Statistical perception of new type of machining error in chaotic optics and adaptive decision method of polishing parameters

Abstract

The present application relates to a kind of new type of chaotic optical processing error statistical perception and polishing parameter self-adaptive decision method, comprising the following steps: S1, for sub-aperture ultra-precision polishing, chaotic error is caused by the removal function of polishing tool each dwell time position under the influence of complex error source, establish the chaotic error perception model based on statistical significance;S2, establish the extraction method of polishing tool head chaotic error rate C, so that chaotic error perception model is used for actual polishing processing guidance;S3, according to chaotic error perception model and the extraction method of polishing tool head chaotic error rate C, combined with intelligent mathematical means to establish polishing parameter self-adaptive decision model, realize high-precision high-efficiency polishing processing.The beneficial effect is high efficiency, good consistency and determinacy, quantitative control in polishing process.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of modern advanced optical processing technology, in particular to a new type of chaotic optical processing error statistical perception and polishing parameter adaptive decision method. BACKGROUND

[0002] With the continuous development of modern optical technology, various high-precision optical components are more and more widely used in super strong laser system, lithography exposure system and astronomical observation system and other important national optical engineering fields; At present, the demand for ultra-precision optical elements is huge. For example, the world's largest laser system, the NIF device of the United States includes 7360 types of meter-level optical elements; The world's largest astronomical telescope, the E-ELT optical telescope of Europe, has a main mirror diameter of 39 meters, which is composed of nearly 800 meter-level off-axis parabolic hexagonal segments. And the error precision of all frequency bands is high, the NIF device of the United States has strict index requirements for errors in the frequency range of 0-100mm-1, aiming to prevent laser damage, prevent focal spot splitting and imaging blur. Although the introduction of sub-aperture polishing technology has greatly improved the efficiency and certainty of optical processing; But with the further improvement of precision and efficiency requirements, the perception and compensation ability of a large part of the polishing error which is difficult to quantify and predict, has chaotic characteristics and complex sources is insufficient, and the low convergence efficiency of the full frequency domain error caused by the decision method of the current artificial experience, even the phenomenon of repeated non-convergence has become a bottleneck problem restricting the production capacity and precision requirements of thousands of components at present.

[0003] Chaos theory is a method of qualitative thinking and quantitative analysis, which is used to explore the behavior in dynamic system that cannot be explained and predicted by single data relationship, but must be explained and predicted by overall and continuous data relationship.

[0004] The present application is aimed at the field of sub-aperture ultra-precision polishing, and the technical problems of low convergence efficiency of full frequency domain error and even repeated non-convergence caused by the decision method of current artificial experience, and the technical improvement of the statistical perception of optical processing error and the adaptive decision method of polishing parameter. SUMMARY

[0005] The purpose of the present application is to provide a sub-aperture ultra-precision polishing method with high efficiency, good consistency and certainty, and quantitative control of polishing process.

[0006] To achieve the above purpose, the technical scheme adopted by the present application is a new type of chaotic optical processing error statistical perception and polishing parameter adaptive decision method, comprising the following steps:

[0007] S1, for sub-aperture ultra-precision polishing, chaotic error is caused by the removal function of each dwell time position of the polishing tool under the influence of complex error sources, a chaotic error perception model based on statistics is established;

[0008] S2, a polishing tool head chaotic error rate C extraction method is established, so that the chaotic error perception model is used for actual polishing processing guidance;

[0009] S3, according to the chaotic error perception model and the polishing tool head chaotic error rate C extraction method, a polishing parameter adaptive decision model is established by combining intelligent mathematical means, and high-precision and high-efficiency polishing processing is realized.

[0010] Preferably, the chaotic error perception model established in step S1 is as follows:

[0011] E res ≈||E theory ||2+C·||Q||2

[0012] ≈||E initial +H-T*TIF||2+C·||T*TIF||2

[0013] with C=σ 2 (δ),

[0014] Wherein, E res is the residual error of the workpiece, E inital is the initial surface shape before polishing, H is the removal depth, T is the dwell time, TIF is the removal function, E theory is the residual error after deconvolution algorithm optimization, C is the variance of chaotic error distribution δ, that is, the chaotic error rate of the polishing tool head, and the symbol σ represents the standard deviation.

[0015] Preferably, the mathematical derivation step of the chaotic error perception model in step S1 is as follows:

[0016] S11, according to the principle of computer-controlled sub-aperture polishing, the chaotic error distribution under the unit removal amount is defined as a random distribution δ, and the removal error is defined as δ·(T*TIF). The actual polishing result E res is represented as: E res =||(E initial +H-(1+δ)·(T*TIF)||2;

[0017] S12, L2 norm optimization is adopted, and the formula is further simplified as:

[0018] S13, since the random distribution δ is difficult to be further separated and quantified, the formula is further simplified as: Wherein, N represents the number of elements in the matrix;

[0019] S14, in the case of good TIF volume removal rate calibration, the expected value of δ is close to 0, the formula

[0020]

[0021] Further simplified as: Where Q represents the theoretical material removal distribution;

[0022] S15, according to the characteristics that the main space period of the chaotic error distribution is less than that of the surface error, there is a local area of appropriate size, the change of the surface error is much smaller than that of the chaotic error, at this time, the surface error is approximately constant, and the formula is further simplified as: Where N is the total size of material removal, that is, the number of elements in the matrix, and M is the size of the appropriate local area;

[0023] S16, according to the law of large numbers, for a specific statistical distribution of chaotic error, if the sampling density is large enough, the size of the sub-aperture polishing tool is much smaller than that of the workpiece, that is, the appropriate local area is small enough N / M is large enough, the polishing tool head chaotic error rate C is close to the variance of the chaotic error distribution δ, which is regarded as a constant in a specific polishing process, different processing technologies correspond to different C, and the polishing tool head chaotic error rate C is expressed as: Where the symbol σ represents the standard deviation.

[0024] Preferably, the polishing tool head chaotic error rate C extraction method in step S2 is as follows: Where E inital and E final represent the actual measured surface shapes before and after polishing respectively; H set represents the material removal amount set according to experience during polishing; E initial +H set -T*TIF is the theoretical polishing residual error, that is, the residual error after optimization of the deconvolution algorithm.

[0025] Preferably, the polishing parameter adaptive decision model in step S3 is as follows: Where d represents the path spacing, the proportional estimation method is used to efficiently and accurately calculate the residence time T(d), H is the removal depth, and TIF is the removal function.

[0026] Preferably, the polishing parameter adaptive decision specifically includes the following steps:

[0027] S31, first, calculate the optimal removal depth H under the combination of different path spacings d and removal functions TIF;

[0028] S32, according to the intermediate frequency criterion, select the optimal path spacing d;

[0029] S33, selecting the optimal removal function TIF according to a low frequency criterion.

[0030] Preferably, the step S31 uses the golden section method to solve the optimal removal depth H.

[0031] Preferably, the step S32 is as follows: Where comb represents a comb function matrix, T msf represents the dwell time after eliminating the 0 frequency value, T interval represents the dwell time of different path intervals, T mean represents the spatial domain average value of T interval ; MSF actual <MSF theory The medium frequency ripple will be suppressed when the coefficient D is set artificially, which can be adjusted according to the actual polishing requirements.

[0032] Preferably, the coefficient D is set to 0.1.

[0033] Preferably, the step S33 low frequency criterion is as follows: comparing the RMS of the surface shape before and after polishing, calculating the low frequency convergence efficiency For the same surface shape, the greater the low frequency convergence efficiency R, the better the low frequency polishing ability of the removal function TIF.

[0034] The present application has the following beneficial effects: 1. In the field of sub-aperture ultra-precision polishing, due to the existence of a large part of polishing errors which are difficult to quantify and predict, have chaotic characteristics and complex sources, it is difficult to realize quantitative control of the polishing process. Based on the existing convolution convergence theory, the coupling form of the expectation and variance of the chaotic error in the polishing model is proved by innovatively using statistical theory, a chaotic error perception model based on statistics is proposed, and the statistical quantitative prediction of the evolution of polishing errors of various tools in each polishing cycle is realized; 2. In addition, based on the polishing tool and machining parameter adaptive decision algorithm of the chaotic error adaptive perception model, the synchronous improvement of polishing precision and efficiency is realized; 3. The shortcomings of the existing error perception method are made up, and the problems of low polishing efficiency, difficulty in ensuring the consistency and certainty of the polishing result, and difficulty in realizing quantitative control of the polishing process are solved. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 A new statistical perception of chaotic optical machining error and adaptive decision method of polishing parameters.

[0036] Figure 2 A derivation step diagram of a chaotic error perception model based on statistical significance.

[0037] Figure 3 is a new type of adaptive decision-making method flow chart of polishing parameters of chaotic optical processing.

[0038] Figure 4 is a new type of adaptive decision-making method of polishing parameters of chaotic optical processing MSF in the frequency criterion actual solving schematic diagram.

[0039] Figure 5 is a new type of statistical perception and adaptive decision-making method of polishing parameters of chaotic optical processing error polishing processing result schematic diagram.

CONCRETE IMPLEMENTATION

[0040] The application will be further described below in conjunction with the embodiments and with reference to the accompanying drawings.

[0041] Embodiment 1

[0042] This embodiment realizes a new type of statistical perception and adaptive decision-making method of polishing parameters of chaotic optical processing error.

[0043] This embodiment is based on the existing convolution convergence theory, and innovatively uses statistical theory to deduce and prove the coupling form of the expectation and variance of the chaotic error in the polishing model, proposes a statistical chaotic error perception method, and realizes the statistical quantitative prediction of the evolution of the polishing error of various tools in each polishing cycle; on this basis, this embodiment proposes a polishing tool and processing parameter adaptive decision-making algorithm based on adaptive perception of chaotic error, to realize the synchronous improvement of polishing precision and efficiency; the present application makes up for the shortcomings of the existing chaotic error perception method, and solves the problems of low polishing efficiency, difficulty in ensuring the consistency and certainty of the polishing result, and difficulty in realizing quantitative control of the polishing process of the existing polishing parameter decision-making method.

[0044] Figure 1 is a new type of statistical perception and adaptive decision-making method of polishing parameters of chaotic optical processing error step diagram. As shown in the accompanying Figure 1 , the steps of this embodiment are as follows:

[0045] 1. A chaotic error perception model based on statistical significance is established,

[0046] E res ≈||E theory ||2+C·||Q||2

[0047] ≈||E initial +H-T*TIF||2+C·||T*TIF||2

[0048] with C=σ 2 (δ)

[0049] where Eres E is the residual error of the workpiece inital H is the initial surface shape before polishing, H is the removal depth, T is the dwell time, TIF is the removal function, E theory E is the residual error after optimization of the deconvolution algorithm, C is the variance of the chaotic error distribution δ (the error rate of the tool head used for polishing). The variance C can be regarded as a constant in a specific polishing process, different processing techniques correspond to different variances C, and the symbol σ represents the standard deviation.

[0050] Figure 2 is a step diagram based on the statistical sense of chaotic error perception model. As shown in the accompanying Figure 2 Although the randomization characteristics of chaotic errors make it difficult to compensate for systematic errors, the present embodiment is based on the statistical theory and finds that chaotic errors have stable expectation and variance, thereby realizing the effective perception of chaotic errors in the statistical sense. The mathematical derivation is as follows:

[0051] According to the principle of computer-controlled sub-aperture polishing, essentially, chaotic errors are caused by the removal rate change of each dwell time position under the influence of complex error sources, and it can be considered that chaotic errors increase linearly with the increase of material removal amount. The present embodiment defines the chaotic error distribution under unit removal amount as a random distribution δ, and then the removal error can be defined as δ·(T*TIF), and the actual polishing result E res can be expressed as:

[0052] E res =||(E initial +H-(1+δ)·(T*TIF)||2

[0053] The above formula adopts L2 norm optimization, and the formula can be further simplified as:

[0054] E res =||E theory -δ·(T*TIF)||2

[0055] with E theory =E initial +H-T*TIF

[0056] Since the random distribution δ is difficult to be further separated and quantified, the above formula can be further simplified as:

[0057]

[0058] Where N represents the number of elements in the matrix. In the case of good TIF volume removal rate calibration, the expectation value of δ is usually close to 0. Therefore, the formula can be further simplified as:

[0059]

[0060] with

[0061] Q i = T*TIF

[0062] where Q represents the theoretical material removal distribution. According to the characteristic that the main space period of the chaotic error distribution is less than that of the surface error, there exists a local region of proper size, in which the variation of the surface error is much less than that of the chaotic error. At this time, the surface error can be approximated as a constant, and the formula can be further simplified as:

[0063]

[0064] with

[0065] where N is the total size of the material removal (the total number of elements in the matrix), and M is the size of the proper local region. According to the law of large numbers, for a certain statistical distribution of the chaotic error, if the sampling density is large enough, the size of the sub-aperture tool is much smaller than that of the workpiece (the local region is small enough, and N / M is large enough), C is close to the variance of the chaotic error distribution δ. The variance C can be regarded as a constant in a certain polishing process, and different processing techniques correspond to different variances C, which can be expressed as:

[0066]

[0067] where the symbol σ represents the standard deviation.

[0068] Thus far, the derivation of the chaotic error perception model described in the present embodiment is completed. The model can accurately predict the RMS (Root Mean Square) level after each tool processing, and the processing evolution process under each process parameter can be well mastered.

[0069] 2. The present embodiment establishes a method for extracting the chaotic error rate C, so that the chaotic error perception model can be used for actual optical processing guidance:

[0070]

[0071] E inital and E final represent the actual measured surface shapes before and after polishing, respectively; H set represents the material removal amount set according to experience during the polishing process; E initial + H set -T*TIF is the theoretical polishing residual error (residual error after optimization by the deconvolution algorithm).

[0072] 3. The embodiment is based on the chaotic error perception model and the extraction method of chaotic error rate C, and further combines intelligent mathematical means to realize a new type of high-precision and high-efficiency polishing parameter adaptive decision model to realize high-precision and high-efficiency processing, so as to get rid of the dilemma of high dependence on human experience, low error convergence precision and low efficiency in the whole frequency band under the existing process framework. The model optimization equation is as follows:

[0073] min E opt =||E initial +H-T(d)*TIF||2+C·||T(d)*TIF||2

[0074] with

[0075] s.t T(d) min <T(d)<T(d) max

[0076] Where d represents the path spacing, the proportional estimation method is used to calculate the residence time T(d) efficiently and accurately, and the above formula contains three variables: removal depth H, path spacing d and removal function TIF.

[0077] Figure 3 It is a new type of chaotic optical processing and polishing parameter adaptive decision method flow chart. As shown in the accompanying Figure 3 , the decision model first calculates the optimal removal depth H under the combination of different path spacing d and removal function TIF, and then selects the optimal path spacing d and removal function TIF according to the proposed medium frequency criterion and low frequency criterion. The golden section method is used to solve the removal depth H in the embodiment.

[0078] The medium frequency criterion is defined in the decision model to select the optimal path spacing d. First, for the medium frequency error, due to the constraint condition of the polishing machine, the path spacing cannot be infinitely small (the path spacing cannot be infinitely small); in addition, the larger the path spacing, the larger the removal amount of a single path point, and under the premise that most path points are constrained by speed, the actual removal amount is smaller, and the polishing efficiency is higher, but according to the convolution theorem, when the path spacing is larger, more obvious medium frequency error will occur. Therefore, it is of great significance to realize the adaptive decision of path spacing to ensure high-precision polishing without obvious medium frequency error. The medium frequency criterion in the embodiment is as follows:

[0079] MSF actual =T msf *TIF

[0080] with T msf =T interval -T mean

[0081] Tinterval =T·comb

[0082] Q S = (T·comb)*TIF

[0083] st T min ≤T≤T max

[0084] Where comb represents the comb function matrix. T msf T represents the dwell time after eliminating the 0 frequency value; interval T represents the dwell time at different path intervals. mean T represents interval The average value of the spatial domain. Figure 4 MSF is a mid-frequency criterion in a novel adaptive decision-making method for chaotic optical processing and polishing parameters. actual Solution diagram. (See attached diagram) Figure 4 As shown, MSF actual The solution method is such that if the following condition is met, the intermediate frequency ripple will be suppressed.

[0085] MSF actual <MSF theory ·D

[0086] The coefficient D is set manually and can be adjusted according to the actual polishing requirements. In the model, D is set to 0.1.

[0087] This embodiment defines a low-frequency criterion for selecting the optimal TIF. The low-frequency criterion is as follows:

[0088]

[0089] For the same surface shape, the larger the R value, the better the low-frequency polishing capability of the TIF, which is worth considering. Under the premise of satisfying the MSF constraint, the TIF with the best convergence efficiency R is usually selected as the optimal TIF to be used.

[0090] 4. Figure 5 This is a schematic diagram of the polishing result, illustrating a novel statistical sensing and adaptive decision-making method for polishing parameters in chaotic optical processing. (See attached diagram.) Figure 5 As shown, the adaptive decision model for polishing parameters proposed in this embodiment enables the processing of optical components with extremely high precision. In addition, besides existing methods that improve tool determinism (using tools with smaller error rates: magnetorheology, ion beams, etc.), adjusting the volume removal rate of the same tool with the same error rate has proven to be another effective method. High-precision shaping with a surface accuracy of 1.788 nm can be achieved by polishing with a small robot-based grinding head.

[0091] Example 2

[0092] The embodiment realizes a new type of statistical sensing of chaotic optical processing error and adaptive decision of polishing parameters. The embodiment performs a set of experiments on a robot polisher based on the embodiment 1 to further verify the accuracy of the proposed chaotic error sensing model and the effectiveness of the high-precision and high-efficiency polishing parameter adaptive decision model.

[0093] The polishing equipment selects ASEA Brown Boveris (ABB) IRB 6620 industrial robot, and the flange end of the robot is provided with a polishing tool. Two K9 optical elements with a diameter of 100 mm and a thickness of 10 mm are used as experimental workpieces (measuring 80% effective aperture), one for chaotic error extraction and one for model verification. The tool is composed of foamed silica gel and damping cloth; the polishing liquid is 10% concentration cerium oxide. Given the initial surface shape and a series of selected TIF of the industrial robot small polishing tool, the polishing parameters such as the best tool head, the best path spacing and the best removal depth are decided based on the polishing parameter adaptive decision model to guide the polishing.

[0094] The specific experimental steps of the embodiment are as follows:

[0095] 1. Perform chaotic error rate C extraction of the industrial robot small polishing tool:

[0096]

[0097] E inital and E final respectively represent the actual measured surface shapes before and after polishing; H set represents the material removal amount set according to experience (traditional convolution model) during polishing; E initial +H set -T*TIF is the theoretical polishing residual error (residual error after optimization by deconvolution algorithm). The error rate of the industrial robot small polishing tool used in the embodiment is 9.9%.

[0098] 2. The embodiment realizes high-precision and high-efficiency polishing based on the chaotic error sensing model and the polishing parameter adaptive decision model, and the specific equation is as follows:

[0099] min E opt =||E initial +H-T(d)*TIF||2+C·||T(d)*TIF||2

[0100] with

[0101] C≈σ 2 (δ)

[0102] s.t T(d) minT(d) < T(d) max

[0103] E res E is the residual error of the workpiece, H is the removal depth, T is the dwell time, TIF is the removal function, E inital E is the residual error of the workpiece, H is the removal depth, T is the dwell time, TIF is the removal function, E theory E is the residual error of the workpiece, H is the removal depth, T is the dwell time, TIF is the removal function, E opt E is the residual error of the workpiece, H is the removal depth, T is the dwell time, TIF is the removal function, E

[0104] The specific steps of the present example are first to calculate the optimal removal depth H under different path spacing d and removal function TIF combinations according to the above formula, and then select the optimal path spacing d and removal function TIF according to the proposed medium frequency criterion and low frequency criterion.

[0105] The present embodiment uses the golden section method to solve the removal depth H, and defines the medium frequency criterion for selecting the optimal path spacing d, which is as follows:

[0106] MSF actual = T msf *TIF

[0107] with T msf = T interval -T mean

[0108] T interval = T·comb

[0109] Q S = (T·comb)*TIF

[0110] s.t T min ≤ T ≤ T max

[0111] comb represents the comb function matrix. T msf represents the dwell time after eliminating the 0 frequency value; T interval represents the dwell time of different path intervals; T mean represents the spatial domain average value of T interval . The solution method of MSF actual is shown in the attached Figure 4 If the following conditions are met, the medium frequency ripple is suppressed. If the following conditions are met, the medium frequency ripple is suppressed.

[0112] MSF actual <MSF theory ·D

[0113] The coefficient D is artificially set and can be adjusted according to actual polishing requirements. In the model, D is set to 0.1.

[0114] The embodiment defines a low-frequency criterion for selecting the optimal TIF, and the low-frequency criterion is as follows

[0115]

[0116] For the same surface shape, the larger the R is, the better the low-frequency polishing capability of the TIF is, and it is worth considering. Under the premise of meeting the MSF constraint condition, the TIF with the best convergence efficiency R is usually selected as the optimal TIF to be used.

[0117] Each polishing step is guided by the best TIF, the best path interval and the best removal depth finally determined according to the above steps. After 4 polishing processes, the processing result is shown in FIG. 4. Figure 5 The polishing parameter decision result is as follows: the TIF with a rotation speed of 500 rpm and a volume removal rate of 0.0423 mm 3 / min is used for polishing processes 1 and 2, the removal depths are 0.056λ and 0.037λ respectively, and the path intervals are both 1 mm; the TIF with a rotation speed of 250 rpm and a volume removal rate of 0.0211 mm 3 / min is used for polishing process 3, the removal depth is 0.021λ, and the path interval is 1 mm; the TIF with a rotation speed of 250 rpm and a volume removal rate of 0.0106 mm 3 / min is used for polishing process 4, the removal depth is 0.016λ, and the path interval is 1 mm.

[0118] Those skilled in the art can understand that all or part of the steps of the above embodiment can be completed by hardware, or by a program instructing related hardware, and the program can be stored in a computer readable storage medium, which can be a magnetic disc, an optical disc, a read-only memory (ROM) or a random access memory (RAM) and the like.

[0119] The above only describes the preferred embodiments of the present application, and it should be noted that those skilled in the art can make several improvements and supplements without departing from the principles of the present application, and these improvements and supplements should also be considered as the protection scope of the present application.

Claims

1. A novel statistical sensing and adaptive decision-making method for polishing parameters in chaotic optical processing, characterized in that... Includes the following steps: S1. For sub-aperture ultra-precision polishing, chaotic error is caused by the removal function of each dwell time position of the polishing tool under the influence of complex error sources. A chaotic error perception model based on statistical significance is established. S2. Establish a method for extracting the chaotic error rate C of the polishing tool head, so that the chaotic error perception model can be used to guide actual polishing processes. S3. Based on the chaotic error perception model and the method for extracting the chaotic error rate C of the polishing tool head, and combined with intelligent mathematical methods, an adaptive decision-making model for polishing parameters is established to achieve high-precision and high-efficiency polishing.

2. The novel statistical sensing and adaptive decision-making method for polishing parameters of chaotic optical processing errors according to claim 1, characterized in that... The chaotic error-aware model established in step S1 is as follows: Among them, E res E represents the residual error of the workpiece. inital The initial surface shape before polishing is given by H, the removal depth is given by T, the dwell time is given by TIF, and E is given by E. theory C represents the residual error after optimization by the deconvolution algorithm, C represents the variance of the chaotic error distribution δ, which is the chaotic error rate of the polishing tool head, σ represents the standard deviation, and Q represents the theoretical material removal distribution.

3. The novel statistical sensing and adaptive decision-making method for polishing parameters of chaotic optical processing errors according to claim 2, characterized in that... The mathematical derivation steps of the chaotic error-aware model in step S1 are as follows: S11. Based on the computer-controlled sub-aperture polishing principle, the chaotic error distribution under unit removal amount is defined as a random distribution δ, and the removal error is defined as δ·(T*TIF). The actual polishing result E res Represented as: E res =||(E initial +H-(1+δ)·(T*TIF)||2; S12. Using L2 norm optimization, the formula is further simplified to: S13. Since the random distribution δ is difficult to further separate and quantify, the formula is further simplified to: Where N represents the number of elements in the matrix; S14. When the TIF volume removal rate is well calibrated, the expected value of δ is close to 0, and the formula is further simplified to: Where Q represents the theoretical material removal distribution; S15. Based on the characteristic that the principal spatial period of the chaotic error distribution is less than the principal spatial period of the surface shape error, there exists a local region of appropriate size where the change in surface shape error is much smaller than the change in chaotic error. In this case, the surface shape error is approximately constant, and the formula is further simplified to: Where N is the total size of material removal, i.e. the number of elements in the matrix, and M is the size of the appropriate local region; S16. According to the law of large numbers, for a specific statistical distribution of chaotic errors, if the sampling density is large enough, the size of the sub-aperture polishing tool is much smaller than the size of the workpiece, that is, the appropriate local area is small enough and N / M is large enough, the chaotic error rate C of the polishing tool head is close to the variance of the chaotic error distribution δ, which is regarded as a constant in a specific polishing process. Different processing techniques correspond to different C. The chaotic error rate C of the polishing tool head is expressed as: The symbol σ represents the standard deviation.

4. The novel statistical sensing and adaptive decision-making method for polishing parameters of chaotic optical processing errors according to claim 3, characterized in that... The method for extracting the chaotic error rate C of the polishing tool head in step S2 is as follows: Where E inital and E final The actual measured surface shapes before and after polishing are shown respectively; H set E represents the amount of material removed during the polishing process, set based on experience. initial +H set -T*TIF represents the theoretical polishing residual error, which is the residual error after optimization by the deconvolution algorithm.

5. The novel statistical sensing and polishing parameter adaptive decision-making method for chaotic optical processing errors according to claim 4, characterized in that... The adaptive decision model for polishing parameters in step S3 is as follows: Where d represents the path spacing, the proportional estimation method is used to calculate the dwell time T(d) efficiently and accurately, H is the removal depth, and TIF is the removal function.

6. The novel statistical sensing and adaptive decision-making method for polishing parameters of chaotic optical processing errors according to claim 5, characterized in that... Step S3, adaptive decision-making for polishing parameters, specifically includes the following steps: S31. First, calculate the optimal removal depth H under different path spacing d and TIF removal function combinations; S32. Based on the intermediate frequency criterion, select the optimal path spacing d; S33. Based on the low-frequency criterion, select the optimal removal function TIF.

7. The novel statistical sensing and adaptive decision-making method for polishing parameters of chaotic optical processing errors according to claim 6, characterized in that: Step S31 uses the golden section method to solve for the optimal removal depth H.

8. The novel statistical sensing and polishing parameter adaptive decision-making method for chaotic optical processing errors according to claim 6, characterized in that... The intermediate frequency criterion for step S32 is as follows: Where comb represents the comb function matrix, T msf T represents the dwell time after eliminating the 0 frequency value. interval T represents the dwell time at different path intervals. mean T represents interval Spatial domain mean; MSF actual <MSF theory • When the frequency ripple is at time D, it will be suppressed. The coefficient D is set manually and can be adjusted according to the actual polishing requirements.

9. The novel statistical sensing and adaptive decision-making method for polishing parameters of chaotic optical processing errors according to claim 8, characterized in that: The coefficient D is set to 0.

1.

10. A novel statistical sensing and polishing parameter adaptive decision-making method for chaotic optical processing errors according to claim 6, characterized in that... Step S33 Low-frequency criterion is as follows: Compare the RMS of the face shape before and after polishing, and calculate the low-frequency convergence efficiency. For the same surface shape, the larger the low-frequency convergence efficiency R, the better the low-frequency polishing capability of the removal function TIF.

Images