An express delivery distribution optimization method of urban public transport and unmanned aerial vehicle cooperation

Abstract

This invention relates to the field of collaborative transportation using public transport and drones, and discloses an optimized method for express delivery allocation in urban public transport and drone collaborative transportation. The steps are as follows: collecting urban traffic flow information data; extracting clustered bus trajectory data features, constructing a multiple linear regression equation to predict road segment travel time, and continuously optimizing the equation based on actual results to reduce prediction errors; obtaining the operating time of public transport in each city, the Euclidean distance from each POI point to the integrated transfer point, and constructing a two-layer optimization framework using a three-way optimization algorithm on the outer layer and a genetic algorithm on the inner layer to realize the express delivery allocation scheme; loading express deliveries according to the allocation scheme for each batch, and determining whether to perform transfer based on real-time traffic data. This further improves the overall efficiency and intelligence level of urban last-mile logistics.

Description

Technical Field

[0001] This invention relates to the field of collaborative transportation between public transport and drones, and specifically to an optimized method for express delivery allocation in collaborative transportation between urban public transport and drones. Background Technology

[0002] With the widespread adoption of online shopping, the express delivery industry has experienced explosive growth, but bottlenecks in last-mile delivery efficiency have become increasingly apparent. Currently, express delivery is still mainly done manually, and door-to-door orders, which account for only about 20% of the total volume, consume more than half of the delivery time, highlighting the "last mile" problem.

[0003] To address this issue, a delivery method combining public transportation and drones has emerged. Drones can traverse ground road network obstacles, achieving efficient point-to-point transportation; urban public transportation, with its extensive network of stations, can serve as resupply, take-off, landing, and transfer nodes for drones. The two complement each other, forming a three-dimensional collaborative model of "ground public transportation + low-altitude drones." Furthermore, using public transportation stations as delivery pick-up points eliminates the need for reconstruction and planning, further reducing manpower and financial resources.

[0004] However, while research and practice on collaborative transportation between urban public transport and drones have begun in the existing technology, such as the Chinese patent CN115423406A, which describes a regional collaborative delivery system and method combining drones and buses, enabling self-coordinated implementation of bus stops and delivery routes based on the interaction between the system and an app, the existing technology has not been specifically fine-tuned for traffic environment perception and prediction capabilities due to the increasingly developed transportation network and the numerous divisions of bus stops and delivery routes. Furthermore, it does not consider different traffic flow patterns (such as morning peak, evening peak, and off-peak periods) and weather factors. A single optimization algorithm struggles to simultaneously handle both weight parameter optimization and specific allocation scheme search, easily getting trapped in local optima and failing to guarantee optimal global delivery efficiency. Summary of the Invention

[0005] Purpose of the invention: To address the problems mentioned in the background art, this invention discloses an optimized method for express delivery allocation using collaborative transportation of urban public transport and drones. By coupling data-driven modeling with a two-layer optimization framework, it achieves efficient integration of urban public transport and drones, thereby improving the overall efficiency and intelligence level of urban last-mile logistics.

[0006] Technical solution

[0007] This invention discloses an optimized express delivery allocation method for collaborative transportation using urban public transport and unmanned aerial vehicles (UAVs), the method comprising the following steps:

[0008] S1: Collect urban traffic flow information data;

[0009] S2: Extract features from clustered bus trajectory data, construct a multiple linear regression equation to predict road segment travel time, and continuously optimize the equation based on actual results to reduce prediction error;

[0010] S3: Obtain the bus operating time of each city, the Euclidean distance from each POI point to the integrated transfer point, and construct a two-layer optimization framework using the Three Kingdoms optimization algorithm in the outer layer and the genetic algorithm in the inner layer to realize the express delivery allocation scheme.

[0011] S4: Load packages according to the distribution plan for each batch of packages, and determine whether to connect them based on real-time traffic data.

[0012] Furthermore, S1 collects urban traffic flow information data, including traffic flow data related to the urban road network, urban bus operation trajectory data, and urban bus operation data; it also acquires POI data including but not limited to companies, schools, commercial centers, and communities; and it acquires express delivery data for the day, including but not limited to delivery location, urgency, volume, weight, and type data; express delivery storage devices are installed on the top of urban buses, and pick-up points are set up within the city.

[0013] Furthermore, based on the urban traffic flow information data, S2 divides the time periods into different traffic flow patterns, sets road intervals within each time period, extracts urban public transport trajectory data features within those intervals, performs PCA dimensionality reduction, and then clusters the data. Based on traffic volume, relevant factors in the cluster type, and characteristic time, an n+1 order linear regression equation is designed:

[0014]

[0015] in Time taken to travel through this section of road; : Coefficients of each influencing factor in the cluster type; The nth influencing factor in cluster type; : Regression equation coefficients; : The influence coefficient of traffic volume in the interval; Indicates the error term; q: the traffic volume of urban buses passing through this section.

[0016] Furthermore, the outer layer of the dual-layer optimization framework described in S3 uses the Three Kingdoms optimization algorithm to optimize the weight allocation of express delivery priority evaluation factors, while the inner layer uses a genetic algorithm. Collaborative optimization is achieved through a bidirectional feedback mechanism between the outer and inner layer algorithms. The outer Three Kingdoms optimization algorithm outputs a set of weight vectors. The inner layer genetic algorithm calculates the priority score of all express deliveries based on this weight, generates a specific multi-batch allocation scheme, and calculates the total idle time of the drones corresponding to this scheme. The idle time is fed back as a fitness value to the outer three-way optimization algorithm to evaluate the current weight vector. The advantages and disadvantages of each factor are analyzed, and TKOA is driven to perform the next round of weight iteration optimization. This cycle continues until the termination condition is met.

[0017] Furthermore, the specific optimization steps of the two-layer optimization framework are as follows:

[0018] Initialize the population for the Three Kingdoms optimization algorithm, with each individual representing a set of weight vectors. These correspond to the weight, volume, destination, and urgency of the express delivery, respectively.

[0019] For each weight vector The priority score for each package is calculated. Based on the priority score, a genetic algorithm is used to allocate packages in multiple batches. Each batch corresponds to one operating shift of a city bus. The total idle time of the drones in the allocation scheme is calculated and used as the weight vector. Fitness evaluation;

[0020] The iterative optimization of the Three Kingdoms optimization algorithm is performed, including dynamic faction reorganization, cross-faction learning, and competition mechanisms; a termination condition is determined, and if satisfied, the optimal weight vector and the corresponding delivery allocation scheme are output. The objective function of the two-layer optimization model is:

[0021]

[0022] in, In weight Total idle time of drones For the first The first integrated connection point Arrival time, This refers to the time for a single drone pickup. To determine the total number of connection points, To be assigned to the pickup point The number of express deliveries;

[0023] The formula for calculating priority scores is:

[0024]

[0025] in, , , , They represent express delivery. The weight, volume, destination, and urgency priority of the data are all normalized, and the weights satisfy the following conditions: ;

[0026] Furthermore, the Three Kingdoms optimization algorithm employs a dynamic faction reorganization mechanism: the population is divided into three factions: Wei, Shu, and Wu, and in each iteration... In each generation, individuals are reorganized based on the average fitness of each faction, migrating some individuals from low-fitness factions to high-fitness factions. A cross-faction learning mechanism is employed, allowing individuals to learn from the optimal solutions of other factions. Simultaneously, a historical experience database is established, retaining the optimal weight vectors from previous generations.

[0027] The iteration termination condition includes at least one of the following: reaching the maximum number of iterations. ;continuous The optimal fitness improvement is less than the threshold. The total idle time of drones is lower than the preset target value. .

[0028] Furthermore, the constraints of the genetic algorithm are: capacity constraint: the total weight of express deliveries allocated to each shift does not exceed the maximum load capacity of the bus. The total volume does not exceed the maximum capacity. Time window constraint: The express delivery must be delivered within the specified time window; Accessibility constraint: The city bus route must be close to the corresponding integrated transfer point for the express delivery.

[0029] The fitness evaluation calculation process includes:

[0030] For each shift, determine the arrival time of each express delivery based on the bus route; group by integrated transfer point and sort the arrival times of each point; calculate the time interval between adjacent arrivals, deduct the drone pickup time, and obtain the actual idle time; sum up the idle time of all drones; impose penalties on schemes that violate the constraints.

[0031]

[0032] in, , , These are the penalty values ​​for exceeding weight, volume, and time limits, respectively. , , The penalty coefficient is... It is a very small positive number.

[0033] Furthermore, the connection determination described in S4 specifically involves transmitting the real-time operation information of the bus to the central processor and then transmitting it to the drone at the integrated connection point in real time to determine whether the delivery will be completed within the predetermined time period. If it is, the delivery will proceed as scheduled; otherwise, the delivery will be deemed a failure, and the information will be transmitted to the central processor to recalculate the delivery information.

[0034] Beneficial effects:

[0035] 1. This invention employs a two-layer optimization framework to achieve bidirectional collaborative optimization of express delivery priority weights and specific allocation schemes. It comprehensively considers multiple attributes such as express delivery weight, volume, destination, and urgency, aiming to minimize the total idle time of drones. This minimizes the risk of drones getting trapped in local optima and only picking up the nearest express delivery, instead approaching the global optimum and ensuring the optimality of the overall delivery plan. Furthermore, the algorithm's greatest advantage lies in its ability to provide high-quality, executable scheduling schemes within a limited time, even in complex systems with nonlinearity, multiple constraints, multiple objectives, and dynamic disturbances.

[0036] 2. This invention integrates multi-source data such as urban road network traffic flow, bus operation trajectories, weather, and day of the week, and combines feature extraction, cluster analysis, and multiple linear regression to establish a road segment travel time prediction model. This model can accurately adapt to different traffic flow patterns and weather changes, such as peak hours and off-peak hours, providing a reliable time benchmark for energy-saving drone shuttle services. Furthermore, the system collects bus operation data in real time and dynamically updates the prediction model, possessing the ability to adaptively handle uncertainties such as traffic fluctuations and weather changes. When a shuttle task fails to be completed within the scheduled time, the system can promptly transmit delivery information back to the central processor for rescheduling, improving the overall system's fault tolerance and execution reliability. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating the specific process of the method of the present invention;

[0038] Figure 2 This is a map showing bus routes and demand locations according to an embodiment of the present invention;

[0039] Figure 3 This is a public transport connection distribution map according to an embodiment of the present invention;

[0040] Figure 4 This is a trajectory diagram of a drone according to an embodiment of the present invention. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] like Figure 1 As shown, this invention discloses an optimized express delivery allocation method using collaborative transportation of urban public transport and drones. The method steps are as follows:

[0043] Step 1: Basic data collection and urban logistics system construction, specifically including:

[0044] Acquire traffic flow data, urban public transport route data, and urban public transport operation data related to the urban road network; obtain POI data, including but not limited to, from companies, schools, commercial centers, and communities, through Python web scraping; and obtain express delivery data for the day, including but not limited to delivery location, urgency, volume, weight, and type data.

[0045] In areas with high express delivery demand, set up integrated transfer points, namely drone parcel lockers, to integrate warehousing and drone transfer functions, for parcel collection and distribution, drone charging and battery replacement, and support rapid battery swapping after drones return.

[0046] Drone-based parcel storage devices are installed on the roofs of city buses to handle long-distance last-mile delivery. Each city bus loads parcels onto the drone storage device before departing from its starting station. City buses depart according to a fixed timetable and travel along fixed routes. When a city bus approaches a nearby integrated transfer point, the drone predicts and picks up and delivers the parcels. When a city bus arrives at its final stop, the drone reloads the parcels and resumes its journey according to the timetable. The bus routes and demand points selected in this embodiment are as follows: Figure 2 As shown.

[0047] Step 2: Based on the traffic flow information of the urban road network, divide the time periods of different traffic flow patterns, set appropriate road intervals within the time periods of different traffic flow patterns, extract the trajectory data features of urban buses within their intervals, and perform clustering.

[0048] Let the weather conditions on a certain day be... , In this system, 0 represents sunny, 1 represents light rain, 2 represents heavy rain, 3 represents light snow, 4 represents heavy snow, and 5 represents fog, haze, wind, or other conditions. Due to significant differences in daily city bus routes throughout the week, the seven days of the week are represented as follows: Meanwhile, excluding special holidays, urban public transport trajectory information of the same type and the same route was selected for modeling and analysis.

[0049] MATLAB was used to model and analyze urban public transport trajectory information. The modeling process followed a pipeline model of "feature extraction → standardization → dimensionality reduction → clustering," ensuring that each step was optimized based on the output of the previous step. The specific steps are as follows:

[0050] (1) Construction of feature matrix

[0051] Calculate feature values ​​for each road segment and construct a two-dimensional matrix of "road segment × feature" as input data for cluster analysis.

[0052] (2) Data standardization

[0053] The feature matrix is ​​standardized using z-scores so that the mean of each feature is 0 and the standard deviation is 1.

[0054] (3) Principal component analysis

[0055] PCA is performed on the standardized features to extract the main feature components, reducing the data dimensionality while retaining most of the information.

[0056] (4) Determining the optimal number of clusters

[0057] The elbow rule was used to analyze the variation of the sum of squares within clusters under different numbers of clusters, and to determine the optimal number of clusters.

[0058] (5) K-means clustering execution

[0059] Using the PCA-reduced data as input, the K-means clustering algorithm is executed to classify road segments into different categories.

[0060] (6) Results Analysis and Verification

[0061] Perform statistical analysis and visualization on the clustering results to verify the clustering effect.

[0062] An n+1 order linear regression equation is designed based on traffic volume, cluster type, and characteristic time to predict road segment travel time. PCA dimensionality reduction can reduce the error caused by too many dependent variables in linear regression. By continuously updating data in practical applications to optimize the prediction results, the error in urban public transport arrival time is reduced.

[0063]

[0064] In the formula:

[0065] Time taken to travel through this section of road;

[0066] : Coefficients of each influencing factor in the cluster type;

[0067] : Regression equation coefficients;

[0068] : The influence coefficient of traffic volume in the interval;

[0069] Error term, which is independent and normally distributed;

[0070] q: Traffic volume of city buses passing through this section.

[0071] Step 3: Obtain the bus operating hours for each city and the Euclidean distance from each POI to the integrated transfer point. Set the specific parameters for the data as follows:

[0072] City bus parameters: bus number for each route, maximum load capacity (kg), maximum volume (dm³). 3 ), the predicted arrival time (s) of each city bus route at each section;

[0073] Drone parameters: maximum payload (kg), maximum flight distance (m), average flight speed (v / m), average battery swapping time (s), average docking time (s);

[0074] Package parameters: Package number, weight (kg), volume (dm³) 3 ), distance (m).

[0075] This embodiment calculates based on the urban bus routes, integrated connection points, and demand points in this area. The acquired urban bus route information, express delivery data, and drone data are input into a two-layer optimization framework for calculation, and express delivery allocation is performed based on the optimal data results. Specific package information is shown in Table 1. Since the total simulation time in this embodiment is relatively short and the difference in receiving time between express deliveries is small, subjective urgency of the deliveries is not considered in this embodiment. Urban bus parameters: maximum load capacity (300kg), maximum volume (500dm²). 3 The parameters of the UAV are as follows: maximum payload (25kg), average flight speed (60km / h), average connection time (10s), and comprehensive connection point data are shown in Table 2.

[0076] Table 1

[0077] FID Shape * longitude latitude Weight (kg) <![CDATA[Volume (dm 3 )]]> 2 point 119.0298 33.53331 3 9 3 point 119.0353 33.53166 4 6 4 point 119.0293 33.52212 6 1 10 point 119.0312 33.54136 2 5 27 point 119.0777 33.55912 3 8 28 point 119.0713 33.57112 5 6 41 point 118.9999 33.58002 5 15 50 point 119.0518 33.60747 1 11 51 point 118.9921 33.6054 3 16 52 point 118.9816 33.6064 3 15 56 point 119.0609 33.57692 3 5 63 point 119.013 33.56476 3 13 64 point 119.0055 33.56247 1 9 65 point 119.0052 33.56558 4 9 75 point 119.0493 33.59175 1 1 79 point 119.008 33.59632 3 8

[0078] Table 2

[0079] Shape * FID longitude latitude point Integrated transfer point 118.9985 33.57365 point Integrated transfer point 119.0122 33.60087 point Integrated transfer point 119.0249 33.53444 point Integrated transfer point 119.066 33.56071 point Integrated transfer point 119.0592 33.58736

[0080] This paper proposes a two-layer optimization framework to address the urban public transport-drone delivery allocation scheme under different traffic flow patterns and time periods. The outer layer uses a three-way optimization algorithm to optimize the weight allocation of delivery priority evaluation factors, while the inner layer uses a genetic algorithm. Collaborative optimization is achieved through a bidirectional feedback mechanism between the outer and inner algorithms. The three-way optimization algorithm optimizes the weight allocation of delivery priority evaluation factors, while the genetic algorithm optimizes the specific allocation scheme for each batch of deliveries. The overall objective is to minimize the total idle time of drones within the region, outputting the optimal weight vector and the corresponding delivery allocation scheme. The specific details are as follows:

[0081] The daily delivery demand is allocated to multiple city buses, and a single city bus can make multiple deliveries per day. A two-layer optimization framework is adopted to solve the city bus-drone delivery allocation problem. The outer layer uses the Three Kingdoms Optimization Algorithm (TKOA) to optimize the weight allocation of delivery priority evaluation factors, while the inner layer uses a Genetic Algorithm (GA). Collaborative optimization is achieved through a two-way feedback mechanism between the outer and inner algorithms. The outer TKOA algorithm outputs a set of weight vectors. The inner layer uses a genetic algorithm (GA) to calculate the priority scores of all packages based on these weights, thereby generating a specific multi-batch allocation scheme and calculating the total idle time of the drones corresponding to that scheme. This idle time is fed back as a fitness value to the outer TKOA layer to evaluate the current weight vector. The advantages and disadvantages are determined, and TKOA is driven to perform the next round of weight iteration optimization. This process is repeated until the termination condition is met. The specific steps are as follows:

[0082] Initialize the population for the Three Kingdoms optimization algorithm, with each individual representing a set of weight vectors. These correspond to the weight, volume, destination, and urgency of the express delivery, respectively.

[0083] For each weight vector Calculate the priority score for each package. Based on the priority scores, use a genetic algorithm to allocate packages in multiple batches, with each batch corresponding to one bus trip in the city. Calculate the total idle time of the drones in the allocation scheme, using it as the weight vector. Fitness evaluation;

[0084] The iterative optimization of the Three Kingdoms optimization algorithm is performed, including dynamic faction reorganization, cross-faction learning, and competition mechanisms; a termination condition is determined, and if satisfied, the optimal weight vector and the corresponding delivery allocation scheme are output. The objective function of the two-layer optimization model is:

[0085]

[0086] in, In weight Total idle time of drones For the first The first integrated connection point Arrival time, This refers to the time for a single drone pickup. To determine the total number of connection points, To be assigned to the pickup point The number of express deliveries.

[0087] The formula for calculating priority scores is:

[0088]

[0089] in, , , , They represent express delivery. The weight, volume, destination, and urgency of the shipment are prioritized, and all attributes are normalized with weights that satisfy the following conditions: .

[0090] The Three Kingdoms optimization algorithm employs a dynamic faction reorganization mechanism, including:

[0091] (1) Divide the population into three camps: Wei, Shu, and Wu;

[0092] (2) Each iteration In each generation, individuals are reorganized based on the average fitness of each faction, and some individuals from low-fitness factions are migrated to high-fitness factions.

[0093] (3) A cross-faction learning mechanism is adopted, allowing individuals to learn from the optimal solutions of other factions;

[0094] (4) Establish a historical experience database and retain the optimal weight vectors of each generation.

[0095] Termination conditions include at least one of the following:

[0096] (1) Reaching the maximum number of iterations ;

[0097] (2) Continuous The optimal fitness improvement is less than the threshold. ;

[0098] (3) The total idle time of the drone is lower than the preset target value. .

[0099] The chromosome encoding in the genetic algorithm adopts a batch-based three-dimensional encoding structure:

[0100]

[0101] in, This refers to the total number of bus trips throughout the day. Indicates the first Each package is assigned to a set of package numbers for each shift, and each package can be assigned to at most one shift.

[0102] The constraints of genetic algorithms include:

[0103] (1) Capacity constraint: The total weight of express delivery allocated for each shift shall not exceed the maximum load capacity of the bus. The total volume does not exceed the maximum capacity. .

[0104] (2) Time window constraint: The express delivery must be delivered within the specified time window;

[0105] (3) Accessibility constraint: Urban bus routes must be close to the comprehensive transfer point corresponding to the express delivery.

[0106] The fitness evaluation calculation process includes:

[0107] (1) For each shift, determine the arrival time of each express delivery according to the bus route;

[0108] (2) Group by integrated connection points and sort by arrival time of each point;

[0109] (3) Calculate the time interval between adjacent arrivals, deduct the time for the drone to pick up the package, and obtain the actual idle time;

[0110] Add up the idle time of all drones;

[0111] Penalties will be imposed on schemes that violate the constraints:

[0112]

[0113] in, , , These are the penalty values ​​for exceeding weight, volume, and time limits, respectively. , , The penalty coefficient is... It is a very small positive number.

[0114] The results after running the program in Matlab are as follows:

[0115] Optimal weight vector: weight , , The calculation results for the connection between urban public transport and drones are shown in Table 3.

[0116] Table 3

[0117] Group number FID line Connection point Bus arrival time (s) Drone takeoff time (s) Meeting distance (m) End longitude Ending latitude Delivery distance (m) 1 2 3 3 1438.4 1374.9 1057.7 119.029832 33.533307 470 2 3 4 3 1933.2 1777.3 2597.9 119.035345 33.531655 1013 3 4 1 3 8300.8 8268 547.1 119.029307 33.522119 1427 4 10 2 3 3047.3 3014.9 540.7 119.031225 33.541364 966 5 27 4 4 1458 1417 683.9 119.077735 33.559121 1095 6 28 5 4 776.8 726 846.1 119.071253 33.571119 1252 7 41 5 1 2084.7 1999.1 1426.2 118.999909 33.580019 719 8 50 1 5 1198.1 912 4769 119.051795 33.607468 2336 9 51 1 2 6060.3 6002.3 966.8 118.99209 33.605399 1926 10 52 3 2 5365.4 5097.5 4465.8 118.981581 33.606398 2897 11 56 3 5 3026.6 2937.3 1487.7 119.060949 33.57692 1170 12 63 4 1 5265.4 5054.7 3510.9 119.01297 33.564763 1667 13 64 3 1 6149.6 5974.6 2916.1 119.005541 33.562474 1402 14 65 1 1 6908.6 6859.8 812.7 119.005247 33.565576 1096 15 75 2 5 2497.6 2211.1 4774.5 119.049311 33.591748 1037 16 79 4 2 4691.4 4541.8 2494.1 119.007995 33.596322 637

[0118] Step 4: Load the packages according to the specific allocation plan for each batch of express deliveries to obtain the initial feasible solution of this plan. This embodiment conducts a joint simulation of a network of 5 bus routes and 16 delivery demand data points, including pick-up point selection, package allocation, and dynamic pick-up mechanism. To ensure smooth simulation operation, the simulation is performed with a time limit of one round trip for these 5 city bus routes. The final delivery result obtained after running the simulation with MATLAB is shown in the figure below. Figure 3 As shown, the drone trajectory map is as follows: Figure 4As shown. In the specific implementation process, the real-time operation information of urban buses is transmitted to the central processor and then transmitted to the drones at the integrated transfer points in real time. This determines whether the delivery has been completed within the predetermined time period. If it has, the delivery will proceed according to the predetermined trajectory; otherwise, the task is deemed a failure, and the information is transmitted to the central processor to recalculate the delivery information.

[0119] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent transformations or modifications made in accordance with the spirit and essence of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones, characterized in that, The method includes the following steps: S1: Collect urban traffic flow information data; S2: Extract features from clustered bus trajectory data, construct a multiple linear regression equation to predict road segment travel time, and continuously optimize the equation based on actual results to reduce prediction error; S3: Obtain the bus operating time of each city, the Euclidean distance from each POI point to the integrated transfer point, and construct a two-layer optimization framework using the Three Kingdoms optimization algorithm in the outer layer and the genetic algorithm in the inner layer to realize the express delivery allocation scheme. S4: Load packages according to the distribution plan for each batch of packages, and determine whether to connect them based on real-time traffic data.

2. The method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones according to claim 1, characterized in that, S1 collects urban traffic flow information data, including traffic flow data related to the urban road network, urban public transport operation trajectory data, and urban public transport operation data; it also acquires POI data including but not limited to companies, schools, commercial centers, and communities; and it acquires express delivery data for the day, including but not limited to delivery location, urgency, volume, weight, and type data; express delivery storage devices are installed on the top of urban buses, and pick-up points are set up within the city.

3. The method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones according to claim 2, characterized in that, Based on the urban traffic flow information data, S2 divides the time periods into different traffic flow patterns, sets road intervals within each time period, extracts urban public transport trajectory data features within each interval, performs clustering after PCA dimensionality reduction, and designs an n+1 order linear regression equation based on traffic volume, relevant factors in the cluster type, and characteristic time: ； in Time taken to travel through this section of road; : Coefficients of each influencing factor in the cluster type; The nth influencing factor in cluster type; : Regression equation coefficients; : The influence coefficient of traffic volume in the interval; Indicates the error term; q: Traffic volume of city buses passing through this section.

4. The method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones according to claim 1, characterized in that, The dual-layer optimization framework described in S3 employs a Three Kingdoms optimization algorithm in the outer layer to optimize the weight allocation of express delivery priority evaluation factors, while the inner layer uses a genetic algorithm. Collaborative optimization is achieved through a bidirectional feedback mechanism between the outer and inner algorithms. The outer Three Kingdoms optimization algorithm outputs a set of weight vectors. The inner layer genetic algorithm calculates the priority score of all express deliveries based on this weight, generates a specific multi-batch allocation scheme, and calculates the total idle time of the drones corresponding to this scheme. The idle time is fed back as a fitness value to the outer three-way optimization algorithm to evaluate the current weight vector. The advantages and disadvantages of each factor are analyzed, and TKOA is driven to perform the next round of weight iteration optimization. This cycle continues until the termination condition is met.

5. The method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones according to claim 4, characterized in that, The specific optimization steps of the two-layer optimization framework are as follows: Initialize the population for the Three Kingdoms optimization algorithm, with each individual representing a set of weight vectors. These correspond to the weight, volume, destination, and urgency of the express delivery, respectively. For each weight vector The priority score for each package is calculated. Based on the priority score, a genetic algorithm is used to allocate packages in multiple batches. Each batch corresponds to one operating shift of a city bus. The total idle time of the drones in the allocation scheme is calculated and used as the weight vector. Fitness evaluation; The iterative optimization of the Three Kingdoms optimization algorithm is performed, including dynamic faction reorganization, cross-faction learning, and competition mechanisms; a termination condition is determined, and if satisfied, the optimal weight vector and the corresponding delivery allocation scheme are output. The objective function of the two-layer optimization model is: ； in, In weight Total idle time of drones For the first The first integrated connection point Arrival time, This refers to the time for a single drone pickup. To determine the total number of connection points, To be assigned to the pickup point The number of express deliveries; The formula for calculating priority scores is: ； in, , , , They represent express delivery. The weight, volume, destination, and urgency priority of the data are all normalized, and the weights satisfy the following conditions: .

6. The method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones according to claim 5, characterized in that, The Three Kingdoms optimization algorithm employs a dynamic faction reorganization mechanism: the population is divided into three factions: Wei, Shu, and Wu, and each iteration... In each generation, individuals are reorganized based on the average fitness of each faction, and some individuals from low-fitness factions are migrated to high-fitness factions. A cross-faction learning mechanism is adopted, allowing individuals to learn from the optimal solutions of other factions. At the same time, a historical experience database is established to retain the optimal weight vectors from previous generations. The iteration termination condition includes at least one of the following: reaching the maximum number of iterations. ;continuous The optimal fitness improvement is less than the threshold. ; The total idle time of the drones is lower than the preset target value .

7. The method for optimizing express delivery allocation through collaborative transportation of urban public transport and drones according to claim 5, characterized in that, The constraints of the genetic algorithm are: capacity constraint: the total weight of express deliveries allocated to each shift does not exceed the maximum load capacity of the bus. The total volume does not exceed the maximum capacity. Time window constraint: The package must be delivered within the specified time window; Accessibility constraint: Urban public transport routes must be close to the comprehensive transfer points corresponding to express delivery services; The fitness evaluation calculation process includes: For each shift, determine the arrival time of each express delivery based on the bus route; group by integrated transfer point and sort the arrival times of each point; calculate the time interval between adjacent arrivals, deduct the drone pickup time, and obtain the actual idle time; sum up the idle time of all drones; impose penalties on schemes that violate the constraints. ； in, , , These are the penalty values ​​for exceeding weight, volume, and time limits, respectively. , , The penalty coefficient is... It is a very small positive number.

8. The method for optimizing express delivery allocation using urban public transport and drone collaborative transportation as described in claim 1, characterized in that, The connection determination described in S4 specifically involves transmitting the real-time operation information of the bus to the central processor and then transmitting it in real-time to the drone at the integrated connection point to determine whether the delivery will be completed within the predetermined time period. If it is, the delivery will proceed as scheduled; otherwise, the delivery will be deemed a failure, and the information will be transmitted to the central processor to recalculate the delivery information.

Images