CN104202188A - Method for carrying out AFDX network path optimization by genetic algorithm - Google Patents

Abstract

The invention discloses a method for carrying out AFDX network path optimization by a genetic algorithm. The method is virtual link path optimization carried out on an AFDX network configured with a VL path. The method disclosed by the invention comprises the following steps of: establishing a connection matrix and a virtual link path group of a switch at first; and then carrying out chromosome coding on a virtual link path; finally carrying out genetic operations of crossing, variation and selection on a chromosome group, obtaining the optimal chromosome in case of meeting an end condition, and extracting out a valid virtual link path, wherein the path is used as the optimized virtual link path. According to the method disclosed by the invention, path optimization for the virtual link VL in the configured AFDX network is solved, and real-time path optimization for the virtual link VL in the configured AFDX network is carried out by applying the genetic algorithm, thus improving the message transmission real-time performance of the AFDX network.

Description

A kind of method that adopts genetic algorithm to carry out the optimization of AFDX network path

Technical field

The invention belongs to avionics system communication network optimize design field, more particularly, refer to a kind of method that adopts genetic algorithm to carry out the optimization of AFDX network path.

Background technology

Avionic full-duplex switched-type Ethernet (Avionics Full Duplex Switched Ethernet, AFDX) is to transform through applicability on the basis of industrial standard Ethernet, and is applicable to the interconnected network technology of avionics system.AFDX interconnection technique has obtained successful Application in passenger vehicle A380 and Boeing 787 aloft.

AFDX network carries out message data flow transmission by virtual link (Virtual Link, VL), and AFDX network provides definite bandwidth and definite route for VL.The path effects of VL the transmission delay of virtual link bearer messages, also affects the flow equalization of AFDX network.

The 39th the 6th phase of volume of " BJ University of Aeronautics & Astronautics's journal " June in 2013 discloses " time is triggered design and the real time analysis of AFDX network ", author Liu Cheng.Fig. 8 of the document discloses a comparatively traditional AFDX network topology structure (being cited as shown in Fig. 1), have 8 switches and formed network backbone, on each switch, connect 8 end systems, one has 64 end systems, and link bandwidth is configured to 10Mbit/s.

Summary of the invention

In order to reduce in avionic full-duplex switched-type Ethernet AFDX, the transmission delay of virtual link bearer messages, the present invention proposes a kind of method that adopts genetic algorithm to carry out the optimization of AFDX network path.The method has solved the path optimization of virtual link VL in the AFDX network having configured, be that application genetic algorithm is carried out real-time route optimization to the VL path of AFDX network, thereby improved the transmission of messages real-time (being that the propagation delay time is short) of AFDX network.

The present invention is a kind of method that adopts genetic algorithm to carry out the optimization of AFDX network path, and described AFDX network refers to the AFDX network that disposes VL path; It is characterized in that including the following step:

Step 1: build switch connection matrix;

Carry out the connection initialization between all switches to disposing the AFDX network in VL path, obtain n × n switch connection matrix n is the number of the switch in AFDX network;

Step 2: build virtual link path population;

Virtual link path population scale U={u is set ₁, u ₂..., u _x; Switch connection matrix SC to step 1 chooses and meets U={u according to path exclusion condition ₁, u ₂..., u _xsource-object-switch path $R-u_x=\left(S_a^A\stackrel{u_x}{\→}{S}_b^B\right);$

Step 3: build chromosome;

Source-object-switch path that step 2 is obtained in the identification number of switch as gene, the encoded chromosome population CH={h that obtains ₁, h ₂..., h _x;

Step 4: calculate chromosomal fitness function value;

Described fitness function Fitness=AVL _max+ time _postpone, wherein, maximum link load factor ${\mathrm{AVL}}_{\max }=\underset{n}{\max }\frac{\underset{\mathrm{VL}\∈C_{\mathrm{nn}}}{\mathrm{\Σ}}\frac{r_{\max }}{t_{\mathrm{BAG}}}}{100\×{10}^6}\×100\%,$ Message transmission delay time rate

Step 5: genetic manipulation;

Current chromosome population is designated as CH _i, previous generation chromosome population is designated as CH _i-1, chromosome population of future generation is designated as CH _i+1;

Step 501, chromosomal the first cross modal:

The mode that uses single-point to intersect, crosspoint is selected in public branch exchange place, if there are multiple public branch exchanges, select at random one as crosspoint, then the part after exchange pairing crosspoint, two new chromosomes that generate, and described new chromosome is added in the population of virtual link of future generation path; If duplicate gene in new chromosome after intersecting, delete the switch of any one repetition, and the new chromosome of having deleted after repeated exchanged machine is added in the population of virtual link of future generation path.

Step 502, chromosomal the second cross modal:

The mode that uses single-point to intersect, if there is not public branch exchange, in the middle of selecting, chromosome intersects, the then part after exchange pairing crosspoint, two new chromosomes that generate, and described new chromosome is added in the population of virtual link of future generation path; If duplicate gene in new chromosome after intersecting, delete the switch of any one repetition, and add in the population of virtual link of future generation path and add in the population of virtual link of future generation path having deleted new chromosome after repeated exchanged machine.

Step 503, chromosomal the first variant form:

The mode that uses single-point variation, change point is selected in except source switch with object switch any switch place, select with described switch exist link-attached another switch of physics as variation result, obtain new chromosome, and described new chromosome added in the population of virtual link of future generation path; If duplicate gene in new chromosome after variation, delete the switch of any one repetition, and the new chromosome of having deleted after repeated exchanged machine is added in the population of virtual link of future generation path.

Step 504, chromosomal the second variant form:

The mode that uses single-point variation, change point is selected in except source switch with object switch any switch place, directly described switch is deleted, obtain new chromosome, and described new chromosome added in the population of virtual link of future generation path.

Step 505, chromosomal selection mode:

Chromosomal selection mode is by the chromosome population CH when former generation _i={ h ₁, h ₂..., h _xand by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponedescribed in calculating in each chromosomal fitness function value and choose x minimum chromosome as population CH of future generation _i+1.

Step 6: judge whether to meet hereditary end condition;

Judge the fitness function value Fh of described minimum _minimumafter continuous 3 generation genetic manipulations, described Fh _minimumwhile variation, stop genetic manipulation; If do not meet, to CH={h ₁, h ₂..., h _xin every item chromosome carry out genetic manipulation.

The advantage that the present invention adopts genetic algorithm to carry out the optimization of AFDX network path is:

1. genetic algorithm is incorporated in the AFDX network that disposes VL path, is more conducive to find in real time the optimal path of virtual link.Can improve the transmission of messages real-time (being that the propagation delay time is short) of AFDX network.

2. apply chromosome in genetic algorithm carries out switch identification number coding to virtual link path, this is conducive to carry out intersection, the mutation operation of genetic algorithm.

3. adopt binary system to carry out encode to switch connection matrix, effectively provide basis for estimation for virtual link path code.

Brief description of the drawings

Fig. 1 is the AFDX schematic network structure having built.

Fig. 2 is AFDX network topology structure schematic diagram of the present invention.

Fig. 3 is that the present invention adopts genetic algorithm to carry out the flow chart of AFDX network path optimization.

Fig. 4 is the AFDX network topology structure schematic diagram of embodiment 1.

Embodiment

Below in conjunction with drawings and Examples, the present invention is described in further detail.

For the AFDX network topology structure shown in Fig. 1, AFDX network packet contains n switch, adopts set form to be expressed as SWITH={S ₁, S ₂..., S _a..., S _b..., S _n; Described SWITH={S ₁, S ₂..., S _a..., S _b..., S _nmiddle S ₁represent first switch, S ₂represent second switch, S _arepresent a switch, S _brepresent b switch, S _nrepresenting n switch, is also last switch; For general knowledge explanation, S _aalso referred to as one of them any switch, S _balso referred to as another any switch wherein, S _nalso referred to as any one switch, a, b, the identification number that n is switch, n is also the sum of switch in AFDX network.In Fig. 1, n=8, has S ₁, S ₂, S ₃, S ₄, S ₅, S ₆, S ₇, S ₈.

For the AFDX network topology structure shown in Fig. 1, AFDX network packet contains m end system, adopts set form to be expressed as ES={E ₁, E ₂..., E _c..., E _d..., E _m; Described ES={E ₁, E ₂..., E _c..., E _d..., E _mmiddle E ₁represent first end system, E ₂represent second end system, E _crepresent c end system, E _drepresent d end system, E _mrepresenting m end system, is also last end system; For general knowledge explanation, E _calso referred to as one of them arbitrary end system, E _dalso referred to as another arbitrary end system wherein, E _malso referred to as any one end system, c, d, the identification number that m is end system.

For the AFDX network topology structure shown in Fig. 1, AFDX network packet contains q bar virtual link, adopts set form to be expressed as VL={L ₁, L ₂..., L _e..., L _f..., L _q; Described VL={L ₁, L ₂..., L _e..., L _f..., L _qmiddle L ₁represent Article 1 virtual link, L ₂represent Article 2 virtual link, L _erepresent e article of virtual link, L _frepresent f article of virtual link, L _qrepresenting q article of virtual link, is also the last item virtual link; For general knowledge explanation, L _ealso referred to as any virtual link wherein, L _falso referred to as another any virtual link wherein, L _qalso referred to as any virtual link, e, f, the identification number that q is virtual link.

The employing genetic algorithm that the present invention proposes is carried out the method for path optimization, and described AFDX network refers to the AFDX network that disposes VL path, and concrete path optimization includes step:

Step 1: build switch connection matrix

Carry out the physical connection initialization between all switches to disposing the AFDX network in VL path, obtain the n × n switch connection matrix that adopts matrix form to express

C _nnrepresent n switch S _nwith n switch S _nconnection, represent the connection between switch self; In like manner, C ₁₁represent the 1st switch S ₁with the 1st switch S ₁connection, C ₂₂represent the 2nd switch S ₂with the 2nd switch S ₂connection, C _aarepresent a switch S _awith a switch S _aconnection, C _bbrepresent b switch S _bwith b switch S _bconnection, C _abrepresent a switch S _awith b switch S _bconnection.

C ₁₂represent the 1st switch S ₁with the 2nd switch S ₂connection, C _1nrepresent the 1st switch S ₁with n switch S _nconnection;

C ₂₁represent the 2nd switch S ₂with the 1st switch S ₁connection, C _2nrepresent the 2nd switch S ₂with n switch S _nconnection;

C _n1represent n switch S _nwith the 1st switch S ₁connection, C _n2represent n switch S _nwith the 2nd switch S ₂connection.

In the present invention, the assignment condition of switch connection matrix SC has three kinds, the first is that the connection between switch self is designated as 0, and the second is between switch, not have the connection of physical link to be designated as 0, and the third has the connection of physical link to be designated as 1 between switch.Switch connection matrix as shown in Figure 1

$\mathrm{SC}=\left[\begin{array}{cccccccc}{C}_{11}& C_{12}& C_{13}& C_{14}& C_{15}& C_{16}& C_{17}& C_{18}\\ C_{21}& C_{22}& C_{23}& C_{24}& C_{25}& C_{26}& C_{27}& C_{28}\\ C_{31}& C_{32}& C_{33}& C_{34}& C_{35}& C_{36}& C_{37}& C_{38}\\ C_{41}& {C_{42}}_{}& C_{43}& C_{44}& C_{45}& C_{46}& C_{47}& C_{48}\\ C_{51}& C_{52}& C_{53}& C_{54}& C_{55}& C_{56}& C_{57}& C_{58}\\ C_{61}& C_{62}& C_{63}& C_{64}& C_{65}& C_{66}& C_{67}& C_{68}\\ C_{71}& C_{72}& C_{73}& C_{74}& C_{75}& C_{76}& C_{77}& C_{78}\\ C_{81}& C_{82}& C_{83}& C_{84}& C_{85}& C_{86}& C_{87}& {C_{88}}_{}\end{array}\right]=\left[\begin{array}{cccccccc}0& 1& 1& 0& 0& 0& 0& 1\\ 1& 0& 0& 1& 0& 0& 0& 1\\ 1& 0& 0& 1& 1& 0& 1& 1\\ 0& 1& 1& 0& 0& 1& 1& 1\\ 0& 0& 1& 0& 0& 1& 1& 0\\ 0& 0& 0& 1& 1& 0& 1& 0\\ 0& 0& 1& 1& 1& 1& 0& 0\\ 1& 1& 1& 1& 0& 0& 0& 0\end{array}\right].$

In the present invention, application binary " 0,1 " encode represents the physical connection of switch, is in order comprehensively to represent the connection state between switch in AFDX network.

Step 2: build virtual link path population

(A) virtual link path population scale U={u is set ₁, u ₂..., u _x; According in AFDX network for the setting of virtual link path population scale U, the number in the virtual link path from source end system A to destination system B.Usually, population number x=3～20.

U ₁represent the mark in the effective virtual link of Article 1 path;

U ₂represent the mark in the effective virtual link of Article 2 path;

U _xrepresent the mark in the effective virtual link of the last item path, for general knowledge explanation, also referred to as any effective virtual link path.

(B) switch matrix to step 1 choose and meet U={u according to path exclusion condition ₁, u ₂..., u _xsource-object-switch path

represent the switch S of carrying source end system A _a;

represent the switch S of carrying destination system B _b;

For example, in Fig. 2: in the time of value population number x=6, source end system A is connected to switch S ₁upper, destination system B is connected to switch S _mupper, there is the effective virtual link of the Article 1 selecting according to path exclusion condition path $R-u_1=\left(S_a^A\stackrel{u_1}{\→}{S}_b^B\right)=S_1-S_3-S_5-S_a-S_b-S_m;$

The effective virtual link of Article 2 path $R-u_2=\left(S_a^A\stackrel{u_2}{\→}{S}_b^B\right)=S_1-S_2-S_4-S_6-S_m;$

The effective virtual link of Article 3 path $R-u_3=\left(S_a^A\stackrel{u_3}{\→}{S}_b^B\right)=S_1-S_8-S_4-S_7-S_6-S_m;$

The effective virtual link of Article 4 path $R-u_4=\left(S_a^A\stackrel{u_4}{\→}{S}_b^B\right)=S_1-S_3-S_5-S_6-S_b-S_m;$

The effective virtual link of Article 5 path $R-u_5=\left(S_a^A\stackrel{u_5}{\→}{S}_b^B\right)=S_1-S_8-S_4-S_6-S_m;$

The effective virtual link of Article 6 path $R-u_6=\left(S_a^A\stackrel{u_6}{\→}{S}_b^B\right)=S_1-S_2-S_4-S_8-S_3-S_5-S_6-S_m.$

(C) in the present invention, described path exclusion condition includes three kinds of situations:

Eliminating situation in the first path refers to that source end system A and destination system B are connected on same switch, do not choose this effective virtual link path, referred to as single switch path.

The second path eliminating situation refer to source end system A to destination system B in the virtual link path of process, have the same switch of repetition, do not choose this effective virtual link path, there is loop referred to as path.

The third path eliminating situation refer to source end system A to destination system B in the virtual link path of process, between switch, there is no physical link, do not choose this effective virtual link path, referred to as without corresponding physical link.

For example, situation is got rid of in the second path, and source end system A is connected to switch S ₁upper, destination system B is connected to switch S _mupper, the path of the switch that passes through is S ₁-S ₃-S ₅-S ₃-S ₇-S ₆-S _m, due at switch S ₃on there is loop situation, get rid of described S ₁-S ₃-S ₅-S ₃-S ₇-S ₆-S _mpath.

For example, situation is got rid of in the third path, and source end system A is connected to switch S ₁upper, destination system B is connected to switch S _mupper, the path of the switch that passes through is S ₁-S ₃-S ₆-S _b-S _m, get rid of described S ₁-S ₃-S ₆-S _b-S _mpath.Because switch S ₃with switch S ₆between there is no physical link, i.e. C in switch matrix SC ₃₆connection code be " 0 ".

Step 3: build chromosome

Source-object-switch path that step 2 is obtained in the identification number of switch as gene, the encoded chromosome population CH={h that obtains ₁, h ₂..., h _x; Then perform step four;

H ₁represent first chromosomal mark;

H ₂represent second chromosomal mark;

H _xrepresent last chromosomal mark, for general knowledge explanation, also referred to as any one chromosome;

$R-u_1=\left(S_a^A\stackrel{u_1}{\→}{S}_b^B\right)=S_1-S_3-S_5-S_a-S_b-S_m$ The encoded chromosome obtaining is h ₁=1-3-5-a-b-m;

$R-u_2=\left(S_a^A\stackrel{u_2}{\→}{S}_b^B\right)=S_1-S_2-S_4-S_6-S_m$ The encoded chromosome obtaining is h ₂=1-2-4-6-m;

$R-u_3=\left(S_a^A\stackrel{u_3}{\→}{S}_b^B\right)=S_1-S_8-S_4-S_7-S_6-S_m$ The encoded chromosome obtaining is h ₃=1-8-4-7-6-m;

$R-u_4=\left(S_a^A\stackrel{u_4}{\→}{S}_b^B\right)=S_1-S_3-S_5-S_6-S_b-S_m$ The encoded chromosome obtaining is h ₄=1-3-5-6-b-m;

$R-u_5=\left(S_a^A\stackrel{u_5}{\→}{S}_b^B\right)=S_1-S_8-S_4-S_6-S_m$ The encoded chromosome obtaining is h ₅=1-8-4-6-m.

$R-u_6=\left(S_a^A\stackrel{u_6}{\→}{S}_b^B\right)=S_1-S_2-S_4-S_8-S_3-S_5-S_6-S_m$ The encoded chromosome obtaining is h ₆=1-2-4-8-3-5-6-m.

In the present invention, the chromosome in application genetic algorithm carries out the identification number coding of switch to virtual link path, and this is conducive to carry out intersection, the mutation operation of genetic algorithm.

Step 4: calculate chromosomal fitness function value

For equilibrium disposed VL path AFDX network flow and reduce propagation delay time of message, the present invention adopts maximum link load factor AVL _maxpropagation delay time rate time with message _postponesum is as fitness function Fitness=AVL _max+ time _postpone; Then perform step five.

In the present invention, ${\mathrm{AVL}}_{\max }=\underset{n}{\max }\frac{\underset{\mathrm{VL}\∈C_{\mathrm{nn}}}{\mathrm{\Σ}}\frac{r_{\max }}{t_{\mathrm{BAG}}}}{100\×{10}^6}\×100\%,$ R _maxrepresent the virtual link VL={L in AFDX network ₁, L ₂..., L _e..., L _f..., L _qmaximum bag long, t _bAGrepresent the virtual link VL={L in AFDX network ₁, L ₂..., L _e..., L _f..., L _qthe bandwidth allocation interval of transmission of messages, n represents the sum of switch in AFDX network, C _nnrepresent n switch S _nwith n switch S _nconnection.

In the present invention, delay _qrepresent any virtual link L that Adoption Network calculation method obtains _qthe message transmission delay time, M _qrepresent any virtual link L _qmessage transmission delay boundary value.

What network calculus method adopted is " electric light and control " the 15th the 9th phase of volume, in September, 2008, and author Yang Yun, bear Hua Gang, the content of Section 3 in " calculating the network calculus method that AFDX postpones ", " utilizes the network delay of the theoretical AFDX of calculating of network calculus ".

In the present invention, according to fitness function Fitness=AVL _max+ time _postponecalculate CH={h ₁, h ₂..., h _xin each chromosomal fitness function value be designated as FCH={Fh ₁, Fh ₂..., Fh _x, and from described FCH={Fh ₁, Fh ₂..., Fh _xin select minimum fitness function value Fh _minimum.

In the present invention, calculate all chromosomal fitness function values, the fitness function value of different virtual link paths differs greatly, if the method that directly uses roulette to select will cause some chromosomes almost can not be copied in the next generation and go, has had a strong impact on the diversity of population.The less chromosome representing of fitness function value is more excellent.

Step 5: genetic manipulation

In the present invention, when the chromosome population of former generation is designated as CH _i, previous generation chromosome population is designated as CH _i-1, chromosome population of future generation is designated as CH _i+1.

Step 501, chromosomal the first cross modal:

In the present invention, the mode that uses single-point to intersect: (A) first crosspoint is selected in public branch exchange place, then to the chromosome population CH when former generation _iin chromosome carry out exchange pairing, the part after exchange pairing crosspoint fills in chromosome separately, thus two new chromosomes that generate; Finally described new chromosome is added to virtual link of future generation path population CH _i+1in.

In the present invention, the mode that uses single-point to intersect: (B) first crosspoint is selected in public branch exchange place, then to the chromosome population CH when former generation _iin chromosome carry out exchange pairing, the part after chiasma just enters in chromosome separately, thus two new chromosomes that generate; If duplicate gene (being that loop appears in path) in two described new chromosomes, delete the switch of any one repetition, and the new chromosome of having deleted after repeated exchanged machine is added to virtual link of future generation path population CH _i+1in.

In the present invention, described public branch exchange place refers to except source switch with object switch switch in addition.If there are multiple public branch exchanges, select at random the crosspoint of a public branch exchange as genetic manipulation.

For example: chromosome h ₂=1-2-4-6-m and chromosome h ₃=1-8-4-7-6-m selects switch S ₄as crosspoint, after intersecting, obtain two new chromosomes,

For example: chromosome h ₃=1-8-4-7-6-m and chromosome h ₆=1-2-4-8-3-5-6-m selects switch S ₄as crosspoint, after intersecting, obtain two new chromosomes, ? the middle switch S repeating that occurred ₈, delete the switch S of a rear repetition ₈, and by the new chromosome of having deleted after repeated exchanged machine add in the population of virtual link of future generation path.

Step 502, chromosomal the second cross modal:

In the present invention, the mode that uses single-point to intersect: (A) first crosspoint is not selected to be in public branch exchange place, then to the chromosome population CH when former generation _iin chromosome select the chromosome in the middle of being positioned to intersect, the part after exchange pairing crosspoint fills in chromosome separately, thus two new chromosomes that generate; Finally described new chromosome is added to virtual link of future generation path population CH _i+1in.

In the present invention, the mode that uses single-point to intersect: (B) first crosspoint is not selected to be in public branch exchange place, then to the chromosome population CH when former generation _iin chromosome select the chromosome in the middle of being positioned to intersect, the part after exchange pairing crosspoint fills in chromosome separately, thus two new chromosomes that generate; If duplicate gene (being that loop appears in path) in two described new chromosomes, delete the switch of any one repetition, and add and in the population of virtual link of future generation path, add virtual link of future generation path population CH having deleted new chromosome after repeated exchanged machine _i+1in.

For example: chromosome h ₁=1-3-5-a-b-m and chromosome h ₅=1-8-4-6-m selects that middle switch of described chromosome (at h ₁in=1-3-5-a-b-m, be switch S ₅, at h ₅in=1-8-4-6-m, be switch S ₄) as crosspoint, after intersecting, obtain two new chromosomes,

Step 503, chromosomal the first variant form:

In the present invention, use the mode of single-point variation: (A) to the chromosome population CH when former generation _ifirst change point is selected in except source switch with object switch any switch place in addition, selects to exist link-attached another switch of physics as variation result with described switch, obtains new chromosome after variation, then by described variation afterwards new chromosome add virtual link of future generation path population CH _i+1.

In the present invention, use the mode of single-point variation: (B) to the chromosome population CH when former generation _ifirst change point is selected in except source switch with object switch any switch place in addition, selects to exist link-attached another switch of physics as variation result with described switch new chromosome after acquisition variation; If duplicate gene (being that loop appears in path) in new chromosome after variation, delete the switch of any one repetition, then the new chromosome of having deleted after repeated exchanged machine is added to virtual link of future generation path population CH _i+1in.

For example: chromosome h ₄=1-3-5-6-b-m selects switch S ₅make a variation, switch S ₅variation is switch S ₇, obtain new chromosome and be

For example: chromosome h ₆=1-2-4-8-3-5-6-m selects switch S ₈make a variation, switch S ₈variation is switch S ₂, obtain new chromosome and be delete the switch S of a rear repetition ₂, the new chromosome that has obtained deleting after repeated exchanged machine is

Step 504, chromosomal the second variant form:

In the present invention, use the mode of single-point variation, to the chromosome population CH when former generation _ifirst change point is selected in except source switch with object switch any switch place in addition, directly deletes described switch, obtains new chromosome after variation, then by described variation afterwards new chromosome add virtual link of future generation path population CH _i+1in.

For example: chromosome h ₁=1-3-5-a-b-m selects switch S _amake a variation, directly delete switch S _aobtaining new chromosome is

In the example of enumerating in the present invention, new chromosome to add virtual link of future generation path population CH _i+1in select operation.

For the variation in the present invention, there is a kind of chromosome only containing two genes, be source switch and object switch, if in initial population, not containing this chromosome, above crossover and mutation operation can not produce this chromosome yet, and this chromosome is probably optimal solution, therefore here to mutation operation correct, this chromosome may be occurred after variation.After random selection change point, not necessarily to change gene, also can directly delete gene.

Step 505, chromosomal selection mode:

In the present invention, chromosomal selection mode is by the chromosome population CH when former generation _i={ h ₁, h ₂..., h _xand by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponedescribed in calculating in each chromosomal fitness function value and choose x minimum chromosome as population CH of future generation _i+1.

In the example of enumerating in the present invention, chromosomal selection mode is by the chromosome population CH when former generation _i={ h ₁, h ₂, h ₃, h ₄, h ₅, h ₆and by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponedescribed in calculating in each chromosomal fitness function value and choose six minimum chromosomes as population CH of future generation _i+1.

Step 6: judge whether to meet hereditary end condition

Judge the fitness function value Fh of described minimum _minimumafter continuous 3 generation genetic manipulations, described Fh _minimumwhile variation, stop genetic manipulation; If do not meet, to CH={h ₁, h ₂..., h _xin every item chromosome re-start the genetic manipulation of step 5.

Described termination genetic manipulation refers to that the chromosome of selecting fitness function value minimum in generation population is in the end as optimum results, from chromosome, extract the switch of source end system A to destination system B process, thereby obtain effective virtual link path.

embodiment 1

The present embodiment is to carry out emulation based on matlab (version number 7.13) platform; Matlab is a kind of advanced techniques computational language and interactive environment for algorithm development, data visualization, data analysis and numerical computations.

Shown in Figure 4, to having configured in the AFDX network in VL path, there are 7 virtual links: i.e. L ₁configuration path be S ₁-S ₂-S ₄-S ₆, source end system is E ₁, destination system is E ₆; L ₂configuration path be S ₂-S ₄-S ₆, source is system E ₂, destination system is E ₆; L ₃configuration path be S ₃-S ₇-S ₆, source end system is E ₃, destination system is E ₆; L ₄configuration path be S ₄-S ₆, source end system is E ₄, destination system is E ₆; L ₅configuration path be S ₅-S ₆, source end system is E ₅, destination system is E ₆; L ₆configuration path be S ₇-S ₆, source end system is E ₇, destination system is E ₆; L ₇configuration path be S ₈-S ₃-S ₅-S ₆, source end system is E ₈, destination system is E ₆.Configuration r _max=4000bits, t _bAG=4ms, M _q=1ms.

Choose virtual link L ₁be optimized its source end system E ₁(it is connected to switch S ₁on), destination system E ₆(it is connected to switch S ₆on), from source end system E ₁to destination system E ₆path be S ₁-S ₂-S ₄-S ₆, L before Adoption Network calculation is optimized ₁the propagation delay time of bearer messages is 629.6 microseconds.

The switch matrix building according to Fig. 4 $\mathrm{SC}=\left[\begin{array}{cccccccc}0& 1& 1& 0& 0& 0& 0& 1\\ 1& 0& 0& 1& 0& 0& 0& 1\\ 1& 0& 0& 1& 1& 0& 1& 1\\ 0& 1& 1& 0& 0& 1& 1& 1\\ 0& 0& 1& 0& 0& 1& 1& 0\\ 0& 0& 0& 1& 1& 0& 1& 0\\ 0& 0& 1& 1& 1& 1& 0& 0\\ 1& 1& 1& 1& 0& 0& 0& 0\end{array}\right];$

While making population number x=3, the effective virtual link of the Article 1 path selecting according to path exclusion condition $R-u_1=\left(S_1^{E_1}\stackrel{u_1}{\→}{S}_6^{E_6}\right)=S_1-S_2-S_4-S_6,$ The encoded chromosome that obtains is h ₁=1-2-4-6;

The effective virtual link of Article 2 path $R-u_2=\left(S_1^{E_1}\stackrel{u_2}{\→}{S}_6^{E_6}\right)=S_1-S_8-S_4-S_6,$ The encoded chromosome that obtains is h ₂=1-8-4-6;

The effective virtual link of Article 3 path $R-u_3=\left(S_1^{E_1}\stackrel{u_3}{\→}{S}_6^{E_6}\right)=S_1-S_8-S_2{-S}_4-S_7-S_6,$ The encoded chromosome that obtains is h ₃=1-8-2-4-7-6.

According to fitness function Fitness=AVL _max+ time _postponecalculate CH ₁={ h ₁, h ₂, h ₃in each chromosomal fitness function value be designated as FCH={Fh ₁, Fh ₂, Fh ₃}={ 0.9096,0.868,1.022}, selects minimum fitness function value Fh _minimum=0.868.

The operation of first generation genetic manipulation, to CH ₁={ h ₁, h ₂, h ₃carry out genetic manipulation, by chromosome h ₁=1-2-4-6 and chromosome h ₃=1-8-2-4-7-6 is with switch S ₂for crosspoint intersects, obtain chromosome h ₂=1-8-4-6 selects switch S ₈for change point makes a variation, switch S ₈variation is switch S ₃, obtain new chromosome and be by the chromosome population CH when former generation ₁={ h ₁, h ₂, h ₃and by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponecalculate each chromosomal fitness function value is respectively choose the wherein chromosome h of 3 fitness function value minimums ₁, h ₂, as population CH of future generation ₂chromosome, now, CH ₂={ h ₁, h ₂, h ₃the chromosome h of population ₁=1-2-4-6, chromosome h ₂=1-8-4-6, chromosome h ₃=1-3-4-6.Fh in this generation population _minimum=0.868, now do not meet hereditary end condition.

The operation of second generation genetic manipulation, to CH ₂={ h ₁, h ₂, h ₃carry out genetic manipulation, by chromosome h ₁=1-2-4-6 and chromosome h ₂=1-8-4-6 is with switch S ₄for crosspoint intersects, obtain chromosome h ₃=1-3-4-6 selects switch S ₄for change point makes a variation, switch S ₄variation is switch S ₃, obtain new chromosome and be by the chromosome population CH when former generation ₂={ h ₁, h ₂, h ₃and by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponecalculate each chromosomal fitness function value is respectively choose the wherein chromosome h of 3 fitness function value minimums ₂, h ₃, as population CH of future generation ₃chromosome, now, CH ₃={ h ₁, h ₂, h ₃the chromosome h of population ₁=1-8-4-6, chromosome h ₂=1-3-4-6, chromosome h ₃=1-8-4-6.Fh in this generation population _minimum=0.868, now do not meet hereditary end condition.

The operation of third generation genetic manipulation, to CH ₃={ h ₁, h ₂, h ₃carry out genetic manipulation, by chromosome h ₁=1-8-4-6 and chromosome h ₃=1-8-4-6 is with switch S ₄for crosspoint intersects, obtain chromosome h ₂=1-3-4-6 selects switch S ₃for change point makes a variation, switch S ₃variation is switch S ₂, obtain new chromosome and be by the chromosome population CH when former generation ₃={ h ₁, h ₂, h ₃and by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponecalculate each chromosomal fitness function value is respectively choose the wherein chromosome h of 3 fitness function value minimums ₁, h ₂, h ₃as population CH of future generation ₄chromosome, now, CH ₄={ h ₁, h ₂, h ₃the chromosome h of population ₁=1-8-4-6, chromosome h ₂=1-3-4-6, chromosome h ₃=1-8-4-6.Fh in this generation population _minimum=0.868, now after continuous 3 generation genetic manipulations, Fh _minimumdo not change, stop genetic manipulation, be i.e. selected population CH ₄={ h ₁, h ₂, h ₃in the chromosome of fitness function value minimum as optimum results, be h here ₁=1-8-4-6, has source end system E ₁to destination system E ₆the switch of process is S ₁-S ₈-S ₄-S ₆, by path S ₁-S ₈-S ₄-S ₆as virtual link L ₁path as optimum results.

Have and optimize front L ₁path be S ₁-S ₂-S ₄-S ₆path as shown in triangle as hollow in dotted line in Fig. 4, the L after genetic manipulation is optimized ₁path be S ₁-S ₈-S ₄-S ₆path as shown in triangle as solid in dotted line in Fig. 4.L after optimizing ₁the propagation delay time of bearer messages is 588 microseconds, with optimize before 629.6 microseconds compared with, its propagation delay time has obtained reducing, the real-time of transmission of messages has obtained enhancing.

Claims (5)

1. adopt genetic algorithm to carry out a method for AFDX network path optimization, described AFDX network refers to the AFDX network that disposes VL path; It is characterized in that including the following step:

Step 1: build switch connection matrix;

Carry out the connection initialization between all switches to disposing the AFDX network in VL path, obtain n × n switch connection matrix n is the number of the switch in AFDX network;

Step 2: build virtual link path population;

Virtual link path population scale U={u is set ₁, u ₂..., u _x, the switch matrix SC of step 1 is chosen and meets U={u according to path exclusion condition ₁, u ₂..., u _xsource-object-switch path

Step 3: build chromosome;

Source-object-switch path that step 2 is obtained in the identification number of switch as gene, the encoded chromosome population CH={h that obtains ₁, h ₂..., h _x;

Step 4: calculate chromosomal fitness function value;

Described fitness function Fitness=AVL _max+ time _postpone, wherein, maximum link load factor ${\mathrm{AVL}}_{\max }=\underset{n}{\max }\frac{\underset{\mathrm{VL}\∈C_{\mathrm{nn}}}{\mathrm{\Σ}}\frac{r_{\max }}{t_{\mathrm{BAG}}}}{100\×{10}^6}\×100\%,$ Message transmission delay time rate

Step 5: genetic manipulation;

Current chromosome population is designated as CH _i, previous generation chromosome population is designated as CH _i-1, chromosome population of future generation is designated as CH _i+1;

Step 501, chromosomal the first cross modal:

The mode that uses single-point to intersect, crosspoint is selected in public branch exchange place, if there are multiple public branch exchanges, select at random one as crosspoint, then the part after exchange pairing crosspoint, two new chromosomes that generate, and described new chromosome is added in the population of virtual link of future generation path; If duplicate gene in new chromosome after intersecting, delete the switch of any one repetition, and the new chromosome of having deleted after repeated exchanged machine is added in the population of virtual link of future generation path;

Step 502, chromosomal the second cross modal:

The mode that uses single-point to intersect, crosspoint is not selected to be in public branch exchange place, selects the chromosome in the middle of being positioned to intersect, the then part after exchange pairing crosspoint, two new chromosomes that generate, and described new chromosome is added in the population of virtual link of future generation path; If duplicate gene in new chromosome after intersecting, delete the switch of any one repetition, and add in the population of virtual link of future generation path and add in the population of virtual link of future generation path having deleted new chromosome after repeated exchanged machine.

Step 503, chromosomal the first variant form:

The mode that uses single-point variation, change point is selected in except source switch with object switch any switch place in addition, selects to exist link-attached another switch of physics as variation result with described switch, obtains new chromosome, and described new chromosome is added in the population of virtual link of future generation path; If duplicate gene in new chromosome after variation, delete the switch of any one repetition, and the new chromosome of having deleted after repeated exchanged machine is added in the population of virtual link of future generation path.

Step 504, chromosomal the second variant form:

The mode that uses single-point variation, change point is selected in except source switch with object switch any switch place in addition, directly deletes described switch, obtains new chromosome, and described new chromosome is added in the population of virtual link of future generation path.

Step 505, chromosomal selection mode:

Chromosomal selection mode is by the chromosome population CH when former generation _i={ h ₁, h ₂..., h _xand by the new chromosome population of intersect-variation generation mix, obtain mixed population then pass through Fitness=AVL _max+ time _postponedescribed in calculating in each chromosomal fitness function value and choose x minimum chromosome as population CH of future generation _i+1.

Step 6: judge whether to meet hereditary end condition;

Judge the fitness function value Fh of described minimum _minimumafter continuous 3 generation genetic manipulations, described Fh _minimumwhile variation, stop genetic manipulation; If do not meet, to CH={h ₁, h ₂..., h _xin every item chromosome carry out genetic manipulation.

2. the method that employing genetic algorithm according to claim 1 is carried out the optimization of AFDX network path, it is characterized in that: the assignment condition of described switch connection matrix SC has three kinds, the first is that the connection between switch self is designated as 0, the second is between switch, not have the connection of physical link to be designated as 0, and the third has the connection of physical link to be designated as 1 between switch.

3. the method that employing genetic algorithm according to claim 1 is carried out the optimization of AFDX network path, is characterized in that: the described path exclusion condition in step 2 includes three kinds of situations;

Eliminating situation in the first path refers to that source end system A and destination system B are connected on same switch, do not choose this virtual link path;

The second path eliminating situation refer to source end system A to destination system B in the virtual link path of process, have the same switch of repetition, do not choose this virtual link path;

The third path eliminating situation refer to source end system A to destination system B in the virtual link path of process, between switch, there is no physical link, do not choose this virtual link path.

4. the method that employing genetic algorithm according to claim 1 is carried out the optimization of AFDX network path, is characterized in that: in virtual link path construction process, the value of population number is 3～20 virtual link paths.

5. the method that employing genetic algorithm according to claim 1 is carried out the optimization of AFDX network path, is characterized in that: can reduce the propagation delay time of the virtual link message of AFDX network, reduce to reach 5～15%.