CN102542404A - Aircraft crew scheduling method and system realizing the same - Google Patents

Abstract

The invention discloses an aircraft crew scheduling method, which comprises the following steps: (1) acquiring original data meeting requirements of a user from a user interface and a database, and analyzing, converting and arranging the original data to acquire all essential data meeting required forms; (2) carrying constraint control on the essential data according to a scheduling rule, generating all probably scheduled task combinations for each aircraft crew, namely generating a relative task string LoW for each aircraft crew, and evaluating and correcting the task strings LoW; (3) converting all corrected task strings into a task column and establishing a constraint matrix; (4) solving the constraint matrix by utilizing an algorithm of disassembling a large-scale problem formed by integer programming problems into sub problems; and (5) outputting calculation results. The invention further discloses a system for realizing the aircraft crew scheduling method, wherein the system comprises a data input module, a model generation module, a solving module, a control module, a basic module and the like. According to the system for realizing the aircraft crew scheduling method, disclosed by the invention, requirements on equitable distribution of flight number plane, aircraft crew scheduling and flight task are met through cooperation of all modules.

Description

A kind of unit scheduling method and realize the system of this unit scheduling method

Technical field

The present invention relates to civil aviaton's technical field, particularly relate to a kind of unit scheduling method, also relate to the system that realizes this unit scheduling method.

Background technology

Recent two decades comes; Along with ground sustainable growth of air market scale and airline scale ground constantly enlarge; The unit problem of arranging an order according to class and grade that with the shot array is essential characteristic has become the significant challenge that the operation management of large aerospace company is faced, and the traditional hand operating type can't provide feasible scheme at the appointed time.

At present; Domestic operator mainly arranges technology and free hand modification adjustment to accomplish the unit work of arranging an order according to class and grade through introducing external DSS automatically; This mode has been alleviated the matter of great urgency of aviation enterprise operation management effectively, but still has following weak point in the long run:

(1) unit arrange an order according to class and grade be one the scheduling human resources system; System with the scheduling machine compares; It has very complicated relatedly with other subsystems of society, therefore is that the foreign technology of principal market very is not fit to the actual environment that China domestic operator is faced with America and Europe.

The practical experience in (2) ten years shows; External correlation technique its domestic application effect is poor; Manual setting complexity height and workload are big, have expended great amount of manpower, in addition; The irrationality of scheduling causes or has increased the weight of the unit contradiction between management and running department and crew and the crew of arranging an order according to class and grade, the final like this whole enterprise competitive power of having damaged.

In addition; Two Chinese invention patent applications (application number 200810167327.2, application number 200610170019.6) disclose Workforce Management and method respectively; Though adopted the different techniques scheme, the two all is that the network call center telephonist is arranged an order according to class and grade, and not only has stronger professional specific aim; And in the process of arranging an order according to class and grade, need correction repeatedly to predict the outcome, and be not suitable for the mathematical model that data are huge or characteristics fluctuation is less.

The unit technology of arranging an order according to class and grade is a kind of professional system to type airline, because the complicacy of this technology makes it still not have application at home, still is in space state.Therefore, independent research is the urgent task that China is faced in civil aviaton's technical field towards the unit technology of arranging an order according to class and grade of domestic large aerospace company.

Summary of the invention

First purpose of the present invention provides a kind of unit scheduling method; Can effectively solve aircrew in the daily operation management work of the large aerospace company problem of arranging an order according to class and grade, and when accomplishing flight planning, make fair and reasonable Task Distribution scheme for the aircrew.

First purpose of the present invention can realize that a kind of unit scheduling method is characterized in that may further comprise the steps through following measure:

(1) obtains the raw data of The profile from user interface and database, raw data is resolved, changes, put in order, to obtain whole basic datas of the form of meeting the requirements;

(2) based on the rule of arranging an order according to class and grade basic data is retrained control,, be each crew and generate corresponding task string LoW, again task string LoW is estimated correction for each crew generates all the task combinations that possibly arrange;

(3) the whole task string LoW after will proofreading and correct are converted into the task row and set up constraint matrix.

(4) adopting the extensive PROBLEM DECOMPOSITION that integer programming problem is formed is the algorithm of subproblem, and constraint matrix is found the solution;

(5) output result of calculation.

The present invention is through constraint control and estimate correction; Set up the constraint matrix that meets domestic actual application environment; And the extensive PROBLEM DECOMPOSITION that proposes integer programming problem is formed is the algorithm of subproblem; Realized improving unit arrange an order according to class and grade the result, improve the purpose of system running speed; Can handle the various aeronautical data amount of domestic large aerospace company well, solve aircrew in the daily operation management work of the large aerospace company problem of arranging an order according to class and grade effectively, and when accomplishing flight planning, made fair and reasonable Task Distribution scheme for the aircrew.

The rule of arranging an order according to class and grade in the above step (2) comprises flight and rule, license qualification rule, the quality rule etc. of having a rest, and these rules obtain according to the relevant laws and regulations and the quality of arranging an order according to class and grade.

As one embodiment of the present invention; In said step (2); It is to adopt the mode of calculating target factor that each task string LoW is estimated correction, and computing formula is:

s＞1

ψ _kThe ideal value of-Di k item element

χ _kThe actual value of-Di k item element

The present invention is above to be calculated the target factor formula and is drawn by the following unit integer programming model of arranging an order according to class and grade:

$\left(1\right),\min \mathrm{\Σ}{c}_j\·x_j+M\·\underset{k}{\mathrm{\Σ}}{y}_k$

$\left(2\right),s.t.{\mathrm{\Σ}}_j{a}_{\mathrm{ij}}\·x_j+y_i^F+y_i^M={\mathrm{\δ}}_i\mathrm{for}\∀t_i\∈T$

$\left(3\right),{\mathrm{\Σ}}_j{b}_{\mathrm{kj}}\·x_j=1\mathrm{for}\∀k\∈C$

$\left(5\right),\left\{\begin{array}{c}{\mathrm{\Σ}}_j{q}_{\mathrm{ij}}\·x_j+y_i^B\≤{\mathrm{\δ}}_i^B\mathrm{for}\∀t_i\∈T\\ y_i^D\≤{\mathrm{\δ}}_i^D\\ y_i^B+y_i^D-y_i^E-y_i^M=0\end{array}\right.$

$\left(6\right),\left\{\begin{array}{c}{\mathrm{\&Sigma;}}_j{u}_{\mathrm{ij}}\&CenterDot;x_j+y_i^M-s_i^{E1}-s_i^{E2}+{\mathrm{\&sigma;}}_i=0\mathrm{for}\&ForAll;i\&Element;Q\&SubsetEqual;T\\ s_i^{\mathrm{<figure-callout\; id="1"\; label="E"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">E</figure-callout>}1}=s_i^{E2}\end{array}\right.$

(7)x _j∈{0，1}，yinteger

T representes all set of tasks to be allocated, and C representes the set that the aircrew constitutes, t _iRepresent i task, c _jThe target factor of representing j bar LoW, M representes the target factor y of surplus variable _kThe expression decision variable, task t _iUnderstaffing's number, a _IjRepresent whether j bar LoW comprises i task,

Male sex understaffing's number of representing i task,

Women understaffing's number of representing i task, δ _iRepresent the staffing of i task, b _KjIf expression j bar LoW belongs to personnel k, then b _KjEqual 1, otherwise be 0; P _kExpression has the set of the task formation of certain attribute, P _k∈ T; p _IjExpression is if j bar LoW comprises task t _i, and t _iBelong to set P _k, and its corresponding personnel have certain specific qualification, then p _IjEqual 1, otherwise be 0; q _IjExpression is if j bar LoW comprises task t _i, and its corresponding personnel accomplish this task, then q according to this rank _IjEqual 1, otherwise be 0; Expression task t _iOther staffing of fooled front-seat class,

Expression task t _iOn the staffing that can demote and carry out. The expression decision variable, task t _iFooled front-seat class not vacant post number; The expression decision variable, task t _iThe vacant post number of last degradation; u _IjRepresent that j bar LoW comprises task i, and personnel are the male sex under it; Q representes all or part of set that comprises the task formation of spending the night.σ _iBe illustrated in the walkthrough has been male sex's quantity of i Task Distribution.

indicates slack variables.

The computing formula of the ideal value of element of the present invention is:

${\mathrm{\&psi;}}_k=\underset{j=1}{\overset{H}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_{k,j}{r}_j$

η _{K, j}The fate of letting fly away of representing k people j group

r _jThe target ratio of representing the j group;

Concrete calculation procedure is following:

1. according to qualification, qualifications and record of service factor, personnel's set is divided, personnel distribute the identical equalization task amount of treating when letting fate fly away and equate in same group;

2. statistics is treated the equalization task amount: promptly deduct manual walkthrough personnel in the whole balanced interval equalization task amount of treating of arranging: the statistics row of treating flight arranging an order according to class and grade interval in the equalization task amount of treating of generation; This statistical value is added up according to task rank; The interval beginning of statistical equilibrium is to the interval beginning of arranging an order according to class and grade; The row that can arrange personnel treats the equalization task amount; If abandon treating the equalization task amount because of individual reason, then be designated as this people in row's task, this statistical value is added up operational staff's rank as task rank; Balanced interval all tasks are sued for peace by rank;

3. add up the fate of letting fly away that all can the row personnel belong to each fair group;

4. calculate target ratio;

5. calculate ideal value.

The present invention comes to calculate respectively target ratio according to the relation between each group task, wherein, and ξ ₁..., ξ _HWhen representing that respectively the 1st to H group personnel only arrange this rank flight task, fate spends the night in the executable world; η ₁..., η _HThat representes the 1st to H group personnel respectively lets the fate summation fly away; The concrete steps of calculating target ratio are following:

A. world task of spending the night of each group is identical, suppose to require letting fly away under the identical situation of fate, and the world that i the organizes fate weight of spending the night is λ _i, then

$r_i={\mathrm{\&lambda;}}_i\mathrm{\&xi;}/\underset{j=1}{\overset{H}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_j{\mathrm{\&eta;}}_j$

Wherein, r _iThe target ratio of representing the i group, ξ=ξ ₁=...=ξ _H

B. world task of spending the night of each group is non-intersect, but some group personnel can carry out the world of some other group task of spending the night, otherwise, then can not, concrete steps are following:

B.1, find all satisfactory group right: the inside comprises two groups, and wherein lineup person can arrange the task of another group, supposes to have h group right, k the group to the lining, k ₁Group personnel can arrange k ₂Group task is being let fly away under the identical situation of fate, and two groups the world fate weight of spending the night does

B.2, whether all groups are to satisfying

k=1 in inspection; ..., h;

A) if satisfy;

k=1 then; ...; H, change over to step b.3;

B) if do not satisfy, find arbitrary ungratified group to hypothesis l,

$r=\frac{{\mathrm{\&xi;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}+{\mathrm{\&xi;}}_{l_2}}{{\mathrm{\&lambda;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}{\mathrm{\&eta;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}+{\mathrm{\&lambda;}}_{l_2}{\mathrm{\&eta;}}_{l_2}}$

${\mathrm{<figure-callout\; id="1"\; label="r\; l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">r</figure-callout>}}_{l_{}}={\mathrm{\&lambda;}}_{l_1}r,$ $r_{l_2}={\mathrm{\&lambda;}}_{l_2}r$

${\mathrm{\&xi;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}={\mathrm{<figure-callout\; id="1"\; label="r\; l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">r</figure-callout>}}_{l_{}}{\mathrm{\&eta;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1},$ ${\mathrm{\&xi;}}_{l_2}=r_{l_2}{\mathrm{\&eta;}}_{l_2}$

Repeat this step;

B.3, calculate end;

C. world task of spending the night of each group is non-intersect, but some group personnel can and must carry out the part world of other group task of spending the night, then

$r_i=\frac{{\mathrm{\&xi;}}_i+\mathrm{\&Delta;}{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},$ i＝1，...，H

Wherein, Δ ξ _iRepresent the i group world fate variable quantity that spends the night, if reduce, then this is worth for negative; This value must knownly maybe can be asked, otherwise can't carry out equilibrium;

D. some group is spent the night, and task intersects but inequality, and concrete computation process is following:

D.1, for all two groups of i that intersect, j supposes ξ ₀Represent two groups of worlds that can fly task fate that spends the night,

A) if $\frac{{\mathrm{\&xi;}}_i/{\mathrm{\&eta;}}_i}{({\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_j}\&le;\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j},$ Then $\begin{array}{c}{r}_i=\frac{{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},r_j=\frac{{\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0}{{\mathrm{\&eta;}}_j}\\ {\mathrm{\&xi;}}_i={\mathrm{\&xi;}}_i,{\mathrm{\&xi;}}_j={\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0\end{array}$

B) if $\frac{({\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_i}{{\mathrm{\&xi;}}_j/{\mathrm{\&eta;}}_j}\&le;\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j}<\frac{{\mathrm{\&xi;}}_i/{\mathrm{\&eta;}}_i}{({\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_j},$ Then $\begin{array}{c}r=\frac{{\mathrm{\&xi;}}_1+{\mathrm{\&xi;}}_2-{\mathrm{\&xi;}}_0}{{\mathrm{\&lambda;}}_i{\mathrm{\&eta;}}_i+{\mathrm{\&lambda;}}_j{\mathrm{\&eta;}}_j}\\ r_i={\mathrm{\&lambda;}}_{i}r,r_j={\mathrm{\&lambda;}}_{j}r\end{array}$

C) if $\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j}<\frac{({\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_i}{{\mathrm{\&xi;}}_j/{\mathrm{\&eta;}}_j},$ Then $\begin{array}{c}{r}_i=\frac{{\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0}{{\mathrm{\&eta;}}_i},r_j=\frac{{\mathrm{\&xi;}}_j}{{\mathrm{\&eta;}}_j}\\ {\mathrm{\&xi;}}_i={\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0,{\mathrm{\&xi;}}_j={\mathrm{\&xi;}}_j\end{array}$

D.2, repeating step d.1,, do not have crossing two groups up to finding;

E. each group task of spending the night is non-intersect, and each group personnel can not carry out the task of other group, then

$r_i=\frac{{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},$ i＝1，...，H。

The present invention sets up constraint matrix and may further comprise the steps in said step (3):

(3) a. is provided with the N value;

(3) b. calculates separate

of current relaxation problem

(3) c. selects an employee i according to current lax separating

, only keeps j LoW and the empty LoW of i;

If a) do not have the employee to be locked and N＞0, N=N-1 changes step (3) b. over to;

B) as if N=0, $\mathrm{\&epsiv;}=\mathrm{Max}\{{\stackrel{\&OverBar;}{x}}_{\mathrm{Ij}}:{\stackrel{\&OverBar;}{x}}_{\mathrm{Ij}}<1\},$ Change step (3) d. over to;

(3) d. is if ε＜MINEPS then changes step (3) e.; Otherwise, change step (3) a. over to;

(3) e. calculates the integer solution

of current problem

(3) f. separates according to current, selects an employee i, and the invalid LoW of i is made as effectively again, changes step (3) e. over to; If there is not such employee, change step (5) over to.

Step of the present invention (3) b. calculates separating and adopting following steps of current relaxation problem; Remember that current primal problem is (P), its dual problem is (D):

(3) b.1, if no initial solution, be that each employee selects the LoW that only comprises the LoW of a task and do not comprise any task, add whole slack variables and surplus variable formation initial problem, find the solution this initial problem, obtain initial feasible solution.

(3) b.2, if no initial dual solution is established initial dual solution π=0;

(3) b.3, with having the row of smallest reduction cost and the row in the initial solution, construct initial subproblem (SP);

(3) b.4, find the solution current subproblem (SP) with simplicial method, acquisition optimum solution

and dual solution ρ;

(3) b.5, if ρ is feasible with respect to (D); Then and ρ are respectively original and dual optimal solutions; Otherwise, change over to step (3) b.6;

(3) b.6: make π '=θ π+(1-θ) ρ, wherein

(3) b.7: the whole nonbasic variables in the deletion (SP);

(3) b.8, select the preceding n minimum with respect to the reduction cost of π ' ₁Individual variable joins (SP);

(3) b.9, all with respect to the reduction cost of ρ less than 0 variable in, select n ₂The minimum variable of individual reduction cost with respect to π ' joins (SP), change over to step (3) b.4.

In said step (3) c.; Select an employee i according to current lax separating ; If exist only to keep j LoW and the empty LoW of i; All the other are labeled as invalid; Note is made potential (i)=j, and is called employee i and has locked j LoW.

In said step (3) e., establish the LoW that current integer solution is selected for employee i, note is made select (i), if select (i) ≠ potential (i), and do not visited by this step, then can recover it and be labeled as invalid LoW.

Second purpose of the present invention provides the system that realizes above-mentioned unit scheduling method.

Second purpose of the present invention can realize that a kind of system that realizes above-mentioned unit scheduling method is characterized in that comprising through following measure:

Data input module is used to obtain the raw data that satisfies customer requirements and this raw data is resolved, changes and be organized into the basic data of the form of meeting the requirements;

The model generation module; According to unit arrange an order according to class and grade the rule basic data is retrained control; Generate all the task combinations that to arrange for each crew; Be each crew and generate corresponding task string LoW, again task string LoW is estimated correction, all the task string LoW after proofreading and correct are converted into the task row and set up constraint matrix;

Find the solution module, being used to adopt the extensive PROBLEM DECOMPOSITION that integer programming problem is formed is that the algorithm of subproblem is found the solution constraint matrix, and output result of calculation;

Control module, the operation that is used for control data load module, model generation module and finds the solution module;

Basic module, the generation module that is used to supply a model, required parameter, solver, data layout and data structure when finding the solution the module operation;

Said control module is respectively with said data input module, model generation module and find the solution module and be connected; Said basic module respectively with the model generation module, find the solution module and be connected; The operation of each module of said control module control, said data input module are obtained the raw data that satisfies customer requirements and the basic data of the form of meeting the requirements is resolved, changes and be organized into to this raw data, after said model generation module obtains this basic data; According to unit arrange an order according to class and grade the rule basic data is retrained control; For each crew generates all the task combinations that possibly arrange, promptly generate corresponding task string LoW, again task string LoW is estimated correction; All task string LoW after proofreading and correct are converted into the task row and set up constraint matrix, and said finding the solution found the solution and result of calculation exported after module obtains constraint matrix; Said data input module with find the solution the said basic module of module run time call.

As one embodiment of the present invention; Said data input module is included as the Subscriber Interface Module SIM and the DBM that raw data is provided that the user provides the window with different choice; The user selects through Subscriber Interface Module SIM, and said control module is called DBM so that the raw data that satisfies customer requirements to be provided according to user's requirement.

The window of selection that data input module according to the invention provides comprises following menu: the time period of arranging an order according to class and grade, the rank of arranging an order according to class and grade, the type of arranging an order according to class and grade, collapsing rule etc.

The present invention has following embodiment; Said model generation module comprises LoW generation module, rule inspection module, target factor computing module and model conversion module; Said LoW generation module is connected with said rule inspection module, target factor computing module and basic module respectively; Said rule inspection module also is connected with basic module, and the target factor computing module also is connected with basic module, the model conversion module also respectively with find the solution module, basic module is connected; After said LoW generation module obtains this basic data; The unit that rule-based inspection module provides is arranged an order according to class and grade rule to the constraint of basic data; Generate all the task combinations that to arrange for each crew; Be each crew and generate corresponding task string LoW, said target factor computing module is estimated correction to task string LoW, and all the task string LoW after said model conversion module will be proofreaied and correct are converted into the task row and set up constraint matrix.

The present invention also has following embodiment; Said basic module comprises that being used to each module the systematic parameter module of operational factor is provided, being used for the original data conversion of database is the basic data management module of the basic data of the form that meets the requirements, the virtual data layer module that is used to raw data and the master tool library module of based process is provided, is used to that the virtual solver module of solver interface is provided and is used to provide uniform data format and data structure; Said systematic parameter module is with basic data management module, target factor computing module, model conversion module and find the solution module and be connected; Said basic data management module also is connected with master tool library module, virtual data layer module respectively; Said master tool library module also is connected with virtual data layer module, basic data management module, said virtual solver module with find the solution module and be connected.

The present invention also comprises the peripheral module that is used to write down and feed back total system ruuning situation, and said peripheral module is connected with control module, LoW generation module and basic data management module respectively.

Compared with prior art, the present invention has following significant effect:

(1) the present invention proofreaies and correct through constraint control and evaluation; Set up the constraint matrix that meets domestic actual application environment; And to propose the extensive PROBLEM DECOMPOSITION that integer programming problem forms be that the extensive PROBLEM DECOMPOSITION of algorithm of subproblem is the algorithm of subproblem; Realized improving unit arrange an order according to class and grade the result, improve the purpose of system running speed, very be fit to the actual environment that China domestic operator is faced.

(2) effect of the present invention is good; Can satisfy flight planning, save human resources greatly, because scheduling rationally; Can avoid or alleviate the unit contradiction between management and running department and crew and the crew of arranging an order according to class and grade, strengthen the whole enterprise competitive power.

(3) the present invention can handle the various aeronautical data amount of domestic large aerospace company well; System maintenance is easy; Solved aircrew in the daily operation management work of the large aerospace company problem of arranging an order according to class and grade effectively, and when accomplishing flight planning, made fair and reasonable Task Distribution scheme for the aircrew.

(4) the present invention adopts high-efficient algorithm; With big PROBLEM DECOMPOSITION is the subproblem that is easy to find the solution; The remarkable speed of convergence of accelerating algorithm; So that the unit that in a few minutes, obtains comprising thousands of variablees the lax of problem of arranging an order according to class and grade separated, and than the algorithm that external related system adopted, adapts to the various aeronautical data amount of domestic large aerospace company better.

(5) the present invention adopts the lax method of separating of LP of finding the solution in solution procedure, compares with interior point method with the existing simplicial method that adopts, and is more suitable for finding the solution of constraint matrix of the present invention.

(6) the present invention is finding the solution on the speed fast 10 times, and internal memory is practiced thrift more than 50%; The method of finding the solution integer programming problem has superperformance, this method obtains separate and optimal L P is lax between separating difference often less than 5%.

(7) practice result shows, can in 3 hours, accomplish the arrangement of 1 month about 9000 task of 2000 employees, and the flight planning completion rate reaches more than 98%; According to sample survey, about 40% aircrew thinks that the result fairness of arranging an order according to class and grade, 55% aircrew think basically fair.

Description of drawings

Below in conjunction with accompanying drawing and specific embodiment invention is described in further detail.

Fig. 1 is a FB(flow block) of the present invention;

Fig. 2 is a target factor calculation process block diagram;

Fig. 3 is the process flow diagram that the present invention finds the solution constraint matrix;

Fig. 4 is the process flow diagram that the present invention finds the solution the LP problem;

Fig. 5 is that system of the present invention forms synoptic diagram.

Embodiment

Shown in Fig. 1～4, be a kind of unit scheduling method of the present invention, specifically comprise following calculation procedure:

(1) obtains the raw data of The profile from user interface and database, raw data is resolved, changes, put in order, to obtain whole basic datas of the form of meeting the requirements;

(2) based on the rule of arranging an order according to class and grade basic data is retrained control,, be each crew and generate corresponding task string LoW, again task string LoW is estimated correction for each crew generates all the task combinations that possibly arrange;

(3) the whole task string LoW after will proofreading and correct are converted into the task row and set up constraint matrix;

(4) adopting the extensive PROBLEM DECOMPOSITION that integer programming problem is formed is the algorithm of subproblem, and constraint matrix is found the solution;

(5) output result of calculation.

The rule of arranging an order according to class and grade in the above step (2) comprises flight and rule, license qualification rule, the quality rule etc. of having a rest, and these rules obtain according to the relevant laws and regulations and the quality of arranging an order according to class and grade.

In the step (2); It is to adopt the mode of calculating target factor that each task string LoW is estimated correction, and computing formula is:

s＞1

ψ _kThe ideal value of-Di k item element

χ _kThe actual value of-Di k item element

More than calculating the target factor formula is drawn by the following unit integer programming model of arranging an order according to class and grade:

$\left(1\right),\min \mathrm{\&Sigma;}{c}_j\&CenterDot;x_j+M\&CenterDot;\underset{k}{\mathrm{\&Sigma;}}{y}_k$

$\left(2\right),s.t.{\mathrm{\&Sigma;}}_j{a}_{\mathrm{ij}}\&CenterDot;x_j+y_i^F+y_i^M={\mathrm{\&delta;}}_i\mathrm{for}\&ForAll;t_i\&Element;T$

$\left(3\right),{\mathrm{\&Sigma;}}_j{b}_{\mathrm{kj}}\&CenterDot;x_j=1\mathrm{for}\&ForAll;k\&Element;C$

$\left(5\right),\left\{\begin{array}{c}{\mathrm{\&Sigma;}}_j{q}_{\mathrm{ij}}\&CenterDot;x_j+y_i^B\&le;{\mathrm{\&delta;}}_i^B\mathrm{for}\&ForAll;t_i\&Element;T\\ y_i^D\&le;{\mathrm{\&delta;}}_i^D\\ y_i^B+y_i^D-y_i^E-y_i^M=0\end{array}\right.$

$\left(6\right),\left\{\begin{array}{c}{\mathrm{\&Sigma;}}_j{u}_{\mathrm{ij}}\&CenterDot;x_j+y_i^M-s_i^{E1}-s_i^{E2}+{\mathrm{\&sigma;}}_i=0\mathrm{for}\&ForAll;i\&Element;Q\&SubsetEqual;T\\ s_i^{E1}=s_i^{E2}\end{array}\right.$

(7)x _j∈{0，1}，yinteger

Formula (1) is an objective function, the corresponding target factor of each task string LoW, and it is the fairness measurement of corresponding employee being carried out this LoW; In addition, when reaching justice, also to minimize like surplus variables such as understaffing's numbers.Therefore, formula (1) is formed by two, and first is the fairness variable, and second is other decision variable, and the minimum value of two additions constitutes objective function.

Formula (2) is that task covers constraint, and except that the virtual task that generates, all tasks must be full.

Formula (3) is an assignment constraints, and everyone can only and must select a LoW;

Formula (4) is qualification constraint, and every qualification stipulated by business rule, like the number that must have certain qualification for some task less than certain numerical value.

Formula (5) is the degradation constraint, and high level aircrew can demote and execute the task, but the quantity that each task can be demoted is limited, and this quantitative value is stipulated by concrete business rule.

Formula (6) is the constraint of spending the night, and the constraint content is stipulated by business, like some mission requirements male sex's that spends the night the even number that adds up to.

Formula (7) is the span constraint of decision variable.

T representes all set of tasks to be allocated, and C representes the set that the aircrew constitutes, t _iRepresent i task, c _jThe target factor of representing j bar LoW, M representes the target factor y of surplus variable _kThe expression decision variable, task t _iUnderstaffing's number, a _IjRepresent whether j bar LoW comprises i task, Male sex understaffing's number of representing i task, Women understaffing's number of representing i task, δ _iRepresent the staffing of i task, b _KjIf expression j bar LoW belongs to personnel k, then b _KjEqual 1, otherwise be 0; P _kExpression has the set of the task formation of certain attribute, P _k∈ T; p _IjExpression is if j bar LoW comprises task t _i, and t _iBelong to set P _k, and its corresponding personnel have certain specific qualification, then p _IjEqual 1, otherwise be 0; q _IjExpression is if j bar LoW comprises task t _i, and its corresponding personnel accomplish this task, then q according to this rank _IjEqual 1, otherwise be 0;

Expression task t _iOther staffing of fooled front-seat class,

Expression task t _iOn the staffing that can demote and carry out.

The expression decision variable, task t _iFooled front-seat class not vacant post number;

The expression decision variable, task t _iThe vacant post number of last degradation; u _IjRepresent that j bar LoW comprises task i, and personnel are the male sex under it; Q representes all or part of set that comprises the task formation of spending the night.σ _iBe illustrated in the walkthrough has been male sex's quantity of i Task Distribution.

indicates slack variables.

Wherein, the computing formula of the ideal value of element is:

${\mathrm{\&psi;}}_k=\underset{j=1}{\overset{H}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_{k,j}{r}_j$

η _{K, j}The fate of letting fly away of representing k people j group

r _jThe target ratio of representing the j group;

Concrete calculation procedure is following:

1. according to qualification, qualifications and record of service factor, personnel's set is divided, personnel distribute the identical equalization task amount of treating when letting fate fly away and equate in same group;

2. statistics is treated the equalization task amount: promptly deduct manual walkthrough personnel in the whole balanced interval equalization task amount of treating of arranging: the statistics row of treating flight arranging an order according to class and grade interval in the equalization task amount of treating of generation; This statistical value is added up according to task rank; The interval beginning of statistical equilibrium is to the interval beginning of arranging an order according to class and grade; The row that can arrange personnel treats the equalization task amount; If abandon treating the equalization task amount because of individual reason, then be designated as this people in row's task, this statistical value is added up operational staff's rank as task rank; Balanced interval all tasks are sued for peace by rank;

3. add up the fate of letting fly away that all can the row personnel belong to each fair group;

4. calculate target ratio;

5. calculate ideal value.

Come to calculate respectively target ratio according to the relation between each group task, wherein, ξ ₁..., ξ _HWhen representing that respectively the 1st to H group personnel only arrange this rank flight task, fate spends the night in the executable world; η ₁..., η _HThat representes the 1st to H group personnel respectively lets the fate summation fly away; The concrete steps of calculating target ratio are following:

A. world task of spending the night of each group is identical, suppose to require letting fly away under the identical situation of fate, and the world that i the organizes fate weight of spending the night is λ _i, then

$r_i={\mathrm{\&lambda;}}_i\mathrm{\&xi;}/\underset{j=1}{\overset{H}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_j{\mathrm{\&eta;}}_j$

Wherein, r _iThe target ratio of representing the i group, ξ=ξ ₁=...=ξ _H

B. world task of spending the night of each group is non-intersect, but some group personnel can carry out the world of some other group task of spending the night, otherwise, then can not, concrete steps are following:

B.1, find all satisfactory group right: the inside comprises two groups, and wherein lineup person can arrange the task of another group, supposes to have h group right, k the group to the lining, k ₁Group personnel can arrange k ₂Group task is being let fly away under the identical situation of fate, and two groups the world fate weight of spending the night does

B.2, whether all groups are to satisfying

k=1 in inspection; ..., h;

A) if satisfy;

k=1 then; ...; H, change over to step b.3;

B) if do not satisfy, find arbitrary ungratified group to hypothesis l,

$r=\frac{{\mathrm{\&xi;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}+{\mathrm{\&xi;}}_{l_2}}{{\mathrm{\&lambda;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}{\mathrm{\&eta;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}+{\mathrm{\&lambda;}}_{l_2}{\mathrm{\&eta;}}_{l_2}}$

$r_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}={\mathrm{\&lambda;}}_{l_1}r,$ $r_{l_2}={\mathrm{\&lambda;}}_{l_2}r$

${\mathrm{\&xi;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1}={\mathrm{<figure-callout\; id="1"\; label="r\; l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">r</figure-callout>}}_{l_{}}{\mathrm{\&eta;}}_{{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">l</figure-callout>}}_1},$ ${\mathrm{\&xi;}}_{l_2}=r_{l_2}{\mathrm{\&eta;}}_{l_2}$

Repeat this step;

B.3, calculate end;

C. world task of spending the night of each group is non-intersect, but some group personnel can and must carry out the part world of other group task of spending the night, then

$r_i=\frac{{\mathrm{\&xi;}}_i+\mathrm{\&Delta;}{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},$ i＝1，...，H

Wherein, Δ ξ _iRepresent the i group world fate variable quantity that spends the night, if reduce, then this is worth for negative; This value must knownly maybe can be asked, otherwise can't carry out equilibrium;

D. some group is spent the night, and task intersects but inequality, and concrete computation process is following:

D.1, for all two groups of i that intersect, j supposes ξ ₀Represent two groups of worlds that can fly task fate that spends the night,

A) if $\frac{{\mathrm{\&xi;}}_i/{\mathrm{\&eta;}}_i}{({\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_j}\&le;\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j},$ Then $\begin{array}{c}{r}_i=\frac{{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},r_j=\frac{{\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0}{{\mathrm{\&eta;}}_j}\\ {\mathrm{\&xi;}}_i={\mathrm{\&xi;}}_i,{\mathrm{\&xi;}}_j={\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0\end{array}$

B) if $\frac{({\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_i}{{\mathrm{\&xi;}}_j/{\mathrm{\&eta;}}_j}\&le;\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j}<\frac{{\mathrm{\&xi;}}_i/{\mathrm{\&eta;}}_i}{({\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_j},$ Then $\begin{array}{c}r=\frac{{\mathrm{\&xi;}}_1+{\mathrm{\&xi;}}_2-{\mathrm{\&xi;}}_0}{{\mathrm{\&lambda;}}_i{\mathrm{\&eta;}}_i+{\mathrm{\&lambda;}}_j{\mathrm{\&eta;}}_j}\\ r_i={\mathrm{\&lambda;}}_{i}r,r_j={\mathrm{\&lambda;}}_{j}r\end{array}$

C) if $\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j}<\frac{({\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_i}{{\mathrm{\&xi;}}_j/{\mathrm{\&eta;}}_j},$ Then $\begin{array}{c}{r}_i=\frac{{\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0}{{\mathrm{\&eta;}}_i},r_j=\frac{{\mathrm{\&xi;}}_j}{{\mathrm{\&eta;}}_j}\\ {\mathrm{\&xi;}}_i={\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0,{\mathrm{\&xi;}}_j={\mathrm{\&xi;}}_j\end{array}$

D.2, repeating step d.1,, do not have crossing two groups up to finding;

E. each group task of spending the night is non-intersect, and each group personnel can not carry out the task of other group, then

$r_i=\frac{{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},$ i＝1，...，H。

In step (3), setting up constraint matrix may further comprise the steps:

(3) a. is provided with the N value;

(3) b. calculates separate

of current relaxation problem

(3) c. selects an employee i according to current lax separating

, only keeps j LoW and the empty LoW of i;

If a) do not have the employee to be locked and N＞0, N=N-1 changes step (3) b. over to;

B) as if N=0, $\mathrm{\&epsiv;}=\mathrm{Max}\{{\stackrel{\&OverBar;}{x}}_{\mathrm{Ij}}:{\stackrel{\&OverBar;}{x}}_{\mathrm{Ij}}<1\},$ Change step (3) d. over to;

(3) d. is if ε＜MINEPS then changes step (3) e.; Otherwise, change step (3) a. over to;

(3) e. calculates the integer solution of current problem

(3) f. separates according to current, selects an employee i, and the invalid LoW of i is made as effectively again, changes step (3) e. over to; If there is not such employee, change step (5) over to.

Wherein, Step (3) b. calculates separating

and adopting following steps of current relaxation problem; Remember that current primal problem is (P), its dual problem is (D):

(3) b.1, if no initial solution, be that each employee selects the LoW that only comprises the LoW of a task and do not comprise any task, add whole slack variables and surplus variable formation initial problem, find the solution this initial problem, obtain initial feasible solution.

(3) b.2, if no initial dual solution is established initial dual solution π=0;

(3) b.3, with having the row of smallest reduction cost and the row in the initial solution, construct initial subproblem (SP);

(3) b.4, find the solution current subproblem (SP) with simplicial method, acquisition optimum solution and dual solution ρ;

(3) b.5, if ρ is feasible with respect to (D); Then

and ρ are respectively original and dual optimal solutions; Otherwise, change over to step (3) b.6;

(3) b.6, make π '=θ π+(1-θ) ρ, wherein

(3) b.7, the whole nonbasic variables in the deletion (SP);

(3) b.8, select the preceding n minimum with respect to the reduction cost of π ' ₁Individual variable joins (SP);

(3) b.9, all with respect to the reduction cost of ρ less than 0 variable in, select n ₂The minimum variable of individual reduction cost with respect to π ' joins (SP), change over to step (3) b.4.

In step (3) c.; Select an employee i according to current lax separating

; If exist

only to keep j LoW and the empty LoW of i; All the other are labeled as invalid; Note is made potential (i)=j, and is called employee i and has locked j LoW.

In step (3) e., establish the LoW that current integer solution is selected for employee i, note is made select (i), if select (i) ≠ potential (i), and do not visited by this step, then can recover it and be labeled as invalid LoW.

As shown in Figure 5, a kind of system that realizes above-mentioned unit scheduling method comprises:

Data input module is used to obtain the raw data that satisfies customer requirements and this raw data is resolved, changes and be organized into the basic data of the form of meeting the requirements;

The model generation module; According to unit arrange an order according to class and grade the rule basic data is retrained control; Generate all the task combinations that to arrange for each crew; Be each crew and generate corresponding task string LoW, again task string LoW is estimated correction, all the task string LoW after proofreading and correct are converted into the task row and set up constraint matrix;

Find the solution module, being used to adopt the extensive PROBLEM DECOMPOSITION that integer programming problem is formed is that the algorithm of subproblem is found the solution constraint matrix, and output result of calculation;

Control module, the operation that is used for control data load module, model generation module and finds the solution module;

Basic module, the generation module that is used to supply a model, required parameter, solver, data layout and data structure when finding the solution the module operation;

Control module is respectively with data input module, model generation module and find the solution module and be connected; Basic module respectively with the model generation module, find the solution module and be connected; The operation of each module of control module control, data input module are obtained the raw data that satisfies customer requirements and the basic data of the form of meeting the requirements is resolved, changes and be organized into to this raw data, after the model generation module obtains this basic data; According to unit arrange an order according to class and grade the rule basic data is retrained control; For each crew generates all the task combinations that possibly arrange, promptly generate corresponding task string LoW, again task string LoW is estimated correction; All task string LoW after proofreading and correct are converted into the task row and set up constraint matrix, find the solution and find the solution and result of calculation is exported after module obtains constraint matrix; Data input module with find the solution module run time call basic module.

In the present embodiment; Data input module is included as the Subscriber Interface Module SIM and the DBM that raw data is provided that the user provides the window with different choice; The user selects through Subscriber Interface Module SIM, and control module is called DBM so that the raw data that satisfies customer requirements to be provided according to user's requirement.

Subscriber Interface Module SIM provides different selections for the user; Can arrange an order according to class and grade according to various conditions and requirement; Provide the different results that arranges an order according to class and grade, the window of the selection that provides comprises following menu: the time period of arranging an order according to class and grade, the rank of arranging an order according to class and grade, the type of arranging an order according to class and grade, collapsing rule etc., the user can be provided with this.

The model generation module comprises LoW generation module, rule inspection module, target factor computing module and model conversion module; The LoW generation module is connected with rule inspection module, target factor computing module and basic module respectively; Rule inspection module also is connected with basic module; The target factor computing module also is connected with basic module, the model conversion module also respectively with find the solution module, basic module is connected; After the LoW generation module obtains this basic data; The unit that rule-based inspection module provides is arranged an order according to class and grade rule to the constraint of basic data; Generate all the task combinations that to arrange for each crew; Be each crew and generate corresponding task string LoW, the target factor computing module is estimated correction to task string LoW, and all the task string LoW after the model conversion module will be proofreaied and correct are converted into the task row and set up constraint matrix.

Basic module comprises that being used to each module the systematic parameter module of operational factor is provided, being used for the original data conversion of database is the basic data management module of the basic data of the form that meets the requirements, the virtual data layer module that is used to raw data and the master tool library module of based process is provided, is used to that the virtual solver module of solver interface is provided and is used to provide uniform data format and data structure; The systematic parameter module is with basic data management module, target factor computing module, model conversion module and find the solution module and be connected; The basic data management module also is connected with master tool library module, virtual data layer module respectively; The master tool library module also is connected with virtual data layer module, basic data management module, virtual solver module with find the solution module and be connected.Find the solution module and in solution procedure, call relevant solver.The parameter that the systematic parameter module provides comprises system's initial parameter and user's input parameter, and checking module, target factor computing module, model conversion module and find the solution module for rule provides each module required parameter.

Interpretational criteria is set and is imported into related coefficient by the systematic parameter module, the target factor computing module is estimated (marking) to the LoW that generates, and evaluation result is the reference that the LoW that generates is proofreaied and correct and optimizes.

Virtual data layer module provides the abstraction interface of concrete data, through the extraction to the reading of data, denoising and useful information, for each algorithm is found the solution module and systematic parameter provides uniform data format and data structure.Make up the data that need through modes such as view, Query Results.Virtual solver module provides the virtual interface of mathematical programming solver; For systematic parameter and derivation algorithm provide unified standard interface; Be transferred to concrete mathematical programming solver, the interface that makes the upper strata call need not to understand structure, the method for concrete solver, makes its transparence.The master tool library module provides the data processing class on basis, like Time Calculation, format conversion etc., data management module can effectively be managed on master tool to virtual data.

System also comprises the peripheral module that is used to write down and feed back total system ruuning situation; Peripheral module is connected with control module, LoW generation module and basic data management module respectively; Peripheral module is to the record and the feedback of total system ruuning situation; Operational process to total system is monitored, and through the ruuning situation of journal file register system, can provide form to supply the user to inquire about and monitor.

Below be the spend the night concrete calculation procedure of deviate (being the poor of element ideal value and actual value) of the world:

With southern aviation is example, and personnel have ranks such as DA, HA, FAT, AT, SG, and wherein, part DA and HA can part-time SG, are designated as DA/SG, HA/SG; Part SG can part-time AT, is designated as SG/AT.DA, HA, FAT, AT are the personnel of main cabin portion, and DA/SG, HA/SG, SG, SG/AT are the personnel of security department.These philtrums have some of the staff need bear more ground task, and it is less to let fate fly away, and this just need let fly away under the fate identical, distribute the more multinational border fate that spends the night, and this part person is the set meal personnel.

1. according to business department's requirement, DA, HA, FAT, AT, DA/SG, HA/SG, set meal SG are carried out equilibrium with set meal SG/AT, non-set meal SG is balanced with SG/AT.

All can be flown the spend the night personnel of flight of the world divides into groups as follows:

1) can fly the set meal personnel of the international long-range flight that spends the night

2) can only fly the spend the night set meal personnel of flight of Southeast Asia

3) can fly non-set meal DA, HA, FAT, AT and DA/SG, the HA/SG of the international long-range flight that spends the night

4) can only fly Southeast Asia spend the night non-set meal DA, HA, FAT, AT and DA/SG, the HA/SG of flight

5) can fly the non-set meal SG and the SG/AT of the international long-range flight that spends the night

6) can only fly Southeast Asia the spend the night non-set meal SG and the SG/AT of flight

2. statistics the world that interval DA, HA, FAT, full-time AT post the produce fate y that spends the night that arranges an order according to class and grade ₁To arranging an order according to class and grade between the interval beginning, fate y spends the night in the world of having arranged in balanced interval beginning for statistics DA, HA, FAT, AT, DA/SG, HA/SG rank personnel ₂Statistics the world that interval all SG posies the produce fate y that spends the night that arranges an order according to class and grade ₃To arranging an order according to class and grade between the interval beginning, fate y spends the night in the world of having arranged in balanced interval beginning for statistics SG, SG/AT, DA/SG, HA/SG rank personnel ₄

3. add up the fate of letting fly away that all can the row personnel belong to each fair group: statistics DA, HA, FAT, AT, DA/SG, HA/SG rank personnel are at the balanced interval fate of effectively letting fly away; Everyone has 6 groups of values of letting fly away; I value representation belongs to the fate of letting fly away of i group; If never belong to this group, then reorganization value is 0;

4. calculate target ratio: require the set meal personnel be non-set meal personnel a doubly, the people that long-range qualification is arranged be have only the medium or short range qualification the people b doubly.DA, HA, FAT, AT, DA/SG, HA/SG rank personnel's target ratio is calculated as follows:

$r=\frac{\mathrm{<figure-callout\; id="1"\; label="y\; ax"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">y</figure-callout>}}{{\mathrm{ax}}_{}}$

r ₁＝ar；r ₂＝abr；r ₃＝r； $r_4=\frac{1}{b}r$

$r_6=\frac{{\mathrm{<figure-callout\; id="1"\; label="y"\; filenames="HDA0000118048800000031.png"\; state="\{\{state\}\}">y</figure-callout>}}_1+y_2+y_3+y_4-x_7{r}_1-x_8{r}_2-x_9{r}_3-x_{10}{r}_4}{{\mathrm{bx}}_5+x_6}$

r ₅＝br ₆

Wherein, y=y ₁+ y ₂

x _i(i=1,2,3,4,5,6): DA, HA, FAT, AT, DA/SG, HA/SG rank personnel i group let fly away fate and;

x ₇: DA/SG, HA/SG, SG, SG/AT belong to the 1st group let the fate summation fly away

x ₈: DA/SG, HA/SG, SG, SG/AT belong to the 2nd group let the fate summation fly away

x ₉: DA/SG, HA/SG, SG, SG/AT belong to the 3rd group let the fate summation fly away

x ₁₀: DA/SG, HA/SG, SG, SG/AT belong to the 4th group let the fate summation fly away

r _i: i group people's target ratio;

5. calculate ideal value and deviate: for certain scheme of arranging an order according to class and grade, adding up the world that this steward will carry out in the whole balanced interval fate that spends the night is z, and this scheme world deviation of spending the night is:

$z-{\mathrm{\&Sigma;}}_{i=1}^6{X}_i{r}_i$

Wherein, X _iThe fate of letting fly away of representing this people i group.

The present invention arranges month about more than 9500 flight tasks for about 2000 stewards and security official; System can draw result of calculation in 5 hours; Coverage rate (number that number/flight of unit Workforce Management arrangement of the present invention needs) is greater than 90%; Fate fairness (the every month fate that spends the night that fairness index=system arranges for the steward/moon target fate) is spent the night near 87% in the world, and implementation rate (task of task sum/automatic scheduling system arrangement of carrying out the automatic scheduling system arrangement in the actual motion is total) is near 53%.

Embodiment of the present invention is not limited thereto; According to foregoing of the present invention; Ordinary skill knowledge and customary means according to this area; Do not breaking away under the above-mentioned basic fundamental thought prerequisite, the present invention can also make modification, replacement or the change of other various ways, all drops within the rights protection scope of the present invention.

Claims (10)

1. unit scheduling method is characterized in that may further comprise the steps:

(1) obtains the raw data of The profile from user interface and database, raw data is resolved, changes, put in order, to obtain whole basic datas of the form of meeting the requirements;

(2) based on the rule of arranging an order according to class and grade basic data is retrained control,, be each crew and generate corresponding task string LoW, again task string LoW is estimated correction for each crew generates all the task combinations that possibly arrange;

(3) the whole task string LoW after will proofreading and correct are converted into the task row and set up constraint matrix;

(4) adopting the extensive PROBLEM DECOMPOSITION that integer programming problem is formed is the algorithm of subproblem, and constraint matrix is found the solution;

(5) output result of calculation.

2. unit scheduling method according to claim 1; It is characterized in that: in said step (2); Each task string LoW is estimated the mode of proofreading and correct employing calculating target factor, and computing formula is:

s＞1

ψ _kThe ideal value of-Di k item element

χ _kThe actual value of-Di k item element

Calculate the target factor formula and be by the following unit integer programming model of arranging an order according to class and grade and draw:

$\left(1\right),\min \mathrm{\&Sigma;}{c}_j\&CenterDot;x_j+M\&CenterDot;\underset{k}{\mathrm{\&Sigma;}}{y}_k$

$\left(2\right),s.t.{\mathrm{\&Sigma;}}_j{a}_{\mathrm{ij}}\&CenterDot;x_j+y_i^F+y_i^M={\mathrm{\&delta;}}_i\mathrm{for}\&ForAll;t_i\&Element;T$

$\left(3\right),{\mathrm{\&Sigma;}}_j{b}_{\mathrm{kj}}\&CenterDot;x_j=1\mathrm{for}\&ForAll;k\&Element;C$

$\left(5\right),\left\{\begin{array}{c}{\mathrm{\&Sigma;}}_j{q}_{\mathrm{ij}}\&CenterDot;x_j+y_i^B\&le;{\mathrm{\&delta;}}_i^B\mathrm{for}\&ForAll;t_i\&Element;T\\ y_i^D\&le;{\mathrm{\&delta;}}_i^D\\ y_i^B+y_i^D-y_i^E-y_i^M=0\end{array}\right.$

$\left(6\right),\left\{\begin{array}{c}{\mathrm{\&Sigma;}}_j{u}_{\mathrm{ij}}\&CenterDot;x_j+y_i^M-s_i^{E1}-s_i^{E2}+{\mathrm{\&sigma;}}_i=0\mathrm{for}\&ForAll;i\&Element;Q\&SubsetEqual;T\\ s_i^{E1}=s_i^{E2}\end{array}\right.$

(7)x _j∈{0，1}，yinteger

T representes all set of tasks to be allocated, and C representes the set that the aircrew constitutes, t _iRepresent i task, c _jThe target factor of representing j bar LoW, M representes the target factor y of surplus variable _kThe expression decision variable, task t _iUnderstaffing's number, a _IjRepresent whether j bar LoW comprises i task,

Male sex understaffing's number of representing i task,

Women understaffing's number of representing i task, δ _iRepresent the staffing of i task, b _KjIf expression j bar LoW belongs to personnel k, then b _KjEqual 1, otherwise be 0; P _kExpression has the set of the task formation of certain attribute, P _k∈ T; p _IjExpression is if j bar LoW comprises task t _i, and t _iBelong to set P _k, and its corresponding personnel have certain specific qualification, then p _IjEqual 1, otherwise be 0; q _IjExpression is if j bar LoW comprises task t _i, and its corresponding personnel accomplish this task, then q according to this rank _IjEqual 1, otherwise be 0;

Expression task t _iOther staffing of fooled front-seat class, Expression task t _iOn the staffing that can demote and carry out.

The expression decision variable, task t _iFooled front-seat class not vacant post number;

The expression decision variable, task t _iThe vacant post number of last degradation; u _IjRepresent that j bar LoW comprises task i, and personnel are the male sex under it; Q representes all or part of set that comprises the task formation of spending the night.σ _iBe illustrated in the walkthrough has been male sex's quantity of i Task Distribution.

indicates slack variables.

3. unit scheduling method according to claim 2 is characterized in that: the computing formula of the ideal value of element is:

${\mathrm{\&psi;}}_k=\underset{j=1}{\overset{H}{\mathrm{\&Sigma;}}}{\mathrm{\&eta;}}_{k,j}{r}_j$

η _{K, j}The fate of letting fly away of representing k people j group

r _jThe target ratio of representing the j group;

Concrete calculation procedure is following:

1. according to qualification, qualifications and record of service factor, personnel's set is divided, personnel distribute the identical equalization task amount of treating when letting fate fly away and equate in same group;

2. statistics is treated the equalization task amount: promptly deduct manual walkthrough personnel in the whole balanced interval equalization task amount of treating of arranging: the statistics row of treating flight arranging an order according to class and grade interval in the equalization task amount of treating of generation; This statistical value is added up according to task rank; The interval beginning of statistical equilibrium is to the interval beginning of arranging an order according to class and grade; The row that can arrange personnel treats the equalization task amount; If abandon treating the equalization task amount because of individual reason, then be designated as this people in row's task, this statistical value is added up operational staff's rank as task rank; Balanced interval all tasks are sued for peace by rank;

3. add up the fate of letting fly away that all can the row personnel belong to each fair group;

4. calculate target ratio;

5. calculate ideal value.

4. unit scheduling method according to claim 3 is characterized in that: come to calculate respectively target ratio according to the relation between each group task, wherein, ξ ₁..., ξ _HWhen representing that respectively the 1st to H group personnel only arrange this rank flight task, fate spends the night in the executable world; η ₁..., η _HThat representes the 1st to H group personnel respectively lets the fate summation fly away; The concrete steps of calculating target ratio are following:

A. world task of spending the night of each group is identical, suppose to require letting fly away under the identical situation of fate, and the world that i the organizes fate weight of spending the night is λ _i, then

$r_i={\mathrm{\&lambda;}}_i\mathrm{\&xi;}/\underset{j=1}{\overset{H}{\mathrm{\&Sigma;}}}{\mathrm{\&lambda;}}_j{\mathrm{\&eta;}}_j$

Wherein, r _iThe target ratio of representing the i group, ξ=ξ ₁=...=ξ _H

B. world task of spending the night of each group is non-intersect, but some group personnel can carry out the world of some other group task of spending the night, otherwise, then can not, concrete steps are following:

B.1, find all satisfactory group right: the inside comprises two groups, and wherein lineup person can arrange the task of another group, supposes to have h group right, k the group to the lining, k ₁Group personnel can arrange k ₂Group task is being let fly away under the identical situation of fate, and two groups the world fate weight of spending the night does

B.2, whether all groups are to satisfying

k=1 in inspection; ..., h;

A) if satisfy;

k=1 then; ...; H, change over to step b.3;

B) if do not satisfy, find arbitrary ungratified group to hypothesis l,

$r=\frac{{\mathrm{\&xi;}}_{l_1}+{\mathrm{\&xi;}}_{l_2}}{{\mathrm{\&lambda;}}_{l_1}{\mathrm{\&eta;}}_{l_1}+{\mathrm{\&lambda;}}_{l_2}{\mathrm{\&eta;}}_{l_2}}$

$r_{l_1}={\mathrm{\&lambda;}}_{l_1}r,$ $r_{l_2}={\mathrm{\&lambda;}}_{l_2}r$

${\mathrm{\&xi;}}_{l_1}=r_{l_1}{\mathrm{\&eta;}}_{l_1},$ ${\mathrm{\&xi;}}_{l_2}=r_{l_2}{\mathrm{\&eta;}}_{l_2}$

Repeat this step;

B.3, calculate end;

C. world task of spending the night of each group is non-intersect, but some group personnel can and must carry out the part world of other group task of spending the night, then

$r_i=\frac{{\mathrm{\&xi;}}_i+\mathrm{\&Delta;}{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},$ i＝1，...，H

Wherein, Δ ξ _iRepresent the i group world fate variable quantity that spends the night;

D. some group is spent the night, and task intersects but inequality, and concrete computation process is following:

D.1, for all two groups of i that intersect, j supposes ξ ₀Represent two groups of worlds that can fly task fate that spends the night,

A) if $\frac{{\mathrm{\&xi;}}_i/{\mathrm{\&eta;}}_i}{({\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_j}\&le;\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j},$ Then $\begin{array}{c}{r}_i=\frac{{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},r_j=\frac{{\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0}{{\mathrm{\&eta;}}_j}\\ {\mathrm{\&xi;}}_i={\mathrm{\&xi;}}_i,{\mathrm{\&xi;}}_j={\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0\end{array}$

B) if $\frac{({\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_i}{{\mathrm{\&xi;}}_j/{\mathrm{\&eta;}}_j}\&le;\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j}<\frac{{\mathrm{\&xi;}}_i/{\mathrm{\&eta;}}_i}{({\mathrm{\&xi;}}_j-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_j},$ Then $\begin{array}{c}r=\frac{{\mathrm{\&xi;}}_1+{\mathrm{\&xi;}}_2-{\mathrm{\&xi;}}_0}{{\mathrm{\&lambda;}}_i{\mathrm{\&eta;}}_i+{\mathrm{\&lambda;}}_j{\mathrm{\&eta;}}_j}\\ r_i={\mathrm{\&lambda;}}_{i}r,r_j={\mathrm{\&lambda;}}_{j}r\end{array}$

C) if $\frac{{\mathrm{\&lambda;}}_i}{{\mathrm{\&lambda;}}_j}<\frac{({\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0)/{\mathrm{\&eta;}}_i}{{\mathrm{\&xi;}}_j/{\mathrm{\&eta;}}_j},$ Then $\begin{array}{c}{r}_i=\frac{{\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0}{{\mathrm{\&eta;}}_i},r_j=\frac{{\mathrm{\&xi;}}_j}{{\mathrm{\&eta;}}_j}\\ {\mathrm{\&xi;}}_i={\mathrm{\&xi;}}_i-{\mathrm{\&xi;}}_0,{\mathrm{\&xi;}}_j={\mathrm{\&xi;}}_j\end{array}$

D.2, repeating step d.1,, up to there not being crossing two groups;

E. each group task of spending the night is non-intersect, and each group personnel can not carry out the task of other group, then

$r_i=\frac{{\mathrm{\&xi;}}_i}{{\mathrm{\&eta;}}_i},$ i＝1，...，H。

5. according to each described unit scheduling method of claim 1～4, it is characterized in that: in said step (3), set up constraint matrix and may further comprise the steps:

(3) a. is provided with the N value;

(3) b. calculates separate

of current relaxation problem

(3) c. selects an employee i according to current lax separating

, only keeps j LoW and the empty LoW of i;

If a) do not have the employee to be locked and N＞0, N=N-1 changes step (3) b. over to;

B) as if N=0, $\mathrm{\&epsiv;}=\mathrm{Max}\{{\stackrel{\&OverBar;}{x}}_{\mathrm{Ij}}:{\stackrel{\&OverBar;}{x}}_{\mathrm{Ij}}<1\},$ Change step (3) d. over to;

(3) d. is if ε＜MINEPS then changes step (3) e.; Otherwise, change step (3) a. over to;

(3) e. calculates the integer solution

of current problem

(3) f. separates according to current, selects an employee i, and the invalid LoW of i is made as effectively again, changes step (3) e. over to; If there is not such employee, change step (5) over to.

6. unit scheduling method according to claim 5; It is characterized in that: described step (3) b. calculates separating

and adopting following steps of current relaxation problem; Remember that current primal problem is (P), its dual problem is (D):

(3) b.1, if no initial solution, be that each employee selects the LoW that only comprises the LoW of a task and do not comprise any task, add whole slack variables and surplus variable formation initial problem, find the solution this initial problem, obtain initial feasible solution;

(3) b.2, if no initial dual solution is established initial dual solution π=0;

(3) b.3, with having the row of smallest reduction cost and the row in the initial solution, construct initial subproblem (SP);

(3) b.4, find the solution current subproblem (SP) with simplicial method, acquisition optimum solution and dual solution ρ;

(3) b.5, if ρ is feasible with respect to (D); Then

and ρ are respectively original and dual optimal solutions; Otherwise, change over to step (3) b.6;

(3) b.6, make π '=θ π+(1-θ) ρ, wherein

(3) b.7, the whole nonbasic variables in the deletion (SP);

(3) b.8, select the preceding n minimum with respect to the reduction cost of π ' ₁Individual variable joins (SP);

(3) b.9, all with respect to the reduction cost of ρ less than 0 variable in, select n ₂The minimum variable of individual reduction cost with respect to π ' joins (SP), change over to step (3) b.4.

7. system that realizes the said unit scheduling method of claim 1 is characterized in that comprising:

Data input module is used to obtain the raw data that satisfies customer requirements and this raw data is resolved, changes and be organized into the basic data of the form of meeting the requirements;

The model generation module; According to unit arrange an order according to class and grade the rule basic data is retrained control; Generate all the task combinations that to arrange for each crew; Be each crew and generate corresponding task string LoW, again task string LoW is estimated correction, all the task string LoW after proofreading and correct are converted into the task row and set up constraint matrix;

Find the solution module, being used to adopt the extensive PROBLEM DECOMPOSITION that integer programming problem is formed is that the algorithm of subproblem is found the solution constraint matrix, and output result of calculation;

Control module, the operation that is used for control data load module, model generation module and finds the solution module;

Basic module, the generation module that is used to supply a model, required parameter, solver, data layout and data structure when finding the solution the module operation;

Said control module is respectively with said data input module, model generation module and find the solution module and be connected; Said basic module respectively with the model generation module, find the solution module and be connected; The operation of each module of said control module control, said data input module are obtained the raw data that satisfies customer requirements and the basic data of the form of meeting the requirements is resolved, changes and be organized into to this raw data, after said model generation module obtains this basic data; According to unit arrange an order according to class and grade the rule basic data is retrained control; For each crew generates all the task combinations that possibly arrange, promptly generate corresponding task string LoW, again task string LoW is estimated correction; All task string LoW after proofreading and correct are converted into the task row and set up constraint matrix, and said finding the solution found the solution and result of calculation exported after module obtains constraint matrix; Said data input module with find the solution the said basic module of module run time call.

8. system according to claim 7; It is characterized in that: said data input module is included as the Subscriber Interface Module SIM and the DBM that raw data is provided that the user provides the window with different choice; The user selects through Subscriber Interface Module SIM, and said control module is called DBM so that the raw data that satisfies customer requirements to be provided according to user's requirement.

9. system according to claim 8; It is characterized in that: said model generation module comprises LoW generation module, rule inspection module, target factor computing module and model conversion module; Said LoW generation module is connected with said rule inspection module, target factor computing module and basic module respectively; Said rule inspection module also is connected with basic module, and the target factor computing module also is connected with basic module, the model conversion module also respectively with find the solution module, basic module is connected; After said LoW generation module obtains this basic data; The unit that provides according to the rule inspection module constraint of rule of arranging an order according to class and grade to basic data; Generate all the task combinations that to arrange for each crew; Be each crew and generate corresponding task string LoW, said target factor computing module is estimated correction to task string LoW, and all the task string LoW after said model conversion module will be proofreaied and correct are converted into the task row and set up constraint matrix.

10. system according to claim 9; It is characterized in that: said basic module comprise be used to each module the systematic parameter module of operational factor be provided, be used for the original data conversion of database be the basic data of the form that meets the requirements the basic data management module, be used to raw data the master tool library module of based process be provided, be used to provide the virtual solver module of solver interface, the peripheral module that is used to that the virtual data layer module of uniform data format and data structure is provided and is used to write down and feed back total system ruuning situation; Said systematic parameter module is with basic data management module, target factor computing module, model conversion module and find the solution module and be connected; Said basic data management module also is connected with master tool library module, virtual data layer module respectively; Said master tool library module also is connected with virtual data layer module, basic data management module, said virtual solver module with find the solution module and be connected; Said peripheral module is connected with control module, LoW generation module and basic data management module respectively.

Images