A shared bicycle scheduling method based on station clustering
Through the shared bicycle scheduling method based on site clustering, the problems of vehicle shortage and surplus are solved, resource distribution is optimized, computing efficiency and scheduling efficiency are improved, and efficient resource utilization is achieved.
Patent Information
- Application Number
- CN202211678229.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-26
- Publication Date
- 2025-09-16
- Estimated Expiration
- 2042-12-26
AI Technical Summary
Existing shared bicycle scheduling methods fail to effectively solve the problems of vehicle shortage and surplus, resulting in uneven resource distribution, and the global optimization calculation cost is too high and inefficient.
A shared bicycle scheduling method based on site clustering is adopted. By dividing the system into multiple planning periods, the sites are clustered using a clustering algorithm, and cluster adjustments are made based on distance and average loss to form an independent intra-cluster scheduling plan. The vehicle scheduling is optimized by combining geographic location and user flow information.
It effectively reduces the losses caused by uneven resource distribution, improves overall computing efficiency, and improves vehicle scheduling efficiency and resource utilization.
Smart Images
Figure CN115964647B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to shared bicycle scheduling, and in particular to a shared bicycle scheduling method based on site clustering. Background Art
[0002] Bike-sharing is a widely used, environmentally friendly, and flexible mode of urban transportation. However, its flexibility also means that user mobility can lead to uneven distribution of system resources. To avoid this loss, operating companies must systematically dispatch vehicles using carriers while the system is running, a process known as dynamic scheduling.
[0003] Existing scheduling methods reduce losses by minimizing vehicle shortages within the system, but they ignore the losses caused by vehicle surpluses and parking shortages. Furthermore, when used in large-scale, real-world shared bike systems, global optimization methods are often computationally expensive and inefficient. Summary of the Invention
[0004] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a shared bicycle scheduling method based on site clustering, which takes into account the problems of vehicle shortage and surplus, thereby truly minimizing the losses caused by uneven resource allocation, and takes into account geographical location and user traffic. When applied to the vehicle dynamic scheduling planning method, it can effectively improve efficiency.
[0005] The object of the present invention is achieved through the following technical solution: a shared bicycle scheduling method based on site clustering, comprising the following steps:
[0006] S1. Divide the operation period of the shared bicycle system into multiple planning periods;
[0007] S2. For the tth planning period, construct a shared bicycle system model;
[0008] S3. During the tth planning period, cluster the bike-sharing system stations using a clustering algorithm. The clustering process first divides the stations into clusters based on distance, and then uses the average loss as a measure of the scheduling demand within the cluster to adjust the clusters.
[0009] S4. Perform scheduling planning for shared bicycles in each cluster formed during the t-th planning period.
[0010] S5. Repeat steps S2 to S4 in different planning periods to complete the scheduling plan of shared bicycles in each planning period.
[0011] The beneficial effects of the present invention are: the present invention reduces the losses caused by uneven resource distribution while improving the overall computing efficiency through a dynamic vehicle scheduling planning method that jointly considers the problems of bicycle shortage and surplus and a clustering algorithm that jointly considers geographical location and user traffic. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 is a flow chart of the method of the present invention;
[0013] Figure 2 The following is a specific clustering flow chart. DETAILED DESCRIPTION
[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the following.
[0015] like Figure 1 As shown, a shared bicycle scheduling method based on site clustering includes the following steps:
[0016] S1. Divide the operation period of the shared bicycle system into multiple planning periods;
[0017] S2. For the tth planning period, construct a shared bicycle system model;
[0018] S201. Given system resource information, including:
[0019] Station set S: contains the capacity of each station in the shared bicycle system. The capacity of a station is the number of parking spaces in the station. In the station set, the capacity of station s is recorded as The number of vehicles at planning period t is denoted as
[0020] S202. Given user traffic information, including user traffic within the planning period t in K training scenarios from historical data, is the number of users who travel from station s to station s′ within the planning period t in the kth training scenario;
[0021] S203. Loss is defined as the number of times user needs cannot be met with existing resources, specifically including:
[0022] Vehicle loss, that is, the loss caused by the user not finding an available vehicle, is the vehicle loss calculation that occurs at station s in the kth training scenario; when no scheduling is performed, it is calculated as
[0023]
[0024] Parking space loss, that is, the loss caused by the user not finding a parking space to return the car, is the parking space loss that occurs at station s in the kth training scenario; when no scheduling is performed, it is calculated as
[0025]
[0026] The total loss is the sum of vehicle loss and parking space loss.
[0027] S3. During the tth planning period, cluster the bike-sharing system stations using a clustering algorithm. The clustering process first divides the stations into clusters based on distance, and then uses the average loss as a measure of the scheduling demand within the cluster to adjust the clusters.
[0028] To effectively plan vehicle scheduling in a large-scale shared bicycle system, a clustering algorithm is proposed. By performing scheduling planning within the formed clusters, the problem size is reduced and computational efficiency is improved.
[0029] The core idea of the clustering algorithm is to make the scheduling requirements within each cluster equivalent by jointly considering the geographical location and user traffic. In this way, while improving efficiency through local scheduling, it can also ensure the scheduling effect of reducing losses.
[0030] The loss of a single station, i.e. the loss incurred at a station when no scheduling is performed, is used as a measure of the scheduling demand of a station. To measure the scheduling demand of a station set, for any station set P, the average loss is defined as
[0031]
[0032] Among them, P +，k is,the set of stations with non-zero clustering loss in scenario k;
[0033] like Figure 2 The specific process is as follows:
[0034] Step 1: Enter system resources and user traffic information; specify the number of generated clusters as C and the number of abnormal station sets as 1;
[0035] Step 2: Pre-set the generated C clusters to be empty sets and the abnormal station set O to be empty sets;
[0036] Step 3: Calculate the average system loss
[0037]
[0038] Among them, S +,k is,the set of stations with non-zero system loss in patrol training scenario k.,The average loss of this system will be used as the comparison criterion in the clustering process;
[0039] Step 4: On average, select C stations in the system and set them as the center stations of C clusters. For example, if 100 stations are divided into 10 clusters, the 1st, 11th, ..., 91st stations can be selected as the center stations of 10 clusters respectively.
[0040] Step 5: Define the distance between station s and any cluster c as the distance between station s and the intermediate station of cluster c; based on this, allocate all stations in the system to the cluster with the closest distance to them.
[0041] Step 6: For each cluster c, calculate the average loss
[0042]
[0043] where P +,k is the set of stations with non - zero cluster loss in scenario k; compare AL c with the system average loss AL g ; if AL c is greater than AL g , then successively move the outermost station in cluster c, that is, the station with the farthest distance from its central station in cluster c, into the abnormal station set O until the average loss of the remaining stations in cluster c is less than AL g .
[0044] Step 7: Allocate the abnormal stations in the abnormal station set O to the cluster with the smallest current average loss one by one. If the average loss of this cluster exceeds the system average loss AL g after receiving the abnormal station, move the abnormal station back to the abnormal station set.
[0045] Step 8: Allocate the abnormal stations returned to the abnormal station set O in Step 7 to the cluster with the smallest current average loss one by one.
[0046] In the embodiment of this application, assume that after the end of Step 7, there are still three abnormal stations, namely stations 3, 6, and 8, in the abnormal station set. Assume that there are currently four clusters, 1, 2, 3, and 4, and the current average loss values of these four clusters have the size relationship of AL3 < AL2 < AL1 < AL4. Then Step 8 is to first allocate station 3 to AL3 (because the average loss value of cluster 3 is the smallest). Assume that after the allocation, because adding station 3 to cluster 3, the current average loss value size relationship of these four clusters becomes AL2 < AL1 < AL3 < AL4; then the next step is to allocate station 6 to cluster 2. At this time, calculate the new average loss value of each cluster and sort them. If the relationship is still AL2 < AL1 < AL3 < AL4, then the last station 8 will still be allocated to cluster 2. And so on until all the abnormal stations in the abnormal station set are allocated to one of the clusters. However, different from Step 7, in Step 7, after any allocation, if the average loss of the cluster after receiving the abnormal station exceeds the system average loss AL g , the abnormal station is moved back to the abnormal station set, but in Step 8, the AL g value limit is no longer considered, and the remaining stations are directly allocated to the cluster with the smallest current AL
[0047] Step 9. Output the final C clusters.
[0048] During the clustering process, the clustering algorithm first divides the clusters based on distance, and then uses the average loss as a measure of the scheduling demand within the cluster to adjust the cluster. While all stations are clustered, the scheduling demand of each cluster is guaranteed to be as balanced as possible to ensure the effectiveness of the scheduling within the cluster.
[0049] S4. Perform scheduling planning for shared bicycles in each cluster formed during the t-th planning period.
[0050] Each cluster formed by the clustering algorithm is treated as an independent system, and planning and scheduling are performed within each cluster. The dynamic vehicle scheduling problem within a cluster is formulated as a mixed-integer nonlinear programming problem. Based on resource distribution and user traffic information, the goal is to minimize the total loss of the cluster in K training scenarios. Conventional methods are used to solve the problem and obtain the scheduling solution.
[0051] The user traffic trends within the same planning period are similar every day. Therefore, the scheduling scheme that minimizes the total loss in the training scenario is a general scheduling scheme. Specifically:
[0052] S401. Treat each cluster as an independent system and perform planning and scheduling within each cluster based on resource distribution and user traffic information.
[0053] S402. Among the C clusters, the system resource information of the cth cluster includes:
[0054] Station Collection c : Contains the capacity of each station in the c-th shared bicycle cluster. The capacity of the station is the number of parking spaces in the station. c In the example, the capacity of station s is recorded as
[0055] The vehicle set V contains the capacity of each vehicle in the c-th shared bicycle cluster. The capacity of the vehicle refers to the maximum number of vehicles it can carry. In the vehicle set, the capacity of the vehicle v is recorded as
[0056] Vehicle distribution: including represents the number of vehicles at station s in the cth shared bicycle cluster at planning period t;
[0057] Carrier distribution: including indicator variables The value of is 0 or 1. It means that at the beginning of planning period t, vehicle v is located at station s; Indicates that vehicle v is not located at station s during the planning period t;
[0058] The station location information is contained in the matrix H, which has S c OK, S c Columns, elements in the matrix H s,s′ represents the distance between two stations s, s′;
[0059] In the embodiment of the present application, taking a system with three stations as an example (i.e., c=3), the matrix H is a 3*3 matrix. The elements of this matrix represent the distance between any two stations. For example, the element H in the matrix 1,1 The distance from station 1 to station 1 is 0. The element H in the matrix 1,2 It is the distance between station 1 and station 2. Assuming that the distance between station 1 and station 2 is 10km according to the measurement in the system, then H 1,2 =10; Similarly, the element H in the matrix 2,1 It is the distance from station 2 to station 1, H 2,1 =H 1,2 =10, and the same applies to the rest of the elements in the matrix.
[0060] S403. Given user traffic information, including user traffic in the c-th shared bicycle cluster within the planning period t in K training scenarios from historical data, is the number of users who travel from station s to station s′ in the cth shared bicycle cluster within the planning period t in the kth training scenario;
[0061] S404. Define the loss in the cth shared bicycle cluster as the number of times that user demand cannot be met with existing resources, specifically including:
[0062] Vehicle loss, that is, the loss caused by the user not finding an available vehicle, is the vehicle loss calculation that occurs at station s in the c-th shared bicycle cluster in the k-th training scenario; when no scheduling is performed, it is calculated as
[0063]
[0064] Parking space loss, that is, the loss caused by the user not finding a parking space to return the car, is the parking space loss at station s in the c-th shared bicycle cluster in the k-th training scenario; when no scheduling is performed, it is calculated as
[0065]
[0066] The total loss in the c-th shared bicycle cluster is the sum of vehicle loss and parking space loss in the c-th shared bicycle cluster;
[0067] S405. Given the relevant variables of the scheduling plan:
[0068] is the number of vehicles picked up by dispatch vehicle v at station s;
[0069] is the number of vehicles returned by dispatch vehicle v at station s;
[0070] indicator variables Represents the dispatch vehicle v picking up the vehicle at station s;
[0071] indicator variables Let v return the vehicle at station s on behalf of the dispatcher.
[0072] S406. Dynamic vehicle scheduling problem in the c-th shared bicycle cluster:
[0073] This problem is a mixed integer nonlinear programming problem. Based on resource distribution and user traffic information, the goal is to minimize the total loss of the cth cluster in K training scenarios. The scheduling planning problem is as follows:
[0074] (1) Minimizing the total loss after target scheduling: The goal is to minimize the total loss after scheduling; the total loss is the sum of vehicle losses and parking space losses in all training scenarios and all stations:
[0075]
[0076] (2) Calculate the total vehicle loss after dispatch, that is, the difference between the number of vehicles required by users and the number of vehicles available at the station after dispatch:
[0077]
[0078] In each training scenario, each station, ensure vehicle loss Not less than the difference between the number of vehicles required by the user and the number of vehicles available at the station after dispatching; in the minimization problem, this is equivalent to calculating the vehicle loss as the difference between the number of vehicles required by the user and the number of vehicles available at the station after dispatching;
[0079] The number of vehicles required by the user is the number of vehicles going from this station to each station sum;
[0080] The number of vehicles available at the station after dispatching is the original number of vehicles at this station The sum of the net number of vehicles brought in at this station by each vehicle; the net number of vehicles brought in at this station by a vehicle is the number of vehicles picked up by the vehicle at this station. Subtract the number of returned vehicles
[0081] (3) Calculate the total parking space loss after scheduling, that is, the difference between the number of parking spaces required by users and the number of available parking spaces at the station after scheduling:
[0082]
[0083] In each training scenario and each station, ensure the loss of parking space Not less than the difference between the number of parking spaces required by the user and the number of available parking spaces at the station after dispatching;
[0084] In the minimization problem, it is equivalent to calculating the parking space loss as the difference between the number of parking spaces required by users and the number of available parking spaces at the station after scheduling;
[0085] The number of parking spaces required by the user is the number of vehicles heading to this station from each station sum;
[0086] The number of available parking spaces at the station after dispatching is the original number of vehicles at this station The sum of the net outgoing numbers of each transport vehicle at this station; the net outgoing number of a transport vehicle at this station is the number of vehicles returned at this station. Subtract the number of cars picked up
[0087] (4) Ensure that the number of vehicles picked up by the transport vehicle does not exceed the transport vehicle capacity: Ensure that at each station, for each transport vehicle, if it picks up a vehicle at this station Number of cars picked up No more than the original number of vehicles at the station or vehicle capacity
[0088]
[0089] (5) Ensure that the number of vehicles picked up by transport vehicles does not exceed the number of vehicles available at the pick-up station: Ensure that the total number of vehicles picked up by all transport vehicles at each station, that is, the number of vehicles picked up by each transport vehicle The sum of the number of vehicles available at the station shall not exceed the number of vehicles available at the station
[0090]
[0091] (6) Ensure that the number of returned transport vehicles does not exceed the number of available parking spaces at the return station: Ensure that, at each station, the total number of returned transport vehicles, i.e., the number of returned vehicles per transport vehicle, is equal to The sum of the number of available parking spaces at the station shall not exceed the station capacity. and the original number of vehicles Difference:
[0092]
[0093] (7) Ensure that the number of vehicles picked up by the transport vehicle is equal to the number of vehicles returned, that is, all the vehicles picked up are returned: Ensure that at each station, for each transport vehicle, if the transport vehicle is returned at this station Number of car returns Equal to the number of cars it picks up, that is, the number of cars it picks up at all stations That is, to ensure that the transport vehicle returns all the vehicles it has picked up;
[0094]
[0095] (8) The driving distance of the carrier during dispatching shall not exceed the limit value H max : Ensure that for each transport vehicle, if it picks up the vehicle at station s and returns it at station s' ( but ), then the distance between station s and station s' is H s,s′ Not exceeding the limit value H max :
[0096]
[0097] (9) Ensure that the transport vehicle only picks up the vehicle at one station. We assume that a transport vehicle can only complete one scheduling task in a planning period: For each transport vehicle, ensure that the vehicle pickup indicator variable The sum of the variables at each station is 1, that is, only one vehicle-picking indicator variable is 1, and the rest are 0.
[0098]
[0099] (10) Limit the distance traveled by the transport vehicle when picking up the vehicle, that is, it can only be collected from the nearest station G v Pick up the vehicle: For each transport vehicle, ensure that it is collected at the nearest station G v External station pickup variables The sum is 0, that is, the variable for picking up a vehicle at a station outside the set of adjacent stations is 0:
[0100]
[0101] (11) Ensure that the transport vehicle is returned only at one station: For each transport vehicle, ensure that the return indicator variable The sum at each station is 1, that is, only one return indicator variable is 1, and the rest are 0:
[0102]
[0103] (12) Ensure that vehicle losses and parking space losses are non-negative, and the number of vehicles picked up and replaced by the transport vehicle is non-negative and does not exceed the capacity of the transport vehicle. The indicator variable takes the value of 0 or 1.
[0104]
[0105]
[0106]
[0107] S407. Solve the planning problem and get the variables In the embodiment of the present application, the scheduling problem is a mixed integer linear programming problem, which can be solved by a classical method such as the branch and bound method.
[0108] Finally, it should be noted that the foregoing descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art will readily appreciate that modifications may be made to the methods described in the foregoing embodiments, such as by changing the names of the methods. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.
Claims
1. A shared bicycle scheduling method based on site clustering, characterized by: The following steps are involved: S1. Divide the operation period of the shared bicycle system into multiple planning periods; S2. For the tth planning period, construct a shared bicycle system model; The step S2 comprises: S201. Given system resource information, including: Station set S: contains the capacity of each station in the shared bicycle system. The capacity of a station is the number of parking spaces in the station. In the station set, the capacity of station s is recorded as The number of vehicles at planning period t is denoted as S202. Given user traffic information, including user traffic within the planning period t in K training scenarios from historical data, is the number of users who travel from station s to station s′ within the planning period t in the kth training scenario; S203. Loss is defined as the number of times user needs cannot be met with existing resources, specifically including: Vehicle loss, that is, the loss caused by the user not finding an available vehicle, is the vehicle loss calculation that occurs at station s in the kth training scenario; when no scheduling is performed, it is calculated as Parking space loss, that is, the loss caused by the user not finding a parking space to return the car, is the parking space loss that occurs at station s in the kth training scenario; when no scheduling is performed, it is calculated as The total loss is the sum of vehicle loss and parking space loss; S3. During the tth planning period, cluster the bike-sharing system stations using a clustering algorithm. The clustering process first divides the stations into clusters based on distance, and then uses the average loss as a measure of the scheduling demand within the cluster to adjust the clusters. The step S3 comprises: Step 1: Enter system resources and user traffic information; specify the number of generated clusters as C and the number of abnormal station sets as 1; Step 2: Pre-set the generated C clusters to be empty sets and the abnormal station set O to be empty sets; Step 3: Calculate the average system loss Among them, S +,k is,the set of stations with non-zero loss in the system in training scenario k.,The average loss of this system will be used as the comparison criterion in the clustering process; Step 4: On average, select c stations in the system and set them as the central stations of C clusters; Step 5. Define the distance between station s and any cluster c as the distance between station s and the middle station of cluster c. Based on this, assign all stations in the system to the cluster closest to them. Step 6: For any cluster c, calculate the average loss Among them, P +，k is the set of stations with non-zero cluster loss in scenario k; AL c and the system average loss AL g Compare, if AL c Greater than AL g , then the most edge station in cluster c, that is, the station farthest from its central station in cluster c, is moved into the abnormal station set O in sequence until the average loss of the remaining stations in cluster c is less than AL g ; Step 7: Assign the abnormal stations in the abnormal station set o one by one to the cluster with the smallest average loss. If the average loss of this cluster after receiving the abnormal station exceeds the system average loss AL g , then the abnormal station is moved back to the abnormal station set; Step 8: Assign the abnormal stations returned to the abnormal station set O in step 7 to the cluster with the smallest average loss one by one; Step 9: Output the final C clusters; S4. Perform scheduling planning for shared bicycles in each cluster formed during the t-th planning period. S5. Repeat steps S2 to S4 in different planning periods to complete the scheduling plan of shared bicycles in each planning period.
2. A shared bicycle scheduling method based on site clustering according to claim 1, characterized in that: The step S4 comprises: S401. Treat each cluster as an independent system and perform planning and scheduling within each cluster based on resource distribution and user traffic information. S402. Among the C clusters, the system resource information of the cth cluster includes: Station Collection c : Contains the capacity of each station in the c-th shared bicycle cluster. The capacity of the station is the number of parking spaces in the station. c In the example, the capacity of station s is recorded as The vehicle set V contains the capacity of each vehicle in the c-th shared bicycle cluster. The capacity of the vehicle refers to the maximum number of vehicles it can carry. In the vehicle set, the capacity of the vehicle v is recorded as Vehicle distribution: including represents the number of vehicles at station s in the cth shared bicycle cluster at planning period t; Carrier distribution: including indicator variables The value of is 0 or 1. It means that at the beginning of planning period t, vehicle v is located at station s; Indicates that vehicle v is not located at station s during the planning period t; Station location information is contained in the matrix H, which has S c OK, S c Columns, elements in the matrix H s,s′ represents the distance between two stations s, s′; S403. Given user traffic information, including user traffic in the c-th shared bicycle cluster within the planning period t in K training scenarios from historical data, is the number of users who travel from station s to station s′ in the cth shared bicycle cluster within the planning period t in the kth training scenario; S404. Define the loss in the cth shared bicycle cluster as the number of times that user demand cannot be met with existing resources, specifically including: Vehicle loss, that is, the loss caused by the user not finding an available vehicle, is the vehicle loss calculation that occurs at station s in the c-th shared bicycle cluster in the k-th training scenario; when no scheduling is performed, it is calculated as Parking space loss, that is, the loss caused by the user not finding a parking space to return the car, is the parking space loss at station s in the c-th shared bicycle cluster in the k-th training scenario; when no scheduling is performed, it is calculated as The total loss in the c-th shared bicycle cluster is the sum of vehicle loss and parking space loss in the c-th shared bicycle cluster; S405. Given the relevant variables of the scheduling plan: is the number of vehicles picked up by dispatch vehicle v at station s; is the number of vehicles returned by dispatch vehicle v at station s; indicator variables Represents the dispatch vehicle v picking up the vehicle at station s; indicator variables Represents the dispatch vehicle v returning at station s; S406. Dynamic vehicle scheduling problem in the c-th shared bicycle cluster: This problem is a mixed integer nonlinear programming problem. Based on resource distribution and user traffic information, the goal is to minimize the total loss of the cth cluster in K training scenarios. The scheduling planning problem is as follows: (1) Minimizing the total loss after target scheduling: The goal is to minimize the total loss after scheduling; the total loss is the sum of vehicle losses and parking space losses in all training scenarios and all stations: (2) Calculate the total vehicle loss after dispatch, that is, the difference between the number of vehicles required by users and the number of vehicles available at the station after dispatch: In each training scenario, each station, ensure vehicle loss Not less than the difference between the number of vehicles required by the user and the number of vehicles available at the station after dispatching; in the minimization problem, this is equivalent to calculating the vehicle loss as the difference between the number of vehicles required by the user and the number of vehicles available at the station after dispatching; The number of vehicles required by the user is the number of vehicles going from this station to each station sum; The number of vehicles available at the station after dispatching is the original number of vehicles at this station The sum of the net number of vehicles brought in at this station by each vehicle; the net number of vehicles brought in at this station by a vehicle is the number of vehicles picked up by the vehicle at this station. Subtract the number of returned vehicles (3) Calculate the total parking space loss after scheduling, that is, the difference between the number of parking spaces required by users and the number of available parking spaces at the station after scheduling: In each training scenario and each station, ensure the loss of parking space Not less than the difference between the number of parking spaces required by the user and the number of available parking spaces at the station after dispatching; In the minimization problem, it is equivalent to calculating the parking space loss as the difference between the number of parking spaces required by users and the number of available parking spaces at the station after scheduling; The number of parking spaces required by the user is the number of vehicles heading to this station from each station sum; The number of available parking spaces at the station after dispatching is the original number of vehicles at this station The sum of the net outgoing numbers of each transport vehicle at this station; the net outgoing number of a transport vehicle at this station is the number of vehicles returned at this station. Subtract the number of cars picked up (4) Ensure that the number of vehicles picked up by the transport vehicle does not exceed the transport vehicle capacity: Ensure that at each station, for each transport vehicle, if it picks up a vehicle at this station Number of cars picked up No more than the original number of vehicles at the station or vehicle capacity (5) Ensure that the number of vehicles picked up by transport vehicles does not exceed the number of vehicles available at the pick-up station: Ensure that the total number of vehicles picked up by all transport vehicles at each station, that is, the number of vehicles picked up by each transport vehicle The sum of the number of vehicles available at the station shall not exceed the number of vehicles available at the station (6) Ensure that the number of returned transport vehicles does not exceed the number of available parking spaces at the return station: Ensure that, at each station, the total number of returned transport vehicles, i.e., the number of returned vehicles per transport vehicle, is equal to The sum of the number of available parking spaces at the station shall not exceed the station capacity. and the original number of vehicles Difference: (7) Ensure that the number of vehicles picked up by the transport vehicle is equal to the number of vehicles returned, that is, all the vehicles picked up are returned: Ensure that at each station, for each transport vehicle, if the transport vehicle is returned at this station Number of car returns Equal to the number of cars it picks up, that is, the number of cars it picks up at all stations That is, to ensure that the transport vehicle returns all the vehicles it has picked up; (8) The driving distance of the carrier during dispatching shall not exceed the limit value H max : Ensure that for each transport vehicle, if it picks up the vehicle at station s and returns it at station s' ( but ), then the distance H between station s and station s' s,s 'Does not exceed the limit value H max : (9) Ensure that the transport vehicle only picks up the vehicle at one station. Assume that a transport vehicle can only complete one scheduling task in a planning period: For each transport vehicle, ensure that the vehicle pickup indicator variable The sum of the variables at each station is 1, that is, only one vehicle-picking indicator variable is 1, and the rest are 0. (10) Limit the distance traveled by the transport vehicle when picking up the vehicle, that is, it can only be collected from the nearest station G v Pick up the vehicle: For each transport vehicle, ensure that it is collected at the nearest station G v External station pickup variables The sum is 0: (11) Ensure that the transport vehicle is returned only at one station: For each transport vehicle, ensure that the return indicator variable The sum at each station is 1, that is, only one return indicator variable is 1, and the rest are 0: (12) Ensure that vehicle losses and parking space losses are non-negative, and the number of vehicles picked up and replaced by the transport vehicle is non-negative and does not exceed the capacity of the transport vehicle. The indicator variable takes the value of 0 or 1. S407. Solve the planning problem and get the variables And scheduling is performed based on the solved variables.