Online car-hailing routing and rebalancing scheduling method and system based on rolling time domain

Abstract

The invention belongs to the field of intelligent transportation, and particularly relates to an online car-hailing routing and rebalancing scheduling method and system based on a rolling time domain, and the method comprises the steps: obtaining an online car-hailing operation region and an operation time period, and dividing the operation region and the operation time period into a plurality of service regions and a plurality of time domain sequences; dividing the time domain sequence to form a vehicle routing time domain window, a demand prediction time domain window and a rebalance scheduling time domain window, wherein two adjacent time domain windows are partially overlapped; acquiring a vehicle state and a travel request of a passenger; in the vehicle routing time domain window, based on a real-time vehicle state and a travel request, a vehicle routing model is formed by taking # imgabs0 # as a target, and a vehicle path is optimized; forming a vehicle rebalance model by taking # imgabs1 # as a target based on vehicles in an idle state in a rebalance scheduling time domain window so as to obtain a cross-regional vehicle scheduling plan and updated regional vehicle distribution; and performing data acquisition of the next time sequence to form a window rolling closed loop.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent transportation, and particularly relates to a method and system for online car-hailing routing and rebalancing scheduling based on a rolling time domain. Background Art

[0002] As an important part of modern transportation services, the operation efficiency and service quality of online car-hailing directly affect the user experience and the market competitiveness of enterprises. However, there are significant disconnections between vehicle routing planning and rebalancing scheduling in existing online car-hailing scheduling methods, resulting in low overall scheduling efficiency. Specifically, vehicle routing planning usually only performs local optimization for immediate demands and fails to form a collaborative mechanism with rebalancing scheduling; while rebalancing scheduling lacks the linkage support for dynamically adjusting real-time routing during execution. This fragmented operation mode not only increases the ineffective driving loss of vehicles but also makes it difficult to maintain the supply-demand balance between regions, thus restricting the improvement of operation efficiency and service quality. The present invention proposes a method for online car-hailing routing and rebalancing scheduling based on a rolling time domain, which combines short-term demand prediction to achieve a balance between global optimization and real-time response, and improve operation efficiency and service quality. Summary of the Invention

[0003] The present invention is precisely proposed based on the above-mentioned needs of the prior art. The technical problem to be solved by the present invention is to provide a method and system for online car-hailing routing and rebalancing scheduling based on a rolling time domain to improve operation efficiency.

[0004] To solve the above problems, the technical solutions provided by the present invention include:

[0005] Provided is a method for online car-hailing routing and rebalancing scheduling based on a rolling time domain, including: obtaining the operation area and operation time period of online car-hailing; dividing the operation area of online car-hailing into multiple non-overlapping service areas, and dividing the operation time period of online car-hailing into multiple non-overlapping time domain sequences; dividing the time domain sequences to form a vehicle routing time domain window, a demand prediction time domain window, and a rebalancing scheduling time domain window, with partial overlap between adjacent two time domain windows; obtaining the vehicle status of each online car-hailing vehicle and the travel requests of passengers in each service area during the operation time period, where the vehicle status includes idle or occupied, and the travel request includes a pick-up point and a drop-off point; within the vehicle routing time domain window, based on the real-time vehicle status and travel requests, to form a vehicle routing model with the goal of optimizing the vehicle path, where c ij is a parameter representing the driving cost of the vehicle from node i to node j; is a decision variable, indicating that vehicle k in service area r∈R travels from node i to node j, otherwise z ijk =0, P r is the travel request in service area r, Vr is the set of all vehicles in service area r; and simultaneously satisfies where C k represents the maximum passenger capacity of vehicle k, represents the time when vehicle k arrives at node i in each service area r ∈ R, represents the travel time of vehicle k from node i to node j within service area r, T max represents the maximum allowable travel time; the prediction time domain window is divided into multiple time intervals, and based on historical vehicle states and travel requests, the future travel demand is predicted using a short-term prediction model; within the rebalancing scheduling time domain window, based on the idle vehicles, with as the objective, a vehicle rebalancing model is formed to obtain a cross-regional vehicle scheduling plan and the updated regional vehicle distribution, where: r' ≠ r, δ r' is the service level difference variable of service area r', is a decision variable, represents that vehicle k is dispatched from service area r to service area r', otherwise represents the time when vehicle k is dispatched from service area r to service area r'; W is the weight parameter for the importance degree of balancing the service level difference δ r' and the vehicle dispatching time , and simultaneously satisfies: Data collection for the next time series is performed to form a rolling closed loop of the vehicle routing time domain window, the demand prediction time domain window, and the rebalancing scheduling time domain window.

[0006] Preferably, in the k-th time domain sequence, the start time of the vehicle routing time domain window is expressed as where is the start time of the rebalancing scheduling time domain window in the (k - 1)-th time domain sequence, t C is the length of the rebalancing scheduling time domain window, Δt C-A represents the overlapping length between the rebalancing scheduling time domain window of time domain sequence k - 1 and the vehicle routing time domain window of time domain sequence k.

[0007] Preferably, in the k-th time domain sequence, the start time of the demand prediction time domain window is expressed as t A is the length of the vehicle routing time domain window, Δt A-B represents the overlapping length between the vehicle routing time domain window of time domain sequence k and the demand prediction time domain window of time domain sequence k.

[0008] Preferably, in the k-th time domain sequence, the start time of the rebalancing scheduling time domain window is expressed as t B is the length of the demand prediction time domain window, and Δt B-C represents the overlapping length between the demand prediction time domain window of the time domain sequence k and the rebalancing scheduling time domain window of the time domain sequence k.

[0009] Preferably, the time domain sequence is expressed as t = t A + t B + t C - Δt A-B - Δt B-C .

[0010] There is also provided a ride-hailing routing and rebalancing scheduling system based on a rolling time domain, including: an acquisition module for acquiring the operation area and operation time period of ride-hailing; an operation area division module for dividing the operation area of ride-hailing into a plurality of non-overlapping service areas; an operation time period division module for dividing the operation time period of ride-hailing into a plurality of non-overlapping time domain sequences; a time domain sequence division module for dividing the time domain sequence into a vehicle routing time domain window, a demand prediction time domain window, and a rebalancing scheduling time domain window, with partial overlap between adjacent two time domain windows; a data acquisition module for acquiring the vehicle status of each ride-hailing vehicle and the travel requests of passengers in each service area during the operation time period, where the vehicle status includes idle or occupied, and the travel requests include the boarding point and the drop-off point; a vehicle routing window processing module, within the vehicle routing time domain window, based on the real-time vehicle status and travel requests, taking as the objective to form a vehicle routing model and optimize the vehicle path, where c ij is a parameter representing the driving cost of the vehicle from node i to node j; is a decision variable, represents that vehicle k in service area r ∈ R travels from node i to node j, otherwise z ijk = 0, P r is the travel request in service area r, and V r is the set of all vehicles in service area r; and simultaneously satisfying where C k represents the maximum passenger capacity of vehicle k, represents the time when vehicle k arrives at node i in each service area r ∈ R, represents the driving time of vehicle k from node i to node j within service area r, and T maxdenotes the maximum allowable driving time; a prediction time-domain window processing module divides the prediction time-domain window into multiple time intervals, and based on the historical vehicle state and travel requests, predicts future travel demands using a short-term prediction model; a rebalancing scheduling time-domain window processing module, based on idle vehicles, forms a vehicle rebalancing model with as the target to obtain a cross-regional vehicle scheduling plan and an updated regional vehicle distribution, where: r'≠r, δ r' is the service level difference variable of service area r', is a decision variable, denotes that vehicle k is scheduled from service area r to service area r', otherwise denotes the time when vehicle k is scheduled from service area r to service area r'; W is the weight parameter for the importance of balancing the service level difference δ r' and the vehicle scheduling time and satisfies:

[0011] Compared with the prior art, the present invention divides the operation area and operation time period of online car-hailing, divides the divided time-domain sequence into time-domain windows of multiple different stages, and there is partial overlap between adjacent windows. Different models are established within the corresponding time-domain windows of different stages to conform to the optimal solution within the corresponding windows. The total driving cost is minimized within the vehicle routing time-domain window, and the vehicle scheduling cost is minimized within the rebalancing scheduling window. Thus, the global online car-hailing is optimized according to the demand, and a collaborative mechanism is formed between vehicle routing and rebalancing scheduling for dynamic adjustment, acting and supporting each other. At the same time, combined with short-term demand prediction, the balance between global optimization and real-time response is achieved, thereby improving the operation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments recorded in the embodiments of this specification. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0013] Figure 1 is the step flowchart of a method for online car-hailing routing and rebalancing scheduling based on a rolling time domain according to the present invention;

[0014] Figure 2 is the schematic diagram of the time-domain sequence division window in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0016] In the description of the embodiments of the present invention, it should be noted that unless otherwise clearly defined and limited, the term "connected" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection. It may be a mechanical connection or an electrical connection. It may be directly connected or indirectly connected through an intermediate medium. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0017] The terms "top", "bottom", "above...", "below", and "on..." described throughout the text are relative positions with respect to the components of the device, such as the relative positions of the top and bottom substrates inside the device. It can be understood that the device is multifunctional and is independent of its orientation in space.

[0018] To facilitate the understanding of the embodiments of the present invention, the following will further explain with specific embodiments in conjunction with the accompanying drawings. The embodiments do not constitute a limitation to the embodiments of the present invention.

[0019] Embodiment 1

[0020] This embodiment provides a rolling horizon-based online car-hailing routing and rebalancing scheduling method.

[0021] As Figure 1 shown, the rolling horizon-based online car-hailing routing and rebalancing scheduling method includes:

[0022] Obtain the operation area and operation time period of the online car-hailing.

[0023] Divide the operation area of the online car-hailing into multiple non-overlapping service areas, and divide the operation time period of the online car-hailing into multiple non-overlapping time domain sequences.

[0024] All service areas form a set R, and r ∈ R represents the serial number of a certain service area.

[0025] All time domain sequences form a set K, and k ∈ K represents the serial number of a certain time domain sequence.

[0026] Divide the time domain sequence into three time domain windows, and adjacent time domain windows partially overlap.

[0027] In chronological order, the three time-domain windows are the vehicle routing time-domain window, the demand forecasting time-domain window, and the rebalancing scheduling time-domain window. The vehicle routing time-domain window is used for real-time path planning, the demand forecasting time-domain window is used for predicting future demands, and the rebalancing scheduling time-domain window is used for cross-regional vehicle scheduling.

[0028] Specifically, as Figure 2 shown, the start time of the vehicle routing time-domain window is and the window length is t A ; the start time of the demand forecasting time-domain window is and the window length is t B ; the start time of the rebalancing scheduling time-domain window is and the window length is t C .

[0029] Among them, the start times of the above three time-domain windows satisfy:

[0030]

[0031] Among them, Δt C-A represents the overlapping length between the rebalancing scheduling time-domain window of time-domain sequence k - 1 and the vehicle routing time-domain window of time-domain sequence k; Δt A-B represents the overlapping length between the vehicle routing time-domain window of time-domain sequence k and the demand forecasting time-domain window of time-domain sequence k; Δt B-C represents the overlapping length between the demand forecasting time-domain window of time-domain sequence k and the rebalancing scheduling time-domain window of time-domain sequence k.

[0032] The window lengths of each time-domain window satisfy: t = t A + t B + t C - Δt A-B - Δt B-C .

[0033] In chronological order, execute time-domain sequences k = 1, 2,..., |K| in sequence. For time-domain sequence k, execute according to the following specific steps.

[0034] Obtain the vehicle status of each online car-hailing vehicle and the travel requests of passengers in each service area during the operation period. The vehicle status includes idle or occupied, and the travel request includes the boarding point and the drop-off point.

[0035] Among them, V r is the set of all vehicles in service area r; N r is the set of all passengers in service area r.

[0036] The vehicle status is denoted as where: Indicates that the vehicle is in an idle state, Indicates that the vehicle is in a passenger-carrying state.

[0037] The travel request includes the passenger boarding point and the alighting point which is the set of boarding and alighting points for all passengers in service area r.

[0038] Within the vehicle routing time window, based on the real-time vehicle state and travel requests, a vehicle routing model is formed with the goal of

[0039] where c ij is a parameter representing the driving cost of the vehicle from node i to node j; is a decision variable, indicating that vehicle k travels from node i to node j within service area r ∈ R, otherwise

[0040] By setting the above goal, the total driving cost is minimized.

[0041] In addition, the vehicle routing model also needs to satisfy:

[0042]

[0043] By the above settings, it is ensured that the boarding and alighting points of each passenger must be visited.

[0044]

[0045] where C k represents the maximum passenger capacity of vehicle k. By the above settings, the continuity of the vehicle route is ensured, and the number of passengers carried by the vehicle does not exceed its capacity limit.

[0046]

[0047] By the above settings, it is ensured that only idle vehicles can be assigned to pick up passengers.

[0048]

[0049] where, represents the time when vehicle k arrives at node i within each service area r ∈ R, represents the travel time of vehicle k from node i to node j within service area r, and T max represents the maximum allowed travel time. By the above settings, it is ensured that the total travel time of the vehicle does not exceed the allowed maximum value.

[0050] Divide the prediction time domain window into multiple time intervals, and based on the historical vehicle status and travel requests, predict the future travel demand using a short-term prediction model.

[0051] With ΔT as an interval, divide the prediction time domain window into T equal-length time intervals t1, t2,..., t T , and within each service area r ∈ R, statistically count the travel demand of passengers within each time interval in chronological order Generate the future passenger travel demand d based on the short-term traffic demand prediction model t+1 , that is:

[0052]

[0053] where f(·) represents a set of various functions, and the functions can adopt historical average models, multi-source linear regression models, random forest models, and graph convolutional neural networks.

[0054] Within the rebalancing scheduling time domain window, based on the idle vehicles, form a vehicle rebalancing model with as the goal to obtain the cross-regional vehicle scheduling plan and the updated regional vehicle distribution.

[0055] where: r' ≠ r, δ r' is the service level difference variable of service area r', reflecting the difference between the actual vehicle scheduling situation and the ideal uniform scheduling state. The smaller δ r' , the closer the service level of area r' is to the ideal situation; is a decision variable, represents that vehicle k is scheduled from service area r to service area r', otherwise represents the time when vehicle k is scheduled from service area r to service area r'; W is the weight parameter for balancing the importance of the service level difference δ r' and the vehicle scheduling time .

[0056] Execute the cross-service area scheduling of idle online car-hailing vehicles through the above settings.

[0057] In addition, the vehicle rebalancing model also needs to satisfy:

[0058]

[0059] Make each vehicle scheduled at most once through the above limitations.

[0060]

[0061] Through the above limitations, the number of vehicles dispatched to the service area r' does not exceed the predicted demand of the service area r'.

[0062]

[0063]

[0064] Through the above limitations, in an ideal situation, vehicles are evenly dispatched to each area.

[0065]

[0066] Through the above limitations, when the demand of each service area and the total number of available vehicles do not exactly match, the difference in service levels among service areas is minimized, and through linearization is performed.

[0067] Data collection for the next time series is carried out to form a rolling closed-loop of the vehicle routing time domain window, the demand prediction time domain window, and the rebalancing scheduling time domain window.

[0068] Embodiment 2

[0069] This embodiment provides an online car-hailing routing and rebalancing scheduling system based on a rolling time domain.

[0070] The online car-hailing routing and rebalancing scheduling system based on a rolling time domain includes:

[0071] An acquisition module, configured to acquire the operation area and operation time period of the online car-hailing.

[0072] An operation area division module, configured to divide the operation area of the online car-hailing into multiple non-overlapping service areas.

[0073] An operation time period division module, configured to divide the operation time period of the online car-hailing into multiple non-overlapping time domain sequences.

[0074] A time domain sequence division module, configured to divide the time domain sequence into a vehicle routing time domain window, a demand prediction time domain window, and a rebalancing scheduling time domain window, and there is partial overlap between adjacent two time domain windows.

[0075] A data collection module, which acquires the vehicle status of each online car-hailing and the travel requests of passengers in each service area during the operation time period, where the vehicle status includes idle or occupied, and the travel request includes the boarding point and the alighting point.

[0076] A vehicle routing window processing module, within the vehicle routing time domain window, based on the real-time vehicle status and travel requests, with as the goal, forms a vehicle routing model and optimizes the vehicle path, where c ijis a parameter representing the driving cost of a vehicle from node i to node j; is a decision variable, indicating that vehicle k travels from node i to node j within service area r ∈ R, otherwise P r is a travel request in service area r, V r is the set of all vehicles in service area r; simultaneously satisfying where C k represents the maximum passenger capacity of vehicle k, represents the time when vehicle k arrives at node i within each service area r ∈ R, represents the driving time of vehicle k from node i to node j within service area r, T max represents the maximum allowed driving time.

[0077] The prediction time domain window processing module divides the prediction time domain window into multiple time intervals, and based on the historical vehicle states and travel requests, predicts the future travel demands using a short-term prediction model.

[0078] The rebalancing scheduling time domain window processing module, based on the idle vehicles, forms a vehicle rebalancing model with as the goal to obtain a cross-regional vehicle scheduling plan and the updated regional vehicle distribution, where: r' ≠ r, δ r' is the service level difference variable of service area r', is a decision variable, indicating that vehicle k is dispatched from service area r to service area r', otherwise represents the time when vehicle k is dispatched from service area r to service area r'; W is the weight parameter of the importance degree of balancing the service level difference δ r' and the vehicle dispatching time simultaneously satisfying:

[0079] The specific embodiments described above further elaborate on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for online car-hailing routing and rebalancing based on rolling time domain, characterized in that: include: Obtain the online car-hailing service operation areas and operation hours; Divide the online car-hailing operation area into multiple non-overlapping service areas, and divide the online car-hailing operation period into multiple non-overlapping time domain sequences; The time domain sequence is divided into vehicle routing time domain window, demand forecasting time domain window and rebalancing scheduling time domain window, and there is partial overlap between two adjacent time domain windows; Obtain the vehicle status of each online-hailing vehicle and the travel requests of passengers in each service area during the operation period, wherein the vehicle status includes idle or carrying passengers, and the travel requests include the pick-up point and the drop-off point; In the vehicle routing time domain window, based on the real-time vehicle status and travel requests, The vehicle routing model is formed for the goal to optimize the vehicle path, where c ij is a parameter, which represents the driving cost of the vehicle from node i to node j; is the decision variable, Indicates that vehicle k in service area r∈R travels from node i to node j, otherwise P r is the travel request in service area r, V r is the set of all vehicles in the service area r; and Among them, C k represents the maximum passenger capacity of vehicle k, represents the time it takes for vehicle k to arrive at node i in each service area r∈R, represents the travel time of vehicle k from node i to node j in service area r, T max represents the maximum allowed travel time, Divide the prediction time window into multiple time intervals, and predict future travel demand based on the short-term prediction model according to the historical vehicle status and travel requests; In the rebalancing scheduling time window, based on the idle state of the vehicle, The vehicle rebalancing model is formed for the goal to obtain the cross-regional vehicle scheduling plan and the updated regional vehicle distribution, where: r'≠r, δ r' is the service level difference variable of service area r', is the decision variable, indicates that vehicle k is dispatched from service area r to service area r', otherwise represents the time it takes for vehicle k to be dispatched from service area r to service area r'; W is the equilibrium service level difference δ r' Vehicle dispatch time The weight parameter of importance satisfies: Carry out data collection for the next time series to form a rolling closed loop of the vehicle routing time domain window, the demand forecasting time domain window and the rebalancing scheduling time domain window.

2. The method for online car-hailing routing and rebalancing based on rolling time domain according to claim 1 is characterized in that: In the kth time domain sequence, the starting time of the vehicle routing time domain window is It is expressed as, in is the starting time of the rebalancing scheduling time domain window in the k-1th time domain sequence, t C is the length of the rebalancing scheduling time window, Δt C-A Represents the overlapping length of the rebalancing scheduling time domain window of time domain sequence k-1 and the vehicle routing time domain window of time domain sequence k.

3. The method for online car-hailing routing and rebalancing based on rolling time domain according to claim 2 is characterized in that: In the kth time domain sequence, the starting time of the demand forecast time domain window is It is expressed as, t A is the length of the vehicle routing time domain window, Δt A-B Represents the overlapping length of the vehicle routing time domain window of time domain sequence k and the demand forecast time domain window of time domain sequence k.

4. The method for online car-hailing routing and rebalancing based on rolling time domain according to claim 3 is characterized in that: In the kth time domain sequence, the start time of the rebalancing scheduling time domain window is It is expressed as, t B is the length of the demand forecast time window, Δt B-C Represents the overlapping length of the demand forecast time domain window of time domain sequence k and the rebalancing scheduling time domain window of time domain sequence k.

5. The method for online car-hailing routing and rebalancing based on rolling time domain according to claim 4 is characterized in that: The time domain sequence is represented by t=t A +t B +t C -Δt A-B -Δt B-C .

6. A network car-hailing routing and rebalancing scheduling system based on rolling time domain, characterized in that: include: The acquisition module is used to obtain the online car-hailing operation area and operation period; The operation area division module is used to divide the online car-hailing operation area into multiple non-overlapping service areas; The operation period division module is used to divide the online car-hailing operation period into multiple non-overlapping time domain sequences; A time domain sequence division module is used to divide the time domain sequence into a vehicle routing time domain window, a demand forecasting time domain window and a rebalancing scheduling time domain window, and two adjacent time domain windows overlap at least partially; A data collection module is used to obtain the vehicle status of each online car-hailing vehicle and the travel requests of passengers in each service area during the operation period, wherein the vehicle status includes idle or carrying passengers, and the travel requests include the pick-up point and the drop-off point; The vehicle routing window processing module is based on the real-time vehicle status and travel request within the vehicle routing time domain window. The vehicle routing model is formed for the goal to optimize the vehicle path, where c ij is a parameter, which represents the driving cost of the vehicle from node i to node j; is the decision variable, Indicates that vehicle k in service area r∈R travels from node i to node j, otherwise P r is the travel request in service area r, V r is the set of all vehicles in the service area r; and Among them, C k represents the maximum passenger capacity of vehicle k, represents the time it takes for vehicle k to arrive at node i in each service area r∈R, represents the travel time of vehicle k from node i to node j in service area r, T max Indicates the maximum allowed travel time; The prediction time domain window processing module divides the prediction time domain window into multiple time intervals, and predicts future travel demand based on the short-term prediction model according to the historical vehicle status and travel requests; The rebalancing scheduling time domain window processing module is based on idle vehicles. The vehicle rebalancing model is formed for the goal to obtain the cross-regional vehicle scheduling plan and the updated regional vehicle distribution, where: r'≠r, δ r' is the service level difference variable of service area r', is the decision variable, indicates that vehicle k is dispatched from service area r to service area r', otherwise represents the time it takes for vehicle k to be dispatched from service area r to service area r'; W is the equilibrium service level difference δ r' Vehicle dispatch time The weight parameter of importance satisfies:

Images