CN112862414A - Collaborative distribution route optimization method based on cluster traveler problem - Google Patents
Collaborative distribution route optimization method based on cluster traveler problem Download PDFInfo
- Publication number
- CN112862414A CN112862414A CN202110390408.4A CN202110390408A CN112862414A CN 112862414 A CN112862414 A CN 112862414A CN 202110390408 A CN202110390408 A CN 202110390408A CN 112862414 A CN112862414 A CN 112862414A
- Authority
- CN
- China
- Prior art keywords
- node
- path
- self
- cluster
- nodes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000005457 optimization Methods 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 31
- 230000007246 mechanism Effects 0.000 claims abstract description 81
- 238000005192 partition Methods 0.000 claims abstract description 6
- 238000003780 insertion Methods 0.000 claims description 3
- 230000037431 insertion Effects 0.000 claims description 3
- 238000013439 planning Methods 0.000 abstract description 3
- 238000004364 calculation method Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012913 prioritisation Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/083—Shipping
- G06Q10/0835—Relationships between shipper or supplier and carriers
- G06Q10/08355—Routing methods
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a collaborative distribution path optimization method based on cluster traveler problems, which comprises the steps of obtaining area parameters of an area to be analyzed; constructing a multi-cluster traveler problem model with coverage; carrying out partition balance optimization; randomly constructing an initial path; optimizing the initial path by adopting a disturbance mechanism; re-optimizing the subsequent path; and repeating the steps to set conditions to obtain a final cooperative distribution path optimization result. The method can effectively balance the workload of each distributor, provide the optimal path planning for the distributors, effectively reduce the total distribution cost, and has high reliability and good practicability.
Description
Technical Field
The invention belongs to the field of path planning, and particularly relates to a collaborative distribution path optimization method based on a cluster traveler problem.
Background
With the development of mobile internet, e-commerce platforms are emerging continuously, the scale of online shopping users is continuously enlarged, and convenience is brought to daily life of people. Meanwhile, express services which are not classified with network shopping are also rapidly increased. Express delivery business has gradually penetrated people's daily life, has brought huge business opportunity for each enterprise. The last kilometer of delivery refers to that after online shopping, an express company delivers goods from a local sorting center to a customer, and the last mile of the whole express process is the delivery. Meanwhile, the last kilometer of distribution is the key point of urban intelligent logistics construction, and is also an important research scene in the field of path optimization.
The path optimization Problem is often transformed to solve a Traveling Salesman Problem (TSP) or a variant thereof. Among them, the Cluster Traveling Salesman Problem (CTSP) is often used to solve the regionalized last mile delivery optimization Problem. In the study of baniasar P et al, it was considered to solve the generalized CTSP problem: the method has the main idea that the generalized CTSP problem is converted into a common TSP model, and a heuristic algorithm is adopted to solve on the basis; further, the application of the proposed solution to the last mile delivery is described, where the delivery truck carries drones to complete the package delivery task, i.e. when the delivery truck is transferred from one customer to another, no one will take the package from the truck and serve more customers in close proximity, while each drone returns to the truck after the corresponding delivery. Under this kind of delivery mode, can effectively reduce the travel time of truck to can reach the mesh of optimizing unmanned aerial vehicle idle rate. However, this model has a very limited number of customers to service due to the limitations of the drone in delivering packages. In addition, the mode ignores the resources of the self-picking point in real life, and the whole distribution process deviates from the actual scene. The study by phong H N et al focused on solving prioritized CTSP: in the text, each client is prioritized according to the degree of urgency, and the client nodes are classified according to the priorities. However, the strict priority system not only increases the total distribution cost, but also limits the solution space, thereby affecting the calculation of the optimal solution. To address this problem, Phuong H N et al set up a d slack mechanism for prioritization that provides that two client nodes will only be serviced in strict order if the difference in priority is greater than d. Experimental results prove that the finally obtained path is similar to the TSP problem solution scheme.
However, the existing research still has the following problems: (1) neglecting existing self-extracting point resources, so that the established model has deviation from an actual distribution scene; (2) only one delivery route is aimed at, and an effective large-area coordinated delivery scheme cannot be provided. These problems lead to the existing solutions being difficult to apply to actual distribution scenarios and having poor practicability.
Disclosure of Invention
The invention aims to provide a collaborative distribution path optimization method which is fit for an actual distribution scene, high in reliability and good in practicability and based on the problem of cluster travelers.
The cooperative distribution path optimization method based on the cluster traveler problem provided by the invention comprises the following steps:
s1, acquiring area parameters of an area to be analyzed;
s2, constructing a multi-cluster traveling salesman problem model with coverage according to the area parameters obtained in the step S1;
s3, carrying out partition balance optimization according to the total distribution request in the region;
s4, randomly constructing an initial path;
s5, optimizing the initial path in the step S4 by adopting a disturbance mechanism;
s6, re-optimizing the subsequent path;
and S7, repeating the steps S5 and S6 until the set conditions are reached, and obtaining the final cooperative distribution path optimization result.
The step S2 of constructing a multi-cluster travel dealer problem model with coverage specifically includes constructing an objective function with the minimum delivery cost as a target, with the limiting conditions of the number of dispatchers, the number of customer traversals, the number of self-service points traversals, path continuity, sub-trip constraints, the self-service point capacity limitations, and the self-service point service range:
constraint conditions are as follows:
ui-uj+(n-m+1)xij≤n-m-1,1≤i≠j≤n
wherein V is a set of nodes, and V ═ VD∪VS∪VC，VDIs a warehouse, VSFor a set of self-pick-up points (SDF), VCA set of client nodes; x is the number ofijTo decide a variable, x is taken if and only if edge (i, j) is on the wayij1, otherwise xij＝0；cijThe path cost corresponding to edge (i, j); x is the number of0jThe variable is a 0-1 variable, and when the value is 1, the variable indicates that a path from the warehouse 0 to the node j exists, and when the value is 0, the variable indicates that the path from the warehouse 0 to the node j does not exist; m is the number of couriers; x is the number ofi0The variable is a 0-1 variable, and when the value is 1, the path returning from the node i to the warehouse 0 exists, and when the value is 0, the path returning from the node i to the warehouse 0 does not exist; z is a radical ofilFor decision variables, z is taken if and only if the customer node i is covered by the self-lifting point node lil1, otherwise zil＝0；uiIs the load of the path before traversing node i; n is the number of the client nodes needing to be served; v. ofbCapacity limit for self-proposed point b; rbIs the service radius of the self-proposed point b.
Performing partition balance optimization in step S3, specifically, performing optimization by using the following steps:
A. in h clusters, the number of nodes in each cluster is recorded as NC1,NC2,...,NCh(ii) a In each cluster, calculating the sum D of the distances from each node to other nodes in the clusteriCalculating the number N of neighbors of each node in the range with the radius of RiR is the service radius of the self-picking point; calculating the weight of the node:
wherein a is a weight parameter corresponding to the sum of distances, and b is a weight parameter corresponding to the number of neighbors; selecting the node with the minimum weight value as the central point of the cluster; finally, the center point set of each cluster is obtained as { C1,C2,...,Ch}；
B. The maximum number of nodes which can be served by each distributor is calculated by the following formula
In the formula NCiThe number of nodes in the ith cluster; h is the number of clusters; m is the number of couriers;
C. selecting the center points of m clusters with the most nodes in the h clusters as the initial centroids of the classes, and recording the initial number L in each class1,L2,...,Lm；
D. Calculating Euclidean distances from the other central points j to each initial centroid i;
E. and D, selecting the central point with the shortest distance according to the Euclidean distance obtained in the step D, and performing the following operations and judgments:
if the number of the nodes of the current class after the cluster j is integrally added into the corresponding class is not more than
Adding the cluster j into the corresponding class integrally, and updating the number of the classes to be Li+Ncj；
If the number of the nodes of the current class is more than the number of the nodes of the corresponding class after the cluster j is integrally added to the corresponding class
Adding the cluster j into the next class with the shortest distance as a whole, and judging the number of corresponding nodes until the cluster j is added into one class;
F. repeating the step E until all the nodes are completely distributed;
G. calculating the variance among the number of nodes which need to be served by each courier and judging: if the variance is smaller than the set threshold, terminating the distribution process; otherwise, calculating the center points of all classes and setting the center points as the mass centers;
H. and D, repeating the step D to the step G until the variance of the number of the clients distributed by each courier is smaller than a set threshold value.
The step S4 is to randomly construct an initial path, specifically to randomly construct an internal path in each original cluster, and then connect the clusters together according to the shortest distance principle, thereby constructing an initial path.
The step S5, which is to optimize the initial path of the step S4 by using a perturbation mechanism, specifically, the following steps are adopted for optimization:
after the current path is subjected to iterative optimization, the following judgment is carried out:
if the cost of the path for k times is not reduced and the current disturbance mechanism is the mechanism a, modifying the current disturbance mechanism into the mechanism b, and performing subsequent iterative optimization;
if the cost of the path for k times is not reduced and the current disturbance mechanism is the mechanism b, modifying the current disturbance mechanism into the mechanism a, and performing subsequent iterative optimization;
if the times that the path cost is not reduced do not reach the set k times, keeping the current disturbance mechanism unchanged, and performing subsequent iterative optimization;
the mechanism a is as follows: judging the number of clusters in the current class:
when the number of the clusters is larger than or equal to a set value, randomly selecting four edges among the classes in the initial path, and reconnecting the four edges according to the minimum cost principle to obtain a path shorter than the previous path;
when the number of the clusters is smaller than a set value, randomly selecting two clusters, and selecting insertion positions for the two clusters again according to a minimum cost principle;
the mechanism b is as follows: judging the number of clusters in the current class:
when the number of the clusters is larger than or equal to a set value, randomly selecting three edges among classes in the initial path, and reconnecting the three edges according to a minimum cost principle to obtain a path shorter than the previous path;
when the number of the clusters is smaller than a set value, maintaining a mechanism a;
and optimizing the initial path by adopting the technical scheme until the specified iteration times are reached.
The re-optimization of the subsequent path in step S6 is specifically to re-optimize the subsequent path by using a neighborhood random search mechanism.
The subsequent path is re-optimized by adopting a neighborhood random search mechanism, and specifically, the re-optimization is performed by adopting the following steps:
randomly selecting one of the following 4 neighborhood mechanisms, and performing neighborhood search on the existing path; meanwhile, if the selected neighborhood mechanism does not optimize the current path, randomly selecting one of the remaining unselected neighborhood mechanisms to perform neighborhood search; if the selected neighborhood mechanism optimizes the current path, the current neighborhood mechanism is continuously adopted for searching;
the conditions for stopping optimization are as follows: all neighborhood searching mechanisms are selected;
the 4 neighborhood mechanisms are:
1)1-shift mechanism: randomly selecting nodes i and j, and judging:
if node i is a client node:
if the node i and the node j are both client nodes, directly moving the node i to the front of the node j;
if the node i is a client node and the node j is a self-picking point, whether the node i meets the requirement of being put into the node j is judged:
if the node i meets the requirement of being placed into the node j, placing the node i into the node j;
otherwise, directly placing the node i in front of the node j;
if the node i is a self-pick-up point, recording CiCalculating an intermediate variable beta-i% (C) for the number of nodes stored in the current self-extracting point seti+1) and judging: if β is 0, the node i takes the whole set of self-extracting points; otherwise, the node i sequentially takes the nodes in the corresponding self-extracting point set and directly places the obtained node i in front of the node j;
2) DropCheap mechanism: randomly selecting a cluster, taking out the node with the maximum consumption in the path of the cluster, and then directly inserting the selected node into the position with the minimum consumption;
3)2-Swap mechanism:
if the node i is a client node: then the following determination is made:
if the node i and the node j are both client nodes, directly exchanging the node i with the node j;
if the node j is a self-extracting point, then C is recordedjFor the node stored in the current self-picking point setAnd calculating the intermediate variable β ═ j% (C)j+1) and judging: if β is 0, the node j takes the whole set of self-extracting points; otherwise, the node j sequentially takes the nodes in the corresponding self-extracting point set, and if the node i meets the requirement of putting the node j into the node j, the node i and the node j are exchanged; otherwise, exchanging the node i with the whole self-extracting point set;
if the node i is a self-extracting point, the following judgment is carried out:
if the node j is a client node, directly exchanging the node i with the node j;
if the node j is a self-extracting point, then C is recordedjCalculating the number of nodes stored in the current self-extracting point set and calculating an intermediate variable beta-j% (C)j+1) and judging: if β is 0, then the entire set of self-lifting points for node i is swapped with the entire set of self-lifting points for node j; otherwise, the node j sequentially takes the nodes in the corresponding self-extracting point set, and judges the node i: if the node i is the whole self-extracting point set, exchanging the two self-extracting point sets; if the points in the two sets are selected, respectively judging whether the node i and the node j can be placed in the self-picking point set of the opposite side, if the node i can be placed in the self-picking point set of the node j and the node j can be placed in the self-picking point set of the node i, exchanging the node i and the node j, and if not, exchanging the self-picking point set of the node i and the self-picking point set of the node j;
4)2-opt mechanism:
if the node i and the node j are in the same cluster, directly turning over the sequence between the two points;
if the node i and the node j are not in the same cluster, finding a demarcation point m between the clusters1，…，mnWithin the same cluster, nodes i to m1Turning over and putting m1And mn-1Turn over inside cluster, mnTurning to a node j; and then the path sequence among the clusters is reversed.
The cooperative delivery path optimization method based on the cluster traveler problem comprises the steps of constructing an objective function and a constraint condition to establish a basic mathematical model, carrying out balanced distribution on the workload of each deliverer according to an input network example, then obtaining an initial path scheme of each deliverer, and then carrying out disturbance operation and neighborhood search operation on the initial path in an iterative manner, thereby obtaining an optimal tour path; therefore, the method can effectively balance the workload of each distributor, provide the optimal path planning for the distributors, effectively reduce the total distribution cost, and has high reliability and good practicability.
Drawings
FIG. 1 is a schematic process flow diagram of the process of the present invention.
Fig. 2 is a schematic view of an application scenario of the method of the present invention.
Detailed Description
Fig. 1 is a schematic view of a method flow of the method of the present invention, and fig. 2 is a schematic view of an application scenario of the method of the present invention.
The cooperative distribution path optimization method based on the cluster traveler problem provided by the invention comprises the following steps:
s1, acquiring area parameters of an area to be analyzed;
s2, constructing a multi-cluster traveling salesman problem model with coverage according to the area parameters obtained in the step S1; specifically, the number of distributors, the number of client traversal times, the number of self-service point traversal times, the path continuity, the sub-trip constraint, the self-service point capacity limitation and the self-service point service range are taken as limiting conditions, and the objective function is constructed by taking the minimum distribution cost as the target:
constraint conditions are as follows:
ui-uj+(n-m+1)xij≤n-m-1,1≤i≠j≤n
wherein V is a set of nodes, and V ═ VD∪VS∪VC，VDIs a warehouse, VSFor a set of self-pick-up points (SDF), VCA set of client nodes; x is the number ofijTo decide a variable, x is taken if and only if edge (i, j) is on the wayij1, otherwise xij＝0；cijThe path cost corresponding to edge (i, j); x is the number of0jThe variable is a 0-1 variable, and when the value is 1, the variable indicates that a path from the warehouse 0 to the node j exists, and when the value is 0, the variable indicates that the path from the warehouse 0 to the node j does not exist; m is the number of couriers; x is the number ofi0The variable is a 0-1 variable, and when the value is 1, the path returning from the node i to the warehouse 0 exists, and when the value is 0, the path returning from the node i to the warehouse 0 does not exist; z is a radical ofilFor decision variables, z is taken if and only if the customer node i is covered by the self-lifting point node lil1, otherwise zil＝0；uiIs the load of the path before traversing node i; n is a customer node needing serviceThe number of the cells; v. ofbCapacity limit for self-proposed point b; rbThe service radius of the self-extracting point b;
s3, carrying out partition balance optimization according to the total distribution request in the region; specifically, the method comprises the following steps:
A. in h clusters, the number of nodes in each cluster is recorded as NC1,NC2,...,NCh(ii) a In each cluster, calculating the sum D of the distances from each node to other nodes in the clusteriCalculating the number N of neighbors of each node in the range with the radius of RiR is the service radius of the self-picking point; calculating the weight of the node:
wherein a is a weight parameter corresponding to the sum of distances, and b is a weight parameter corresponding to the number of neighbors; selecting the node with the minimum weight value as the central point of the cluster; finally, the center point set of each cluster is obtained as { C1,C2,...,Ch}；
B. The maximum number of nodes which can be served by each distributor is calculated by the following formula
In the formula NCiThe number of nodes in the ith cluster; h is the number of clusters; m is the number of couriers;
C. selecting the center points of m clusters with the most nodes in the h clusters as the initial centroids of the classes, and recording the initial number L in each class1,L2,...,Lm；
D. Calculating Euclidean distances from the other central points j to each initial centroid i;
E. and D, selecting the central point with the shortest distance according to the Euclidean distance obtained in the step D, and performing the following operations and judgments:
if it will collectThe number of the nodes of the current class after the group j is integrally added into the corresponding class is not more than
Adding the cluster j into the corresponding class integrally, and updating the number of the classes to be Li+Ncj；
If the number of the nodes of the current class is more than the number of the nodes of the corresponding class after the cluster j is integrally added to the corresponding class
Adding the cluster j into the next class with the shortest distance as a whole, and judging the number of corresponding nodes until the cluster j is added into one class;
F. repeating the step E until all the nodes are completely distributed;
G. calculating the variance among the number of nodes which need to be served by each courier and judging: if the variance is smaller than the set threshold, terminating the distribution process; otherwise, calculating the center points of all classes and setting the center points as the mass centers;
H. and D, repeating the step D to the step G until the variance of the number of the clients distributed by each courier is smaller than a set threshold value.
S4, randomly constructing an initial path; specifically, an internal path is randomly constructed in each original cluster, and then the clusters are connected together according to the shortest distance principle, so that an initial path is established;
s5, optimizing the initial path in the step S4 by adopting a disturbance mechanism; specifically, the method comprises the following steps:
after the current path is subjected to iterative optimization, the following judgment is carried out:
if the cost of the path for k times is not reduced and the current disturbance mechanism is the mechanism a, modifying the current disturbance mechanism into the mechanism b, and performing subsequent iterative optimization;
if the cost of the path for k times is not reduced and the current disturbance mechanism is the mechanism b, modifying the current disturbance mechanism into the mechanism a, and performing subsequent iterative optimization;
if the times that the path cost is not reduced do not reach the set k times, keeping the current disturbance mechanism unchanged, and performing subsequent iterative optimization;
the mechanism a is as follows: judging the number of clusters in the current class:
when the number of the clusters is larger than or equal to a set value (such as 8), randomly selecting four edges among classes in the initial path, and reconnecting the four edges according to the minimum cost principle to obtain a path shorter than the previous path;
when the number of the clusters is smaller than a set value (for example, 8), randomly selecting two clusters, and selecting insertion positions for the two clusters again according to a minimum cost principle;
the mechanism b is as follows: judging the number of clusters in the current class:
when the number of the clusters is more than or equal to a set value (such as 3), randomly selecting three edges among the classes in the initial path, and reconnecting the three edges according to the minimum cost principle to obtain a path shorter than the previous path;
when the number of the clusters is smaller than a set value (for example, 3), maintaining the mechanism a;
and optimizing the initial path by adopting the technical scheme until the specified iteration times are reached.
S6, re-optimizing the subsequent path; specifically, a neighborhood random search mechanism is adopted to re-optimize a subsequent path;
in specific implementation, the following steps are adopted for re-optimization:
randomly selecting one of the following 4 neighborhood mechanisms, and performing neighborhood search on the existing path; meanwhile, if the selected neighborhood mechanism does not optimize the current path, randomly selecting one of the remaining unselected neighborhood mechanisms to perform neighborhood search; if the selected neighborhood mechanism optimizes the current path, the current neighborhood mechanism is continuously adopted for searching;
the conditions for stopping optimization are as follows:
all neighborhood searching mechanisms are selected;
the 4 neighborhood mechanisms are:
1)1-shift mechanism: randomly selecting nodes i and j, and judging:
if node i is a client node:
if the node i and the node j are both client nodes, directly moving the node i to the front of the node j;
if the node i is a client node and the node j is a self-picking point, whether the node i meets the requirement of being put into the node j is judged:
if the node i meets the requirement of being placed into the node j, placing the node i into the node j;
otherwise, directly placing the node i in front of the node j;
if the node i is a self-pick-up point, recording CiCalculating an intermediate variable beta-i% (C) for the number of nodes stored in the current self-extracting point seti+1) and judging: if β is 0, the node i takes the whole set of self-extracting points; otherwise, the node i sequentially takes the nodes in the corresponding self-extracting point set and directly places the obtained node i in front of the node j;
2) DropCheap mechanism: randomly selecting a cluster, taking out the node with the maximum consumption in the path of the cluster, and then directly inserting the selected node into the position with the minimum consumption;
3)2-Swap mechanism:
if the node i is a client node: then the following determination is made:
if the node i and the node j are both client nodes, directly exchanging the node i with the node j;
if the node j is a self-extracting point, then C is recordedjCalculating the number of nodes stored in the current self-extracting point set and calculating an intermediate variable beta-j% (C)j+1) and judging: if β is 0, the node j takes the whole set of self-extracting points; otherwise, the node j sequentially takes the nodes in the corresponding self-extracting point set, and if the node i meets the requirement of putting the node j into the node j, the node i and the node j are exchanged; otherwise, exchanging the node i with the whole self-extracting point set;
if the node i is a self-extracting point, the following judgment is carried out:
if the node j is a client node, directly exchanging the node i with the node j;
if the node j is a self-extracting point, then C is recordedjCalculating the number of nodes stored in the current self-extracting point set and calculating an intermediate variable beta ═ j% ((Cj+1) and judging: if β is 0, then the entire set of self-lifting points for node i is swapped with the entire set of self-lifting points for node j; otherwise, the node j sequentially takes the nodes in the corresponding self-extracting point set, and judges the node i: if the node i is the whole self-extracting point set, exchanging the two self-extracting point sets; if the points in the two sets are selected, respectively judging whether the node i and the node j can be placed in the self-picking point set of the opposite side, if the node i can be placed in the self-picking point set of the node j and the node j can be placed in the self-picking point set of the node i, exchanging the node i and the node j, and if not, exchanging the self-picking point set of the node i and the self-picking point set of the node j;
4)2-opt mechanism:
if the node i and the node j are in the same cluster, directly turning over the sequence between the two points;
if the node i and the node j are not in the same cluster, finding a demarcation point m between the clusters1，…，mnWithin the same cluster, nodes i to m1Turning over and putting m1And mn-1Turn over inside cluster, mnTurning over to j; and then the path sequence among the clusters is reversed.
And S7, repeating the steps S5 and S6 until the set conditions are reached, and obtaining the final cooperative distribution path optimization result.
Claims (7)
1. A collaborative distribution route optimization method based on cluster traveler problems comprises the following steps:
s1, acquiring area parameters of an area to be analyzed;
s2, constructing a multi-cluster traveling salesman problem model with coverage according to the area parameters obtained in the step S1;
s3, carrying out partition balance optimization according to the total distribution request in the region;
s4, randomly constructing an initial path;
s5, optimizing the initial path in the step S4 by adopting a disturbance mechanism;
s6, re-optimizing the subsequent path;
and S7, repeating the steps S5 and S6 until the set conditions are reached, and obtaining the final cooperative distribution path optimization result.
2. The method for optimizing a collaborative distribution path based on a trunking traveler problem according to claim 1, wherein the step S2 is to construct a multi-trunking traveler problem model with coverage, specifically to construct an objective function with the objective of minimizing distribution cost by taking the number of distributors, the number of customer traversals, the number of self-service point traversals, path continuity, sub-trip constraints, self-service point capacity limitations, and self-service point service scope as constraints:
constraint conditions are as follows:
ui-uj+(n-m+1)xij≤n-m-1,1≤i≠j≤n
wherein V is a set of nodes, and V ═ VD∪VS∪VC，VDIs a warehouse, VSFor a set of self-pick-up points (SDF), VCA set of client nodes; x is the number ofijTo decide a variable, x is taken if and only if edge (i, j) is on the wayij1, otherwise xij＝0；cijThe path cost corresponding to edge (i, j); x is the number of0jThe variable is a 0-1 variable, and when the value is 1, the variable indicates that a path from the warehouse 0 to the node j exists, and when the value is 0, the variable indicates that the path from the warehouse 0 to the node j does not exist; m is the number of couriers; x is the number ofi0The variable is a 0-1 variable, and when the value is 1, the path returning from the node i to the warehouse 0 exists, and when the value is 0, the path returning from the node i to the warehouse 0 does not exist; z is a radical ofilFor decision variables, z is taken if and only if the customer node i is covered by the self-lifting point node lil1, otherwise zil＝0；uiIs the load of the path before traversing node i; n is the number of the client nodes needing to be served; v. ofbCapacity limit for self-proposed point b; rbIs the service radius of the self-proposed point b.
3. The collaborative distribution path optimization method based on the cluster traveler problem according to claim 2, wherein the partition balancing optimization in step S3 is specifically performed by adopting the following steps:
A. in h clusters, the number of nodes in each cluster is recorded as NC1,NC2,...,NCh(ii) a In each cluster, calculating the sum D of the distances from each node to other nodes in the clusteriCalculating each node in halfNumber of neighbors N within the range of a radius RiR is the service radius of the self-picking point; calculating the weight of the node:
wherein a is a weight parameter corresponding to the sum of distances, and b is a weight parameter corresponding to the number of neighbors; selecting the node with the minimum weight value as the central point of the cluster; finally, the center point set of each cluster is obtained as { C1,C2,...,Ch}；
B. The maximum number of nodes which can be served by each distributor is calculated by the following formula
In the formula NCiThe number of nodes in the ith cluster; h is the number of clusters; m is XXXXXXX;
C. selecting the center points of m clusters with the most nodes in the h clusters as the initial centroids of the classes, and recording the initial number L in each class1,L2,...,Lm；
D. Calculating Euclidean distances from the other central points j to each initial centroid i;
E. and D, selecting the central point with the shortest distance according to the Euclidean distance obtained in the step D, and performing the following operations and judgments:
if the number of the nodes of the current class after the cluster j is integrally added into the corresponding class is not more than
Adding the cluster j into the corresponding class integrally, and updating the number of the classes to be Li+Ncj；
If the number of the nodes of the current class is more than the number of the nodes of the corresponding class after the cluster j is integrally added to the corresponding class
Adding the cluster j into the next class with the shortest distance as a whole, and judging the number of corresponding nodes until the cluster j is added into one class;
F. repeating the step E until all the nodes are completely distributed;
G. calculating the variance among the number of nodes which need to be served by each courier and judging: if the variance is smaller than the set threshold, terminating the distribution process; otherwise, calculating the center points of all classes and setting the center points as the mass centers;
H. and D, repeating the step D to the step G until the variance of the number of the clients distributed by each courier is smaller than a set threshold value.
4. The method for optimizing a collaborative distribution path based on a cluster traveler problem according to claim 3, wherein the step S4 is implemented by randomly constructing an initial path, specifically by randomly constructing an internal path in each original cluster, and then connecting the clusters according to a shortest distance principle, thereby establishing the initial path.
5. The collaborative distribution path optimization method based on the trunking traveler problem of claim 4, wherein the step S5 is to optimize the initial path of the step S4 by using a perturbation mechanism, specifically by using the following steps:
after the current path is subjected to iterative optimization, the following judgment is carried out:
if the cost of the path for k times is not reduced and the current disturbance mechanism is the mechanism a, modifying the current disturbance mechanism into the mechanism b, and performing subsequent iterative optimization;
if the cost of the path for k times is not reduced and the current disturbance mechanism is the mechanism b, modifying the current disturbance mechanism into the mechanism a, and performing subsequent iterative optimization;
if the times that the path cost is not reduced do not reach the set k times, keeping the current disturbance mechanism unchanged, and performing subsequent iterative optimization;
the mechanism a is as follows: judging the number of clusters in the current class:
when the number of the clusters is larger than or equal to a set value, randomly selecting four edges among the classes in the initial path, and reconnecting the four edges according to the minimum cost principle to obtain a path shorter than the previous path;
when the number of the clusters is smaller than a set value, randomly selecting two clusters, and selecting insertion positions for the two clusters again according to a minimum cost principle;
the mechanism b is as follows: judging the number of clusters in the current class:
when the number of the clusters is larger than or equal to a set value, randomly selecting three edges among classes in the initial path, and reconnecting the three edges according to a minimum cost principle to obtain a path shorter than the previous path;
when the number of the clusters is smaller than a set value, maintaining a mechanism a;
and optimizing the initial path by adopting the technical scheme until the specified iteration times are reached.
6. The cooperative distribution path optimization method based on the cluster traveler problem of claim 5, wherein the subsequent path is re-optimized in step S6, specifically by using a neighborhood random search mechanism.
7. The collaborative distribution path optimization method based on the cluster traveler problem according to claim 6, characterized in that the subsequent path is re-optimized by using a neighborhood random search mechanism, specifically by using the following steps:
randomly selecting one of the following 4 neighborhood mechanisms, and performing neighborhood search on the existing path; meanwhile, if the selected neighborhood mechanism does not optimize the current path, randomly selecting one of the remaining unselected neighborhood mechanisms to perform neighborhood search; if the selected neighborhood mechanism optimizes the current path, the current neighborhood mechanism is continuously adopted for searching;
the conditions for stopping optimization are as follows:
all neighborhood searching mechanisms are selected;
the 4 neighborhood mechanisms are:
1)1-shift mechanism: randomly selecting nodes i and j, and judging:
if node i is a client node:
if the node i and the node j are both client nodes, directly moving the node i to the front of the node j;
if the node i is a client node and the node j is a self-picking point, whether the node i meets the requirement of being put into the node j is judged:
if the node i meets the requirement of being placed into the node j, placing the node i into the node j;
otherwise, directly placing the node i in front of the node j;
if the node i is a self-pick-up point, recording CiCalculating an intermediate variable beta-i% (C) for the number of nodes stored in the current self-extracting point seti+1) and judging: if β is 0, the node i takes the whole set of self-extracting points; otherwise, the node i sequentially takes the nodes in the corresponding self-extracting point set and directly places the obtained node i in front of the node j;
2) DropCheap mechanism: randomly selecting a cluster, taking out the node with the maximum consumption in the path of the cluster, and then directly inserting the selected node into the position with the minimum consumption;
3)2-Swap mechanism:
if the node i is a client node: then the following determination is made:
if the node i and the node j are both client nodes, directly exchanging the node i with the node j;
if the node j is a self-extracting point, then C is recordedjCalculating the number of nodes stored in the current self-extracting point set and calculating an intermediate variable beta-j% (C)j+1) and judging: if β is 0, the node j takes the whole set of self-extracting points; otherwise, the node j sequentially takes the nodes in the corresponding self-extracting point set, and if the node i meets the requirement of putting the node j into the node j, the node i and the node j are exchanged; otherwise, exchanging the node i with the whole self-extracting point set;
if the node i is a self-extracting point, the following judgment is carried out:
if the node j is a client node, directly exchanging the node i with the node j;
if the node j is a self-extracting point, then C is recordedjCalculating the number of nodes stored in the current self-extracting point set and calculating an intermediate variable beta-j% (C)j+1) and judging: if β is 0, then the entire set of self-lifting points for node i is swapped with the entire set of self-lifting points for node j; otherwise, the node j sequentially takes the nodes in the corresponding self-extracting point set, and judges the node i: if the node i is the whole self-extracting point set, exchanging the two self-extracting point sets; if the points in the two sets are selected, respectively judging whether the node i and the node j can be placed in the self-picking point set of the opposite side, if the node i can be placed in the self-picking point set of the node j and the node j can be placed in the self-picking point set of the node i, exchanging the node i and the node j, and if not, exchanging the self-picking point set of the node i and the self-picking point set of the node j;
4)2-opt mechanism:
if the node i and the node j are in the same cluster, directly turning over the sequence between the two points;
if the node i and the node j are not in the same cluster, finding a demarcation point m between the clusters1，…，mnWithin the same cluster, nodes i to m1Turning over and putting m1And mn-1Turn over inside cluster, mnTurning over to j; and then the path sequence among the clusters is reversed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110390408.4A CN112862414B (en) | 2021-04-12 | 2021-04-12 | Collaborative distribution path optimization method based on cluster travel business problem |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110390408.4A CN112862414B (en) | 2021-04-12 | 2021-04-12 | Collaborative distribution path optimization method based on cluster travel business problem |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112862414A true CN112862414A (en) | 2021-05-28 |
| CN112862414B CN112862414B (en) | 2023-12-05 |
Family
ID=75992522
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110390408.4A Active CN112862414B (en) | 2021-04-12 | 2021-04-12 | Collaborative distribution path optimization method based on cluster travel business problem |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112862414B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114925937A (en) * | 2022-06-27 | 2022-08-19 | 东北大学 | Stage scene point cloud scanning site selection and path planning method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413209A (en) * | 2013-07-17 | 2013-11-27 | 西南交通大学 | Method for selecting multi-user and multi-warehouse logistics distribution path |
| CN105740972A (en) * | 2016-01-22 | 2016-07-06 | 合肥工业大学 | Drop and pull transport coordination path planning method |
| CN105956681A (en) * | 2016-04-15 | 2016-09-21 | 合肥工业大学 | Drop-and-pull transport dynamic path planning method based on receding horizon optimization |
| CN108108847A (en) * | 2017-12-29 | 2018-06-01 | 合肥工业大学 | A kind of paths planning method of electric business logistics last one kilometer dispatching |
| CN110782086A (en) * | 2019-10-24 | 2020-02-11 | 山东师范大学 | Rescue vehicle distribution path optimization method and system with unmanned aerial vehicle |
-
2021
- 2021-04-12 CN CN202110390408.4A patent/CN112862414B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103413209A (en) * | 2013-07-17 | 2013-11-27 | 西南交通大学 | Method for selecting multi-user and multi-warehouse logistics distribution path |
| CN105740972A (en) * | 2016-01-22 | 2016-07-06 | 合肥工业大学 | Drop and pull transport coordination path planning method |
| CN105956681A (en) * | 2016-04-15 | 2016-09-21 | 合肥工业大学 | Drop-and-pull transport dynamic path planning method based on receding horizon optimization |
| CN108108847A (en) * | 2017-12-29 | 2018-06-01 | 合肥工业大学 | A kind of paths planning method of electric business logistics last one kilometer dispatching |
| CN110782086A (en) * | 2019-10-24 | 2020-02-11 | 山东师范大学 | Rescue vehicle distribution path optimization method and system with unmanned aerial vehicle |
Non-Patent Citations (1)
| Title |
|---|
| 李腾;冯珊;宋君;刘金芳;: ""货到人"拣选系统机器人任务分配的鲁棒双层规划模型", 运筹与管理, no. 12, pages 29 - 38 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114925937A (en) * | 2022-06-27 | 2022-08-19 | 东北大学 | Stage scene point cloud scanning site selection and path planning method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112862414B (en) | 2023-12-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Cheng et al. | Cooperation-aware task assignment in spatial crowdsourcing | |
| Liu et al. | Mobility-aware dynamic taxi ridesharing | |
| Gao et al. | Optimal multi-taxi dispatch for mobile taxi-hailing systems | |
| Zheng et al. | Online trichromatic pickup and delivery scheduling in spatial crowdsourcing | |
| Wolfson et al. | Fairness versus optimality in ridesharing | |
| Pan et al. | Ridesharing: simulator, benchmark, and evaluation | |
| Al Chami et al. | A lexicographic approach for the bi-objective selective pickup and delivery problem with time windows and paired demands | |
| Velasco et al. | A non-dominated sorting genetic algorithm for a bi-objective pick-up and delivery problem | |
| CN109919354B (en) | Car pooling optimization method, index structure and system based on social big data | |
| Lei et al. | An improved variable neighborhood search for parallel drone scheduling traveling salesman problem | |
| CN111860957B (en) | Multi-vehicle-type vehicle path planning method considering secondary distribution and balancing | |
| CN110705741A (en) | Multi-distribution center vehicle path optimization method based on improved ant colony algorithm | |
| Shahzaad et al. | Robust composition of drone delivery services under uncertainty | |
| Raj et al. | A branch-and-price approach for the parallel drone scheduling vehicle routing problem | |
| CN112801347A (en) | Multi-target city two-stage distribution planning method based on mobile transfer station and crowdsourcing | |
| CN109840625B (en) | Courier group path navigation method | |
| Lin et al. | VShare: A wireless social network aided vehicle sharing system using hierarchical cloud architecture | |
| CN112862414B (en) | Collaborative distribution path optimization method based on cluster travel business problem | |
| Wang et al. | Multi-objective location-routing problem with simultaneous pickup and delivery for urban distribution | |
| CN109800911B (en) | Unified navigation method for delivery paths of multiple couriers | |
| Li et al. | Efficient adaptive matching for real-time city express delivery | |
| Wang et al. | Two‐Stage Solution for Meal Delivery Routing Optimization on Time‐Sensitive Customer Satisfaction | |
| CN115619304A (en) | Logistics node site selection planning method based on clustering algorithm | |
| Costa et al. | Last mile delivery considering time-dependent locations | |
| Benjaafar et al. | Human in the loop automation: Ride-hailing with remote (tele-) drivers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |