CN105260234A - Independent task list based process dynamic execution optimization method - Google Patents

Abstract

The invention discloses an independent task list based process dynamic execution optimization method. The method comprises the following steps: (1) establishing a process model oriented for execution optimization; 2) solving a task distribution rate based on global optimization; and 3) adopting an independent task list based process execution optimization system structure. According to the method, different efficiencies of resource processing activities in a many-to-many resource and activity support relationship are considered, so that global task distribution optimization can be performed according to a certain goal; and a mixed distribution policy combining a task optimization distribution rate with an idle resource priority, and a re-distribution policy are adopted, so that the situation that one idle resource occurs in process execution and processable tasks exist in an independent task list before other resources due to only adoption of an independent task list mode is eliminated, the resource utilization rate is effectively increased, and the process execution efficiency is further improved.

Description

Process dynamics based on independent task list performs optimization method

Technical field

The present invention relates to computer technology, infotech and systems engineering field, in particular, relate to a kind of process scheduling/execution optimization method.

Background technology

The ordered set (such as customer order processing procedure, parts process of manufacture etc.) of process to be enterprise be a series of activities relevant in logic that a certain target carries out.Movable execution needs the support of resource, and resource can be any entity as personnel, physical equipment, file, application system/program etc.In order to quick customer in response demand; adapt to dynamically changeable market environment; the agility of raising process, robustness; usually can require the degree of specialization of reduction resource, (employee grasps multiple technical ability as required to improve the flexibility degree of process; various rolls can be born; machining center can process polytype part or multiple working procedure etc.), cause the support relation that there is a large amount of multi-to-multis during the course between activity and resource.Therefore all there is the scheduling/execution optimization problem of process in enterprise in the way to manage centered by process, and namely enterprise needs rationally to determine the movable problem being performed by which kind of resource and when performed in the execution of process.Scheduling/execution optimum theory method has had extensive research in the application in the fields such as Job Shop Scheduling, project management and grid computing, and therefore process scheduling/execution optimization has many theoretical methods that can use for reference.But due to the many Stochastic sum uncertain factors in the complicacy (as also exist selection and loop structure) of enterprise process structure and implementation thereof, (arrival as task instances is that continuous print is random, the movable execution time is also random etc.) scheduling/execution optimum theory method of making these traditional generally directly can not be effectively applied to the execution of enterprise process, current heuristic mutation operations/execution the optimization method usually adopted in process performs based on shared queue, such as: identical for active instance (task) type is placed in a shared queue, when there being resource space idle, by FIFO (prerequisite variable), SPT (shortest processing time first), EDD (expiration time is preferential the earliest), MST (minimum slack time is preferential), LNS (maximum succeeding activity quantity is preferential), the priority of the heuristic rules such as LFJ (minimum flexible priority of task) or predefined gives task matching the resource of qualified/ability execution.When there is the manageable task of certain resource, the method for shared queue does not normally allow this resource free time, to belong to non-delayed scheduling.Great majority have the advantages such as algorithm is simple and quick, applied widely based on the resource distribution/method for scheduling task of heuristic rule/priority, but what it was more paid close attention to is local optimum, do not consider the efficiency of resource process different task and the task matching optimization of the overall situation.

Summary of the invention

Object of the present invention is just to overcome the shortcoming that the heuristic process scheduling/execution optimization method based on shared queue exists, and provides a kind of execution optimization method that can be used in business procedure/workflow under resource and movable multi-to-multi support relation adopting independent task list mode and rescheduling strategy to combine.

The present invention is achieved through the following technical solutions above-mentioned purpose: a kind of process dynamics based on independent task list performs optimization method, comprises the steps:

The first step: optimize process of establishing model towards execution.The processing speed of the activity involved at least needing description process to perform of this process model, resource, the logic control relation between activity and activity, the support relation between activity and resource and resource process active instance (task);

Second step: solve task matching rate based on global optimization.Set up the task matching Optimum analyses model based on the overall situation according to the process model set up and different optimization aim, then adopt corresponding majorized function in MATLAB Optimization Toolbox to solve, obtain the task matching rate based on global optimization;

3rd step: adopt the process based on independent task list to perform optimization system structure.

Process based on independent task list performs optimization system structure and comprises: task matching and task choosing.The function of task matching is: how to give corresponding resource task matching when active instance (task) produces, namely task is put in the task list of corresponding resource and goes; The function of task choosing is: when how resource space idle selects an active instance (task) to process from independent task list.Task matching is associated by independent task list in process performs with task choosing, simultaneously operation, co-ordination.

Described method for allocating tasks is described below:

(1) if the partition coefficient resource that is greater than 0 (having right of distribution) is all busy, and partition coefficient be 0 (ex distribution) but have the resource of disposal right available free, so give ex distribution according to the fast priority principle of processing speed task matching, have disposal right, idle, that processing speed is the fastest resource.

(2) if having the idling-resource of right of distribution, the partition coefficient size so according to them between the idling-resource that these have right of distribution recalculates partition coefficient, then presses probability Random assignment according to new partition coefficient.

(3) if all have the resource of disposal right all busy, be so randomly assigned to by probability the resource that has right of distribution according to partition coefficient.

Described task choosing method is described below:

(1) judge that whether the task list of oneself is empty, if then do not forwarded to (2) for sky; If be sky, forward to (3);

(2) from the task list of oneself, a tasks carrying is selected according to heuristic rule (as: SPT, EDD, MST etc.);

(3) judge in other task list, whether there is its accessible task, as then do not forwarded to (4), if any then forwarding to (5);

(4) resource is waited for, until the task list of oneself is not empty, forwards to (2);

(5) from exist, select an accessible tasks carrying according to heuristic rule (as SPT, EDD, MST etc.) other task list that load is maximum.

Beneficial effect of the present invention is: relative to traditional formal style based on shared queue process scheduling/execution optimization method, consider the resource of multi-to-multi and the different efficiency of resource process activity under Activity supporting relation, can set the goal according to one and carry out the task matching optimization of the overall situation; Relative to the existing process implementation strategy based on independent task list, have employed by task optimization partition coefficient in conjunction with the idle preferential mixed allocation strategy of resource and code reassignment strategy, considering under the prerequisite that the task optimization of the overall situation distributes, overcome that simple to adopt independent task list in process performs, there will be a certain resource idle and there is the situation of its manageable task in independent task list before other resource, effectively improve the utilization factor of resource, thus further development execution efficiency.

Accompanying drawing explanation

Fig. 1 be one from the abstract process model building-block of logic out of actual manufacturing enterprise business procedure.

Fig. 2 is the process execution optimization system structure based on independent task list.

Fig. 3 is task matching based on independent task list and task choosing algorithm.

Embodiment

Below in conjunction with certain manufacture process, the invention will be further described:

The first step: optimize process of establishing model towards execution

One comprises from the abstract process model out of actual manufacturing enterprise business procedure:

Movable: A={a1, a2, a3, a4, a5, a6, a7, a8};

Resource: R={r1, r2, r3, r4};

Logic control relation between activity and activity as shown in Figure 1;

Resource and movable support relation:

$U=\left\{\begin{array}{c}<a1,r1>,<a1,r2>,<a1,r4>,<a2,r3>,<a2,r4>,<a3,r1>,<a3,r3>,<a4,r2>,<a4,r4>,<a5,r2>,\\ <a5,r3>,<a6,r1>,<a6,r4>,<a7,r1>,<a7,r2>,<a7,r3>,<a8,r2>,<a8,r3>,<a8,r4>\end{array}\right\}$

The processing speed of resource process active instance (task):

μ _a1,r1＝14，μ _a1,r2＝8，μ _a1,r4＝8，μ _a2,r3＝6，μ _a2,r4＝14，μ _a3,r1＝8，μ _a3,r3＝5，μ _a4,r2＝12，μ _a4,r4＝6，μ _a5,r2＝8，μ _a5,r3＝12，μ _a6,r1＝10，μ _a6,r4＝18，μ _a7,r1＝12，μ _a7,r2＝4，μ _a7,r3＝8，μ _a8,r2＝2，μ _a8,r3＝4，μ _a8,r4＝3。

Second step: solve task matching rate based on global optimization

Hypothetical target is the productive capacity of maximization procedure , so according to the process model set up, the mathematical model of its task matching optimization is:

Objective function:

Constraint condition: $\underset{r\&Element;R_a}{\&Sigma;}{\mathrm{\&xi;}}_{a,r}=1,\&ForAll;a\&Element;A$

$0\&le;{\mathrm{\&xi;}}_{a,r}\&le;1,\&ForAll;<a,r>\&Element;U$

Wherein:

ξ _a,rfor the task optimization partition coefficient required, namely distribute to the original probability of resource r when the active instance (task) of a produces.

A _r＝{a|〈a,r〉∈U},R _a＝{r|〈a,r〉∈U}

When process performs one time, movable a expects that the number of times performed is calculated as follows: f _a1=1, f _a2=1, f _a3=0.4, f _a4=0.6, f _a5=1, f _a7=1.25, f _a8=1.

The mathematical model of above task matching optimization is a nonlinear optimal problem, and the fmaxmin () function of available MATLAB Optimization Toolbox solves.

3rd step: adopt the process based on independent task list to perform optimization system structure

Process based on independent task list performs optimization system structure as shown in Figure 2, and it comprises: task matching and task choosing.The function of task matching is: how to give corresponding resource task matching when active instance (task) produces, namely task is put in the independent task list of corresponding resource and goes; The function of task choosing is: when how resource space idle selects an active instance (task) to process from independent task list.Task matching is associated by independent task list in process performs with task choosing, simultaneously operation, collaborative work, as shown in Figure 3.Described in Fig. 3 " the free time, recalculate partition coefficient according to their partition coefficient size between the resource that has right of distribution " computing method as follows: if for the example (task) of certain movable a, current idle, the resource that has right of distribution is respectively the corresponding allocation rate obtained by second step is respectively partition coefficient after so recalculating is respectively: " IWL " expression " independent task list " described in Fig. 3, " PT " expression " accessible task ".

Above-described embodiment is preferred embodiment of the present invention; it is not the restriction to technical solution of the present invention; as long as without the technical scheme that creative work can realize on the basis of above-described embodiment, all should be considered as falling within the scope of the rights protection of patent of the present invention.

Claims (1)

1. the process dynamics based on independent task list performs optimization method, it is characterized in that: comprise the steps:

The first step: optimize process of establishing model towards execution; The processing speed of the activity involved at least needing description process to perform of this process model, resource, the logic control relation between activity and activity, the support relation between activity and resource and resource process active instance (task);

Second step: solve task matching rate based on global optimization; Set up the task matching Optimum analyses model based on the overall situation according to the process model set up and different optimization aim, then adopt corresponding majorized function in MATLAB Optimization Toolbox to solve, obtain the task matching rate based on global optimization;

3rd step: adopt the process based on independent task list to perform optimization system structure

Process based on independent task list performs optimization system structure and comprises: task matching and task choosing.The function of task matching is: how to give corresponding resource task matching when active instance (task) produces, task is put in the task list of corresponding resource and goes; The function of task choosing is: when how resource space idle selects an active instance (task) to process from independent task list.Task matching is associated by independent task list in process performs with task choosing, simultaneously operation, co-ordination;

Described method for allocating tasks is described below:

(1) if the partition coefficient resource that is greater than 0 (having right of distribution) is all busy, and partition coefficient be 0 (ex distribution) but have the resource of disposal right available free, so give ex distribution according to the fast priority principle of processing speed task matching, have disposal right, idle, that processing speed is the fastest resource;

(2) if having the idling-resource of right of distribution, the partition coefficient size so according to them between the idling-resource that these have right of distribution recalculates partition coefficient, then presses probability Random assignment according to new partition coefficient;

(3) if all have the resource of disposal right all busy, be so randomly assigned to by probability the resource that has right of distribution according to partition coefficient;

Described task choosing method is described below:

(1) judge that whether the task list of oneself is empty, if then do not forwarded to (2) for sky; If be sky, forward to (3);

(2) from the task list of oneself, a tasks carrying is selected according to heuristic rule (as: shortest processing time first, the earliest expiration time priority scheduling preferential, minimum slack time);

(3) judge in other task list, whether there is its accessible task, as then do not forwarded to (4), if any then forwarding to (5);

(4) resource is waited for, until the task list of oneself is not empty, forwards to (2);

(5) from exist, select an accessible tasks carrying according to heuristic rule (as preferential in shortest processing time first, the earliest expiration time, minimum slack time priority scheduling) other task list that load is maximum.