CN114003011B - Multi-load AGVS deadlock prevention task scheduling method - Google Patents
Multi-load AGVS deadlock prevention task scheduling method Download PDFInfo
- Publication number
- CN114003011B CN114003011B CN202111293432.2A CN202111293432A CN114003011B CN 114003011 B CN114003011 B CN 114003011B CN 202111293432 A CN202111293432 A CN 202111293432A CN 114003011 B CN114003011 B CN 114003011B
- Authority
- CN
- China
- Prior art keywords
- station
- agv
- trailer
- empty
- agvs
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM]
- G05B19/41865—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS], computer integrated manufacturing [CIM] characterised by job scheduling, process planning, material flow
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/30—Nc systems
- G05B2219/32—Operator till task planning
- G05B2219/32252—Scheduling production, machining, job shop
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
-
- 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
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/60—Electric or hybrid propulsion means for production processes
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Mathematical Physics (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Manufacturing & Machinery (AREA)
- Quality & Reliability (AREA)
- Automation & Control Theory (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
- General Factory Administration (AREA)
Abstract
The invention provides a multi-load AGVS deadlock prevention task scheduling method, and belongs to the technical field of automatic guided vehicle systems. Firstly, counting buffer space information of each station trailer and an idle AGV set in a system; then, generating a plurality of station feeding sequence schemes according to the station feeding sequence generation rule, and complementing the initial population according to the individual generation rule; then, grouping the idle AGVs according to an AGV grouping strategy; then, evaluating individuals in the population based on the empty trailer warehouse-returning task scheduling decision, the deadlock avoidance strategy and the full trailer delivery task scheduling decision; and finally, generating a new generation station feeding sequence population by using an intelligent optimization algorithm, and outputting an optimal individual as an optimal scheduling scheme through a compromise strategy when the preset iteration times are reached. The method can effectively avoid the first type of deadlock problem of the multi-load AGVS, and simultaneously improves the convergence rate of the algorithm while obtaining excellent solutions based on optimization of the initial population generation and algorithm improvement in population evolution.
Description
Technical Field
The invention relates to the technical field of automatic guided vehicle systems, in particular to a multi-load AGVS anti-deadlock task scheduling method.
Background
An AGV System (AGVS) consisting of a plurality of automatic guided vehicles (Automatic Guided Vehicle, AGVs) is a multi-mobile robot System for material distribution, has the advantages of high flexibility and automation degree, low running noise, strong System expansion capability and the like, becomes a marked intelligent logistics device, and is more widely applied to the fields of warehouse logistics, manufacturing workshops, port transportation and the like. The method is used for solving a plurality of problems such as guide path network design, task scheduling, traffic management, path planning and the like when the AGVS planning design and the optimization control are performed. The AGVS task scheduling is always a hotspot in the research field of AGVS, and a plurality of scheduling algorithms exist, but a plurality of scheduling algorithms take the traditional single-load AGVS as a research object. In recent years, the multi-load AGVS has received wide attention from industry and academia due to the advantages of strong single-vehicle transportation capability, low traffic jam rate and the like.
The existing multi-load AGVS task scheduling problem does not consider the constraint of the number of trailers that can be accommodated by each workstation. This constraint belongs to the buffer capacity constraint and has been demonstrated in the study of single load AGVS: ignoring the constraint causes a system deadlock phenomenon, and finally, the effectiveness of a scheduling rule is lost, which is a key and difficult problem to be solved when AGVS task scheduling is performed. The triggering condition of the AGVS deadlock phenomenon is closely related to the material distribution flow and the application environment. The existing research on the first type of deadlock takes single-load AGVS under the application environment of a job shop as a research object, and at present, research reports on the phenomenon of the first type of deadlock of multi-load AGVS are not yet seen. Aiming at the multi-load AGVS material distribution flow under the complex product assembly manufacturing application environment, the deadlock phenomenon triggering condition is ascertained, and a corresponding deadlock avoidance strategy is designed and integrated into the multi-load AGVS task scheduling method.
Therefore, the invention provides a multi-load AGVS anti-deadlock task scheduling method.
Disclosure of Invention
In order to solve the problems, the invention provides a multi-load AGVS anti-deadlock task scheduling method aiming at the problem of scheduling of large-scale auxiliary material distribution multi-load AGVS tasks in a vehicle assembly application environment.
In order to achieve the above purpose, the present invention provides the following technical solutions.
A multi-load AGVS deadlock prevention task scheduling method comprises the following steps:
s1: counting buffer space information of the trailers at each station and an idle AGV set;
s2: generating a plurality of station feeding sequence schemes according to the station feeding sequence generation rules, taking the station feeding sequence schemes as part of individuals in the initial population of the intelligent optimization algorithm, and completing the initial population according to the individual generation rules;
s3: according to an AGV grouping strategy, grouping idle AGVs in the system to form an AGV set for executing an empty trailer warehouse-returning taskAnd AGV set for performing full trailer delivery tasks +.>
S4: each individual in the population was evaluated: firstly, based on an individual station feeding sequence scheme, according to the scheduling decision of the empty trailer warehouse-returning task, the method comprises the following steps ofAGVs are distributed to the empty trailer warehouse-returning tasks to form an empty trailer warehouse-returning task scheduling scheme; then, based on the station feeding sequence scheme, according to the deadlock avoidance strategyWhether the full trailer delivery task causes the execution AGV to become a blocked AGV is determined in a slightly sequential manner, if yes, the task is skipped, if no, the execution AGV is stopped according to the full trailer delivery task scheduling decision, and the execution AGV is stopped>AGVs are distributed for the tasks to form a full-load trailer distribution task scheduling scheme; finally, calculating an objective function of the individual according to the formed empty trailer warehouse-returning task scheduling scheme and the full trailer delivery task scheduling scheme, and taking the objective function value as the individual evaluation;
s5: and (4) taking an intelligent optimization algorithm as an optimization flow, taking an initial population as a starting population, continuously evolving to generate a new generation station feeding sequence population according to an evolution strategy based on the evaluation of individuals, outputting an optimal individual as an optimal scheduling scheme through a compromise strategy when the preset iteration times are reached, and otherwise, entering step S4.
Further, the station feeding sequence generation rule is one or more of a plurality of priority stations with current empty trailers, a plurality of priority stations with current non-empty trailers, a plurality of priority stations with current part sleeves, a multi-attribute rule based on the current state of the system, a plurality of priority stations with the number of trailers in the next feeding of the station, a plurality of priority stations with the number of non-empty trailers in the next feeding of the station, a plurality of priority stations with the number of part sleeves in the next feeding of the station, and a multi-attribute rule based on the state of the system in the next feeding of the station.
Further, the stations currently empty with a plurality of trailers are prioritized, specifically:
wherein: ∈ indicates that the rule will be for each station W i According to the utility value thereofDetermining the station feeding sequence from high to low; />Indicating station W i The number of empty trailers at the current time; />Indicating AGV G m From station W i The number of empty trailers loaded; n (N) G The AGV total number in the system; />Indicating station W i The number of empty trailers allocated to carry AGVs but not yet mounted;
the station is preferential to the ones with less number of currently non-empty trailers, specifically:
wherein: ∈ indicates that the rule will be for each station W i According to the utility value thereofDetermining the station feeding sequence from low to high; station W i The number of non-empty trailers at the current time; />Indicating AGV G m The current delivery to the station W i A number of full trailers;indicating that the AGV has been transported without being delivered to station W i A number of full trailers;
the station is preferential to the ones with less current parts, specifically:
wherein:indicating station W i The number of auxiliary materials at the current moment; k (K) i Indicating station W i The maximum auxiliary material sleeve number which can be loaded by each full trailer;
the multi-attribute rule based on the current state of the system specifically comprises the following steps:
the number of trailers is more than one when the next feeding is performed at the station, specifically:
wherein: tau (0, i) represents the AGV traveling directly from the auxiliary material inventory area to the job site P i Average time of (2); ζ is the production beat of the vehicle assembly line;
the fewer priority of the non-empty trailers is given when the station is fed next time, specifically:
wherein: ceil () represents a rounding up operation;
the station is preferential to the few part sets when the next material is fed, and specifically comprises the following steps:
the multi-attribute rule based on the system state when the station is fed next time is specifically as follows:
further, the deadlock avoidance strategy is to ensure that the system is in any timeThe residual space of the trailer at any station is not negative, so that the AGV is prevented from being blocked in the system; when the station W i The remaining space R of the trailer which can be accommodated i When it is 0, it continues to be station W i The AGV is allocated to the full trailer to be allocated, the execution AGV becomes a blocking AGV, and the station W is stopped i And (3) distributing AGVs for the to-be-distributed full trailer distribution tasks, wherein the blocking AGVs are AGVs for bearing the full trailer distribution tasks of stations with the residual space of the trailer buffer zone being 0.
Further, the empty trailer warehouse-returning task scheduling decision comprises the following steps:
s4.1.1: initializing, inputting a station feeding sequenceAnd an idle AGV set for performing an empty trailer return task +.>Will->Each AGV G of (1) m The task start execution time of AGV G is set to 0 m Is left in the execution sequence of the empty trailer warehouse-returning task>Initializing to be empty, and turning S4.1.2 after completion;
s4.1.2: sequentially taking station feeding sequence SQ W The next station number in (1) is assumed to be W i Station W i The current task of returning the empty trailer to the warehouse to be allocated isTurn S4.1.3;
s4.1.3: calculation ofAfter all the AGVs are distributed and the empty trailer is mounted, the AGVs can reach a station W i Task->Assigned to the earliest arriving station W i AGV G of (V) m Namely, the following steps: />With AGV G m Distribution relation of-> Turn S4.1.4;
s4.1.4: predicting AGV G m Arrive at station W i Time of (2)Station W at this moment i The number of unassigned empty trailers in possession +.>
Then determine G m Requiring slave stations W i Number of empty trailers returned
Wherein: c (C) G The maximum number of trailers that can be mounted by the AGV; turn S4.1.5;
s4.1.5: updating SQ W Andinformation about the station W i From SQ W Deleted in (b), if G m Full load then disconnect it from +.>Middle deletion, transfer S4.1.6;
s4.1.6: check SQ W Andif->And->None of them is empty, go to step S4.1.2, otherwise, go to S4.1.7;
s4.1.7: completing the scheduling decision of the empty trailer warehouse-returning task, and outputting eachAnd->
Further, the full trailer delivery task scheduling decision comprises the following steps:
s4.2.1: initializing, inputting a station feeding sequence SQ W And for distributing empty AGV sets for full trailersWill->Each AGV G of (1) m The task start execution time is set to G m Loading and unloading station P running from its parking position to the auxiliary material stock area 0 The execution sequence of the delivery tasks of the full trailer is +.>Initializing to be empty, and turning S4.2.2 after completion;
s4.2.2: sequentially taking station feed supplement sequence SQ W The next station number in (1) is assumed to be W i Station W i The current task of returning the fully loaded trailer to the warehouse to be allocated isTurn S4.2.3;
s4.2.3: calculation ofEach AGV not fully loaded in the system can complete the delivery task of all the fully loaded trailers and then drive the trailer to the station W i At the moment of unloading the trailer, task ∈ ->Assigned to the station W which can be reached earliest i AGV G of (V) m Namely, the following steps: />With AGV G m Distribution relation of-> Turn S4.2.4;
s4.2.4: calculation station W i The remaining space R of the trailer which can be accommodated i G m The number of full trailers that can currently be mounted is also set to ensure executionIs not blocked and meets the AGV mounting capacity constraint, G m For the station W i The number of full trailers distributed +.>Must not exceed W i The remaining space R of the trailer which can be accommodated i And AGV canThe remaining trailer capacity on load, namely:
if it isDescription of the station W i The required full trailer is already distributed, turn S4.2.5, if->Description of the station W i There is also trailer surplus space, turn S4.2.3, station W i And a full trailer delivery task is addedIs a AGV of (2);
s4.2.5: updating SQ W Andinformation about the station W i From SQ W Deleted in (b), if G m Full load then disconnect it from +.>Middle deletion, transfer S4.2.6;
s4.2.6: check SQ W Andif SQ W And->All are not empty, turn S4.2.2, otherwise, turn S4.2.7;
s4.2.7: the dispatching decision of the delivery task of the full-load trailer is completed, and each delivery task is outputAnd->
Further, the objective function is one or more of minimizing a mission delivery path and maximizing a waiting downtime remaining time.
Further, the intelligent optimization algorithm adopts an NSGA-II genetic algorithm, generates new individuals through an intersection algorithm and a mutation algorithm, and generates a new generation station feeding sequence population through evolution, and specifically comprises the following steps:
assuming a population size of N P First, the parent population is combined with N generated by genetic manipulation P Combining the new individuals into candidate population, wherein as all optimization targets are minimized by default when NSGA-II is subjected to non-dominant ranking, the optimization targets are 2N in the candidate population P The individuals are subjected to non-dominant sorting, and the optimized objective function values of the individuals are subjected to dispersion standardization; individual bodyThe objective function values are respectively adjusted as follows:
wherein:respectively obtaining the maximum value and the minimum value of two objective functions of all individuals of the candidate population;
2N in candidate population by using function value after dispersion normalization adjustment P The individuals are subjected to non-dominant ranking, the ranking level and the crowding degree of each individual are determined, all the individuals are ranked according to the ranking level from small to large, and the individuals with the same ranking level are ranked according to the crowding degree from large to small;
ordering and the likePerforming neighborhood search operation on individuals with rank 1, and sorting any one of the individuals with rank 1Station codes of two positions are arbitrarily exchanged to obtain a neighborhood individual of the individual, and the length of the chromosome is N B Is a member of the group consisting of N B ×(N B -1)/2 neighborhood individuals;
randomly generating N P Decoding and evaluating each neighborhood individual, ifIf the non-dominant individual is governed by a certain neighborhood individual, the individual is directly updated to be a better neighborhood individual so as to accelerate the non-dominant individual to converge towards a better direction; non-dominant sorting is carried out on candidate population after neighborhood searching, and the optimal N is selected P Individuals constitute a new generation of populations.
Further, the compromise strategy includes:
the decoding scheduling scheme corresponding to the individual realizes AGVS scheduling: selecting an individual with the longest remaining time from the downtime in the non-dominant front; the individual with the shortest mission path in the non-dominant front is selected.
Further, the multi-load AGVS deadlock prevention task scheduling method has the following constraint conditions:
AGV G m the empty trailer warehouse-returning task and the full trailer delivery task cannot be simultaneously executed:
the number of trailers mounted by the AGV at each time cannot exceed the maximum number of trailers which can be mounted by the AGV:
station trailer buffer capacity limitation:
only when the empty AGV travels from its stop to the parts inventory area, the full trailer starts to load:
AGVs deliver full-load hanging time, deliver in proper order:
the idle AGV travels from its parked position to the corresponding station and the first empty trailer starts to load:
when the AGV loads the empty trailer, the following steps are sequentially carried out:
after the AGV loads all the empty trailers, the AGV runs to a part stock area to unload all the empty trailers:
wherein Ω G ＝{G m M is N and m is more than or equal to 1 and less than or equal to N G A collection of multi-load AGVs; omega shape W ＝{W i I is E N and i is more than or equal to 1 and less than or equal to N w -a set of assembly stations; omega shape P ＝{P j I j is E N and 0.ltoreq.j.ltoreq.N w A collection of loading and unloading work sites for each assembly work station; d (P) i ,P j ) Is station point P i To station point P j Is the most significant of (3)A short distance; v is the running speed of the AGV; s is S m For AGV G m The total path of all tasks to be walked is completed; τ (i, j) is AGV from station point P i Directly run to the working site P j The average time of (a) is d i,j /v；The average number of work stations required to be accessed for single delivery of all AGVs; />Is AGVG m Executing the full trailer delivery mission->A mounting starting time of the device; />For AGV G m Executing the full trailer delivery mission->Is the end of the unloading moment;AGV G m executing the empty trailer return task->A mounting starting time of the device; />For AGV G m Executing the empty trailer return task->Ending the unloading moment of (c).
The invention has the beneficial effects that:
aiming at the problem of scheduling the multi-load AGVS tasks during large-scale auxiliary material distribution in the vehicle assembly application environment, the invention provides a multi-load AGVS deadlock prevention task scheduling method. Firstly, the method designs an anti-deadlock strategy for guaranteeing an AGV without blocking in a system; secondly, in order to improve the quality of the initial population of the intelligent optimization algorithm, a heuristic rule base for generating high-quality individuals of the initial population part is established; in order to effectively reduce the size of the solution space, the near-optimal solution can be conveniently and rapidly found. An empty trailer warehouse-returning task scheduling decision and a full trailer delivery task scheduling decision are designed to decide an anti-deadlock task scheduling scheme meeting all constraint conditions; in order to accelerate the convergence rate of the intelligent optimization algorithm, a population evolution mechanism with elite retention strategy and neighborhood search is designed; the method has the advantages of both an intelligent optimization algorithm and a heuristic scheduling rule, has high response speed, can predict and evaluate the actual regulation and control effect of a regulation and control scheme during scheduling, has stronger global optimizing capability, and can meet the requirement of system multi-objective optimization deadlock prevention coordination control.
Drawings
FIG. 1 is a flowchart of an overall multi-load AGVS deadlock prevention task scheduling method according to an embodiment of the present invention;
FIG. 2 is a flowchart of an empty trailer return task scheduling decision in accordance with an embodiment of the present invention;
FIG. 3 is a flow chart of a full trailer delivery task scheduling decision in accordance with an embodiment of the present invention;
FIG. 4 is a schematic diagram of a crossover operator according to an embodiment of the present invention;
FIG. 5 is a schematic diagram of a mutation operator according to an embodiment of the present invention;
FIG. 6 is a schematic view of a blocked AGV according to an embodiment of the invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Although decision variables of the multi-load AGVS task scheduling problem facing vehicle assembly and manufacture are more, the basic purpose is to ensure sufficient auxiliary materials of all assembly stations, so that the invention combines the advantages of an intelligent optimization algorithm and heuristic scheduling rules, and provides a multi-load AGV deadlock prevention task scheduling method with station feeding sequence optimized as an entry point, and the whole flow is shown in figure 1. Firstly, generating a station feeding sequence scheme set through a constructed heuristic generation rule of the station feeding sequence, and taking the station feeding sequence scheme set as part of individuals in an initial population of an intelligent optimization algorithm; then, an intelligent optimization algorithm is used as a main control optimization flow, and new station material supplementing sequence populations are continuously evolved. When the individual in the station feeding sequence set needs to be evaluated, the distribution relation of each task and the AGV is determined by a heuristic assignment flow, the number of empty or full trailers needed to be loaded/distributed by each station is respectively determined according to an anti-deadlock strategy, and a scheduling scheme corresponding to the station feeding sequence scheme is generated, so that the evaluation of the scheme individual is realized.
A multi-load AGVS deadlock prevention task scheduling method comprises the following steps:
s1: and counting the buffer space information of each station trailer and the idle AGV set in the system.
S2: generating a plurality of station feeding sequence schemes according to the station feeding sequence generation rules, taking the station feeding sequence schemes as part of individuals in the initial population of the intelligent optimization algorithm, and complementing the initial population according to the individual generation rules.
S3: according to an AGV grouping strategy, grouping idle AGVs in the system to form an AGV set for executing an empty trailer warehouse-returning taskAnd AGV set for performing full trailer delivery tasks +.>
S4: each individual in the population was evaluated. Firstly, based on an individual station feeding sequence scheme, according to the scheduling decision of the empty trailer warehouse-returning task, the method comprises the following steps ofAGVs are distributed for the empty trailer warehouse returning task to form an empty trailerA trailer returning task scheduling scheme; then, based on the station feeding sequence scheme, whether the execution AGV becomes a blocking AGV or not is sequentially judged according to the deadlock avoidance strategy, if so, the task is skipped, and if not, the execution AGV is judged according to the dispatching decision of the full trailer delivery task, and the execution AGV is judged according to the deadlock avoidance strategy>AGVs are distributed for the tasks to form a full-load trailer distribution task scheduling scheme; and finally, calculating an objective function of the individual according to the formed empty trailer warehouse-returning task scheduling scheme and the full trailer delivery task scheduling scheme, and taking the objective function value as the individual evaluation.
S5: and (4) taking an intelligent optimization algorithm as a main control optimization flow, taking an initial population as a starting population, continuously evolving to generate a new generation station feeding sequence population based on evaluation according to an evolution strategy, outputting an optimal individual as an optimal scheduling scheme through a compromise strategy when the preset iteration times are reached, and otherwise, entering step S4.
Further, the design of the deadlock prevention strategy comprises the following steps:
establishing a mathematical model with minimized task delivery distance and maximized waiting and stopping production residual time as comprehensive optimization targets, determining constraint conditions, and determining a deadlock prevention strategy of the non-blocking AGV by analyzing deadlock phenomenon triggering conditions, namely ensuring that the residual space of a trailer at any station in the system is not negative at any moment. As shown in FIG. 6, AGV G 1 Bear station W 47 Is full of trailer delivery tasks but due to station W 47 At present, no trailer buffer space exists, and at this time, if no other AGVs execute station W 47 Empty trailer warehouse-returning task, station W 47 Will always not release the buffer space, AGV G 1 Only along the shortest loop path where the unloading point is located, a condition known as the occlusion of the AGV. If all AGVs in the system are blocked, all AGVs will always be unable to unload, a condition known as a system deadlock.
Wherein, the average delivery path of the tasks is minimized:
wherein P is 0 A loading and unloading work site for the auxiliary material stock area; n (N) G The AGV total number in the system; p (G) m ) For AGV G m Is the current dynamic position of (a);for AGV G m The execution sequence of the delivery task of the full-load trailer at this time is +.>For AGV G m An nth full-load trailer delivery task which needs to be executed at the time; />For AGV G m The number of full-load trailer delivery tasks to be executed at this time; /> For AGV G m The execution sequence of the empty trailer warehouse-returning task at this time; />For AGV G m The nth empty trailer to be delivered is returned to the warehouse; />For AGV G m The number of empty trailer warehouse-returning tasks to be executed at this time; />For AGV G m This time needsThe target station of the delivery task of the nth full trailer; />For AGV G m The station where the nth empty trailer warehouse-returning task to be executed is located;
the maximum remaining time for waiting and stopping production is as follows:
wherein: for the station W i The number of auxiliary materials at the time t, wherein t is 0 and represents the current time; and xi is the production beat of the vehicle assembly line.
The constraint conditions are as follows:
AGV G m the empty trailer warehouse-returning task and the full trailer delivery task cannot be simultaneously executed:
the number of trailers mounted by the AGV at each time cannot exceed the maximum number of trailers which can be mounted by the AGV:
station trailer buffer capacity limitation:
only when the empty AGV travels from its stop to the parts inventory area, the full trailer starts to load:
AGVs deliver full-load hanging time, deliver in proper order:
the idle AGV travels from its parked position to the corresponding station and the first empty trailer starts to load:
when the AGV loads the empty trailer, the following steps are sequentially carried out:
after the AGV loads all the empty trailers, the AGV runs to a part stock area to unload all the empty trailers:
wherein Ω G ＝{G m M is N and m is more than or equal to 1 and less than or equal to N G A collection of multi-load AGVs; omega shape W ＝{W i I is E N and i is more than or equal to 1 and less than or equal to N w And is a set of assembly stations, where N W The total number of stations in the system; omega shape P ＝{P j I j is E N and 0.ltoreq.j.ltoreq.N w A collection of loading and unloading work sites for each assembly work station; d (P) i ,P j ) Is station point P i To station point P j Is the shortest distance to the vehicle; v is the running speed of the AGV; s is S m For AGV G m The total path of all tasks to be walked is completed; τ (i, j) is AGV from station point P i Directly run to the working site P j The average time of (a) is d i,j /v；The average number of work stations required to be accessed for single delivery of all AGVs; c (C) G The maximum number of trailers that can be mounted by the AGV; c (C) i For the station W i The number of trailers that can be accommodated; k (K) i For the station W i The maximum auxiliary material sleeve number which can be loaded by each full trailer; />For the station W i The number of non-empty trailers at time t; />For the station W i The number of empty trailers at time t;for the station W i Is a full trailer delivery task; />For the station W i Is a blank trailer warehouse-returning task; />For AGV G m Executing the full trailer delivery mission->A mounting starting time of the device; />For AGV G m Performing full trailer delivery tasksIs the end of the unloading moment; />AGV G m Executing the empty trailer return task->A mounting starting time of the device;for AGV G m Executing the empty trailer return task->Is the end of the unloading moment; SQ (SQ) W ＝{W i …W j …W k The sequence of station feeding sequences; w (W) i A sequence numbering each station; />For all blocking AGVs, wherein +.>The number of AGVs blocked in the system;a set of all active AGVs, wherein +.>The number of the AGVs in the system is calculated; /> For a set of all free AGVs, wherein,the number of idle AGVs in the system; />For the set of free AGVs to be used for the delivery of full trailers, wherein the +.>To be used to deliver an empty AGV for a full trailer, it is common to: for the set of free AGVs to be used for the delivery of empty trailers, wherein +.>To be used to deliver an empty trailer, the number of empty AGVs is typically:
furthermore, in view of the great influence of the advantages and disadvantages of the initial population on the convergence speed of the intelligent optimization algorithm, in order to improve the quality of the initial population, the invention selects some state attribute indexes of the system to calculate the utility value of the storage state of each station auxiliary material, and designs a heuristic rule for generating the station material supplementing sequence. Heuristic rules can be divided into single attribute rules and multi-attribute rules according to different numbers of selected state attribute indexes. In the multi-load AGVS manufactured for vehicle assembly, the optional state attribute indicators mainly include: station W i Number of non-empty trailersStation W i Number of empty trailers->Station W i Parts sleeve number->Station W i Distance d from parts inventory area i,j Etc. By selecting the state attribute indexes at different time points or combining a plurality of state attribute indexes, the method can constructA plurality of heuristic rules for generating station feeding sequences, partial rules and corresponding stations W i Utility value->The calculation method of (2) is as follows:
the number of currently empty trailers at the station is more than one:
wherein: ∈ indicates that the rule will be for each station W i According to the utility value thereofDetermining the station feeding sequence from high to low;indicating station W i The number of empty trailers allocated to carry AGVs but not yet mounted;
the station is preferential to the ones with less non-empty trailers:
wherein: ∈ indicates that the rule will be for each station W i According to the utility value thereofDetermining the station feeding sequence from low to high;indicating that the AGV has been transported without being delivered to station W i A number of full trailers;
the priority of the station with less current part sets is as follows:
multi-attribute rules based on the current state of the system:
the number of trailers is higher when the next feeding is performed at the station:
/>
the number of non-empty trailers is low when the station is fed next time:
wherein: ceil () represents a rounding up operation;
the priority of fewer parts in the next feeding process of the station is as follows:
multi-attribute rule based on system state when station next feed supplement:
each rule generates a station feeding sequence according to the corresponding utility value, the station feeding sequence is used as the individual of the initial population, and other individuals of the initial population are generated completely randomly.
Empty trailer warehouse-returning task scheduling decision
Determined from idle AGV grouping decisionsIn a given station feed sequence SQ W As each empty spaceThe priority order of the trailer returning tasks is that the empty trailer returning tasks are allocated to +.>Idle AGVs in (a) to decide +.>And->In the decision process, an empty trailer warehouse-returning task is distributed to an AGV which reaches the station first; predicting the real-time number of empty trailers of a workstation when the AGV reaches a working site, and determining the warehouse returning task of each empty trailer>The number of empty trailers to be mounted is +.>The method comprises the following specific steps:
s4.1.1: initializing, inputting a station feeding sequence SQ W (i.e., station feed order scheme) and free AGV set for performing empty trailer return tasksWill->Each AGV G of (1) m The task start execution time of AGV G is set to 0 m Is left in the execution sequence of the empty trailer warehouse-returning task>Initialized to null and completed back to S4.1.2.
S4.1.2: sequentially taking station feeding sequence SQ W The next station number in (1) is assumed to be W i Station W i The current task of returning the empty trailer to the warehouse to be allocated isAnd S4.1.3.
S4.1.3: calculation ofAfter all the AGVs are distributed and the empty trailer is mounted, the AGVs can reach a station W i Task->Assigned to the earliest arriving station W i AGV G of (V) m Namely, the following steps: />With AGV G m Distribution relation of-> And S4.1.4.
S4.1.4: predicting AGV G m Arrive at station W i Time of (2)Station W at this moment i The number of unassigned empty trailers in possession +.>/>
Then determine G m Requiring slave stations W i Number of empty trailers returned
Wherein: c (C) G Maximum number of trailers mountable for AGV
And S4.1.5.
S4.1.5: updating SQ W Andinformation about the station W i From SQ W Deleted in (b), if G m Full load then disconnect it from +.>Middle deletion, transfer S4.1.6.
S4.1.6: check SQ W Andif SQ W And->None are empty, go to step S4.1.2, otherwise go to S4.1.7.
S4.1.7: completing the scheduling decision of the empty trailer warehouse-returning task, and outputting eachAnd->
Full trailer delivery task scheduling decision
Determined from idle AGV grouping decisionsIn a given station feed sequence SQ W As the priority order of the delivery tasks of the full trailer, the delivery tasks of the full trailer are sequentially allocated to +.>Idle AGVs in (a) to decide +.>And->In the decision process, in order to reduce the running distance of the AGVs, the distance between stations of the trailer is used as the +.>Distribution order->The basis of (2); determining the number of full-load trailers required to be distributed at each station by adopting anti-deadlock strategy>The method comprises the following specific steps:
s4.2.1: initializing, inputting a station feeding sequence SQ W (i.e., station feed sequence scheme) and for dispensing empty AGVs sets of full trailersWill->Each AGV G of (1) m The task start execution time is set to G m Loading and unloading station P running from its parking position to the auxiliary material stock area 0 The execution sequence of the delivery tasks of the full trailer is +.>Initialized to null and completed back to S4.2.2.
S4.2.2: sequentially taking station feed supplement sequence SQ W The next station number in (1) is assumed to be W i Station W i Full-load trailer to be distributed currentlyThe job of returning to the warehouse isAnd S4.2.3.
S4.2.3: calculation ofEach AGV not fully loaded in the system can complete the delivery task of all the fully loaded trailers and then drive the trailer to the station W i At the moment of unloading the trailer, task ∈ ->Assigned to the station W which can be reached earliest i AGV G of (V) m Namely, the following steps: />With AGV G m Distribution relation of-> And S4.2.4.
S4.2.4: calculation station W i The remaining space R of the trailer which can be accommodated i G m The number of full trailers that can currently be mounted is also set to ensure executionIs not blocked and meets the AGV mounting capacity constraint, G m For the station W i The number of full trailers distributed +.>Must not exceed W i The remaining space R of the trailer which can be accommodated i And the remaining trailer capacity that the AGV can mount, namely:
if it isDescription of the station W i The required full trailer is already distributed, turn S4.2.5, if->Description of the station W i There is also trailer surplus space, turn S4.2.3, station W i And a full trailer delivery task is addedIs provided.
S4.2.5: updating SQ W Andinformation about the station W i From SQ W Deleted in (b), if G m Full load then disconnect it from +.>Middle deletion, transfer S4.2.6.
S4.2.6: check SQ W Andif SQ W And->All are not empty then go to S4.2.2, otherwise go to S4.2.7.
S4.2.7: the dispatching decision of the delivery task of the full-load trailer is completed, and each delivery task is outputAnd->
Evolutionarily generating new generation station feed supplement sequence population
The intelligent optimization algorithm in the invention can adopt ant colony algorithm, wolf colony algorithm, black hole algorithm, genetic algorithm and other algorithms, and the implementation case uses non-dominant ordering genetic algorithm as an example to generate new individuals through crossover algorithm and mutation algorithm, and the method is as follows:
(1) Crossover operator
Randomly selecting two different individuals from the father population after non-dominant ranking to enter a crossing operation, wherein the selection steps of any one of the individuals are as follows:
firstly, determining a ranking level of an individual, wherein the probability delta (r) of the ranking level r of the individual being selected is as follows:
δ(r)＝δ(1)ρ (r-1)
wherein: ρ is a probability factor, and the value range is: 0< ρ <1; delta (1) is the probability that an individual of rank 1 is selected, which needs to be calculated from the highest rank max (r) of the non-dominated ordering of the population and the probability factor ρ, as follows:
δ(1)＝(1-ρ)/[(1-ρ max(r) )]
the method can ensure that the lower the individual sorting level is, the higher the sorting probability is, and the sum of the sorting probabilities of all the sorting levels is 1.
And secondly, sorting all individuals of the sorting level r according to the crowdedness of the individuals, and determining the selected individuals by adopting a roulette method.
The schematic diagram of the crossover operator is shown in fig. 4, assuming that two Parent individuals are respectively Parent1 and Parent2, two crossover positions Pos1 and Pos2 are randomly selected, the station sequences between the crossing points of Parent1 are rearranged according to the sequence of the station sequences in Parent2, and the rearranged Parent1 is used as Child individual Child.
(2) Mutation operator
An example of mutation operation is shown in fig. 5, in which two crossing positions are randomly selected and two position codes are exchanged, the mutated individual needs to be verified whether it appears in the parent population, and if the individual appears in the parent population, the crossing and mutation operation needs to be performed again.
Race evolution mechanism
Assuming a population size of N P First, the parent population is combined with N generated by genetic manipulation P Combining the new individuals into candidate population, wherein all optimization targets are minimized by default when NSGA-II is subjected to non-dominant ranking, so as to achieve 2N in the candidate population P The individuals are subjected to non-dominant ranking, and the optimized objective function values of the individuals are subjected to dispersion normalization to obtain the individualsFor example, the objective function values are respectively adjusted as follows:
wherein:respectively obtaining the maximum value and the minimum value of two objective functions of all individuals of the candidate population;
2N in candidate population by using function value after dispersion normalization adjustment P The individuals are subjected to non-dominant ranking, the ranking level and the crowding degree of each individual are determined, all the individuals are ranked according to the ranking level from small to large, and the individuals with the same ranking level are ranked according to the crowding degree from large to small;
performing neighborhood search operation on individuals with ranking level 1, and performing neighborhood search operation on any one of individuals with ranking level 1Station codes of two positions are arbitrarily exchanged to obtain a neighborhood individual of the individual, and the length of the chromosome is N B Is a member of the group consisting of N B ×(N B -1)/2 neighborhood individuals;
randomly generating N P Decoding and evaluating each neighborhood individual, ifIf the non-dominant individual is governed by a certain neighborhood individual, the individual is directly updated to be a better neighborhood individual so as to accelerate the non-dominant individual to converge towards a better direction; non-dominant sorting is carried out on candidate population after neighborhood searching, and the optimal N is selected P Individuals constitute a new generation of populations.
Compromise strategy
And the decoding scheduling scheme corresponding to the individual realizes AGVS scheduling. Such as: in order to avoid the assembly line from entering a production stopping state due to material waiting, the productivity of the assembly line is improved, and an individual with the longest residual time from the production stopping in the non-dominant front (Pareto front) can be selected. In order to reduce the path and time of executing tasks by the AGVS and improve the operation efficiency of the AGVS, an individual with the shortest task path in the non-dominant front (Pareto front) can be selected.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (10)
1. The multi-load AGVS deadlock prevention task scheduling method is characterized by comprising the following steps of:
s1: counting buffer space information of the trailers at each station and an idle AGV set;
s2: generating a plurality of station feeding sequence schemes according to the station feeding sequence generation rules, taking the station feeding sequence schemes as part of individuals in the initial population of the intelligent optimization algorithm, and completing the initial population according to the individual generation rules;
s3: according to an AGV grouping strategy, grouping idle AGVs in the system to form an AGV set for executing an empty trailer warehouse-returning taskAnd AGV set for performing full trailer delivery tasks +.>
S4: each individual in the population was evaluated: firstly, based on an individual station feeding sequence scheme, according to the scheduling decision of the empty trailer warehouse-returning task, the method comprises the following steps ofAGVs are distributed to the empty trailer warehouse-returning tasks to form an empty trailer warehouse-returning task scheduling scheme; then, based on the station feeding sequence scheme, whether the execution AGV becomes a blocking AGV is sequentially judged according to the deadlock avoidance strategy, if so, the task is skipped, and if not, the execution AGV is judged according to the dispatching decision of the full trailer delivery taskAGVs are distributed for the tasks to form a full-load trailer distribution task scheduling scheme; finally, calculating an objective function of the individual according to the formed empty trailer warehouse-returning task scheduling scheme and the full trailer delivery task scheduling scheme, and taking the objective function value as the individual evaluation;
s5: and (4) taking an intelligent optimization algorithm as an optimization flow, taking an initial population as a starting population, continuously evolving to generate a new generation station feeding sequence population according to an evolution strategy based on the evaluation of individuals, outputting an optimal individual as an optimal scheduling scheme through a compromise strategy when the preset iteration times are reached, and otherwise, entering step S4.
2. The multi-load AGVS anti-deadlock task scheduling method according to claim 1, wherein the station feed sequence generating rule is one or more of a multi-attribute rule based on a current state of the system, a multi-trailer number priority when feeding next time of the station, a part sleeve number priority when feeding next time of the station, and a multi-attribute rule based on a system state when feeding next time of the station.
3. The method for scheduling the deadlock prevention tasks of the multi-load AGVS according to claim 2, wherein the stations currently have a plurality of empty trailers with priority, specifically:
wherein: ∈ indicates that the rule will be for each station W i According to the utility value thereofDetermining the station feeding sequence from high to low; />Indicating station W i The number of empty trailers at the current time; />Indicating AGV G m From station W i The number of empty trailers loaded; n (N) G The AGV total number in the system; />Indicating station W i The number of empty trailers allocated to carry AGVs but not yet mounted;
the station is preferential to the ones with less number of currently non-empty trailers, specifically:
wherein: ∈ indicates that the rule will be for each station W i According to the utility value thereofDetermining the station feeding sequence from low to high; station W i Number of non-empty trailers at the current time；/>Indicating AGV G m The current delivery to the station W i A number of full trailers; />Indicating that the AGV has been transported without being delivered to station W i A number of full trailers;
the station is preferential to the ones with less current parts, specifically:
wherein:indicating station W i The number of auxiliary materials at the current moment; k (K) i Indicating station W i The maximum auxiliary material sleeve number which can be loaded by each full trailer;
the multi-attribute rule based on the current state of the system specifically comprises the following steps:
the number of trailers is more than one when the next feeding is performed at the station, specifically:
wherein: tau (0, i) represents the AGV traveling directly from the auxiliary material inventory area to the job site P i Average time of (2); ζ is the production beat of the vehicle assembly line;
the fewer priority of the non-empty trailers is given when the station is fed next time, specifically:
wherein: ceil () represents a rounding up operation;
the station is preferential to the few part sets when the next material is fed, and specifically comprises the following steps:
the multi-attribute rule based on the system state when the station is fed next time is specifically as follows:
4. the method for scheduling multi-load AGVS anti-deadlock tasks according to claim 1, wherein the deadlock avoidance strategy is to ensure that the remaining space of the trailer at any station in the system is not negative at any time, thereby avoiding blocking the AGV in the system; when the station W i The remaining space R of the trailer which can be accommodated i When it is 0, it continues to be station W i The AGV is allocated to the full trailer to be allocated, the execution AGV becomes a blocking AGV, and the station W is stopped i And (3) distributing AGVs for the to-be-distributed full trailer distribution tasks, wherein the blocking AGVs are AGVs for bearing the full trailer distribution tasks of stations with the residual space of the trailer buffer zone being 0.
5. The multi-load AGVS anti-deadlock task scheduling method according to claim 1, wherein the empty trailer return task scheduling decision comprises the steps of:
s4.1.1: initializing, inputting a station feeding sequence SQ W And an idle AGV set for performing an empty trailer return taskWill->Each AGV G of (1) m Setting the task start execution time of (1) to 0, AGVG m Is left in the execution sequence of the empty trailer warehouse-returning task>Initializing to be empty, and turning S4.1.2 after completion;
s4.1.2: sequentially taking station feeding sequence SQ W The next station number in (1) is assumed to be W i Station W i The current task of returning the empty trailer to the warehouse to be allocated isTurn S4.1.3;
s4.1.3: calculation ofAfter all the AGVs are distributed and the empty trailer is mounted, the AGVs can reach a station W i Task->Assigned to the earliest arriving station W i AGV G of (V) m Namely, the following steps: />With AGV G m Distribution relation of (3)Turn S4.1.4;
s4.1.4: predicting AGV G m Arrive at station W i Time of (2)Station W at this moment i Owned noDistributing empty trailer number->
Then determine G m Requiring slave stations W i Number of empty trailers returned
Wherein: c (C) G The maximum number of trailers that can be mounted by the AGV; turn S4.1.5;
s4.1.5: updating SQ W Andinformation about the station W i From SQ W Deleted in (b), if G m Full load then disconnect it from +.>Middle deletion, transfer S4.1.6;
s4.1.6: check SQ W Andif SQ W And->None of them is empty, go to step S4.1.2, otherwise, go to S4.1.7;
s4.1.7: completing the scheduling decision of the empty trailer warehouse-returning task, and outputting eachAnd->
6. The multi-load AGVS deadlock prevention task scheduling method according to claim 1, wherein the full trailer delivery task scheduling decision comprises the steps of:
s4.2.1: initializing, inputting a station feeding sequence SQ W And for distributing empty AGV sets for full trailersWill beEach AGV G of (1) m The task start execution time is set to G m Loading and unloading station P running from its parking position to the auxiliary material stock area 0 The execution sequence of the delivery tasks of the full trailer is +.>Initializing to be empty, and turning S4.2.2 after completion;
s4.2.2: sequentially taking station feed supplement sequence SQ W The next station number in (1) is assumed to be W i Station W i The current task of returning the fully loaded trailer to the warehouse to be allocated isTurn S4.2.3;
s4.2.3: calculation ofEach AGV not fully loaded in the system can complete the delivery task of all the fully loaded trailers and then drive the trailer to the station W i At the moment of unloading the trailer, task ∈ ->Assigned to the station W which can be reached earliest i AGV G of (V) m Namely, the following steps: />With AGV G m Distribution relation of->Turn S4.2.4;
s4.2.4: calculation station W i The remaining space R of the trailer which can be accommodated i G m The number of full trailers that can currently be mounted is also set to ensure executionIs not blocked and meets the AGV mounting capacity constraint, G m For the station W i The number of full trailers distributed +.>Must not exceed W i The remaining space R of the trailer which can be accommodated i And the remaining trailer capacity that the AGV can mount, namely:
if it isDescription of the station W i The required full trailer is already distributed, turn S4.2.5, if->Description of the station W i There is also trailer surplus space, turn S4.2.3, station W i And a step of executing the full trailer delivery task is added>Is a AGV of (2);
s4.2.5: updating SQ W Andinformation about the station W i From SQ W Deleted in (b), if G m Full load then disconnect it from +.>Middle deletion, transfer S4.2.6;
s4.2.6: check SQ W Andif SQ W And->All are not empty, turn S4.2.2, otherwise, turn S4.2.7;
s4.2.7: the dispatching decision of the delivery task of the full-load trailer is completed, and each delivery task is outputAnd->
7. The multi-load AGVS anti-deadlock task scheduling method according to claim 1, wherein the objective function is one or more of minimizing task delivery distance and maximizing standby downtime.
8. The multi-load AGVS deadlock prevention task scheduling method according to claim 1, wherein the intelligent optimization algorithm adopts NSGA-ii genetic algorithm, generates new individuals through crossover algorithm and mutation algorithm, and the evolution generates new generation station feed order population, specifically comprising the following steps:
assuming a population size of N P First, the parent population is combined with N generated by genetic manipulation P Combining the new individuals into candidate population, wherein as all optimization targets are minimized by default when NSGA-II is subjected to non-dominant ranking, the optimization targets are 2N in the candidate population P The individuals are subjected to non-dominant sorting, and the optimized objective function values of the individuals are subjected to dispersion standardization; individual bodyThe objective function values are respectively adjusted as follows:
wherein:respectively obtaining the maximum value and the minimum value of two objective functions of all individuals of the candidate population;
2N in candidate population by using function value after dispersion normalization adjustment P The individuals are subjected to non-dominant ranking, the ranking level and the crowding degree of each individual are determined, all the individuals are ranked according to the ranking level from small to large, and the individuals with the same ranking level are ranked according to the crowding degree from large to small;
performing neighborhood search operation on individuals with ranking level 1, and performing neighborhood search operation on any one of individuals with ranking level 1Station codes of two positions are arbitrarily exchanged to obtain a neighborhood individual of the individual, and the length of the chromosome is N B Is a member of the group consisting of N B ×(N B -1)/2 neighborhood individuals;
randomly generating N P Decoding and evaluating each neighborhood individual, ifIf the non-dominant individual is governed by a certain neighborhood individual, the individual is directly updated to be a better neighborhood individual so as to accelerate the non-dominant individual to converge towards a better direction; non-dominant sorting is carried out on candidate population after neighborhood searching, and the optimal N is selected P Individuals constitute a new generation of populations.
9. The multi-load AGVS anti-deadlock task scheduling method according to claim 1, wherein the compromise policy comprises:
the decoding scheduling scheme corresponding to the individual realizes AGVS scheduling: selecting an individual with the longest remaining time from the downtime in the non-dominant front; the individual with the shortest mission path in the non-dominant front is selected.
10. The multi-load AGVS deadlock prevention task scheduling method according to any one of claims 1-9, wherein the multi-load AGVS deadlock prevention task scheduling method has the following constraint conditions:
AGV G m the empty trailer warehouse-returning task and the full trailer delivery task cannot be simultaneously executed:
the number of trailers mounted by the AGV at each time cannot exceed the maximum number of trailers which can be mounted by the AGV:
station trailer buffer capacity limitation:
only when the empty AGV travels from its stop to the parts inventory area, the full trailer starts to load:
AGVs deliver full-load hanging time, deliver in proper order:
the idle AGV travels from its parked position to the corresponding station and the first empty trailer starts to load:
when the AGV loads the empty trailer, the following steps are sequentially carried out:
after the AGV loads all the empty trailers, the AGV runs to a part stock area to unload all the empty trailers:
wherein Ω G ＝{G m M is N and m is more than or equal to 1 and less than or equal to N G A collection of multi-load AGVs; omega shape W ＝{W i I is E N and i is more than or equal to 1 and less than or equal to N w -a set of assembly stations; omega shape P ＝{P j I j is E N and 0.ltoreq.j.ltoreq.N w A collection of loading and unloading work sites for each assembly work station; d (P) i ，P j ) Is station point P i To station point P j Is the shortest distance to the vehicle; v is the running speed of the AGV; s is S m For AGV G m The total path of all tasks to be walked is completed; τ (i, j) is AGV from station point P i Directly run to the working site P j The average time of (a) is d i，j /v；The average number of work stations required to be accessed for single delivery of all AGVs; />For AGV G m Performing full trailer delivery tasksA mounting starting time of the device; />For AGV G m Executing the full trailer delivery mission->Is the end of the unloading moment;AGV G m executing the empty trailer return task->A mounting starting time of the device; />For AGV G m Executing the empty trailer return task->Ending the unloading moment of (c).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111293432.2A CN114003011B (en) | 2021-11-03 | 2021-11-03 | Multi-load AGVS deadlock prevention task scheduling method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111293432.2A CN114003011B (en) | 2021-11-03 | 2021-11-03 | Multi-load AGVS deadlock prevention task scheduling method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114003011A CN114003011A (en) | 2022-02-01 |
| CN114003011B true CN114003011B (en) | 2023-08-15 |
Family
ID=79927192
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111293432.2A Active CN114003011B (en) | 2021-11-03 | 2021-11-03 | Multi-load AGVS deadlock prevention task scheduling method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114003011B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004280213A (en) * | 2003-03-13 | 2004-10-07 | Japan Science & Technology Agency | Distributed path planning device and method, and distributed path planning program |
| CN107544513A (en) * | 2017-09-29 | 2018-01-05 | 盐城工学院 | The anti-deadlock dispatching method of more carrying capacity automatic guide vehicles and device |
| CN109508010A (en) * | 2018-12-17 | 2019-03-22 | 盐城工学院 | A kind of perspective anti-deadlock dynamic allocation method of multiple-mobile-robot system grid point based on grating map |
| CN110084512A (en) * | 2019-04-26 | 2019-08-02 | 河海大学常州校区 | A kind of multi-robot Task Allocation towards intelligent warehousing system |
| CN110264120A (en) * | 2019-05-06 | 2019-09-20 | 盐城品迅智能科技服务有限公司 | A kind of intelligent storage route planning system and method based on more AGV |
| CN111474926A (en) * | 2020-03-24 | 2020-07-31 | 浙江中烟工业有限责任公司 | Waste smoke recovery method based on multiple AGV time window path optimization algorithm |
| AU2020101761A4 (en) * | 2020-08-11 | 2020-09-17 | Nanjing University Of Science & Technology | Method for planning path of parking agv based on improved dijkstra algorithm |
| CN112396328A (en) * | 2020-11-20 | 2021-02-23 | 河海大学常州校区 | Multi-objective task allocation scheduling optimization method for multi-load carrying robot group |
| CN112783172A (en) * | 2020-12-31 | 2021-05-11 | 重庆大学 | AGV and machine integrated scheduling method based on discrete whale optimization algorithm |
-
2021
- 2021-11-03 CN CN202111293432.2A patent/CN114003011B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004280213A (en) * | 2003-03-13 | 2004-10-07 | Japan Science & Technology Agency | Distributed path planning device and method, and distributed path planning program |
| CN107544513A (en) * | 2017-09-29 | 2018-01-05 | 盐城工学院 | The anti-deadlock dispatching method of more carrying capacity automatic guide vehicles and device |
| CN109508010A (en) * | 2018-12-17 | 2019-03-22 | 盐城工学院 | A kind of perspective anti-deadlock dynamic allocation method of multiple-mobile-robot system grid point based on grating map |
| CN110084512A (en) * | 2019-04-26 | 2019-08-02 | 河海大学常州校区 | A kind of multi-robot Task Allocation towards intelligent warehousing system |
| CN110264120A (en) * | 2019-05-06 | 2019-09-20 | 盐城品迅智能科技服务有限公司 | A kind of intelligent storage route planning system and method based on more AGV |
| CN111474926A (en) * | 2020-03-24 | 2020-07-31 | 浙江中烟工业有限责任公司 | Waste smoke recovery method based on multiple AGV time window path optimization algorithm |
| AU2020101761A4 (en) * | 2020-08-11 | 2020-09-17 | Nanjing University Of Science & Technology | Method for planning path of parking agv based on improved dijkstra algorithm |
| CN112396328A (en) * | 2020-11-20 | 2021-02-23 | 河海大学常州校区 | Multi-objective task allocation scheduling optimization method for multi-load carrying robot group |
| CN112783172A (en) * | 2020-12-31 | 2021-05-11 | 重庆大学 | AGV and machine integrated scheduling method based on discrete whale optimization algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 自动导引车系统避免死锁的多属性任务调度方法;管贤平;戴先中;;中国机械工程(第19期);73-78 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114003011A (en) | 2022-02-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107506956B (en) | Based on improvement particle cluster algorithm supply chain production and transport coordinated dispatching method and system | |
| CN110084545B (en) | Integrated scheduling method of multi-lane automatic stereoscopic warehouse based on mixed integer programming model | |
| CN109214608B (en) | Vehicle scheduling optimization method | |
| CN108985597A (en) | A kind of dynamic logistics dispatching method | |
| CN113359702B (en) | Intelligent warehouse AGV operation optimization scheduling method based on water wave optimization-tabu search | |
| CN107578197A (en) | The uncertain mix flow vehicles dispatching system optimization of region method of demand | |
| CN112978165A (en) | Intelligent stereoscopic warehouse and warehouse-out control method, device and system thereof | |
| CN111860957B (en) | Multi-vehicle-type vehicle path planning method considering secondary distribution and balancing | |
| CN115481897A (en) | AGV unmanned warehouse equipment optimization configuration method | |
| Jiang | Solving multi-robot picking problem in warehouses: a simulation approach | |
| CN110263970B (en) | Hybrid discrete artificial bee colony algorithm for solving automatic rail trolley feeding scheduling problem | |
| CN114003011B (en) | Multi-load AGVS deadlock prevention task scheduling method | |
| CN112989696A (en) | Automatic picking system goods location optimization method and system based on mobile robot | |
| CN111160654B (en) | Transportation path optimization method for reducing total cost based on fuzzy C-means-simulated annealing algorithm | |
| Jin et al. | Dynamic scheduling of mobile-robotic warehouse logistics system | |
| CN114707707A (en) | Method and system for scheduling AGV task based on improved genetic algorithm | |
| CN115358455A (en) | Electric vehicle path optimization method considering battery replacement and hybrid time window constraints | |
| CN116523221A (en) | Optimal scheduling method and system for intelligent warehouse picking task | |
| CN112508478B (en) | Flexible logistics distribution task allocation method based on self-organizing automated guided vehicle | |
| CN115907066A (en) | Cement enterprise vehicle scheduling method based on hybrid sparrow intelligent optimization algorithm | |
| Li et al. | Two-stage multi-AGV path planning based on speed pre-allocation | |
| Fang et al. | Heterogeneous-vehicle distribution logistics planning for assembly line station materials with multiple time windows and multiple visits | |
| CN114462764A (en) | Dispatching method of multilayer multi-port hoister | |
| Zhang et al. | Design and Optimization of AGV Scheduling Algorithm in Logistics System Based on Convolutional Neural Network | |
| Li et al. | Coordinated control of multi-zone AVS/RS, conveyors and pick-up operations in warehouse system |
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 |