CN114186924B

CN114186924B - Collaborative distribution path planning method and device, electronic equipment and storage medium

Info

Publication number: CN114186924B
Application number: CN202111416765.XA
Authority: CN
Inventors: 孙智彬; 梁爽; 陈彦如; 李秀燕
Original assignee: Beijing Sinoiov Vehicle Network Technology Co ltd
Current assignee: Beijing Sinoiov Vehicle Network Technology Co ltd
Priority date: 2021-11-25
Filing date: 2021-11-25
Publication date: 2023-07-25
Anticipated expiration: 2041-11-25
Also published as: CN114186924A

Abstract

The invention provides a collaborative distribution path planning method, a collaborative distribution path planning device, electronic equipment and a storage medium, wherein the collaborative distribution path planning method comprises the following steps: acquiring distribution information; establishing a collaborative distribution model according to the distribution information; acquiring rural road punishment cost, and processing the distribution information according to the rural road punishment cost; according to the processed distribution information and the collaborative distribution model, an initial solution algorithm is applied, and an initial solution of a collaborative distribution path is calculated; and iterating the initial solution by using an optimization algorithm to obtain an optimal solution of the collaborative distribution path. In this way, in the collaborative distribution path, part of client points are served by the rural supermarket collection point, and rural road punishment cost is introduced on the basis of rural scene road information, so that the road sections where trucks are not easy to pass are replaced by unmanned aerial vehicles, the overall distribution cost is reduced, and the distribution efficiency is greatly improved.

Description

Collaborative distribution path planning method and device, electronic equipment and storage medium

Technical Field

The present invention relates to the field of path planning technologies, and in particular, to a collaborative distribution path planning method, device, electronic equipment, and storage medium.

Background

With the development of modern agriculture, the living standard of farmers is improved, and the effect of rural logistics is increasingly revealed. Compared with other areas, the rural areas have the characteristics of poor road conditions, wide areas, more logistics main bodies, small scale and the like. Based on cost constraints, rural areas are currently generally distributed in segmented blocks by a truck in combination with a proxy for pickup, which is inefficient.

With the development of technology, in recent years, the co-delivery of unmanned aerial vehicles and trucks is attracting attention, and a certain research result is also being formed. Because the unmanned aerial vehicle has no requirement on the terrain, the unmanned aerial vehicle has high adaptability to rural scenes naturally, and the truck serves as a mobile warehouse for the unmanned aerial vehicle, so that the cruising ability of the unmanned aerial vehicle can be improved; therefore, each truck is provided with an unmanned aerial vehicle, and the distribution efficiency of the unmanned aerial vehicle can be greatly improved. Unfortunately, unmanned aerial vehicle-truck collaborative distribution research for rural logistics is few at present, and special situations of rural scenes cannot be comprehensively considered.

At present, the research in rural areas only uses an unmanned aerial vehicle-truck collaborative distribution scheme in urban areas to carry out path planning, so that distribution efficiency is lower.

Disclosure of Invention

The invention solves the problems that the present rural area only uses the unmanned aerial vehicle-truck collaborative distribution scheme of urban areas to carry out path planning, the consideration of rural specific scenes is lacked, and the distribution efficiency is low.

In order to solve the above-mentioned problems, the present invention first provides a collaborative distribution path planning method, which includes:

acquiring distribution information, wherein the distribution information at least comprises: rural scene road information, customer point information, truck information, unmanned aerial vehicle information and collection point information;

establishing a collaborative distribution model according to the distribution information;

acquiring rural road punishment cost, and processing the distribution information according to the rural road punishment cost;

according to the processed distribution information and the collaborative distribution model, an initial solution algorithm is applied, and an initial solution of a collaborative distribution path is calculated;

and iterating the initial solution by using an optimization algorithm to obtain an optimal solution of the collaborative distribution path.

In this way, in the collaborative distribution path, part of customer points are served by the rural supermarket collection agent, and based on rural scene road information, rural road punishment cost is introduced, and delivery service is replaced by unmanned aerial vehicles on a road section where trucks are not easy to pass, so that the overall distribution cost is reduced, and the distribution efficiency is greatly improved.

Preferably, establishing a collaborative distribution model according to the distribution information includes:

determining an objective function of the collaborative distribution;

Determining constraints of customer points, trucks, unmanned aerial vehicles and alternative collection points; determining a time constraint of a joint point of the unmanned aerial vehicle and the truck in the distribution network;

and establishing the collaborative distribution model according to the constraint and the objective function.

Preferably, according to the processed delivery information and the collaborative delivery model, an initial solution algorithm is applied to calculate an initial solution of a collaborative delivery path, including:

distributing the client points in the range of the substituted receiving points to obtain opened substituted receiving points and unassigned client points;

constructing an initial truck delivery route by using a greedy insertion algorithm for the turned-on pickup points and unassigned clients;

dividing client points into a point set to be allocated for unmanned aerial vehicle service according to a cost comparison principle and unmanned aerial vehicle bearing capacity;

determining unmanned aerial vehicle service clients, routes and launching points and landing points of the unmanned aerial vehicle based on a principle that the mutual waiting time of the trucks and the unmanned aerial vehicles is shortest, a limiting condition of the maximum flight capacity of the unmanned aerial vehicle, and the front-back sequence of the launching points and the landing points of the unmanned aerial vehicle;

and for the client of the unmanned aerial vehicle, which is concentrated in the points to be distributed and does not meet the limiting condition of the unmanned aerial vehicle when the unmanned aerial vehicle route is constructed, reinserting the points to be distributed into the truck delivery route by using a greedy algorithm, so as to obtain the initial solution of the collaborative delivery route.

Preferably, the iterating the initial solution using an optimization algorithm to obtain an optimal solution of the collaborative distribution path includes:

selecting a destructive operator and a repair operator according to the weight, wherein the destructive operator is a client point destructive operator or a substitute point destructive operator;

sequentially executing a destructive operator and a repair operator, wherein in the execution of the destructive operator, a collecting point destructive operator is executed in a specified iteration period, and client point destructive operators are executed in the rest iteration periods; updating an operator score according to the obtained quality of the new solution after each iteration period is finished and updating the new solution according to the simulated annealing rule;

each iteration, recording the total iteration times, the iteration times of using a client point damage operator, the update weight iteration times and the new solution quality non-improvement iteration times;

if the total iteration number is smaller than a first threshold value, re-executing the operator selected according to the weight;

if the iteration times of the client point damage operators are smaller than a second threshold value, selecting the client point damage operators, otherwise, selecting the substitute point damage operators, and resetting the iteration times of the client point damage operators to 0; if it is

If the number of the updating weight iterations is equal to a third threshold, updating the weight of the operator according to the operator score, resetting the number of the updating weight iterations to 0 and returning to the operator according to the weight selection;

If the new solution quality is not improved and the iteration number is equal to a fourth threshold value, executing all local search operators in the local search operator set, and returning to the process of selecting the destruction and repair operators according to the weight after the execution is finished;

and stopping iteration and outputting the current solution as an optimal solution if the total iteration number is equal to a first threshold or the new solution quality is not improved and the iteration number is equal to a fifth threshold.

Preferably, the optimization algorithm stopping condition is: the total iteration times are equal to a first threshold value, or the new solution quality is not improved, the iteration times are equal to a fifth threshold value, and one of the two is satisfied.

Preferably, the condition of using the local search operator is that the new solution quality does not increase the number of iterations equal to the fourth threshold.

Preferably, the local search operator set includes: unmanned aerial vehicle-truck exchange operator, take over point-unmanned aerial vehicle exchange operator, unmanned aerial vehicle emission-landing point exchange converter, two-point exchange operator.

Next, a collaborative distribution path planning apparatus is provided, which includes:

an information acquisition unit configured to acquire delivery information including at least: rural scene road information, customer point information, truck information, unmanned aerial vehicle information and collection point information;

A model building unit for building a collaborative distribution model according to the distribution information;

an information processing unit for acquiring rural road punishment costs and processing the distribution information according to the rural road punishment costs;

an initial solution calculation unit, configured to apply an initial solution algorithm to calculate an initial solution of a collaborative distribution path according to the processed distribution information and the collaborative distribution model;

and the optimizing unit is used for iterating the initial solution by using an optimizing algorithm to obtain an optimal solution of the collaborative distribution path.

Again, there is provided an electronic device comprising a computer readable storage medium storing a computer program and a processor, which when read and executed by the processor, implements a collaborative distribution path planning method as described above.

Finally, a computer readable storage medium is provided, which stores a computer program which, when read and run by a processor, implements a collaborative distribution path planning method as described above.

In this way, in the collaborative distribution path, part of client points are served by the rural supermarket collection point, and rural road punishment cost is introduced on the basis of rural scene road information, so that the road sections where trucks are not easy to pass are replaced by unmanned aerial vehicles, the overall distribution cost is reduced, and the distribution efficiency is greatly improved.

Drawings

FIG. 1 is a flow chart of a collaborative distribution path planning method according to an embodiment of the present invention;

FIG. 2 is a flowchart of a collaborative distribution path planning method S20 according to an embodiment of the present invention;

FIG. 3 is a flowchart of a collaborative distribution path planning method S40 according to an embodiment of the present invention;

FIG. 4 is a flowchart of a collaborative distribution path planning method S50 according to an embodiment of the present invention;

FIG. 5 is a block diagram illustrating a cooperative distribution path planning apparatus according to an embodiment of the present invention;

fig. 6 is a block diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order that the above objects, features and advantages of the invention will be readily understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.

The embodiment of the application provides a collaborative distribution path planning method, which can be executed by a collaborative distribution path planning device, wherein the collaborative distribution path planning device can be integrated in electronic equipment such as a computer, a server, a computer and the like. FIG. 1 is a flow chart of a collaborative distribution path planning method according to an embodiment of the present invention; the collaborative distribution path planning method comprises the following steps:

s10, obtaining distribution information, wherein the distribution information at least comprises: rural scene road information, customer point information, truck information, unmanned aerial vehicle information and collection point information;

the obtained delivery information may be data in a standard computing example (during experiment or verification), or may be data actually collected; the rural scene road data, the position information and the like can be obtained by crawling data through a map; corresponding delivery information can also be obtained through the GPS data of the vehicle.

S20, establishing a collaborative distribution model according to the distribution information;

Wherein the collaborative distribution model may be an operations study model, which may be composed of constraints and objective functions for a plurality of actual conditions;

s30, acquiring rural road punishment cost, and processing the distribution information according to the rural road punishment cost;

the rural road punishment cost can be determined according to historical data or map information, and can also be comprehensively determined according to other actual conditions; it can also be derived by cyclic feedback.

And processing the distribution information to apply unit punishment cost to each client point according to the position of the client point.

Preferably, the unit road penalty cost is sequentially increased outwards by the geometric center of the client point set until 1.85 times of the travel cost; the cost of the road penalty for between client points is road length x 1/2.

Among them, because the topography of most rural areas is: the middle part is village and town, and the periphery is suburban or mountain area, so the data processing is based on the following modes: the unit road penalty cost is sequentially increased outward by the geometric center of the client point set to 1.85 times the travel cost. That is, the unit road penalty cost of the client point at the town center is close to 0, and the unit road penalty cost of the client point at the mountain area is close to 1.85 times the travel cost.

The road punishment cost between the client points is the road length multiplied by 1/2 (the unit road punishment cost of the client point i and the unit road punishment cost of the client point j), so that the road condition of the client is sequentially excessive from good road condition to poor road condition.

S40, according to the processed distribution information and the collaborative distribution model, an initial solution algorithm is applied, and an initial solution of a collaborative distribution path is calculated;

and S50, iterating the initial solution by using an optimization algorithm to obtain an optimal solution of the collaborative distribution path.

The steps S40-S50 are adaptive large-scale neighborhood search algorithms, that is, after an initial solution is obtained by using an initial algorithm, the initial solution is input into the adaptive large-scale neighborhood search algorithm to be optimized.

In this way, in the collaborative distribution path, part of customer points are served by the rural supermarket collection agent, and based on rural scene road information, rural road punishment cost is introduced, and delivery service is replaced by unmanned aerial vehicles on a road section where trucks are not easy to pass, so that the overall distribution cost is reduced, and the distribution efficiency is greatly improved. Preferably, as shown in fig. 2, the step S20 of establishing a collaborative distribution model according to the distribution information includes:

S21, determining an objective function of collaborative distribution;

the objective function of the collaborative distribution can be divided into four parts: the first part is the travel cost of the truck; the second part is the travel cost of the unmanned aerial vehicle; the third part is punishment cost of the truck running under different road conditions in rural areas; and the fourth part is the operation cost of the collection point.

Wherein, penalty cost can be understood as: when the road condition of the truck to a certain customer point is poor, the fuel consumption of the truck is increased due to the increase of turning and braking times, and penalty cost is applied to the travel of the truck. Due to the addition of punishment cost, unmanned aerial vehicles are selected to carry out distribution service under the action of a minimized target under the condition of poor road conditions.

Preferably, the point of sale operating cost comprises two sub-parts: one part is fixed cost and the other part is compensation cost, here it is assumed that the compensation cost is proportional to the distance of the customer from the point of collection.

The compensation cost can be understood as: when the customer is far away from the point of collection (still within the service range of the point of collection), the customer needs to spend more effort for self-lifting, and a certain compensation cost is paid to the customer, so that the customer can go to the point of collection for self-lifting.

S22, determining constraints of customer points, trucks, unmanned aerial vehicles and alternative collection points; determining a time constraint of a truck and unmanned aerial vehicle joint in a distribution network;

preferably, the constraining portion may comprise four portions, respectively: constraints on customer points, constraints on trucks, constraints on drones (this section includes constraints on truck-drone junctions), and constraints on proxy destinations.

Preferably, the constraints on the client points are: each client must be accessed only once, and the access modes can be selected as follows: the truck, the unmanned aerial vehicle and the collection point are self-lifting.

Preferably, the constraints on the truck are: requiring the truck to accurately leave and return to the distribution center once; avoiding the repeated running of the truck on a certain road section and eliminating the constraint of a truck sub-loop; since trucks must leave a client point after accessing the client point to continue to participate in the distribution system, constraints require that trucks at the access point must leave the point.

Preferably, the constraints on the drone are: because the truck plays a role of a mobile warehouse of the unmanned aerial vehicle while participating in distribution, the unmanned aerial vehicle can only emit from the point i when the truck accesses the point i and the point k, and the unmanned aerial vehicle can be recovered by the truck after the truck accesses the point j and drops at the point k; the special condition that the unmanned aerial vehicle transmits from the warehouse, the constraint on the unmanned aerial vehicle route is the same as the previous constraint mode; the launching and landing of the unmanned aerial vehicle have a sequence, and the launching of the unmanned aerial vehicle required to travel for a certain time is constrained to be required before landing; requiring that the unmanned aerial vehicle can be launched and landed from any node at most once; since the truck serves as a mobile warehouse, the unmanned aerial vehicle needs to be launched from the truck and dropped onto the truck, so that the access sequence of the truck needs to be constrained, and the truck needs to access the unmanned aerial vehicle launching point first and then the unmanned aerial vehicle drop point.

Preferably, the constraints on the proxy harvest are: only when the customer is served by the proxy recipient, the proxy recipient can be turned on, in order to avoid special situations: the proxy receiving point serves as a transmitting point or a landing point of the unmanned aerial vehicle under the condition of not serving the client point; the client can carry out self-lifting to the collection point, and can carry out self-lifting to the collection point only when the client is in the service range of the opened collection point; only when the truck passes the pickup point, the pickup point can be started, namely, the pickup point can provide service for the customer only after the truck is delivered to the pickup point; the package carrying capacity of the proxy receipts is limited, and the sum of the demands of the clients of the proxy receipts service cannot exceed the maximum capacity of the proxy receipts.

Preferably, the time constraint of the distribution network can be further included, and since the unmanned aerial vehicle and the truck perform cooperative distribution, the truck needs to retract the unmanned aerial vehicle for completing the distribution within a specified time (i.e. meeting the flight capability of the unmanned aerial vehicle), so that the time required for the unmanned aerial vehicle to travel in the distribution network, i.e. the time taken by the unmanned aerial vehicle to launch from the truck to the unmanned aerial vehicle to drop onto the truck, needs to be calculated. The content that this part needs to calculate is: in the unmanned plane path, the unmanned plane reaches the time of the transmitting point, reaches the time of the customer point of service, reaches the time of the falling point; and the time when the truck arrives at the launching point in the unmanned aerial vehicle path, and the time when the truck leaves the launching point and falls off the unmanned aerial vehicle; the unmanned aerial vehicle and the truck have the same departure time at the launching point because the truck immediately goes to the next access point while the unmanned aerial vehicle launches on the truck; the truck leaves the drop point together with the unmanned aerial vehicle after receiving the unmanned aerial vehicle, so the time for the truck to leave the drop point is the same as the time for the unmanned aerial vehicle to leave the drop point; using the time obtained above, the time spent by the drone from the launch to the service client point and the time spent by the landing from the service client point in the drone path can be calculated; after the two times are obtained, the unmanned aerial vehicle flight capacity can be constrained by combining the weight of the unmanned aerial vehicle battery and the effective load: this is because the power consumed by the drone is approximately linear with battery weight and payload. Only when the energy consumption required for a single trip is within the flight capabilities of the drone will the drone perform the trip. The node at which the unmanned aerial vehicle finishes traveling is that the unmanned aerial vehicle falls on a truck, because if the unmanned aerial vehicle reaches a falling point first, the time of the unmanned aerial vehicle waiting for hovering needs to be calculated;

Only when the unmanned aerial vehicle is recovered at the landing point, the unmanned aerial vehicle can be arranged for the next customer service, so that the order of the last trip landing point and the next trip launching point of the unmanned aerial vehicle needs to be restrained; the package bearing capacity of the unmanned aerial vehicle is limited, and the unmanned aerial vehicle can serve the client point only when the demand of the client point is smaller than the maximum load of the unmanned aerial vehicle, and the unmanned aerial vehicle cannot meet the bearing capacity of the unmanned aerial vehicle and is served by a truck or a collecting point.

S23, establishing the collaborative distribution model according to constraint and objective function;

wherein the collaborative distribution model may be derived from a combination of constraints and objective functions.

In this way, a collaborative distribution model is established through an objective function and constraint, so that the problem of collaborative distribution of the truck and the unmanned vehicle is scientifically described; penalty costs of different road conditions are introduced into the objective function, so that the matching degree of the whole model and actual distribution is greatly increased; by introducing constraints such as road conditions and the like into the constraints, the matching degree of the model is further improved, and the accuracy of path planning is greatly improved.

Preferably, the objective function is:

wherein, the related model symbols and variables are defined as follows:

Preferably, as shown in fig. 3, S40, according to the processed delivery information and the collaborative delivery model, an initial solution algorithm is applied to calculate an initial solution of a collaborative delivery path, including:

s41, distributing the client points in the substituted receiving point range to the substituted receiving point; obtaining an opened generation receiving point and an unassigned client point;

and if the receiving point is provided with the receiving point and the client is not served, closing the receiving point after the step is finished.

The formula is:

wherein p is _ik Representing the probability that client point i is assigned to the proxy receiving point k, r _S Service scope for the substitute receiving point, d _ik N is the distance from the client point i to the collection point k, N is the client point set ^S Is a collection of proxy collection points.

By the step, two types of points of the opened generation and collection points and scattered client points which are not distributed to the generation and collection points can be obtained.

S42, constructing an initial truck delivery route by using a greedy insertion algorithm aiming at the opened pickup points and the unassigned clients;

s43, dividing the client points into a point set to be allocated for unmanned aerial vehicle service according to a cost comparison principle and unmanned aerial vehicle bearing capacity limit;

s44, determining unmanned aerial vehicle delivery clients, routes and launching points and landing points based on the principle that the mutual waiting time of the trucks and the unmanned aerial vehicles is shortest, the limiting condition of the maximum flight capacity of the unmanned aerial vehicles, and the front-back sequence of the launching points and the landing points of the unmanned aerial vehicles;

S45, for the client of the unmanned aerial vehicle service point to be distributed, which does not meet the unmanned aerial vehicle limiting condition when the unmanned aerial vehicle route is built, reinserting the client into the truck delivery route by using a greedy algorithm, and obtaining an initial solution of the collaborative delivery route.

In the steps S43-S45, all scattered client points on the truck route need to be processed, and the service points of the unmanned aerial vehicle are divided, namely, the service points are selected according to the cost comparison principle and the bearing capacity limitation of the unmanned aerial vehicle, and if the cost of service by using the unmanned aerial vehicle is lower, the client points are divided into the point set to be allocated of the unmanned aerial vehicle. After the processing is finished, aiming at the client points to be distributed of each unmanned aerial vehicle, traversing all truck stopping points in sequence, and meeting the following conditions: 1) The driving mileage of the unmanned aerial vehicle does not exceed the maximum flying capacity of the unmanned aerial vehicle; 2) The order of the next launch point of the truck to access the unmanned aerial vehicle must be based on the principle that the waiting time of the truck and the unmanned aerial vehicle is shortest when the truck accesses the previous drop point, so as to determine the unmanned aerial vehicle distribution path. And (3) circulating the point set to be distributed of the unmanned aerial vehicle, and taking the point set to be distributed of the truck route if the client point does not meet the two conditions.

Preferably, if the insertion of the client point to be allocated by the unmanned aerial vehicle at any position of the truck path in the manner of the unmanned aerial vehicle path violates the two conditions, the client point to be allocated is reinserted into the truck path through a greedy insertion algorithm, and the truck provides the delivery service.

In this way, an initial solution of the collaborative distribution path may be obtained, thereby facilitating subsequent iterations.

Preferably, as shown in fig. 4, S50, iterating the initial solution using an optimization algorithm to obtain an optimal solution of the collaborative distribution path, including:

s51, selecting a destructive operator and a repair operator according to the weight, wherein the destructive operator is a client point destructive operator or a substitute point destructive operator;

the client point damage operator is used for damaging the client point, and the collecting point damage operator is used for damaging the collecting point.

Here, the destruction operator is destroyed for the proxy and the client, so that a better destruction effect is achieved, and the destruction can be performed more completely.

Wherein the client point destructive operator is selected from the first destructive operator set; the substitute-receiving point destructive operator is selected from the second destructive operator set, and the repair operator is selected from the repair operator set; each operator has own weight, and a destructive operator (client point destructive operator or substitute point destructive operator) and a repair operator can be randomly selected from the operators according to the weight, wherein the larger the operator weight is, the higher the probability of being selected is.

S52, sequentially executing a destructive operator and a repair operator, wherein in the execution of the destructive operator, the destructive operator of the collecting point is executed in a specified iteration period, and the destructive operators of the client point are executed in the rest iteration periods;

the main framework of the ALNS algorithm is as follows: setting the inner layer iteration times and the outer layer iteration times. Executing a client point damage operator when the inner layer iteration number is smaller than N; and when the inner layer iteration number is equal to N, executing a destructive operator replacing the collection point, and resetting the inner layer iteration number to 0. After the destructive operator is executed, the repair operator is executed, and after the repair operator is executed, the outer layer iteration times are increased by 1.

For the proxy receipt damage operator, as long as the operation of closing the proxy receipt is adopted in the operator, all clients of the closed proxy receipt service will be added into the damage list. The proxy receipts that are turned on in this stage will be inserted into the path of the truck according to the greedy insertion principle, and the client points in the damage list will be reassigned in the repair stage. The corruption for the proxy endpoint may be considered as a combined corruption-repair operator because the opening and closing of the proxy endpoint is not re-operated during subsequent repairs.

After the disruption (for the customer site or the generation site) operation is completed, all elements in the distribution system can be divided into: the method comprises the steps of a damaged truck path, a client point to be allocated to the unmanned aerial vehicle (if a deleted truck service point is just a launch or landing point of the unmanned aerial vehicle, removing the unmanned aerial vehicle path, dividing the unmanned aerial vehicle service client point into a set of points to be allocated to the unmanned aerial vehicle), a damage list, an unopened generation collection point and a generation collection point which is opened in the truck path.

In the repair operator stage, only the client points in the damage list and the points to be distributed of the unmanned aerial vehicle are considered to be inserted into the distribution system, and closing and opening of the collection points are not operated any more.

And S53, updating a new solution according to the rule of simulated annealing after each iteration period is finished, and updating an operator score according to the quality of the new solution.

And receiving a new solution according to a simulated annealing criterion, wherein the scoring rule of the solution is as follows: (a, B, C) = (5, 10, 20). Wherein A, B, C can represent better than the history optimal solution, better than the current solution, worse than the current solution respectively.

S54, recording the total iteration times, the iteration times of using a client point damage operator, the iteration times after updating the weight and the iteration times of which the new solution quality is not improved for each iteration;

the total iteration number may be the outer layer iteration number, and each time the destruction operator (client point destruction operator or proxy point destruction operator) and repair operator are executed, it is considered that one iteration is performed.

S55, if the total iteration times are smaller than a first threshold value, re-executing the operator and the repair operator according to the weight;

s56, if the iteration times of the client point damage operators are smaller than a second threshold, selecting the client point damage operators, otherwise, selecting the collecting point damage operators, and resetting the iteration times of the client point damage operators to 0;

S57, if the update weight iteration number is equal to a third threshold value, updating the weight of the operator according to the operator score, resetting the update weight iteration number to 0 and returning to the operator according to the weight selection;

s58, if the new solution quality is not improved and the iteration number is equal to a fourth threshold value, executing all local search operators in the local search operator set, and returning to the process of selecting the destruction and repair operators according to the weight after the execution is finished;

when the new solution quality is not improved over multiple iteration cycles, it may be a local optimum. In this case, a local search operator is performed, thereby avoiding a locally optimal situation.

And S59, stopping iteration and outputting the current solution as an optimal solution if the total iteration number is equal to a first threshold value or the iteration number with the unrisen new solution quality is equal to a fifth threshold value.

In this way, the initial solution is iteratively optimized through the ALNS algorithm, so as to obtain the current optimal cooperative distribution path.

It should be noted that the optimal solution is only the best solution in the current search range, and is not necessarily the globally optimal solution.

The conditions for performing the local search are: the new solution quality fails to promote the number of iterations equal to a fourth threshold. That is, if the quality of the new solution is not optimized in none of the consecutive fourth threshold iterations, this means that local optimality may occur.

The iteration stop condition is as follows: the total iteration times are equal to a first threshold value, or the iteration times of which the new solution quality cannot be improved are equal to a fifth threshold value, and one of the two thresholds is met.

The second condition for the iteration to stop is that if the quality of the current solution is not optimized in none of the consecutive fifth threshold iterations, there is less likelihood that the solution will continue to improve quality. Or the iteration times are too many, so that the iteration is also forcedly stopped in order to avoid occupying too much calculation time.

Wherein the fifth threshold is greater than the fourth threshold.

Preferably, the second set of destruction operators comprises: randomly opening the destructive operator, randomly closing the destructive operator, closing the destructive operator by the minimum number of clients, opening the destructive operator by the maximum number of clients, and closing the destructive operator by the maximum number of clients.

Randomly starting a destruction operator: and randomly selecting one collection point from the unopened collection points to be opened.

Randomly closing a destruction operator: and randomly selecting one collection point from the opened collection points to close.

Randomly closing the open break operator: one of the turned-on generation-collection points is randomly selected to be turned off, and the other generation-collection point is turned on. The newly opened generation-collection points are not based on random selection, but are based on distance, and if the generation-collection points which are not opened are closer to the generation-collection points which are closed, the generation-collection points are more likely to be opened.

Minimum number of customers off maximum number of customers on destruction operator: and selecting the generation receiving point with the least service clients to close, and adding the service clients thereof into the damage list. And meanwhile, in the unopened generation receiving points, selecting the generation receiving point with the largest number of client points in the service range for opening. The destruction operator may also be referred to as a point of origin swap operator.

Maximum number of customers closing the destruction operator: and selecting the generation receiving point with the largest service client number to close, and adding the service client into the damage list.

Preferably, the first set of destruction operators comprises: random destructive operators, greedy destructive operators with perturbations.

Random destruction operator: the client points are randomly selected for deletion until the number of client point removals is reached.

Greedy destroy operator: the customer points that maximize the total cost are deleted until the customer point removal number is reached.

Greedy destruction operator with perturbation: similar to the greedy destructor, the perturbation coefficients need to be multiplied when calculating the cost. To increase randomness, the perturbation coefficients are random numbers uniformly distributed at [0.7,1 ].

Preferably, the repair operator set includes: the greedy inserts repair operators, two-stage greedy repair operators, two-stage repair operators and greedy punishment repair operators.

Greedy insert repair operator: first, the client points in the damage list are scored. I.e., the score of the customer in the service range of the substituted-receipt point is lower than the score of the customer not in the service range of the substituted-receipt point; the more distant a client point is from the shortest distance of a client of an existing path, the higher the client point score. In the insertion process, the insertion is performed from high to low according to the client score. Three ways are inserted: (a) inserting into a truck path; (b) inserting a drone path; (c) inserting a proxy endpoint service scope; the cost of the three modes is calculated, and the service mode and the insertion position with the lowest cost are selected.

Two-stage greedy repair operator: in the first stage, if the client point in the damage list is in the range of the opened collecting point service, the client is inserted into the collecting point service. Otherwise, the client points are inserted into the truck path according to a greedy strategy. In the second phase, for a customer just inserted into the truck path, if two conditions are satisfied at the same time: (a) unmanned aerial vehicle maximum load bearing requirements and flying ability requirements; (b) The delivery cost of using the drone is lower than the delivery cost of the truck, and it is inserted into the drone path.

Two-stage repair operator: unlike the two-stage greedy repair operator, the (1) client point insertion increases distance multiplied by noise for increased randomness when inserting the client point into the truck path Is in the range of (0.8,1.2). (2) In the second stage, the client point only needs to meet the maximum bearing requirement and the flying capability requirement of the unmanned aerial vehicle, namely, the client point is randomly inserted into the unmanned aerial vehicle path.

Greedy penalty repair operator: the operator first calculates a penalty value for the client points in the destruction list according to the following formula:penalty valueHigh client point priority insertion. In the repair process, the repair is performed according to a two-stage greedy repair operator.

The local search operator set includes two types of operators: local search operators for service style (first three operators), local search operators for truck paths (last two operators).

Unmanned aerial vehicle-truck exchange operator: for all customer points served by the drone, the cost is calculated as they are served by the truck. If the cost of service by the truck is lower than the cost of service by the drone, the drone path is deleted and the customer point is inserted into the truck path. This operator is designed to avoid that the optimal service of the client point of the drone after the truck path is changed is not the drone.

The point-truck swap operator: for all customer points served by the collection point, the cost is calculated as they are served by the truck. If the cost of service by the truck is lower than the cost of service by the point of return, the point is inserted into the truck path. For the case that only one client point is served by the point of origin, when calculating the cost of the client point when served by the point of origin, the fixed operating cost of the point of origin needs to be added. The operator is designed to avoid: the customer point position of the collection point service is in the path from the last point to the collection point of the truck, and the situation that the cost is increased possibly caused by the collection point service isolated point can be avoided.

Collection-unmanned aerial vehicle exchange operator: for all customer points served by the collection point, if they are within the delivery capability of the drone, the cost of their service by the drone is calculated. If the cost of service by the drone is lower than the cost of the point of collection, the customer is added to the drone path.

The unmanned aerial vehicle transmits and falls the point exchange operator, change the access sequence of the truck through exchanging the launching point and landing point of the unmanned aerial vehicle, in order to find a better truck access sequence;

and (3) a two-point exchange operator, wherein any two points except the unmanned aerial vehicle transmitting point and the landing point in the truck path are randomly selected, and the access sequence is exchanged.

In this way, the truck carries the drone and the package from the delivery center, the launch of the drone is performed at the launch point of the drone, and customer points located in environments with poor road conditions are more likely to be served by the drone due to the road penalty costs imposed on the truck. After the unmanned aerial vehicle is launched by the truck, the delivery work is continued, and if the on pickup point exists, the truck needs to send the package to the pickup point. Based on the flying capabilities of the drone, the truck needs to receive the landed drone at the landing point of the drone. This process follows the cost minimization principle.

In order to verify the feasibility of the model and the planning method provided by the invention, experiments were performed, and all codes were programmed with python 3.7. The final experimental data are not shown in detail here, and according to the experimental results, the distribution network model has advantages in cost compared with other collaborative distribution models; compared with other heuristic algorithms, the path planning algorithm has great advantages in time and cost, and most of examples can find an accurate solution in a short time.

The embodiment of the application provides a collaborative distribution path planning device, which is used for executing the collaborative distribution path planning method according to the above content of the invention, and the collaborative distribution path planning device is described in detail below.

As shown in fig. 5, the collaborative distribution path planning apparatus includes:

an information acquisition unit 101 for acquiring delivery information including at least: rural scene road information, customer point information, truck information, unmanned aerial vehicle information and collection point information;

a model building unit 102 for building a collaborative distribution model according to the distribution information;

an information processing unit 103 for acquiring rural road penalty costs and processing the distribution information according to the rural road penalty costs;

an initial solution calculation unit 104, configured to apply an initial solution algorithm to calculate an initial solution of a collaborative distribution path according to the processed distribution information and the collaborative distribution model;

and an optimizing unit 105, configured to iterate the initial solution using an optimizing algorithm, to obtain an optimal solution of the collaborative distribution path.

Preferably, the model building unit 102 is further configured to: determining an objective function of the collaborative distribution; determining constraints of customer points, trucks, unmanned aerial vehicles and alternative collection points; determining a time constraint of a joint point of the unmanned aerial vehicle and the truck in the distribution network; and establishing the collaborative distribution model according to the constraint and the objective function.

Preferably, the initial solution calculation unit 104 is further configured to: distributing the client points in the collection point substitution range to the collection point substitution; obtaining an opened generation receiving point and an unassigned client point; constructing an initial truck route using a greedy insertion algorithm to access on-coming points and unassigned customer points; and dividing the client points into the point sets to be distributed of the unmanned aerial vehicle service according to the cost comparison principle and the unmanned aerial vehicle bearing capacity limit. Based on the principle that the waiting time of the truck and the unmanned aerial vehicle is shortest, the limiting condition of the maximum flight capacity of the unmanned aerial vehicle, the front-back sequence of the launching point and the landing point of the unmanned aerial vehicle, the unmanned aerial vehicle delivery client, the route, the launching point and the landing point of the unmanned aerial vehicle are determined, and the initial solution of the collaborative delivery route is obtained.

Preferably, the initial solution calculation unit 104 is further configured to: if the insertion of the client points of the point set to be distributed by the unmanned aerial vehicle at any position of the truck path in a manner of the unmanned aerial vehicle path violates the flight capacity constraint of the unmanned aerial vehicle, the client points are inserted into the truck path through a greedy algorithm in a manner of truck service.

Preferably, the optimizing unit 105 is further configured to: selecting a destructive operator (client point destructive operator or substitute point destructive operator) and a repairing operator according to the weight; in the iteration, the destruction operator and the repair operator are performed sequentially. In the execution of the destructive operator, the destructive operator of the collecting point is executed in a specified iteration period, and the destructive operators of the client point are executed in the rest iteration periods; updating the new solution according to the rule of simulated annealing after each iteration period is finished, and updating the operator score according to the obtained quality of the new solution. And (3) recording the total iteration times, the iteration times of using the client point damage operators, the iteration times after updating the weights and the iteration times of which the new solution quality is not improved for each iteration. If the total iteration times are smaller than a first threshold value, the destruction and repair operation is continuously executed according to the weight selection operator; and if the number of iterations of using the client point breaking operator is equal to the second threshold, selecting the substitute point breaking operator according to the weight, and resetting the number of iterations of the client point breaking operator to 0. If the iteration number after the weight updating is equal to a third threshold value, updating the weight of the operator, resetting the iteration number after the weight updating and returning to the operator selection according to the weight; if the iteration times of the new solution quality which is not improved is equal to a fourth threshold value, executing all local search operators in the local search operator set, returning to the operator according to the weight selection after executing all the local search operators, and continuing to execute the destruction and repair operation; and stopping iteration and outputting the current solution as an optimal solution if the total iteration number is equal to a first threshold or the iteration number with the new solution quality not improved is equal to a fifth threshold.

Preferably, the information processing unit 103 is further configured to: sequentially increasing unit road punishment cost outwards by using the geometric center of the client point set until the cost is 1.85 times of travel cost; the cost of the road penalty for between client points is road length x 1/2.

Preferably, the optimization algorithm stopping condition is: the total iteration times are equal to a first threshold value, or the iteration times of which the new solution quality is not improved are equal to a fifth threshold value, and one of the two values is satisfied.

Preferably, the condition for using the local search operator is that the number of iterations for which the new solution quality is not improved is equal to the fourth threshold.

Preferably, the first set of destruction operators comprises: randomly opening the destructive operator, randomly closing the destructive operator, closing the destructive operator by the minimum number of clients, opening the destructive operator by the maximum number of clients, and closing the destructive operator by the maximum number of clients.

Preferably, the second set of destruction operators comprises: random destructive operators, greedy destructive operators with perturbations.

Preferably, the client point destruction operator is a destruction operator for a client point.

Preferably, the proxy damage operator is a damage operator for a proxy.

An embodiment of the present application provides an electronic device, as shown in fig. 6, which includes a computer readable storage medium 301 storing a computer program and a processor 302, where the computer program is read and executed by the processor to implement a collaborative distribution path planning method as described above.

Embodiments of the present application provide a computer readable storage medium storing a computer program which, when read and executed by a processor, implements a collaborative distribution path planning method as described above.

The technical solution of the embodiment of the present invention may be embodied in essence or a part contributing to the prior art or all or part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be an air conditioner, a refrigeration apparatus, a personal computer, a server, or a network device, etc.) or processor to perform all or part of the steps of the method of the embodiment of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk, etc.

It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

In this application, all embodiments are described in a related manner, and identical and similar parts of the embodiments are all referred to each other, and each embodiment is mainly described in the differences from other embodiments. For relevance, reference is made to the description of the foregoing embodiments.

Although the present invention is disclosed above, the present invention is not limited thereto. Various changes and modifications may be made by one skilled in the art without departing from the spirit and scope of the invention, and the scope of the invention should be assessed accordingly to that of the appended claims.

Claims

1. A collaborative distribution path planning method, comprising:

establishing a collaborative distribution model according to the distribution information, including:

determining an objective function of the collaborative distribution; the objective function of the collaborative distribution is divided into four parts: the first part is the travel cost of the truck; the second part is the travel cost of the unmanned aerial vehicle; the third part is punishment cost of the truck running under different road conditions in rural areas; the fourth part is the collection-point operation cost, which comprises two sub-parts: one part is fixed cost, and the other part is compensation cost, wherein the compensation cost is compensation cost paid to a self-service customer by a point of sale;

establishing the collaborative distribution model according to the constraints of the client points, the trucks, the unmanned aerial vehicles and the proxy receiving points and/or the time constraints of the combining points of the unmanned aerial vehicles and the trucks in the distribution network and an objective function;

according to the processed delivery information and the collaborative delivery model, an initial solution algorithm is applied to calculate an initial solution of a collaborative delivery path, and the method comprises the following steps:

determining unmanned aerial vehicle delivery clients, routes and launching points and landing points of the unmanned aerial vehicles based on the principle that the mutual waiting time of the trucks and the unmanned aerial vehicles is shortest, the limiting condition of the maximum flight capacity of the unmanned aerial vehicles and the front-back sequence of the launching points and the landing points of the unmanned aerial vehicles;

For the clients of unmanned aerial vehicle service points to be distributed, which are concentrated in the unmanned aerial vehicle route construction process and do not meet the unmanned aerial vehicle limiting conditions, reinserting the points to be distributed into the truck delivery route by using a greedy insertion algorithm, so as to obtain an initial solution of the collaborative delivery route;

iterating the initial solution by using an optimization algorithm to obtain an optimal solution of the collaborative distribution path, including:

sequentially executing a destructive operator and a repair operator, wherein in the execution of the destructive operator, a collecting point destructive operator is executed in a specified iteration period, and client point destructive operators are executed in the rest iteration periods;

setting inner layer iteration times and outer layer iteration times, and executing a client point damage operator when the inner layer iteration times are smaller than a specified threshold value; when the inner layer iteration number is equal to a specified threshold value, executing a destructive operator replacing the collection point, and resetting the inner layer iteration number to 0; after the destructive operator is executed, a repair operator is executed, and after the repair operator is executed, the outer layer iteration times are increased by 1;

updating an operator score according to the obtained quality of the new solution after each iteration period is finished and updating the new solution according to the simulated annealing rule;

if the total iteration number is smaller than a first threshold value, re-executing the operator according to the weight selection;

if the iteration times of the client point damage operators are smaller than a second threshold value, selecting the client point damage operators, otherwise, selecting the substitute point damage operators, and resetting the iteration times of the client point damage operators to 0;

if the update weight iteration number is equal to a third threshold value, updating the weight of the operator according to the operator score, resetting the update weight iteration number to 0 and returning to select the operator according to the weight;

if the new solution quality is not improved and the iteration number is equal to a fourth threshold value, executing all local search operators in the local search operator set, and returning to select a destruction operator and a repair operator according to the weight after the execution is finished;

if the total iteration times are equal to a first threshold value or the new solution quality is not improved, and the iteration times are equal to a fifth threshold value, stopping iteration and outputting the current solution as an optimal solution;

the repair operator is selected from a repair operator set; the repair operator set comprises a greedy insertion repair operator, a two-stage greedy repair operator, a two-stage repair operator and a greedy punishment repair operator;

The greedy insertion repair operator scores the client points in the damage list and inserts the client points into the collaborative distribution path from high to low according to the client scores;

in the first stage, if the client point in the damage list is in the opened service range of the substituted receiving point, inserting the client point into the substituted receiving point service, otherwise, inserting the client point into a truck path according to a greedy strategy; in the second stage, if the client point inserted into the truck path meets the maximum bearing requirement and the flight capability requirement of the unmanned aerial vehicle at the same time, the delivery cost of using the unmanned aerial vehicle is lower than the truck delivery cost, and the client point is inserted into the unmanned aerial vehicle path;

when the two-stage repair operator inserts the first-stage client point into the truck path, the client point inserts an increased distance to multiply noise, and only meets the maximum bearing requirement and the flying capability requirement of the unmanned aerial vehicle in the second-stage client point, and is randomly inserted into the unmanned aerial vehicle path;

and calculating a punishment value of the greedy punishment repair operator for the client points in the damage list, and preferentially inserting the client points with high punishment values into the collaborative distribution path.

2. The collaborative distribution path planning method of claim 1, wherein the set of local search operators comprises: unmanned aerial vehicle-truck exchange operator, take over point-unmanned aerial vehicle exchange operator, unmanned aerial vehicle emission-landing point exchange converter, two-point exchange operator.

3. The collaborative distribution path planning method of claim 2 wherein said point of origin breaker is selected from a second set of breaker operators;

the second set of destruction operators comprises: randomly opening the destructive operator, randomly closing the destructive operator, closing the destructive operator with the minimum number of clients, opening the destructive operator with the maximum number of clients, and closing the destructive operator with the maximum number of clients;

the random on destroy operator: randomly selecting one collection point from unopened collection points to be opened;

the random shutdown destruction operator: randomly selecting one collection point from the opened collection points to close;

the random off-on destruction operator: randomly selecting one of the opened generation collection points to close, and opening the other generation collection point; the newly opened generation collection points are not based on random selection, but are based on distance, and if the non-opened generation collection points are closer to the closed generation collection points, the generation collection points are more likely to be opened;

Minimum number of customers off maximum number of customers on destruction operator: selecting the generation receiving point with the least service clients to close, and adding the service clients thereof into the damage list; meanwhile, in the unopened generation collection points, the generation collection point with the largest number of clients in the service range is selected for opening;

maximum number of customers closing the destruction operator: selecting the generation receiving point with the largest service client number to close, and adding the service client into the damage list;

the client point damage operator is selected from a first damage operator set;

the first set of destruction operators comprises: randomly destroying operators, greedy destroying operators, and carrying out greedy destroying operators with disturbance;

the random disruption operator: randomly selecting client points to delete until the number of client point removal is reached;

the greedy destruction operator: deleting the client point which maximizes the total cost until the number of client point removal is reached;

the perturbed greedy destruction operator: multiplying the perturbation coefficients in calculating the cost.

4. A cooperative dispensing path planning apparatus, comprising:

A model building unit for building a collaborative distribution model according to the distribution information, comprising: determining an objective function of the collaborative distribution; the objective function of the collaborative distribution is divided into four parts: the first part is the travel cost of the truck; the second part is the travel cost of the unmanned aerial vehicle; the third part is punishment cost of the truck running under different road conditions in rural areas; the fourth part is the collection-point operation cost, which comprises two sub-parts: one part is fixed cost, and the other part is compensation cost, wherein the compensation cost is compensation cost paid to a self-service customer by a point of sale; determining constraints of customer points, trucks, unmanned aerial vehicles and alternative collection points; determining a time constraint of a joint point of the unmanned aerial vehicle and the truck in the distribution network; establishing the collaborative distribution model according to the constraints of the client points, the trucks, the unmanned aerial vehicles and the proxy receiving points and/or the time constraints of the combining points of the unmanned aerial vehicles and the trucks in the distribution network and an objective function;

an initial solution calculation unit, configured to apply an initial solution algorithm according to the processed delivery information and the collaborative delivery model, and calculate an initial solution of a collaborative delivery path, including:

the optimizing unit is configured to iterate the initial solution using an optimizing algorithm to obtain an optimal solution of the collaborative distribution path, and includes:

Setting inner layer iteration times and outer layer iteration times, and executing a client point damage operator when the inner layer iteration times are smaller than a specified threshold value; when the inner layer iteration number is equal to a specified threshold value, executing a destructive operator replacing the collection point, and resetting the inner layer iteration number to 0; after the destructive operator is executed, a repair operator is executed, and after the repair operator is executed, the outer layer iteration times are increased by 1; updating an operator score according to the obtained quality of the new solution after each iteration period is finished and updating the new solution according to the simulated annealing rule;

5. An electronic device comprising a computer readable storage medium storing a computer program and a processor, the computer program implementing the collaborative distribution path planning method of any one of claims 1-3 when read and executed by the processor.

6. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when read and run by a processor, implements the collaborative distribution path planning method according to any of claims 1-3.