A multi-vehicle cooperative meal delivery path planning method suitable for an old-age canteen

Abstract

The application discloses a kind of multi-cargo collaborative meal arrangement path planning methods suitable for old-age canteen, by constructing with minimization total distribution time, minimization distribution cost, maximum meal arrangement punctuality and maximum meal temperature compliance rate as target multi-objective optimization function, it is ensured from optimization target that the parallel optimization orientation of multiple key performance indicators;In constraint level, the characteristics of vehicle, old-age meal arrangement scene requirement and temperature control demand are included;By setting the reciprocal of the above multi-objective function as fitness function, and embedding the constraint checking mechanism into the crossover and mutation operation of genetic algorithm, the genetic algorithm can continuously select the feasible solution that meets all constraints and is superior in each target in iteration;The advantage is to meet the comprehensive demand of collaborative optimization of distribution efficiency, cost and meal quality in old-age meal assistance scene.

Description

Technical Field

[0001] This invention relates to a meal delivery route planning method, and more particularly to a multi-vehicle collaborative meal delivery route planning method suitable for elderly canteens. Background Technology

[0002] Meal delivery for the elderly is a socialized service designed to provide timely and warm meals to senior citizens, with clear quality requirements for delivery timeliness and food temperature. Currently, this service generally uses electric delivery vehicles for ground delivery, but this model faces the following prominent bottlenecks in actual operation: traffic congestion affects delivery timeliness; extending services to remote areas requires the construction of expensive transfer stations, increasing operating costs; and there are service blind spots in special terrain areas such as mountainous areas and waterways.

[0003] While drone delivery effectively overcomes terrain limitations, its limited single-drone range and payload capacity make it difficult to independently support large-scale, continuous meal delivery tasks. Existing route planning research and patent solutions mostly focus on optimizing single objectives (such as shortest time or lowest cost). For example, patent CN110046857A only involves single-drone scheduling and does not achieve coordination with ground vehicles; while patent CN113139678A involves vehicle-drone coordination, it does not incorporate the critical quality constraint of meal temperature into the optimization model. Therefore, existing technologies are insufficient to meet the comprehensive needs of elderly meal assistance scenarios for the coordinated optimization of delivery efficiency, cost, and meal temperature preservation. Summary of the Invention

[0004] The technical problem this invention aims to solve is to provide a multi-vehicle collaborative meal delivery route planning method suitable for elderly canteens. This method can meet the comprehensive needs of elderly meal assistance scenarios for the coordinated optimization of delivery efficiency, cost, and meal temperature preservation quality.

[0005] The technical solution adopted by this invention to solve the above-mentioned technical problems is as follows: a multi-vehicle collaborative meal delivery route planning method suitable for elderly canteens, wherein the multi-vehicles include meal delivery vehicles and drones, and includes the following steps: Step S1: Obtain the planning information and parameters required for multi-vehicle collaborative meal delivery in the elderly canteen. The planning information and parameters include: demand points and elderly canteen information, environmental and temperature parameters, vehicle attribute parameters, and key variables for path planning. Step S2: Construct a multi-objective optimization function with the objectives of minimizing total delivery time, minimizing delivery cost, maximizing on-time meal preparation rate, and maximizing meal temperature compliance rate; Step S3: Based on the characteristics of the vehicle, the requirements of the elderly meal delivery scenario, and the temperature control requirements, construct the constraints for multi-vehicle collaborative meal delivery; Step S4: Based on the multi-objective optimization function in step S2 and the constraints in step S3, the genetic algorithm is improved to obtain an improved genetic algorithm. The improvement includes: defining the reciprocal of the multi-objective optimization function as the fitness function of the genetic algorithm, and incorporating the mechanism for verifying and processing the constraints constructed in step S3 into the crossover and mutation operations of the genetic algorithm. Step S5: Iteratively optimize the meal delivery path using the improved genetic algorithm to obtain the global optimal solution, which is the result of the collaborative meal delivery path planning between the UAV and the meal delivery vehicle.

[0006] Compared with existing technologies, the advantages of this invention are as follows: First, at the modeling level, by constructing a multi-objective optimization function with the objectives of minimizing total delivery time, minimizing delivery cost, maximizing on-time delivery rate, and maximizing food temperature compliance rate, the parallel optimization of multiple key performance indicators is ensured from the optimization objectives. Second, at the constraint level, vehicle characteristics, requirements of elderly meal delivery scenarios, and temperature control needs are incorporated. Finally, at the solution level, by setting the reciprocal of the above multi-objective function as the fitness function and embedding the constraint verification mechanism into the crossover and mutation operations of the genetic algorithm, the genetic algorithm can continuously select feasible solutions that simultaneously satisfy all constraints and are balanced and superior in each objective during iteration. Thus, this technical closed loop of "multi-objective modeling - physical constraint embedding - algorithm synchronous optimization" enables this invention to obtain a drone and meal delivery vehicle collaborative delivery solution that performs well in multiple dimensions such as time, cost, on-time delivery, and temperature preservation in a single solution, thereby meeting the comprehensive needs of elderly meal assistance scenarios for the collaborative optimization of delivery efficiency, cost, and food temperature preservation quality.

[0007] Furthermore, in step S1, the planning information and parameters include: (I) Demand Point Information and Senior Citizens' Dining Hall Information: The demand point information includes the total number of demand points N (unit: points), the first demand point, and the second demand point. i Geographic coordinates of each demand point Meal demand and expected delivery time window ,in, , The upper limit of the expected delivery time, Expected minimum delivery time; the information about the senior citizens' canteen includes the geographical coordinates of the canteen's location. Meal preparation completion time (Unit: s) and initial temperature of the food (Unit: °C) (ii) Environmental and temperature parameters: including ambient temperature (Unit: °C) Food temperature decay coefficient (unit: ) and preset minimum safe temperature for meals (Unit: °C) (III) Vehicle attribute parameters: including the total number of vehicles M (unit: vehicles), of which the number of catering vehicles is... (Unit: vehicles), the number of drones is (Unit: vehicles), satisfying M = + Number all vehicles from 1 to M, where The catering trucks are numbered 1 to , The drone's serial number is +1 to M, numbered The vehicle is called a vehicle. , =1, 2, ..., M; Vehicle The attributes include: rated load capacity (Unit: kg) Movement speed (Unit: m / s), Energy cost per unit mileage Maximum driving range (Unit: m); For catering carts, the thermal insulation performance coefficient is also included. , For drones, this also includes the drone's maximum delivery radius. and thermal insulation performance coefficient , ; (iv) Key variables for path planning: including vehicles For the i Task allocation variables for each demand point Vehicle Demand point access sequence Total driving / flight mileage , No. The actual delivery time and food temperature at each demand point If the vehicle Responsible for the For each demand point, then Otherwise, it is 0.

[0008] Furthermore, the multi-objective optimization function constructed in step S2 Equation (1) is used to express this as: (1) In equation (1), The weighting coefficient for total delivery time The weighting factor for total delivery cost Weighting coefficients for on-time meal delivery rate The weighting coefficients for the food temperature compliance rate are all greater than 0 and satisfy the following conditions: Its specific value is preset according to the priority of the delivery service; The total delivery time is calculated using the formula shown in equation (2): (2) in, For vehicles Departure time; For vehicles To the point of demand The delivery distance is calculated based on the road network distance for food delivery vehicles and the straight-line distance for drones. For vehicles movement speed, For vehicles Total waiting time during the delivery process; The total delivery cost is calculated using the formula shown in equation (3): (3) The formula for calculating the on-time delivery rate is shown in equation (4): (4) in, Used to represent demand points Is the actual delivery time within If inside, then ,otherwise ; The formula for calculating the food temperature compliance rate is shown in equation (5): (5) in, Used to represent demand points Food delivery temperature Is it greater than or equal to? If so, then ,otherwise ; The calculation formula is shown in equation (6): (6) In equation (6), To be responsible for the demand points i The thermal insulation performance coefficient of a certain vehicle used for food delivery. This indicates the distance a vehicle travels from the cafeteria to the point of demand. Travel time for all route segments traversed For a certain vehicle to reach the demand point Previous cumulative waiting time, e is the base of the natural logarithm.

[0009] Furthermore, the constraints constructed in step S3 include load constraints, endurance constraints, drone delivery radius constraints, time window constraints, task allocation constraints, and food temperature constraints.

[0010] Furthermore, the load constraint, the endurance constraint, the drone delivery radius constraint, the time window constraint, the task allocation constraint, and the food temperature constraint are respectively shown in equations (7) to (12): (7) (8) (9) (10) (11) (12) in, Indicate demand points The actual delivery time.

[0011] Furthermore, the specific process of iteratively optimizing the meal preparation path using the improved genetic algorithm in step S5 is as follows: Step S5.1: Initialize the parameters of the improved genetic algorithm: Set the population size. Maximum number of iterations Crossover probability Probability of mutation Number of elite individuals retained ; Step S5.2: Set the population coding rule to integer-type two-layer coding: Each individual is represented by a two-layer coding method. The first layer is the task assignment code, which is a sequence A of length N. The first layer of sequence A is the task assignment code. i The data A[i] represents the demand point. i The assigned vehicle number, i.e., the demand point i The food is assigned to vehicle A[i] for delivery, where A[i]∈{1, 2, ..., M}; the second layer is path sequence encoding, which is a list R, where the k-th column of list R is the ordered access sequence of all demand points assigned to vehicle k, i.e., the vehicle... Demand point access sequence ; Step S5.3: Initial population generation: Under the constraints described above, randomly generate... Individual task assignment codes are generated, and based on these codes, all assigned demand points for each vehicle are randomly arranged to generate a path sequence code. The currently generated code is then used to... Individuals constitute the 0th generation population, i.e., the initial population; Step S5.4: Calculate the fitness value of each individual in the 0th generation population using equation (13): (13) Among them, the higher the fitness value, the better the overall performance of the individual; Step S5.5: Set the iteration variable t and initialize t to 1; Step S5.6: Perform the t-th iteration to obtain the t-th generation population. The specific process is as follows: Step S5.6.1: Perform the selection operation: First, select the generation with the highest fitness from the (t-1)th generation population. Each individual is directly included in the t-th generation population. Individuals are selected to preserve superior genes; then, a tournament selection process is conducted: each time, two individuals are randomly selected from the (t-1)th generation population, and the one with higher fitness is selected for the mating pool. This process is repeated until the number of individuals in the mating pool reaches a certain threshold. This ensures the direction of evolution while maintaining population diversity; Step S5.6.2: Perform crossover and mutation operations on the individuals in the mating pool: Step S5.6.2.1: Take the individuals in the mating pool as parent individuals and randomly pair up the parent individuals; Step S5.6.2.2: For each pair of parent individuals, generate a random number in the interval [0, 1) for that pair of parent individuals. If this number is greater than or equal to... If the condition is met, then the crossover operation is not performed on the parent individuals; otherwise, the crossover operation is performed. The specific process of the crossover operation is as follows: The parent individuals are referred to as Parent Individual 1 and Parent Individual 2. First, a two-point crossover is performed on the first-level task allocation code: two different crossover points are randomly selected, and the vehicle number sequences of Parent Individual 1 and Parent Individual 2 between these two crossover points are swapped to generate the first-level task allocation codes for Child Individual 1 and Child Individual 2. Then, a sequential crossover is performed on the second-level path sequence code: a vehicle is randomly selected, and the demand point access sequences of that vehicle in Parent Individual 1 and Parent Individual 2 are obtained respectively, denoted as sequences S1 and S2. Two different position indices are randomly selected in sequence S1, and all elements between these two indices are defined as the reserved fragment FG. The process for Child Individual 1 with respect to that vehicle is then constructed. The demand point access sequence follows these rules: First, all demand point numbers in the reserved fragment FG are placed in their original order. Then, starting from the beginning of the demand point access sequence for this vehicle in sequence S2, the demand point numbers in the demand point access sequence for this vehicle in S2 that are not in F are appended to the demand point access sequence for the offspring individual one for this vehicle in the order in which they appear in S2. The demand point access sequence for the offspring individual two for this vehicle is constructed using a symmetric rule, that is, the fragments in S2 are retained and filled in the order of S1. For other vehicles that are not randomly selected, the offspring individual one directly inherits the corresponding demand point access sequence of the parent individual one, and the offspring individual two directly inherits the corresponding demand point access sequence of the parent individual two. Step S5.6.2.3: For all offspring individuals obtained in step S5.6.2.2, generate a random number in the interval [0, 1) for each offspring individual. If this random number is greater than or equal to... If the result is positive, the mutation operation will not be performed; otherwise, the mutation operation will be performed to update the device. The specific process of mutation operation is as follows: In the first-level task allocation encoding of the offspring individual, a demand point is randomly selected and reassigned to another vehicle that meets its vehicle type and does not violate the constraints after allocation; In the second-level path sequence encoding, a demand point access sequence of a vehicle is randomly selected and its local path is perturbed. The local path perturbation is as follows: reverse the access order of a continuous interval in the demand point access sequence, swap the positions of two demand points in the demand point access sequence, or insert a demand point in the demand point access sequence into another random position; Step S5.6.3: Perform real-time verification on all offspring individuals generated by the crossover operation according to the constraints; if an offspring individual does not meet the constraints, discard it. Step S5.6.4: Calculate the fitness value of each retained offspring individual; Step S5.6.5: If the number of offspring individuals retained is not less than Then select the one with the highest fitness. Each offspring individual forms the t-th generation population. If the number of offspring individuals retained is less than [a certain number], then [the following is a separate, unrelated statement:] ...an individual, if the number of offspring individuals retained is less than [a certain number]... Then all the retained offspring individuals are considered as individuals in the t-th generation population. At this point, the number of individuals in the t-th generation population is insufficient. From the (t-1)th generation population, excluding the first generation with the highest fitness From the individuals other than the first individual, selection continues in descending order of fitness until the number of individuals in the t-th generation equals the number of individuals in the population. At this point, the t-th generation population is obtained. Step S5.7: Determine if the current value of t is equal to If it equals, then the iteration ends, and the 1st iteration is... The task assignment code and path order code of the individual with the highest fitness value in the population are used as the output of the optimal path planning; if they are not equal, the value of t is updated by adding 1 to the current value of t, and then the process returns to step S5.6 for the next iteration. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the multi-vehicle collaborative meal delivery path planning method of the present invention; Figure 2 This is a schematic diagram illustrating the process of iteratively optimizing the meal preparation path using the improved genetic algorithm described in this invention. Detailed Implementation

[0013] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0014] Example 1: As Figure 1 As shown, a multi-vehicle collaborative meal delivery route planning method suitable for elderly canteens, where the multi-vehicles include meal delivery vehicles and drones, includes the following steps: Step S1: Obtain the planning information and parameters required for multi-vehicle collaborative meal delivery in the elderly canteen. The planning information and parameters include: demand points and elderly canteen information, environmental and temperature parameters, vehicle attribute parameters, and key variables for path planning. Step S2: Construct a multi-objective optimization function with the objectives of minimizing total delivery time, minimizing delivery cost, maximizing on-time meal preparation rate, and maximizing meal temperature compliance rate; Step S3: Based on the characteristics of the vehicle, the requirements of the elderly meal delivery scenario, and the temperature control requirements, construct the constraints for multi-vehicle collaborative meal delivery; Step S4: Based on the multi-objective optimization function in step S2 and the constraints in step S3, the genetic algorithm is improved to obtain an improved genetic algorithm. The improvements include: defining the reciprocal of the multi-objective optimization function as the fitness function of the genetic algorithm, and incorporating the mechanism for verifying and processing the constraints constructed in step S3 into the crossover and mutation operations of the genetic algorithm. Step S5: Iteratively optimize the meal delivery path using an improved genetic algorithm to obtain the global optimal solution, which is the result of the collaborative meal delivery path planning between the drone and the meal delivery vehicle.

[0015] In this embodiment, firstly, at the modeling level, a multi-objective optimization function is constructed with the objectives of minimizing total delivery time, minimizing delivery cost, maximizing on-time delivery rate, and maximizing food temperature compliance rate. This ensures parallel optimization of multiple key performance indicators from the perspective of optimization objectives. Secondly, at the constraint level, vehicle characteristics, requirements of elderly meal delivery scenarios, and temperature control needs are incorporated. Finally, at the solution level, by setting the reciprocal of the above multi-objective function as the fitness function and embedding the constraint verification mechanism into the crossover and mutation operations of the genetic algorithm, the genetic algorithm can continuously select feasible solutions that simultaneously satisfy all constraints and are balanced and superior in each objective during iteration. Thus, this technical closed loop of "multi-objective modeling - physical constraint embedding - algorithm synchronous optimization" enables the present invention to obtain a drone and meal delivery vehicle collaborative delivery solution that performs well in multiple dimensions such as time, cost, on-time delivery, and temperature preservation in a single solution. This can meet the comprehensive needs of elderly meal assistance scenarios for the collaborative optimization of delivery efficiency, cost, and food temperature preservation quality.

[0016] Example 2: This example is basically the same as Example 1, except that in this example, the planning information and parameters in step S1 include: (a) Demand Point Information and Senior Citizens' Dining Hall Information: Demand point information includes the total number of demand points N, the number of demand points, and the number of demand points N. i Geographic coordinates of each demand point Meal demand and expected delivery time window ,in, , The upper limit of the expected delivery time, Minimum expected delivery time; information about the senior citizens' dining hall includes the geographical coordinates of the dining hall's location. Meal preparation completion time and the initial temperature of the food ; (ii) Environmental and temperature parameters: including ambient temperature (Unit: °C) Food temperature decay coefficient (unit: ) and preset minimum safe temperature for meals ; (III) Vehicle attribute parameters: including the total number of vehicles M (unit: vehicles), of which the number of catering trucks is... (Unit: vehicles), the number of drones is (Unit: vehicles), satisfying M = + Number all vehicles from 1 to M, where The catering trucks are numbered 1 to , The drone's serial number is +1 to M, numbered The vehicle is called a vehicle. , = 1, 2, ..., M; Vehicle The attributes include: rated load capacity (Unit: kg) Movement speed (Unit: m / s), Energy cost per unit mileage Maximum driving range (Unit: m); For catering carts, the thermal insulation performance coefficient is also included. , For drones, this also includes the drone's maximum delivery radius. (Unit: m) and thermal insulation performance coefficient , ; (iv) Key variables for path planning: including vehicles For the i Task allocation variables for each demand point Vehicle Demand point access sequence Total driving / flight mileage , No. The actual delivery time and food temperature at each demand point If the vehicle Responsible for the For each demand point, then Otherwise, it is 0.

[0017] Example 3: This example is basically the same as Example 2, except that: in this example, the multi-objective optimization function constructed in step S2... Equation (1) is used to express this as: (1) In equation (1), The weighting coefficient for total delivery time The weighting factor for total delivery cost Weighting coefficients for on-time meal delivery rate The weighting coefficients for the food temperature compliance rate are all greater than 0 and satisfy the following conditions: Its specific value is preset according to the priority of the delivery service; The total delivery time (in seconds) is calculated using the formula shown in equation (2): (2) in, For vehicles Departure time (in seconds); For vehicles To the point of demand The delivery distance (unit: m) is calculated based on the road network distance for food delivery vehicles and the straight-line distance for drones. For vehicles The speed of movement (unit: m / s). For vehicles Total waiting time during the delivery process (in seconds); The total delivery cost (unit: yuan) is calculated using the formula shown in equation (3): (3) The formula for calculating the on-time delivery rate is shown in equation (4): (4) in, Used to represent demand points Is the actual delivery time within If inside, then ,otherwise ; The formula for calculating the food temperature compliance rate is shown in equation (5): (5) in, Used to represent demand points Food delivery temperature Is it greater than or equal to? If so, then ,otherwise ; The calculation formula is shown in equation (6): (6) In equation (6), To be responsible for the demand points i The thermal insulation performance coefficient of a certain vehicle used for food delivery. (Unit: s) represents the distance a vehicle travels from the cafeteria to the point of demand. Travel time for all route segments traversed For a certain vehicle to reach the demand point Previous cumulative waiting time (unit: seconds). e is the base of the natural logarithm.

[0018] Example 4: This example is basically the same as Example 3, except that: in this example, the constraints constructed in step S3 include load constraints, endurance constraints, drone delivery radius constraints, time window constraints, task allocation constraints, and food temperature constraints.

[0019] Example 5: This example is basically the same as Example 4, except that: in this example, the load constraint, endurance constraint, UAV delivery radius constraint, time window constraint, task allocation constraint, and food temperature constraint are as shown in equations (7) to (12) respectively: (7) (8) (9) (10) (11) (12) in, Indicate demand points Actual delivery time (unit: seconds).

[0020] Example 6: This example is basically the same as Example 5, except that: in this example, as Figure 2 As shown, the specific process of iteratively optimizing the meal preparation route using the improved genetic algorithm in step S5 is as follows: Step S5.1: Initialize the parameters of the improved genetic algorithm: Set the population size Maximum number of iterations Crossover probability Probability of mutation Number of elite individuals retained ; Step S5.2: Set the population coding rule to integer-type two-layer coding: Each individual is represented by a two-layer coding method. The first layer is the task assignment code, which is a sequence A of length N. The first layer of sequence A is the task assignment code. i The data A[i] represents the demand point. i The assigned vehicle number, i.e., the demand point i The food is assigned to vehicle A[i] for delivery, where A[i]∈{1, 2, ..., M}; the second layer is the path sequence encoding, which is a list R, where the data in the k-th column of list R is the data assigned to the vehicle. An ordered sequence of accesses to all demand points, i.e., the vehicle Demand point access sequence ; Step S5.3: Initial population generation: Under the constraints, randomly generate... Individual task assignment codes are generated, and based on these codes, all assigned demand points for each vehicle are randomly arranged to generate a path sequence code. The currently generated code is then used to... Individuals constitute the 0th generation population, i.e., the initial population; Step S5.4: Calculate the fitness value of each individual in the 0th generation population using equation (13): (13) Among them, the higher the fitness value, the better the overall performance of the individual; Step S5.5: Set the iteration variable t and initialize t to 1; Step S5.6: Perform the t-th iteration to obtain the t-th generation population. The specific process is as follows: Step S5.6.1: Perform the selection operation: First, select the generation with the highest fitness from the (t-1)th generation population. Each individual is directly included in the t-th generation population. Individuals are selected to preserve superior genes; then, a tournament selection process is conducted: each time, two individuals are randomly selected from the (t-1)th generation population, and the one with higher fitness is selected for the mating pool. This process is repeated until the number of individuals in the mating pool reaches a certain threshold. This ensures the direction of evolution while maintaining population diversity; Step S5.6.2: Perform crossover and mutation operations on the individuals in the mating pool: Step S5.6.2.1: Take the individuals in the mating pool as parent individuals and randomly pair up the parent individuals; Step S5.6.2.2: For each pair of parent individuals, generate a random number in the interval [0, 1) for that pair of parent individuals. If this number is greater than or equal to... If the condition is met, then the crossover operation is not performed on the parent individuals; otherwise, the crossover operation is performed. The specific process of the crossover operation is as follows: The parent individuals are referred to as Parent Individual 1 and Parent Individual 2. First, a two-point crossover is performed on the first-level task allocation code: two different crossover points are randomly selected, and the vehicle number sequences of Parent Individual 1 and Parent Individual 2 between these two crossover points are swapped to generate the first-level task allocation codes for Child Individual 1 and Child Individual 2. Then, a sequential crossover is performed on the second-level path sequence code: a vehicle is randomly selected, and the demand point access sequences of that vehicle in Parent Individual 1 and Parent Individual 2 are obtained respectively, denoted as sequences S1 and S2. Two different position indices are randomly selected in sequence S1, and all elements between these two indices are defined as the reserved fragment FG. The process of constructing the child individual 1's code for that vehicle is then performed. The demand point access sequence for the vehicle follows these rules: First, all demand point numbers in the reserved fragment FG are placed in their original order. Then, starting from the beginning of the demand point access sequence for the vehicle in sequence S2, the demand point numbers in the demand point access sequence for the vehicle in S2 that are not in F are appended to the demand point access sequence for the offspring individual one for that vehicle in the order in which they appear in S2. The demand point access sequence for the offspring individual two for the vehicle is constructed using a symmetric rule, that is, the fragment in S2 is retained and filled in the order of S1. For other vehicles that are not randomly selected, the offspring individual one directly inherits the corresponding demand point access sequence of the parent individual one, and the offspring individual two directly inherits the corresponding demand point access sequence of the parent individual two. Step S5.6.2.3: For all offspring individuals obtained in step S5.6.2.2, generate a random number in the interval [0, 1) for each offspring individual. If this random number is greater than or equal to... If the result is positive, the mutation operation will not be performed; otherwise, the mutation operation will be performed to update the device. The specific process of mutation operation is as follows: In the first-level task allocation coding of the offspring individual, a demand point is randomly selected and reassigned to another vehicle that meets its vehicle type and does not violate the constraints after allocation; In the second-level path sequence coding, a demand point access sequence of a vehicle is randomly selected and its local path is perturbed. The local path perturbation is as follows: reverse the access order of a continuous interval in the demand point access sequence, swap the positions of two demand points in the demand point access sequence, or insert a demand point in the demand point access sequence into another random position; Step S5.6.3: Perform immediate verification on all offspring individuals generated by the crossover operation according to the constraints; if an offspring individual does not meet the constraints, discard it. Step S5.6.4: Calculate the fitness value of each retained offspring individual; Step S5.6.5: If the number of offspring individuals retained is not less than Then select the one with the highest fitness. Each offspring individual forms the t-th generation population. If the number of offspring individuals retained is less than [a certain number], then [the following is a separate, unrelated statement:] ...an individual, if the number of offspring individuals retained is less than [a certain number]... Then all the retained offspring individuals are considered as individuals in the t-th generation population. At this point, the number of individuals in the t-th generation population is insufficient. From the (t-1)th generation population, excluding the first generation with the highest fitness From the individuals other than the first individual, selection continues in descending order of fitness until the number of individuals in the t-th generation equals the number of individuals in the population. At this point, the t-th generation population is obtained. Step S5.7: Determine if the current value of t is equal to If it equals, then the iteration ends, and the 1st iteration is... The task assignment code and path order code of the individual with the highest fitness value in the population are used as the output of the optimal path planning; if they are not equal, the value of t is updated by adding 1 to the current value of t, and then the process returns to step S5.6 for the next iteration.

[0021] To verify the performance of the multi-vehicle collaborative meal delivery route planning method for elderly canteens of the present invention, simulation verification was conducted in the following experimental environment: Windows 10 operating system, 16GB RAM, Intel(R) Core(TM) i5-12500H CPU, Python 3.8 programming language, and the SUMO (Simulation of Urban MObility) traffic simulation platform were used to construct the multi-vehicle collaborative meal delivery route planning method for elderly canteens of the present invention. This method was then compared with the traditional vehicle delivery mode, i.e., the benchmark method in which delivery vehicles sequentially complete all order deliveries along a fixed or optimized route connecting all customer points. The simulation results are shown in Table 1. Table 1 Analysis of the data in Table 1 shows that, compared with the traditional delivery model, the multi-vehicle collaborative meal delivery route planning method for elderly canteens of the present invention exhibits significant advantages in all three evaluation indicators: travel time is reduced by approximately 29.8%, indicating that the multi-vehicle collaborative meal delivery route planning method for elderly canteens of the present invention can effectively shorten the longest delivery time and improve overall timeliness; the total travel distance is reduced by approximately 14.5%, reflecting the effect of the multi-vehicle collaborative meal delivery route planning method for elderly canteens of the present invention in terms of mileage saving; and transportation costs are reduced by approximately 25.7%, verifying the superiority of the multi-vehicle collaborative meal delivery route planning method for elderly canteens of the present invention in terms of energy consumption and operating cost control.

[0022] The simulation results above demonstrate that the multi-vehicle collaborative meal delivery route planning method of the present invention, applicable to elderly canteens, can simultaneously improve delivery timeliness, save driving mileage, and reduce transportation costs in a real traffic simulation environment, fully meeting the collaborative optimization needs for delivery efficiency and economy in elderly meal assistance scenarios.

Claims

1. A multi-vehicle collaborative meal delivery path planning method suitable for elderly canteens, characterized in that, The multi-vehicle system includes a food delivery vehicle and a drone, and includes the following steps: Step S1: Obtain the planning information and parameters required for multi-vehicle collaborative meal delivery in the elderly canteen. The planning information and parameters include: demand points and elderly canteen information, environmental and temperature parameters, vehicle attribute parameters, and key variables for path planning. Step S2: Construct a multi-objective optimization function with the objectives of minimizing total delivery time, minimizing delivery cost, maximizing on-time meal preparation rate, and maximizing meal temperature compliance rate; Step S3: Based on the characteristics of the vehicle, the requirements of the elderly meal delivery scenario, and the temperature control requirements, construct the constraints for multi-vehicle collaborative meal delivery; Step S4: Based on the multi-objective optimization function in step S2 and the constraints in step S3, the genetic algorithm is improved to obtain an improved genetic algorithm. The improvement includes: defining the reciprocal of the multi-objective optimization function as the fitness function of the genetic algorithm, and incorporating the mechanism for verifying and processing the constraints constructed in step S3 into the crossover and mutation operations of the genetic algorithm. Step S5: Iteratively optimize the meal delivery path using the improved genetic algorithm to obtain the global optimal solution, which is the result of the collaborative meal delivery path planning between the UAV and the meal delivery vehicle.

2. The multi-vehicle collaborative meal delivery path planning method for elderly canteens according to claim 1, characterized in that, In step S1, the planning information and parameters include: (I) Demand Point Information and Senior Citizens' Dining Hall Information: The demand point information includes the total number of demand points N (unit: points), the first demand point, and the second demand point. i Geographic coordinates of each demand point Meal demand and expected delivery time window ,in, , The upper limit of the expected delivery time (unit: seconds). Expected delivery time limit (unit: seconds); the information about the senior citizens' canteen includes the geographical coordinates of the canteen's location. Meal preparation completion time (Unit: s) and initial temperature of the food (Unit: °C) (ii) Environmental and temperature parameters: including ambient temperature (Unit: °C) Food temperature decay coefficient (unit: ) and preset minimum safe temperature for meals (Unit: °C) (III) Vehicle attribute parameters: including the total number of vehicles M (unit: vehicles), of which the number of catering trucks is... (Unit: vehicles), the number of drones is (Unit: vehicles), satisfying M = + Number all vehicles from 1 to M, where The catering trucks are numbered 1 to , The drone's serial number is +1 to M, numbered The vehicle is called a vehicle. , = 1, 2, ..., M; Vehicle The attributes include: rated load capacity (Unit: kg) Movement speed (Unit: m / s), Energy cost per unit mileage Maximum driving range (Unit: m); For catering carts, the thermal insulation performance coefficient is also included. , For drones, this also includes the drone's maximum delivery radius. and thermal insulation performance coefficient , ; (iv) Key variables for path planning: including vehicles For the i Task allocation variables for each demand point Vehicle Demand point access sequence Total driving / flight mileage , No. The actual delivery time and food temperature at each demand point If the vehicle Responsible for the For each demand point, then Otherwise, it is 0.

3. The multi-vehicle collaborative meal delivery path planning method for elderly canteens according to claim 2, characterized in that, The multi-objective optimization function constructed in step S2 Equation (1) is used to express this as: （1） In equation (1), The weighting coefficient for total delivery time The weighting factor for total delivery cost Weighting coefficients for on-time meal delivery rate The weighting coefficients for the food temperature compliance rate are all greater than 0 and satisfy the following conditions: Its specific value is preset according to the priority of the delivery service; The total delivery time is calculated using the formula shown in equation (2): (2) in, For vehicles Departure time; For vehicles To the point of demand The delivery distance is calculated based on the road network distance for food delivery vehicles and the straight-line distance for drones. For vehicles movement speed, For vehicles Total waiting time during the delivery process; The total delivery cost is calculated using the formula shown in equation (3): (3) The formula for calculating the on-time delivery rate is shown in equation (4): (4) in, Used to represent demand points Is the actual delivery time within If inside, then ,otherwise ; The formula for calculating the food temperature compliance rate is shown in equation (5): (5) in, Used to represent demand points Food delivery temperature Is it greater than or equal to? If so, then ,otherwise ; The calculation formula is shown in equation (6): (6) In equation (6), To be responsible for the demand points i The thermal insulation performance coefficient of a certain vehicle used for food delivery. This indicates the distance a vehicle travels from the cafeteria to the point of demand. Travel time for all route segments traversed For a certain vehicle to reach the demand point Previous cumulative waiting time, e is the base of the natural logarithm.

4. The multi-vehicle collaborative meal delivery path planning method for elderly canteens according to claim 3, characterized in that, The constraints constructed in step S3 include load constraints, endurance constraints, drone delivery radius constraints, time window constraints, task allocation constraints, and food temperature constraints.

5. The multi-vehicle collaborative meal delivery path planning method for elderly canteens according to claim 4, characterized in that, The load constraint, the endurance constraint, the drone delivery radius constraint, the time window constraint, the task allocation constraint, and the food temperature constraint are shown in equations (7) to (12) respectively: (7) (8) (9) (10) (11) (12) in, Indicate demand points The actual delivery time.

6. The multi-vehicle collaborative meal delivery path planning method for elderly canteens according to claim 5, characterized in that, The specific process of iteratively optimizing the meal preparation path using the improved genetic algorithm in step S5 includes: Step S5.1: Set algorithm parameters, including population size, maximum number of iterations, crossover probability, mutation probability, and number of elite individuals retained; Step S5.2: Individuals are represented by an integer-type two-layer coding method, where the first layer of coding represents the vehicle number assigned to each demand point, and the second layer of coding represents the sequential order of each vehicle accessing the demand point; Step S5.3: Under the premise of satisfying the above constraints, randomly generate an initial population; Step S5.4: Calculate the fitness value of each individual based on the fitness function; Step S5.5: Iteratively update the population through selection, crossover, and mutation operations, wherein: the selection operation adopts a combination of elite retention and tournament selection; the crossover operation includes performing two-point crossover on the task assignment code and performing sequential crossover on the path order code; the mutation operation includes vehicle reassignment on the task assignment code and local path perturbation on the path order code; during the crossover and mutation process, the constraints are checked in real time, and individuals that do not meet the constraints are discarded. Step S5.6: Repeat steps S5.4 to S5.5 until the maximum number of iterations is reached, and output the individual with the highest fitness as the optimal path planning scheme.

Images