A method and system for allocating a ride-hailing order based on an artificial intelligence algorithm

Abstract

The application discloses a kind of based on artificial intelligence algorithm's online car-hailing order distribution method and system, it is related to intelligent scheduling technical field, including obtaining order data and driver data to be distributed;Adopt dynamic search radius and multi-state driver screening strategy to determine candidate driver set, the estimated arrival time is calculated in combination with real-time traffic condition;Establish multi-factor order acceptance willingness evaluation model, calculate the order acceptance probability of each order-driver combination;Build comprehensive matching degree evaluation model, calculate the matching degree of all order-driver combinations, generate matching degree matrix;Global optimization algorithm is handled matching degree matrix, and the order-driver distribution result is output;Order is pushed to the driver in order-driver distribution result and monitored response state, and re-distribution processing is executed to unresponsive condition.The application improves the success rate and matching accuracy of order distribution by multi-factor order acceptance willingness evaluation and global optimization algorithm, reduces order waiting time, and improves overall distribution efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent scheduling, in particular to a ride-hailing order allocation method and system based on artificial intelligence algorithm. BACKGROUND

[0002] With the rapid development of mobile internet technology, ride-hailing services have become an important urban travel mode. The order allocation technology of ride-hailing platforms has gone through a development process from simple rules to intelligent algorithms. Early allocation methods mainly adopt distance-based matching strategies, such as shortest distance allocation, first-come-first-served rules, etc., which can realize basic order-driver matching functions. With the expansion of platform scale, some methods begin to consider factors such as driver state, historical score, etc., and improve the allocation effect through weighted algorithms.

[0003] However, the existing order allocation methods still have some technical problems: in the aspect of driver order acceptance willingness evaluation, most methods fail to comprehensively analyze the historical behavior patterns, current working state and order characteristics of the driver, resulting in inaccurate prediction of order acceptance success rate; in the aspect of global optimization, there is a lack of effective matrix processing and algorithm selection mechanism, making it difficult to achieve optimal allocation in large-scale order scenarios; in the aspect of dynamic processing, there is a lack of systematic processing procedures for re-allocation after allocation failure. SUMMARY

[0004] In view of the above existing problems, the present application is proposed.

[0005] Therefore, the present application provides a ride-hailing order allocation method and system based on artificial intelligence algorithm, which solves the technical problems of inaccurate order acceptance willingness evaluation, poor global optimization effect and imperfect dynamic re-allocation processing in the existing order allocation methods.

[0006] To solve the above technical problems, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a ride-hailing order allocation method based on artificial intelligence algorithm, which comprises:

[0008] Obtain the order data in the order pool and the driver data in the driver pool, and perform data preprocessing;

[0009] Adopt dynamic search radius and multi-state driver screening strategy to determine the candidate driver set for each order to be allocated, and calculate the estimated arrival time of each order-driver combination based on real-time traffic conditions;

[0010] Establish a multi-factor order acceptance willingness evaluation model, calculate the order acceptance probability of each order-driver combination based on the historical service data, current working state and order characteristics of the candidate driver;

[0011] A comprehensive matching degree evaluation model is constructed, and the matching degrees of all order-driver combinations are calculated based on the estimated arrival time and the order acceptance probability to generate a matching degree matrix;

[0012] The matching degree matrix is processed by using a global optimization algorithm, and an order-driver assignment result is output;

[0013] The orders are pushed to the drivers in the order-driver assignment result, and the response state is monitored, and re-assignment processing is performed for unresponsive cases.

[0014] As a preferred scheme of the order allocation method for online car-hailing based on artificial intelligence algorithm, wherein:

[0015] An initial search radius is set with the order starting point coordinates of the order to be allocated as the center, and the final search radius is determined according to the number of drivers within the initial search radius;

[0016] Within the final search radius, hierarchical screening is performed based on the service state of the driver to form a preliminary driver set;

[0017] The upper limit of the number of candidate drivers is determined according to the order priority of the order to be allocated, and a corresponding number of drivers are selected from the preliminary driver set to form a candidate driver set.

[0018] As a preferred scheme of the order allocation method for online car-hailing based on artificial intelligence algorithm, wherein:

[0019] The order starting point coordinates of the order to be allocated and the current coordinates of each candidate driver in the candidate driver set are obtained;

[0020] The path planning starting point is determined based on the service state of the driver, a heuristic path search algorithm is used for path planning, and multiple candidate paths are generated;

[0021] The real-time traffic condition data of each road section on the candidate path is obtained, and the estimated travel time of each candidate path is calculated in combination with the path length;

[0022] For each order-driver combination, the candidate path with the shortest estimated travel time is selected, the state transition time is determined according to the current service state of the candidate driver, and the estimated arrival time of each order-driver combination is calculated.

[0023] As a preferred scheme of the order allocation method for online car-hailing based on artificial intelligence algorithm, wherein:

[0024] Construct a multi-dimensional driver order acceptance willingness evaluation system based on historical behavior factors, order matching factors, work status factors, and fatigue status factors, and set weight coefficients for each factor;

[0025] Obtain historical service data of candidate drivers, extract driver service characteristics, and extract order characteristics based on the order data to be assigned;

[0026] Based on historical order acceptance rates in historical service data, determine historical behavioral factors;

[0027] Calculate the order matching factor based on the degree of matching between driver service characteristics and order characteristics;

[0028] The working status factor is determined based on the driver's service status, and the fatigue status factor is calculated based on the continuous online time in the driver's working status.

[0029] A weighted fusion algorithm is used to calculate the order acceptance probability value for each order-driver combination by weighting each factor according to its weight coefficient.

[0030] As a preferred embodiment of the ride-hailing order allocation method based on artificial intelligence algorithms of the present invention, the construction of a comprehensive matching degree evaluation model includes:

[0031] Obtain the current rating and historical complaint rate of candidate drivers, and calculate the service quality score;

[0032] Based on order distribution characteristics and driver resource status, service quality thresholds, time response upper limits, and order acceptance probability lower limits are set to construct tiered screening conditions.

[0033] A tiered filtering mechanism is adopted to filter order-driver combinations based on tiered filtering conditions, retaining valid matching combinations;

[0034] Based on the estimated arrival time, order acceptance probability, and service quality score, a weighted fusion algorithm is used to calculate the comprehensive matching degree value for each valid matching combination.

[0035] A sparse matrix framework is constructed, and the comprehensive matching degree value is filled into the corresponding order-driver location to generate a matching degree matrix.

[0036] As a preferred embodiment of the ride-hailing order allocation method based on artificial intelligence algorithms of the present invention, the method includes: processing the matching degree matrix using a global optimization algorithm, including:

[0037] Select the appropriate global optimization algorithm based on the dimension and sparsity of the matching degree matrix;

[0038] The matching degree matrix is ​​weighted according to order priority;

[0039] The selected global optimization algorithm is used to process the weighted matching degree matrix, and the order-driver allocation results are output.

[0040] Perform conflict detection and feasibility verification on the order-driver allocation results, identify orders that were not successfully allocated, and mark them as pending reassignment.

[0041] As a preferred embodiment of the ride-hailing order allocation method based on artificial intelligence algorithms of the present invention, the order priority includes high priority, medium priority and low priority.

[0042] Secondly, the present invention provides a ride-hailing order allocation system based on artificial intelligence algorithms, comprising:

[0043] The data acquisition and preprocessing module is used to acquire the order data to be assigned from the order pool and the driver data from the driver pool, and to perform data preprocessing.

[0044] The candidate driver matching module is used to determine a set of candidate drivers for each order to be assigned by adopting a dynamic search radius and a multi-state driver screening strategy, and to calculate the estimated arrival time of each order-driver combination by combining real-time traffic conditions.

[0045] The order acceptance willingness assessment module is used to establish a multi-factor order acceptance willingness assessment model. Based on the candidate driver's historical service data, current work status and order characteristics, it calculates the order acceptance probability of each order-driver combination.

[0046] The comprehensive matching evaluation module is used to build a comprehensive matching evaluation model. Based on the estimated arrival time and order acceptance probability, it calculates the matching degree of all order-driver combinations in the current batch and generates a matching degree matrix.

[0047] The global optimization allocation module is used to process the matching degree matrix using a global optimization algorithm and output the order-driver allocation results;

[0048] The order push monitoring module is used to push orders to drivers in the order-driver assignment results and monitor their response status, and perform reassignment processing for non-response situations.

[0049] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the ride-hailing order allocation method based on artificial intelligence algorithm as described in the first aspect of the present invention.

[0050] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the ride-hailing order allocation method based on artificial intelligence algorithm as described in the first aspect of the present invention.

[0051] The beneficial effects of this invention are as follows: By constructing a multi-factor order acceptance willingness evaluation model, this invention can accurately predict the probability of drivers accepting orders, improve the order allocation success rate, and reduce driver order rejection. Through dynamic search radius and multi-state driver screening strategies, it achieves precise location and efficient screening of candidate drivers, improving driver resource utilization efficiency and system response speed. By combining a heuristic route planning algorithm based on real-time traffic conditions, it provides high-precision arrival time prediction, improving passenger waiting experience and enhancing service quality. Through a global optimization algorithm to process the matching degree matrix, it achieves optimal matching and allocation of orders and drivers, maximizing overall operational efficiency and reducing empty mileage and waiting time. Through closed-loop feedback and reassignment mechanisms, it ensures the timely processing of unresponsive orders, improving system stability and reliability, and ensuring that passenger travel needs are effectively met. Attached Figure Description

[0052] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0053] Figure 1 This is a flowchart of a ride-hailing order allocation method based on artificial intelligence algorithms.

[0054] Figure 2 A flowchart for constructing a multi-factor order acceptance willingness assessment model for ride-hailing order allocation based on artificial intelligence algorithms.

[0055] Figure 3 This is a flowchart of the global optimization algorithm for ride-hailing order allocation based on artificial intelligence algorithms.

[0056] Figure 4 This is a module connection diagram of a ride-hailing order allocation system based on artificial intelligence algorithms. Detailed Implementation

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0058] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0059] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0060] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a ride-hailing order allocation method based on artificial intelligence algorithms, the flowchart of which is shown below. Figure 1 As shown, the method includes the following steps:

[0061] S1: Obtain the order data to be assigned from the order pool and the driver data from the driver pool, and perform data preprocessing.

[0062] Specifically, the order data to be assigned includes the order number, order start coordinates, order end coordinates, order generation time, and order priority. Driver data includes the driver number, driver's current coordinates, driver service status, and driver work status. Driver work status includes continuous online time, and driver service status includes idle, on the way to pick up passengers, carrying passengers, about to complete an order, and resting. The statuses of on the way to pick up passengers, carrying passengers, and about to complete an order are determined based on a comprehensive assessment of GPS location data, order execution progress, and driver behavior patterns.

[0063] Furthermore, order priority is determined through a dynamic scoring mechanism, which comprehensively considers user-dimensional factors (user level, historical ride frequency), order-dimensional factors (estimated fare, trip distance, origin and destination location type), time-dimensional factors (travel time period), and supply and demand-dimensional factors (current regional order density, number of drivers). A weighted fusion algorithm is used to calculate a comprehensive score, and the weight coefficients are adjusted according to the real-time supply and demand situation, ultimately classifying them into high priority, medium priority, and low priority.

[0064] Furthermore, considering the high concurrency of ride-hailing orders, the order pool employs a hybrid triggering mechanism. This mechanism sets processing time and order quantity thresholds. When the number of orders in the pool reaches the order quantity threshold or the processing time threshold is reached since the last processing, all pending orders in the current pool are retrieved as a processing batch. The order quantity and processing time thresholds are determined based on order concurrency and processing performance requirements. During peak periods, the order quantity threshold is appropriately increased to reduce processing frequency, while during off-peak periods, the processing time threshold is shortened to improve response speed, ensuring a high batch triggering frequency to guarantee a good user experience.

[0065] It should be noted that the order pool and driver pool adopt a near real-time update mechanism. The order pool periodically receives newly generated orders and removes orders that have been assigned. The driver pool periodically updates drivers' online / offline status, location changes, and service status changes. During batch processing, the driver's service status is updated immediately after each order assignment to avoid duplicate assignments within the same batch.

[0066] S2: Employs a dynamic search radius and multi-state driver screening strategy to determine a set of candidate drivers for each order to be assigned, and calculates the estimated arrival time of each order-driver combination in conjunction with real-time traffic conditions.

[0067] Specifically, step S2 includes:

[0068] S2.1: Set the initial search radius centered on the starting coordinates of the orders to be assigned, and determine the final search radius based on the number of drivers within the initial search radius.

[0069] The initial search radius is set at 1-3 kilometers, determined based on regional road density and average driver response time to ensure drivers can reach the order origin within a reasonable timeframe. The number of drivers within the initial search radius is counted; if the number of drivers is insufficient, the search radius is expanded; if the number of drivers is excessive, the search radius is reduced, until the search radius reaches its boundary or the number of drivers meets the requirements. The final search radius is then determined, with the boundary range defined based on the operational service area.

[0070] S2.2: Within the final search radius, perform hierarchical filtering based on driver service status to form a preliminary set of drivers.

[0071] In one embodiment, the tiered screening process is as follows: Drivers in an idle state are directly included, and drivers nearing completion of their orders with an estimated completion time less than a preset completion time threshold are also included, forming an initial driver set. The completion time threshold is determined based on the upper limit of order waiting time.

[0072] S2.3: Determine the upper limit of the number of candidate drivers based on the order priority of the orders to be assigned, and select the corresponding number of drivers from the initial driver set to form the candidate driver set.

[0073] Furthermore, orders with higher priority are assigned more candidate drivers to improve the matching success rate. From the initial set of drivers, a corresponding number of drivers are selected based on a combination of driver service status priority (idle status first, followed by near completion status) and distance to form a candidate driver set.

[0074] Preferably, by using a dynamic search radius and a multi-state driver screening strategy, the present invention can adaptively adjust the search range according to the driver distribution density in different regions. At the same time, it can perform differentiated screening based on driver service status and order priority. Under the premise of controlling the size of the candidate driver set, it ensures that high-priority orders receive more sufficient matching resources. This avoids matching failures caused by an excessively small search range and prevents the computational burden caused by an excessively large candidate set.

[0075] S2.4: Obtain the starting coordinates of the order to be assigned, and the current coordinates of each candidate driver in the candidate driver set.

[0076] S2.5: Determine the starting point for route planning based on the driver's service status, and use a heuristic path search algorithm to generate multiple candidate routes.

[0077] Preferably, the current service status of each candidate driver in the candidate driver set is determined, where drivers in an idle state use their current coordinates as the starting point for path planning, and drivers in a near-completed order state use the current order's endpoint coordinates as the starting point for path planning. Based on the determined path planning starting point and the order's starting point coordinates for the pending order, a starting-endpoint coordinate pair for path search is constructed. The A* heuristic search algorithm, combined with the real-time road network topology, is used to perform path search for each starting-endpoint coordinate pair. During the path search process, a multi-path generation mechanism is employed, generating multiple different candidate paths for each order-driver combination by adjusting the search strategy and expanding node selection. The generated candidate paths are validated for validity, and invalid paths containing prohibited sections, construction sections, or abnormal detours are eliminated, forming the final candidate path set.

[0078] S2.6: Obtain real-time traffic data for each segment of the candidate path and calculate the estimated travel time for each candidate path based on the path length.

[0079] Furthermore, the segment composition information of each candidate path in the candidate path set is extracted to obtain basic attribute data such as road type, length, and speed limit of each segment; the real-time traffic data interface is called to obtain real-time traffic condition data such as current traffic flow, average driving speed, and congestion level of each segment on the candidate path; based on the basic attribute data of the segments and the real-time traffic condition data, the estimated travel time of each segment is determined by a segmented calculation method, wherein the estimated travel time is calculated by dividing the segment length by the actual driving speed; the estimated travel times of all segments on the same candidate path are accumulated to obtain the estimated travel time of each candidate path.

[0080] Preferably, this invention determines the starting point for differentiated route planning based on the driver's service status, which accurately reflects the driver's actual schedulable status and avoids route planning deviations. It employs the A* heuristic search algorithm combined with real-time road network topology for multi-path generation, ensuring path diversity while reducing computational complexity and improving system response speed. By acquiring real-time traffic condition data in segments and using segment-level refined time calculations, the accuracy of estimated arrival times is improved, reducing matching failures caused by time estimation deviations.

[0081] S2.7: For each order-driver combination, select the candidate route with the shortest estimated travel time, determine the state transition time based on the current service status of the candidate drivers, and calculate the estimated arrival time for each order-driver combination.

[0082] Specifically, the calculation of state transition time includes: drivers in an idle state do not require additional time compensation; drivers in a state about to complete their order need to have their estimated completion time added. The estimated arrival time is calculated as follows: using the current time as the starting point, adding the estimated travel time of the selected candidate route, and adding the state transition time of the corresponding driver service status, the estimated arrival time of the order-driver combination is finally obtained.

[0083] In practical applications, drivers in different service states have varying available time due to the need to handle their current tasks. Idle drivers can immediately respond to new orders, while drivers nearing completion need to wait for their current order to finish. By employing a state transition time compensation mechanism, the actual reachability time of each driver can be more accurately reflected, improving the accuracy of time estimation.

[0084] S3: Establish a multi-factor order acceptance willingness assessment model, and calculate the order acceptance probability for each order-driver combination based on the candidate driver's historical service data, current work status and order characteristics.

[0085] Specifically, the flowchart for constructing a multi-factor order acceptance willingness assessment model is as follows: Figure 2 As shown, it includes:

[0086] S3.1: Construct a multi-dimensional driver willingness to accept orders assessment system based on historical behavior factors, order matching factors, work status factors, and fatigue status factors, and set weight coefficients for each factor.

[0087] Specifically, a driver's willingness to accept orders is mainly influenced by four dimensions: past experience, order attractiveness, current ability, and physical condition. Historical behavior factors reflect the driver's order-accepting habits and preferences formed based on past experience; order matching factors reflect whether the order matches the driver's preferences, including attractive factors such as distance, region, and time of day; work status factors reflect whether the driver currently has the ability to accept orders; and fatigue factors reflect the negative impact of long working hours on order-accepting willingness.

[0088] Furthermore, these four factors comprehensively evaluate drivers' willingness to accept orders from different perspectives. They include historical experience references, real-time status assessments, and the attractiveness of the orders themselves. Compared with traditional distance-first or random allocation methods, the multi-factor order acceptance willingness assessment model can more accurately predict the probability of drivers accepting orders and improve the matching success rate of order allocation.

[0089] Furthermore, the weighting coefficients are set according to the degree of influence of each factor on the willingness to place orders. In one embodiment, the weight of the work status factor is set to 0.35, the weight of the order matching factor is set to 0.3, the weight of the historical behavior factor is set to 0.2, and the weight of the fatigue status factor is set to 0.15. The weighting coefficients can be dynamically adjusted according to time period characteristics, driver type, and regional characteristics.

[0090] S3.2: Obtain historical service data of candidate drivers, extract driver service features, and extract order features based on the order data to be assigned.

[0091] It should be noted that historical service data includes historical order acceptance rate, preferred order acceptance distance range, preferred service area, preferred service time period, and preferred order value range, which are obtained by statistically analyzing drivers' historical service records. The historical order acceptance rate is the ratio of the number of successful orders accepted by a driver within a preset time window to the total number of orders pushed to the driver; the preferred order acceptance distance range is obtained based on statistical analysis of the driver's historical order acceptance distance distribution; the preferred service area is determined based on the geographical distribution of the driver's historical service trajectory; the preferred service time period is obtained by statistically analyzing the driver's service activity during each time period; and the preferred order value range is obtained based on statistical analysis of the distribution of the order value of the driver's historical orders.

[0092] Furthermore, driver service characteristics include distance preference characteristics, region preference characteristics, time period preference characteristics, and revenue preference characteristics; order characteristics include order distance characteristics, order region characteristics, order time period characteristics, and order value characteristics.

[0093] S3.3: Determine historical behavior factors based on historical order acceptance rates in historical service data.

[0094] In one embodiment, a weighted order acceptance rate is obtained by applying time decay weighting to the driver's historical service data for different time periods. The ratio of the weighted order acceptance rate to the platform's average order acceptance rate is calculated as a relative order acceptance tendency index, which is then normalized to form a historical behavior factor.

[0095] S3.4: Calculate the order matching factor based on the degree of matching between driver service characteristics and order characteristics.

[0096] Furthermore, similarity scores between driver service features and order features are calculated separately in terms of distance, region, time period, and revenue. These similarity scores are then weighted and fused according to preset weights. The resulting weighted and fused comprehensive similarity score is normalized and mapped to a preset factor range to form an order matching factor. The preset weights are determined based on the importance of each dimension to the driver's order-accepting decision, with higher weights for distance and region, and moderate weights for time period and revenue.

[0097] S3.5: Determine the working status factor based on the driver's service status, and calculate the fatigue status factor based on the continuous online time in the driver's working status.

[0098] Among them, the driver service status is "Idle," with a work status factor of 1.0, indicating that the driver is fully capable of accepting orders immediately. The driver service status is "About to complete an order," and the work status factor is inversely proportional to the remaining completion time; the shorter the remaining completion time, the higher the work status factor. In addition, the fatigue status factor decreases with the increase of continuous online time, to reflect the negative impact of driver fatigue on their willingness to accept orders.

[0099] S3.6: A weighted fusion algorithm is used to calculate the order acceptance probability value for each order-driver combination by weighting each factor according to its weight coefficient.

[0100] S4: Construct a comprehensive matching evaluation model, calculate the matching degree of all order-driver combinations based on the estimated arrival time and order acceptance probability, and generate a matching degree matrix.

[0101] Specifically, step S4 includes:

[0102] S4.1: Obtain the current rating and historical complaint rate of candidate drivers, and calculate the service quality score.

[0103] Among them, driver ratings and service complaint rates are obtained from driver profiles, and the service quality score is obtained by weighting and combining driver ratings and service complaint rates.

[0104] S4.2: Based on order distribution characteristics and driver resource status, set service quality thresholds, time response upper limits, and order acceptance probability lower limits to construct tiered screening conditions.

[0105] Specifically, the ratio of the number of orders in the current batch to the number of available drivers is calculated as the supply-demand ratio coefficient. This coefficient is then matched against a pre-defined threshold mapping table to determine the corresponding service quality threshold, upper limit for time response, and lower limit for order acceptance probability, forming tiered screening criteria for candidate drivers. The threshold mapping table is built based on the platform's historical order allocation data. By analyzing the allocation performance under different supply-demand ratios in the platform's historical order allocation data, the service quality threshold, upper limit for time response, and lower limit for order acceptance probability corresponding to each supply-demand ratio range are determined.

[0106] Ideally, a dynamic matching mechanism between the supply-demand ratio coefficient and the threshold mapping table is used to achieve an intelligent balance between matching quality and allocation success rate, effectively solving the problem that fixed screening criteria are difficult to adapt to changes in supply and demand.

[0107] S4.3: A tiered filtering mechanism is adopted to filter order-driver combinations based on tiered filtering conditions, retaining valid matching combinations.

[0108] Furthermore, the estimated arrival time, order acceptance probability, and service quality score are standardized; each order-driver combination is checked one by one to see if it meets the tiered screening criteria, and combinations that meet the criteria are retained as valid matching combinations.

[0109] S4.4: Based on the estimated arrival time, order acceptance probability, and service quality score, a weighted fusion algorithm is used to calculate the comprehensive matching degree value of each valid matching combination.

[0110] The weighting coefficients are determined based on the degree of influence of each indicator on the matching effect. In one embodiment, the corresponding weighting coefficients are determined by statistically analyzing the correlation between each indicator and the final order success rate in historical allocation data.

[0111] Preferably, by integrating three core dimensions—estimated arrival time, order acceptance probability, and service quality score—a shift from a single indicator to a multi-dimensional comprehensive evaluation is achieved. Estimated arrival time ensures passengers' timeliness needs, order acceptance probability ensures the success rate of allocation, and service quality score improves passengers' service experience. The synergistic effect of these three factors significantly improves the accuracy and comprehensiveness of order-driver matching, avoiding the local optima problem caused by a single indicator.

[0112] S4.5: Construct a sparse matrix framework, fill the comprehensive matching degree value into the corresponding order-driver location, and generate a matching degree matrix.

[0113] Specifically, the overall matching score of a valid match combination is entered into the corresponding order-driver location, and zero is entered into the other locations.

[0114] S4.6: Perform integrity verification on the generated matching degree matrix.

[0115] Specifically, the system checks rows with all zeros in the matching matrix and gradually relaxes the tiered screening criteria to re-screen candidate drivers for orders without valid matches until each order obtains at least one non-zero matching combination. In extreme cases, such as when there are no available drivers around the order or the location is too remote, if a valid matching combination still cannot be obtained, the order is marked as unassignable in this batch and automatically added to the next batch for processing.

[0116] Preferably, this invention proposes an adaptive hierarchical screening matching evaluation mechanism. By setting hierarchical screening conditions and calculating a comprehensive matching value, it achieves accurate evaluation of order-driver combinations. This matching evaluation mechanism can intelligently adjust the screening strategy according to real-time supply and demand conditions, improving the allocation success rate while ensuring matching quality, and significantly enhancing the adaptability and overall performance of the order allocation system.

[0117] S5: Use a global optimization algorithm to process the matching degree matrix and output the order-driver allocation results.

[0118] Specifically, the flowchart of the global optimization algorithm is as follows: Figure 3 As shown, it includes:

[0119] S5.1: Select the appropriate global optimization algorithm based on the dimension and sparsity of the matching degree matrix.

[0120] Specifically, when the matching degree matrix is ​​small or highly sparse, an exact algorithm (such as the Hungarian algorithm or the KM algorithm) is selected, while when the matching degree matrix is ​​large and lowly sparse, a heuristic algorithm (such as the genetic algorithm) is selected.

[0121] S5.2: Weight the matching degree matrix according to order priority.

[0122] Specifically, for high-priority orders, all matching degree values ​​in the corresponding row of the matching degree matrix are adjusted using a weighting factor, while the original matching degree values ​​for other orders remain unchanged. The weighting factor is dynamically calculated based on the order waiting time; the longer the waiting time, the larger the weighting factor.

[0123] S5.3: The selected global optimization algorithm is used to process the weighted matching degree matrix and output the order-driver allocation results.

[0124] Specifically, the weighted matching degree matrix is ​​used as the algorithm input. The optimization objective is to maximize the overall matching degree, with the constraint that each order is assigned a unique driver and each driver is assigned at most one order. A global optimization algorithm is started to solve the problem, and the calculation stops when the algorithm termination condition is met (the exact algorithm completes its solution or the heuristic algorithm reaches its maximum number of iterations). The algorithm output is analyzed to extract the successfully assigned order-driver pairings and generate a list of order-driver assignment results.

[0125] S5.4: Perform conflict detection and feasibility verification on the order-driver allocation results, identify orders that were not successfully allocated, and mark them as pending reassignment.

[0126] Furthermore, the concurrent allocation process is checked for conflicts such as duplicate driver assignments. For conflicting order-driver pairs, conflict resolution is performed based on order priority and matching degree. The real-time online status and service capabilities of each driver at the time of allocation are queried synchronously to verify whether the assigned driver is still available to accept orders and within service range. Order-driver pairs with detected conflicts or abnormal driver status are removed from the order-driver allocation result list, and the corresponding orders are remarked as pending allocation. The number of successfully allocated orders and the number of orders awaiting reallocation are counted, an allocation result report is generated, and the final valid order-driver allocation result is output.

[0127] Preferably, this invention proposes a globally optimized intelligent order allocation mechanism. Through a complete processing flow including adaptive algorithm selection, dynamic priority weighting, global optimization solution, and result verification, it achieves intelligent allocation of ride-hailing orders. This intelligent order allocation mechanism improves computational efficiency while ensuring the priority allocation of high-priority orders and maximizing the overall allocation effect. Result verification ensures the feasibility of the allocation scheme, comprehensively improving the success rate of order allocation, service quality, and system stability.

[0128] S6: Push orders to drivers in the order-driver assignment results and monitor response status. Perform reassignment processing for non-response cases.

[0129] Specifically, the corresponding orders are pushed to drivers in the order-driver allocation results in parallel, and the driver response status is monitored. If a driver actively rejects an order or fails to accept an order within the time limit, the unresponsive driver resources are released and a reassignment process is triggered. Orders that fail to be reassigned are automatically added to the next batch for processing.

[0130] The parallel push strategy sends push messages to each successfully assigned order-driver pair simultaneously, with a preset response time window. Furthermore, the reassignment process employs a multi-level degradation strategy: first, it checks other valid matching combinations for the order in the matching matrix, then selects the driver with the highest matching degree who has not yet been assigned a driver for the push.

[0131] It should be noted that orders marked as pending reassignment will be reassigned directly. Unresponsive drivers refer to drivers who actively refuse orders or fail to accept orders within the time limit.

[0132] Furthermore, when orders that fail to be allocated are added to the next batch, their order priorities are dynamically adjusted as follows: Priority is determined based on the cumulative waiting time of the order. Orders whose cumulative waiting time does not exceed the first threshold retain their original priority; orders whose cumulative waiting time exceeds the first threshold but is below the second threshold are upgraded to medium priority; and orders whose cumulative waiting time exceeds the second threshold are upgraded to high priority. The first and second thresholds are determined based on the platform's historical response time distribution and service quality standards.

[0133] Preferably, the present invention employs a dynamic reallocation mechanism and a priority adjustment strategy, which reduces order waiting time and improves overall allocation efficiency.

[0134] This embodiment also provides a ride-hailing order allocation system based on artificial intelligence algorithms, and the module connection diagram is as follows. Figure 4 As shown, the system includes:

[0135] The data acquisition and preprocessing module is used to acquire the order data to be assigned from the order pool and the driver data from the driver pool, and to perform data preprocessing.

[0136] The candidate driver matching module is used to determine a set of candidate drivers for each order to be assigned by adopting a dynamic search radius and a multi-state driver screening strategy, and to calculate the estimated arrival time of each order-driver combination by combining real-time traffic conditions.

[0137] The order acceptance willingness assessment module is used to establish a multi-factor order acceptance willingness assessment model. Based on the candidate driver's historical service data, current work status and order characteristics, it calculates the order acceptance probability of each order-driver combination.

[0138] The comprehensive matching evaluation module is used to build a comprehensive matching evaluation model. Based on the estimated arrival time and order acceptance probability, it calculates the matching degree of all order-driver combinations in the current batch and generates a matching degree matrix.

[0139] The global optimization allocation module is used to process the matching degree matrix using a global optimization algorithm and output the order-driver allocation results;

[0140] The order push monitoring module is used to push orders to drivers in the order-driver assignment results and monitor their response status, and perform reassignment processing for non-response situations.

[0141] This embodiment also provides a computer device applicable to the ride-hailing order allocation method based on artificial intelligence algorithms, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the ride-hailing order allocation method based on artificial intelligence algorithms as proposed in the above embodiment.

[0142] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0143] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the ride-hailing order allocation method based on artificial intelligence algorithms as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0144] In summary, this invention, by constructing a multi-factor order acceptance willingness evaluation model, can accurately predict the probability of drivers accepting orders, improve the order allocation success rate, and reduce driver order rejections. Through dynamic search radius and multi-state driver screening strategies, it achieves precise location and efficient screening of candidate drivers, improving driver resource utilization efficiency and system response speed. By combining a heuristic route planning algorithm based on real-time traffic conditions, it provides high-precision arrival time estimates, improving passenger waiting experience and enhancing service quality. Through a global optimization algorithm processing the matching degree matrix, it achieves optimal matching and allocation of orders and drivers, maximizing overall operational efficiency and reducing empty mileage and waiting time. Through closed-loop feedback and reassignment mechanisms, it ensures the timely processing of unresponsive orders, improving system stability and reliability, and guaranteeing the effective satisfaction of passenger travel needs.

[0145] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for allocating ride-hailing orders based on artificial intelligence algorithms, characterized in that: include: Retrieve the order data to be assigned from the order pool and the driver data from the driver pool, and perform data preprocessing; A dynamic search radius and multi-state driver screening strategy is adopted to determine a set of candidate drivers for each order to be assigned, and the estimated arrival time of each order-driver combination is calculated in combination with real-time traffic conditions. Establish a multi-factor order acceptance willingness assessment model, and calculate the order acceptance probability for each order-driver combination based on the candidate driver's historical service data, current work status and order characteristics; Construct a comprehensive matching degree evaluation model, calculate the matching degree of all order-driver combinations based on the estimated arrival time and the order acceptance probability, and generate a matching degree matrix; The matching degree matrix is ​​processed using a global optimization algorithm, and the order-driver allocation results are output. Push orders to drivers in the order-driver assignment results and monitor their response status; perform reassignment processing for non-response cases. The establishment of the multi-factor order acceptance willingness assessment model includes: Construct a multi-dimensional driver order acceptance willingness evaluation system based on historical behavior factors, order matching factors, work status factors, and fatigue status factors, and set weight coefficients for each factor; Obtain historical service data of candidate drivers, extract driver service characteristics, and extract order characteristics based on the order data to be assigned; Based on historical order acceptance rates in historical service data, determine historical behavioral factors; Calculate the order matching factor based on the degree of matching between the driver service characteristics and the order characteristics; The working status factor is determined based on the driver's service status, and the fatigue status factor is calculated based on the continuous online time in the driver's working status. A weighted fusion algorithm is used to calculate the order acceptance probability value for each order-driver combination by weighting each factor according to the weight coefficients. The construction of the comprehensive matching degree evaluation model includes: Obtain the current rating and historical complaint rate of candidate drivers, and calculate the service quality score; Based on order distribution characteristics and driver resource status, service quality thresholds, time response upper limits, and order acceptance probability lower limits are set to construct tiered screening conditions. A tiered filtering mechanism is adopted to filter order-driver combinations based on the tiered filtering conditions, retaining valid matching combinations; Based on the estimated arrival time, order acceptance probability, and service quality score, a weighted fusion algorithm is used to calculate the comprehensive matching degree value for each valid matching combination. A sparse matrix framework is constructed, and the comprehensive matching degree value is filled into the corresponding order-driver location to generate a matching degree matrix.

2. The ride-hailing order allocation method based on artificial intelligence algorithms as described in claim 1, characterized in that: The process of determining the candidate driver set for each order to be assigned includes: The initial search radius is set with the coordinates of the starting point of the order to be assigned as the center, and the final search radius is determined based on the number of drivers within the initial search radius; Within the final search radius, drivers are filtered in layers based on their service status to form a preliminary set of drivers. The upper limit of the number of candidate drivers is determined based on the order priority of the orders to be assigned, and the corresponding number of drivers are selected from the initial set of drivers to form the candidate driver set.

3. The ride-hailing order allocation method based on artificial intelligence algorithms as described in claim 1, characterized in that: The calculation of the estimated arrival time for each order-driver combination, based on real-time traffic conditions, includes: Get the starting coordinates of the orders to be assigned, and the current coordinates of each candidate driver in the candidate driver set; The starting point for route planning is determined based on the driver's service status, and a heuristic path search algorithm is used for route planning to generate multiple candidate routes. Obtain real-time traffic data for each segment of the candidate route, and calculate the estimated travel time for each candidate route based on the route length. For each order-driver combination, select the candidate route with the shortest estimated travel time, determine the state transition time based on the current service status of the candidate drivers, and calculate the estimated arrival time for each order-driver combination.

4. The ride-hailing order allocation method based on artificial intelligence algorithms as described in claim 1, characterized in that: The process of using a global optimization algorithm to process the matching degree matrix includes: Select the appropriate global optimization algorithm based on the dimension and sparsity of the matching degree matrix; The matching degree matrix is ​​weighted according to order priority; The selected global optimization algorithm is used to process the weighted matching degree matrix, and the order-driver allocation results are output. Perform conflict detection and feasibility verification on the order-driver allocation results, identify orders that were not successfully allocated, and mark them as pending reassignment.

5. The ride-hailing order allocation method based on artificial intelligence algorithms as described in claim 4, characterized in that: The order priorities include high priority, medium priority, and low priority.

6. A ride-hailing order allocation system based on an artificial intelligence algorithm, based on the ride-hailing order allocation method based on an artificial intelligence algorithm as described in any one of claims 1 to 5, characterized in that: include: The data acquisition and preprocessing module is used to acquire the order data to be assigned from the order pool and the driver data from the driver pool, and to perform data preprocessing. The candidate driver matching module is used to determine a set of candidate drivers for each order to be assigned by adopting a dynamic search radius and a multi-state driver screening strategy, and to calculate the estimated arrival time of each order-driver combination by combining real-time traffic conditions. The order acceptance willingness assessment module is used to establish a multi-factor order acceptance willingness assessment model. Based on the candidate driver's historical service data, current work status and order characteristics, it calculates the order acceptance probability of each order-driver combination. The comprehensive matching degree evaluation module is used to construct a comprehensive matching degree evaluation model, calculate the matching degree of all order-driver combinations in the current batch based on the estimated arrival time and the order acceptance probability, and generate a matching degree matrix. The global optimization allocation module is used to process the matching degree matrix using a global optimization algorithm and output the order-driver allocation results. The order push monitoring module is used to push orders to drivers in the order-driver assignment results and monitor their response status, and perform reassignment processing for non-response situations.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the ride-hailing order allocation method based on artificial intelligence algorithms as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the ride-hailing order allocation method based on artificial intelligence algorithms as described in any one of claims 1 to 5.

Images