CN101477224B - Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning - Google Patents

Abstract

The invention discloses a genetic programming-based reconstruction method for axial nonuniform strain of the Bragg gratings and belongs to the field of nonuniform strain reconstruction. The method comprises the following steps: acquiring a structural response signal, randomly generating a Bragg grating axial nonuniform strain distribution expression, calculating a simulated reflection spectrum of a Bragg grating, calculating a fitness function, optimizing the nonuniform strain distribution expression through the reproduction, intersection and variation operations of the genetic programming, and finally, repeating the last two steps till a preset maximum number of generations is reached. The method uses both a genetic programming algorithm and a modified T-array reflection spectrum formulation to reconstruct the grating axial nonuniform strain distribution expression, expresses a function expression in form of a binary tree without making any assumption concerning the grating axial strain distribution in any form in advance during the random generation of the strain distribution expression and optimizes any individual expression through the genetic manipulation of the binary tree. The method can accelerate convergence and improve calculation efficiency.

Description

Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning

Technical field

The present invention relates to a kind of heterogeneous strain reconstruction method, relate in particular to a kind of Bragg optical grating axial heterogeneous strain reconstruction method, belong to the heterogeneous strain reconstruction field based on genetic planning.

Background technology

In recent years, optical fiber Bragg raster (FBG) is as the important devices of optical fiber communication and Fibre Optical Sensor, and it has caused the great interest of people in the health monitoring Application for Field.When physical quantitys such as the stress of FBG environment of living in, strain, temperature change, can cause the variation of grating cycle or optical fiber effective refractive index, cause the reflectance spectrum shape of FBG to change,, just can obtain the situation of change of measured physical quantity by measuring the variation of reflectance spectrum shape.

Strain and strain gradient that reflectivity information reconstruct optical grating axial by grating reflection spectrum bears are very thorny engineering indirect problems.The solution of inverse problems method that some are commonly used, as gradient descent algorithm reconstruct Strain Distribution effectively, and efficient is not high; Yet correlative study shows, heuristic intelligent algorithm shows uniqueness in the finding the solution of indirect problem and efficient optimization is found the solution performance.This method is divided into uniform plurality of sections with grating, every section Strain Distribution is considered as constant, utilize the strain value of each grating section of genetic manipulation reconstruct such as selection, intersection and variation of genetic algorithm, the essence of its work is to approach continuous Strain Distribution with the Strain Distribution of Discrete Distribution, has guaranteed that algorithm has higher reconstruct speed.After this, intelligent algorithms such as simulated annealing, adaptive modeling annealing and simulated annealing evolution algorithm are used to the axial heterogeneous strain distribution reconstruct of FBG in succession, the basic thought of these work all is based on the measured length chromosomal inheritance and the evolution of gene expression, therefore the strain value of reconstruct optical grating axial piecemeal causes the partial loss of strain information easily; When the raster-segment number more also can cause the search volume excessive, influence the precision and the speed of strain identification; Also having a kind of method hypothesis Strain Distribution is the form of quadratic polynomial, with the undetermined coefficient in this polynomial expression of improved Simulated Anneal Algorithm Optimize, and then the heterogeneous strain that obtains whole optical grating axial distributes, this method is comparatively effective to strain reconstruction linear, quadratic distribution, but obviously is difficult to be applicable to the Strain Distribution reconstruct problem of complicated function forms such as sine with big strain gradient, high-order moment; Generally speaking, said method has all been done the hypothesis of certain form in advance to the Strain Distribution of optical grating axial, and the Strain Distribution that exists in the practical structures is an arbitrary form, without any prior imformation can be used to suppose the Strain Distribution form, therefore all there is the limitation of himself in existing method.

Last century, genetic planning (GP) algorithm that proposes of professor Koza and genetic algorithm maximum different of the nineties Stanford Univ USA were individual form differences, the individuality of genetic algorithm is the character string of a fixed length, and the individuality of GP is a function expression, uses tree construction nonlinear, random length to represent.At present genetic planning is used widely in the fields such as synthetic, Symbolic Regression, music and image generation of design automatically, pattern-recognition, robot control, neural network structure, and in the health monitoring field correlative study is not arranged as yet.

Summary of the invention

The present invention proposes a kind of Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning for the Strain Distribution expression formula of utilizing automatic design of genetic planning and optimization Bragg optical grating axial under existing conventional common apparatus condition.

A kind of Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning comprises the steps:

(1) gather the structural response signal:

Connect wideband light source to spectrometer, scanning survey light obtains the spectrum of incident light; Connect the end of broadband light to the Bragg grating, the Bragg grating other end connects spectrometer, scans the transmitted light of Bragg grating, obtains the spectrum of transmitted light; It is the transmission spectrum of unit that the spectrum of transmitted light deducts the dB that the spectrum of incident light obtains, and this transmission spectrum sampled point is converted into reflectivity, finally obtains Bragg grating reflection spectrum;

(2) generate Bragg optical grating axial heterogeneous strain distribution and expression formula at random:

The initial controlled variable of genetic planning is set, the Bragg optical grating axial heterogeneous strain distribution and expression formula population that utilizes genetic planning to generate at random to represent with the binary tree form, wherein: the tree structure of genetic planning is made up of the element among collection of functions F and the full stop collection T, and collection of functions F comprises sign of operation and mathematical function conditional expression; Full stop collection T comprises input state variable constant and does not have the ginseng function, initial population is made up of numerous individualities, each individuality all is to be generated by the combination of the arbitrary element random alignment among collection of functions F and the full stop collection T, after selecting root node, according to the variable number that is sent, determine the number of branches that grows, again from collection of functions F and full stop collection T and concentrate and to select the caudal knot point of an element as branch by equally distributed random device: if what select is element the collection of functions F, then repeat above-mentioned selection course; If what select is element among the full stop collection T, then this branch just stops growing, and has promptly generated body one by one after all branches all stop growing;

(3) simulated reflections of calculating the Bragg grating is composed:

Utilize improved T matrix method earlier, the Bragg grating is divided into the M equal portions, each part is a cross-talk grating;

A. calculate the grating cycle of Bragg grating any position behind the stand under load:

$\widetilde{\mathrm{\Λ}}\left(z\right)={\mathrm{\Λ}}_0\frac{{[1+(1-p_e)\mathrm{\ϵ}\left(z\right)]}^2}{1+(1-p_e)\mathrm{\ϵ}\left(z\right)-(1-p_e)z{\mathrm{\ϵ}}^{\′}\left(z\right)}$

Wherein: z is a Bragg optical grating axial coordinate, p _eBe elasto-optical coefficient, ε (z), ε ' (z) are respectively the strain and the strain gradient at Bragg optical grating axial coordinate z place, Λ ₀Intrinsic light grid cycle for the Bragg grating;

B. calculate direct current from coupled systemes number and AC coupling coefficient:

Direct current is from the coupled systemes number: $\widehat{\mathrm{\σ}}\left(z\right)=\frac{2\mathrm{\π}}{\mathrm{\λ}}(n_{\mathrm{eff}}+{\stackrel{\‾}{\mathrm{\δn}}}_{\mathrm{eff}})-\frac{\mathrm{\π}}{\widetilde{\mathrm{\Λ}}\left(z\right)}$

Wherein: n _EffBe effective refractive index, λ is a wavelength, Be the index modulation degree of depth;

AC coupling coefficient: $k=\frac{\mathrm{\π}}{\mathrm{\λ}}\mathrm{\υ}{\stackrel{\‾}{\mathrm{\δn}}}_{\mathrm{eff}},$ Wherein: υ is the fringe visibility of variations in refractive index;

C. the transport property of every section uniform grating is with corresponding transmission matrix F _iExpression:

$F_i=\left[\begin{array}{cc}{F}_i^1-F_i^2& -F_i^3\\ F_i^3& F_i^1+F_i^2\end{array}\right]$

Wherein: $F_i^1=\cosh \left({\mathrm{\γ}}_i\mathrm{\Δz}\right),$ $F_i^2=j\frac{{\widehat{\mathrm{\σ}}}_i}{{\mathrm{\γ}}_i}\sinh \left({\mathrm{\γ}}_i\mathrm{\Δz}\right),$ $F_i^3=j\frac{k}{{\mathrm{\γ}}_i}\sinh \left({\mathrm{\γ}}_i\mathrm{\Δz}\right),$ ${\mathrm{\γ}}_i=\sqrt{k^2-{\widehat{\mathrm{\σ}}}_i^2},$ j ²＝-1；

D. calculate the simulated reflections spectrum of Bragg grating:

Strain value ε by every cross-talk grating _iCalculate the transmission matrix F of every cross-talk grating _i, can draw the transport property of whole Bragg grating:

$\left[\begin{array}{c}{R}_M\\ S_M\end{array}\right]=F\left[\begin{array}{c}{R}_0\\ S_0\end{array}\right]$

Wherein: F=F ₁F ₂F _M, R _i, S _iBe respectively the forward direction and the amplitude of back of i section grating to transmission mode;

The Bragg grating hop count M of segmentation satisfies: $M=\frac{{2n}_{\mathrm{eff}}L}{{\mathrm{\λ}}_B},$ Wherein: λ _BBe the Bragg wavelength of grating, L is a Bragg fiber grating length;

Then the simulated reflections of Bragg grating is composed: $r\left(\mathrm{\λ}\right)={\left|\frac{S_M}{R_M}\right|}^2;$

(4) calculate fitness function:

Set up fitness function with the Euclidean distance of testing in the step (1) between the Strain Distribution expression formula corresponding simulating reflectance spectrum that the Bragg optical grating reflection is composed and step (3) obtains that records:

T _n＝‖r _n-r _o‖

Wherein: r _oBe the Bragg optical grating reflection spectrum that experiment records, r _nBe the reflectance spectrum of n heterogeneous strain distribution function expression formula correspondence in the population, T _nBe n individual fitness value, n is a sequence number individual in the population, and its value is to taking turns value between default population individual amount successively 1; (5) optimize heterogeneous strain distribution and expression formula by the duplicating of genetic planning, intersection and mutation operation:

A. dynamically adjust replication rate, crossing-over rate and aberration rate:

Each two individualities of picked at random select the high individuality of fitness as first individuality that needs genetic manipulation therein, dynamically adjust replication rate, crossing-over rate and aberration rate according to the fitness f of first individuality by following formula:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-(p_{c1}-p_{c2})(f-f_{\mathrm{avg}})/(f_{\max }-f_{\mathrm{avg}})& f\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f<f_{\mathrm{avg}}\end{array}\right.$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-(p_{m1}-p_{m2})(f_{\max }-f)/(f_{\max }-f_{\mathrm{avg}})& f\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f<f_{\mathrm{avg}}\end{array}\right.$

p _r＝1-p _c-p _m

Wherein: fitness f ∈ T _n, f _MaxBe current population maximum adaptation degree value, f _AvgBe the average fitness value of current population, p _rBe replication rate, p _cBe crossing-over rate, p _mBe aberration rate, p _C1Be the crossing-over rate upper limit, p _C2Be crossing-over rate lower limit, p _M1Be the aberration rate upper limit, p _M2Be the aberration rate lower limit;

B. carry out genetic manipulation:

Random number rand between producing one 0～1 is respectively according to the Probability p of duplicating, intersecting and making a variation _r, p _cAnd p _mSelect to determine the type of genetic manipulation: if rand ∈ (0, p _r], then first individuality is carried out replicate run; If rand ∈ (p _r, p _c], then adopt above-mentioned system of selection to select one second individuality again and carry out interlace operation with first individuality; If rand ∈ (p _c, p _m], then first individuality is carried out mutation operation, reach the population of new generation of default number up to generation;

(6) repeat step (4) and step (5), till reaching default maximum genetic algebra, the highest individuality of fitness value is the heterogeneous strain distribution and expression formula that will obtain in last colony in generation.

The present invention provides the heterogeneous strain reconstruction method of Bragg optical grating axial in a kind of engineering structure damage active monitoring.This method has been used Bragg grating sensing technique and genetic programming algorithm, comprehensively adopts the Strain Distribution expression formula of genetic programming algorithm and improved T matrix reflectance spectrum row formula reconstruct optical grating axial.The dynamic parameter method to set up that the present invention proposes: crossing-over rate p _cWith aberration rate p _mChange with fitness, when each individual fitness of population reaches unanimity or during local optimum, crossing-over rate p _cWith aberration rate p _mIncrease; And when colony's fitness relatively disperses, crossing-over rate p _cWith aberration rate p _mReduce, this can improve speed of convergence, avoids precocious.The inventive method is also taked to be provided with deeply more flexibly than existing dynamic tree, and its rule can guarantee to set when dark when best individuality tree is lower than current restriction deeply, and dynamic tree is provided with deeply that can to change into preferably individual so far tree during evolution automatically dark.This method does not need in advance the Strain Distribution of optical grating axial to be done any type of hypothesis, and the Strain Distribution of optimization is a continuous function, thereby has avoided segmentation optimization can only obtain the drawback of extreme position strain value; Raster-segment is just in order to calculate the reflectance spectrum of Strain Distribution correspondence, and then the convenient fitness value that calculates, and for population and individual evolution provide foundation, so what of raster-segment number can directly not have influence on the precision of Strain Distribution reconstruct.To sum up, the present invention has the rapid convergence of adding, improves counting yield, reliability height, precision height, can access the strain value of Bragg optical grating axial optional position.

Fig. 1 is the inventive method process flow diagram.

Description of drawings

Fig. 2 is a tree structure example schematic diagram among the present invention: what represent among the figure is 2x+ (3-y/5) tree structure.

Fig. 3 intersects example schematic diagram among the present invention: (a) the individual synoptic diagram of parent; (b) offspring individual synoptic diagram.

Fig. 4 is the example schematic diagram that makes a variation among the present invention: (a) synoptic diagram before the variation; (b) variation back synoptic diagram.

As shown in Figure 1, a kind of Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning comprises the steps:

Embodiment

(1) gather the structural response signal:

Connect wideband light source to spectrometer, scanning survey light obtains the spectrum of incident light; Connect the end of broadband light to the Bragg grating, the Bragg grating other end connects spectrometer, scans the transmitted light of Bragg grating, obtains the spectrum of transmitted light; It is the transmission spectrum of unit that the spectrum of transmitted light deducts the dB that the spectrum of incident light obtains, and this transmission spectrum sampled point is converted into reflectivity, finally obtains Bragg grating reflection spectrum;

(2) generate Bragg optical grating axial heterogeneous strain distribution and expression formula at random:

The initial controlled variable of genetic planning is set, comprise that initial maximal tree is set is dark and maximal tree is dark, the Bragg optical grating axial heterogeneous strain distribution and expression formula population that utilizes genetic planning to generate at random to represent with the binary tree form, wherein: the tree structure of genetic planning is made up of the element among collection of functions F and the full stop collection T, tree structure as shown in Figure 2, collection of functions F comprises sign of operation and mathematical function conditional expression; Full stop collection T comprises input state variable constant and does not have the ginseng function, initial population is made up of numerous individualities, each individuality all is to be generated by the combination of the arbitrary element random alignment among collection of functions F and the full stop collection T, after selecting root node, according to the variable number that is sent, determine the number of branches that grows, again from collection of functions F and full stop collection T and concentrate and to select the caudal knot point of an element as branch by equally distributed random device: if what select is element the collection of functions F, then repeat above-mentioned selection course; If what select is element among the full stop collection T, then this branch just stops growing, and has promptly generated body one by one after all branches all stop growing;

(3) simulated reflections of calculating the Bragg grating is composed:

Utilize improved T matrix method earlier, the Bragg grating is divided into the M equal portions, each part is a cross-talk grating;

A. calculate the grating cycle of Bragg grating any position behind the stand under load:

$\widetilde{\mathrm{\&Lambda;}}\left(z\right)={\mathrm{\&Lambda;}}_0\frac{{[1+(1-p_e)\mathrm{\&epsiv;}\left(z\right)]}^2}{1+(1-p_e)\mathrm{\&epsiv;}\left(z\right)-(1-p_e)z{\mathrm{\&epsiv;}}^{\&prime;}\left(z\right)}$

Wherein: z is a Bragg optical grating axial coordinate, p _eBe elasto-optical coefficient, ε (z), ε ' (z) are respectively the strain and the strain gradient at Bragg optical grating axial coordinate z place, Λ ₀Intrinsic light grid cycle for the Bragg grating;

B. calculate direct current from coupled systemes number and AC coupling coefficient:

Direct current is from the coupled systemes number: $\widehat{\mathrm{\&sigma;}}\left(z\right)=\frac{2\mathrm{\&pi;}}{\mathrm{\&lambda;}}(n_{\mathrm{eff}}+{\stackrel{\&OverBar;}{\mathrm{\&delta;n}}}_{\mathrm{eff}})-\frac{\mathrm{\&pi;}}{\widetilde{\mathrm{\&Lambda;}}\left(z\right)}$

Wherein: n _EffBe effective refractive index, λ is a wavelength,

Be the index modulation degree of depth;

AC coupling coefficient: $k=\frac{\mathrm{\&pi;}}{\mathrm{\&lambda;}}\mathrm{\&upsi;}{\stackrel{\&OverBar;}{\mathrm{\&delta;n}}}_{\mathrm{eff}},$ Wherein: υ is the fringe visibility of variations in refractive index;

C. the transport property of every section uniform grating is with corresponding transmission matrix F _iExpression:

$F_i=\left[\begin{array}{cc}{F}_i^1-F_i^2& -F_i^3\\ F_i^3& F_i^1+{\mathrm{<figure-callout\; id="2"\; label="F\; i"\; filenames="H2009100284572E00012.png,H2009100284572E00021.png"\; state="\{\{state\}\}">F</figure-callout>}}_i^{}\end{array}\right]$

Wherein: $F_i^1=\cosh \left({\mathrm{\&gamma;}}_i\mathrm{\&Delta;z}\right),$ ${\mathrm{<figure-callout\; id="2"\; label="F\; i"\; filenames="H2009100284572E00012.png,H2009100284572E00021.png"\; state="\{\{state\}\}">F</figure-callout>}}_i^{}=j\frac{{\widehat{\mathrm{\&sigma;}}}_i}{{\mathrm{\&gamma;}}_i}\sinh \left({\mathrm{\&gamma;}}_i\mathrm{\&Delta;z}\right),$ ${\mathrm{<figure-callout\; id="3"\; label="F\; i"\; filenames="H2009100284572E00012.png,H2009100284572E00021.png"\; state="\{\{state\}\}">F</figure-callout>}}_i^{}=j\frac{k}{{\mathrm{\&gamma;}}_i}\sinh \left({\mathrm{\&gamma;}}_i\mathrm{\&Delta;z}\right),$ ${\mathrm{\&gamma;}}_i=\sqrt{k^2-{\widehat{\mathrm{\&sigma;}}}_{\mathrm{<figure-callout\; id="2"\; label="i"\; filenames="H2009100284572E00012.png,H2009100284572E00021.png"\; state="\{\{state\}\}">i</figure-callout>}}^2},$ j ²＝-1；

D. calculate the simulated reflections spectrum of Bragg grating:

Strain value ε by every cross-talk grating _iCalculate the transmission matrix F of every cross-talk grating _i, can draw the transport property of whole Bragg grating:

$\left[\begin{array}{c}{R}_M\\ S_M\end{array}\right]=F\left[\begin{array}{c}{R}_0\\ S_0\end{array}\right]$

Wherein: F=F ₁F ₂F _M, R _i, S _iBe respectively the forward direction and the amplitude of back of i section grating to transmission mode;

The Bragg grating hop count M of segmentation satisfies: $M=\frac{{2n}_{\mathrm{eff}}L}{{\mathrm{\&lambda;}}_B},$ Wherein: λ _BBe the Bragg wavelength of grating, L is a Bragg fiber grating length;

Then the simulated reflections of Bragg grating is composed: $r\left(\mathrm{\&lambda;}\right)={\left|\frac{S_M}{R_M}\right|}^2;$

(4) calculate fitness function:

Set up fitness function with the Euclidean distance of testing in the step (1) between the Strain Distribution expression formula corresponding simulating reflectance spectrum that the Bragg optical grating reflection is composed and step (3) obtains that records:

T _n＝‖r _n-r _o‖

Wherein: r _oBe the Bragg optical grating reflection spectrum that experiment records, r _nBe the reflectance spectrum of n heterogeneous strain distribution function expression formula correspondence in the population, T _nBe n individual fitness value, n is a sequence number individual in the population, and its value is to taking turns value between default population individual amount successively 1;

(5) optimize heterogeneous strain distribution and expression formula by the duplicating of genetic planning, intersection and mutation operation:

A. dynamically adjust replication rate, crossing-over rate and aberration rate:

Each two individualities of picked at random select the high individuality of fitness as first individuality that needs genetic manipulation therein, dynamically adjust replication rate, crossing-over rate and aberration rate according to the fitness f of first individuality by following formula:

$p_c=\left\{\begin{array}{cc}{p}_{c1}-(p_{c1}-p_{c2})(f-f_{\mathrm{avg}})/(f_{\max }-f_{\mathrm{avg}})& f\&GreaterEqual;f_{\mathrm{avg}}\\ p_{c1}& f<f_{\mathrm{avg}}\end{array}\right.$

$p_m=\left\{\begin{array}{cc}{p}_{m1}-(p_{m1}-p_{m2})(f_{\max }-f)/(f_{\max }-f_{\mathrm{avg}})& f\&GreaterEqual;f_{\mathrm{avg}}\\ p_{m1}& f<f_{\mathrm{avg}}\end{array}\right.$

p _r＝1-p _c-p _m

Wherein: fitness f ∈ T _n, f _MaxBe current population maximum adaptation degree value, f _AvgBe the average fitness value of current population, p _rBe replication rate, p _cBe crossing-over rate, p _mBe aberration rate, p _C1Be the crossing-over rate upper limit, p _C2Be crossing-over rate lower limit, p _M1Be the aberration rate upper limit, p _M2Be the aberration rate lower limit;

B. carry out genetic manipulation:

Random number rand between producing one 0～1 is respectively according to the Probability p of duplicating, intersecting and making a variation _r, p _cAnd p _mSelect to determine the type of genetic manipulation: if rand ∈ (0, p _r], then first individuality is carried out replicate run; If rand ∈ (p _r, p _c], then adopt above-mentioned system of selection to select one second individuality again and carry out interlace operation with first individuality; If rand ∈ (p _c, p _m], then first individuality is carried out mutation operation, reach the population of new generation of default number up to generation; It is dark that dynamic constraints tree is set in this step, it be intersect, the new tree that produces is no more than the dark individuality of dynamic constraints tree deeply and is kept behind the mutation operation, surpassing dark and its fitness of dynamic constraints tree is not that the highest individuality is directly eliminated; When the individuality of new generation is that the highest individual and its tree of fitness is lower than maximal tree when dark deeply, dynamic tree is dark to be provided with that to change into preferably individual so far tree during evolution automatically dark;

Wherein: duplicate: the high parent individuality of selecting of fitness is not copied in the colony of future generation with not adding conversion;

Intersect: as shown in Figure 3, the node of two individualities of picked at random, the subtree that will link to each other with node exchanges, and obtains two new individualities;

Variation: as shown in Figure 4, the node in certain individuality of picked at random is replaced this subtree below node with the subtree that produces at random;

(6) repeat step (4) and step (5), till reaching default maximum genetic algebra, the highest individuality of fitness value is the heterogeneous strain distribution and expression formula that will obtain in last colony in generation.

Claims (2)

1. the Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning is characterized in that comprising the steps:

(1) gather the structural response signal:

Connect wideband light source to spectrometer, scanning survey light obtains the spectrum of incident light; Connect the end of broadband light to the Bragg grating, the Bragg grating other end connects spectrometer, scans the transmitted light of Bragg grating, obtains the spectrum of transmitted light; It is the transmission spectrum of unit that the spectrum of transmitted light deducts the dB that the spectrum of incident light obtains, and this transmission spectrum sampled point is converted into reflectivity, finally obtains Bragg grating reflection spectrum;

(2) generate Bragg optical grating axial heterogeneous strain distribution and expression formula at random:

The initial controlled variable of genetic planning is set, the Bragg optical grating axial heterogeneous strain distribution and expression formula population that utilizes genetic planning to generate at random to represent with the binary tree form, wherein: the tree structure of genetic planning is made up of the element among collection of functions F and the full stop collection T, and collection of functions F comprises sign of operation and mathematical function conditional expression; Full stop collection T comprises input state variable constant and does not have the ginseng function, initial population is made up of numerous individualities, each individuality all is to be generated by the combination of the arbitrary element random alignment among collection of functions F and the full stop collection T, after selecting root node, according to the variable number that is sent, determine the number of branches that grows, again from collection of functions F and full stop collection T and concentrate and to select the caudal knot point of an element as branch by equally distributed random device: if what select is element the collection of functions F, then repeat above-mentioned selection course; If what select is element among the full stop collection T, then this branch just stops growing, and has promptly generated body one by one after all branches all stop growing;

(3) simulated reflections of calculating the Bragg grating is composed:

Utilize improved T matrix method earlier, the Bragg grating is divided into the M equal portions, each part is a cross-talk grating;

A. calculate the grating cycle of Bragg grating any position behind the stand under load:

Wherein: z is a Bragg optical grating axial coordinate, p _eBe elasto-optical coefficient, ε (z), ε ' (z) are respectively the strain and the strain gradient at Bragg optical grating axial coordinate z place, Λ ₀Intrinsic light grid cycle for the Bragg grating;

B. calculate direct current from coupled systemes number and AC coupling coefficient:

Direct current is from the coupled systemes number:

Wherein: n _EffBe effective refractive index, λ is a wavelength,

Be the index modulation degree of depth; AC coupling coefficient:

Wherein: υ is the fringe visibility of variations in refractive index;

C. the transport property of every section uniform grating is with corresponding transmission matrix F _iExpression:

Wherein:

j ²=-1;

D. calculate the simulated reflections spectrum of Bragg grating:

Strain value ε by every cross-talk grating _iCalculate the transmission matrix F of every cross-talk grating _i, can draw the transport property of whole Bragg grating:

Wherein: F=F ₁F ₂... F _M, R _i, S _iBe respectively the forward direction and the amplitude of back of i section grating to transmission mode;

The Bragg grating hop count M of segmentation satisfies:

Wherein: λ _BBe the Bragg wavelength of grating, L is a Bragg fiber grating length;

Then the simulated reflections of Bragg grating is composed:

(4) calculate fitness function:

Set up fitness function with the Euclidean distance of testing in the step (1) between the Strain Distribution expression formula corresponding simulating reflectance spectrum that the Bragg optical grating reflection is composed and step (3) obtains that records:

T _n＝||r _n-r _o||

Wherein: r _oBe the Bragg optical grating reflection spectrum that experiment records, r _nBe the reflectance spectrum of n heterogeneous strain distribution function expression formula correspondence in the population, T _nBe n individual fitness value, n is a sequence number individual in the population, and its value is to taking turns value between default population individual amount successively 1;

(5) optimize heterogeneous strain distribution and expression formula by the duplicating of genetic planning, intersection and mutation operation:

A. dynamically adjust replication rate, crossing-over rate and aberration rate:

Each two individualities of picked at random select the high individuality of fitness as first individuality that needs genetic manipulation therein, dynamically adjust replication rate, crossing-over rate and aberration rate according to the fitness f of first individuality by following formula:

p _r＝1-p _c-p _m

Wherein: fitness f ∈ T _n, f _MaxBe current population maximum adaptation degree value, f _AvgBe the average fitness value of current population, p _rBe replication rate, p _cBe crossing-over rate, p _mBe aberration rate, p _C1Be the crossing-over rate upper limit, p _C2Be crossing-over rate lower limit, p _M1Be the aberration rate upper limit, p _M2Be the aberration rate lower limit;

B. carry out genetic manipulation:

Random number rand between producing one 0～1 is respectively according to the Probability p of duplicating, intersecting and making a variation _r, p _cAnd p _mSelect to determine the type of genetic manipulation: if rand ∈ (0, p _r], then first individuality is carried out replicate run; If rand ∈ (p _r, p _c], then adopt above-mentioned system of selection to select one second individuality again and carry out interlace operation with first individuality; If rand ∈ (p _c, p _m], then first individuality is carried out mutation operation, reach the population of new generation of default number up to generation;

(6) repeat step (4) and step (5), till reaching default maximum genetic algebra, the highest individuality of fitness value is the heterogeneous strain distribution and expression formula that will obtain in last colony in generation.

2. the Bragg optical grating axial heterogeneous strain reconstruction method based on genetic planning according to claim 1, it is dark to it is characterized in that in the b execution genetic manipulation dynamic constraints tree being set in the step (5), it be intersect, the new tree that produces is no more than the dark individuality of dynamic constraints tree deeply and is kept behind the mutation operation, surpassing dark and its fitness of dynamic constraints tree is not that the highest individuality is directly eliminated; When the individuality of new generation is that the highest individual and its tree of fitness is lower than maximal tree when dark deeply, dynamic tree is dark to be provided with that to change into preferably individual so far tree during evolution automatically dark.

Images