CN110362378A - A kind of method for scheduling task and equipment - Google Patents
A kind of method for scheduling task and equipment Download PDFInfo
- Publication number
- CN110362378A CN110362378A CN201810315075.7A CN201810315075A CN110362378A CN 110362378 A CN110362378 A CN 110362378A CN 201810315075 A CN201810315075 A CN 201810315075A CN 110362378 A CN110362378 A CN 110362378A
- Authority
- CN
- China
- Prior art keywords
- task
- factor
- population
- generation
- evolutionary generation
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
- G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
- G06F9/5061—Partitioning or combining of resources
- G06F9/5072—Grid computing
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/44—Arrangements for executing specific programs
- G06F9/455—Emulation; Interpretation; Software simulation, e.g. virtualisation or emulation of application or operating system execution engines
- G06F9/45533—Hypervisors; Virtual machine monitors
- G06F9/45558—Hypervisor-specific management and integration aspects
- G06F2009/4557—Distribution of virtual machine instances; Migration and load balancing
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Mathematical Physics (AREA)
- Physiology (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Genetics & Genomics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of method for scheduling task and equipment, the described method includes: generating the initial population of the task and processing combination of nodes that handle task schedule to processing node according to multiple tasks to be processed and multiple processing nodes for handling task；According to current evolutionary generation and maximum evolutionary generation, construct the input factor for controlling the initial population cross and variation, according to required time and expense that task is handled in processing node, fitness function of the building for calculating task and the fitness for handling combination of nodes；New differential evolution algorithm is generated according to the input factor and the fitness function；The initial population is optimized by the new differential evolution algorithm, acquisition task handles the task schedule in combining the time required to node is handled into combination with the set of the optimum combination of the processing node optimum combination characterization and overhead-optimized.
Description
Technical field
The present invention relates to field of computer technology more particularly to a kind of method for scheduling task and equipment.
Background technique
In recent years, with the high speed development of cloud computing technology, the task schedule in cloud computing platform becomes cloud computing application
One of research hotspot, task schedule refers to the various demands according to task, by using appropriate strategy different task point
It is fitted in cloud computing platform on suitable resource node and executes.
The use cost of different computing resources is different in cloud computing platform, and the strong computing resource of usual processing capacity makes
It is higher with cost, however, the task scheduling approach of major part cloud computing platform will only shorten the deadline of task at present
As research emphasis, it is unfavorable for the performance boost of cloud computing platform, so, how to select between optimal task and computing resource
Scheduling scheme be current urgent problem.
Summary of the invention
The embodiment of the present invention provides a kind of method for scheduling task and equipment, can be realized and selects optimal task and calculate to provide
The purpose of scheduling scheme between source.
In a first aspect, a kind of method for scheduling task provided in an embodiment of the present invention, comprising:
According to multiple tasks to be processed and multiple processing nodes for handling task, generate task schedule everywhere
The initial population of task and processing combination of nodes that reason node is handled；
According to current evolutionary generation and maximum evolutionary generation, construct for controlling the defeated of the initial population cross and variation
Enter the factor, according to required time and expense that task is handled in processing node, building is saved for calculating task and processing
The fitness function of the combined fitness of point；
New differential evolution algorithm is generated according to the input factor and the fitness function；
The initial population is optimized by the new differential evolution algorithm, obtains task and processing node most
Excellent combined set, the task schedule in combination is handled node into combination and handled by optimum combination characterization to be taken
Between and overhead-optimized.
It, can be first according to current evolutionary generation and maximum using above-mentioned method for scheduling task provided in an embodiment of the present invention
Evolutionary generation, constructs the input factor for controlling the initial population cross and variation, and according to task processing node into
The required time and expense of row processing, fitness function of the building for calculating task and the fitness for handling combination of nodes,
Then new differential evolution algorithm is generated using the input factor and fitness function of building, and then is calculated using new differential evolution
Method optimizes the initial population, obtain by task schedule to handle node handled the time required to and expense most
Excellent task and processing combination of nodes realize that selection is optimal so time and the expense of processing node processing task can be taken into account
Task and computing resource between scheduling scheme purpose, and promoted task processing platform performance.
Optionally, the input factor includes intersecting the factor, the current evolutionary generation of basis and maximum evolutionary generation,
The input factor for controlling the initial population cross and variation is constructed, specifically:
Intersect factor CR according to current evolutionary generation t and maximum evolutionary generation T, building:
CR=CRmin+(CRmax-CRmin)(t/T)2；
Wherein, CRminFor preset intersection factor minimum value, CRmaxFor preset intersection factor maximum value.
Above-mentioned optional mode describes the intersection factor CR of building, and at the initial stage of evolution, the value of CR is smaller, is more conducive to protect
Hold population diversity and ability of searching optimum；As the increase of evolutionary generation namely the number of iterations increase, it is continuous to intersect factor CR
Increase, convergence rate can be increased, so, there is the probability for guaranteeing to search globally optimal solution while accelerating convergence rate
Beneficial effect, and then can further promote the probability for choosing the scheduling scheme between optimal task and computing resource,
And promote the performance of task processing platform.
Optionally, the input factor includes zoom factor, the current evolutionary generation of basis and maximum evolutionary generation,
The input factor for controlling the initial population cross and variation is constructed, specifically:
According to the current evolutionary generation t and maximum evolutionary generation T, zoom factor F is constructed:
λ=e1-T/(T+1-t)；
F=F0*2λ；
Wherein, F0For preset zoom factor initial value, λ is the evolutionary generation factor.
Above-mentioned optional mode describes the intersection factor zoom factor F of building, and at the initial stage of evolution, F has the larger value, larger
Zoom factor F can be scanned in the overall situation so that population can keep diversity of individuals in the early stage, avoid precocity；
With the increase of the number of iterations namely in later stage of evolution, the value of F constantly reduces, and is conducive to the excellent information for retaining solution, so, tool
There is the further probability for increasing and searching globally optimal solution, and convergence speed of the algorithm can also be accelerated, improves algorithm
The effect of efficiency.
Optionally, described that the initial population is optimized by the new differential evolution algorithm, obtain task with
Handle the set of the optimum combination of node, comprising:
The initial population is optimized by the new differential evolution algorithm, and is held in each optimization process
The following operation of row, obtains optimal target population until reaching maximum evolutionary generation T, wherein the optimal target population is
For task and processing node optimum combination set:
All offspring individuals generated during suboptimization according to the intersection factor CR are obtained, by all filial generations
Individual is added to the newborn individual set during suboptimization；
Calculate the adaptation of each individual in the fitness of each individual in the newborn individual set and previous generation population
Degree, and the newborn individual set and previous generation population are shunk according to the zoom factor F, acquisition is leaned on by fitness
The next-generation population that the individual of preceding predetermined number is constituted.
Above-mentioned optional mode, which describes, is added to all offspring individuals that each evolution namely iterative process generate often
It participates in the newborn individual set of secondary evolution in the competition of previous generation population, and then selects the next-generation population evolved every time, institute
To have the advantages that be able to maintain population diversity, and then also further improve the probability for searching globally optimal solution.
Second aspect provides a kind of task scheduling equipment, comprising:
Population generation unit, for according to multiple tasks to be processed and multiple processing nodes for handling task,
Generate the initial population of the task and processing combination of nodes that handle task schedule to processing node；
Construction unit, for constructing for controlling described initial kind according to current evolutionary generation and maximum evolutionary generation
The input factor of group's cross and variation constructs based on according to required time and expense that task is handled in processing node
The fitness function of the fitness of calculation task and processing combination of nodes, it is raw according to the input factor and the fitness function
The differential evolution algorithm of Cheng Xin；
Optimize unit, for optimizing by the new differential evolution algorithm to the initial population, obtains task
It is characterized with the set optimum combination of the optimum combination of processing node and the task schedule in combining is handled into node into combination
The time required to being handled and overhead-optimized.
Optionally, the input factor includes intersecting the factor, and the construction unit is also used to:
Intersect factor CR according to the current evolutionary generation t and maximum evolutionary generation T, building:
CR=CRmin+(CRmax-CRmin)(t/T)2；
Wherein, CRminFor preset intersection factor minimum value, CRmaxFor preset intersection factor maximum value.
Optionally, the input factor includes zoom factor, and the construction unit is also used to:
According to the current evolutionary generation t and maximum evolutionary generation T, zoom factor F is constructed:
λ=e1-T/(T+1-t)；
F=F0*2λ；
Wherein, F0For preset zoom factor initial value, λ is the evolutionary generation factor.
Optionally, the optimization unit is also used to:
The initial population is optimized by the new differential evolution algorithm, and is held in each optimization process
The following operation of row, obtains optimal target population until reaching maximum evolutionary generation T, wherein the optimal target population is
For task and processing node optimum combination set:
All offspring individuals generated during suboptimization according to the intersection factor CR are obtained, by all filial generations
Individual is added to the newborn individual set during suboptimization；
Calculate the adaptation of each individual in the fitness of each individual in the newborn individual set and previous generation population
Degree, and the newborn individual set and previous generation population are shunk according to the zoom factor F, acquisition is leaned on by fitness
The next-generation population that the individual of preceding predetermined number is constituted.
The technical effect of task scheduling equipment provided by the present application may refer to each implementation of above-mentioned first aspect
Technical effect, details are not described herein again.
The third aspect provides a kind of equipment, comprising:
At least one processor, and
The memory being connect at least one described processor；
Wherein, the memory is stored with the instruction that can be executed by least one described processor, described at least one
The instruction that device is stored by executing the memory is managed, method as described in relation to the first aspect is executed.
Fourth aspect provides a kind of computer readable storage medium:
A kind of method for scheduling task provided in an embodiment of the present invention, can be according to current evolutionary generation and maximum evolution generation
Number, constructs the input factor of new differential evolution algorithm, and the required time handled according to task in processing node with
And expense, the fitness function of new differential evolution algorithm is constructed, and then optimize by using new differential evolution algorithm,
Obtain by task schedule to processing node handled the time required to and expense OPTIMAL TASK with handle combination of nodes, so
Time and the expense that processing node processing task can be taken into account, realize the scheduling selected between optimal task and computing resource
The purpose of scheme, and promote the performance of task processing platform.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, embodiment will be described below
Needed in attached drawing be briefly described, it should be apparent that, the accompanying drawings in the following description is only of the invention some
Embodiment.
Fig. 1 is a kind of flow chart of method for scheduling task provided in an embodiment of the present invention；
Fig. 2 is a kind of structural schematic diagram of task scheduling equipment provided in an embodiment of the present invention；
Fig. 3 is the structural schematic diagram of another equipment provided in an embodiment of the present invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical solution of the present invention is clearly and completely described, it is clear that described embodiment is skill of the present invention
A part of the embodiment of art scheme, instead of all the embodiments.Based on the embodiment recorded in present specification, this field is general
Logical technical staff every other embodiment obtained without creative efforts, belongs to the technology of the present invention side
The range of case protection.
The high speed development of Internet technology has driven the development of cloud computing technology, as the research heat in cloud computing platform
The task schedule of one of point refers to: according to the various demands of task, different tasks being assigned to cloud by using appropriate strategy
In execute on suitable resource node.
The use cost of different computing resources is different in cloud computing platform, in general, the strong computing resource of processing capacity
Higher operating costs, and the use cost of the weaker computing resource of processing capacity is lower, however, in major part cloud computing at present
Task scheduling approach only using the deadline for shortening task as research emphasis, have ignored computing resource in cloud computing platform
Use cost, be unfavorable for the performance boost of cloud computing platform, so, how to select between optimal task and computing resource
Scheduling scheme is current urgent problem.
For this purpose, the embodiment of the invention provides a kind of method for scheduling task and equipment, it is optimal to solve shortage selection at present
Task and computing resource between scheduling scheme the problem of.
It should be noted that a kind of method for scheduling task provided in an embodiment of the present invention and equipment, in addition to can be applied to
Except the task schedule of cloud computing platform, it is also applied to other any type for needing to carry out task schedule in the prior art
Application platform in.In the present embodiment, specifically for applying in cloud computing platform, to the above-mentioned side of the embodiment of the present invention
Method is described in detail.
Here, being briefly described to the differential evolution algorithm being related in the embodiment of the present invention.
Differential evolution algorithm (Differential Evolution, vehicle economy), and can be described as Differential Evolution Algorithm or micro-
Divide evolution algorithm, is a kind of emerging evolutionary computation technique, it is initially to be mentioned by American scholar Storn and Price in nineteen ninety-five
A kind of bionic intelligence computational algorithm of simulation natural evolution rule out, differential evolution algorithm are mainly used for solving continuous variable
Global Optimal Problem.
Differential evolution algorithm mainly includes variation (Mutation), intersects (Crossover), selection (Selection) three
Kind operation, basic thought is to utilize two randomly selected from group since a certain initial population being randomly generated
Random variation source of the difference vector of body as third individual, it is individual with third according to certain rules after difference vector is weighted
Vector summation, to generate new variation individual, which is to make a variation.Then, variation individual is pre-determined with some
Target individual carries out parameter mixing, generates test individual, this process is to intersect.If the fitness value of test individual is better than mesh
The fitness value of individual is marked, then tests individual in the next generation and replaces target individual, otherwise target individual still preserves, the behaviour
Alternatively.In the evolutionary process of every generation, each individual vector is primary as target individual, and algorithm passes through constantly iteration
It calculates, retains defect individual, eliminate worst individual, guiding search process is approached to globally optimal solution.
The embodiment of the present invention is described in further detail with reference to the accompanying drawings of the specification.It should be appreciated that described herein
Embodiment only for the purpose of illustrating and explaining the present invention and is not intended to limit the present invention.
Referring to FIG. 1, being a kind of method for scheduling task provided in an embodiment of the present invention, comprising:
Step S101: according to multiple tasks to be processed and multiple processing nodes for handling task, generating will appoint
Business is dispatched to the initial population of the task that processing node is handled and processing combination of nodes.
In embodiments of the present invention, energy can be calculated for having in cloud computing platform for handling the processing node of task
Any kind of computing resource of power, such computing resource such as virtual machine, processor etc., here, not doing any restrictions.
The multiple processing sections of method in the embodiment of the present invention determining multiple tasks to be processed and for handling task
After point, so that it may according to above-mentioned multiple processing tasks and multiple processing nodes, generate by task schedule to handle node into
The task of row processing and the initial population of processing combination of nodes are entirely searched in order to ensure population can cover in practical applications
Random function is selected usually to generate initial population in rope space, and the embodiment of the present invention can also be by selecting random function next life
At the initial population of the task and processing combination of nodes that handle task schedule to processing node.
In the task of generation and during handling the initial population of combination of nodes, population scale, Ye Jisheng can also be set
At initial population in include individual quantity, in embodiments of the present invention, the individual of the initial population of generation refers to task
With processing combination of nodes.
Table one:
| Pij(t) | Virtual machine 1 | Virtual machine 2 | Virtual machine 3 | Virtual machine 4 | Virtual machine 5 |
| Task 1 | P11(0) | P12(0) | P13(0) | P14(0) | P15(0) |
| Task 2 | P21(0) | P22(0) | P23(0) | P24(0) | P25(0) |
| Task 3 | P31(0) | P32(0) | P33(0) | P34(0) | P35(0) |
For example, setting population scale is 10, multiple tasks to be processed include: task 1, task 2, task 3；For handling
Multiple processing nodes of task, including virtual machine 1, virtual machine 2, virtual machine 3, virtual machine 4, virtual machine 5 are raw by random function
At include task and to handle the initial population of combination of nodes can be wantonly 10 individuals shown in table one.
Wherein, the x in table oneij(t) single individual namely some task in the initial population generated are indicated and is handled
Combination of nodes, t indicate that evolutionary generation, the t value 0 in initial population, i indicate mission number, and j indicates virtual machine label, such as x11
(0), task 1 is dispatched to the task that virtual machine 1 is handled combines with virtual machine in initial population is indicated.
Step S102: according to current evolutionary generation and maximum evolutionary generation, building is handed over for controlling the initial population
The input factor for pitching variation, according to required time and expense that task is handled in processing node, building is appointed for calculating
The fitness function of business and the fitness of processing combination of nodes.
In practical applications, above-mentioned steps 102 can execute after step 101, can also hold before step S101
Row, can also be performed simultaneously with step S101, not do any restrictions herein.
Preferably, the input factor of the building in above-mentioned steps 102 includes intersecting because of the period of the day from 11 p.m. to 1 a.m, and the basis works as evolution
Algebra and maximum evolutionary generation, construct the input factor for controlling the initial population cross and variation, specifically:
According to current evolutionary generation t and maximum evolutionary generation T, new intersection factor CR is constructed:
CR=CRmin+(CRmax-CRmin)(t/T)2Formula (1)；
Wherein, CRminFor preset intersection factor minimum value, CRmaxFor preset intersection factor maximum value.
In practical applications, intersect factor CR can be described as crossover probability CR again, in the crossover process of differential evolution algorithm, hand over
Fork factor CR will affect the search capability and convergence rate of differential evolution algorithm, intersect factor CR value range be usually [0.5,
1], intersect factor CR value range it is, of course, also possible to adjust according to specific application scenarios, here, to intersect factor CR value
Range be [0.5,1] for, then, preset intersection factor minimum value CRminCan value be 0.5, the preset intersection factor is most
Big value CRmaxCan value be 1.
According to preset intersection factor minimum value CRmin, preset intersection factor maximum value CRmax, current evolutionary generation t with
And maximum evolutionary generation T, construct intersection factor CR new as shown in above-mentioned formula (1), the new intersection factor CR with into
The variation of the algebra of change namely the variation of the number of iterations and update.
For example, when current evolutionary generation t value is 1, according in the current first time evolutionary process of above-mentioned formula (1)
New intersection factor CR value be CRmin+(CRmax-CRmin)(1/T)2；When current evolutionary generation t value is T, according to upper
Stating the new intersection factor CR value known to formula (1) in current the T times evolutionary process is CRmax。
Due to the intersection factor CR new as shown in above-mentioned formula (1) in embodiments of the present invention, rebuild, into
The value at the initial stage of change, CR is smaller, is more conducive to keep population diversity and ability of searching optimum；With the increase of evolutionary generation,
The value of CR is bigger, and algorithm is more scanned for locally, and convergence rate is also faster.Therefore the embodiment of the present invention by with
The number of iterations increases, and constantly increases the method for intersecting factor CR, carrys out improved differential evolution algorithm, and then reach quickening convergence rate
While guarantee to search the probability of globally optimal solution.
Preferably, when the input factor of the building in above-mentioned steps 102 includes zoom factor, the basis works as evolution
Algebra and maximum evolutionary generation, construct the input factor for controlling the initial population cross and variation, specifically:
According to the current evolutionary generation t and maximum evolutionary generation T, new zoom factor F is constructed:
λ=e1-T/(T+1-t)Formula (2)；
F=F0*2λFormula (3)；
Wherein, F0For preset zoom factor initial value, λ is the evolutionary generation factor.
In practical applications, zoom factor can be described as controlling elements again, and zoom factor controls differential evolution algorithm variation
Search range in the process constructs the new contracting as shown in formula (3) with adaptive ability in embodiments of the present invention
The value of putting factor F namely new zoom factor F can be adaptive with the variation of evolutionary generation namely the variation of the number of iterations
It updates, specifically, F has the larger value, and biggish zoom factor F can be scanned in the overall situation, and then be made at the initial stage of algorithm
Diversity of individuals can be kept in the early stage by obtaining population, avoid precocity；With the increase of the number of iterations namely in the algorithm later period, F's
Value constantly reduces, and the solution of Evolution of Population to later period has more excellent information, and small zoom factor F can retain the excellent of solution
Information avoids optimal solution from being destroyed, and increases the probability for searching globally optimal solution, and can also accelerate the convergence speed of algorithm
Degree, improves the efficiency of algorithm.
Further, in embodiments of the present invention, the required time that can also be handled according to task in processing node
And expense, fitness function of the building for calculating task and the fitness for handling combination of nodes, specifically, can be according to reality
Border needs to be arranged the weight for the required time that task is handled in processing node, and setting task is at processing node
The weight of the required expense of reason, for example, the weight of the weight of required time and the expense can be disposed as 50%；It can
To set 80% for the weight of required time, 20% is set by the weight of required expense；It can also be by the power of required time
It resets and is set to 40%, set 60% etc. for the weight of required expense, here, it is set as 80% with the weight of required time, it will
The weight of required expense is set as 20%, fitness letter of the building for calculating task and the fitness for handling combination of nodes
Number are as follows:
Fitness function=0.8* task size * processing node basic frequency+0.2* completes required by task time * processing section
Point horse-power formula (4)；
Step S103: new differential evolution algorithm is generated according to the input factor and the fitness function.
In embodiments of the present invention, construct the new intersection factor as shown in formula (1), as shown in formula (3)
After new zoom factor and the fitness function as shown in formula (4), so that it may based on the new intersection factor, new scaling because
Son and fitness function generate new differential evolution algorithm.
Step S104: optimizing the initial population by the new differential evolution algorithm, obtains task and place
Manage the set of the optimum combination of node.
Wherein, the task schedule in combination is handled node into combination and handled by optimum combination characterization is taken
Between and overhead-optimized.
Using the new differential evolution algorithm of generation, to the task and processing for handling task schedule to processing node
The initial population of combination of nodes optimizes, to obtain the set of task with the optimum combination for handling node.
Preferably, new differential evolution algorithm is being used, to generation by task schedule to handling what node was handled
When task and the initial population of processing combination of nodes optimize, the executable following operation of every suboptimization is maximum until reaching
Evolutionary generation T obtains optimal target population:
Step A: all offspring individuals generated during suboptimization according to the intersection factor CR are obtained, by the institute
The newborn individual set for thering is offspring individual to be added to during suboptimization；
Step B: each individual in the fitness of each individual and previous generation population is calculated in the newborn individual set
Fitness, and the newborn individual set and previous generation population are shunk according to the zoom factor F, are obtained by fitting
The next-generation population that the individual of the forward predetermined number of response is constituted.
Wherein, the optimal target population of acquisition is the set of the optimum combination of task and processing node.
Namely in each evolutionary process, the value of the CR to evolve every time is updated according to formula (1), according to each evolutionary process
In updated CR value and previous generation population generate multiple offspring individuals, multiple offspring individuals of generation are all added to
In newborn individual set in each evolutionary process, each individual in newborn individual set then is calculated using above-mentioned formula (4)
Fitness and previous generation population in each individual fitness, and according to zoom factor F to the newborn individual set with
And previous generation population is shunk, and the next-generation population being made of the individual of the forward predetermined number of fitness is obtained.
For example, being optimized according to the initial population that new evolution algorithm is 10 to scale, in first time evolutionary process
Namely t value is 1, initial population is used as previous generation population namely parent population, brings the value of t namely 1 into formula (1) acquisition the
The value namely CR of the CR once to evolvemin, then utilize CRminCross and variation operation is carried out to previous generation population, generates new
Body namely offspring individual.
All offspring individuals of generation are added in the newborn individual set in first time evolutionary process, adaptability is utilized
Function formula (4) calculates separately the suitable of each individual in the fitness of each individual in newborn individual set and previous generation population
Response, be calculated by formula (3) evolve for the first time by zoom factor F, and using F pairs of the zoom factor is calculated
Newborn individual set and previous generation population are shunk, and to exclude the small individual of fitness, and then are obtained larger by fitness
The individual of predetermined number form next-generation population, wherein above-mentioned predetermined number by calculate to zoom factor F determine,
This is not just discussed excessively.
Then it carries out second to evolve, in the second evolutionary process, t value is 2, brings the value of t into formula (1) and obtains second
The value of the CR of secondary evolution, the value of the CR then to evolve using second (namely obtain previous generation population after evolving for the first time
The previous generation population that next-generation population is evolved as second) cross and variation operation is carried out, generate new individual namely filial generation
Body.
Likewise, all offspring individuals of generation are added in the newborn individual set in second of evolutionary process, benefit
The fitness of each individual and previous generation population in the newborn individual set of second of evolution are calculated separately with fitness function
In each individual fitness, by formula (3) be calculated second of evolution by zoom factor F, and using calculating
Newborn individual set and previous generation population are shunk to zoom factor F, to exclude the small individual of fitness, and then obtained
By the biggish predetermined number of fitness individual form the next generation population, and so on until reaching maximum evolutionary generation T, obtain
Optimal target population namely the set for task and the optimum combination of processing node.
So the above-mentioned method for scheduling task provided is provided through the invention, it can according to current evolutionary generation and most
Macroevolution algebra constructs the input factor of new differential evolution algorithm, and the institute handled according to task in processing node
It takes time and expense, constructs the fitness function of new differential evolution algorithm, and then by using new differential evolution algorithm
Optimize, obtain by task schedule to processing node handled the time required to and expense OPTIMAL TASK with handle node
Combination is realized so time and the expense of processing node processing task can be taken into account and selects optimal task and computing resource
Between scheduling scheme purpose, and promoted task processing platform performance.
Optionally, the new differential evolution algorithm of above-mentioned basis can be realized with program language during concrete practice,
The initial population of the handling task schedule to processing node of the task and processing combination of nodes to generation optimizes, and obtains
The set of the task of obtaining and the optimum combination of processing node, the program language selected can be C language, Java language, C Plus Plus etc.
Any type of program language, does not do any restrictions herein.
Based on the same inventive concept, a kind of task scheduling equipment, the task tune of the equipment are provided in the embodiment of the present invention
The specific implementation of degree method can be found in the description of above method embodiment part, and overlaps will not be repeated, referring to FIG. 2, should
Equipment includes:
Population generation unit 20, for according to multiple tasks to be processed and multiple processing sections for handling task
Point generates the initial population of the task and processing combination of nodes that handle task schedule to processing node；
Construction unit 21, for according to current evolutionary generation and maximum evolutionary generation, building to be described initial for controlling
The input factor of population cross and variation, according to required time and expense that task is handled in processing node, building is used for
The fitness function of the fitness of calculating task and processing combination of nodes, according to the input factor and the fitness function
Generate new differential evolution algorithm；
Optimization unit 22 is appointed for being optimized by the new differential evolution algorithm to the initial population
It is engaged in that the task schedule in combining is handled to section into combination with the set of the optimum combination of the processing node optimum combination characterization
Point handled the time required to and overhead-optimized.
Optionally, the input factor includes intersecting the factor, and the construction unit is also used to:
Intersect factor CR according to the current evolutionary generation t and maximum evolutionary generation T, building:
CR=CRmin+(CRmax-CRmin)(t/T)2；
Wherein, CRminFor preset intersection factor minimum value, CRmaxFor preset intersection factor maximum value.
Optionally, the input factor includes zoom factor, and the construction unit is also used to:
According to the current evolutionary generation t and maximum evolutionary generation T, zoom factor F is constructed:
λ=e1-T/(T+1-t)；
F=F0*2λ；
Wherein, F0For preset zoom factor initial value, λ is the evolutionary generation factor.
Optionally, the optimization unit is also used to:
The initial population is optimized by the new differential evolution algorithm, and is held in each optimization process
The following operation of row, obtains optimal target population until reaching maximum evolutionary generation T, wherein the optimal target population is
For task and processing node optimum combination set:
All offspring individuals generated during suboptimization according to the intersection factor CR are obtained, by all filial generations
Individual is added to the newborn individual set during suboptimization；
Calculate the adaptation of each individual in the fitness of each individual in the newborn individual set and previous generation population
Degree, and the newborn individual set and previous generation population are shunk according to the zoom factor F, acquisition is leaned on by fitness
The next-generation population that the individual of preceding predetermined number is constituted.
The technical effect of task scheduling equipment provided by the present application may refer to each implementation of above-mentioned first aspect
Technical effect, details are not described herein again.
Based on the same inventive concept, a kind of equipment is provided in the embodiment of the present invention, referring to FIG. 3, the equipment includes:
At least one processor 30, and
The memory 31 being connect at least one described processor；
Wherein, the memory is stored with the instruction that can be executed by least one described processor, described at least one
The instruction that device is stored by executing the memory is managed, method for scheduling task as described above is executed.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, complete hardware embodiment, complete software embodiment or reality combining software and hardware aspects can be used in the present invention
Apply the form of example.Moreover, it wherein includes the computer of computer usable program code that the present invention, which can be used in one or more,
The shape for the computer program product implemented in usable storage medium (including but not limited to magnetic disk storage and optical memory etc.)
Formula.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
Mind and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (10)
1. a kind of method for scheduling task characterized by comprising
According to multiple tasks to be processed and multiple processing nodes for handling task, generates and save task schedule to processing
The initial population of task and processing combination of nodes that point is handled；
According to current evolutionary generation and maximum evolutionary generation, construct input for controlling the initial population cross and variation because
Son, according to required time and expense that task is handled in processing node, building is for calculating task and processing node group
The fitness function of the fitness of conjunction；
New differential evolution algorithm is generated according to the input factor and the fitness function；
The initial population is optimized by the new differential evolution algorithm, obtain task and handles the optimal set of node
The set of conjunction, the time required to the optimum combination characterization is handled processing node of the task schedule in combination into combination
And overhead-optimized.
2. the method as described in claim 1, which is characterized in that the input factor includes intersecting the factor, and the basis is current
Evolutionary generation and maximum evolutionary generation, construct the input factor for controlling the initial population cross and variation, specifically:
Intersect factor CR according to current evolutionary generation t and maximum evolutionary generation T, building:
CR=CRmin+(CRmax-CRmin)(t/T)2；
Wherein, CRminFor preset intersection factor minimum value, CRmaxFor preset intersection factor maximum value.
3. method according to claim 2, which is characterized in that the input factor includes zoom factor, and the basis is current
Evolutionary generation and maximum evolutionary generation, construct the input factor for controlling the initial population cross and variation, specifically:
According to the current evolutionary generation t and maximum evolutionary generation T, zoom factor F is constructed:
λ=e1-T/(T+1-t)；
F=F0*2λ；
Wherein, F0For preset zoom factor initial value, λ is the evolutionary generation factor.
4. method as claimed in claim 3, which is characterized in that it is described by the new differential evolution algorithm to described initial
Population optimizes, and obtains task and handles the set of the optimum combination of node, comprising:
The initial population is optimized by the new differential evolution algorithm, and be performed both by each optimization process with
Lower operation obtains optimal target population until reaching maximum evolutionary generation T, wherein the optimal target population is to appoint
The set of business and the optimum combination of processing node:
All offspring individuals generated during suboptimization according to the intersection factor CR are obtained, by all offspring individuals
The newborn individual set being added to during suboptimization；
The fitness of each individual in the fitness of each individual in the newborn individual set and previous generation population is calculated, and
The newborn individual set and previous generation population are shunk according to the zoom factor F, obtained forward by fitness
The next-generation population that the individual of predetermined number is constituted.
5. a kind of task scheduling equipment characterized by comprising
Population generation unit, for generating according to multiple tasks to be processed and multiple processing nodes for handling task
The initial population of task and processing combination of nodes that task schedule to processing node is handled；
Construction unit, for according to current evolutionary generation and maximum evolutionary generation, building to be handed over for controlling the initial population
The input factor for pitching variation, according to required time and expense that task is handled in processing node, building is appointed for calculating
The fitness function of business and the fitness of processing combination of nodes generates new according to the input factor and the fitness function
Differential evolution algorithm；
Optimize unit, for optimizing by the new differential evolution algorithm to the initial population, obtains task and place
The set of the optimum combination of node is managed, the task schedule in combination is handled node progress by the optimum combination characterization into combination
Processing required time and overhead-optimized.
6. equipment as claimed in claim 5, which is characterized in that the input factor includes intersecting the factor, the construction unit
It is also used to:
Intersect factor CR according to current evolutionary generation t and maximum evolutionary generation T, building:
CR=CRmin+(CRmax-CRmin)(t/T)2；
Wherein, CRminFor preset intersection factor minimum value, CRmaxFor preset intersection factor maximum value.
7. equipment as claimed in claim 6, which is characterized in that the input factor includes zoom factor, the construction unit
It is also used to:
According to the current evolutionary generation t and maximum evolutionary generation T, zoom factor F is constructed:
λ=e1-T/(T+1-t)；
F=F0*2λ；
Wherein, F0For preset zoom factor initial value, λ is the evolutionary generation factor.
8. equipment as claimed in claim 7, which is characterized in that the optimization unit is also used to:
The initial population is optimized by the new differential evolution algorithm, and be performed both by each optimization process with
Lower operation obtains optimal target population until reaching maximum evolutionary generation T, wherein the optimal target population is to appoint
The set of business and the optimum combination of processing node:
All offspring individuals generated during suboptimization according to the intersection factor CR are obtained, by all offspring individuals
The newborn individual set being added to during suboptimization；
The fitness of each individual in the fitness of each individual in the newborn individual set and previous generation population is calculated, and
The newborn individual set and previous generation population are shunk according to the zoom factor F, obtained forward by fitness
The next-generation population that the individual of predetermined number is constituted.
9. a kind of equipment characterized by comprising
At least one processor, and
The memory being connect at least one described processor；
Wherein, the memory is stored with the instruction that can be executed by least one described processor, at least one described processor
By executing the instruction of the memory storage, method according to any of claims 1-4 is executed.
10. a kind of computer readable storage medium, it is characterised in that:
The computer-readable recording medium storage has computer instruction, when the computer instruction is run on computers,
So that computer executes such as method of any of claims 1-4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810315075.7A CN110362378A (en) | 2018-04-10 | 2018-04-10 | A kind of method for scheduling task and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810315075.7A CN110362378A (en) | 2018-04-10 | 2018-04-10 | A kind of method for scheduling task and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110362378A true CN110362378A (en) | 2019-10-22 |
Family
ID=68212868
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810315075.7A Pending CN110362378A (en) | 2018-04-10 | 2018-04-10 | A kind of method for scheduling task and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110362378A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110727505A (en) * | 2019-12-17 | 2020-01-24 | 杭州连银科技有限公司 | Distributed task scheduling and service monitoring system capable of being hot-loaded |
| CN114862216A (en) * | 2022-05-16 | 2022-08-05 | 中国银行股份有限公司 | Method and device for determining agile project scheduling scheme |
| CN115204525A (en) * | 2022-09-14 | 2022-10-18 | 中科航迈数控软件(深圳)有限公司 | Processing task scheduling method, device, terminal and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080307399A1 (en) * | 2007-06-05 | 2008-12-11 | Motorola, Inc. | Gene expression programming based on hidden markov models |
| CN105760929A (en) * | 2016-03-11 | 2016-07-13 | 浙江工业大学 | Layered global optimization method based on DFP algorithm and differential evolution |
| CN107885601A (en) * | 2017-10-27 | 2018-04-06 | 重庆邮电大学 | A kind of cloud computing method for scheduling task based on difference and ant group algorithm |
-
2018
- 2018-04-10 CN CN201810315075.7A patent/CN110362378A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080307399A1 (en) * | 2007-06-05 | 2008-12-11 | Motorola, Inc. | Gene expression programming based on hidden markov models |
| CN105760929A (en) * | 2016-03-11 | 2016-07-13 | 浙江工业大学 | Layered global optimization method based on DFP algorithm and differential evolution |
| CN107885601A (en) * | 2017-10-27 | 2018-04-06 | 重庆邮电大学 | A kind of cloud computing method for scheduling task based on difference and ant group algorithm |
Non-Patent Citations (4)
| Title |
|---|
| EDWIN C. SHI等: "Differential Evolution with Adaptive Population Size", 《PROCEEDINGS OF THE 19TH INTERNATIONAL CONFERENCE ON DIGITAL SIGNAL PROCESSING》 * |
| JANEZ BREST等: "Population size reduction for the differential evolution algorithm", 《APPLIED INTELLIGENCE》 * |
| RYOJI TANABE等: "Improving the Search Performance of SHADE Using Linear Population Size Reduction", 《2014 IEEE CONGRESS ON EVOLUTIONARY COMPUTATION (CEC)》 * |
| ZHE ZHENG等: "A Multi-objective Optimization Scheduling Method Based on the Improved Differential Evolution Algorithm in cloud Computing", 《INTERNATIONAL CONFERENCE ON CLOUD COMPUTING AND SECURITY》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110727505A (en) * | 2019-12-17 | 2020-01-24 | 杭州连银科技有限公司 | Distributed task scheduling and service monitoring system capable of being hot-loaded |
| CN110727505B (en) * | 2019-12-17 | 2020-04-10 | 杭州连银科技有限公司 | Distributed task scheduling and service monitoring system capable of being hot-loaded |
| CN114862216A (en) * | 2022-05-16 | 2022-08-05 | 中国银行股份有限公司 | Method and device for determining agile project scheduling scheme |
| CN115204525A (en) * | 2022-09-14 | 2022-10-18 | 中科航迈数控软件(深圳)有限公司 | Processing task scheduling method, device, terminal and storage medium |
| CN115204525B (en) * | 2022-09-14 | 2022-12-16 | 中科航迈数控软件(深圳)有限公司 | Processing task scheduling method, device, terminal and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Hybrid whale optimization algorithm enhanced with Lévy flight and differential evolution for job shop scheduling problems | |
| Shen et al. | Mathematical modeling and multi-objective evolutionary algorithms applied to dynamic flexible job shop scheduling problems | |
| Ashouraie et al. | Priority-based task scheduling on heterogeneous resources in the Expert Cloud | |
| Lin et al. | Minimizing makespan and total flowtime in permutation flowshops by a bi-objective multi-start simulated-annealing algorithm | |
| Sharma et al. | Enhancing the food locations in an artificial bee colony algorithm | |
| Sathappan et al. | Modified genetic algorithm for multiobjective task scheduling on heterogeneous computing system | |
| CN110362378A (en) | A kind of method for scheduling task and equipment | |
| CN112784362A (en) | Hybrid optimization method and system for unmanned aerial vehicle-assisted edge calculation | |
| CN113821318B (en) | Internet of things cross-domain subtask combination collaborative computing method and system | |
| Pooranian et al. | Hybrid metaheuristic algorithm for job scheduling on computational grids | |
| CN110969362A (en) | Multi-target task scheduling method and system under cloud computing system | |
| CN114741955A (en) | Multi-objective optimization task scheduling method based on security cloud | |
| CN111813500A (en) | Multi-target cloud workflow scheduling method and device | |
| CN115293623A (en) | Training method and device for production scheduling model, electronic equipment and medium | |
| CN117271101A (en) | Operator fusion method and device, electronic equipment and storage medium | |
| Deniziak et al. | Cost optimization of real-time cloud applications using developmental genetic programming | |
| CN113504998A (en) | Method, device and equipment for determining task scheduling scheme | |
| Sharma et al. | An efficient task scheduling of multiprocessor using genetic algorithm based on task height | |
| CN115169754B (en) | Energy scheduling method and device, electronic equipment and storage medium | |
| Eswari et al. | Modified multi-objective firefly algorithm for task scheduling problem on heterogeneous systems | |
| CN116450366A (en) | Multi-objective optimal scheduling method, equipment and medium for satellite measurement, operation and control resources | |
| Lattarulo et al. | Application of the multi-objective alliance algorithm to a benchmark aerodynamic optimization problem | |
| CN113220437B (en) | Workflow multi-target scheduling method and device | |
| Aydın et al. | A configurable generalized artificial bee colony algorithm with local search strategies | |
| Chen et al. | Chaotic differential evolution algorithm for resource constrained project scheduling problem |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191022 |
|
| RJ01 | Rejection of invention patent application after publication |