CN116258311A - Medical unmanned aerial vehicle scheduling model and solving algorithm for considering sample timeliness - Google Patents
Medical unmanned aerial vehicle scheduling model and solving algorithm for considering sample timeliness Download PDFInfo
- Publication number
- CN116258311A CN116258311A CN202211454497.5A CN202211454497A CN116258311A CN 116258311 A CN116258311 A CN 116258311A CN 202211454497 A CN202211454497 A CN 202211454497A CN 116258311 A CN116258311 A CN 116258311A
- Authority
- CN
- China
- Prior art keywords
- unmanned aerial
- aerial vehicle
- node
- algorithm
- label
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 claims abstract description 24
- 238000007689 inspection Methods 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims abstract description 17
- 230000014509 gene expression Effects 0.000 claims description 28
- 230000008859 change Effects 0.000 claims description 12
- 230000009977 dual effect Effects 0.000 claims description 10
- 230000009467 reduction Effects 0.000 claims description 7
- 238000005265 energy consumption Methods 0.000 claims description 6
- 230000002829 reductive effect Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000003780 insertion Methods 0.000 claims description 5
- 230000037431 insertion Effects 0.000 claims description 5
- 238000002372 labelling Methods 0.000 claims description 4
- 230000007423 decrease Effects 0.000 claims description 2
- 238000012217 deletion Methods 0.000 claims description 2
- 230000037430 deletion Effects 0.000 claims description 2
- 230000000670 limiting effect Effects 0.000 claims description 2
- 230000036961 partial effect Effects 0.000 claims description 2
- 238000005192 partition Methods 0.000 claims description 2
- 238000013138 pruning Methods 0.000 claims description 2
- 230000009471 action Effects 0.000 claims 1
- 230000002238 attenuated effect Effects 0.000 claims 1
- FFBHFFJDDLITSX-UHFFFAOYSA-N benzyl N-[2-hydroxy-4-(3-oxomorpholin-4-yl)phenyl]carbamate Chemical compound OC1=C(NC(=O)OCC2=CC=CC=C2)C=CC(=C1)N1CCOCC1=O FFBHFFJDDLITSX-UHFFFAOYSA-N 0.000 claims 1
- 101150067055 minC gene Proteins 0.000 claims 1
- 230000000452 restraining effect Effects 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract description 20
- 230000001360 synchronised effect Effects 0.000 abstract description 2
- 230000005611 electricity Effects 0.000 description 5
- 150000007523 nucleic acids Chemical class 0.000 description 5
- 102000039446 nucleic acids Human genes 0.000 description 5
- 108020004707 nucleic acids Proteins 0.000 description 5
- 238000005457 optimization Methods 0.000 description 2
- 230000036962 time dependent Effects 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000012567 medical material Substances 0.000 description 1
- 230000001550 time effect Effects 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06311—Scheduling, planning or task assignment for a person or group
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- Operations Research (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- Quality & Reliability (AREA)
- Development Economics (AREA)
- Mathematical Physics (AREA)
- Pure & Applied Mathematics (AREA)
- Game Theory and Decision Science (AREA)
- Computational Mathematics (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a medical unmanned aerial vehicle scheduling model and a solving algorithm for considering sample timeliness, wherein the model establishment comprises the following processes: 1) Defining state parameters and variables of the unmanned aerial vehicle; 2) Determining an objective function of the model, wherein the objective function of the model is used for representing the time for minimizing the completion of all the inspection orders by the unmanned aerial vehicle; 3) Constraint conditions are determined. According to the invention, the medical unmanned aerial vehicle distribution taking sample timeliness into consideration under the field replacement scene is introduced, the optimal power replacement position of the unmanned aerial vehicle is decided according to the information such as sample inspection information, load, the residual capacity of the unmanned aerial vehicle and the like, the synchronous carrying out of sample collection and power replacement is realized, and the efficiency is improved. Under the constraint of sample timeliness, the overall detection efficiency and the service quality are improved. According to the invention, the problem is modeled as a mixed integer linear programming model, a branch pricing algorithm is designed for solving, and the mode of combining an accurate algorithm and a heuristic algorithm is adopted, so that the solving speed can be improved, and the problem can be solved to the optimal state.
Description
Technical Field
The invention relates to a medical unmanned aerial vehicle scheduling model for considering sample timeliness and a solving algorithm thereof.
Technical Field
In actual medical unmanned aerial vehicle scheduling, there are often scheduling demands of multiple community service points at the same time. Each community service point can generate scheduling requirements of samples to be detected, and each piece of scheduling requirement information comprises timeliness requirements of the samples to be detected and corresponding delivery unit hospitals. In the scheduling of nucleic acid detection sample unmanned aerial vehicles, energy consumption is a key factor affecting unmanned aerial vehicle mileage. In the conventional unmanned aerial vehicle delivery or truck-unmanned aerial vehicle combined delivery scheduling problem, it is often assumed that the battery power after each take-off of the unmanned aerial vehicle can always meet the delivery of all orders until the unmanned aerial vehicle returns to a warehouse or a truck. In addition, it is often assumed that the unmanned aerial vehicle performs power exchange or charging after completing the distribution of an order every time, and the actual residual capacity of the battery is not considered. This mode, while capable of performing scheduling tasks, is inefficient in battery usage. And each time of power change has power change time, which leads to the increase of the whole inspection time and influences the distribution efficiency and the timeliness of the samples.
By setting up nucleic acid service points in all communities in the city, the medical staff collects the nucleic acid samples to be detected on site and then sends the nucleic acid samples to all detection institutions for nucleic acid detection. After each community service point collects samples to be detected, unmanned aerial vehicles are used for sequentially conveying the samples of each community to a specified detection mechanism for detection. Further, since the validity of the sample gradually decreases over time, the timeliness of the sample is considered herein, thereby ensuring the quality of the sample for inspection. According to the invention, medical unmanned aerial vehicle distribution taking sample timeliness into consideration under the field change prospect is introduced. Each community service point is provided with a matched unmanned station, and the unmanned aerial vehicle can collect samples to be detected at each unmanned station. The unmanned aerial vehicle collects samples to be detected at a plurality of community service points by considering capacity constraint and energy consumption constraint of the unmanned aerial vehicle. According to the information such as sample sending and detecting information, load, unmanned aerial vehicle's residual capacity, etc. come the best position of trading of decision unmanned aerial vehicle to realize collecting the sample and trade the synchronous going on of electricity, raise the efficiency. Under the constraint of sample timeliness, the overall detection efficiency and the service quality are improved. The invention models the problem as a mixed integer linear programming model, and designs a branch pricing algorithm to solve the problem. Wherein the main problem is a set coverage model and the sub-problem is a constrained basic shortest path problem. In the solving of the sub-problem, a mode of combining an accurate algorithm and a heuristic algorithm is adopted. The designed label algorithm and heuristic algorithm can be effectively solved.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention aims to provide a medical unmanned aerial vehicle scheduling model and a solving algorithm for considering sample timeliness.
The medical unmanned aerial vehicle scheduling model for considering sample timeliness is established by the following steps:
3.1 symbol definition:
u: = {1,..k } unmanned numbering set;
h: = {1,..h } set of initial positions of drones;
c: = { h+1,.,. H+n } community service point set, i.e., a set of demand generation points for sample collection to be tested;
d: = { h+n+1,..h+2n } sample is sent to the set of unit sites;
e: = { h+2n+1} unmanned virtual endpoint;
n: set of all nodes of = { H } C } D };
a: the set of arcs between nodes = { (i, j) |i E n\ { E }, j E n\h, i +.j };
g: = (N, a) directed graph set;
d i : load demand for unmanned aerial vehicle inspection;
L i : unmanned aerial vehicle leaves node i and carries load;
f: unmanned aerial vehicle dead weight;
l: maximum load (kg) of the unmanned aerial vehicle;
[e i ,l i ]: the unmanned aerial vehicle sends out the time window of examining the demand, unmanned aerial vehicle must accomplish the taking and delivering of medical supplies in the time window; wherein,,
e i ,l i respectively representing the earliest and latest start service times;
[η min ,η max ]: sample timeliness; wherein eta min ,η max Representing the minimum and maximum time that the sample is effective, respectively;
t ij : time of flight from node i to node j;
ρ: full electric quantity of the unmanned aerial vehicle;
alpha: an energy consumption factor;
t bs : the power change time of the unmanned aerial vehicle;
delta: sample timeliness factor;
gamma: energy-consuming time factor
M: a very large positive integer (e.g., M is greater than or equal to 100000);
3.2 variables:
y i : if the unmanned aerial vehicle performs power conversion operation at the node i, y is i =1, otherwise z i =0;
3.3 model objective function:
the problem is defined in a directed graph g= (N, a), where n= {1, 2..h+2n } is the set of all nodes, a= { (i, j) |i E n\e, j E n\h, i not equal j } is the set of arcs between nodes. We define h= { 1..h } as the drone initial position, u= { 1..k } as the drone number set. C= { h+1,..h+n } is defined as the community service point set, i.e. the demand generation point set for sample collection to be measured. D= { h+n+1,..h+2n } is a collection of sample unit sites, each requirement containing one sample collection site and unit site. E=k+2n+1 is defined as the drone virtual endpoint.
There is a sample load demand for each service pointWherein the load for the demand generation point is defined as +.>The requirement of the corresponding detection unit is +.>Setting the requirement of an initial node of the unmanned aerial vehicle>Each requirement has a requirement of time window, the time window of requirement i is defined as +.>The unmanned aerial vehicle must accomplish the fetching and delivering of medical supplies in the prescribed time. Furthermore, each sample has its own timeliness requirement, defined +.>For sample timeliness of unmanned aerial vehicle k reaching point i, each sampleThe product must meet the requirements of timelinessDefining L as the maximum effective load of the unmanned aerial vehicle, and F as the self-frame weight of the unmanned aerial vehicle. Definitions->And (5) loading for the unmanned aerial vehicle to leave the node i. />For the time of arrival of unmanned plane k at node i (i.e. +.>Representing the time at which drone k starts serving node i), the arrival time of each node must satisfy the constraints of the time window. ρ is the amount of electricity that the unmanned aerial vehicle is full. Defining eta as time unit factor, delta as time effect factor, t ij T is the time from node i to node j bs Is the power change time of the unmanned aerial vehicle. Definitions->For the power consumption of the unmanned aerial vehicle to reach node i, < >>And (5) consuming electricity for the unmanned aerial vehicle to leave the node i. y is i E {0,1} is a decision variable of whether the unmanned aerial vehicle is powered on at the service point i, when y i When=1, it means that the unmanned plane is powered up at service point i, at this time +.>Otherwise->Definitions->As decision variables, when->Representing that the unmanned aerial vehicle k flies through the arc (i, j); otherwise, indicating that the unmanned plane k does not fly through the arc(i, j). Based on the symbol definition, the section establishes an unmanned aerial vehicle scheduling model taking energy consumption constraint into consideration under the field change scene, and the specific model is expressed as follows:
the objective function in this design modeling is to represent minimizing the time for the drone to complete all of the inspection orders.
3.4 determining constraint conditions:
constraint expression (2) indicates that the sample for any one demand must be serviced by the same drone. Constraint expression (3) indicates that the drone must start from the starting position. Constraint expression (4) ensures that each sample is collected and sent for inspection and that only one drone is serviced. Constraint expression (5) represents a flow-level constraint for a node. Constraint expression (6) indicates that each drone must eventually return to the virtual junction, i.e., to the drone virtual destination. Constraint expressions (7) - (8) represent the time-dependent constraint of the samples after sample decay. Constraint expression (9) represents the time order of access before and after the node. Constraint expression (10) represents unmanned payload constraint logic before and after a node. Constraint expression (11) indicates that the same inspection-required unmanned aerial vehicle must go to the community service point to collect the sample before being inspected. Constraint expression (12) represents a power constraint for the node access order. Constraint expressions (13) - (16) represent logical constraints of the post-power-change electric quantity. Constraint expression (17) represents a time window constraint of the unmanned access node. Constraint expression (18) represents a node load upper limit constraint. Constraint expressions (19) - (20) represent variable type constraints.
3.5 design of branch pricing algorithm:
the mixed integer linear programming models (1) - (20) designed above can be directly solved by using a commercial solver CPLEX, but CPLEX can only solve the sample-sending-inspection unmanned aerial vehicle scheduling problem of community service points under a small scale. When the number of community service points is medium-large, the solver CPLEX cannot obtain a viable solution in a given time. Based on this, a branch pricing algorithm is designed to solve the medical drone scheduling problem of medium and large scale. In the branch pricing algorithm, the questions are first modeled as a set coverage model and pricing sub-questions. In the pricing sub-problem, a pattern of combining a precise algorithm and a heuristic algorithm is designed. In the process of solving the sub-problem, the heuristic algorithm can effectively improve the solving speed, quicken finding out the path with the reduced cost as the negative number, and then add the path into the main problem until the path with the reduced cost as the negative number cannot be found out in the sub-problem. In this step, the accurate algorithm based on dynamic programming can ensure that the sub-problem is solved to the optimum, but the solving speed of the accurate algorithm is not fast. The heuristic algorithm has a faster solving speed, is convenient for quickly finding a path with a reduction cost of negative number, but cannot guarantee that the solving is optimal. The mode of combining the accurate algorithm and the heuristic algorithm can not only improve the solving speed, but also ensure that the sub-problems are solved to the optimal.
The aggregate coverage model modeling process is as follows:
3.5.1 symbol definition:
TABLE 1 set overlay model symbols and meanings
3.5.2 set coverage model:
the aggregate coverage model is constructed as follows: since the solution set R corresponding to the path R is very large, the solution cannot be directly performed at this time. With the column-based generation (Column Generation) framework, the aggregate coverage model is first considered in one partial solution space, and then the solution space aggregate is gradually expanded by solving the pricing sub-problem. And limiting the solution set in the set coverage model based on the limitation to obtain a limited main problem model. Define R' as a subset of the set Ω, at this pointVariable e in the subject problem to be restricted r Relaxation, linear relaxation models (21) - (24) of the following constrained main problem are obtained. />
(RMLP)min C r e r #(21)
Definition of the definitionFor the dual variable of constraint (22), +.>As a dual variable of constraint (23), it can be analyzed that the set-based overlay model is better than the set partition model.
3.5.3 pricing sub-problem:
in the pricing sub-problem, it is necessary to find all paths for which the reduce cost is negative, whereThe path is then added to the constrained main problem, and the sub-problem is updated by the dual variables obtained after solving the constrained main problem. The solution is repeated until all paths with the reduction cost negative are found. Solving the pricing sub-problem is effectively solving a basic shortest path problem with resource constraints (Elementary Shortest Path Problem with Resource Constraints ESPPRC).
According to the definition above, the reduce cost of path r in the sub-problem can be constructed by the dual variables in RMLP. In the solving of the sub-problem, the cost of the arcs in the sub-problem is updated by the dual variables. The reduce cost defining path r is:
wherein:
update formula (25) is rewritten as the following formula (28):
A' represents a set of arcs between nodes of a child problem;
n' represents the set of all nodes of the child problem.
After the pricing sub-problem is built, a design algorithm is needed to solve. The combination of a label algorithm based on dynamic programming and two heuristic algorithms is designed for the ESPPRC problem.
3.6 sub-problem labeling algorithm:
in the current broad sense, ESPPRC is an NP-hard problem. In this context, a combination of dynamic programming and heuristic algorithms is used for solving. First, a label algorithm based on dynamic programming concept is designed. The solution framework algorithm can be expressed as: from one source node, each node along the way generates several labels. In the expansion process, the same node has a plurality of different labels, and the dominant rule is needed to be used for pruning, so that the search space is reduced.
Firstly, defining a label L, wherein the information of the label L mainly comprises the node position of the current label: p (L); t (L) is the time of arrival of the tag L at the current location; w (L) is the accumulated load of the label L; b (L) is the current residual capacity of the label L; c (L) is the accumulated consumption cost of the label L; s (L) is the sample timeliness of label L. In addition, O (L) is defined as a node set of the label L, wherein only the picking of goods is finished and the delivery of goods is not finished yet; v (L) is a set of nodes that have completed service; UR (L) is a node where the label L cannot be extended. Pseudo code for the tag algorithm as shown in algorithm 1 below, the extension and dominance rules of the tag algorithm will be set forth in the following section.
3.6.1 expansion of tags:
when the tag L is extended back to the node j, the tag L may be extended toward the node j if the following condition is satisfied in the extension. If and only if the following conditions hold:
t(L)+t p(L),j ≤l j #(29)
w(L)+d j ≤L#(30)
b(L)-ep p(L),j ≥0#(31)
s(L)-sc p(L),j ≥η min #(32)
wherein ep a(L),j Representing power consumption from the current tag position p (L) to node j, sc p(L),j Indicating the timeliness, t, of the post-sample from the current tag position p (L) to node j p(L),j Meaning the time from the tag p (L) to the node j; equation (29) represents extending the time window constraint that needs to be satisfied by the node; equation (30) indicates that the upper load limit of the unmanned aerial vehicle is not allowed to be exceeded in the expansion; equation (31) represents that the power constraint needs to be satisfied in the expansion; equation (32) represents the sample timeliness constraint required in the extension.
If an extension from the current label L to node j is a viable extension, then a new label L 'will be generated at node j, the information of label L' being as follows:
p(L′)=j#(42)
t(L′)=max{p j ,t(L)+t p(L),j }#(43)
l(L′)=l(L)+d j #(44)
c(L′)=c(L)+t p(L),j #(45)
b(L′)=b(L)-ec p(L),j #(46)
formulas (42) - (46) set the node location, time, load, cost, and power of the new tag; formulas (47) and (48) are used to update the sets O and V of labels L'.
3.6.2 tag dominance rules:
in the process of solving the pricing sub-problem by the label setting algorithm, a large number of labels are generated on each node. If the labels on the nodes are not pruned reasonably, the difficulty of solving the sub-problems is increased. According to the label information designed in the prior art, each label carries information such as own resources and the like. On the same node, the label deletion is stopped by designing the label dominant rule. The reasonable label dominant rule can effectively reduce the solving difficulty of the sub-problem, thereby improving the solving efficiency.
p(L)=p(L′)#(49)
c(L)≤c(L′)#(50)
t(L)≤t(L′)#(51)
b(L)≥b(L′)#(52)
Equation (49) indicates that the current positions of the tag L and the tag L' agree; equation (50) shows that label L costs less than label L' ; Equation (51) shows that the accumulated time of tag L is shorter than tag L'; equation (52) indicates that the remaining power of the tag L is greater than the tag L'.
Tag algorithm pseudocode:
3.7 sub-problem heuristic:
when the pricing sub-problem is solved, the problem can be solved to the optimal degree by using the label algorithm, but the solving speed of the label algorithm is slow. In order to improve the speed of solving the sub-problem, a heuristic algorithm is designed to solve. It is well known that heuristic algorithms, while capable of increasing the solution speed, cannot guarantee that all paths with negative reduction costs in the sub-problem are found. According to the respective advantages of the label algorithm and the heuristic algorithm, a mode of combining the label algorithm and the heuristic algorithm is adopted. In the initial stage of solving, a heuristic algorithm is used for solving. As long as the heuristic can find paths for which the reduce cost is negative, the found paths are added to the restricted master question. With the sub-problem being solved, the heuristic starts to converge at this time, and a column with a negative reduction cost cannot be found out quickly. But in this case, there may still be a path where the reduced cost is negative, but this cannot be found using heuristic algorithms. At this point, the labeling algorithm begins solving the sub-problem, and the labeling algorithm is used to ensure that all paths with all reduce costs negative are found.
In solving the sub-problem, firstApplied to the solution is a heuristic algorithm. The nodes are randomly selected from the path for removal and then reinserted in a random manner. The removal factor and the insertion factor are generated in a random manner, and the reception criterion is adopted by the Metropolis standard. Given an initial solution sol and a search scale scal, an initial optimal solution is first set as the initial solution sol best =sol, define the current solution sol cur =sol. Algorithm iterations are then started. Definition cur sol For the temporary solution, then randomly de-cur sol Selecting the scal nodes for removing operation. Inserting the removed node into the path in a random manner, and comparing target obj (sol) of sub-problem heuristic after the insertion is completed best ) And obj (cur) sol ). If obj (sol) best )>obj(cur sol ) Then the current optimal solution sol best =cur sol . Otherwise, adopting Metropolis standard to receive the current solution, repeating the above steps until the algorithm meets the termination condition, and then outputting the optimal solution sol best . Heuristic pseudocode is shown below.
Heuristic algorithm pseudocode:
the branch pricing algorithm is composed of a column generating algorithm and a branch delimitation algorithm, and after the column generating algorithm is solved, a linear relaxation solution is obtained. The solution at this point is a solution at the root node in branch-and-bound, often not an integer solution. For an integer programming problem, branches are typically used to process when non-integer solutions are obtained. In the branching process, by branching a non-integer solution into two sub-problems, then solving again based on the sub-problems until an integer solution is obtained. In branch-and-bound solutions, each of the integer and non-integer solutions is an upper bound and a lower bound. And solving through continuous iteration of the algorithm until the gap of the upper and lower boundaries is 0, wherein the obtained integer solution is the optimal solution. The solution flow of the branch pricing algorithm is shown in fig. 1 below, and is mainly composed of column generation and branch delimitation.
Description of the drawings:
FIG. 1 is a flow chart of a branch pricing algorithm;
fig. 2 is a schematic illustration of medical drone scheduling.
Description of the preferred embodiments
The invention will be further described with reference to examples and drawings, to which reference is made by way of illustration, but not limitation, for the understanding of those skilled in the art.
An unmanned plane distribution network optimization solving algorithm for urban instant distribution has the following modeling process:
4.1 symbol definition:
u: = {1,..k } unmanned numbering set;
h: = {1,..h } set of initial positions of drones;
c: = { h+1,.,. H+n } community service point set, i.e., a set of demand generation points for sample collection to be tested;
d: = { h+n+1,..h+2n } sample is sent to the set of unit sites;
e: = { h+2n+1} unmanned virtual endpoint;
n: set of all nodes of = { H } P } D };
a: the set of arcs between nodes = { (i, j) |i E n\ { E }, j E n\h, i +.j };
g: = (N, a) directed graph set;
d i : load demand for unmanned aerial vehicle inspection;
L i : unmanned aerial vehicle leaves node i and carries load;
f: unmanned aerial vehicle dead weight;
l: maximum load (kg) of the unmanned aerial vehicle;
[e i ,l i ]: the unmanned aerial vehicle sends out the time window of examining the demand, unmanned aerial vehicle must accomplish the taking and delivering of medical supplies in the time window; wherein,,
e i ,l i respectively representing the earliest and latest start service times;
[η min ,η max ]: sample timeliness; wherein eta min ,η max Representing the minimum and maximum time that the sample is effective, respectively;
t ij : time of flight from node i to node j;
ρ: full electric quantity of the unmanned aerial vehicle;
alpha: an energy consumption factor;
t bs : the power change time of the unmanned aerial vehicle;
delta: sample timeliness factor;
gamma: energy-consuming time factor
M: a very large positive integer (e.g., M is greater than or equal to 100000);
4.2 variables:
y i : if the unmanned aerial vehicle performs power conversion operation at the node i, y is i =1, otherwise z i =0;
4.3 model objective function:
the objective function in this design modeling is to represent minimizing the time for the drone to complete all of the inspection orders.
4.4 constraint:
constraint (2) indicates that for any one desired sample must be serviced by the same drone ;
Constraint (3) indicates that the drone must start from the starting position ;
Constraint (4) ensures that each sample is collected and submitted and that only one unmanned aerial vehicle is serviced ;
Constraint (5) represents a flow-leveling constraint for a node ;
Constraint (6) indicates that each drone must eventually return to the virtual meeting point ;
Constraints (7) - (8) represent the time-dependent constraints of the sample after sample decay ;
Constraint (9) represents the chronological order of access before and after a node ;
Constraint (10) represents unmanned aerial vehicle payload constraint logic around a node ;
Constraint (11) indicates that the same unmanned aerial vehicle with inspection requirements must go to a community service point to collect samples before inspection ;
Constraint (12) represents a power constraint for node access order ;
Constraints (13) - (16) represent logical constraints of post-power-change power ;
Constraint (17) represents a time window constraint of the unmanned aerial vehicle access node ;
Constraint (18) represents a node load upper limit constraint ;
Constraints (19) - (20) represent variable type constraints.
4.5 aggregate coverage model:
according to the symbol definition, the following set coverage model is constructed:
(RMLP)min C r e r #(21)
4.5 pricing sub-problem:
4.6 expansion of tags:
t(L)+t p(L),j ≤l j #(29)
w(L)+d j ≤L#(30)
b(L)-ep p(L),j ≥0#(31)
s(L)-sc p(L),j ≥η min #(32)
example 1
As an example, this embodiment 1 was performed to verify the scientificity and effectiveness of the above-described solution algorithm for unmanned aerial vehicle distribution network optimization for urban instant distribution:
as shown in fig. 2 below, the community service point C1 needs to send the sample to the detection unit H1 for detection, and at this time, the service point C5 needs to send the sample to be detected to the detection unit H1 for detection, but at this time, the location of the service point C5 is temporarily free from the unmanned aerial vehicle for detection service. The drone D1 starts from the service point C1, then goes to the C5 service point, and finally sends the samples of the C1 and C5 service points to the H1 detection unit for detection. Because the residual electric quantity of the unmanned aerial vehicle D1 can meet the service of the present time, the unmanned aerial vehicle D1 does not need to perform power conversion service at a C5 service point. Similarly, the service points C2 and C3 have the requirement that a sample to be detected is sent to an H2 detection unit for inspection, at the moment, the unmanned aerial vehicle D2 starts from the service point C2, the electric quantity after reaching the service point C3 cannot support the unmanned aerial vehicle D2 to fly to the H2 detection unit, so that the unmanned aerial vehicle D2 changes electricity at the service point C3, and the timeliness requirement of the sample is not influenced after the time accumulation of the electricity change. The C4 service point needs the H2 detection unit to send a batch of medical detection articles to the service point, so that the unmanned aerial vehicle D3 starts from the detection unit H2, carries medical materials to fly to the C4 service point, and meets the time window requirement of the C4 service point. In the medical unmanned aerial vehicle scheduling process, time windows of requirements and timeliness constraints of samples need to be considered. The flight route and the power change of the unmanned aerial vehicle can influence the completion time of the requirements, and the power change must be considered while the path planning is carried out.
Claims (7)
1. A medical unmanned aerial vehicle scheduling model for taking sample timeliness into account, characterized in that the model is built up by the following procedures:
1) The unmanned aerial vehicle state parameters and variables are defined, and parameter symbols are defined as follows:
u: = {1,..k } unmanned numbering set;
h: = {1,..h } set of initial positions of drones;
c: a= { h+1, …, h+n } community service point set, namely a demand generation point set for collecting a sample to be tested;
d: = { h+n+1, …, h+2n } sample is sent to the unit site collection;
e: = { h+2n+1} unmanned virtual endpoint;
n: set of all nodes of = { H } C } D };
a: an arc set between nodes = { (i, j) |i E n\ { E }, j E n\h, i +.j };
g: = (N, a) directed graph set;
d i load demand of unmanned aerial vehicle sending inspection;
L i the unmanned aerial vehicle leaves the node i and carries the load;
f, unmanned aerial vehicle dead weight;
l is the maximum load (kg) of the unmanned aerial vehicle;
[e i ,l i ]the unmanned aerial vehicle send out the time window of examining the demand, unmanned aerial vehicle must finish the taking and delivering of the medical supplies in the time window; wherein e i ,l i Respectively representing the earliest and latest start service times;
[η min ,η max ]sample timeliness; wherein eta min ,η max Representing the minimum and maximum time that the sample is effective, respectively;
t ij time of flight from node i to node j;
rho is full electric quantity of the unmanned aerial vehicle;
alpha is an energy consumption factor;
t bs the power change time of the unmanned aerial vehicle;
delta is the sample timeliness factor;
gamma energy-consuming time factor
M: a very large positive integer;
y i If the unmanned aerial vehicle performs power conversion operation at the node i, y i =1, otherwise z i =0;
2) Determining an objective function of the model:
the objective function of the model is to represent that the time for the unmanned aerial vehicle to complete all the inspection orders is minimized, and the operation mode is shown in a formula (1):
3) Determining constraint conditions including the following constraint process
1. The sample which is restricted to any one delivery requirement must be served by the same unmanned aerial vehicle, namely, each delivery requirement has a pick-up point and a delivery point, and the two positions must be served by one unmanned aerial vehicle;
2. the restraining unmanned aerial vehicle must start from a starting position;
3. restricting the collection and the inspection of each sample to be detected to be all and only one unmanned aerial vehicle to serve;
4. constraining the flow of the node balance;
5. constraining each unmanned aerial vehicle to finally return to the virtual unmanned aerial vehicle terminal;
6. the effectiveness of the sample to be tested gradually decreases along with the time, so that the timeliness of the sample after the sample is attenuated is restrained;
7. the time sequence of the front and back access of the nodes is restrained, namely the earliest service time of the next node is larger than or equal to the service time of the previous node and the flight time of the unmanned aerial vehicle;
8. establishing effective load constraint logic of unmanned aerial vehicles before and after nodes;
9. unmanned aerial vehicles which restrict the same inspection requirement can be inspected after the unmanned aerial vehicles go to a community service point to collect samples;
10. constraining the electric quantity of the node access sequence;
11. establishing logic constraint of electric quantity after power change of the unmanned aerial vehicle;
12. establishing time window constraint of access nodes of the unmanned aerial vehicle;
13. and establishing node load upper limit constraint.
2. A medical unmanned aerial vehicle scheduling model for taking sample timeliness into account as claimed in claim 1, wherein in step 3), the basic constraints of the course of action of the unmanned aerial vehicle are as shown in equations (2) - (6):
expression (2) indicates that for any one desired sample must be serviced by the same drone; expression (3) indicates that the drone must start from the starting position; expression (4) ensures that each sample is collected and sent for inspection and that only one drone is serviced; expression (5) represents the flow balance constraint for the node. Expression (6) indicates that each drone must eventually return to the virtual junction, i.e., to the drone virtual destination.
3. A medical unmanned aerial vehicle scheduling model for taking sample timeliness into account as claimed in claim 1, wherein in step 3), the basic constraints on sample timeliness are as shown in formulas (7) - (8):
expressions (7) - (8) represent the timeliness constraint of the samples after sample decay.
4. A medical unmanned aerial vehicle scheduling model for taking sample timeliness into account as claimed in claim 1, wherein in step 3) the basic constraints on the time sequence, load and power of the unmanned aerial vehicle access nodes are shown in equations (9) - (18), and the constraints on the type of variables involved are shown in equations (19) - (20):
expression (9) represents the chronological order of the access before and after the node,representing the time when the unmanned plane k starts to serve the node i;
expression (10) represents unmanned aerial vehicle payload constraint logic before and after the node;
expression (11) shows that the same inspection-for-purpose unmanned aerial vehicle must first go to a community service point to collect a sample and then can be inspected;
expression (12) represents a power constraint of the node access order;
expressions (13) - (16) represent logical constraints of the amount of power after power conversion;
expression (17) represents a time window constraint of the unmanned access node;
expression (18) represents a node load upper limit constraint;
expressions (19) - (20) represent variable type constraints.
5. A solution algorithm for a medical drone dispatch model taking sample timeliness into account as claimed in claim 1, characterised by comprising the following process:
a branch pricing algorithm is designed to solve the scheduling problem of the medical unmanned aerial vehicle in a medium-large scale; in a branch pricing algorithm, firstly, the medical unmanned aerial vehicle scheduling model is optimized and modeled again to form a set coverage model and a pricing sub-problem; in the pricing sub-problem, a mode of combining an accurate algorithm and a heuristic algorithm is designed for solving;
1) The aggregate coverage model modeling process is as follows:
first, set of overlay model symbols and meanings
R: all feasible path sets;
C r : the cost of path r;
n ir : the number of times node i is accessed by path r;
e r : a variable representing 0 or 1, 1 if path r is selected, and 0 otherwise;
second, build a set coverage model:
because the solution set R corresponding to the path R is huge, the solution cannot be directly carried out at the moment; adopting a frame based on column generation, firstly considering a set coverage model in a partial solution space, and then gradually expanding a solution space set by solving a pricing sub-problem; limiting the solution set in the set coverage model based on the limitation to obtain a limited main problem model; define R' as a subset of set R, at which timeVariable e in the subject problem to be restricted r Relaxation, linear relaxation model equations (21) - (24) get the following constrained main problem:
(RMLP)minC r e r #(21)
definition of the definitionFor the dual variable of constraint (22), +.>As the dual variables of constraint (23), the set-based coverage model is better than the set partition model;
third, pricing sub-problem:
in the pricing sub-problem, it is necessary to find paths where all reduce costs are negative, where reduceThen adding the path into the limited main problem, and updating the sub-problem through the dual variables obtained after solving the limited main problem; repeatedly solving until all paths with the reduction cost as negative numbers are found; solving the pricing sub-problem is actually solving a basic shortest path problem with resource constraints;
according to the definition, the reduce cost of the path r in the sub-problem can be constructed through the dual variables in the RMLP; in the solving of the sub-problem, updating the cost of the arcs in the sub-problem by means of the dual variables; the reduce cost defining path r is:
wherein:
update formula (25) is rewritten as the following formula (28):
A' represents a set of arcs between nodes of a child problem;
n' represents a set of all nodes of the child problem;
2) And solving the pricing sub-problem based on a method of combining a label algorithm and a heuristic algorithm of dynamic programming.
6. A solution algorithm for a medical drone dispatch model taking sample timeliness into account as claimed in claim 5, wherein the process of the labeling algorithm of the sub-problem is as follows:
designing a label algorithm based on a dynamic programming idea, wherein the solution framework algorithm can be expressed as follows: starting from a source node, each node along the way generates a plurality of labels; in the expansion process, a plurality of different labels are arranged on the same node, and a dominant rule is needed to be used for pruning, so that the search space is reduced;
firstly, defining a label L, wherein the information of the label L mainly comprises the node position of the current label: p (L); t (L) is the time of arrival of the tag L at the current location; w (L) is the accumulated load of the label L; b (L) is the current residual capacity of the label L; c (L) is the accumulated consumption cost of the label L; s (L) is the sample timeliness of the label L; in addition, O (L) is defined as a node set of the label L, wherein only the picking of goods is finished and the delivery of goods is not finished yet; v (L) is a set of nodes that have completed service; UR (L) is a node whose label L cannot be extended;
1) Expansion of the tag:
when the tag L extends back to the node j, the tag L may extend to the node j if and only if the following condition is met in the extension:
t(L)+t p(L),j ≤l j #(29)
w(L)+d j ≤L#(30)
b(L)-ep p(L),j ≥0#(31)
s(L)-sc p(L),j ≥η min #(32)
wherein ep a(L),j Representing power consumption from the current tag position p (L) to node j, sc p(L),j Indicating the timeliness, t, of the post-sample from the current tag position p (L) to node j p(L),j Meaning the time from the tag p (L) to the node j; equation (29) represents extending the time window constraint that needs to be satisfied by the node; equation (30) indicates that the upper load limit of the unmanned aerial vehicle is not allowed to be exceeded in the expansion; equation (31) represents that the power constraint needs to be satisfied in the expansion; equation (32) represents the sample timeliness constraint required in the extension;
if an extension from the current label L to node j is a viable extension, then a new label L 'will be generated at node j, the information of label L' being as follows:
p(L′)=j#(42)
t(L′)=max{p j ,t(L)+t p(L),j }#(43)
l(L′)=l(L)+d j #(44)
c(L')=c(L)+t p(L),j #(45)
b(L′)=b(L)-ec p(L),h #(46)
formulas (42) - (46) set the node location, time, load, cost, and power of the new tag; formulas (47) and (48) are used to update the sets O and V of the label L';
2) Tag dominance rules:
in the process of solving the pricing sub-problem by the label setting algorithm, a large number of labels are generated on each node, and if the labels on the nodes are not reasonably pruned, the difficulty of solving the sub-problem is increased; according to the label information designed in the prior art, each label is provided with information of own resources; on the same node, label deletion is carried out by designing a label dominant rule; the reasonable label dominant rule can effectively reduce the solving difficulty of the sub-problem, thereby improving the solving efficiency;
p(L)=p(L′)#(49)
c(L)≤c(L')#(50)
t(L)≤t(L')#(51)
b(L)≥b(L')#(52)
equation (49) indicates that the current positions of the tag L and the tag L' agree; equation (50) indicates that label L is less costly than label L'; equation (51) shows that the accumulated time of tag L is shorter than tag L'; equation (52) indicates that the remaining power of the tag L is greater than the tag L'.
7. A solution algorithm for a medical drone scheduling model taking sample timeliness into account as claimed in claim 6, characterized in that the procedure of the heuristic algorithm of the sub-problem is as follows:
when the pricing sub-problem is solved, the problem can be solved to the optimal state by using a label algorithm, but the solving speed of the label algorithm is slow; in order to improve the speed of solving the sub-problem, a heuristic algorithm is designed to solve, and although the heuristic algorithm can improve the solving speed, paths with negative numbers of all reduce costs in the sub-problem cannot be found; according to the advantages of the label algorithm and the heuristic algorithm, a mode of combining the label algorithm and the heuristic algorithm is adopted; in the initial stage of solving, firstly, a heuristic algorithm is used for solving, and the found path is added into the limited main problem as long as the heuristic algorithm can find the path with the reduced cost as a negative number; with the continuous solving of the sub-problem, the heuristic starts to converge at this time, and a column with a negative number of the reduce cost cannot be quickly found out; in this case, there may still be a path with a negative reduction cost, but this cannot be found using heuristic algorithms; at this time, the label algorithm starts to solve the sub-problem, and the label algorithm is utilized to ensure that all paths with the reduction cost as negative numbers are all found;
when solving the sub-problem, the method is firstly applied to solving a heuristic algorithm; randomly selecting nodes from the paths to remove, and then reinserting the removed nodes in a random mode; the removal factor and the insertion factor are generated in a random manner, and the receiving criterion adopts the Metropolis standard; given an initial solution sol and a search scale scal, an initial optimal solution is first set as the initial solution sol best =sol, define the current solution sol cur =sol; then starting algorithm iteration; definition cur sol For the temporary solution, then randomly de-cur sol Selecting scal nodes to perform removing operation; inserting the removed node into the path in a random manner, and comparing target obj (sol) of sub-problem heuristic after the insertion is completed best ) And obj (cur) sol ) The method comprises the steps of carrying out a first treatment on the surface of the If obj (sol) best )>obj(cur sol ) Then the current optimal solution sol best =cur sol Otherwise, receiving the current solution by adopting a Metropolis standard; repeating the above steps until the algorithm meets the termination condition, and then outputting the optimal solution sol best ;
The heuristic algorithm has the following operation processes:
input: initial feasible solution: a sol; search scale: scal;
1) Initializing an optimal solution and a current solution: sol best =sol,cur sol =sol;
2) While loop, algorithm termination condition is not satisfied:
3) Current decor sol Copying: cur's' sol =cur sol ;
4) For solving cur' sol Performing a removal operation;
5) For solving cur' sol Performing an insertion operation;
6) Target obj (sol) if sub-problem heuristics best) >obj(cur' sol ) Then the current optimal solution is updated: sol best ='cur sol ;
7) If step 6) is not satisfied, then the current solution cur 'is received using the Metropolis criterion' sol ;
And (3) outputting: optimal solution sol best
The branch pricing algorithm is composed of a column generating algorithm and a branch delimitation algorithm, and after the column generating algorithm is solved, a linear relaxation solution is obtained. The solution at this point is a solution at the root node in branch-and-bound, often not an integer solution. For an integer programming problem, branches are typically used to process when non-integer solutions are obtained. In the branching process, by branching a non-integer solution into two sub-problems, then solving again based on the sub-problems until an integer solution is obtained. In branch-and-bound solutions, each of the integer and non-integer solutions is an upper bound and a lower bound. And solving through continuous iteration of the algorithm until the gap of the upper and lower boundaries is 0, wherein the obtained integer solution is the optimal solution.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211454497.5A CN116258311A (en) | 2022-11-21 | 2022-11-21 | Medical unmanned aerial vehicle scheduling model and solving algorithm for considering sample timeliness |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211454497.5A CN116258311A (en) | 2022-11-21 | 2022-11-21 | Medical unmanned aerial vehicle scheduling model and solving algorithm for considering sample timeliness |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116258311A true CN116258311A (en) | 2023-06-13 |
Family
ID=86679924
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211454497.5A Pending CN116258311A (en) | 2022-11-21 | 2022-11-21 | Medical unmanned aerial vehicle scheduling model and solving algorithm for considering sample timeliness |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116258311A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116777063A (en) * | 2023-06-20 | 2023-09-19 | 广东工业大学 | Two-dimensional boxing method based on one-tool cutting constraint and branch pricing algorithm |
-
2022
- 2022-11-21 CN CN202211454497.5A patent/CN116258311A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116777063A (en) * | 2023-06-20 | 2023-09-19 | 广东工业大学 | Two-dimensional boxing method based on one-tool cutting constraint and branch pricing algorithm |
CN116777063B (en) * | 2023-06-20 | 2024-02-27 | 广东工业大学 | Two-dimensional boxing method based on one-tool cutting constraint and branch pricing algorithm |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
James et al. | Online vehicle routing with neural combinatorial optimization and deep reinforcement learning | |
James | Two-stage request scheduling for autonomous vehicle logistic system | |
Meng et al. | The multi-visit drone routing problem for pickup and delivery services | |
CN107766994A (en) | A kind of shared bicycle dispatching method and scheduling system | |
Cheung et al. | Dynamic routing model and solution methods for fleet management with mobile technologies | |
Shi et al. | Memory-based ant colony system approach for multi-source data associated dynamic electric vehicle dispatch optimization | |
CN103345504B (en) | The operator construction method of single star scheduling | |
CN110929925B (en) | Electric vehicle charging station planning method based on environment temperature and user behaviors | |
Grzybowska et al. | Decision support system for real-time urban freight management | |
Pan et al. | Ridesharing: simulator, benchmark, and evaluation | |
Jiang et al. | A branch-and-price algorithm for large-scale multidepot electric bus scheduling | |
Lowalekar et al. | Zone path construction (zac) based approaches for effective real-time ridesharing | |
CN116258311A (en) | Medical unmanned aerial vehicle scheduling model and solving algorithm for considering sample timeliness | |
Wang et al. | Optimization of ride-sharing with passenger transfer via deep reinforcement learning | |
CN106600180A (en) | Method and system of transportation route optimization of logistics management system | |
Garcia et al. | Hybrid approach for the public transportation time dependent orienteering problem with time windows | |
Guan et al. | Ridesharing in urban areas: Multi-objective optimisation approach for ride-matching and routeing with commuters’ dynamic mode choice | |
Thirugnanasambandam et al. | Experimental analysis of ant system on travelling salesman problem dataset TSPLIB | |
Zhang et al. | A generalized multi-depot vehicle routing problem with replenishment based on LocalSolver | |
CN116384606A (en) | Scheduling optimization method and system based on cooperative distribution of vehicle unmanned aerial vehicle | |
Zhen et al. | Vehicle routing for customized on-demand bus services | |
Shafahi et al. | A matching-based heuristic algorithm for school bus routing problems | |
Jovanovic et al. | Matheuristic fixed set search applied to electric bus fleet scheduling | |
Agarwal et al. | Scalable algorithms for bicriterion trip-based transit routing | |
CN113988570A (en) | Multi-objective evolutionary algorithm-based tourism bus scheduling optimization method |
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 |