graph. At the same time, for each matching relationship Calculate its corresponding matching weight Matching weight It is used to comprehensively represent factors such as order revenue, driver pick-up distance cost, and passenger waiting time cost, and can be expressed as a linear weighted or nonlinear combination of the above factors, so as to provide a unified quantitative evaluation basis for subsequent matching solutions.

[0057] The matching solution and order dispatching unit is used to perform matching optimization solutions and complete order dispatching on the executable bipartite graph. Specifically, the matching solution and order dispatching unit is used to perform matching optimization solutions and complete order dispatching on the executable bipartite graph. Above, under the constraints that each order node can match at most one driver node and each driver node can serve at most one order, a matching optimization solution is executed to obtain a one-to-one matching result between orders and drivers. The matching optimization solution can be implemented using a weighted bipartite graph matching algorithm, an approximate matching algorithm, or a heuristic matching strategy to meet the real-time requirements of online order dispatch. After obtaining the matching result, the matching solution and order dispatch execution unit sends the matching result to the order dispatch execution interface during the order dispatch process. Complete the order assignment, thereby determining the parameters within the pick-up area. Under the constraints, stable and efficient order matching and execution can be achieved.

[0058] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, making the system and its various devices, modules, and units of this invention function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered as a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered as structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0059] In the description of this application, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this application.

[0060] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method for matching ride-hailing orders with a dynamically adjustable pick-up range, characterized in that, Includes the following steps: Step S1: Obtain the order status information and driver status information of the ride-hailing platform at the current operating time. The order status information includes the order origin location and the order arrival time. The driver status information includes the driver's current location and service status. Step S2: Based on the order status information and driver status information, construct a potential matching relationship between the order node and the driver node, wherein the pick-up range is used as a distance constraint to determine whether a matching relationship can be formed between the order node and the driver node; Step S3: Within a decision cycle with a coarser time granularity, based on the platform's current supply and demand status, historical operating information, and business constraints, construct decision inputs for large language model reasoning, and call the large language model to dynamically adjust the connection range, generating corresponding executable connection range parameters; Step S4: Within a finer time granularity dispatch cycle, potential matching relationships are filtered according to the pick-up range parameters to generate an executable matching structure that satisfies the pick-up range constraints. Based on the matching optimization algorithm, a one-to-one matching decision between the order and the driver is made on the executable matching structure.

2. The ride-hailing order matching method with dynamically adjustable pick-up range according to claim 1, characterized in that, Step S1 includes: Step S1.1: Collect order origin location, order arrival time, and order service attribute information from the ride-hailing platform's operating system; Step S1.2: Collect real-time location information, service status information, and historical service capacity parameters of idle drivers from the driver dispatch system; Step S1.3: Perform time alignment and spatial unification processing on the order information and driver information to generate basic state inputs for constructing potential matching relationships between orders and drivers.

3. The ride-hailing order matching method with dynamically adjustable pick-up range according to claim 1, characterized in that, Step S2 includes: Step S2.1: Based on order status information and driver status information, establish a set of order nodes and a set of driver nodes, and assign a unique identifier to each order node and driver node; Step S2.2: For each order node and each driver node, calculate the candidate pick-up distance based on their spatial location; Step S2.3: Generate a set of potential matching edges between order nodes and driver nodes based on the candidate pick-up distance, and organize the set of potential matching edges into a bipartite graph data structure.

4. The ride-hailing order matching method with dynamically adjustable pick-up range according to claim 1, characterized in that, Step S3 includes: Step S3.1: Within the preset pick-up range decision period, extract supply and demand statistical features for adjusting the pick-up range from the platform operation data. The supply and demand statistical features include the number of orders to be matched, the number of idle drivers, the regional supply and demand ratio, the average pick-up distance statistics, and the average waiting time statistics. Step S3.2: Normalize and summarize the supply and demand statistical characteristics and historical order dispatch information to form a status input for adjusting the pick-up range, and load platform business constraint parameters; based on preset prompt word engineering rules, organize the status input and business constraint parameters into prompt word input for large language model inference; Step S3.3: Based on the input of the prompt word, call the large language model to perform inference, generate the pick-up range adjustment result, and output the executable pick-up range parameters after constraint verification.

5. The ride-hailing order matching method with dynamically adjustable pick-up range according to claim 1, characterized in that, Step S4 includes: Step S4.1: Based on the executable pick-up range parameter, perform distance threshold filtering on the potential matching edge set, retain the potential matching edges whose candidate pick-up distance does not exceed the pick-up range, and obtain the executable matching edge set that satisfies the pick-up range constraint; Step S4.2: Construct an executable bipartite graph based on the executable matching edge set, and calculate the matching weights for the matching edges in the executable bipartite graph; Step S4.3: Perform matching optimization on the executable bipartite graph to obtain a one-to-one matching result between the order and the driver, and send the matching result to the dispatch execution module to complete the order assignment.

6. A ride-hailing order matching system with dynamically adjustable pick-up range, characterized in that, include: The status awareness and preprocessing module is used to collect order status information and driver status information during the operation of the ride-hailing platform. It performs time synchronization and spatial unified processing of order origin location, order arrival time, driver current location and driver service status, and generates basic status data for subsequent matching and decision-making. The matching structure construction module is used to calculate the candidate pick-up distance between the order and the driver based on the basic state data, construct the potential matching relationship between the order node and the driver node, and organize the potential matching relationship into a matching structure that can be filtered and solved. The dynamic adjustment decision module for pick-up range is used to construct a large language model inference input based on the platform's supply and demand status characteristics, historical order dispatching information and business constraints within a preset pick-up range decision period, and call the large language model to dynamically adjust the pick-up range, generating executable pick-up range parameters that take effect in subsequent order dispatching periods. The order matching and dispatching module is used to filter potential matching relationships based on the executable pick-up range parameters within the dispatching cycle, construct an executable matching structure that satisfies the pick-up range constraints, and solve the one-to-one matching result between the order and the driver based on the matching optimization algorithm on the executable matching structure, thereby completing the order assignment.

7. The ride-hailing order matching system with dynamically adjustable pick-up range according to claim 6, characterized in that, The state awareness and preprocessing module includes: The order status collection unit is used to collect the starting location information, order arrival time, and order service attribute information of the orders to be matched; The driver status acquisition unit is used to collect real-time location information, driver service status, and order acceptance capability information of idle drivers. The status alignment and verification unit is used to perform time synchronization and spatial unification processing of order status information and driver status information, and to filter and verify abnormal status data according to preset rules.

8. The ride-hailing order matching system with dynamically adjustable pick-up range according to claim 6, characterized in that, The matching structure construction module includes: The node construction unit is used to create a set of order nodes and a set of driver nodes, and to assign a unique identifier to each order node and driver node. The distance calculation unit is used to calculate the candidate pick-up distance between the order node and the driver node; The matching relationship generation unit is used to generate potential matching relationships based on the candidate meeting distances and organize them into a matching relationship structure in the form of a bipartite graph.

9. The ride-hailing order matching system with dynamically adjustable pick-up range according to claim 6, characterized in that, The dynamically adjustable pick-up range decision module includes: The supply and demand characteristic statistics unit is used to collect supply and demand characteristic information of the platform's operating status during the decision-making period of the receiving area; The adjustment input construction unit is used to normalize and combine the supply and demand characteristic information and historical order dispatch operation information, and to construct the large language model prompt word input based on the prompt word engineering rules; The pick-up range decision unit is used to call the large language model for reasoning, generate the pick-up range adjustment result, and output the executable pick-up range parameters after business constraint verification.

10. The ride-hailing order matching system with dynamically adjustable pick-up range according to claim 6, characterized in that, The order matching and dispatching module includes: The matching relationship filtering unit is used to filter potential matching relationships based on the executable access range parameter using a distance threshold, and generate executable matching relationships that meet the constraints. An executable structural building unit is used to construct an executable bipartite graph based on executable matching relationships and to calculate matching weights for matching edges. The matching solution and order dispatching unit is used to perform matching optimization on the executable bipartite graph, obtain one-to-one matching results, and complete order dispatch.

Images