Demand response type unmanned bus scheduling method under multi-area view angle

Abstract

The invention discloses a demand response type unmanned bus scheduling method under a multi-region view angle. The method comprises the following steps: dividing an external bus trip space into K sub-regions, and determining the total number of unmanned buses capable of being scheduled in each sub-region; in the initial distribution stage of vehicle dispatching control, the storage capacity of each node is detected one by one from inside to outside in the central area so as to carry out unmanned small bus distribution; in the secondary adjustment stage, the number of unmanned buses at the nodes is judged one by one from outside to inside in the outermost area, and the unmanned buses are preferentially distributed to the inner layer; the condition that the delivery state judgment of the unmanned buses in all the sub-areas meets a balance equation is guaranteed, and rule setting is conducted on secondary adjustment; and according to the Markov process characteristics, evaluating a car dispatching scheme of the unmanned car, so that the operation cost of the unmanned car in each sub-region meets an expected formula. According to the invention, it is ensured that a plurality of sub-regions can meet external travel requirements in real time during linkage operation of the unmanned minibus.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a demand-responsive unmanned minibus dispatching method from a multi-region perspective. Background Art

[0002] Demand Responsive Transit (DRT) has been widely promoted in many cities around the world as an operation mode that can overcome the insufficient responsiveness of traditional public transportation systems. Traditional public transportation is usually set on predetermined routes and fixed schedules, which leads to its own lack of flexibility in operation and difficulty in adapting to the dynamic needs of passengers. Compared with traditional public transportation, demand-responsive buses adopt a more non-intensive operation mode in terms of price, time, route and station. Generally speaking, demand-responsive buses can be divided into two modes according to their flexibility: semi-flexibility mode and full-flexibility mode. Semi-flexible demand-responsive buses refer to those vehicles that do not need to cross fixed stations, and can also adjust routes according to passenger demand between fixed stations. In the bus service mode, the schedule is relatively flexible. For full-flexibility demand-responsive buses, buses do not need to follow any predefined schedules and routes. Bus services are launched to dynamically generate routes to meet the needs of passengers, taking into account their preferences, including boarding and alighting times and locations. Similar to the services provided by city taxis, it can provide services based on passenger needs rather than operating on fixed routes or schedules. The vehicle's schedule and route will also be adjusted according to real-time passenger requirements, providing flexible travel options without being restricted by fixed stations.

[0003] However, no matter what operating mode a demand-responsive bus chooses, the effectiveness of its system depends largely on an efficient vehicle scheduling strategy. In particular, a fully flexible mode relies more on real-time adjustments to passenger demand, requiring complex algorithms to dynamically optimize routes and schedules to ensure minimal waiting time and high service quality while balancing operating costs. In the prior art, although some research has been conducted on the scheduling of driverless buses, there has been no research on vehicle scheduling for minibuses that compete directly with taxi operating modes, and no consideration has been given to how to achieve efficient scheduling technology in multi-level regional nesting. Summary of the invention

[0004] The purpose of the present invention is to provide a demand-responsive driverless minibus scheduling method from a multi-region perspective, ensuring that multiple sub-regions can instantly meet external travel needs when operating driverless minibuses in a coordinated manner.

[0005] The technical solution adopted by the present invention is:

[0006] A demand-responsive driverless minibus dispatching method from a multi-region perspective includes the following steps:

[0007] Step 1: Introduce a multi-level inventory system management model, set the vehicles required for external bus travel demand forecasting to follow the Poisson distribution, and divide the external bus travel into sub-areas, and then determine the total number of driverless minibuses that can be dispatched in each layer of sub-area; Layer sub-regions are recorded as nodes , characterizing the control from the first Layer to The management node for the number of dispatches of all driverless minibuses in the layer sub-area;

[0008] Furthermore, the specific steps for determining the total number of driverless minibuses that can be dispatched in each layer of sub-area are as follows:

[0009] Step 1-1, calculate and obtain any node The total number of driverless minibuses that can be dispatched is calculated as follows:

[0010] , ;

[0011] in, Representation Node The number of driverless minibuses available Indicates the number of spare driverless minibuses reserved in each sub-area to deal with emergencies;

[0012] Steps 1-2, summarize to get any time The total number of driverless minibuses in each sub-area is expressed as .

[0013] Specifically, Representation Node If there are unmanned minibuses available, Indicates the number of available driverless minibuses in the entire operating system. In actual operation, each sub-region will reserve a portion of capacity to deal with emergencies. This number of driverless minibuses can be included in the net inventory of system capacity as public information. The total number of driverless minibuses in each sub-region can be expressed as , where the total number of driverless minibuses in any k-th sub-area is , .

[0014] Step 2: According to the network association in the management mode of the multi-level inventory system, the vehicle dispatch control process of each node is divided into the initial allocation stage of effective weight and the secondary adjustment stage; in the initial allocation stage, the storage capacity of each node is detected one by one from the central area "from the inside to the outside" to make a preliminary allocation of driverless minibuses; and then in the secondary adjustment stage, the number of driverless minibuses in each sub-area is determined by threshold value one by one from the outermost area "from the outside to the inside", and the driverless minibuses are allocated to the inner layer first;

[0015] Furthermore, the specific steps of the preliminary allocation stage are:

[0016] Step 2-1-1, set the vehicle dispatching decision trigger threshold for each sub-area, and the threshold relationship between the sub-areas satisfies:

[0017] (1)

[0018] in, For the outermost The vehicle dispatch determination trigger threshold for each sub-area;

[0019] Step 2-1-2, dynamic tracking at any time The storage capacity of driverless minibuses in the sub-region nodes , and in Determine the storage capacity of driverless minibuses in the corresponding sub-area from the inside to the outside Is it established? If yes, start the vehicle dispatching work and execute step 2-1-3; otherwise, make the next sub-area judgment and execute step 2-1-2;

[0020] Step 2-1-3, allocating driverless minibuses based on the fast routing decision model of the Open Jackson network;

[0021] The following relationship is satisfied between the nodes in the fast routing decision model:

[0022] (2)

[0023] That is, during allocation, the number of effective driverless minibuses at each node must meet the following requirements:

[0024] (3).

[0025] In the formula, The constraint threshold of the number of available driverless minibuses in sub-area k is determined by the dispatch threshold of sub-area 1. The difference between the dispatching decision trigger threshold of sub-area k+1 is determined, and 0 means that the driverless minibus is dispatched from the outermost sub-area.

[0026] Specifically, in the initial allocation stage, the present invention sets each sub-area to respond to each public transportation travel demand according to the judgment principle of "first come, first served". The trigger threshold for dispatching vehicles is , then the threshold relationship between each sub-region satisfies:

[0027] (1)

[0028] The present invention dynamically tracks any time The storage capacity of driverless minibuses in the sub-region nodes , and in and The dispatching work is started when the demand center area is reached. In order to give priority to meeting the needs of the central area, the allocation rule of the driverless minibus follows the order of "from inside to outside" to detect the storage capacity of each node one by one. Based on this, the present invention deconstructs the allocation process of the driverless minibus into a fast routing decision model based on the Open Jackson network, and sets the following relationship between each node:

[0029] (2)

[0030] Under the role of routing selection, the allocation at this stage is equivalent to ensuring that the number of effective driverless minibuses at each node meets:

[0031] (3)

[0032] In the formula, The constraint threshold of the number of available driverless minibuses in sub-area k is determined by the dispatch threshold of sub-area 1. The difference between the dispatching decision trigger threshold of sub-area k+1 is determined, and 0 means that the driverless minibus is dispatched from the outermost sub-area.

[0033] According to formula (1), the threshold satisfies the relationship Formula (3) shows that for any node They all have corresponding effective thresholds when determining vehicle dispatch. . As external parking requirements change, Layer sub-region nodes greater than the threshold , it means that these sub-areas have stored enough minibuses, and in the There is still a lack of unmanned public buses at the layer nodes.

[0034] Furthermore, the method of the secondary adjustment stage is:

[0035] Step 2-2-1, from the outermost The storage capacity of driverless minibuses in each sub-area is analyzed in turn;

[0036] Step 2-2-2, determine whether the storage capacity of driverless minibuses in the current sub-area is less than the dispatch determination trigger threshold of the corresponding sub-area; if so, set the node The extracted vehicles are assigned to the current sub-area; otherwise, execute step 2-2-3;

[0037] Step 2-2-3, determine whether all sub-areas from the outermost to the sub-area i where the vehicle is separated have been judged to be complete; if so, end the allocation adjustment of the sub-area i where the vehicle is currently separated and wait for the next adjustment; otherwise, execute step 2-2-2.

[0038] Specifically, in the secondary adjustment stage, the present invention will follow the order of "from outside to inside" to determine the number of driverless minibuses at the node one by one. A driverless minibus needs to be extracted from the node To Node If a large number of vehicles have been assigned between nodes, the extracted driverless minibus will skip these intermediate nodes and be directly assigned to node , to meet the rule of prioritizing inner layer allocation. Generally speaking, from the Jackson network node Extract a minibus and assign it to the node , its purpose is to protect the nodes after routing The number of driverless minibuses is greater than the minimum preset value:

[0039] (4)

[0040] in, Furthermore, in order to determine the downstream node to which the vehicle is to be dispatched, the present invention starts parsing from node K. ,like , then the extracted driverless minibus is assigned to node K, otherwise the node K-1 is parsed , until node i+1. Similar to the initial allocation stage, when , the model will be assigned to nodes first The function of formula (4) is similar to that of formula (3), that is, it is expected that the routing strategy of the driverless minibus in the secondary adjustment stage can satisfy , , and the vehicle from the node Deploy to nodes During the process, the number of driverless minibuses in the middle layer sub-area will remain constrained.

[0041] Step 3: further set rules for the secondary adjustment to ensure that the state determination of the driverless minibus delivery in each sub-area satisfies the equilibrium equation; the equilibrium equation specifically refers to the state of the number of driverless minibuses in the sub-area The probability distribution of the case The following formula constraints should be satisfied:

[0042] (11)

[0043] (12)

[0044] (13)

[0045] (14)

[0046] Wherein, formula (11) means that in the sub-area The situation when there is an unmanned minibus responding to passenger demand on the sub-area; Formula (12) represents the situation when there is at least one unmanned minibus responding to passenger demand on sub-area 1; Formula (13) refers to the situation when there is at least one unmanned minibus responding to passenger demand on sub-area and When there is an unmanned minibus responding to passenger demand; Formula (14) refers to the situation in which There is an unmanned minibus dispatched to the sub-area To respond to passenger needs; represents the dispatch rate of driverless minibuses; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; represents the consumption rate of driverless minibuses used to serve the public transportation travel demand in sub-area 1; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; , are all unit vectors, representing the number of dispatched driverless minibuses with one in sub-areas 1, m, i, and K, respectively; is the identity matrix, which refers to the case when there is more than one driverless minibus in sub-area 1; This refers to the situation where a driverless minibus in area 1 is transferred to other sub-areas; , Refers to sub-areas in different situations and The number of valid driverless minibuses; the specific conditions are defined as follows:

[0047] (15)

[0048] (16);

[0049] Respectively represent the corresponding area 1, , The number of valid driverless minibuses is as follows:

[0050] (1) When a driverless minibus in sub-area 1 is decided to be sent to a downstream node;

[0051] if for If both are true, then ;

[0052] Otherwise, for any , ;

[0053] (2) When sub-region , When a driverless minibus is decided to be sent to a downstream node;

[0054] if ,but ;

[0055] when hour, ;

[0056] (3) When a driverless minibus in sub-area K-1 is decided to be sent to a downstream node;

[0057] like ,but ;

[0058] otherwise .

[0059] Among them, the parameters Characterize the constraint threshold of routing selection in the K-layer sub-area, let Indicates sub-area To Sub-Area The set between , then the number of valid driverless minibuses can be expressed as:

[0060] (5)

[0061] In Indicates the number of valid driverless minibuses in each sub-area. Definition is the state set where the number of driverless minibuses in the sub-area meets the threshold constraint, and its formula is:

[0062] (6)

[0063] make Indicates the secondary adjustment stage from the node Assign to Node The routing strategy of , at this time, the number of vehicles in the sub-area changes from Evolved into ;otherwise .

[0064] Furthermore, in order to solve the product form solution of the steady-state distribution, when setting the constraint threshold of the routing selection strategy of each sub-area, the number of driverless minibuses should be further satisfied. The state of will be given by the following set (7)

[0065] In the conditions Under the action of , the steady-state solution of the Markov process can be obtained by the following calculation

[0066] (8)

[0067] in is a normalization constant, which can be taken as 1 in general. The generated state vector will satisfy the equilibrium equations, and both the input rate and the output rate are state independent.

[0068] Step 4: Evaluate the dispatching scheme of driverless minibuses according to the characteristics of the Markov process. The expected operating cost of driverless minibuses in each sub-region satisfies the following formula:

[0069] (10)

[0070] in, Characterize the constraint threshold of routing selection in the K-layer sub-area; represents the number of driverless minibuses that are effective in each sub-area under this constraint; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; In Indicates the status of the number of valid driverless minibuses in each sub-area; represents the vacancy cost rate of the driverless minibus, represents the oversaturation service cost rate of driverless minibuses; The cost of dispatching bicycles to each sub-area ; is the identity matrix, representing the situation where an unmanned minibus is dispatched from the dispatch center to sub-area i; (•) + It means comparing the calculation result in the brackets with 0 and taking only 0 or positive values.

[0071] Specifically, in order to ensure the stability of Open Jackson, the present invention sets the dispatch rate of the driverless minibus The vehicle consumption rate is less than that required by external public transport travel demand Therefore, the state evolution of the number of driverless minibuses in each sub-region can be regarded as a Markov process. , and let

[0072] (9)

[0073] in, represents the probability distribution of a Markov process.

[0074] Since for any time , both have , it can be seen that when the vacancy cost rate of the driverless minibus is set and oversaturated service cost rate , then the bicycle dispatching cost in each sub-area is The expected operating cost of the driverless minibus can be calculated using the following formula:

[0075] (10).

[0076] in, represents the vacancy cost rate of the driverless minibus, represents the oversaturation service cost rate of driverless minibuses; Dispatching costs for each sub-area ; Indicates the vehicle dispatch decision trigger threshold for the first sub-area in the inner layer; represents the original number of driverless minibuses; Indicates sub-area The identity matrix when adding a driverless minibus; (•) + It means comparing the calculation result in the brackets with 0 and taking only 0 or positive values.

[0077] The present invention adopts the above technical solution, introduces the multi-level inventory management mode into the dispatch decision process of the driverless minibus, and constructs a fast routing decision model based on the Open Jackson network by deriving the steady-state distribution solution of vehicle allocation in the linkage operation mechanism of the multi-level sub-region driverless minibus, and obtains the optimal allocation plan of the driverless minibus in multiple regions. While ensuring the operating income, the present invention ensures that multiple sub-regions can immediately meet the external travel needs when the driverless minibus is operated in linkage. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments;

[0079] Figure 1 This is a schematic diagram of the evolution characteristics of the monthly smart card swiping situation of the present invention;

[0080] Figure 2 A schematic diagram of the dispatch results of driverless minibuses in each sub-region;

[0081] Figure 3 A schematic diagram comparing the occupancy of driverless minibuses and the dispatch results in each sub-area;

[0082] Figure 4 This is a heat map of demand-responsive bus travel distribution in each sub-region. DETAILED DESCRIPTION

[0083] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0084] The present invention discloses a demand-responsive driverless minibus dispatching method from a multi-region perspective, comprising the following steps:

[0085] Step 1: Introduce a multi-level inventory system management model, set the vehicles required for external bus travel demand forecasting to follow the Poisson distribution, and divide the external bus travel into sub-areas; at the same time, determine the total number of driverless minibuses that can be dispatched in each layer of sub-area; any Layer sub-regions are recorded as nodes , characterizing the control from the first Layer to The management node for the number of dispatches of all driverless minibuses in the layer sub-area;

[0086] Furthermore, the specific steps for determining the total number of driverless minibuses that can be dispatched in each layer of sub-area are as follows:

[0087] Step 1-1, calculate and obtain any node The total number of driverless minibuses that can be dispatched is calculated as follows:

[0088] , ;

[0089] in, Representation Node The number of driverless minibuses available Indicates the number of spare driverless minibuses reserved in each sub-area to deal with emergencies;

[0090] Steps 1-2, summarize to get any time The total number of driverless minibuses in each sub-area is expressed as .

[0091] Specifically, Representation Node If there are unmanned minibuses available, Indicates the number of available driverless minibuses in the entire operating system. In actual operation, each sub-region will reserve a portion of capacity to deal with emergencies. This number of driverless minibuses can be included in the net inventory of system capacity as public information. The total number of driverless minibuses in each sub-region can be expressed as , where the total number of driverless minibuses in any k-th sub-area is , .

[0092] Step 2: According to the network correlation in the management mode of the multi-level inventory system, the vehicle dispatch control process of each node is divided into a preliminary allocation stage of effective rights and a secondary adjustment stage; in the preliminary allocation stage, the storage capacity of each node is detected one by one from the central area "from the inside to the outside" to allocate the unmanned minibuses; and then in the secondary adjustment stage, the number of unmanned minibuses in the node is determined one by one from the outermost area "from the outside to the inside", and the unmanned minibuses are allocated to the inner layer first; specifically, the inner layer means the closer to the unmanned minibus dispatch center; the outer layer means the farther away from the unmanned minibus dispatch center.

[0093] Furthermore, the specific steps of the preliminary allocation stage are:

[0094] Step 2-1-1, set the vehicle dispatching decision trigger threshold for each sub-area, and the threshold relationship between the sub-areas satisfies:

[0095] (1)

[0096] in, For the outermost The vehicle dispatch determination trigger threshold for each sub-area;

[0097] Step 2-1-2, dynamic tracking at any time The storage capacity of driverless minibuses in the sub-region nodes , and in Determine the storage capacity of driverless minibuses in the corresponding sub-area from the inside to the outside Is it established? If yes, start the vehicle dispatching work and execute step 2-1-3; otherwise, make the next sub-area judgment and execute step 2-1-2;

[0098] Step 2-1-3, allocating driverless minibuses based on the fast routing decision model of the Open Jackson network;

[0099] The following relationship is satisfied between the nodes in the fast routing decision model:

[0100] (2)

[0101] That is, during allocation, the number of effective driverless minibuses at each node must meet the following requirements:

[0102] (3).

[0103] In the formula, The constraint threshold of the number of available driverless minibuses in sub-area k is determined by the dispatch threshold of sub-area 1. The difference between the dispatching decision trigger threshold of sub-area k+1 is determined, and 0 means that the driverless minibus is dispatched from the outermost sub-area.

[0104] Specifically, in the initial allocation stage, the present invention sets each sub-area to respond to each public transportation travel demand according to the judgment principle of "first come, first served". The trigger threshold for dispatching vehicles is , then the threshold relationship between each sub-region satisfies:

[0105] (1)

[0106] The present invention dynamically tracks any time The storage capacity of driverless minibuses in the sub-region nodes , and in and The dispatching work is started when the demand center area is reached. In order to give priority to meeting the needs of the central area, the allocation rule of the driverless minibus follows the order of "from inside to outside" to detect the storage capacity of each node one by one. Based on this, the present invention deconstructs the allocation process of the driverless minibus into a fast routing decision model based on the Open Jackson network, and sets the following relationship between each node:

[0107] (2)

[0108] Under the role of routing selection, the allocation at this stage is equivalent to ensuring that the number of effective driverless minibuses at each node meets:

[0109] (3)

[0110] According to formula (1), the threshold satisfies the relationship Formula (3) shows that for any node They all have corresponding effective thresholds when determining vehicle dispatch. . As external parking requirements change, Layer sub-region nodes greater than the threshold , it means that these sub-areas have stored enough minibuses, and in the There is still a lack of unmanned public buses at the layer nodes.

[0111] Furthermore, the method of the secondary adjustment stage is:

[0112] Step 2-2-1, from the outermost The storage capacity of driverless minibuses in each sub-area is analyzed in turn;

[0113] Step 2-2-2, whether the storage capacity of driverless minibuses in the current sub-region is less than the dispatching judgment trigger threshold of the corresponding sub-region; if so, set the node The extracted vehicles are assigned to the current sub-area; otherwise, execute step 2-2-3;

[0114] Step 2-2-3, determine whether all sub-areas from the outermost to the sub-area i where the vehicle is separated have been judged to be complete; if so, end the allocation adjustment of the sub-area i where the vehicle is currently separated and wait for the next adjustment; otherwise, execute step 2-2-2.

[0115] Specifically, in the secondary adjustment stage, the present invention will follow the order of "from outside to inside" to determine the number of driverless minibuses at the node one by one. A driverless minibus needs to be extracted from the node To Node If a large number of vehicles have been assigned between nodes, the extracted driverless minibus will skip these intermediate nodes and be directly assigned to node , to meet the rule of prioritizing inner layer allocation. Generally speaking, from the Jackson network node Extract a minibus and assign it to the node , its purpose is to protect the nodes after routing The number of driverless minibuses is greater than the minimum preset value:

[0116] (4)

[0117] in, Furthermore, in order to determine the downstream node to which the vehicle is to be dispatched, the present invention starts parsing from node K. ,like , then the extracted driverless minibus is assigned to node K, otherwise the node K-1 is parsed , until node i+1. Similar to the initial allocation stage, when , the model will be assigned to nodes first The function of formula (4) is similar to that of formula (3), that is, it is expected that the routing strategy of the driverless minibus in the secondary adjustment stage can satisfy , , and the vehicle from the node Deploy to nodes During the process, the number of driverless minibuses in the middle layer sub-area will remain constrained.

[0118] Step 3: further set rules for the secondary adjustment to ensure that the state determination of the driverless minibus delivery in each sub-area satisfies the equilibrium equation; the equilibrium equation specifically refers to the state of the number of driverless minibuses in the sub-area The probability distribution of the case The following formula constraints should be satisfied:

[0119] (11)

[0120] (12)

[0121] (13)

[0122] (14)

[0123] Wherein, formula (11) means that in the sub-area The situation when there is an unmanned minibus responding to passenger demand on the sub-area; Formula (12) represents the situation when there is at least one unmanned minibus responding to passenger demand on sub-area 1; Formula (13) refers to the situation when there is at least one unmanned minibus responding to passenger demand on sub-area and When there is an unmanned minibus responding to passenger demand; Formula (14) refers to the situation in which There is an unmanned minibus dispatched to the sub-area To respond to passenger needs; represents the dispatch rate of driverless minibuses; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; represents the consumption rate of driverless minibuses used to serve the public transportation travel demand in sub-area 1; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; , are all unit vectors, representing the number of dispatched driverless minibuses with one in sub-areas 1, m, i, and K, respectively; is the identity matrix, which refers to the case when there is more than one driverless minibus in sub-area 1; This refers to the situation where a driverless minibus in area 1 is transferred to other sub-areas; , Refers to sub-areas in different situations and The number of valid driverless minibuses; the specific conditions are defined as follows:

[0124] (15)

[0125] (16);

[0126] Respectively represent the corresponding area 1, , The number of valid driverless minibuses is as follows:

[0127] (1) When a driverless minibus in sub-area 1 is decided to be sent to a downstream node;

[0128] if for If both are true, then ;

[0129] Otherwise, for any , ;

[0130] (2) When sub-region , When a driverless minibus is decided to be sent to a downstream node;

[0131] if ,but ;

[0132] when hour, ;

[0133] (3) When a driverless minibus in sub-area K-1 is decided to be sent to a downstream node;

[0134] like ,but ;

[0135] otherwise .

[0136] in, represents the constraint threshold of the number of driverless minibuses in sub-area k; represents the vehicle dispatch determination trigger threshold of sub-area m; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; In Indicates the status of the number of valid driverless minibuses in each sub-area; =1 means that an unmanned minibus is adjusted from node 1 to node K; express; express; It means that a driverless minibus is adjusted from node 1 to node i; It means that the driverless minibus has not been dispatched and still stays at node m; Indicates that there is an unmanned minibus from the node Adjust to node i; Indicates that there is an unmanned minibus from the node Adjust to Node .

[0137] Specifically, the parameters Characterize the constraint threshold of routing selection in the K-layer sub-area, let Indicates sub-area To Sub-Area The set between , then the number of valid driverless minibuses can be expressed as:

[0138] (5)

[0139] In Indicates the number of valid driverless minibuses in each sub-area. Definition is the state set where the number of driverless minibuses in the sub-area meets the threshold constraint, and its formula is:

[0140] (6)

[0141] make Indicates the secondary adjustment stage from the node Assign to Node The routing strategy of , at this time, the number of vehicles in the sub-area changes from Evolved into ;otherwise .

[0142] Furthermore, in order to solve the product form solution of the steady-state distribution, when setting the constraint threshold of the routing selection strategy of each sub-area, the number of driverless minibuses should be further satisfied. The state of will be given by the following set (7)

[0143] In the conditions Under the action of , the steady-state solution of the Markov process can be obtained by the following calculation

[0144] (8)

[0145] in is a normalization constant, which can be taken as 1 in general. The generated state vector will satisfy the equilibrium equations, and both the input rate and the output rate are state independent.

[0146] Step 4: Evaluate the dispatching scheme of driverless minibuses according to the characteristics of the Markov process. The expected operating cost of driverless minibuses in each sub-region satisfies the following formula:

[0147] (10)

[0148] in, Characterize the constraint threshold of routing selection in the K-layer sub-area; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; In Indicates the status of the number of valid driverless minibuses in each sub-area; represents the vacancy cost rate of the driverless minibus, represents the oversaturation service cost rate of driverless minibuses; The cost of dispatching bicycles to each sub-area . represents the original number of driverless minibuses; Indicates sub-area The identity matrix when adding a driverless minibus; (•) + It means comparing the calculation result in the brackets with 0 and taking only 0 or positive values.

[0149] Specifically, in order to ensure the stability of Open Jackson, the present invention sets the dispatch rate of the driverless minibus The vehicle consumption rate is less than that required by external public transport travel demand Therefore, the state evolution of the number of driverless minibuses in each sub-region can be regarded as a Markov process. , and let

[0150] (9)

[0151] in, represents the probability distribution of a Markov process.

[0152] Since for any time , both have , it can be seen that when the vacancy cost rate of the driverless minibus is set and oversaturated service cost rate , then the bicycle dispatching cost in each sub-area is The expected operating cost of the driverless minibus can be calculated using the following formula:

[0153] (10).

[0154] The method of the present invention is further described in detail below with reference to examples.

[0155] 1. Divide the study area into multiple sub-areas and conduct a survey on the dispatch of driverless minibuses.

[0156] The present invention applies the automatic vehicle location (AVL) data of the Nanjing public transportation system to simulate the origin-destination (OD) point pairs of demand-responsive bus travel. Nanjing is the capital city of Jiangsu Province, China. The study area of ​​the present case is Qixia District, Nanjing, which is located in the north-central part of Nanjing and is connected to another city, Zhenjiang. Qixia District has an area of ​​395.44 square kilometers and a population of 1.0127 million in 2023 (Nanjing Municipal Bureau of Statistics 2023). It has built a modern waterfront urban area with science and technology education, shipping logistics, scenic spots, and mechanical manufacturing as its main functions. It should be pointed out that despite the establishment of more than 1,000 bus stations and 5 subway lines, demand-responsive bus services also began to operate on a pilot basis in the area in June 2024. Citizens only need to place an order through a smartphone application, and the dynamic bus will arrive at the designated location in the shortest time.

[0157] The case of the present invention selects bus AVL data and smart card data in May 2021 as information sources to improve the accuracy of OD estimation. AVL data provides bus service information, including but not limited to detailed information such as bus station name, vehicle ID, route, arrival and departure time. In order to ensure quality and availability, this case implements a data cleaning process to exclude duplicate transaction records and records lacking smart card ID or boarding information. The method of calculating the passenger's alighting point (i.e., destination) assumes that the passenger will choose the nearest station as the alighting point during the transfer process, and corrects the proximity in time and space. Specifically, when a passenger boards the bus at station B1 of a certain route and then boards the bus again at station B2 of another route, the algorithm will search for station A1, which is closest to B2 and meets the transfer conditions, as the passenger's first alighting station. The same mechanism continues to be applied to estimate the alighting station for future trips.

[0158] 2. Record data clearly, process results, and set model parameter information.

[0159] The original AVL data of the case in the present invention was collected in May 2021. After data processing, 2,721,760 valid smart card data remained. Figure 1 The distribution of travel demand for different numbers of buses is shown. In general, the public transportation demand in the study area is significantly higher on weekdays than on non-working days. During the study, this case only selected the first 15,000 rows of data each day to simulate the demand-responsive bus travel in the area. The time recorded in the data is generally from 4:30 to 9:00 am every day. As mentioned above, if the method of the present invention sets too many partition levels, it is easy to lead to high complexity of the model and increase the time consumption of the allocation of driverless minibuses when making preliminary judgments and secondary adjustment decisions. In this case of the present invention, the study area is divided into 5 partitions, and the constraint threshold of each partition is set according to 10% of the number of parking spaces for minibuses in each sub-area. The specific model parameter settings are shown in Table 1.

[0160] Table 1 Basic parameter settings of driverless minibus dispatch model in each sub-region

[0161]

[0162] 3. Summarize various model parameters and realize dynamic judgment of driverless minibus dispatch.

[0163] The model was solved by programming in Python 3.10. When performing sensitivity analysis on a certain parameter, other parameters of the model were temporarily fixed. For example, when analyzing the influencing factors of the steady-state distribution solution of driverless minibuses in each sub-region, although the model parameters The number of driverless minibuses stored at each node will vary depending on the travel demand of each sub-region. , Vehicle dispatch determination trigger threshold and the constraint threshold of driverless minibuses Remains unchanged.

[0164] Figure 2 The dispatch results of driverless minibuses in each sub-area. It can be seen that, on the one hand, the dispatch strategy proposed in this case can save a large number of dispatched vehicles even under the one-to-one vehicle service requirement, thereby saving operating expenses. For some sub-areas, especially those in the inner circle, this is an overflow vehicle dispatch strategy. In response to 15,000 travel demands every day, the number of vehicles dispatched by the method of the present invention is only about 3,800 to 7,700. Among them, there are also obvious differences in the number of vehicles dispatched to each sub-area. The number of driverless minibuses dispatched to sub-areas 4 and 5 (that is, the areas farthest from the city center of Nanjing and with the lowest priority of control nodes) is the largest, and the number of vehicles changes the most. The second is dispatched to sub-areas 1 and 2, which also dispatched a certain number of vehicles with a considerable fluctuation range. The number of driverless minibuses dispatched to sub-area 3 is the least. On the other hand, in terms of the total number of vehicle dispatches, compared with Figure 1 By comparing with the card swiping situation throughout the day, it can be seen that the driverless minibuses dispatched by this strategy are generally consistent with the fluctuations in the total card swipe volume throughout the day. This is not only because the method of the present invention is a vehicle dispatching strategy generated based on demand, and can dynamically compensate for in-service vehicles according to the consumption of each sub-area. At the same time, because the first 15,000 data of each day tested can cover the characteristics of morning rush hour commuting, the evolution trend of the number has a fluctuation characteristic consistent with the evolution trend of the whole day. In order to further explore the efficiency of vehicle dispatching of the method of the present invention, this case compares the results of the number of occupied vehicles and the number of dispatched vehicles in each sub-area in the experimental data, as shown below. Figure 3 shown.

[0165] from Figure 3It can be seen that without changing any influencing parameters, the number of vehicles allocated by the method of the present invention is not much different from the demand number, and only the innermost layer allocates too many driverless minibuses. For the sub-areas of other levels, such as sub-areas 1, 2 and 3, the method of the present invention has shown a significant "saving" function, that is, only a portion of the total external bus travel demand needs to be dispatched to meet the local individual bus travel needs of the sub-area through recycling. First, the fluctuation relationship between the number of driverless minibuses dispatched in each sub-area and the total bus travel demand may not be as obvious as expected. By comparing the evolution of the ratio with the solid line and the dotted line, it can be seen that the number of dispatched vehicles in sub-areas 2, 3 and 4 has a relatively consistent change and fluctuation with the number of travel demands, while the difference in this change and fluctuation from sub-area 1 to sub-area 5 is more obvious. From the perspective of the overall number of services, the dispatched vehicles in sub-area 1 can cover the largest service demand. It only needs to dispatch driverless minibuses that account for 10% to 20% of the total demand to meet the local travel needs. Sub-area 3 is also very efficient. This situation means that the vehicles dispatched from the control center will flow back to sub-areas 1 and 3 after completing the service, so that the driverless minibuses in each sub-area are maintained within the threshold range required in the calculation process.

[0166] To further compare the delivery efficiency, the case of the present invention is based on Figure 4 Characterizes the distribution of sub-areas where driverless minibuses stay after completing their respective services, where the X-axis represents the date, and each row of the Y-axis represents the proportion of vehicles departing from a sub-area and arriving at another sub-area. For example, "S1-S1" refers to the proportion of driverless minibuses departing from sub-area 1 that stay in sub-area 1 after completing their services, while "S1-S2" refers to the proportion of vehicles departing from sub-area 1 that eventually stay in sub-area 2, and so on until "S5-S5". It can be seen that in the case of the present invention, there are obvious similarities and differences in the evolution of public transportation travel demand in sub-areas, which leads to differences in the number of dispatched vehicles. The similarity is that the proportion of vehicles serving each sub-area that eventually stay in their own sub-area to continue serving is the highest, and the farther away from the initial location, the lower the proportion of people's travel. The difference lies in the number of vehicles that diverge from each sub-area to serve that finally stay in each sub-area, and generally confirms the previous discussion, that is, sub-area 1 and sub-area 3 are the places that absorb the most vehicles from other sub-areas, so that they remain like this. Figure 3 The relatively high vehicle utilization rate of driverless minibuses is shown.

[0167] The present invention adopts the above technical solution, introduces the multi-level inventory management mode into the dispatch decision process of the driverless minibus, and constructs a fast routing decision model based on the Open Jackson network by deriving the steady-state distribution solution of vehicle allocation in the linkage operation mechanism of the multi-level sub-region driverless minibus, and obtains the optimal allocation plan of the driverless minibus in multiple regions. While ensuring the operating income, the present invention ensures that multiple sub-regions can immediately meet the external travel needs when the driverless minibus is operated in linkage.

[0168] Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. In the absence of conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The components of the embodiments of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the detailed description of the embodiments of the present application is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians of the art without making creative work are within the scope of protection of the present application.

Claims

1. A demand-responsive driverless minibus dispatching method from a multi-region perspective, characterized by: It includes the following steps: Step 1: Introduce a multi-level inventory system management model, set the vehicles required for external bus travel demand forecasting to follow the Poisson distribution, and divide the external bus travel into sub-areas, and then determine the total number of driverless minibuses that can be dispatched in each layer of sub-area; Layer sub-regions are recorded as nodes , characterizing the control from the first Layer to The management node for the number of dispatches of all driverless minibuses in the layer sub-area; Step 2: According to the network association in the management mode of the multi-level inventory system, the vehicle dispatch control process of each node is divided into the initial allocation stage of effective weight and the secondary adjustment stage; in the initial allocation stage, the storage volume of each node is detected one by one from the central area "from the inside to the outside" to make a preliminary allocation of driverless minibuses; and then in the secondary adjustment stage, the number of driverless minibuses in each sub-area is determined by threshold value one by one from the outermost area "from the outside to the inside", and the driverless minibuses are allocated to the inner layer first; Step 3: further set rules for the secondary adjustment to ensure that the state determination of the driverless minibus delivery in each sub-area satisfies the equilibrium equation; the equilibrium equation specifically refers to the state of the number of driverless minibuses in the sub-area The probability distribution of the case The following formula constraints should be satisfied: (11) (12) (13) (14) Wherein, formula (11) means that in the sub-area The situation when there is an unmanned minibus responding to passenger demand on the sub-area; Formula (12) represents the situation when there is at least one unmanned minibus responding to passenger demand on sub-area 1; Formula (13) refers to the situation when there is at least one unmanned minibus responding to passenger demand on sub-area and When there is an unmanned minibus responding to passenger demand; Formula (14) refers to the situation in which There is an unmanned minibus dispatched to the sub-area To respond to passenger needs; represents the dispatch rate of driverless minibuses; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; represents the consumption rate of driverless minibuses used to serve the public transportation travel demand in sub-area 1; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transport demand; Respectively represent the corresponding sub-area 1, , The number of driverless minibuses in effect; , are all unit vectors, representing the number of dispatched driverless minibuses with one in sub-areas 1, m, i, and K, respectively; is the identity matrix, which refers to the case when there is more than one driverless minibus in sub-area 1; This refers to the situation where a driverless minibus in area 1 is transferred to other sub-areas; , Refers to sub-areas in different situations and The number of effective driverless minibuses; Step 4: Evaluate the dispatching scheme of driverless minibuses based on the characteristics of the Markov process. The expected operating cost of driverless minibuses in each sub-region satisfies the following formula: (10) in, represents the vacancy cost rate of the driverless minibus, represents the oversaturation service cost rate of driverless minibuses; The cost of dispatching bicycles to each sub-area ; Indicates the vehicle dispatch decision trigger threshold for the first sub-area in the inner layer; represents the original number of driverless minibuses; Indicates sub-area The identity matrix when adding a driverless minibus; (•) + It means comparing the calculation result in the brackets with 0 and taking only 0 or positive values.

2. The demand-responsive driverless minibus dispatching method from a multi-region perspective according to claim 1 is characterized by: The specific steps for determining the total number of driverless minibuses that can be dispatched in each layer of sub-area in step 1 are as follows: Step 1-1, calculate and obtain any node The total number of driverless minibuses that can be dispatched is calculated as follows: ， ； in, Representation Node The number of driverless minibuses available Indicates the number of spare driverless minibuses reserved in each sub-area to deal with emergencies; Steps 1-2, summarize to get any time The total number of driverless minibuses in each sub-area is expressed as .

3. The demand-responsive driverless minibus dispatching method from a multi-region perspective according to claim 1 is characterized by: The specific steps of the preliminary allocation stage in step 2 are: Step 2-1-1, set the vehicle dispatching decision trigger threshold for each sub-area, and the threshold relationship between the sub-areas satisfies: (1) in, Indicates the trigger threshold for dispatching vehicles in the first sub-area of ​​the inner layer; and so on. For the outermost The vehicle dispatch determination trigger threshold for each sub-area; Step 2-1-2, dynamic tracking at any time The storage capacity of driverless minibuses in the sub-region nodes , and in Determine the storage capacity of driverless minibuses in the corresponding sub-area from the inside to the outside Is it established? If yes, start the vehicle dispatching work and execute step 2-1-3; otherwise, make the next sub-area judgment and execute step 2-1-2; Step 2-1-3, allocating driverless minibuses based on the fast routing decision model of the Open Jackson network; The following relationship is satisfied between the nodes in the fast routing decision model: (2) That is, during allocation, the number of effective driverless minibuses at each node must meet the following requirements: (3) In the formula, Indicates sub-area The constraint threshold of the number of available driverless minibuses is determined by the dispatching threshold of sub-area 1. and subregions The difference between the dispatching judgment trigger thresholds is determined by , where 0 means dispatching the driverless minibus from the outermost sub-area.

4. The demand-responsive driverless minibus dispatching method from a multi-region perspective according to claim 1 is characterized by: The method for the secondary adjustment phase in step 2 is: Step 2-2-1, from the outermost The storage capacity of driverless minibuses in each sub-area is analyzed in turn; Step 2-2-2, determine whether the storage capacity of driverless minibuses in the current sub-area is less than the vehicle dispatch determination trigger threshold of the corresponding sub-area; if so, set the node The extracted vehicles are assigned to the current sub-area; Otherwise, execute step 2-2-3; Step 2-2-3, determine the sub-area from the outermost area to the separated vehicle Whether all sub-regions between have been judged to be completed; if so, the sub-region of the current vehicle is terminated The allocation is adjusted and wait for the next adjustment; otherwise, execute step 2-2-2.

5. The demand-responsive driverless minibus dispatching method from a multi-region perspective according to claim 1 is characterized by: The specific steps of step 3 are as follows: Step 3-1, let Indicates sub-area To Sub-Area The sub-region To Sub-Area The number of valid driverless minibuses between is expressed as: (5) Step 3-2, Definition is the state set where the number of driverless minibuses in the sub-area meets the threshold constraint, and the corresponding formula is: (6) in, Characterizes the constraint threshold of routing selection in the K-layer sub-area, In Indicates the status of the number of valid driverless minibuses in each sub-area; Step 3-3, let Indicates the secondary adjustment stage from the node Assign to Node The routing strategy of , at this time, the number of vehicles in the sub-area changes from Evolved into ;otherwise .

6. The demand-responsive driverless minibus dispatching method from a multi-region perspective according to claim 5 is characterized by: When determining the constraint threshold of each sub-area routing strategy in step 3-2, the number of driverless minibuses is The state will be represented by the collection Given, set The expression is as follows: (7) In the conditions Under the action of , the steady-state solution of the Markov process is calculated, and the corresponding calculation expression is as follows: (8) in is the normalization constant, Indicates that it is used to serve the sub-area The consumption rate of driverless minibuses for public transportation demand; by the state space The generated state vector satisfies the equilibrium equation, and both the input rate and the output rate are state independent.

7. The demand-responsive driverless minibus dispatching method from a multi-region perspective according to claim 1 is characterized by: The specific rules set in step 3 are as follows: (1) When a driverless minibus in sub-area 1 is decided to be sent to a downstream node; if for If both are true, then ; Otherwise, for any , ; (2) When sub-region , When a driverless minibus is decided to be sent to a downstream node; if ,but ; when hour, ; (3) When a driverless minibus in sub-area K-1 is decided to be sent to a downstream node; like ,but ; otherwise ; in, represents the constraint threshold of the number of driverless minibuses in sub-area k; represents the vehicle dispatch determination trigger threshold of sub-area m; Represents the number of valid driverless minibuses in all sub-areas between 1 and k; represents the number of valid driverless minibuses in all sub-areas between 1 and k; In Indicates the status of the number of valid driverless minibuses in each sub-area; =1 means that an unmanned minibus is adjusted from node 1 to node K; It means that a driverless minibus is adjusted from node 1 to node i; It means that the driverless minibus has not been dispatched and still stays at node m; Indicates that there is an unmanned minibus from the node Adjust to node i; Indicates that there is an unmanned minibus from the node Adjust to Node .

Images