Reservation-based public transportation passenger and freight integrated transportation optimization method based on rendezvous point strategy

Abstract

The present application relates to the field of information and communication technology suitable for passenger and freight transport management purposes, and aims to provide a reservation type of passenger and freight co-transport optimization method based on a meeting point strategy. It includes: receiving basic transport information of passengers and freight in a reservation manner, analyzing the passenger and freight co-transport meeting point generation strategy based on space-time clustering; establishing a two-stage reservation type of passenger and freight co-transport optimization model to seek the path with the lowest cost under the premise of meeting the request; solving based on a genetic-tabu search algorithm, and iteratively optimizing the path scheme through destruction and repair operations; finding the path scheme that meets the cost minimization target, or reaching the preset iteration number. The present application realizes the space-time coordination of passenger boarding and alighting and freight loading and unloading operations, can effectively reduce the vehicle detours caused by scattered service freight requests, and significantly improves the resource utilization efficiency and overall operation efficiency of the bus in the off-peak period on the premise of prioritizing passenger travel experience.

Description

Technical Field

[0001] This invention relates to the field of information and communication technology applicable to passenger and freight transport management, and in particular to a reservation-based public transport optimization method for passenger and freight transport based on a meeting point strategy. Background Technology

[0002] As a public transport service model that dynamically schedules services based on passenger requests, reservation-based public transport features adjustable routes and customizable services, making it suitable for flexibly integrating freight transport tasks while ensuring passenger travel needs are met. However, due to the spatial and temporal dispersion of freight transport requests, directly incorporating them into reservation-based public transport route planning may lead to increased vehicle detours, decreased service efficiency, and even negatively impact passenger experience.

[0003] To overcome the aforementioned problems, the "meeting point strategy" has gained attention in the fields of shared mobility and logistics delivery in recent years. This strategy establishes several relatively fixed spatial locations that facilitate multi-user access, enabling centralized passenger boarding / alighting or cargo loading / unloading, thereby reducing vehicle detours and improving transportation efficiency. Introducing the meeting point strategy into the scenario of pre-booked public transport for both passenger and freight transport is expected to effectively integrate passenger and freight requests without significantly increasing system complexity, thus improving the utilization rate of public transport resources. However, research on applying the "meeting point strategy" to pre-booked public transport for both passenger and freight transport remains relatively scarce. In particular, systematic modeling and optimization methods are lacking in areas such as how to scientifically deploy meeting points, how to coordinate the relationship between passenger and freight requests and vehicle routes, and how to achieve efficient completion of freight tasks while ensuring passenger service quality.

[0004] For example, the applicant previously filed a patent application titled "A Route Planning Method for Passenger-Freight Combined Transport Based on Adaptive Large Neighborhood Search Algorithm" (CN119863182A). Its technical solution, by establishing a route optimization model, focuses on vehicle capacity and time window constraints while meeting all requirements, seeking the lowest-cost route solution. However, this solution focuses on optimizing passenger-freight combined transport route planning using an adaptive large neighborhood search algorithm after receiving basic passenger and freight transport information through reservations. It can be seen that this solution is still limited to the traditional "point-to-point" direct transport mode, meaning vehicles must strictly traverse the original origin and destination coordinates of every passenger or freight item. This rigid stopping requirement forces vehicles to frequently stop and make numerous unnecessary detours to serve scattered orders, significantly sacrificing trunk line transport efficiency. Therefore, this solution fails to utilize the short-distance connection capabilities of passengers and freight items for spatiotemporal aggregation, and cannot break through the efficiency bottleneck of "multi-point stops and high-frequency detours" at the source of demand distribution.

[0005] This invention proposes an optimization method for pre-booked public transport for both passenger and freight services based on a meeting point strategy, in order to solve the above problems. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide an optimization method for reservation-based public transport passenger and freight transportation based on meeting point strategy.

[0007] To solve the technical problem, the solution of the present invention is:

[0008] A reservation-based public transport optimization method for passenger and freight transportation based on a meeting point strategy is provided, including the following steps:

[0009] (1) Receive basic transportation information of passengers and goods by reservation, including at least the number of passengers and goods, origin and destination and time requirements; analyze the generation strategy of passenger and freight rendezvous point based on spatiotemporal clustering by collecting the characteristics of different passenger and freight requests;

[0010] (2) Based on the meeting point strategy, a two-stage reservation-based public transport passenger and freight co-transport optimization model is established to seek the path with the lowest cost under the premise of satisfying the request; the optimization model includes two parts: constraints and objective function. The objective function is divided into two stages: one focusing on operational efficiency and the other focusing on service quality; the request includes at least passenger and freight service requests, feasible meeting point locations, time window restrictions for requesting services, and vehicle capacity.

[0011] (3) Solve the optimization model based on the genetic-taboo search algorithm, and iteratively optimize the path scheme through destruction and repair operations until a path scheme that meets the cost minimization objective is found, or the preset number of iterations is reached.

[0012] Compared with the prior art, the technical advantages of the present invention are:

[0013] 1. This invention proposes a reservation-based public transport optimization method for passenger and freight sharing based on a meeting point strategy. By scientifically deploying relatively fixed meeting points, it spatially aggregates scattered freight requests, achieving spatiotemporal coordination between passenger boarding / alighting and freight loading / unloading operations. This effectively reduces vehicle detours caused by serving scattered freight requests, significantly improving resource utilization efficiency and overall system operational efficiency during off-peak hours while prioritizing passenger travel experience. Numerical experimental results show that the meeting point strategy-based model outperforms the traditional model that does not consider meeting points.

[0014] 2. This invention constructs a two-stage mixed-integer programming model that integrates passenger and freight requests, aiming to scientifically plan vehicle routes and task allocation. In the first stage, the model focuses on improving system operational efficiency, with the core objective of shortening vehicle travel time and reducing station dwell times, quickly filtering feasible passenger and freight requests. In the second stage, the model shifts its optimization focus to improving service quality, striving to reduce passenger on-vehicle time, minimize walking transfers, and reduce freight transport transfer times, aiming to provide passengers with a faster and more comfortable travel experience while ensuring freight efficiency. Throughout the optimization process, the model comprehensively considers practical constraints such as time windows and vehicle capacity, ensuring that the final solution is not only theoretically optimal but also closely matches the actual operational scenario, providing a solid decision-making foundation for the efficient and reliable operation of the public transportation system under the passenger-freight mixed transport mode.

[0015] 3. This invention proposes an efficient hybrid heuristic algorithm to address the high complexity and difficulty of solving the aforementioned model. This algorithm integrates the local optimization capability of tabu search with the global exploration mechanism of genetic algorithms, balancing the depth and breadth of the search through a dynamic adaptive strategy. The algorithm achieves rapid response to dynamic requests in the first stage based on an online insertion and evaluation strategy; and in the second stage, it refines the solution of vehicle paths through global optimization. Numerical experimental results show that the algorithm performs well in terms of solution quality and stability, and can efficiently handle large-scale real-world cases, providing a powerful computational tool for the practical application of the method. Attached Figure Description

[0016] Figure 1 This is a schematic diagram illustrating the determination of the meeting point in this invention.

[0017] Figure 2 This is a schematic diagram of the meeting point strategy of the present invention.

[0018] Figure 3 This is a schematic diagram illustrating the execution process of the pre-booked public transport system for both passenger and freight transport in this invention.

[0019] Figure 4 This is a schematic diagram illustrating the execution process of the solution algorithm in this invention.

[0020] Figure 5 This is a schematic diagram of the passenger-freight co-transport bus route under the meeting point strategy generated in this invention. Detailed Implementation

[0021] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0022] There are two comparable strategy modes for the planning problem of pre-booked public transport for both passengers and goods: the no-meeting-point strategy and the meeting-point strategy. The no-meeting-point mode is the traditional approach commonly used in existing technologies. In this mode, the vehicle must sequentially visit the actual origin / destination of each passenger or cargo, and all requests are completed at their original locations. The vehicle path constitutes a standard "node-unique-passage" vehicle routing problem (such as the scheme used in CN119863182A). The meeting-point mode, on the other hand, pre-defines several spatially fixed "meeting points" in the road network that facilitate the aggregation of multiple requests. Passengers or cargo can reach the nearest meeting point via short-distance transfer methods such as walking, mini-shuttles, or temporary parking before completing boarding / alighting or loading / unloading all at once. In this mode, the vehicle only needs to visit the meeting point without needing to penetrate each original request location, thus effectively reducing detour distance and the number of stops. The difference between the two strategies lies in whether or not "meeting points" are used to achieve spatial aggregation of requests: the no-meeting-point mode can guarantee "zero transfer" of passengers or goods but may increase vehicle mileage; the meeting-point mode trades moderate transfer costs for intensive vehicle routing, thereby reducing operating mileage, compressing fleet size, and improving the utilization rate of public transport resources. This invention constructs a quantitative model to address this problem, and while ensuring the timeliness of passenger travel and freight transport, it collaboratively optimizes request-meeting-point allocation and vehicle routing to achieve the green co-transportation goal of prioritizing passenger transport, coordinating freight transport, and ensuring system efficiency.

[0023] This invention proposes an optimization method for pre-booked public transport passenger and freight services based on a meeting point strategy. It mainly comprises three key parts: meeting point strategy analysis, establishment of a pre-booking optimization model, and construction of a solution algorithm. Specifically, the meeting point strategy is formulated based on the characteristics of different passenger and freight requests to determine the optimal boarding / alighting or loading / unloading locations for each request. The pre-booking optimization model aims to find the lowest-cost service solution while satisfying various request conditions. The solution algorithm is constructed to address the computational complexity of large-scale mixed-integer programming problems by proposing a computational method suitable for the model structure, thereby achieving both accuracy and speed in solving the model.

[0024] The system referred to below in this invention is a pre-booked public transport passenger and freight system, which is essentially a transportation command system integrating several pre-booking terminals and a cloud-based dispatch center. In actual operation, customer travel and freight transportation request information can be collected through a software APP installed on smart terminal devices (such as smartphones, smart tablets, computer terminals, etc.). After being uploaded to the cloud or data center server via the Internet, the system platform built into the server calls the pre-booked public transport passenger and freight optimization model and auxiliary calculation, demonstration, communication and other functional modules to specifically implement the passenger and freight transport meeting point strategy shown in this invention. After mastering the technical solution of this invention, those skilled in the art can implement the layout of the passenger and freight transport system and the implementation process of related models or functional modules based on their skills without any obstacles. Therefore, the detailed implementation details of this invention will not be elaborated further.

[0025] Step 1: Establish and analyze passenger and freight transport convergence point strategies

[0026] The system receives basic transportation information for passengers and goods via reservation, including at least the number of passengers and goods, origin and destination, and time requirements. By collecting the characteristics of different passenger and freight requests, the system analyzes strategies for generating passenger and freight rendezvous points based on spatiotemporal clustering.

[0027] This invention proposes a strategy for generating passenger and freight rendezvous points based on spatiotemporal clustering. This strategy introduces the concept of an acceptable service domain, transforming traditional point-to-point transportation demand into a flexible spatial region. It then utilizes Monte Carlo sampling and dynamic time window pruning methods to calculate the optimal spatiotemporal rendezvous point capable of simultaneously serving multiple requests. The strategy mainly comprises the following three core steps:

[0028] 1. Parameterize basic transportation information to construct service domains.

[0029] Extract key transportation parameters from passenger travel and freight transportation requests, including at least the original coordinates, time window, and request type; configure the connection tolerance radius based on the request type, and transform the original origin and destination of each request into an acceptable service domain.

[0030] The system first receives a set of passenger travel requests within a specified time frame. Collection of cargo transportation requests For any request The system extracts its key transportation parameters:

[0031] Original coordinates: Pick-up / Pick-up point With drop-off / delivery point ;

[0032] Time window: The earliest start time expected by the user. With the latest end time Forming a time window Request type: Identified as "passengers" or "cargo".

[0033] Based on the request type, the system configures a connection tolerance radius for each request. To define the user's spatial flexibility: if For passengers, set (Acceptable walking radius) represents the walking distance that passengers are willing to walk; if For goods, set (Acceptable transport connection radius) represents the acceptable end-point connection distance for goods.

[0034] Accordingly, the origin and destination of each request are transformed into an acceptable service domain. For coordinates The origin of the point whose service domain satisfies:

[0035]

[0036] In the formula, This indicates the coordinates of the potential rendezvous point.

[0037] 2. Calculation of spatial rendezvous points based on Monte Carlo sampling method

[0038] Traverse possible request combinations to find the intersection of all service domains involved in the requests; use the bounding box Monte Carlo sampling method to solve the problem, and obtain spatial candidate meeting points by constructing sampling bounding boxes, random sampling, and intersection verification.

[0039] As shown in Figure 1, to determine the optimal rendezvous point in space, the system traverses possible request combinations, searching for the intersection of all service domains involved in the requests. Given the irregularity of the intersecting regions of multiple circles, this invention employs the bounding box Monte Carlo sampling method for solving this problem.

[0040] Step 2.1: Construct the sampling bounding box

[0041] Calculate combinations The minimum bounding rectangle range of all requested service domains. Its boundary is defined as:

[0042]

[0043]

[0044] Among them, combination This refers to a set of candidate requests, meaning the system assumes that "these requests may be served in the same location and at the same time," and names this group of requests as... ; Rectangular range It refers to combination The smallest bounding rectangle of all requested service domains in the range; , These are the left and right boundaries, respectively. , These are the upper and lower boundaries, respectively; This indicates that the larger of the two values ​​is taken. This indicates that the smaller of the two values ​​is taken.

[0045] Step 2.2: Random Sampling

[0046] exist Generate random sampling points within the range ,in ,and The representative variable follows a uniform distribution.

[0047] Step 2.3: Intersection Validation

[0048] Verification sampling points Does it exist within the service domains of all participating requests simultaneously? That is, for combinations... Any request in It must meet the following conditions:

[0049] ,

[0050] If the above conditions are met, then point It was marked as a spatial candidate rendezvous point.

[0051] 3. Effectiveness Analysis of Pruning Based on Dynamic Time Window

[0052] Feasibility filtering of spatial candidate rendezvous points is performed in the time dimension to generate effective time windows for rendezvous points; the intersection of all request time windows is calculated and a feasibility judgment is made; by verifying the order constraint, it is ensured that the actual drop-off time of any request is later than the boarding time.

[0053] Points that only satisfy spatial distance constraints may not be temporally feasible. The system needs to perform temporal feasibility filtering on spatial candidate rendezvous points to generate effective time windows for the rendezvous points. :

[0054] Step 3.1: Calculate the intersection of time windows

[0055] The service time at the rendezvous point must accommodate the time constraints of all participating requests. (System calculation combination) The earliest service time of the intersection of all request time windows, i.e., the rendezvous point, is... The latest service time is .

[0056] Step 3.2: Feasibility Judgment

[0057] Effective time window at the rendezvous point In the middle, if If so, it means there is a common time period, and the meeting point is valid; if This indicates a time conflict, and the candidate point is eliminated.

[0058] Step 3.2: Verify the association constraints between boarding and alighting.

[0059] If a combination contains both a request to board and another request to alight, or if the boarding and alighting of the same request occur at different meeting points, the system further verifies the order constraints to ensure that any request... The actual time of getting off the bus is later than the time of boarding the bus.

[0060] As shown in Figure 2, with the meeting point strategy, vehicles only need to visit the meeting point and do not need to go to the original location of each requester, which can effectively reduce detour distance and number of stops.

[0061] Step 2: Establish a reservation optimization model

[0062] A two-stage reservation-based public transport optimization model for passenger and freight transport is established based on the meeting point strategy to seek the lowest-cost path while satisfying requests. The optimization model consists of two parts: constraints and an objective function. The objective function is divided into two stages: one focusing on operational efficiency and the other on service quality. The request includes at least passenger and freight service requests, feasible meeting point locations, time window restrictions for requesting services, and vehicle capacity.

[0063] Figure 3 The document illustrates the execution process of the pre-booked public transport system for both passenger and freight transport in this invention. Users submit travel requests and specify travel information through different terminals (such as mobile phones, telephones, or computers); the system analyzes all passenger and freight requests based on a pre-booking optimization model and decides whether to accept the requests; for all accepted requests, vehicle travel tasks and route planning are generated, and the pre-booked public transport system is ultimately used to execute the tasks.

[0064] The optimization model of this invention comprises two parts: constraints and an objective function. The constraints not only cover basic path continuity, flow balance, vehicle transport time windows, and capacity constraints, but also introduce a shared aggregation incentive mechanism specifically for the meeting point strategy in passenger-freight transport. Specifically, unlike the rigid constraint in existing technologies that requires requests to correspond to fixed stations, this model requires that the actual stop point for each request be dynamically selected from its dedicated "spatiotemporal candidate set" (including original pick-up and drop-off points and multiple potential meeting points). Simultaneously, a shared reward factor is introduced into the objective function; when the model detects that multiple independent requests successfully share the same meeting point in spatiotemporal space, a function value reward is given. Through these improvements, this invention transforms the simple vehicle route optimization problem into a joint optimization problem of "dynamic station location selection - path collaborative planning," thereby guiding the algorithm to balance "low-cost end-point connections" and "high efficiency of trunk line transport."

[0065] 1. Path continuity and flow balance constraints

[0066] Equations (1)-(4) below are path continuity and flow balance constraints, used to ensure that vehicles must start from the starting station and eventually reach the destination station, and meet the requirements of arriving at and leaving each requested point.

[0067]

[0068] In the formulas, the symbols refer to: i and j are nodes or locations that provide requests or services; O is the bus origin station; D is the bus terminal station; T is the set of passenger pick-up and drop-off point pairs; and V is the set of all passenger pick-up and drop-off, delivery and meeting points. This refers to any value that the parameter can take, such as This means that for any i, it belongs to V; It is a 0-1 variable; it is 1 if the vehicle passes through path (i, j), and 0 otherwise. It is a 0-1 variable; it is 1 if the vehicle passes through the request point or meeting point i, and 0 otherwise. It is a 0-1 variable; it is 1 if the vehicle passes through path (j, i), and 0 otherwise. It is a 0-1 variable; it is 1 if the vehicle departs from the starting station, and 0 otherwise. This is a 0-1 variable; it is 1 if the vehicle reaches the terminal station, and 0 otherwise. The meanings of the other symbols are the same as above.

[0069] 2. Service Request Constraints:

[0070] Equation (5) is a service request constraint, which ensures that the origin and destination of the same request must be used or not used in the same trip. That is, when the request is transported from the origin, its destination must be part of the total trip; otherwise, neither the origin nor the destination of the request is in this trip. In addition, the setting of Equation (6) ensures that the reservation bus can serve one and only one identical request.

[0071]

[0072] Symbols in various formulas: D n For passenger drop-off or delivery point assembly; y j The variable is 0-1; it is 1 if the vehicle passes through the request point or meeting point j, and 0 otherwise; n refers to the number of passenger and freight requests; N is the set of passenger and freight requests; P n This is the meeting point for picking up or picking up customers; the meanings of the other symbols are the same as above.

[0073] 3. Vehicle transportation time window constraints:

[0074] To ensure vehicles can serve requests within the desired timeframes, appropriate time constraints need to be added to the operation of the pre-booked bus service, as follows:

[0075] Equation (7) is used to restrict vehicles from using the path after the service at request point i is terminated. The time to reach the next service point j must not exceed the service start time of that point. Equation (8) is the transportation service time priority constraint, meaning that for each passenger or cargo request, the service time for the passenger pick-up or cargo pickup process must be earlier than the time for the passenger drop-off or delivery process. Equations (9) and (10) are the time window constraints for requesting the start of service, meaning the time when vehicle service request i begins. No earlier than the specified time. It cannot be later than the specified time. .

[0076]

[0077] Symbols in various formulas: The start time of service for the vehicle at request point i; The dwell time at the request point or meeting point i; Let $ be the travel time of the vehicle between the requested points $i$ and $j$. It is a sufficiently large positive number; The start time of service for the vehicle at request point j; Set the earliest time to start service for request point i; Set the latest start time for service to request point i; the meanings of the other symbols are the same as above.

[0078] 4. Vehicle capacity constraints:

[0079] As a mode of transportation, pre-booked buses need to ensure that the actual demand for transportation does not exceed the vehicle's rated capacity during actual operation. Therefore, the model needs to limit the number of passengers and goods in the vehicle, and equations (11)-(14) constrain the vehicle's movement along the path after it finishes serving the requesting point i. Upon arrival at the next service point j, the passenger and cargo load at the time of departure from j must be equal to the sum of the passenger and cargo load after leaving point i and the passenger pick-up / drop-off or cargo loading / unloading volume at point j. Equation (15) states that the total load of the vehicle at point i must not exceed the vehicle capacity.

[0080]

[0081] Symbols in various formulas: The total number of passengers loaded in the vehicle when it leaves the request point i; The total cargo load of the vehicle when it leaves request point i; The total number of passengers loaded in the vehicle when it leaves the request point j; The total cargo load of the vehicle when it leaves the request point j; Let J be the number of passenger requests for point j. Let J be the quantity of goods requested at request point j. For bus capacity, for example when The symbol 20 indicates that the vehicle can transport 20 passengers, 20 standard parcels of goods, or a combination of 20 passengers and goods; the meanings of the other symbols are the same as above.

[0082] 5. Objective function:

[0083] The optimization model of this invention adopts a two-stage optimization framework. Its core idea is to further optimize service quality while ensuring the basic operational efficiency of the system, and it clearly reflects the principle of "passenger priority".

[0084] The objective function of Phase 1 focuses on operational efficiency, selecting feasible solutions with lower operating costs by minimizing the total vehicle travel time and total service time at the station, which lays the foundation for the efficient operation of the entire system.

[0085] Phase Two, building upon Phase One, shifts the focus of optimization to service quality. Its objective function consists of three key components: the first is the average passenger time on the vehicle, aimed at improving the passenger travel experience; the second and third are the total walking transfer time for passengers and the total transport transfer time for goods, respectively. By incorporating and optimizing the "average passenger time on the vehicle," a metric directly related to passenger experience, into the objective function in Phase Two, rather than merely treating it as a constraint, the value orientation of prioritizing passenger transport over freight transport in the public transportation system is thus implemented.

[0086] Meanwhile, the model uses different levels of convergence points and rewards to promote sharing in both phases, thus we can conclude that:

[0087] The objective function for stage one is:

[0088]

[0089] The objective function for stage two is:

[0090]

[0091] The symbols in each formula are as follows: The time it takes for the passenger to walk to the corresponding meeting point i; MP represents the time it takes for the goods to reach the corresponding rendezvous point i; MP is the set of rendezvous points. To meet passenger pick-up and drop-off point needs; Meet at the passenger pick-up point; , These represent the rewards for using the rendezvous point in each of the two phases; the other symbols have the same meaning.

[0092] The adaptive optimization framework constructed by the above-described method in this invention can accurately respond to multiple decision-making objectives such as "minimizing time cost," "minimizing operating expenses," and "maximizing system benefits." Its core lies in dynamically integrating passenger and freight demand, allocating appropriate meeting points and vehicle routes to each request, thereby achieving efficient coordination and rational allocation of transportation resources at the system level. The ultimate goal is to significantly reduce the overall operating cost of the public transportation system and improve its service efficiency, laying a solid technical foundation for promoting the innovative model of pre-booked public transport for both passengers and freight.

[0093] Step 3: Construct the solution algorithm

[0094] The optimization model is solved using a genetic tabu search algorithm, which iteratively optimizes the path scheme through destruction and repair operations until a path scheme that satisfies the cost minimization objective is found, or the preset number of iterations is reached.

[0095] This invention proposes a two-stage optimization algorithm for passenger-freight collaborative delivery based on a hybrid genetic-tabu search approach. This improved algorithm is deeply customized to address the spatiotemporal characteristics of passenger-freight co-transportation meeting point strategies and can be used to solve passenger-freight collaborative delivery path problems with time windows and capacity constraints. By integrating the global search capability of genetic algorithms with the local optimization capability of tabu search, this algorithm effectively solves the problem of traditional methods easily getting trapped in local optima when dealing with complex constraints. Simultaneously, this algorithm avoids the limitation of traditional path optimization that only sorts fixed stations, constructing a chromosome-based encoding mechanism of "path sequence + station candidates," and introducing a dedicated "shared aggregation reward" evaluation factor into the objective function. Through this structural improvement, the algorithm possesses the ability to simultaneously optimize vehicle driving order and dynamically select the best meeting point from the candidate set, thus effectively solving the "point selection-connection" collaborative optimization problem under complex spatiotemporal constraints.

[0096] The core mechanism of the optimization algorithm adopts a two-stage framework: the first stage determines whether an online decision request is accepted, and the second stage performs global path optimization on accepted requests. Through this phased collaborative optimization process, the algorithm can meet the real-time requirements of online decision-making while ensuring the quality of the solution. Figure 4 illustrates the execution process of the solution algorithm of this invention.

[0097] The steps of the two-stage optimization algorithm based on hybrid genetics-tacit search in this invention are as follows:

[0098] Step 1: Input parameters and initialization

[0099] The parameters required to run the input algorithm include: the maximum number of iterations. Population size Cross rate Variation rate Taboo search maximum iteration Contraindication table size Vehicle capacity The first objective function, which focuses on operational efficiency, accepts a threshold. .

[0100] Step 2: Generate passenger and freight demand and initialize the system.

[0101] Sequentially generate or receive passenger / cargo request streams. Each request includes type, generation time, time window, and information on boarding / unloading or loading / unloading points. Initialize an empty request acceptance list, a list of candidate boarding / loading points, and a list of candidate unloading / unloading points.

[0102] Step 3: Accept the decision-making cycle online

[0103] Determine if there are any unprocessed requests:

[0104] If not, proceed to step 6 to perform optimization; if yes, perform the following sub-steps to process the current request:

[0105] Step 3.1 Transient Assessment

[0106] The currently unprocessed requests are temporarily added to the list of accepted requests, forming a temporary request set.

[0107] Step 3.2 Optimization Solution

[0108] Taking the temporary request set as input, a hybrid genetic-tabu search algorithm is executed (see step 4), where the objective function uses a stage-one parameter that emphasizes operational efficiency (such as a smaller shared reward). Solving for the temporary optimal solution yields the solution. and its objective function value .

[0109] Step 3.3 Accept the decision

[0110] Determine whether the objective function value of the temporary solution is better than the acceptance threshold:

[0111] like If so, the request is formally accepted, permanently added to the acceptance list, and the corresponding candidate point list is updated.

[0112] like If so, the request will be rejected.

[0113] Step 3.4 Process the next request

[0114] Return to the judgment in step 3, process the next unprocessed request, and continue until all requests have been processed.

[0115] Step 4: Hybrid Genetic-Taboo Search Algorithm Flow (Core Solution Method)

[0116] This process is invoked by steps 3 and 1 to optimize a given set of requests.

[0117] Step 4.1: Generate the initial population

[0118] Coding design: Chromosomes are composed of pathway sequence genes and candidate site selection genes.

[0119] Population initialization: Randomly generated The initial population consists of 1 feasible initial chromosome that satisfies the requirement of picking up goods before delivery.

[0120] Step 4.2: Calculate fitness

[0121] For each chromosome in the population, decode its path and evaluate its objective function value (fitness). The objective function includes travel time, time in the vehicle, connection time, waiting time, sharing reward, and vehicle number penalty.

[0122] Step 4.3: Main loop of genetic algorithm ( )

[0123] Determine the current algebra Is it greater than the maximum algebra? :

[0124] If not, proceed to steps 4.3.1 to 4.3.5; if yes, exit the loop and proceed to step 4.4.

[0125] Step 4.3.1: Select Parent Generation

[0126] Based on fitness, select the best individuals from the current population as parents.

[0127] Step 4.3.2: Cross Operation

[0128] With probability Perform a crossover operation (path sequence crossover and candidate point selection crossover) on the selected parent individuals to generate offspring individuals.

[0129] Step 4.3.3: Mutation Operation

[0130] With probability Perform mutation operations (path sequence mutation and candidate point selection mutation) on offspring individuals.

[0131] Step 4.3.4: Tabu Search Local Optimization

[0132] For the offspring individuals generated by crossover mutation, perform tabu search for local optimization (see step 5 for details) and return the optimized individuals.

[0133] Step 4.3.5: Population Renewal and Generation Increment

[0134] The newly optimized individuals are added to the new population to complete one iteration and update. Then return to the judgment in step 4.3.

[0135] Step 4.4: Return the optimal chromosome

[0136] After the main loop of the genetic algorithm ends, the chromosome with the best fitness is selected from the final population as the output of this process.

[0137] Step 5: Taboo Search Local Optimization Process

[0138] This process is invoked in step 4.3.4 for deep optimization of a single chromosome.

[0139] Step 5.1: Initialization

[0140] Set the current solution Given the input chromosome, the current optimal solution Initialize the tabu list to empty, and set the iteration counter. .

[0141] Step 5.2: Taboo Search Loop ( )

[0142] judge Is it greater than :

[0143] If not, proceed to steps 5.2.1 to 5.2.4; if yes, exit the loop and proceed to step 5.3.

[0144] Step 5.2.1: Generate a set of neighborhood operations

[0145] By performing operations such as path mutation and candidate point mutation, a set of neighborhood solutions for the current solution is generated.

[0146] Step 5.2.2: Evaluate the feasibility of neighborhood solutions

[0147] Determine whether the neighborhood solution satisfies constraints such as time window, vehicle capacity, and continuity of shared points:

[0148] If the objective function is satisfied, its objective function value is evaluated; if it is not satisfied, it is considered an infeasible solution and is usually rejected or given a high penalty value.

[0149] Step 5.2.3: Selecting and Accepting the New Solution

[0150] From the feasible neighborhood solutions, select the optimal solution that is not in the tabu list as a candidate solution.

[0151] If a candidate solution is better than the current optimal solution, then update the current solution and the current optimal solution; otherwise, decide whether to update the current solution according to certain rules (such as whether it is better than the current solution).

[0152] Add the corresponding operations to the tabu list and control the size of the tabu list.

[0153] Step 5.2.4: Update the counter

[0154] renew Then return to the judgment in step 5.2.

[0155] Step 5.3: Return the optimized chromosome

[0156] After the tabu search loop ends, return the current optimal solution. As the output of this process.

[0157] Step 6: Global Path Optimization

[0158] The input is the set of requests finally accepted in step 3 and its candidate points.

[0159] The hybrid genetic-tabu search algorithm in step 4 is invoked, where the objective function adopts the parameters of the objective function in step 6 that emphasizes service quality (such as a larger shared reward). By performing fine-grained optimization, the globally optimal path solution can be obtained. .

[0160] Step 7: Output the optimal solution and end.

[0161] Decoding the Global Optimal Solution It outputs detailed optimization information, including: vehicle routes, service times at each station, and meeting point usage.

[0162] Step 4: Verify the effectiveness of the technical solution

[0163] To demonstrate the technical advantages of the path optimization method with / without meeting points in this invention, a comparative calculation was set up for verification.

[0164] In the comparative experiment, the Gurobi solver and the Genetic Tabu Search algorithm (GA-TS) were used to solve the passenger-freight co-transport route optimization model based on the meeting point strategy. For Gurobi, version 11.0 was called using Python, and the solution time limit was set to 5000 seconds. For GA-TS, the specific parameters were set as follows: Genetic Algorithm (GA): Population Size was set to 50, Maximum Iterations was set to 500, Crossover Rate was set to 0.8, and Mutation Rate was set to 0.2. Tabu Search (TS): Embedded as a local search operator in the Genetic Algorithm, its maximum number of iterations per call was set to 50, and the Tabu Size was set to 10 to prevent the algorithm from getting trapped in local optima. Under the same computational conditions, the performance differences between the two methods in solving the route optimization problem were evaluated through comparative experiments. Table 1 shows the results of comparing strategies with and without meeting points, taking an example with a vehicle capacity of 20 and 4 passenger and 4 freight demands.

[0165] Table 1 Comparison of results with / without rendezvous point strategy

[0166]

[0167] As can be seen from the data in Table 1, the meeting point strategy proposed in this invention can significantly reduce the overall objective function value of the system, thereby optimizing the global transportation cost. Particularly for vehicle operations, by setting meeting points, both vehicle travel time and dwell time are effectively reduced, with the decrease in dwell time being particularly significant. This fully verifies the technical effectiveness of this invention in reducing vehicle stop frequency and detour costs through the meeting point strategy, thus greatly improving the efficiency of mainline transportation.

[0168] On the other hand, data comparison trends show that although implementing this strategy increases the "end-point cost" of passenger walking and freight connection, this investment successfully achieves a significant improvement in the overall system efficiency, and the total benefit far outweighs the increased cost in some areas. This indicates that the present invention can utilize the short-distance connection capabilities of passenger and freight vehicles, at the cost of minimal end-point convenience, to solve the technical problem of low system efficiency caused by frequent stops in traditional shared transport models, verifying the practicality and superiority of this strategy in complex passenger and freight shared transport scenarios. Figure 5 illustrates an example of the passenger and freight shared transport bus route under the meeting point strategy of the present invention.

[0169] In summary, this invention addresses the problems of insufficient integration of passenger and freight requests, low utilization of public transport resources, frequent vehicle detours, and difficulty in guaranteeing passenger service quality in existing technologies. This method constructs a hybrid scheduling model that integrates passenger and freight requests, scientifically utilizing meeting points to achieve spatiotemporal coordination between passenger boarding / alighting and cargo loading / unloading. By employing a multi-objective optimization algorithm, it dynamically optimizes vehicle routes and task allocation, improving the overall transportation efficiency of the system. This method effectively reduces vehicle detour distances and operating mileage, lowers the road resource occupation by urban delivery vehicles, and provides theoretical support and decision-making tools for expanding the urban logistics functions of pre-booked public transport systems and realizing a new urban transportation mode of "passenger and freight sharing, green and efficient," thus contributing to the construction of smart cities and sustainable transportation systems.

[0170] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various modifications or variations within the scope of the claims, which do not affect the essence of the present invention.

Claims

1. A reservation-based public transport optimization method for passenger and freight transport based on a meeting point strategy, characterized in that, Includes the following steps: (1) Receive basic transportation information of passengers and goods by reservation, including at least the number of passengers and goods, origin and destination and time requirements; analyze the generation strategy of passenger and freight rendezvous point based on spatiotemporal clustering by collecting the characteristics of different passenger and freight requests; The analysis of the strategy for generating passenger and freight transport meeting points specifically includes: (1.1) Parameterize the basic transportation information and construct the service domain; Extract key transportation parameters from passenger travel and freight transportation requests, including at least the original coordinates, time window, and request type; configure the connection tolerance radius based on the request type, and transform the original origin and destination of each request into an acceptable service domain; (1.2) Calculation of spatial rendezvous points based on Monte Carlo sampling method Traverse possible request combinations to find the intersection of all service domains involved in the requests; use the bounding box Monte Carlo sampling method to solve the problem, and obtain spatial candidate meeting points by constructing sampling bounding boxes, random sampling, and intersection verification; (1.3) Effectiveness analysis of dynamic time window pruning Feasibility filtering of spatial candidate rendezvous points is performed in the time dimension to generate effective time windows for rendezvous points; the intersection of all request time windows is calculated and a feasibility decision is made; by verifying the order constraints, it is ensured that the actual drop-off time of any request is later than the boarding time. (2) Based on the meeting point strategy, a two-stage reservation-based public transport passenger and freight co-transport optimization model is established to seek the path with the lowest cost under the premise of satisfying the request; the optimization model includes two parts: constraints and objective function. The objective function is divided into two stages: one focusing on operational efficiency and the other focusing on service quality; the request includes at least passenger and freight service requests, feasible meeting point locations, time window restrictions for requesting services, and vehicle capacity. The first phase, focusing on operational efficiency, uses a target function that minimizes total vehicle travel time and total service time at stations to screen for feasible solutions with lower operating costs. The second phase, focusing on service quality, consists of three key components: average passenger time in the vehicle, total passenger walking transfer time, and total freight transport transfer time. Different levels of meeting-point rewards are used in both phases to promote sharing. The objective function for stage one is: (16) The objective function for stage two is: (17) In the formula, i and j represent the nodes or locations that provide the request or service; It is a 0-1 variable; it is 1 if the vehicle passes through path (i, j), and 0 otherwise. Let $ be the travel time of the vehicle between the requested points $i$ and $j$. The dwell time at the request point or meeting point i; The variable is 0-1. It is 1 if the vehicle passes through the request point or meeting point i, and 0 otherwise; MP is the set of meeting points; V is the set of all passenger pick-up and drop-off points, delivery points, and meeting points. For the request point The number of passenger requests; The start time of service for the vehicle at request point i; The start time of service for the vehicle at the request point j; The time it takes for the passenger to walk to the corresponding meeting point i; The time it takes for the goods to reach the corresponding rendezvous point i; To meet passenger pick-up and drop-off point needs; Meet at the passenger pick-up point; , These represent the rewards for using the rendezvous point in each of the two phases. (3) Solve the optimization model based on the genetic-taboo search algorithm, and iteratively optimize the path scheme through destruction and repair operations until a path scheme that meets the cost minimization objective is found, or the preset number of iterations is reached.

2. The method according to claim 1, characterized in that, In step (2), the constraints include: path continuity and flow balance constraints, service request constraints, vehicle transport time window constraints, and vehicle capacity constraints.

3. The method according to claim 2, characterized in that, The path continuity and flow balance constraints mean that vehicles must depart from the starting depot and eventually arrive at the destination depot, while meeting the requirements of arriving at and leaving each requested point.

4. The method according to claim 2, characterized in that, The request service constraint means that the origin and destination of the same request must be used or not used in the same trip; that is, when the request is transported from the origin, its destination must be part of the total trip, otherwise the origin and destination of the request are not in this trip.

5. The method according to claim 2, characterized in that, The vehicle transportation time window constraint means that the time it takes for a vehicle to arrive at the next service point must not be greater than the service start time of that point. For each passenger or cargo request, the service time for the passenger pick-up or cargo pickup process must be earlier than the time for the passenger drop-off or delivery process. The time when the vehicle starts the service request should be within the specified time interval.

6. The method according to claim 2, characterized in that, The vehicle capacity constraint means ensuring that the actual number of vehicles needed to transport does not exceed the vehicle's rated capacity.

7. The method according to claim 1, characterized in that, In step (3), the genetic-taboo search algorithm adopts a two-stage framework: the first stage is to decide whether to accept the online request, and the second stage is to optimize the global path for accepted requests; specifically, it includes the following steps: (3.1) Input parameters and initialization (3.2) Generate passenger and freight demand and initialize the system (3.3) Online decision-making cycle Determine if there are any unprocessed requests; if not, proceed to step (3.6) to perform optimization; if yes, perform the following operations to process the current request: transient evaluation, optimization solution, acceptance of decision, and processing of the next unprocessed request; (3.4) Hybrid Genetic-Taboo Search Algorithm Flow The optimization process for a given set of requests includes: generating an initial population, calculating fitness, running the main loop of the genetic algorithm, and returning the optimal chromosome. After the main loop of the genetic algorithm is completed, the chromosome with the best fitness is selected from the final population as the output. (3.5) Tabu Search Local Optimization Process This function is used for deep optimization of a single chromosome, including: initialization, tabu search loop, and returning the optimized chromosome; after the tabu search loop ends, it returns the current optimal solution as the output. (3.6) Global path optimization Input the final set of requests and their candidate points accepted in step (3.3), call the hybrid genetic-tabu search algorithm in step (3.4), use the objective function parameters that emphasize service quality, and perform fine optimization to obtain the globally optimal path scheme; (3.7) Output the optimal solution and end. Decode the global optimal solution and output detailed optimization information, including: vehicle routes, service times at each station, and rendezvous point usage.

Images