A discount-based method and system for dispatching a car

Abstract

The application relates to the technical field of Internet car-hailing, and discloses a discount-based car-hailing order dispatching method, which is characterized in that a passenger defines a maximum tolerable pick-up distance and an expected discount by himself / herself, different customers' tolerance to waiting time is utilized, no discount is given to orders within the global pick-up distance limit range set by a platform, and cars within a dispatching range are matched, a unified discount is determined according to the expected discount of an order and the matching condition of the car and the order, the order and the car within the time window are mobilized, passengers with long waiting time can travel at a lower price, waiting time is exchanged for discount, the matching probability of the order and the car is improved, more orders can be served, and user experience is improved. For the platform, although the platform gives customers a discount, higher profits can be obtained because the dispatching scheme is optimized. The application also provides a system for realizing the above method.

Description

Technical Field

[0001] This invention relates to the field of internet ride-hailing technology, and in particular to a ride-hailing dispatch method and system based on discounts. Background Technology

[0002] With the rapid development of mobile internet technology and the sharing economy model, ride-hailing platforms such as Didi Chuxing and Uber have emerged, and ride-hailing has become an important mode of transportation for the public. The typical operation of ride-hailing platforms is as follows: (1) Passengers and drivers join the ride-hailing platform through terminal devices with ride-hailing software installed; (2) Passengers submit travel information on the platform, such as pick-up location and destination, and generate an online ride-hailing order to send to the platform; (3) Once a new request is received, the platform decides on a suitable driver to pick up and drop off the passenger; (4) When the trip is completed, the platform calculates the passenger's fare and the driver's income. With these platforms, passengers have reduced travel costs and enjoyed fast and convenient transportation, while also bringing more employment opportunities to drivers.

[0003] Traditional travel is primarily non-carpooling, where passengers hire a car on a temporary basis and pay by mileage. In non-carpooling, each driver can only serve a single order at a time. In recent years, ride-hailing platforms have been promoting carpooling services to utilize empty seats in passenger vehicles. Carpooling allows multiple user orders to be assigned to a single vehicle simultaneously. Shared mobility not only reduces user travel costs and increases platform profits but also reduces vehicle emissions and alleviates traffic congestion. Currently, more and more shared mobility platforms are launching carpooling services. In existing models, to minimize driver travel distances while ensuring reasonable passenger waiting times to reduce order cancellation rates, ride-hailing platforms typically set a uniform pick-up distance to constrain passenger-driver matching and driver route planning. The pick-up distance refers to the distance the driver travels from their location when they receive the order to the order's origin. Ultimately, the pick-up distance for each order will not exceed the platform-set pick-up distance; orders exceeding this distance are considered invalid.

[0004] In the existing ride-hailing model, platforms set pick-up distance limits, matching orders with cars within these limits. These limits fail to adequately utilize the varying levels of user tolerance for waiting time, restricting matching to only available vehicles within a limited area and resulting in inefficient resource allocation. Especially during peak hours, these limits reduce the probability of successfully hailing a ride, failing to reduce waiting times and potentially leading to order cancellations due to prolonged wait times. Furthermore, for users, choosing to wait a few extra minutes offers no tangible benefit, suggesting their willingness to extend their wait time may be weak.

[0005] Existing technology provides a method for searching and picking up / dropping off passengers for intercity ride-hailing services, relating to the fields of transportation and the internet. In this method, drivers and passengers can mutually view each other's basic information, eliminating the need to be in unfamiliar environments. The ride-sharing website comprehensively sorts passenger information based on gender and the driver's distance from the passenger's departure point. The resulting list of passenger information is arranged sequentially, allowing drivers to select passengers beyond a purely automatic matching process. Passengers can set their allowed ride duration as a percentage of the total travel time from departure to destination. If another passenger is picked up during the journey, the per-kilometer fee is lower for passengers with longer allowed ride durations. However, this patent does not disclose how the per-kilometer fee is calculated. Summary of the Invention

[0006] The purpose of this invention is to provide a discount-based ride-hailing dispatch method and system that optimizes platform scheduling, improves order success rate, and maximizes platform profits.

[0007] To achieve the above objectives, the present invention provides a discount-based ride-hailing dispatch method, comprising the following steps:

[0008] S1: Obtain an order set O and a vehicle set V within a time window; the orders in order set O include the maximum tolerable pick-up distance and the expected discount;

[0009] S2: Determine whether the orders in order set O are carpooling orders. If they are not carpooling orders, proceed to step S3. If they are carpooling orders, proceed to steps S4 to S7.

[0010] S3: Dispatch non-carpooling orders, including:

[0011] S301: If the order contains pd ij If ≤pd0, no discount will be given and the nearest car will be dispatched to match the order;

[0012] S302: If pd0 <pd ij ≤tol i Based on the principle that the discount factor given by the platform should not be lower than the expected discount factor of the order and maximize the platform's total profit, the optimal discount factor value and the corresponding scheduling result are output; where pd0 is the global pick-up distance limit set by the platform, pd ij It's the pick-up distance, toll i This is the maximum acceptable pick-up distance limit for customers; specifically, based on the principle of maximizing the platform's total profit, the optimal discount factor value and corresponding scheduling results are output. If the optimal discount factor of the order is greater than the expected discount factor or there is no driver matching the order, the ride-hailing failure is returned; otherwise, the ride-hailing success is achieved.

[0013] S4: Schedule orders within the global pick-up range: Find all order-car pairs within the global pick-up distance range and put them into pool0; without giving any discounts, calculate the first immediate profit and schedule orders according to the principle of maximizing the first immediate profit.

[0014] S5: Delete carpooling orders that were scheduled in step S4 and cars that have reached capacity limits in the travel plan from order set O and vehicle set V to obtain new order set O' and new vehicle set V'; in the new order set O' and new vehicle set V', find the expected discount factor ep that satisfies the order-based discount. i For the carpooling problem, considering the pick-up distance and detour constraints, all order-car pairs are placed into a pool, and the second immediate profit is calculated. The order-car pairs in the pool are then scheduled according to the principle of maximizing the second immediate profit to obtain the travel plan.

[0015] S6: Let the discount factor given by the platform be γ. If the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order... i Furthermore, if the platform's total profit is greater than 0, the discount factor γ given by the platform will be considered an effective discount factor γ. 1 ; Calculate each effective discount factor γ 1 Based on the corresponding total platform profit, select the effective discount factor γ that maximizes the total platform profit. k ;

[0016] S7: The effective discount factor γ obtained from step S6 k Find the condition based on the effective discount factor γ. k Given the constraints of the carpooling problem, calculate the third-instance profit for all feasible order-car pairs, schedule the order-car pairs in the pool according to the principle of maximizing the third-instance profit, and output the final route plan for the cars.

[0017] Specifically, the process of determining the optimal discount factor is as follows:

[0018] S302-1: According to pd0 <pd ij ≤tol i The non-carpooling orders and cars in the order set O and vehicle set V are matched to obtain order-car pairs;

[0019] S302-2: Iterate through all order-car pairs, find valid discount factors and add them to the valid discount set D. Sort the discount factors in the valid discount set D in ascending order, and uniformly sample k discount factors from these sorted discount factors. Calculate the scheduling result for these k discount factors and their profit; select the discount factor γ with the highest profit.m ;

[0020] S302-3: In the discount factor γ m Continue sampling evenly in the vicinity and select the discount factor that yields the greatest profit.

[0021] 302-4: Repeat step S302-3 until the discount factor that maximizes profit is the same as the largest discount factor obtained from the previous sampling. Output this discount factor as the optimal discount factor.

[0022] Specifically, in step S302-3, the function FINDPEAK(l,r,k) is used for sampling, where l is the index of the first element in the sampling range, r is the index of the last element, and k is the number of sampling discount factors.

[0023] Specifically, in step S4, let the first immediate profit be P. 1 i,j The calculation is as follows:

[0024] P 1 i,j =rev i -α1*Δtd j (o i )-b0*II(v j (is empty);

[0025] Among them, rev i It is an order. i The price, α1 is the labor cost and fuel cost paid to the driver per unit distance traveled, Δtd j (o i ) is to put the order o i Insert car v j The increased travel distance in the route planning, b0 is the starting price, II(v) j `isempty` is used to determine the car's `v` value. j An indicator function to determine if the vehicle is empty.

[0026] Specifically, in step S5, the second immediate profit is P. 2 i,j The calculation is as follows:

[0027] P 2 i,j =rev i -α1*Δtd j (o i )-b0*II(v j is empty)-ep i *(pd ij -pd0)-∑i′≠i ep i′ *Δpd i′j ;

[0028] Among them, pd ij The platform sets a global pick-up distance limit for pd0, which is the pick-up distance setting. i′≠i ep i′ *Δpd i′j This refers to the insertion of o i This increased the pick-up distance for other orders, resulting in a higher total discount for those orders.

[0029] Specifically, in step S6, if the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order... i And the order is in the travel plan When the profit is zero, the discount factor γ given by the platform is considered an invalid discount factor γ. 0 Invalid discount factor γ for each order 0 and effective discount factor γ 1 The set D' that constitutes the discount factor γ given by the platform;

[0030] Let the total profit of the platform be P(γ), calculated as follows:

[0031] P(γ) = ab - α1 * cd * γ;

[0032] Where a, b, c, and d represent ∑rev i ,∑b0,∑td(plan j ), ∑(pd ij -pd0);

[0033] Iterate through all travel plans obtained in step S5 Calculate each γ for each order 1 and γ 0 And the corresponding Δa, Δb, Δc, and Δd for each discount factor, and then add all discount factors to the effective discount set D';

[0034] Traverse all discount factors in the effective discount set D' in ascending order, calculate the corresponding coefficients a, b, c, d and the corresponding profit; output the platform's total profit as the discount factor γ with the largest value P(γ). k .

[0035] Specifically, in step S7, the third immediate profit is P. 3 i,j The calculation is as follows:

[0036]

[0037] This invention also provides a discount-based ride-hailing dispatch system, comprising:

[0038] The order set and vehicle set acquisition module is used to acquire the order set and vehicle set within a time window;

[0039] The order determination module is used to determine whether an order in the order set is a carpooling order, and sends non-carpooling orders to the non-carpooling dispatch module and carpooling orders to the carpooling dispatch module.

[0040] The non-carpooling dispatch module is used to dispatch vehicles based on the order's pick-up distance, the platform's global pick-up distance limit, and the customer's maximum tolerable pick-up distance limit. When the order's pick-up distance is less than the platform's global pick-up distance limit, no discount is given, and the nearest car is dispatched to match the order. When the order's pick-up distance is greater than the platform's global pick-up distance limit but less than or equal to the customer's maximum tolerable pick-up distance, the module outputs the optimal discount factor value and the corresponding dispatch result, based on the principle that the platform's discount factor should not be lower than the order's expected discount factor and maximize the platform's total profit.

[0041] The carpooling dispatch module is used to dispatch cars based on the pickup distance of the order, the global pickup distance limit set by the platform, and the maximum tolerable pickup distance limit of the customer. It finds all order-car pairs within the global pickup distance range and puts them into pool0 without any discount. It calculates the first immediate profit and dispatches the orders according to the principle of maximizing the first immediate profit. Then, it deletes the dispatched carpooling orders and cars that have reached the capacity limit in the travel plan. For the remaining carpooling orders and cars, the platform applies the same discount factor, which is not less than the expected discount factor of the order, and dispatches them according to the principle of maximizing the platform's profit.

[0042] Specifically, the non-carpooling dispatch module includes an optimal discount factor determination unit. This unit matches non-carpooling orders with cars to obtain order-car pairs based on the order's pick-up distance being greater than the platform's global pick-up distance limit but less than or equal to the customer's maximum tolerable pick-up distance. It then iterates through all order-car pairs, finds valid discount factors, adds them to a valid discount set D, sorts the discount factors in D in ascending order, and uniformly samples k discount factors from these sorted factors. It then calculates the dispatch result for each of these k discount factors and their profit. Finally, it selects the discount factor γ with the highest profit. m ; at discount factor γ m Continue sampling evenly in the vicinity and select the discount factor with the highest profit. Repeat the above operation of sampling in the vicinity of the discount factor with the highest profit until the discount factor with the highest profit in the current sampling is the same as the largest discount factor obtained in the previous sampling. Output this discount factor as the optimal discount factor.

[0043] Specifically, the carpooling dispatch module includes a given discount factor determination module, which is used to find an order-based expected discount factor ep that satisfies the order set and vehicle set of deleted carpooling orders and vehicles that have reached capacity limits in the travel plan. i For the carpooling problem, considering the pick-up distance and detour constraints, all order-car pairs are placed into a pool, and the second immediate profit is calculated. The order-car pairs in the pool are then scheduled according to the principle of maximizing the second immediate profit to obtain the travel plan. Let the discount factor given by the platform be γ. If the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order, i Furthermore, if the platform's total profit is greater than 0, the discount factor γ given by the platform will be considered an effective discount factor γ. 1 ; Calculate each effective discount factor γ 1 Based on the corresponding total platform profit, select the effective discount factor γ that maximizes the total platform profit. k Find the condition based on the effective discount factor γ. k Given the constraints of the carpooling problem, calculate the third-instance profit for all feasible order-car pairs, schedule the order-car pairs in the pool according to the principle of maximizing the third-instance profit, and output the final route plan for the cars.

[0044] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0045] This invention allows passengers to define their maximum tolerable pick-up distance and expected discount. Leveraging the varying levels of customer tolerance for waiting time, it matches orders within the platform's global pick-up distance limit with available vehicles, without offering any discounts. For orders and vehicles within the platform's global pick-up distance limit, it determines a uniform discount based on the order's expected discount and the platform's maximum profit, thus allocating vehicles and vehicles within that time window. This allows passengers with longer wait times to travel at lower prices, essentially "trading waiting time for discounts," increasing the probability of order-vehicle matching, improving order success rates, and enabling more orders to be served, thereby enhancing the user experience. For the platform, although it offers discounts to customers, the optimized scheduling scheme allows for higher profits. Attached Figure Description

[0046] Figure 1 This is a flowchart of a discount-based ride-hailing dispatch method according to an embodiment of the present invention.

[0047] Figure 2 This is a bipartite graph in an embodiment of the present invention, with orders and cars forming one side node respectively.

[0048] Figure 3 This is the pseudocode for step S302 of Embodiment 1 of the present invention, which uses the improved brute-force algorithm IBF-Discount.

[0049] Figure 4 This is the pseudocode for step S302 of Embodiment 3 of the present invention, which uses the improved algorithm O-Discount.

[0050] Figure 5 This is the pseudocode of the greedy algorithm in Embodiment 4 of the present invention.

[0051] Figure 6 This is the pseudocode of the algorithm for determining the discount factor in steps S4 to S6 of Embodiment 4 of the present invention.

[0052] Figure 7 This is a schematic diagram illustrating the impact of the discount factor on the platform's total profit in Embodiment 4 of the present invention.

[0053] Figure 8 This is an order finding function in Embodiment 4 of the present invention. 1 and γ 0 Example diagram.

[0054] Figure 9 yes Figure 8 The table shows the coefficient variation of P(γ) in the examples. Detailed Implementation

[0055] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0056] The terms used in this invention are defined as follows:

[0057] In non-carpooling scenarios:

[0058] Time Window: A time window is a snapshot in time that batches orders. Orders are scheduled within a time window, then a new batch of orders accumulates, which is scheduled in the next time window. Within each time window, the input consists of the order set O = {o...} i} and the car set V = {v j}composition.

[0059] Order: An order is defined as follows This indicates the starting point of the order (pickup location). Indicates the destination, t i It is o i Creation time, rev i It is an order. i The price. tol i This is the maximum permissible pick-up distance that users can tolerate, toli >pd0. pd0 is the global pick-up distance limit set by this platform. ep i This refers to the expected discount amount for each additional kilometer of pick-up distance beyond pd0. i and ep i This is the user's input. The function d(·,·) gives the shortest travel distance between two locations.

[0060] Automobile: The definition of a car is v j = <l j >,l j Indicates v j The location.

[0061] Pick-up distance: Pick-up distance pd ij It refers to o i Before the starting point, car v j The distance traveled is equal to

[0062] Assume that some users can tolerate a pick-up distance exceeding pd0, while still having a maximum tolerable pick-up distance limit tol. i In other words, for that which satisfies pd0 <pd ij ≤tol i For orders placed by PD, the platform needs to offer them a discount as compensation. However, if PD... ij >tol i Even with discounts, users will still cancel their orders.

[0063] Discount: Discount dsc i It is to ensure that the pick-up distance meets pd0. <pd ij ≤tol i orders o i It is linearly related to distances exceeding pd0.

[0064]

[0065] Where γ is the platform's uniform discount factor for each batch of orders, that is, the amount deducted per kilometer after exceeding pd0. If ep i <γ, meaning the platform's discount is insufficient to cause users to extend their pick-up distance, the dispatch result will be o. i Refuse, dsc i =0.

[0066] Driver Fees: The fees paid to the driver include a base fare b0 and driving costs. b0 is a fixed value, α1 is the labor cost and fuel cost paid to the driver per unit distance traveled, and the driving costs are proportional to the distance traveled. The relationship is linear.

[0067] cost j =b0+α1*td j (2)

[0068] Order Profit: Order O i Profit p(o) i )for:

[0069]

[0070] Note that if pd ij >tol i or pd0 <pd ij ≤tol i &ep i >γ, it will be rejected by the user, so p(o i ) = 0.

[0071] Platform Profit: When a batch of orders is processed, the platform profit P is the total profit from all orders.

[0072]

[0073] In the non-carpooling problem, discount-based order scheduling (DBOD-NS) involves determining a platform-given discount factor γ and scheduling the resulting orders within a given time window. The goal is to maximize platform profit, and the scheduling result should satisfy the following constraints:

[0074] I. Pickup Distance Limit: For any dispatched order, the pickup distance must not exceed the user's maximum tolerable pickup distance, i.e.

[0075] II. Discount Restriction: For any dispatched order with a pick-up distance exceeding pd0, the platform's discount factor γ shall not be lower than the order's expected discount factor ep. i ,Right now

[0076] In carpooling scenarios:

[0077] DBOD-RS: Discount-based order scheduling problem in carpooling scenarios

[0078] Time window, order, discount, driver fee: These definitions in DBOD-RS are exactly the same as those in DBOD-NS, so these concepts will not be repeated.

[0079] Automobile: The definition of a car is v j = <l j ,c j plan j >,l jRepresents v j Position, c j Represents v j Passenger capacity, plan j For v j The travel plan is represented as a sequence of road nodes, such as

[0080] Pick-up distance: o i Pick-up distance pd ij It is a car v j arrive The distance traveled. For example, the travel plan 1 for v1 is... The pick-up distance of O1 is PD. 11 for And the distance to O2's pick-up point PD 21 for In DBOD-RS, the platform also sets a global pick-up distance, pd0.

[0081] Detour Ratio: The detour ratio ∈ is the proportion of the shortest path that can be taken by an acceptable additional detour, with a limit of o. i from arrive Total distance, In DBOD-RS, ∈ is set by the platform.

[0082] Platform Profit: Given the scheduling results of a batch of orders, the platform profit P is:

[0083]

[0084] Among them, II(pd) ij >pd0) is used to determine pd ij The indicator function for whether it is greater than pd0. Note that when order scheduling and route planning remain unchanged, P is a decreasing function of γ.

[0085] Discount-based order scheduling in carpooling scenarios, DBOD-RS: Within a time window, given a set of cars V and a set of orders O, discount-based order scheduling in the carpooling problem involves determining the platform's given discount factor γ and providing the order scheduling and car route planning results. The goal is to maximize the platform's profit, and the following constraints should be satisfied:

[0086] I. Pickup Distance Limit: For any dispatched order, the pickup distance must not exceed the user's maximum tolerable pickup distance, i.e.

[0087] II. Discount Restriction: For any dispatched order with a pick-up distance exceeding pd0, the platform's discount factor γ shall not be lower than the order's expected discount factor ep. i,Right now

[0088] III. Detour Ratio Limit: For any dispatched order, the total order distance is limited. Constraints, i.e.

[0089] IV: Capacity Limit: For any dispatched vehicle, the number of orders dispatched to the vehicle must not exceed its capacity at any time. j ,Right now

[0090] Example 1

[0091] like Figure 1 As shown, a preferred embodiment of the present invention provides a discount-based ride-hailing dispatch method, which includes the following steps:

[0092] S1: Obtain an order set O and a vehicle set V within a time window; the orders in order set O include the maximum tolerable pick-up distance and the expected discount;

[0093] S2: Determine whether the orders in order set O are carpooling orders. If they are not carpooling orders, proceed to step S3. If they are carpooling orders, proceed to steps S4 to S7.

[0094] S3: Dispatch non-carpooling orders, including:

[0095] S301: If the order contains pd ij If ≤pd0, no discount will be given and the nearest car will be dispatched to match the order;

[0096] S302: If pd0 <pd ij ≤tol i Based on the principle that the discount factor given by the platform should not be lower than the expected discount factor of the order and maximize the platform's total profit, the optimal discount factor value and the corresponding scheduling result are output; where pd0 is the global pick-up distance limit set by the platform, pd ij It's the pick-up distance, toll i This is the maximum acceptable pick-up distance for customers;

[0097] S4: Schedule orders within the global pick-up range: Find all order-car pairs within the global pick-up distance range and put them into pool0; without giving any discounts, calculate the first immediate profit and schedule orders according to the principle of maximizing the first immediate profit.

[0098] S5: Schedule orders outside the global pick-up range: Delete carpooling orders that were scheduled in step S4 from order set O and vehicle set V, as well as cars that have reached capacity limits in the travel plan, to obtain new order set O' and new vehicle set V'; in the new order set O' and new vehicle set V', find the expected discount factor ep that satisfies the order. i For the carpooling problem, considering the pick-up distance and detour constraints, all order-car pairs are placed into a pool, and the second immediate profit is calculated. The order-car pairs in the pool are then scheduled according to the principle of maximizing the second immediate profit to obtain the travel plan.

[0099] S6: Let the discount factor given by the platform be γ. If the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order... i Furthermore, if the platform's total profit is greater than 0, the discount factor γ given by the platform will be considered an effective discount factor γ. 1 ; Calculate each effective discount factor γ 1 Based on the corresponding total platform profit, select the effective discount factor γ that maximizes the total platform profit. k ;

[0100] S7: The effective discount factor γ obtained from step S6 k Find the condition based on the effective discount factor γ. k Given the constraints of the carpooling problem, calculate the third-instance profit for all feasible order-car pairs, schedule the order-car pairs in the pool according to the principle of maximizing the third-instance profit, and output the final route plan for the cars.

[0101] This embodiment allows passengers to define their maximum tolerable pick-up distance and expected discount. Leveraging the varying levels of customer tolerance for waiting time, orders within the platform's global pick-up distance limit are not discounted, and vehicles within the dispatch range are matched. For orders and vehicles within the platform's global pick-up distance limit, a uniform discount is determined based on the order's expected discount and the platform's maximum profit. This discount is then used to allocate orders and vehicles within that time window, allowing passengers with longer wait times to travel at lower prices—"trading discounts for waiting time"—increasing the probability of order-vehicle matching, improving order success rates, enabling more orders to be served, and enhancing the user experience. For the platform, although it offers discounts to customers, the optimized dispatch scheme allows for higher profits.

[0102] In step S1, a user inputs a maximum tolerable pick-up distance and an expected discount. When the user inputs the maximum tolerable distance, it can be converted into waiting time according to the distance and traffic conditions. When the user inputs the expected discount, optional discounts are provided based on the closest and farthest vehicles from the user, and one maximum tolerable distance corresponds to one discount, thereby avoiding the chaos of excessively low expected discounts.

[0103] In step S3 for non-carpooling scenario scheduling and steps S4 to S7 for carpooling scenario scheduling, if a vehicle receives both a non-carpooling order and a carpooling order at the same time, conflict adjustment is performed in the final stage. For a vehicle that has been dispatched an order in both modes, the assignment with higher revenue is finally selected, then the assignment with lower revenue is canceled, and only the order-vehicle pair is canceled, while the discount factor and other order-vehicle pairs remain unchanged. For orders that fail to be matched within the time window, a prompt of failed ride-hailing is returned to the user. If the user still needs a vehicle, they can hail a ride again and become an order for the next time window.

[0104] In a non-carpooling scenario, a bipartite graph with orders and vehicles forming nodes on two separate sides is constructed as Figure 2 shown. Then, if an order and a vehicle do not violate any constraints, an edge can be drawn between them to indicate that matching is allowed. Any feasible scheduling result can thus be expressed as a subgraph of the bipartite graph. Assume there are n orders and n vehicles, and ep1<ep2<…<ep n . In Figure 2 (a), assume that the pick-up distance of all order-vehicle pairs is greater than pd0. Since γ=0, the bipartite graph has no matching edges. Figure 2 (b) and Figure 2 (d) have additional matching edges because o1 and o2 satisfy both the discount constraint and the pick-up distance constraint. In Figure 2 (c), when γ=a (ep1<a<ep2), compared with Figure 2 (b), no matching edge is added, because except for o1, there are still no orders satisfying the discount constraint. Therefore, the matching condition of the bipartite graph can only be changed when the value of γ is the expected discount factor of an order.

[0105] When γ=ep i , o i may have additional matching edges. However, the matching of two types of orders is not affected by the value of γ. I) For o ij with pd i ≤pd0 for all vehicle-order pairs, we do not need to provide any discount for them. II) For o ij ≥tol i with pd for all vehicle-order pairs, o iEven offering a discount cannot make it effective. Therefore, the expected discount factor for both types of orders cannot increase the edges of the bipartite graph. Only for those orders with at least one satisfying pd0... <pd ij ≤tol i For orders of cars, discounts are needed to make them valid, increasing the matching edges in the bipartite graph, thus increasing their ep i It is a candidate for the optimal discount factor. Therefore, we only need to iterate through the effective discount factors in the discount set D to obtain the optimal value of γ.

[0106] In this embodiment, step S302 employs the improved brute-force algorithm IBF-Discount to find the optimal discount factor that maximizes profit and its corresponding scheduling result. For example... Figure 3 As shown, the algorithm takes order set O and car set V as input and outputs the optimal discount factor γ and the corresponding scheduling result M. In lines 2-6, the algorithm iterates through all car order pairs (o... i ,v j The process begins by identifying valid discount factors and adding them to the valid discount set D. Lines 7-16 iterate through all possible values ​​of γ, using the Hungarian algorithm to calculate the profit for each value. γ ranges from all discount factors in D. Note that lines 10-14 iterate through all order-car pairs to find valid order-car pairs that satisfy both the pick-up distance and discount constraints. Line 17 finds the value of γ that maximizes the profit and the corresponding scheduling result.

[0107] Example 2

[0108] The difference between this embodiment and Embodiment 1 is that this embodiment uses an improved algorithm in step S302.

[0109] The specific process for determining the optimal discount factor in step S302 of this embodiment is as follows:

[0110] S302-1: According to pd0 <pd ij ≤tol i The non-carpooling orders and cars in the order set O and vehicle set V are matched to obtain order-car pairs;

[0111] S302-2: Iterate through all order-car pairs, find valid discount factors and add them to the valid discount set D. Sort the discount factors in the valid discount set D in ascending order, and uniformly sample k discount factors from these sorted discount factors. Calculate the scheduling result for these k discount factors and their profit; select the discount factor γ with the highest profit. m ;

[0112] S302-3: In the discount factor γ mContinue sampling evenly in the vicinity and select the discount factor that yields the highest profit; in this embodiment, the discount factor γ... m Sampling k / 2 discount factors from the vicinity, i.e., within discount factor γ m First sample k / 4 discount factors, then at discount factor γ m Post-sampling with k / 4 discount factors;

[0113] 302-4: Repeat step S302-3 until the discount factor that maximizes profit is the same as the largest discount factor obtained from the previous sampling. Output this discount factor as the optimal discount factor. Alternatively, reduce the number of sampled discount factors to 1 and output this discount factor as the optimal discount factor.

[0114] The improved algorithm involves arranging the n discount factors in ascending order and uniformly sampling k discount factors from this arrangement. The Hungarian algorithm is then used to obtain the scheduling results for these k discount factors and calculate their profits. The discount factor with the highest profit is selected, assuming its index in the discount factor array is c. The next sampling range is then... The number of sampling points is Repeat the above steps until k = 1. Note that the value of k is determined by the number of orders and is usually small. For more than 200 orders, an initial value of 5 for k will provide a good approximation.

[0115] The method for determining the optimal discount factor in this embodiment is more efficient and has a lower time cost than the method in Embodiment 1.

[0116] The other steps in this embodiment are the same as in Embodiment 1, and will not be repeated here.

[0117] Example 3

[0118] The difference between this embodiment and embodiment two is that the sampling methods used in steps S302-2 and S302-3 are different. In this embodiment, step S302 uses the optimized algorithm O-Discount.

[0119] In this embodiment, in step S302-3, the function FINDPEAK(l,r,k) is used for sampling, where l is the index of the first element in the sampling range, r is the index of the last element, and k is the number of sampling discount factors.

[0120] This embodiment uses a different method for searching the optimal discount factor. Specifically, as follows: Figure 4As shown, the function FINDPEAK(l,r,k) represents the sampling process of the discount factor. l is the index of the first element in the sampling range, r is the index of the last element, and k is the number of sampling discount factors. Lines 6-10 show the termination condition of the function recursion: when the number of samples k becomes 1, the corresponding discount factor is the approximately optimal discount factor we are looking for. In lines 11-13, s is the sampling interval. If s = 0, the number of samples k is changed to... Proceed to the next recursion. Line 15 calculates the profit corresponding to the sampled discount factor using the Hungarian algorithm. Line 16 selects the discount factor with the highest profit and updates the sampling range for the next recursion. Line 17 proceeds to the next recursion, and the number of samples becomes...

[0121] This embodiment requires at most a search. The discount factor reduces the time complexity compared to Example 2.

[0122] The other steps in this embodiment are the same as in Embodiment 1, and will not be repeated here.

[0123] Implementation 4

[0124] This embodiment provides a detailed description of discount-based order scheduling (DBOD-RS) in a carpooling scenario, specifically steps S4 to S7.

[0125] DBOD-RS mainly comprises two sub-problems: determining the discount factor γ and order scheduling, which are interrelated. This embodiment first determines the discount factor and then schedules the orders.

[0126] Determining the discount factor involves three stages. Stage 1 assigns no discount and schedules orders within the global pick-up distance. Stage 2 schedules orders beyond the global pick-up distance. First, the rules for determining the discount factor γ are simplified, assigning these orders a discount factor equal to their expected discount factor. A greedy algorithm is then used to determine order scheduling and route planning. Note that the route planning obtained here is used to find an approximate optimal discount factor, not the final route plan. In our setup, the discount factor γ is uniform for each order; therefore, orders scheduled in this stage do not meet the discount constraints of Discount-Based Order Scheduling (DBOD-RS) in a carpooling scenario. In Stage 3, we iterate through all values ​​in the discount factor set D, calculating the total profit for all valid orders in the route planning of Stage 2 for each value. The discount factor that maximizes the profit is the approximate optimal discount factor γ we are looking for. m Note that the discount factor set D includes γ. m All candidate discount factors.

[0127] In order scheduling, we use a greedy algorithm to obtain the final order scheduling and route planning. Here, we assign a discount coefficient of γ to each order. m Greedy algorithms, such as Figure 5 As shown. All order scheduling and route planning are implemented using a greedy algorithm. Given a pool containing many feasible order-car pairs, the greedy algorithm proceeds step-by-step, scheduling at each step to maximize the immediate profit P. ij The greedy algorithm updates the travel plans for each car in a pool using a route planning algorithm based on queue jumping. The algorithm continues until there are no more feasible order-car pairs in the pool, at which point it outputs the travel plans for the scheduled cars. For scheduling a pool of n orders and m cars, the time cost of the greedy algorithm is O(n log n). 2 (m+c 2 q), where O(c) 2 q) represents the time complexity of the path planning algorithm based on queue-jumping operations. c represents the maximum capacity of the cars, and O(q) represents the time cost of the shortest path query.

[0128] The method for determining the discount factor in a carpooling scenario is steps S4 to S6. In this embodiment, the determination of the discount factor in a carpooling scenario is divided into three stages, namely steps S4, S5, and S6.

[0129] (1) S4: Schedule orders within the global pick-up range, such as Figure 6 As shown in lines 2-5 of the document.

[0130] In lines 2-4, find all order-car pairs within the global pick-up distance range and add them to pool0. The instant profit in line 3 is the first instant profit P. 1 i,j The calculation is as follows:

[0131] P 1 i,j =rev i -α1*Δtd j (o i )-b0*II(v j (is empty); (6)

[0132] Among them, rev i It is an order. i The price, α1 is the labor cost and fuel cost paid to the driver per unit distance traveled, Δtd j (o i ) is to put the order o i Insert car v j The increased travel distance in the route planning, b0 is the starting price, II(v) j `isempty` is used to determine the car's `v` value.j The function indicates whether the vehicle is empty. Step S4 dispatches orders based on the principle of maximizing immediate profit, such as... Figure 6 As shown in line 5, this embodiment uses a greedy algorithm to schedule orders.

[0133] (2) S5: Schedule orders outside the global pick-up range, such as Figure 6 As shown in lines 6-10 of the document.

[0134] Line 6 removes dispatched orders and vehicles that have reached capacity limits in the travel plan from order set O and vehicle set V, resulting in a new order set O' and a new vehicle set V'. Lines 7-9 iterate through the new order set O' and new vehicle set V', adding all order-vehicle pairs that satisfy the pick-up distance and detour restrictions in the discount-based order dispatch (DBOD-RS) problem under carpooling scenarios, and place them into a pool. Line 8 shows the immediate profit as the second immediate profit, P. 2 i,j The calculation is as follows:

[0135] P 2 i,j =rev i -α1*Δtd j (o i )-b0*II(v j is empty)-ep i *(pd ij -pd0)-∑ i′≠i ep i′ *Δpd i′j (7)

[0136] Among them, pd ij The platform sets a global pick-up distance limit for pd0, which is the pick-up distance setting. i′≠i ep i′ *Δpd i′j This refers to the insertion of o i The pick-up distance for other orders is increased, resulting in a higher total discount paid to those orders. Step S5 schedules the order-car pairs in the pool according to the principle of maximizing immediate profit, thus obtaining the travel plan. This embodiment uses a greedy algorithm to schedule order-car pairs in the pool, such as... Figure 6 As shown in line 10 of the table. The travel plan obtained in this step is represented as... All travel plans are combined into new_schedule, but all travel plans are invalid.

[0137] (3)S6: Find the optimal discount factor, i.e., the effective discount factor γ. k ,like Figure 6 As shown in lines 11-21 of the document.

[0138] When searching for the optimal discount factor, we first find the set of discount factors D, which is the candidate set of optimal discount factors. Then, we iterate through each discount factor in the set D, calculating the total profit of all valid orders in `new_schedule`. The discount factor that brings the maximum profit in `new_schedule` is the discount factor we have found. Note that valid orders mean that their profits are positive and they satisfy the discount constraint in the defined constraints of the Discount-Based Order Scheduling (DBOD-RS) problem in a carpooling scenario. The reason why we don't need to check other conditions in the "Constraints of the Discount-Based Order Scheduling (DBOD-RS) Problem in a Carpooling Scenario" is that we only remove the origin and destination of invalid orders from `new_schedule`, and the remaining orders will not violate the other conditions in the "Constraints of the Discount-Based Order Scheduling (DBOD-RS) Problem in a Carpooling Scenario". Because, when from plan j Deleting the pick-up location and destination of an order does not violate the capacity limit, pick-up distance limit, and detour ratio limit of the "Discount-Based Order Scheduling (DBOD-RS) Problem in Carpooling Scenarios". Firstly, regarding the capacity limit, deleting an order reduces the number of orders in the travel plan, thus not violating this principle. Regarding the pick-up distance and detour ratio limits, according to the triangle inequality, deleting an order reduces the number of available pick-up locations and destinations. j Access plan j The travel distance of any remaining node will not exceed the travel distance in the original travel plan. Therefore, the pickup distance and detour distance of the remaining orders will not be longer than before the infeasible orders were deleted.

[0139] Specifically, step S6 includes:

[0140] S6-1: Obtain the set of discount factors D': Let the discount factor given by the platform be γ. If the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order, i Furthermore, if the platform's total profit is greater than 0, the discount factor γ given by the platform will be considered an effective discount factor γ. 1 If the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order. i And the order is in the travel plan When the profit is zero, the discount factor γ given by the platform is considered an invalid discount factor γ. 0 Invalid discount factor γ for each order 0 and effective discount factor γ 1 The set D' constitutes the discount factor γ given by the platform.

[0141] We observed that when traversing the discount factors in D in ascending order, Orders in i There are three states: I) If γ <ep i o i Invalid; II) If γ≥ep i &p(o i If )>0, then o i Effective. III) If γ is too large, p(o i )≤0,o i This will be invalidated again. Orders only contribute to total profit when they are in a valid state, and p(o) i The expression for ) is fixed in state II). That is, only when o is set... i The discount factor for changing from state I) to state II) and o i Only by changing the discount factor from state II to state III can p(o) be changed. i The expression for γ. Other values ​​of γ do not affect p(o). i The expression of ).

[0142] Therefore, for new_schedule Orders on the platform i Effective discount factor It is o i The discount factor for changing from an invalid state to an effective state, i.e. From v j The current plan, Insert / delete i The increase / loss of profit is calculated as follows:

[0143]

[0144] in, It is by solving P 2 i,j In formula (7), (pd) ij -pd0)+∑ i′≠i Δpd i′j The simplified result. (Compared to solving P) 2 i,j The formula (7) is different, but the formula (8) gives a uniform discount factor for each order.

[0145] For new_schedule Orders on the platform i Invalid discount factor It is o i The discount factor for changing from an effective state to an ineffective state, i.e. Note that the coefficients in formula (8) are affected by route planning. Therefore, each time we add / delete an order in a travel plan, we should update the coefficients in formula (8) for other valid orders in the same travel plan to accurately calculate the γ of other valid orders. 0 .

[0146] S6-2: Calculate the effective discount factor γ 1 Based on the corresponding total platform profit, select the effective discount factor γ that maximizes the total platform profit. k .

[0147] When determining discount factors, the discount factors that generate the maximum profit are found among the effective discount factors. For example... Figure 7 As shown, Figure 7 This shows how γ affects the platform's total profit P(γ). There are three orders o1, o2, o3, and we assume we have already calculated their effective discount factors. and invalid discount factor The relative order of these discount factors is shown on the coordinate axis. We can observe that the effective discount factors... This can cause a sharp increase in P(γ) because of the increased available orders. For They do not cause a sharp decrease / increase. This is because every time we add / delete an order travel plan, the coefficients of formula (8) for other valid orders of the same travel plan need to be updated. This ensures that the profit loss to the current route when an order is deleted is 0, thus preventing a sharp decrease / increase in P(γ). According to formula (4), P(γ) is determined by any two key discount factors (key discount factors are effective discount factors γ). 1 or invalid discount factor γ 0 The function is decreasing between these intervals because all travel plans remain unchanged during these intervals. In summary, it can be seen that each γ... 1 These are all maxima of P(γ), therefore the discount factor that brings the greatest profit exists among the effective discount factors.

[0148] Therefore, let the total profit of the platform be P(γ), and rewrite formula (5) as follows:

[0149] P(γ)=ab-α1*cd*γ; (9)

[0150] Where a, b, c, and d represent ∑rev i ,∑b0,∑td(plan j ), ∑(pd ij-pd0). If we record Δa, Δb, Δc, and Δd caused by each key discount factor, we can obtain an expression for P(γ) and calculate the total profit for each key discount factor. Then, we can easily find the discount factor that brings the maximum profit. Therefore, to facilitate the calculation of P(γ), we also add invalid discount factors in D.

[0151] like Figure 6 As shown in lines 11-13, all travel plans of new_schedule are traversed, that is, all travel plans obtained in step S5 are traversed. Find each γ for each order 1 and γ 0 And the corresponding Δa, Δb, Δc, and Δd for each key discount factor, and then add all key discount factors to the effective discount set D'. Lines 16-19 iterate through all discount factors in the effective discount set D' in ascending order, calculating the corresponding coefficients a, b, c, d and the corresponding profit. Line 20 finds the discount factor γ that maximizes the platform's total profit P(γ). k The value is then output on line 21.

[0152] like Figure 8 and Figure 9 As shown, a γ-shaped tool for finding orders is provided. 1 and γ 0 The example illustrates how to search for γ for each order on a travel plan. 1 and γ 0 And the corresponding coefficient changes Δa, Δb, Δc, Δd of P(γ) caused by them, such as Figure 9 As shown. Figure 8 (a) shows v1 in new_schedule. There are two orders, o1 and o2, with rev1 = 22.5, rev2 = 27, α1 = 1.5, b0 = 5, and pd0 = 2. We know... Orders listed are invalid, so they are represented by dashed lines. The right-hand axis marks the discount factors we need to search for in ascending order; discount factors already searched are marked in red. Figure 8 In (b), when we search for ep1, we find

[0153] ΔP 1,1 (ep1) = 22.5 – 5 – 1.5 * 9 - 1 * 1 = 3 > 0, therefore Therefore, we can safely add o1 to plan 1. We do this based on ΔP. 1,1 (·)exist Figure 9 The second column records This results in Δa, Δb, Δc, and Δd. Then we can calculate P(γ) = ab - α1*cd*γ = (0 + 22.5) – (0 + 5) - 1.5*(0 + 9) – (0 + 1)*γ = 4 - γ. We use... calculate And add it to the coordinate axis. Figure 8 In (c), when we search for ep2, we find ΔP 2,1 (ep2)>0, so Depend on The Δa, Δb, Δc, and Δd values ​​of P(γ) caused by these values ​​are recorded in Figure 9 In the third column. Then we can calculate P(γ)=(27+22.5)-(5+0)-1.5*(9+7)-(1+3)*γ=20.5-4*γ. We use Calculate Note that when we add o2 in plan 1, we should modify... We use Recalculated exist Figure 8 In (d), when we search At that time, we remove o1 from plan1 and according to ΔP 1,1 (·)exist Figure 9 The coefficient variation of P(γ) is recorded in column 4. Then, we use... Recalculate This remains the same as before. Figure 8 In (e), when we search At that time, we remove o2 from plan 1, and according to ΔP 2,1 (·)exist Figure 9 The coefficient changes of P(γ) are recorded in column 5. We observe that Δa, Δb, Δc, and Δd in columns 2 and 4 (columns 3 and 5) do not correspond perfectly, indicating that the coefficients of formula (8) are affected by path planning. Finally, we obtain the key discount factors of o1 and o2 and the corresponding coefficient changes of P(γ).

[0154] In other words, in step S6 of this embodiment, the plan is traversed. 0 j For all orders, calculate the effective discount γ for each order. 1 and invalid discount γ 0 And the changes in the coefficients of the total profit expression corresponding to these discounts. Effective discount γ 1 This refers to making a plan 0 jThe critical discount for an order to change from an invalid state to an valid state is, specifically, if γ is greater than or equal to the expected discount factor ep of the order. i Furthermore, if the order profit is greater than 0, then γ is used as the effective discount factor for that order. Ineffective discount γ 0 This refers to making a plan 0 j The critical discount that causes an order in the plan to change from an effective state to an invalid state, specifically, makes the plan 0 j The critical discount where the profit of orders in the plan begins to turn negative. Ascending order traversal of plan... 0 j For all orders with valid and invalid discounts, the total profit corresponding to each valid and invalid discount is calculated based on the coefficient changes in the total profit expression for each discount. The discount that maximizes the total profit is then found as the approximate optimal discount γ in the final output. k .

[0155] S7: Order scheduling.

[0156] The first step in order scheduling is to find all feasible orders – car pairs (o i ,v j They are placed into a pool, and their immediate profit is calculated. Feasible order-car pairs are those that satisfy the constraints of the "Discount-Based Order Scheduling in Carpooling Scenario (DBOD-RS) Problem". Note that the discount factor for each order is the output value of Algorithm 4, γ. m The immediate profit is P. 3 i,j The calculation method is as follows:

[0157]

[0158] Then, in this embodiment, a greedy algorithm is used to schedule the order-car pairs in the pool and output the final route plan for the car.

[0159] Other aspects of this embodiment are the same as those of Embodiment 1, and will not be repeated here.

[0160] Example 5

[0161] This embodiment provides a discount-based ride-hailing dispatch system to implement the methods of Embodiment 1, Embodiment 2, Embodiment 3, or Embodiment 4. The system includes:

[0162] The order set and vehicle set acquisition module is used to acquire the order set and vehicle set within a time window;

[0163] The order determination module is used to determine whether an order in the order set is a carpooling order, and sends non-carpooling orders to the non-carpooling dispatch module and carpooling orders to the carpooling dispatch module.

[0164] The non-carpooling dispatch module is used to dispatch vehicles based on the order's pick-up distance, the platform's global pick-up distance limit, and the customer's maximum tolerable pick-up distance limit. When the order's pick-up distance is less than the platform's global pick-up distance limit, no discount is given, and the nearest car is dispatched to match the order. When the order's pick-up distance is greater than the platform's global pick-up distance limit but less than or equal to the customer's maximum tolerable pick-up distance, the module outputs the optimal discount factor value and the corresponding dispatch result, based on the principle that the platform's discount factor should not be lower than the order's expected discount factor and maximize the platform's total profit.

[0165] The carpooling dispatch module is used to dispatch cars based on the pickup distance of the order, the global pickup distance limit set by the platform, and the maximum tolerable pickup distance limit of the customer. It finds all order-car pairs within the global pickup distance range and puts them into pool0 without any discount. It calculates the first immediate profit and dispatches the orders according to the principle of maximizing the first immediate profit. Then, it deletes the dispatched carpooling orders and cars that have reached the capacity limit in the travel plan. For the remaining carpooling orders and cars, the platform applies the same discount factor, which is not less than the expected discount factor of the order, and dispatches them according to the principle of maximizing the platform's profit.

[0166] Specifically, the non-carpooling dispatch module includes an optimal discount factor determination unit. This unit matches non-carpooling orders with cars to obtain order-car pairs based on the order's pick-up distance being greater than the platform's global pick-up distance limit but less than or equal to the customer's maximum tolerable pick-up distance. It then iterates through all order-car pairs, finds valid discount factors, adds them to a valid discount set D, sorts the discount factors in D in ascending order, and uniformly samples k discount factors from these sorted factors. It then calculates the dispatch result for each of these k discount factors and their profit. Finally, it selects the discount factor γ with the highest profit. m ; at discount factor γ m Continue sampling evenly in the vicinity, selecting the discount factor with the highest profit. Repeat the sampling operation near the discount factor with the highest profit until the discount factor with the highest profit in the current sampling is the same as the highest discount factor obtained in the previous sampling. Output this discount factor as the optimal discount factor. Specifically, based on the principle of maximizing platform profits, output the optimal value of the discount factor γ uniformly given by the platform and the corresponding scheduling result. If the expected discount factor ep of the order... i If the value is greater than γ or there is no driver matching the order, the ride request will be returned as a failure; otherwise, the ride request will be successful.

[0167] Additionally, the carpooling dispatch module includes a given discount factor determination module, which is used to find an order-based expected discount factor ep that satisfies the order set of deleted dispatched carpooling orders and the order set and vehicle set of cars that have reached capacity limits in the travel plan. i For the carpooling problem, considering the pick-up distance and detour constraints, all order-car pairs are placed into a pool, and the second immediate profit is calculated. The order-car pairs in the pool are then scheduled according to the principle of maximizing the second immediate profit to obtain the travel plan. Let the discount factor given by the platform be γ. If the discount factor γ given by the platform is greater than or equal to the expected discount factor ep of the order, i Furthermore, if the platform's total profit is greater than 0, the discount factor γ given by the platform will be considered an effective discount factor γ. 1 ; Calculate each effective discount factor γ 1 Based on the corresponding total platform profit, select the effective discount factor γ that maximizes the total platform profit. k Find the condition based on the effective discount factor γ. k Given the constraints of the carpooling problem, this module calculates the third-immediate profit for all feasible order-car pairs, schedules order-car pairs in the pool according to the principle of maximizing the third-immediate profit, and outputs the final route plan for the cars. In other words, the carpooling scheduling module assigns a discount value to each order based on its expected discount factor ep. i The travel plan is obtained by scheduling according to the principle of maximizing immediate profit. 0 j. According to the travel plan 0 j. Determine the approximately optimal discount factor γ k The value of γ. k As a uniform discount factor applied to orders by the platform, allocation is based on the principle of maximizing platform profits. Iterate through the plans. 0 j For all orders, calculate the effective discount γ for each order. 1 and invalid discount γ 0 And the changes in the coefficients of the total profit expression corresponding to these discounts. Effective discount γ 1 This refers to making a plan 0 j The critical discount for an order to change from an invalid state to an valid state is, specifically, if γ is greater than or equal to the expected discount factor ep of the order. i Furthermore, if the order profit is greater than 0, then γ is used as the effective discount factor for that order. Ineffective discount γ 0 This refers to making a plan 0 j The critical discount that causes an order in the plan to change from an effective state to an invalid state, specifically, makes the plan 0j The critical discount where the profit of orders in the plan begins to turn negative. Ascending order traversal of plan... 0 j For all orders with valid and invalid discounts, the total profit corresponding to each valid and invalid discount is calculated based on the coefficient changes in the total profit expression for each discount. The discount that maximizes the total profit is then found as the approximate optimal discount γ in the final output. k .

[0168] In summary, the embodiments of the present invention provide a discount-based ride-hailing dispatch method and system, which provides a ride-hailing mechanism of "exchanging waiting time for discounts", and allows users to define their maximum tolerable pick-up distance and expected discount factor. The platform provides a specific implementation method for a unified discount factor. Specifically, it provides a method for order scheduling and discount factor determination in non-carpooling scenarios, as well as a method for order scheduling and discount factor determination in carpooling scenarios.

[0169] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of the present invention, and these improvements and substitutions should also be considered within the scope of protection of the present invention.

Claims

1. A discount-based method for dispatching a car, characterized in that, Includes the following steps: S1: Obtain an order set O and a vehicle set V within a time window; the orders in order set O include the maximum tolerable pick-up distance and the expected discount; S2: Determine whether the orders in order set O are carpooling orders. If they are not carpooling orders, proceed to step S3. If they are carpooling orders, proceed to steps S4 to S7. S3: Dispatch non-carpooling orders, including: S301: If the order is in the No discount is given and the nearest car is dispatched to match the order. S302: If , according to the principle that the discount factor given by the platform is not lower than the expected discount factor of the order and maximizes the total profit of the platform, output the value of the optimal discount factor and the corresponding scheduling result; wherein, is the global pickup distance limit set by the platform, is the pickup distance, is the maximum tolerable pickup distance limit of the customer; S4: Schedule orders in the global pickup range: Find all the order-car pairs in the global pickup distance range and put them into ; without giving any discount, calculate the first instant profit benefit, and schedule the orders according to the principle of maximizing the first instant profit benefit; in step S4, let the first instant profit benefit be , which is calculated as follows: ； wherein is the price of the order , is the labor and fuel cost paid to the driver per unit of distance traveled, is the travel distance added to the path planning of the car to insert the order , is the base fare, is an indicator function that judges whether the car is empty or not. S5: delete the ride-sharing orders in the order set O and the vehicle set V that have been dispatched in step S4 and the cars that reach the capacity limit in the travel plan to obtain a new order set O' and a new vehicle set V'; in the new order set O' and the new vehicle set V', find the order-car pairs that meet the expectation discount factor based on the order all the order-car pairs of the pickup distance constraint and the detour constraint of the ride-sharing problem, put them into a pool, and calculate a second instant profit yield; dispatch the order-car pairs in the pool according to the principle of maximizing the second instant profit yield to obtain a travel plan ; In step S5, the second instant profit yield is calculated as follows: ； in, Distance for picking up the guest The platform has a global pick-up distance limit. This refers to the insertion This increased the pick-up distance for other orders, resulting in a higher total discount paid to those orders. S6: Let the discount factor given by the platform be... If the platform provides a discount factor Greater than or equal to the expected discount factor of the order And when the platform's total profit is greater than 0, the discount factor given by the platform will be applied. Considered as an effective discount factor ; Calculate each effective discount factor Based on the corresponding total platform profit, select the effective discount factor that maximizes the total platform profit. ; S7: Effective discount factor obtained from step S6 Find the solution based on the effective discount factor. Given the constraints of the carpooling problem, calculate the third-instance profit for all feasible order-car pairs, schedule the order-car pairs in the pool according to the principle of maximizing the third-instance profit, and output the final route plan for the cars.

2. The discount-based ride-hailing dispatch method according to claim 1, characterized in that, In step S302, the specific process for determining the optimal discount factor is as follows: S302-1: According to The non-carpooling orders and cars in the order set O and vehicle set V are matched to obtain order-car pairs; S302-2: Iterate through all order-car pairs, find valid discount factors and add them to the valid discount set D. Sort the discount factors in the valid discount set D in ascending order, and uniformly sample k discount factors from these sorted discount factors. Calculate the scheduling result for these k discount factors and their profit; select the discount factor with the highest profit. ; S302-3: In the discount factor Continue sampling evenly in the vicinity and select the discount factor that yields the greatest profit. 302-4: Repeat step S302-3 until the discount factor that maximizes profit is the same as the largest discount factor obtained from the previous sampling. Output this discount factor as the optimal discount factor.

3. The discount-based ride-hailing dispatch method according to claim 2, characterized in that, In step S302-3, the function FINDPEAK(l,r, k) is used for sampling, where l is the index of the first element in the sampling range, r is the index of the last element, and k is the number of sampling discount factors.

4. The discount-based ride-hailing dispatch method according to claim 1, characterized in that, In step S6, if the discount factor given by the platform... Expected discount factor greater than or equal to the order And the order is in the travel plan When the profit is equal to 0, the discount factor given by the platform will be applied. Considered an invalid discount factor Invalid discount factor for each order and effective discount factor The discount factor given by the platform set ; Let the total profit of the platform be The calculation is as follows: ； Where a, b, c, and d represent , , , ; Iterate through all travel plans obtained in step S5 Calculate each On each order and and the corresponding discount factor a, b、 c. d, then add all discount factors to the valid discount set. ; Traverse the valid discount set in ascending order Calculate the coefficients a, b, c, d and the corresponding profits for all discount factors; output the total platform profit as follows: Maximum discount factor .

5. The discount-based ride-hailing dispatch method according to claim 1, characterized in that, In step S7, the third immediate profit is The calculation is as follows: 。 6. A discount-based ride-hailing dispatch system, used to implement the discount-based ride-hailing dispatch method according to any one of claims 1-5, characterized in that, include: The order set and vehicle set acquisition module is used to acquire the order set and vehicle set within a time window; The order determination module is used to determine whether an order in the order set is a carpooling order, and sends non-carpooling orders to the non-carpooling dispatch module and carpooling orders to the carpooling dispatch module. The non-carpooling dispatch module is used to dispatch vehicles based on the order's pick-up distance, the platform's global pick-up distance limit, and the customer's maximum tolerable pick-up distance limit. When the order's pick-up distance is less than the platform's global pick-up distance limit, no discount is given, and the nearest car is dispatched to match the order. When the order's pick-up distance is greater than the platform's global pick-up distance limit but less than or equal to the customer's maximum tolerable pick-up distance, the module outputs the optimal discount factor value and the corresponding dispatch result, based on the principle that the platform's discount factor should not be lower than the order's expected discount factor and maximize the platform's total profit. The carpooling dispatch module is used to dispatch vehicles based on the order's pick-up distance, the platform's global pick-up distance limit, and the customer's maximum tolerable pick-up distance limit; it identifies all order-vehicle pairs within the global pick-up distance range and places them into the carpooling dispatch system. In the process, without any discounts, the platform calculates the first immediate profit and schedules orders based on maximizing the first immediate profit. Then, it removes the scheduled carpooling orders and cars that have reached capacity limits in the travel plan. For the remaining carpooling orders and cars, the platform applies the same discount factor, which is no less than the expected discount factor of the order, and schedules them based on the principle of maximizing platform profits.

7. The discount-based ride-hailing dispatch system according to claim 6, characterized in that, The non-carpooling dispatch module includes an optimal discount factor determination unit. This unit matches non-carpooling orders with cars to obtain order-car pairs based on the order's pick-up distance being greater than the platform's global pick-up distance limit but less than or equal to the customer's maximum tolerable pick-up distance. It then iterates through all order-car pairs, finds valid discount factors, adds them to a valid discount set D, sorts the discount factors in D in ascending order, and uniformly samples k discount factors from this sorted set. It then calculates the dispatch result for each of these k discount factors and their profit. Finally, it selects the discount factor with the highest profit. ; in discount factor Continue sampling evenly in the vicinity and select the discount factor with the highest profit. Repeat the above operation of sampling in the vicinity of the discount factor with the highest profit until the discount factor with the highest profit in the current sampling is the same as the largest discount factor obtained in the previous sampling. Output this discount factor as the optimal discount factor.

8. The discount-based ride-hailing dispatch system according to claim 6, characterized in that, The carpooling dispatch module includes a given discount factor determination module, which is used to find an order-based expected discount factor that satisfies the carpooling order set and vehicle set that have reached capacity limits in the travel plan, after the carpooling orders have been deleted and dispatched. For the carpooling problem, considering the pick-up distance and detour constraints, all order-car pairs are placed into a pool, and the second immediate profit is calculated. The order-car pairs in the pool are then scheduled according to the principle of maximizing the second immediate profit to obtain the travel plan. Let the discount factor given by the platform be... If the platform provides a discount factor Greater than or equal to the expected discount factor of the order And when the platform's total profit is greater than 0, the discount factor given by the platform will be applied. Considered as an effective discount factor ; Calculate each effective discount factor Based on the corresponding total platform profit, select the effective discount factor that maximizes the total platform profit. Find the solution based on the effective discount factor. Given the constraints of the carpooling problem, calculate the third-instance profit for all feasible order-car pairs, schedule the order-car pairs in the pool according to the principle of maximizing the third-instance profit, and output the final route plan for the cars.

Images