CN102707434B - In-cavity self-adaptive optical beam purification system and method - Google Patents

Abstract

The invention relates to an in-cavity self-adaptive optical beam purification system and a method. The self-adaptive optical beam purification system consists of a parallel light source, a deformable mirror, a Hartmann sensor, a data acquisition and control computer, a high-voltage amplifier, a spectroscope, a focusing lens, a beam quality diagnosis camera and a laser device. During beam purification, the prior surface shape of the deformable mirror is preset at first via the Hartmann sensor, and then the beam is purified in a random optimization control method. By adopting the method, prior data can be utilized effectively, the time for purification of the beam in the cavity can be greatly reduced, and the control stability of the in-cavity beam purification system can be improved.

Description

A kind of in-chamber adaptive optical light beam cleaning system and method

Technical field

The present invention relates to a kind of in-chamber adaptive optical light beam cleaning system and method, be applicable to distortion in chamber taking repeated Static Shift as main, the little resonator cavity of dynamic disturbances distortion absolute value carries out chamber inner light beam purification, belongs to adaptive optics field and field of lasers.

Background technology

In-chamber adaptive optical light beam cleaning system is distortion in a kind of real-time follow-up compensation laserresonator chamber, improves laser works performance, makes a kind of ADAPTIVE OPTICS SYSTEMS of Laser Output Beam energy or beam quality optimum.Since ADAPTIVE OPTICS SYSTEMS is successfully applied to CO first ₂in laser chamber, the compensation of distortion is (referring to Adaptive laser resonator such as R.H.Freeman, 1978, Optics Letters), source---the laserresonator producing from laser, improve near-field beam and far-field distribution form, just aspect the laser instrument such as gas, solid, obtained broad research and application.

In in-chamber adaptive optical light beam cleaning system, can actively change face shape and play the distorting lens that compensates distortion effect, as a high reflective cavity mirror of laserresonator.Different from the control device of traditional adaptive optics based on phase conjugation principle, due to not relation one to one of the distortion correction amount on wavefront distortion and distorting lens in resonator cavity, the quality that in chamber, the disturbing influence of distortion and distoring mirror shape Laser Output Beam with export energy.Therefore the control method of in-chamber adaptive optical light beam cleaning system and control theory are study hotspots always.

At present, the control method of in-chamber adaptive beam cleanup system mainly contains two kinds, taking far field beam quality as optimization aim, without an optimized-type adaptive optics control method for Wavefront sensor, another is taking distortion in chamber as the Phase conjugate control method of correction target.

A photoelectric detector for optimized-type adaptive optics control method (as devices such as CCD, CMOS and photodiodes) detecting laser output intensity distributes, and using from the beam quality of light distribution calculating as optimization aim, optimize the control signal of distorting lens driver according to random optimization control algolithm.Simple with its control system, be easy to the reasons such as adjustment, this method application is more extensive, if the genetic algorithm optimization distorting lens actuator voltage of W.Lubeigt employing in 2004 is (referring to Intracavity adaptive optics optimisation of a grazing-incidence Nd:GdVO4 laser such as W.Lubeigt, 2004OSA/CLE0), the people such as Ping Yang in 2007 adopt the mode coefficient of genetic algorithm optimization distorting lens, (referring to people Intracavity transverse modes controlled by a genetic algorithm based on Zernike mode coefficients such as Ping Yang, 2007, Optics Express) etc. optimized-type adaptive optics control algolithm.Except the control algolithm of above introduction, in in-chamber adaptive optical light beam cleaning system, be also shown in the report of the random optimization such as random paralleling gradient descent algorithm, simulated annealing control algolithm.

In-chamber adaptive optical control method based on Phase conjugate control technology, it is the light wavefront information that distorts or be associated with distortion in chamber in chamber by measuring, then according to wavefront distortion in phase conjugation principle compensated cavity, realize the object of improving laser works state.Patent " utilize in-chamber adaptive optical technology to improve the device of solid state laser beam quality " (Chinese Patent Application No.: 200610011199.3) introduced a kind of with He-Ne light as beacon beam, survey gain media distortion with Hartmann sensor, utilize the adaptive optics of distortion in distorting lens compensated cavity.

Optimized-type adaptive optics control system adopts system optimizing control, objective function using the laser performance index be concerned about as optimized algorithm, using the required control signal of distorting lens as Optimal Parameters, in parameter space, search for optimization solution with iterative manner since an initial value.In the middle of in-chamber adaptive optical light beam cleaning system, the initial value of algorithm search has a significant impact the stability of convergence, speed of convergence and convergence.

For distorting taking Static Shift as main in chamber, and distortion time stability is high, under the little situation of dynamic disturbances distortion amplitude, after the inner light beam cleaning system convergence of chamber, distoring mirror shape can be thought to be superimposed with the disturbance compared with small magnitude near a static face shape, because the Static Shift in chamber has good repeatability, so, while doing chamber inner light beam purification closed-loop control at every turn, the static face shape of distorting lens is basically identical, therefore without start search from distorting lens driver no-voltage as conventional search methods, but can be when last closed loop finishes magnitude of voltage.This method has extraordinary linear dependence at distorting lens driver path increment with driving signal, and minute surface is not set up under the prerequisite with the larger variation of environment temperature generation.But, for some distorting lens as bimorph deformable mirror, due to creep effect and the poor reason of face shape thermal stability of driver, preset voltage can not obtain required distoring mirror shape, the initial value of algorithm search is no longer to start from optimum face shape, therefore magnitude of voltage when preset last closed loop finishes, there is larger uncertainty in laser system duty.

Summary of the invention

Technology of the present invention is dealt with problems and is: overcome the weak point of existing random optimization control method from zero initial value search and preset voltage method, one preset distorting lens priori face shape is rapidly and accurately provided, for providing the chamber inner light beam of an optimum initial value, random optimization control algolithm purifies control method, the method can effectively be utilized priori data, make chamber inner light beam purify shortening greatly consuming time, improve the control degree of stability of chamber inner light beam cleaning system.

Technical solution of the present invention is: a kind of in-chamber adaptive optical light beam cleaning system, as shown in Figure 1, comprise: source of parallel light 1, distorting lens 2, Hartmann sensor 3, data acquisition and control computing machine 4, high-voltage amplifier 5, spectroscope 6, condenser lens 7, beam quality diagnosis camera 8 and laser instrument (9), described distorting lens 2 is as the rear mirror of laser instrument 9 resonator cavitys; The collimation parallel beam that source of parallel light 1 is sent, the parallel beam that incident distorting lens 2, and source of parallel light at a certain angle 1 sends covers the hot spot of laser instrument 9 excited radiation lights on distorting lens 2 minute surfaces; Parallel beam after distorting lens 2 reflections is received by Hartmann sensor 3; The light beam that laser instrument 9 sends is through spectroscope 6 beam splitting, and a branch of is as Laser output; Another bundle focuses into into beam quality diagnosis camera 8 through condenser lens 7, and data acquisition receives the signal of Hartmann sensor 3 and beam quality diagnosis camera 8 and is connected with high-voltage amplifier 5 with control computing machine 4; While carrying out beam cleanup, data acquisition utilizes Hartmann sensor 3 preset priori face shape on distorting lens 2 with control computing machine 4, then the beam quality of measuring according to beam quality diagnosis camera 8, utilize the controlled output voltage of random optimization control method, thereby play the object of optimizing laser instrument 9 output beam qualities.

Described random optimization control algolithm adopts random paralleling gradient descent algorithm, genetic algorithm, climbing method, ant group algorithm or simulated annealing.

A kind of in-chamber adaptive optical light beam purification method, performing step is as follows:

The first step, demarcates Hartmann sensor 3, and before laser instrument 9 starts, data acquisition gathers with controlling computing machine 4 the light spot image data that multiframe Hartmann sensor 3 is exported, and hot spot data is averaging to the initial calibration data of rear conduct, according to formula

calculate the initial centroid position coordinate of the hot spot [X in sub-aperture _c0, Y _c0], and construct the hot spot initial slope matrix G that each facula deviation amount corresponding to sub-aperture forms ₀; Wherein I _ii the signal intensity that pixel-by-pixel basis is received in certain sub-aperture in Hartmann sensor 3, X _iand Y _iit is respectively the coordinate of i pixel;

Second step, experiment measuring distorting lens 2 voltages-slope response matrix R _xy, successively the J of distorting lens 2 driver applied to unit voltage and utilize Hartmann sensor 3 to record hot spot side-play amount matrix simultaneously, voltage-slope response matrix R of structural deformation mirror 2 _xy, and solve R _xygeneralized inverse matrix

The 3rd step, preset distorting lens priori face shape, facula deviation moment matrix corresponding to distorting lens 2 priori face shape is G, while carrying out beam cleanup first, G=G ₀, the voltage vector that distorting lens 2 is applied is $V=R_{\mathrm{xy}}^{+}(G-G_0);$

The 4th step, adopt random optimization control method to optimize far field beam quality, the light distribution data that data acquisition and control computing machine 4 receiving beam quality diagnosis cameras 8 collect, calculate beam quality, according to the control voltage of random optimization control method real-time update distorting lens 2 each drivers, follow the tracks of dynamical distortion in compensation distorting lens 2 chambeies, make laser instrument 9 obtain diffraction limited beam output;

The 5th step, upgrade facula deviation moment matrix G, when the 4th step is optimized laser instrument 9 far field beam qualities, the Beam Wave-Front slope that data acquisition and control computing machine 4 real time record Hartmann sensors 3 are measured, hot spot side-play amount matrix G ' while recording far field beam quality optimum, before this chamber inner light beam purifies and finishes, make G=G ' complete the renewal of facula deviation moment matrix, when chamber inner light beam purifies again, start to carry out from " the 3rd step ".

The present invention's advantage is compared with prior art:

(1) the present invention adopts the facula mass center deviation data of Hartmann sensor output to weigh the relative variation of distoring mirror shape, and without the face shape of accurately trying to achieve distorting lens minute surface, this apparatus structure is simple, and anti-environmental interference ability is strong, less demanding to control system computing velocity.

(2) the present invention carries out the priori face shape of preset distorting lens optimum by Hartmann sensor, overcome the weak point of existing random optimization control method from zero initial value search and preset voltage method, the one optimum priori face of preset distorting lens shape is rapidly and accurately provided, for providing the chamber inner light beam of an optimum initial value, random optimization control algolithm purifies control method, the method can effectively be utilized priori data, make chamber inner light beam purify shortening greatly consuming time, improve the degree of stability that chamber inner light beam purifies.

Brief description of the drawings

Fig. 1 light path schematic diagram of the present invention;

Fig. 2 embodiment of the present invention light path schematic diagram;

Position of components schematic diagram in resonator to be clean in Fig. 3 embodiment of the present invention;

Basic mode hot spot and survey facula position figure on distorting lens driver distribution plan and minute surface in Fig. 4 the present invention.

Embodiment

As shown in Figure 2, the light path schematic diagram of a kind of embodiment of in-chamber adaptive optical light beam cleaning system, the source of parallel light 1 being formed by 650nm semiconductor laser 1.1, microcobjective 1.2, pin hole 1.3, collimation lens 1.4, distorting lens 2, the Hartmann sensor 3 that contracting bundle device 3.1, microlens array 3.2, camera 3.3 form, data acquisition forms with control computing machine 4, high-voltage amplifier 5, spectroscope 6, condenser lens 7, beam quality diagnosis camera 8, laser instrument 9.

The source of parallel light that 650nm semiconductor laser 1.1, microcobjective 1.2, pin hole 1.3 and collimation lens 1.4 form, sends the uniform parallel beam of light intensity with 22 ° of incident distorting lenss 2.Contracting bundle device 3.1 is positioned at distorting lens approximately 300 millimeters of places afterwards, bore is 50 millimeters, coaxial with the 650nm parallel beam after distorting lens reflection, play 30 millimeters of detecting light beams contracting bundle to the bore matching with microlens array 3.2, make light beam after contracting is restrainted enter microlens array 3.2 completely and do not exceed the target surface of microlens array 3.2 and camera 3.3.Camera 3.3 is connected with control computing machine 4 with data acquisition by capture card.

The composition of laser instrument 9 as shown in Figure 3, is followed successively by 2 two lens of 37 unit bimorph deformable mirror, aperture, Nd:YAG gain media, output coupling mirror from left to right.Distorting lens 2 adopts 37 unit bimorph deformable mirrors of Photoelectric Technology Inst., Chinese Academy of Sciences's development, the drive electrode of distorting lens 2 distributes as shown in Figure 4, after the state adjustment of laser instrument 9 and directional light completes, hot spot on distorting lens 2 minute surfaces distributes as shown in Figure 4, the directional light of 650nm is due to certain angle incident distorting lens 2, ovalize on minute surface (solid black lines ellipse), and cover the hot spot (black dotted lines ellipse) of laser instrument 9 basic mode light beams on distorting lens completely.

Light beam that laser instrument 9 sends is divided into two parts through spectroscope 6, and a branch of for the high power portion that sees through is as the output of laser instrument, ingoing power meter, carries out the measurement of power; A branch of partial low-power for reflection has identical light distribution and PHASE DISTRIBUTION with high-power light beam, therefore can weigh by low power optical beam the beam quality of high power-beam.In order effectively to utilize the space of experiment optical table, with a high reflective mirror 11, the light beam of partial low-power being turned back, to incide focal length be in the condenser lens 7 of 300 millimeters, beam quality diagnosis camera 8 is positioned on the focal plane of condenser lens 7, and beam quality diagnosis camera 8 is MV-D1024E CMOS cameras that photon focus company produces.

Data acquisition with control the mainboard of computing machine 4 on have the data collecting card being connected with camera 3.3 and beam quality diagnosis camera 8 respectively, receive the data of camera 3.3 and beam quality diagnosis camera 8; Data acquisition turns with the digital-to-analogue that the D/A card of controlling computing machine 4 use 3 Zhang16 roads is controlled voltage, is connected with high-voltage amplifier 5.High-voltage amplifier 5 is connected with distorting lens 2 by distorting lens data line after the analog voltage signal receiving is amplified.

The face shape of distorting lens 2 is with to incide 650nm directional light wavefront distortion before microlens array 3.2 after distorting lens 2 reflection corresponding one by one, also with through microlens array 3.2 light splitting with to focus on facula deviation on camera 3.3 target surfaces corresponding, therefore can represent by the side-play amount of spot array on camera 3.3 the relative variation of distorting lens 2 mirror shapes.The relative variation of facula deviation amount needs a contrast benchmark, the primary face shape of selecting distorting lens in experiment is as with reference to benchmark, the Hartmann sensor 3 that microlens array 3.2 and camera 3.3 are formed is demarcated, to determine the initial facula deviation moment matrix (also referred to as initial calibration data) of distorting lens.Not making alive of the driver of distorting lens 2 while carrying out initial alignment, it on camera target surface, is the hot spot dot matrix being evenly distributed, camera 3.3 target surfaces are divided into multiple little square region by the straight line of level and numerical value, and each square region is corresponding with a sub-lens of microlens array 3.2; Then, gather multiframe light spot image data,

Be averaging as initial calibration data, then according to formula

calculate every facula mass center position [X in sub-aperture _c0, Y _c0] and each the hot spot initial offset moment matrix G that facula mass center side-play amount corresponding to sub-aperture forms ₀=[X ₀, Y ₀], I in above relational expression _ii the signal intensity that pixel-by-pixel basis is received in certain sub regions, X _iand Y _iit is respectively the coordinate of i pixel in this subregion.

In ADAPTIVE OPTICS SYSTEMS, for discrete driving continuous surface deformable mirror, directly Slope Method is the most simple and effective control method.In this embodiment, adopt direct Slope Method to carry out preset to distorting lens 2 minute surfaces.Directly the basic theories of Slope Method is described below, and establishing distorting lens has n driver, and j driver is R to Hartmann sensor 3 detectors l facula deviation amount response corresponding to sub-aperture _xjand R (l) _yj(l); The corresponding facula deviation moment matrix of target face shape of preset distorting lens is G=[G _x, G _y], the control voltage that j driver will load is V _j; Have m sub-aperture if breathe out Hartmann sensor 3, the region in l sub-aperture is S _l.Each driver is linear superposition on the impact of hot spot side-play amount on the sub-aperture of Hartmann sensor 3, so there are Hartmann sensor 3 l the average gradients in sub-aperture to be:

$\left\{\begin{array}{c}{G}_x\left(l\right)={\mathrm{\Σ}}_{j=1}^n{R}_{\mathrm{xj}}\left(l\right)V_j\\ G_y\left(l\right)={\mathrm{\Σ}}_{j=1}^n{R}_{\mathrm{yj}}\left(l\right)V_j\end{array}\right.$ l＝1,2,3...m

Above formula can be expressed as matrix form:

$\left[\begin{array}{c}{G}_x\left(1\right)\\ G_y\left(1\right)\\ G_x\left(2\right)\\ G_y\left(2\right)\\ ...\\ G_x\left(m\right)\\ G_y\left(m\right)\end{array}\right]=\left[\begin{array}{cccc}{R}_{x1}\left(1\right)& R_{x2}\left(1\right)& ...& R_{\mathrm{xn}}\left(1\right)\\ R_{y1}\left(1\right)& R_{y2}\left(1\right)& ...& R_{\mathrm{yn}}\left(1\right)\\ R_{x1}\left(2\right)& R_{x2}\left(2\right)& ...& R_{\mathrm{xn}}\left(2\right)\\ R_{y1}\left(2\right)& R_{y2}\left(2\right)& ...& R_{\mathrm{yn}}\left(2\right)\\ ...& ...& ...& ...\\ R_{x1}\left(m\right)& R_{x2}\left(m\right)& ...& R_{\mathrm{xn}}\left(m\right)\\ R_{y1}\left(m\right)& R_{y2}\left(m\right)& ...& R_{\mathrm{yn}}\left(m\right)\end{array}\right]\left[\begin{array}{c}{V}_1\\ V_2\\ ...\\ V_n\end{array}\right]+\left[\begin{array}{c}{\mathrm{\ϵ}}_1\\ {\mathrm{\ϵ}}_2\\ {\mathrm{\ϵ}}_3\\ {\mathrm{\ϵ}}_4\\ ...\\ {\mathrm{\ϵ}}_{2m-1}\\ {\mathrm{\ϵ}}_{2m}\end{array}\right]$

ε is dimensionless, R _xyfor voltage-facula deviation amount response matrix.Obtaining so minimum on-load voltage is

for the pseudo inverse matrix of matrix R.Carrying out successively the each driver of distorting lens 2 being applied to unit voltage before closed-loop control, the facula deviation amount in the region that on Hartmann sensor 3, each lenticule is corresponding that records deducts initial facula deviation moment matrix G ₀, construct voltage-facula deviation amount response matrix R, and calculate the pseudo inverse matrix of R and be stored in data acquisition and control in the internal memory of computing machine 4, for computing time is saved in real-time closed-loop control.

While carrying out chamber inner light beam purification, data acquisition and control computing machine 4 need to utilize the priori face shape of Hartmann sensor 3 preset distorting lenss.Suppose the corresponding facula deviation moment matrix of the priori face shape G=[G of distorting lens _x, G _y] (G=G when closed loop first after beam cleanup system building ₀), the voltage vector that distorting lens is applied is

due to the creep effect of distorting lens 2 drivers and the reason such as mirror shape thermal stability is poor, with direct Slope Method, distorting lens driver is applied to voltage control signal for the first time, 2 shape distance objective surface form deviations of distorting lens are larger, in the invention process example, utilize direct Slope Method to carry out iteration control distorting lens minute surface 3 times.

In the invention process example, utilize Hartmann sensor to carry out iteration three times to 3 distorting lens driving voltages, distoring mirror shape is almost close to priori face shape, now, data acquisition and control computing machine 4 are no longer using facula deviation amount as FEEDBACK CONTROL index, the data that start beam quality diagnosis camera 8 to collect are processed, and calculate far field beam beam quality, in the embodiment of the present invention using the Power in the bucket (PIB) of light beam as far field beam quality.Gray-scale value I(x, y that far field beam light intensity and camera are measured) corresponding.Coordinate (the x of intensity peak _max, y _max) be I(x, y) coordinate of gray-scale value maximum point, the radius of bucket is r=15 pixel, light beam gross energy:

E＝∑∑I(x,y)

Energy in bucket:

E _PIB＝∑∑I(x,y) (x-x _max) ²+(y-y _max) ²＜r ²；

Power in the bucket PIB=E _pIB, the then control target using PIB as random paralleling gradient descent algorithm, utilizes each driver control voltage of random paralleling gradient descent algorithm real-time update distorting lens, follows the tracks of dynamical distortion in compensated cavity, makes laser instrument keep optimum duty.The implementation procedure of random paralleling descent algorithm is that hypothesis control system distorting lens driving voltage after k-1 iteration is inferior is in the time of the k time iteration, formation voltage random perturbation vector

after distorting lens driver is applied to positive disturbance, system performance evaluation index change amount is ${\mathrm{\δJ}}_{+}^k=J^k(u_1^{k-1}+{\mathrm{\δu}}_1^k,u_2^{k-1}+{\mathrm{\δu}}_2^k,...,u_{37}^{k-1}+{\mathrm{\δu}}_{37}^k)-J^k(u_1^{k-1},u_2^{k-1},...,u_{37}^{k-1});$ Distorting lens driver is applied after negative disturbance ${\mathrm{\δJ}}_{-}^k=J^k(u_1^{k-1}-{\mathrm{\δu}}_1^k,u_2^{k-1}-{\mathrm{\δu}}_2^k,...,u_{37}^{k-1}-{\mathrm{\δu}}_{37}^k)-J^k(u_1^{k-1},u_2^{k-1},...,u_{37}^{k-1}),$ System performance evaluation index change amount is

so when the k time iteration, the voltage being applied on distorting lens 2 drivers is u ^k=u ^k-1+ γ δ u ^kδ J ^k.System is for the first time when closed loop, and the voltage vector that distorting lens 2 is applied is

because G=G when closed loop first ₀in fact distorting lens 2 is applied to no-voltage, then adopt random optimization control algolithm to control laser remote field beam quality, control procedure is equivalent to traditional random optimization control method, and beam system expends and reaches 12 seconds, just converges to basic mode pattern.

When laser instrument 9 far field beam qualities are optimized, data acquisition is measured the corresponding facula deviation moment matrix of distoring mirror shape in real time with control computing machine 4, facula deviation moment matrix G ' while recording far field beam quality optimum, before control system exits, make G=G ', complete the renewal of facula deviation moment matrix.

Method of the present invention can effectively utilize priori data to carry out chamber inner light beam purification, but system does not have priori data in the time of closed loop first, still start to search for from the primary face shape of distorting lens, therefore can not show due advantage, having for the first time after priori data, control system can make full use of the priori data of last closed loop, and the system closed-loop speed of making is accelerated greatly.

Have for the first time after the priori data of closed-loop control, carried out chamber inner light beam for the second time below and purify experiment, to verify that the present invention utilizes validity and the raising speed of convergence actual effect of the present invention of priori data.The voltage making zero on distorting lens driver, the voltage vector that distorting lens is applied is

utilize Hartmann sensor to carry out iteration three times to distorting lens driving voltage, distoring mirror shape is almost close to the optimum face shape of target, then according to each driver control voltage of random paralleling gradient descent algorithm real-time update distorting lens, control system expends only 0.7s and just converges to basic mode pattern, has reduced 11.3s than the random control method for improving 12s consuming time with common.

Claims (3)

1. an in-chamber adaptive optical light beam cleaning system, it is characterized in that comprising: source of parallel light (1), distorting lens (2), Hartmann sensor (3), data acquisition and control computing machine (4), high-voltage amplifier (5), spectroscope (6), condenser lens (7), beam quality diagnosis camera (8) and laser instrument (9), described distorting lens (2) is as the rear mirror of laser instrument (9) resonator cavity; The collimation parallel beam that source of parallel light (1) is sent, the parallel beam that incident distorting lens (2), and source of parallel light at a certain angle (1) sends covers the hot spot of laser instrument (9) excited radiation light on distorting lens (2) minute surface; Parallel beam after distorting lens (2) reflection is received by Hartmann sensor (3); The light beam that laser instrument (9) sends is through spectroscope (6) beam splitting, and a branch of is as Laser output; Another bundle focuses into into beam quality diagnosis camera (8) through condenser lens (7), and data acquisition receives the signal of Hartmann sensor (3) and beam quality diagnosis camera (8) and is connected with high-voltage amplifier (5) with control computing machine (4); While carrying out beam cleanup, data acquisition utilizes Hartmann sensor (3) the upper preset priori face shape of distorting lens (2) with control computing machine (4), then the beam quality of measuring according to beam quality diagnosis camera (8), utilize the controlled output voltage of random optimization control method, thereby play the object of optimizing laser instrument (9) output beam quality.

2. a kind of in-chamber adaptive optical light beam cleaning system according to claim 1, is characterized in that: described random optimization control algolithm adopts random paralleling gradient descent algorithm, genetic algorithm, climbing method, ant group algorithm or simulated annealing.

3. an in-chamber adaptive optical light beam purification method, is characterized in that adopting in-chamber adaptive optical light beam cleaning system as claimed in claim 1 or 2, and performing step is as follows:

The first step, demarcates Hartmann sensor (3), and before laser instrument (9) start, data acquisition and the light spot image data of controlling computing machine (4) collection multiframe Hartmann sensor (3) output, be averaging the initial calibration data of rear conduct to hot spot data, according to formula

calculate the initial centroid position coordinate of the hot spot [X in sub-aperture _c0, Y _c0], and construct the hot spot initial slope matrix G that each facula deviation amount corresponding to sub-aperture forms ₀; Wherein I _ii the signal intensity that pixel-by-pixel basis is received in certain sub-aperture in Hartmann sensor (3), X _iand Y _iit is respectively the coordinate of i pixel;

Second step, experiment measuring distorting lens (2) voltage-slope response matrix R _xy, successively J driver of distorting lens (2) applied unit voltage and utilizes Hartmann sensor (3) to record hot spot side-play amount matrix simultaneously, voltage-slope response matrix R of structural deformation mirror (2) _xy, and solve R _xygeneralized inverse matrix

The 3rd step, preset distorting lens priori face shape, facula deviation moment matrix corresponding to distorting lens (2) priori face shape is G, while carrying out beam cleanup first, G=G ₀, the voltage vector that distorting lens (2) is applied is

$V=R_{\mathrm{xy}}^{+}(G-G_0);$

The 4th step, adopt random optimization control method to optimize far field beam quality, the light distribution data that data acquisition and control computing machine (4) receiving beam quality diagnosis camera (8) collect, calculate beam quality, according to the control voltage of each driver of random optimization control method real-time update distorting lens (2), follow the tracks of dynamical distortion in compensation distorting lens (2) chamber, make laser instrument (9) obtain diffraction limited beam output;

The 5th step, upgrade facula deviation moment matrix G, when the 4th step is optimized laser instrument (9) far field beam quality, data acquisition and the Beam Wave-Front slope of controlling computing machine (4) real time record Hartmann sensor (3) measurement, hot spot side-play amount matrix G ' while recording far field beam quality optimum, before this chamber inner light beam purifies and finishes, make G=G ' complete the renewal of facula deviation moment matrix, when chamber inner light beam purifies again, start to carry out from " the 3rd step ".

Images