CN101047449A - Waveguide-optical fibre automatic core regulating method and device based on multi-target evolution algorithmic - Google Patents

Abstract

A method for automatically regulating core of waveguide-optical fiber based on multi-object evolution algorithm includes treating discrete genetic screening process by stable culling policy, applying father generation cross and real value variation in searching process, only change two individual bodies in each generation of genetic population and having no relationship to artificial weight-setting in process of realizing multi-object optimization.

Description

Waveguide-optical fibre automatic core regulating method and device based on multi-objective evolutionary algorithm

Technical field

The present invention relates to a kind of waveguide-optical fibre automatic core regulating method and device, belong in the optical communication field, the automatic interconnection technique and the encapsulation technology field of integrated optical wave guide device and monomode fiber array based on multi-objective evolutionary algorithm.

Background technology

In recent years, in the ultrahigh speed transmission experiment of optic communication link net and Access Network Fiber to the home technology, adopted fiber waveguide device more and more.Lead-in light waveguide device in fiber optic network, the binding encapsulation that must solve optical fiber and fiber waveguide.A key technology that wherein will solve is that fiber waveguide device docks with the low-loss of fiber array.International advanced index is that every damaged on end consumption is lower than 0.15dB.Reach this index, the mould field distribution that requires on the one hand fiber waveguide is consistent with optical fiber as far as possible, necessarily requires the optical axis alignment precision of fiber waveguide and optical fiber to be controlled at below the 0.1 μ m on the other hand.

Utilize high precision adjusting rack to adopt the conventional manual operation, specification requirement is very high, and particularly after being coupled into 0.3dB, accurately testing as little varying signal of operation criterion is key problem in technology, and corresponding microoperation is very difficult, so efficient is very low, repeatability is very poor.Adopt automatic technology to realize that fiber waveguide is docked coupling with high efficiency, low-loss between the optical fiber be a kind of effective means, conventional automatic core-adjusting method follows manually operated thinking, and the relative position of finely tuning waveguide optical axis and optical axis by scanning obtains big as far as possible coupling efficiency.Relevant patent documentation has:

A) the automatic aligning of fiber waveguide device and fiber array and channel energy equalizing technique thereof (Wuhan light news Science and Technology Ltd., application number 02115963.7, publication number 1383012)

B) a kind of fiber waveguide device and self-aligning method of fiber array (Wuhan light news Science and Technology Ltd., application number 02115964.5, publication number 1383005) of being used for

Conventional automatic core-adjusting method depends on the accurate test to little varying signal, not only requires the guide rail of adjusting bracket that very high traveling precision is arranged, and also requires detector that very high sensitivity is arranged.Generally speaking, owing to precision reason and various random error, usually cause transferring the repeated bad of core.

In recent years reported that a kind of new barycenter core regulating method is used to address this problem, this method avoids directly seeking peak, adopts the barycenter of measuring near the coupling efficiency distribution of calculating peak value to determine peak.Overcome conventional accent core process from principle and depended on the difficulty that little varying signal is accurately tested.Relevant patent documentation has:

The barycenter core regulating method of c) waveguide-optical fiber automatic butt and used automatic core-adjusting device (Shanghai University of Science and Technology, patent No. ZL03129249.6) thereof

The operation principle of this system is based upon on the feedback servo basis of light signal, shows very strong logicality.Must make between optical fiber-waveguide-optical fiber by initial adjustment and realize logical light, but when the feedback path between computer and power meter entered running status, the automatic core-adjusting program can come into force.The logical light work of initial adjustment still needs manual operation to finish, and this work is very high to operating personnel's specification requirement, relies on operator's qualification, and the time can't lump together.In view of this, proposed a kind of automatic core-adjusting system that adopts genetic algorithm recently, important improvement is that the logical light of initial adjustment has been realized automation, and people's dependence is significantly reduced.Relevant patent documentation has:

D) based on the waveguide-optical fibre automatic core regulating method and the device (Shanghai University of Science and Technology, application number 200410018175.1) thereof of genetic algorithm

But, although hereditary method is not limited by the strong logicality of feedback processing, allow to adopt the parallel work-flow of each dimension of space, but per generation population must contain tens of～hundreds of individual, mean that every hereditary generation must carry out tens of groups of parallel work-flows at least, fiber array and fiber waveguide array transfer the optimizing convergence of core need go through the heredity in nearly 100 generations, and total parallel work-flow reaches thousands of groups, time-consumingly want tens of minutes.In addition, what array butt joint automatic core-adjusting related to is a multi-objective optimization question, has both required each passage that high coupling efficiency is arranged, and wishing again has good homogeneous between each passage coupling efficiency.Because there is the channel spacing error inevitably in discrete array device, when reality was transferred core, the implementation procedure of two targets usually was to conflict mutually, can not be met simultaneously.What is heavy how gently by about weighting in two indexs, but the reasonable setting of weight proportion is still a difficult thing so far, generally can only adopt examination repeatedly to transfer the method for revising gradually.

Summary of the invention

Above-mentioned hereditary method automatic core-adjusting operational ton is excessive in order to solve in the present invention, weight proportion is set the difficult technologies problem, develop a kind of newly based on the automatic core-adjusting method of multi-objective evolutionary algorithm and be applicable to the automatic core-adjusting device of this method.

The present invention is based on following basic principle:

What run into usually in the optical communication system is butt joint between single mode waveguide array and the monomode fiber array, at the butt joint end face, realizes carrier transmission by the coupling of the power between the eigen mode.Having filled up the cured binders for the treatment of that operation wavelength is absorbed little and refractive index match between waveguide and the optical fiber, end face reflection is approaching-50dB, can ignore, single core end face coupling efficiency shows as the overlap integral between the guided mode field distribution of knowing in theory.Be without loss of generality, as shown in Figure 1, the optical axis of establishing the bar waveguide has an angular deviation θ along the z direction between optical axis and bar waveguide optical axis, the two optical axis hearts have deviation delta x and Δ y on perpendicular to the z direction of principal axis, fiber end face and Waveguide end face are Δ z along the axial spacing of z.Consider that some unpredictable factors cause the asymmetric of field distribution of waveguide in the fiber waveguide manufacturing process, use W _XoAnd W _YoRepresent single mode waveguide guided mode girdling the waist on x and y orientation respectively, the with a tight waist of optical fiber mode is W _FoSingle core end face coupling efficiency η can be expressed as:

η＝η _x·η _y (1a)

${\mathrm{\&eta;}}_x=\frac{2\exp [-\frac{k^2({\mathrm{<figure-callout\; id="2"\; label="W\; fo"\; filenames="A20071003898800211.png,A20071003898800212.png"\; state="\{\{state\}\}">W</figure-callout>}}_{\mathrm{fo}}^{}+W_{\mathrm{xo}}^2){\mathrm{\&theta;}}^2}{2{\mathrm{\&eta;}}_{\mathrm{xo}}\left(\mathrm{\&Delta;z}\right)}-\frac{2}{{\mathrm{\&eta;}}_{\mathrm{xo}}\left(\mathrm{\&Delta;z}\right)}(\frac{{\mathrm{\&Delta;x}}^2}{{\mathrm{<figure-callout\; id="2"\; label="W\; fo"\; filenames="A20071003898800211.png,A20071003898800212.png"\; state="\{\{state\}\}">W</figure-callout>}}_{\mathrm{fo}}^{}}+\frac{{(\mathrm{\&Delta;z\&theta;}-\mathrm{\&Delta;x})}^2}{W_{\mathrm{xo}}^2})]}{\sqrt{{\mathrm{\&eta;}}_{\mathrm{xo}}\left(\mathrm{\&Delta;z}\right)}}...\left(1b\right)$

${\mathrm{\&eta;}}_y=\frac{2\exp [-\frac{2{\mathrm{\&Delta;<figure-callout\; id="2"\; label="y"\; filenames="A20071003898800211.png,A20071003898800212.png"\; state="\{\{state\}\}">y</figure-callout>}}^2}{{\mathrm{\&eta;}}_{\mathrm{yo}}\left(\mathrm{\&Delta;z}\right)}(\frac{1}{{\mathrm{<figure-callout\; id="2"\; label="W\; fo"\; filenames="A20071003898800211.png,A20071003898800212.png"\; state="\{\{state\}\}">W</figure-callout>}}_{\mathrm{fo}}^{}}+\frac{1}{W_{\mathrm{yo}}^2})]}{\sqrt{{\mathrm{\&eta;}}_{\mathrm{yo}}\left(\mathrm{\&Delta;z}\right)}}...\left(1c\right)$

In the formula

${\mathrm{\&eta;}}_{\mathrm{xo}}\left(\mathrm{\&Delta;z}\right)={(\frac{W_{\mathrm{xo}}}{W_{\mathrm{fo}}}+\frac{W_{\mathrm{fo}}}{W_{\mathrm{xo}}})}^2+{\left(\frac{2\mathrm{\&Delta;z}}{k{W}_{\mathrm{fo}}{W}_{\mathrm{xo}}}\right)}^2...\left(2a\right)$

${\mathrm{\&eta;}}_{\mathrm{yo}}\left(\mathrm{\&Delta;z}\right)={(\frac{W_{\mathrm{yo}}}{W_{\mathrm{fo}}}+\frac{W_{\mathrm{fo}}}{W_{\mathrm{yo}}})}^2+{\left(\frac{2\mathrm{\&Delta;z}}{k{W}_{\mathrm{fo}}{W}_{\mathrm{yo}}}\right)}^2...\left(2b\right)$

Here, k is the wave number in the medium of light wave between waveguide and optical fiber, and is relevant with the refractive index of medium, reflected the dielectric influence of binding agent.Coupling efficiency η reduces and increases with θ, Δ x, Δ y and Δ z's, when θ, Δ x, Δ y and Δ z are zero, maximum coupling efficiency is arranged:

${\mathrm{\&eta;}}_{\max }=\frac{1}{(\frac{W_{\mathrm{fo}}}{W_{\mathrm{xo}}}+\frac{W_{\mathrm{xo}}}{W_{\mathrm{fo}}})(\frac{W_{\mathrm{fo}}}{W_{\mathrm{yo}}}+\frac{W_{\mathrm{yo}}}{W_{\mathrm{fo}}})}...\left(3\right)$

This moment, theoretical following formula (3) provided 100% end face coupling efficiency if the mould field distribution of waveguide and optical fiber is identical.Although actual conditions are not ideals like this, but theory analysis is pointed out a principle, for given fiber waveguide and optical fiber, obtain high end face coupling efficiency, must make both optical axis alignments as far as possible and keeping parallelism, both end face at interval should be tried one's best little and fill suitable refractive index match medium.

It is the popularization of above-mentioned single core coupling that the multichannel array is coupling on the principle, because actual Waveguide array and fiber array are not the same on the core spacing, can not accomplish that all passages all obtain the highest coupling efficiency simultaneously, practical compromise is to adopt two indexs to exchange the core result to weigh evaluation, one is whether the insertion loss sum of all passages is enough little, another is whether insert the difference of loss maximum and minimum value in all passages enough little, and the latter is used to define uniformity.Because the loss of discrete device self is determined, enters when transferring core butt joint flow process, the factor that loss is inserted in influence mainly is an end face coupling loss, and end face coupling loss is represented with the decibel value of end face coupling efficiency.

The output that can monitor all passages when array is transferred core in real time is best certainly, but because the waveguide channels number is on the increase with the development of device technology, and the restriction that is subjected to the package system cost, transfer core system generally not to be equipped with a large amount of light power meters, real current situation is to adopt the method for transferring core as the illustrative twin-core of Fig. 2, when 1 * 8 Waveguide branching coupler docks coupling with fiber array, the output of getting the 2nd and the 7th paths (can also be other two paths separated by a distance) is as sampled signal, and the butt joint of other passage coupling guarantees by the precision of each passage geometric position of discrete device.For the fiber waveguide array of 0.1 μ m positional precision with less than the fiber array of 0.5 μ m positional precision, adopting in the practicality of twin-core core regulating method does not have obstacle.

Because the end face of multicore array has bigger length-width ratio, end face fine setting at interval must be associated with faces parallel, requires to be close between end face parallel, otherwise because end collision in limit causes fine setting at interval not carry out.Therefore the perpendicularity deviation of the optical axis of commercial waveguide chip or fiber array and its end face generally in 0.2 °, can be controlled by the depth of parallelism of adjusting between the two array end faces to ensure the depth of parallelism between waveguide optical axis and the optical axis.The adjustment of the depth of parallelism and end distance relates to and θ between the end face of exit end _x, θ _yWith the z three-dimensional, the butt joint of array twin-core relates to x, y and θ _zThree-dimensional.At the incident end, since the circular symmetry of single fiber, θ _zAdjustment forfeiture meaning.So the space dimensionality that the twin-core automatic core-adjusting relates to is 11 dimensions, be respectively to be in space five dimensions waveguide chip input one side, that adjust input optical fibre core position, and be in six-dimensional space waveguide chip output one side, that adjust the positioning optical waveguides core position of fiber array.

The importing of multi-objective evolutionary algorithm:

What array butt joint automatic core-adjusting related to is a multi-objective optimization question, has both required each passage that high coupling efficiency is arranged, and wishing again has good homogeneous between each passage coupling efficiency.The multi-objective evolutionary algorithm that the present invention relates to is a kind of based on the noninferior solution set algorithm of stablizing replacement policy, multi-target evolution, adopting does not have the domination ordering rule to each the individual Pareto of the distribution order in the population, after individuality is pressed the classification of Pareto ordinal number, all individualities with identical ordinal number are calculated its crowding distance, the individuality of low Pareto ordinal number means the more composite request of a plurality of targets of convergence, and the individuality of big crowding distance means richer diversity.Compare with traditional genetic algorithm, the single fitness function that the present invention does not adopt foundation to comprise a plurality of weighting targets is estimated the individuality in the population, but eliminate order by the size decision of individual Pareto ordinal number and crowding distance thereof, the individuality that the Pareto order is little and crowding distance is big is endowed more procreation chance.

In order to improve convergence rate, searching process combines that many parents intersect and two of ant group climbing methods are heuristic.Produce new offspring's individuality with many parents cross method earlier to M parental generation is individual, form one M+1 individual transitional population, behind the individuality of position, the superseded end of the size of individual Pareto ordinal number and crowding distance thereof, produce new offspring's individuality with real-valued variation method again, eliminate position, end individuality by identical criterion, obtain thus by M the individual population of forming of future generation.Obviously, when method of the present invention is put to practicality, only need do the machine operation of two individualities to each for evolving, operational ton significantly reduces than traditional genetic algorithm, can significantly improve the efficient of automatic core-adjusting.

Import the mapping that multi-objective evolutionary algorithm mechanism need solve evolution algorithmic genetic model and reality accent core process, certain space relative bearing between waveguide optical axis and the optical axis when describing reality accent core with the individual concept in the evolution algorithmic model; The population notion has reflected the set of the relative bearing of being considered of having living space.The benefit of method of the present invention is to allow to adopt the real coding scheme, precision that has existed when not only having avoided the binary processing continuous function and the contradiction between the search volume, and computational speed improves greatly, therefore represent chromosome with the real number array that contains a plurality of decision variables, be equivalent to the coordinate of physical space.Two indexs that the expectation of accent core process reaches are respectively as two target function values.Through notion mapping like this, array butt joint automatic core-adjusting is converted into the process that multi-objective evolutionary algorithm is optimized two target function values to coupling efficiency and inhomogeneity searching process.

Get 1 * 8 Waveguide branching coupler shown in Figure 2 and be coupled as example with docking of fiber array, the space dimensionality that the twin-core automatic core-adjusting relates to is 11 dimensions.Usually, under micro-supervisory work, initial deviation between waveguide optical axis and optical axis is controlled within the 50 μ m easily, the spacing of Waveguide end face and fiber end face is controlled within the 4 μ m easily, the excursion of determining x axle and y axle parameter thus is [50 μ m, 50 μ m], the excursion of z axle parameter is [0 μ m, 4 μ m].Get θ _zThe hunting zone at angle is ± 5 °.For fear of waveguide chip and fiber array take place up and down or about limit end collision, θ _xAngle and θ _yThe hunting zone at angle is by following formula

$-\frac{z_{\max }}{h}\&le;\sin {\mathrm{\&theta;}}_x\&le;\frac{z_{\max }}{h},-\frac{z_{\max }}{w}\&le;\sin {\mathrm{\&theta;}}_y\&le;\frac{z_{\max }}{w}...\left(4\right)$

Decision, h, w and z here _MaxIt is respectively the largest interval between thickness, width and the both ends of the surface of fiber array.

Above-mentioned space 11 dimensions are corresponding to 11 decision variables of evolution algorithmic, and the decision variable of one group of 11 dimension constitutes body U one by one _i, form population by setting up M group of individuals.Each individuality is the corresponding coupling efficiency of corresponding generation in transferring the core process, coupling efficiency is converted into above-mentioned two target function values estimates.Because what obtain during practical operation is the luminous power of sampled signal, it and insertion are worn to inverse ratio, and then uniformity is directly represented with the difference of two sampled signal power.Two target functions are used respectively

f ₁(U _i)＝-[P ₁(U _i)+P ₂(U _i)] (5)

f ₂(U _i)＝|P ₁(U _i)-P ₂(U _i)| (6)

Definition, P in the formula ₁And P ₂Be respectively to transfer the core position to be in by individual U _iDuring the dimensional orientation of regulation, the Output optical power that records by two sampling channels, they are the insertion loss when adopting decibel to represent with the ratio of input optical power.Obvious P ₁And P ₂Bigger, target function f ₁Littler; P ₁With P ₂Difference littler, target function f ₂Littler.

The allocation rule of all individual Pareto orders is in the population: establishing the individuality of being investigated is U _i, if with another individual U _{J ≠ i}There is following target function value relation

f ₁(U _i)＞f ₁(U _j≠i)and f ₂(U _i)≥f ₂(U _j≠i) (7)

Then individual U _iThe Pareto ordinal number increase by 1, otherwise the ordinal number increment is 0.By with U _iAll individualities in addition compare one by one, and accumulative total obtains individual U _iPareto ordinal number value.So the investigation process spreads all over all individualities, finishes the classification of the individuality of population by the Pareto ordinal number.Obviously, criterion of (7) formula and ordinal number allocation rule have guaranteed that the individuality of low Pareto ordinal number more possesses the tendency that the binocular mark is optimized simultaneously.To having the individuality of identical ordinal number in the population, calculate crowding distance: the k under the identical ordinal number individual U by following rule _jPress

f ₁(U _m)＜f ₁(U _m+1) m＝1，2，L，k-1. (8)

f ₂(U _n)＜f ₂(U _N+1) n=1,2, L, the order of k-1. is arranged with cambic m subscript and n subscript respectively, calculates

$d_1\left(U_m\right)=\left\{\begin{array}{cc}\&infin;& \mathrm{at\; m}=1\mathrm{or\; k}\\ \frac{f_1\left(U_{m+1}\right)-f_1\left(U_{m-1}\right)}{f_1\left(U_k\right)-f_1\left(U_1\right)}& \mathrm{at\; m}\&NotEqual;1\mathrm{or\; k}\end{array}\right....\left(9a\right)$

$d_2\left(U_n\right)=\left\{\begin{array}{cc}\&infin;& \mathrm{at\; n}=1\mathrm{or\; k}\\ \frac{f_2\left(U_{n+1}\right)-f_2\left(U_{n-1}\right)}{f_2\left(U_k\right)-f_2\left(U_1\right)}& \mathrm{at\; n}\&NotEqual;1\mathrm{or\; k}\end{array}\right....\left(9b\right)$

Again the corresponding respectively original i subscript arrangement that is returned to is arranged in arrangement of m subscript and n subscript, so each individual U _iThus a pair of d ₁(U _i) and d ₂(U _i).Individual U _iCrowding distance be defined as

d(U _i)＝d ₁(U _i)+d ₂(U _i) (10)

All individualities in the said process traversal population.Obviously, the crowding distance here is meant that in the two-dimensional space by the decision of two target dimensions, in the population certain is investigated individual and its estimation with the space length between the adjacent individuality of ordinal number.The individuality that crowding distance is big, its two target function values are big, therefore big to the diversity contribution with the ordinal number individuality in the population with difference with two target function values of the adjacent individuality of ordinal number.Physically, diversity helps to transfer the core process to avoid being absorbed in the local convergence of suboptimum.In a word, for any two individualities in the population, the individuality that ordinal number is little is better than the big individuality of ordinal number; For two identical individualities of ordinal number, the individuality that crowding distance is big is better than the little individuality of crowding distance.So far finished evaluation with two parameters of Pareto sum of ranks crowding distance to all individualities in the generation population.

Population genetic is finished by three steps such as intersecting, make a variation and eliminate.Intersect and make a variation and influence the speed of evolving and whether restraining and restraining.The individuality that participate in to intersect the more, evolving, it is bigger to converge to optimum possibility; Before the superior and inferior evaluating of the individuality that participation intersects was more arranged, convergence rate sooner.Adopt many parents to intersect for this reason, from current size is the population of M, select m the individuality that the row of evaluation is preceding, by following interlace operation

$U_0=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_i{U}_i...\left(11\right)$

Obtain a new individual U ₀, weight coefficient satisfies here

$\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_i=1,-0.5\&le;a_i\&le;1.5...\left(12\right)$

Relation.New individual U ₀With transition population that size is M+1 of former M individual formation, calculate new individual U ₀Pareto order and crowding distance, eliminate position, end individuality by these two indexs after, the new population that to obtain a size again be M.In order further to increase the population diversity and to accelerate convergence rate, do real-valued mutation operation again, picked at random body one by one by real-valued variation, increases near a new individuality father's individuality from new population, and the formation size is the transition population of M+1 once more.By calculating, compare Pareto order and crowding distance, behind the individuality of position, superseded end, the side gets the population of future generation that size is M similarly.Adopt so ant group climbing method, by generations of intersection, variation, stablely genetic manipulation such as eliminate, constantly evolve, till arriving the maximum evolutionary generation of setting.The size of maximum evolutionary generation is relevant with the convergence property of multi-objective evolutionary algorithm, and the result that can the reference test operation or the result of referential data l-G simulation test determine that it is more than or equal to 35 that the present invention gets maximum evolutionary generation.

Numerical simulation:

The docking system that provides with Fig. 2 is as simulation model, sampling optical fiber is the 2nd and the 7th passage, for outstanding end face coupling efficiency, be without loss of generality, loss, bending loss and the Y branch coupling loss of ignoring 1 * 8 Waveguide branching coupler, this model reaches 100% ideally at the end face coupling efficiency, and the insertion loss of each passage of Waveguide branching coupler equals the principle loss in 1 minute 8 of 9.031dB.Get the W with a tight waist of conventional monomode fiber under the 1550nm wavelength _Fo=5.63 μ m, the asymmetric factor-alpha=1-(W in the mould field of quartzy single mode bar waveguide _Xo/ W _Yo) generally in 0.1% magnitude, get the W with a tight waist of the quartzy single mode bar waveguide of refractive index contrast Δ=0.3% by 0.4% here _Yo=5.63 μ m, W _Xo=5.61 μ m, input optical power are 0dBm.Initial population produces with the method for completely random, and the population size is 30.

Transfer the convergence property of core process in order to hold the multiple target evolution, consider that earlier the core of fiber array and Waveguide array is the ideal situation of 250 μ m at interval, the spacing between the 2nd and the 7th sampling optical fiber is 1250 μ m.The medium of filling in the gap between waveguide both ends of the surface and the I/O fiber array end face is with reference to the parameter of BP300 ultraviolet binding agent, and the refractive index on the 1550nm wavelength is 1.444, with the refringence of quartz less than 0.01, transmitance is 98%/mm.Done 10 times and repeated emulation, in each hereditary 100 generations, initial population adopts random device to generate.Simulation result is listed in table 1, and it is 9.042dB that two cores insert loss mean value, the loss of deduction principle, and end face coupling loss mean value is about 0.011dB.Owing to set optical fiber and waveguide

The fiber array of table 1. desired pitch docks the coupling Simulation result with the Waveguide array twin-core.

Simulation times	Evolutionary generation	Two cores of optimized individual correspondence insert loss (dB) during convergence
				Core
		2	Core 7
1	100			9.070	9.070
		2	100			9.031	9.031
3	100			9.051	9.051
		4	100			9.034	9.034
5	100			9.046	9.046
		6	100			9.031	9.031
7	100			9.053	9.053
		8	100			9.031	9.031
9	100			9.044	9.044
		10	100			9.032	9.032
On average				9.042	9.042

Channel pitch equates that homogeneity error is 0dB.Fig. 3 has provided the wherein convergence process of an emulation, restrains very fastly in early days, approximately evolves through 35 generations and can enter zone less than 9.1dB, and search efficiency and convergence are all fine.

In the actual product, Waveguide array is owing to adopt the photoetching technique preparation, the array spacing has excellent precision, the preparation of fiber array is adopted and is buried the technology of optical fiber quarter behind the V-shaped groove, the precision control of array spacing is difficulty, table 2 has provided with the position distribution data of array core apart from the core center of a quartz base plate V-shaped groove 8 core fibre arrays of tester actual measurement, compares worst error＜0.35 μ m with 250 μ m theoretical pitch.In conjunction with actual conditions,, made of the alternative ideal fiber array of the measured data of table 2

Table 2. 8 core fibre array core center measured values.

The passage ordinal number	1	2	3	4	5	6	7	8
									X shaft position (μ m)	-250.32	0	249.76	499.86	749.94	999.82	1250.64	1499.89
Y shaft position (μ m)	-0.01	0	-0.32	0.12	-0.02	0.11	0.19	0.30

The twin-core automatic butt emulation of multi-objective evolutionary algorithm, table 3 has been concluded the result of 10 emulation, and the mean value of the insertion loss of the passage 2 of correspondence and passage 7 is respectively 9.050dB and 9.085dB when evolving for 100 generations.

Table 3. actual measurement fiber array docks the coupling Simulation result with the Waveguide array twin-core.

Simulation times	Two cores of correspondence insert loss (dB) when evolving for 100 generations
			Core
	2	Core 7
1			9.032	9.093
	2	9.032			9.095
3			9.063	9.133
	4	9.061			9.046
5			9.033	9.078
	6	9.031			9.092
7			9.082	9.049
	8	9.034			9.094
9			9.059	9.052
	10	9.073			9.117
On average			9.050	9.085

Compare with ideal situation, increased 0.008dB and 0.043dB respectively, reflected of the influence of fiber array core apart from error.The coupling position that the 1st time emulation obtains in the option table 3, the insertion loss of having calculated 8 passages the results are shown in Fig. 4.The mean value of the insertion loss of 8 passages is 9.072dB, the loss of deduction principle, and end face coupling loss mean value is about 0.041dB, and characterizing the maximum of uniformity index and the difference of minimum value is 0.054dB, satisfies practical requirement.The above results provides theoretical foundation for multi-objective evolutionary algorithm automatic core-adjusting system.

Based on above-mentioned principle, a kind of waveguide-optical fibre automatic core regulating method of multiple target algorithm is characterized in, method step is:

1. set up the mapping between waveguide-optical fibre automatic core regulating physical parameter and the multi-objective evolutionary algorithm model, individuality in certain relative space position and the corresponding multi-objective evolutionary algorithm model of direction between waveguide optical axis and the optical axis when wherein transferring core, population in the corresponding multi-objective evolutionary algorithm model of the set of above-mentioned relative space position and direction, chromosome in the corresponding multi-objective evolutionary algorithm model of solid axes parameter, first target function f in the corresponding multi-objective evolutionary algorithm model of luminous power sum of two or more waveguide channels that accent core process records ₁, second target function f in the corresponding multi-objective evolutionary algorithm model of difference of the maximum of the luminous power of two or more waveguide channels that accent core process records and minimum value ₂

2. the initial adjustment of fiber array-fiber waveguide-fiber array system is confirmed: adopt computer control, stepper motor driven mode to adjust input and use the monomode fiber array with monomode fiber array and output, make it respectively deviation of the alignment with fiber waveguide input port and output port in ± 50 mu m ranges;

3. the method with completely random produces initial population Q (t=0), and t represents evolutionary generation here, and number individual in the population is more than 30 or 30, sets maximum evolutionary generation 〉=35;

4. to each individual its target function value that calculates: input is with the monomode fiber array and export with the monomode fiber array and enter the space specified position by each individual regulation with the direction attitude of appointment, and read optical power value with two or more waveguide channels of recorded and stored, this process is carried out successively by the individuality order, utilizes the described optical power value that records to calculate two target function value f of each individuality ₁And f ₂

5. two target function values according to each individuality calculate each individual Pareto order, and Pareto order individuals with same is calculated its crowding distance;

6. be better than the principle that the big individuality of crowding distance in two identical individualities of the big individuality of Pareto ordinal number, Pareto ordinal number is better than the little individuality of crowding distance according to the little individuality of Pareto ordinal number, finish evaluation all individualities in the generation population;

7. adopt intersection, variation and three steps of elimination of the last one to produce population of future generation, the said process iteration is carried out, and till arriving the maximum evolutionary generation of setting, finishes a waveguide-optical fibre automatic core regulating process.

A kind of based on the used automatic core-adjusting device of the waveguide-optical fibre automatic core regulating method of multi-objective evolutionary algorithm, it is characterized in that described device comprises the antidetonation platform, the fiber waveguide support, the accurate sextuple adjusting bracket of two covers, two drivers, two cover monitoring systems, lasing light emitter, light power meter and computer, chip of light waveguide is fixed on the described fiber waveguide support, this fiber waveguide support can be at x, y, z three dimensional realization manual adjustment, the accurate sextuple adjusting bracket of described two covers be respectively applied for fixing and adjust input with and output with the dimensional orientation of fiber array, and lay respectively at the both sides of described fiber waveguide support, described two cover monitoring systems lay respectively at the top and the dead ahead of fiber waveguide support, described fiber waveguide support, two accurate sextuple adjusting brackets of cover and two cover monitoring systems all are fixedly mounted on the described antidetonation platform.

The invention has the beneficial effects as follows: multi-objective evolutionary algorithm automatic core-adjusting method provided by the invention and core adjusting device thereof, improved the deficiency of existing automatic core-adjusting method from mechanism.Remarkable result of the present invention shows with conventional hereditary method automatic core-adjusting and compares, the machine operation number of times significantly reduces, and need not artificially to participate in the setting of target weight, significantly improve operability, improved the automaticity and the operating efficiency of realization fiber array-waveguide device-fiber array system low-loss rapid abutting joint coupling effectively.

Description of drawings

Fig. 1 is the analytic modell analytical model figure of the end face coupling of monomode fiber and the waveguide of single mode bar;

Fig. 2 is that 1 * 8 waveguide Splitter schemes with the coupling of docking of fiber array;

Fig. 3. for multi-objective evolutionary algorithm is transferred core numerical simulation process curve chart;

Fig. 4 is the insertion loss figure of 8 passages;

Fig. 5 is a multi-objective evolutionary algorithm automatic core-adjusting flow chart;

Fig. 6 is the structural framing figure of automatic core-adjusting system.

Embodiment

A kind of waveguide-optical fibre automatic core regulating method of multiple target algorithm is characterized in, the concrete steps of method are:

1. set up the mapping between waveguide-optical fibre automatic core regulating physical parameter and the multi-objective evolutionary algorithm model, individuality in certain relative space position and the corresponding multi-objective evolutionary algorithm model of direction between waveguide optical axis and the optical axis when wherein transferring core, population in the corresponding multi-objective evolutionary algorithm model of the set of above-mentioned relative space position and direction, chromosome in the corresponding multi-objective evolutionary algorithm model of solid axes parameter, first target function f in the corresponding multi-objective evolutionary algorithm model of luminous power sum of two or more waveguide channels that accent core process records ₁, second target function f in the corresponding multi-objective evolutionary algorithm model of difference of the maximum of the luminous power of two or more waveguide channels that accent core process records and minimum value ₂

2. the initial adjustment of fiber array-fiber waveguide-fiber array system is confirmed: adopt computer control, stepper motor driven mode to adjust input and use the monomode fiber array with monomode fiber array and output, make it respectively deviation of the alignment with fiber waveguide input port and output port in ± 50 mu m ranges;

3. the method with completely random produces initial population Q (t=0), and t represents evolutionary generation here, and number individual in the population is more than 30 or 30, sets maximum evolutionary generation 〉=35;

4. to each individual its target function value that calculates: input is with the monomode fiber array and export with the monomode fiber array and enter the space specified position by each individual regulation with the direction attitude of appointment, and read optical power value with two or more waveguide channels of recorded and stored, this process is carried out successively by the individuality order, utilizes the described optical power value that records to calculate two target function value f of each individuality ₁And f ₂

5. two target function values according to each individuality calculate each individual Pareto order, and Pareto order individuals with same is calculated its crowding distance;

6. be better than the principle that the big individuality of crowding distance in two identical individualities of the big individuality of Pareto ordinal number, Pareto ordinal number is better than the little individuality of crowding distance according to the little individuality of Pareto ordinal number, finish evaluation all individualities in the generation population;

7. adopt intersection, variation and three steps of elimination of the last one to produce population of future generation, the said process iteration is carried out, and till arriving the maximum evolutionary generation of setting, finishes a waveguide-optical fibre automatic core regulating process.

Solid axes parameter described in the described step 1 is space 11 dimensions, comprises 6 dimensions of 5 peacekeeping outputs one side of input one side except rotating around optical axis, and this 11 dimension coordinate variable is encoded with the real number array.

Any i individual 11 dimension coordinate variable U from the above mentioned, that encode with any i real number array described in the described step 1 _iConstitute.

First target function f described in the described step 1 ₁(U _i) and second target function f ₂(U _i) obtain by following two formulas respectively:

f ₁(U _i)＝-[P ₁(U _i)+P ₂(U _i)]

f ₂(U _i)＝|P ₁(U _i)-P ₂(U _i)|

In the formula, P ₁(U _i) and P ₂(U _i) be respectively to transfer the core position to be in by individual U _iDuring the dimensional orientation of regulation, the Output optical power that records by two sampling channels.

The Pareto order of the individuality described in the described step 5 is calculated by following rule:

For any i individual U _i, if with another individual U _{J ≠ i}There is following target function value relation

f ₁(U _i)＞f ₁(U _{J ≠ i}) and f ₂(U _i) 〉=f ₂(U _{J ≠ i})

So, individual U _iThe Pareto ordinal number increase by 1, otherwise the ordinal number increment is 0.By this rule, by with U _iAll individualities in addition compare one by one, and accumulative total obtains individual U _iPareto ordinal number value.

The crowding distance of the individuality described in the described step 5 is calculated by following rule:

To having the individuality of identical ordinal number in the population, the k under the identical ordinal number individual U _iPress

f ₁(U _m)＜f ₁(U _m+1) m＝1，2，L，k-1.

f ₂(U _n)＜f ₂(U _n+1) n＝1，2，L，k-1.

Order arrange with cambic m subscript and n subscript respectively, calculate

Again the corresponding respectively original i subscript arrangement that is returned to is arranged in arrangement of m subscript and n subscript, so each individual U _iThus a pair of d ₁(U _i) and d ₂(U _i), individual U _iCrowding distance d (U _i) obtain by following formula

d(U _i)＝d ₁(U _i)+d ₂(U _i)

Elimination of the last one described in the described step 7 is meant in the groups of individuals of rejecting Pareto ordinal number maximum, the individuality of crowding distance minimum.

Intersection described in the described step 7 is calculated by following rule:

Be better than the principle that the big individuality of crowding distance in two identical individualities of the big individuality of Pareto ordinal number, Pareto ordinal number is better than the little individuality of crowding distance according to the little individuality of Pareto ordinal number, from current size is to select m the individuality that the row of evaluation is preceding the population of M, by following calculating

${\mathrm{<figure-callout\; id="0"\; label="U"\; filenames="A20071003898800211.png,A20071003898800212.png"\; state="\{\{state\}\}">U</figure-callout>}}_0=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_i{U}_i$

Obtain a new individual U ₀, weight coefficient a in the formula _iSatisfy

$\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_i=1,-0.5\&le;a_i\&le;1.5$

Relation, new individual U ₀With transition population that size is M+1 of former M individual formation, calculate new individual U by above-mentioned 5 rule ₀Pareto order and crowding distance, eliminate position, end individuality by above-mentioned 7 rule after, the new population that to obtain a size again be M.

Variation described in the described step 7 is obtained by following rule:

Picked at random body one by one from the above-mentioned new population that obtains adopts real number value to change described real-valued coding that should individuality, increases near a new individuality the parent individuality thus, and constituting size once more is the transition population of M+1.Computation rule by the crowding distance of above-mentioned Pareto sum of ranks individuality obtains all individual Pareto order and crowding distances in this transition population, eliminate position, end individuality according to the elimination of the last one rule after, obtaining one by one, body quantity is the population of future generation of M.

The concrete accent core process of implementing is summarised as: adjust the monomode fiber array at two ends under the microimaging monitoring, make it to aim at substantially with the I/O waveguide of fiber waveguide device, start following program then as shown in Figure 5.

Based on the structure of the waveguide-optical fibre automatic core regulating device of multi-objective evolutionary algorithm as shown in Figure 6, comprise the antidetonation platform, the fiber waveguide support, the accurate sextuple adjusting bracket of two covers, two drivers, two cover monitoring systems, lasing light emitter, light power meter and computer, chip of light waveguide is fixed on the described fiber waveguide support, this fiber waveguide support can be at x, y, z three dimensional realization manual adjustment, the accurate sextuple adjusting bracket of described two covers be respectively applied for fixing and adjust input with and output with the dimensional orientation of fiber array, and lay respectively at the both sides of described fiber waveguide support, described two cover monitoring systems lay respectively at the top and the dead ahead of fiber waveguide support, described fiber waveguide support, two accurate sextuple adjusting brackets of cover and two cover monitoring systems all are fixedly mounted on the described antidetonation platform.

On the fiber waveguide support in the middle of chip of light waveguide is fixed in, this fiber waveguide support can be in x, y, z three dimensional realization manual adjustment.Two the sextuple adjusting brackets of precision in both sides place the left and right sides of fiber waveguide support respectively, are respectively applied for dimensional orientation fixing and adjustment input optical fibre array and output optical fibre array.Three-dimensional shifting axle of x, the y of accurate sextuple adjusting bracket, z and the θ that rotates around shifting axle _x, θ _y, θ _zThe Three dimensional rotation axle is by step motor drive.The minimum amount of movement of x, y, the three-dimensional shifting axle of z is 0.05 μ m, θ _xThe minimum rotation amount of rotation axis is 0.0030 degree, θ _yThe minimum rotation amount of rotation axis is 0.0032 degree, θ _zThe minimum rotation amount of rotation axis is 0.0025 degree.Two drivers are used to drive the stepping motor of two sextuple adjusting brackets of precision, and driver links to each other with computer by gpib interface, and the available software programming is controlled.CCD microscope and micropositioning stage, display and illumination cold light source that monitoring system is arranged along x, y orthogonal by two covers are formed, the microexamination when being used for just adjusting.Lasing light emitter is in order to provide stable work light wave, the output interface of lasing light emitter and fiber alignment.Light power meter is used to survey the Output optical power of output optical fibre, is furnished with two input interfaces, can test the power output that comes from two optical fiber simultaneously.Light power meter is equipped with gpib interface, is connected with computer to realize communication.Command system is taken on by computer, transfers the core operation according to software programming control, accepts light power meter in real time and surveys the optical power signals that obtains, and finishes differentiation, memory and feedback operation in real time.Systems soft ware is worked out with the VB language.

Embodiment:

1. the end face of single-core fiber-bar waveguide-single-core fiber system coupling multi-objective evolutionary algorithm automatic core-adjusting and result thereof:

Select the automatic core-adjusting of single-core fiber-bar waveguide-single-core fiber system, the test of investigation end face coupling loss, can get rid of the problem that the waveguide own loss brings, be convenient to reflect more objectively the performance of accent core system aspect hardware and software two.Because be single core coupling, uniformity index is desensitized.Fiber waveguide selects for use Japanese NHK company silica based to bury the waveguide of type single mode bar, and the waveguide core Cross section Design is of a size of 8 * 8 μ m ², refractive index contrast is 0.3%, the both ends of the surface plane lapping.Optical fiber is the plane lapping monomode fiber of SUMITOMO CHEMICAL electrician company.Be equipped with the index-matching fluid of trace between fiber end face that cleaning is dry and Waveguide end face, effectively suppress end face reflection.Initial adjustment is carried out under the x of orthogonal, the micro-supervision of y bidimensional, adopts mouse action, 0.1 μ m stepper drive, easily the core deviation of the alignment with optical fiber and fiber waveguide be controlled at ± 50 mu m ranges in, the spacing of Waveguide end face and fiber end face is controlled within the 4 μ m.Start the multi-objective evolutionary algorithm automatic core-adjusting program of step establishment as stated above this moment, accent core parallel carrying out in two ends finished back record power output, finishes and once transfers the time-consuming 3min of being not more than of core.After transferring core to finish, the output monomode fiber changes multimode fiber into, repeats above-mentioned accent core process.Because the core diameter of multimode fiber reaches 50 μ m, can be similar to and think that multimode fiber has all received the power output of single mode bar waveguide.The decibel difference of the measured value of multimode fiber and monomode fiber is the end face coupling loss of waveguide of single mode bar and monomode fiber.The automatic core-adjusting experiment number is 7 times, the results are shown in table 4.1542.4nm the coupling loss mean value on the wavelength is 0.1136dB, the maximum of single coupling loss is less than 0.13dB, and standard deviation is less than 0.02dB.Show that this system is realizing that aspect high efficiency, the low-loss optical fiber-waveguide automatic butt be highly effective.

The core result of the test is transferred in the end face coupling of table 4. optical fiber-waveguide-fibre system.

The actual measurement number of times	End face coupling loss (dB)
		1	0.1088
2	0.0804
		3	0.1257
4	0.1256
		5	0.1272
6	0.1297
		7	0.0979
The average coupling loss of end face	0.1136
		Standard deviation n-1	0.0187

2. the multi-objective evolutionary algorithm automatic core-adjusting and the result thereof of single fiber array-1 * 8 Waveguide branching coupler-8 fiber array:

Waveguide device is selected silica based 1 * 8 branch coupler of Japanese NHK company for use, and 8 core fibre arrays of the single-core fiber array of input and output select for use secondary rainbow to reach photoelectric device company product.Waveguide end face and fiber array end face are equipped with between end face and wait to solidify BP300 ultraviolet binding agent through plane lapping, and the core spacing nominal value of waveguide and fiber array is 250 μ m.After finishing initial adjustment under the micro-supervision, start multi-objective evolutionary algorithm automatic core-adjusting program by the above-mentioned steps establishment, operation wavelength is 1542.4nm, sample channel is the 2nd and the 7th passage.Repeat to transfer core test 7 times, single transfers the core time all less than 10min.Table 5 has been concluded measured data, and each passage inserts loss all less than 10.7dB, and uniformity index is less than 0.77dB.Show that the automatic core-adjusting system based on multi-objective evolutionary algorithm possesses the feature that realizes low-loss array accent core, good reproducibility at short notice.

The automatic core-adjusting experimental result of table 5. fiber array-1 * 8 Waveguide branching coupler-fiber array system.

The actual measurement number of times	Insert loss (dB)
											Core 1	Core 2	Core 3	Core 4	Core 5	Core 6	Core 7	Core 8	Mean value	max-min
	1	9.7973	9.9875	10.1281	9.9353	10.3947	10.4581	10.1191	10.3041	10.1453											0.6608
2											9.9053	9.9854	10.4215	10.3425	10.1493	10.2786	10.0231	10.3894	10.1867	0.5162
	3	9.8547	9.9733	10.3418	10.5892	10.2879	10.3483	10.0571	10.4018	10.2318											0.7345
4											9.9015	10.1067	10.6284	10.3951	10.1687	10.6619	9.9938	10.4872	10.2929	0.7604
	5	10.0694	9.9829	10.2164	10.3497	9.9548	9.7492	9.9128	10.2478	10.0604											0.4369
6											9.9815	9.9782	10.3491	10.1149	10.6891	10.2176	9.9279	10.3481	10.2008	0.7612
	7	9.7603	10.0223	10.5149	10.0189	10.4973	10.3168	9.9592	10.4952	10.1985											0.7546

Claims (12)

1. waveguide-the optical fibre automatic core regulating method of a multiple target algorithm is characterized in, the method concrete steps are:

(1) sets up mapping between waveguide-optical fibre automatic core regulating physical parameter and the multi-objective evolutionary algorithm model, individuality in certain relative space position and the corresponding multi-objective evolutionary algorithm model of direction between waveguide optical axis and the optical axis when wherein transferring core, population in the corresponding multi-objective evolutionary algorithm model of the set of above-mentioned relative space position and direction, chromosome in the corresponding multi-objective evolutionary algorithm model of solid axes parameter, first target function f in the corresponding multi-objective evolutionary algorithm model of luminous power sum of two or more waveguide channels that accent core process records ₁, second target function f in the corresponding multi-objective evolutionary algorithm model of difference of the maximum of the luminous power of two or more waveguide channels that accent core process records and minimum value ₂

(2) initial adjustment of fiber array-fiber waveguide-fiber array system is confirmed: adopt computer control, stepper motor driven mode to adjust input and use the monomode fiber array with monomode fiber array and output, make it respectively deviation of the alignment with fiber waveguide input port and output port in ± 50 mu m ranges;

(3) method with completely random produces initial population Q (t=0), and t represents evolutionary generation here, and number individual in the population is more than 30 or 30, sets maximum evolutionary generation 〉=35;

(4) to each individual its target function value that calculates: input is with the monomode fiber array and export with the monomode fiber array and enter the space specified position by each individual regulation with the direction attitude of appointment, and read optical power value with two or more waveguide channels of recorded and stored, this process is carried out successively by the individuality order, utilizes the described optical power value that records to calculate two target function value f of each individuality ₁And f ₂

(5) two target function values according to each individuality calculate each individual Pareto order, and Pareto order individuals with same is calculated its crowding distance;

(6) be better than the principle that the big individuality of crowding distance in two identical individualities of the big individuality of Pareto ordinal number, Pareto ordinal number is better than the little individuality of crowding distance according to the little individuality of Pareto ordinal number, finish evaluation all individualities in the generation population;

(7) adopt intersection, variation and three steps of elimination of the last one to produce population of future generation, the said process iteration is carried out, and till arriving the maximum evolutionary generation of setting, finishes a waveguide-optical fibre automatic core regulating process.

2. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1, it is characterized in that, solid axes parameter described in the described step (1) is space 11 dimensions, comprise 6 dimensions of 5 peacekeeping outputs one side of input one side except rotating around optical axis, this 11 dimension coordinate variable is encoded with the real number array.

3. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, any i the individual 11 dimension coordinate variable U from the above mentioned, that encode with any i real number array described in the described step (1) _iConstitute.

4. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, first target function f described in the described step (1) ₁(U _i) and second target function f ₂(U _i) obtain by following two formulas respectively:

f ₁(U _i)＝-[P ₁(U _i)+P ₂(U _i)]

f ₂(U _i)＝|P ₁(U _i)-P ₂(U _i)|

In the formula, P ₁(U _i) and P ₂(U _i) be respectively to transfer the core position to be in by individual U _iDuring the dimensional orientation of regulation, the Output optical power that records by two sampling channels.

5. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, the Pareto order of the individuality described in the described step (5) is calculated by following rule:

For any i individual U _i, if with another individual U _{J ≠ i}There is following target function value relation

f ₁(U _i)＞f ₁(U _{J ≠ i}) and f ₂(U _i) 〉=f ₂(U _{J ≠ i})

So, individual U _iThe Pareto ordinal number increase by 1, otherwise the ordinal number increment is 0.By this rule, by with U _iAll individualities in addition compare one by one, and accumulative total obtains individual U _iPareto ordinal number value.

6. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, the crowding distance of the individuality described in the described step (5) is calculated by following rule:

To having the individuality of identical ordinal number in the population, the k under the identical ordinal number individual U _iPress

f ₁(U _m)＜f ₁(U _m+1) m＝1，2，L，k-1.

f ₂(U _n)＜f ₂(U _N+1) n=1,2, L, the order of k-1. is arranged with cambic m subscript and n subscript respectively, calculates

Again the corresponding respectively original i subscript arrangement that is returned to is arranged in arrangement of m subscript and n subscript, so each individual U _iThus a pair of d ₁(U _i) and d ₂(U _i), individual U _iCrowding distance d (U _i) obtain by following formula

d(U _i)＝d ₁(U _i)+d ₂(U _i)。

7. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, the elimination of the last one described in the described step (7) is meant in the groups of individuals of rejecting Pareto ordinal number maximum, the individuality of crowding distance minimum.

8. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, the intersection described in the described step (7) is calculated by following rule:

Be better than the principle that the big individuality of crowding distance in two identical individualities of the big individuality of Pareto ordinal number, Pareto ordinal number is better than the little individuality of crowding distance according to the little individuality of Pareto ordinal number, from current size is to select m the individuality that the row of evaluation is preceding the population of M, by following calculating

$U_0=\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_i{U}_i$

Obtain a new individual U ₀, weight coefficient a in the formula _iSatisfy $\underset{i=1}{\overset{m}{\mathrm{\&Sigma;}}}{a}_i=1,$ -0.5≤a _i≤ 1.5 relation, new individual U ₀With transition population that size is M+1 of former M individual formation, calculate new individual U by above-mentioned 5 and 6 rule ₀Pareto order and crowding distance, eliminate position, end individuality by above-mentioned 7 rule after, the new population that to obtain a size again be M.

9. waveguide-the optical fibre automatic core regulating method of a kind of multiple target algorithm according to claim 1 is characterized in that, the variation described in the described step (7) is obtained by following rule:

Picked at random body one by one from the above-mentioned new population that obtains adopts real number value to change described real-valued coding that should individuality, increases near a new individuality the parent individuality thus, and constituting size once more is the transition population of M+1.Computation rule by the above-mentioned Pareto and the crowding distance of individuality obtains all individual Pareto order and crowding distances in this transition population, and behind the individuality of position, the superseded end of rule of elimination, obtaining one by one, body quantity is the population of future generation of M.

10. one kind based on the used automatic core-adjusting device of the waveguide-optical fibre automatic core regulating method of multi-objective evolutionary algorithm, it is characterized in that, described device comprises the antidetonation platform, the fiber waveguide support, the accurate sextuple adjusting bracket of two covers, two drivers, two cover monitoring systems, lasing light emitter, light power meter and computer, chip of light waveguide is fixed on the described fiber waveguide support, this fiber waveguide support can be at x, y, z three dimensional realization manual adjustment, the accurate sextuple adjusting bracket of described two covers be respectively applied for fixing and adjust input with and output with the dimensional orientation of fiber array, and lay respectively at the both sides of described fiber waveguide support, described two cover monitoring systems lay respectively at the top and the dead ahead of fiber waveguide support, described fiber waveguide support, two accurate sextuple adjusting brackets of cover and two cover monitoring systems all are fixedly mounted on the described antidetonation platform.

11. according to claim 10ly it is characterized in that the x of described sextuple adjusting bracket, y, the three-dimensional shifting axle of z and the θ that rotates around shifting axle based on the used automatic core-adjusting device of the waveguide-optical fibre automatic core regulating method of multi-objective evolutionary algorithm _x, θ _y, θ _zThe Three dimensional rotation axle is by step motor drive.The minimum amount of movement of x, y, the three-dimensional shifting axle of z is 0.05 μ m, θ _xThe minimum rotation amount of rotation axis is 0.0030 degree, θ _yThe minimum rotation amount of rotation axis is 0.0032 degree, θ _zThe minimum rotation amount of rotation axis is 0.0025 degree, and the stepping motor of two sextuple adjusting brackets is driven respectively by two drivers, and driver links to each other with computer by gpib interface.

12. it is according to claim 10 based on the used automatic core-adjusting device of the waveguide-optical fibre automatic core regulating method of multi-objective evolutionary algorithm, it is characterized in that, CCD microscope and micropositioning stage, display and illumination cold light source that described monitoring system is arranged along x, y orthogonal by two covers are formed, the microexamination when being used for just adjusting.

Images