CN102079125B - Model-free control method for focal distance of injection molded plastic lens - Google Patents

Model-free control method for focal distance of injection molded plastic lens Download PDF

Info

Publication number
CN102079125B
CN102079125B CN 201010599452 CN201010599452A CN102079125B CN 102079125 B CN102079125 B CN 102079125B CN 201010599452 CN201010599452 CN 201010599452 CN 201010599452 A CN201010599452 A CN 201010599452A CN 102079125 B CN102079125 B CN 102079125B
Authority
CN
China
Prior art keywords
control
point
focal length
working point
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
CN 201010599452
Other languages
Chinese (zh)
Other versions
CN102079125A (en
Inventor
陈曦
孔祥松
邵之江
王喆
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CN 201010599452 priority Critical patent/CN102079125B/en
Publication of CN102079125A publication Critical patent/CN102079125A/en
Application granted granted Critical
Publication of CN102079125B publication Critical patent/CN102079125B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Injection Moulding Of Plastics Or The Like (AREA)

Abstract

The invention discloses a model-free control method for the focal distance of an injection molded plastic lens. According to the invention, aiming at the problems such as frequent change of operation parameters, difficulty of process modeling, limited model precision and the like existing in the focal distance control process of the injection molded plastic lens, the model-free control method for the focal distance of the injection molded plastic lens is designed according to the characteristics of the injection molded plastic lens that the cost in the production process is low and the batch can be repeated, thus the control on the focal distance of the plastic lens is realized. By utilizing the control method, the control overhead can be reduced, and the control time is shortened, thereby improving the focal distance control efficiency, and being beneficial to the energy conservation and emission reduction in the production process of injection molded lenses.

Description

Model-free control method for focal length of injection molding plastic lens
Technical Field
The invention relates to the field of focal length control of injection molding plastic lenses, in particular to a model-free focal length control method of injection molding plastic lenses.
Background
The focal length of the lens is linearly related to the diopter of the lens, and the focal length is an important quality index of the plastic optical lens, and the size and the stability of the lens reflect the quality characteristics of the lens. Focal length control is of great importance for quality control of lenses, especially for plastic optical lenses produced by injection molding. Because injection molding is a complex, multistage and nonlinear intermittent production process, and many process parameters influencing the focal length of an injection molding lens cause difficulty in establishing a process model, uncertain factors such as noise and the like often exist in the actual injection molding process, which cause difficulty in efficiently realizing the focal length control of the lens.
From the perspective of product quality control, in the field of focus control of injection molded lenses at present, common methods include: firstly, a trial and error method, which obtains a better control working point by repeatedly adjusting process parameters according to experience by an operator, is simple and easy to implement, but is very time-consuming, inefficient and heavily dependent on the personal experience of the operator; secondly, a test design method, which is to determine a test scheme through test design, carry out a test, collect focal length information of each test point, analyze or fit data, predict an optimal working point through an approximate fitting model to realize lens focal length control; the method has the advantages that the number of tests can be reduced through a well-designed test, and the process characteristics can be discovered as much as possible; the method has the defects that the experimental design method still needs more experience assistance and is difficult to implement on line, and only approximate local optimal working points can be found, so the control precision of the method is not high; third, a model-based focus control method; the method needs a prior model as a basis, and can be divided into two categories, namely a mechanism model and an identification regression model based on test data according to different models depending on the prior model. The method has the advantages of simple implementation, off-line implementation and the like on the basis of the existing model; however, due to the complexity of the injection molding process, the focal length characterization of the plastic lens is difficult to realize, and a mechanism model between the focal length and each process parameter cannot be accurately established. The regression model based on the data generally needs to be obtained through a large number of experiments, the workload is often large, and the extrapolation performance of the model is not ideal; the model is therefore a big bottleneck of the method. Meanwhile, the injection molding process has the characteristic of frequent change of working points, and materials, molds and some process conditions are often required to be changed in the industrial production process. In case of a change of these elements, the original model accuracy cannot be applied at all to a new production process. Therefore, the focus control method based on the model usually needs to consume a lot of time, effort and tests to establish the process model, and the control accuracy of the focus control method is necessarily limited by the corresponding model accuracy, which causes that the efficiency of the focus control method is low in the actual implementation process, and the control accuracy cannot be effectively guaranteed.
In summary, the currently used focus control method often requires sufficient experience of operators and needs a lot of tests and time investment, and is high in control cost, low in implementation efficiency, and not beneficial to energy conservation, emission reduction and environmental protection.
Considering that the injection molding process has the characteristics of low unit production cost, rapidness and easiness in repetition, even if the focal lengths of products produced in the first few or even dozens of batches cannot meet the requirements, as long as the focal lengths can continuously approach the control indexes and reach the requirements after a certain batch, the consumed capital cost and time cost are negligible compared with those of the traditional control method. Therefore, a focal length model-free control method based on a parallel perturbation method is provided according to the low cost and batch repeatability of the production process of the injection molding plastic lens.
The parallel Perturbation method (SPSA) is a nonlinear Stochastic Approximation method proposed by j.c. spider in 1992. (J.C.Spall. multivariable storage optimization Using a simulation optimization, IEEE Transactions on Automatic Control, vol.37, pp. 332-341.) this method is a Gradient-based optimization method, but its Gradient information is approximated by a function value; the method has the advantages that only function value information is needed in the searching process; meanwhile, the parallel perturbation strategy is adopted, so that the information quantity required by gradient estimation can be greatly reduced, and the method is particularly suitable for the multi-dimensional parameter optimization problem, and is convenient to implement and high in efficiency. However, in the actual industrial production process, the method is sensitive to process noise, and under the condition that uncertain factors such as noise exist, the accuracy of gradient estimation is poor, and the working efficiency of the method is low. Therefore, in the invention, the existence of uncertainty in the focal length control process is considered, the improvement is carried out on the basis of the parallel perturbation method, and the model-free control method for the focal length of the injection molding plastic lens is provided. The method has better robustness and noise resistance.
The invention realizes the model-free control of the lens focal length by utilizing the characteristics of low cost and repeatable batch in the production process of the injection molding plastic lens, reduces the control overhead and shortens the control time; the method has important significance for improving the focal length control efficiency of the injection molding lens and realizing energy conservation and emission reduction in the industrial production process.
Disclosure of Invention
The invention aims to provide a model-free and efficient focal length control method aiming at the defects and shortcomings of the existing injection molding plastic lens focal length control method.
The invention adopts the following technical scheme for realizing the aim of the invention: a model-free control method for focal length of an injection molding plastic lens comprises the following steps:
(1) the method is initialized: set of focus modeless control method parameters
Figure DEST_PATH_IMAGE002
Wherein:
Figure DEST_PATH_IMAGE004
is a focal length control index set value,
Figure DEST_PATH_IMAGE006
for the focus target deviation tolerance to be met,for the focus fluctuation variance tolerance,
Figure DEST_PATH_IMAGE010
in order to repeat the number of tests,
Figure DEST_PATH_IMAGE012
in order to repeat the number of measurements,
Figure DEST_PATH_IMAGE014
in order to verify the number of tests,
Figure DEST_PATH_IMAGE016
for the maximum number of test points to be counted,
Figure DEST_PATH_IMAGE018
are parallel perturbation method coefficients. Is provided withIs provided with
Figure DEST_PATH_IMAGE020
The process parameters can be respectively defined as:(ii) a Order to
Figure DEST_PATH_IMAGE024
Characterizing the second one formed by the combination of these process parameters
Figure DEST_PATH_IMAGE026
And controlling the working point. Setting the initial control operating point by the operator
Figure DEST_PATH_IMAGE028
，The number of the process parameters is shown; will be provided with
Figure DEST_PATH_IMAGE030
Normalization is carried out, and the vector of the normalized control working point is
Figure DEST_PATH_IMAGE032
(ii) a Setting control operating point count
Figure DEST_PATH_IMAGE034
。
(2) And (3) controlling the working point iteration based on a parallel perturbation method: the continuous control working point sequence of the production is set as
Figure DEST_PATH_IMAGE036
(ii) a Wherein
Figure DEST_PATH_IMAGE038
The focus estimation has already been carried out,
Figure DEST_PATH_IMAGE040
for a new control operating point vector to be estimated, byThe parallel perturbation method is used for generating,
Figure 230261DEST_PATH_IMAGE040
the specific production method is as follows.
a. Order to
Figure DEST_PATH_IMAGE042
，“"means the symbol is divided by the integer,
Figure DEST_PATH_IMAGE046
is the remainder; if it is not
Figure 163844DEST_PATH_IMAGE046
=0, turn (f); if it is not
Figure 506969DEST_PATH_IMAGE046
=1, go (b); if it is not
Figure 304024DEST_PATH_IMAGE046
=2, go (e).
b. Updating the gain of the parallel perturbation method according to a formula,
Figure DEST_PATH_IMAGE050
The gain is updated.
c. Generation of new ones by Monte Carlo
Figure 536029DEST_PATH_IMAGE020
Dimension random vector-perturbation vector
Figure DEST_PATH_IMAGE052
Wherein each dimension of the vector is randomly generated by a Bernoulli distribution, wherein the probabilities of generating +1, -1 are all 0.5.
d. Generating a positive perturbation point: generating a forward shot point according to the perturbation vector:(ii) a Order to
Figure 568576DEST_PATH_IMAGE040
=
Figure DEST_PATH_IMAGE056
And (4) turning to the step (3).
e. And (3) reverse perturbation point generation:
Figure DEST_PATH_IMAGE058
(ii) a Order to
Figure 205356DEST_PATH_IMAGE040
=
Figure DEST_PATH_IMAGE060
And (4) turning to the step (3).
f. Order to
Figure DEST_PATH_IMAGE062
(ii) a If s =0, let
Figure 297946DEST_PATH_IMAGE040
=
Figure 394078DEST_PATH_IMAGE032
Turning to the step (3); otherwise, estimating at
Figure DEST_PATH_IMAGE064
Approximate gradient at point:
Figure DEST_PATH_IMAGE066
(ii) a Wherein,
Figure DEST_PATH_IMAGE068
is an estimate of the focal length of the last control operating point.
g. Search for the next iteration point along the approximate gradient direction:(ii) a Order to
Figure 853484DEST_PATH_IMAGE040
=
Figure DEST_PATH_IMAGE072
And (4) turning to the step (3).
(3) Model-free estimation preprocessing: control operating point to be estimated
Figure 921935DEST_PATH_IMAGE040
Is a normalized process parameter combination vector; in the pretreatment stage, theThe corresponding physical process parameters can be obtained by carrying out reverse normalization(ii) a After conversion, it is necessary to ensure
Figure 4739DEST_PATH_IMAGE074
Still meeting the physical constraints of the process parameters. Is provided with
Figure DEST_PATH_IMAGE076
Is a feasible process parameter if
Figure DEST_PATH_IMAGE078
Then, then
Figure DEST_PATH_IMAGE080
(ii) a Otherwise, get
Figure DEST_PATH_IMAGE082
Wherein
Figure DEST_PATH_IMAGE084
Is Euclidean norm, i.e. the process parameterDistance in feasible domain
Figure 267837DEST_PATH_IMAGE074
Nearest point to replace
Figure 253111DEST_PATH_IMAGE074
As an alternative to controlling the operating point.
(4) Controlling a working point on-line test: according to the control operating point
Figure DEST_PATH_IMAGE086
Modifying the corresponding technological parameter value on the operation panel of the injection molding machine to ensure that the technological parameter value
Figure DEST_PATH_IMAGE088
Is set to
Figure DEST_PATH_IMAGE090
. Producing plastic lens by injection molding machine, measuring lens focal length by lensometer, and repeating measurement
Figure 375919DEST_PATH_IMAGE012
And recording the corresponding focal length measurement
Figure DEST_PATH_IMAGE092
(ii) a Wherein
Figure DEST_PATH_IMAGE094
The numbers are numbered for the times of the on-line tests,the number of measurements is numbered.
(5) Model-free estimation post-processing: in the model-free post-processing stage of the control working point, the number of repeated tests is initially set according to the model-free control method
Figure 305697DEST_PATH_IMAGE010
Calling (4) to control the working point to perform online test until the test times
Figure 296787DEST_PATH_IMAGE094
To achieveAt this time, the working point can be calculated at the control working point
Figure 302713DEST_PATH_IMAGE074
Focal length estimate of (b):
Figure DEST_PATH_IMAGE098
。
(6) and (3) verifying a control working point: if the operating point is controlled
Figure 985367DEST_PATH_IMAGE074
Estimate of the focal length of
Figure DEST_PATH_IMAGE100
Satisfies the following conditions:
Figure DEST_PATH_IMAGE102
，if the absolute value operation is represented, the focal length index is considered to be initially reached, otherwise, the step (7) is carried out; due to uncertain factors such as noise, additional measures are required
Figure 140667DEST_PATH_IMAGE014
The additional verification test is added, and the step (4) is called
Figure 100533DEST_PATH_IMAGE014
Next, the verification value was obtained by the following formula
Figure DEST_PATH_IMAGE106
: (ii) a If the verification value satisfies
Figure DEST_PATH_IMAGE110
Then the optimum process parameters control the operating point
Figure DEST_PATH_IMAGE112
Turning to (8); otherwise, the step (7) is carried out.
(7) Controlling the working point to count: controlling operating point counting
Figure DEST_PATH_IMAGE114
If, if
Figure DEST_PATH_IMAGE116
Turning to the step (2); otherwise, returning to the step (1) to readjust the optimization goals and parameters.
(8) And (3) implementation of an optimal control working point:
Figure DEST_PATH_IMAGE118
namely the optimal control working point found by the method. According to
Figure DEST_PATH_IMAGE120
Setting relevant technological parameters for controlling focal length on the control panel of the injection molding machine respectively to ensure that the technological parameters
Figure 624530DEST_PATH_IMAGE088
Is set to
Figure DEST_PATH_IMAGE122
(ii) a And performing online implementation on the optimal working point, wherein the focal length of the produced injection molding lens meets the control target.
The invention has the beneficial effects that:
(1) the invention realizes the control of the focal length of the injection molding lens without depending on a model, avoids the great cost and time investment of model establishment and improves the efficiency of focal length control;
(2) the method does not depend on a model, greatly reduces the dependence on the knowledge and information of the known process, does not need human intervention in the optimization control process, can automatically run, is convenient to implement on line, and greatly reduces the implementation cost and difficulty;
(3) the method adopts a parallel perturbation method to realize the gradient approximation of the multivariable problem by using a small amount of tests, thereby improving the efficiency of the model-free control method;
(4) because the injection molding process often has certain randomness, the accurate optimal control working point is difficult to obtain based on the modeling method; the method can effectively overcome noise and model deviation by implementing on line and introducing anti-randomness measures, and finds a real and reliable optimal focal length control working point.
Drawings
FIG. 1 is a schematic diagram of the operation of the present invention;
FIG. 2 is a schematic workflow of the present invention;
FIG. 3 is a schematic flow chart of a parallel perturbation method for generating a sequence of control working points in the present invention;
FIG. 4 shows the control effect of the present invention on the plastic focus control of an injection molding machine.
Detailed Description
The present invention is described in detail below with reference to the attached drawings. FIG. 1 is a schematic diagram of the present invention, in which a parallel perturbation method is used to iteratively generate a control working point sequence, each control working point is subjected to parameter setting by an injection molding machine control panel and is implemented on an injection molding machine on line, model-free estimation is performed by model-free estimation preprocessing, on-line experiments, model-free post-processing and the like and fed back to the parallel perturbation method, the parallel perturbation method updates the control working point sequence according to a feedback result to generate a new control working point to be estimated, the process is continuously repeated until a control target requirement is met and a model-free control method verification link is performed, and at this time, the output of the control method is an optimal control working point, thereby realizing the focus control of the plastic lens.
Fig. 2 is a schematic view of the working process of the present invention. A model-free control method for focal length of an injection molding plastic lens comprises the following steps:
(1) the method is initialized: set of focus modeless control method parametersWherein:
Figure 425575DEST_PATH_IMAGE004
is a focal length control index set value,for the focus target deviation tolerance to be met,
Figure 249361DEST_PATH_IMAGE008
for the focus fluctuation variance tolerance,
Figure 454077DEST_PATH_IMAGE010
in order to repeat the number of tests,
Figure 652977DEST_PATH_IMAGE012
in order to repeat the number of measurements,
Figure 7342DEST_PATH_IMAGE014
in order to verify the number of tests,
Figure 938389DEST_PATH_IMAGE016
for the maximum number of test points to be counted,
Figure 692718DEST_PATH_IMAGE018
are parallel perturbation method coefficients. Is provided with
Figure 882260DEST_PATH_IMAGE020
The process parameters can be respectively defined as:
Figure 343328DEST_PATH_IMAGE022
(ii) a Order to
Figure 507594DEST_PATH_IMAGE024
Characterizing the second one formed by the combination of these process parameters
Figure 437635DEST_PATH_IMAGE026
And controlling the working point. Setting the initial control operating point by the operator
Figure 916020DEST_PATH_IMAGE028
，The number of the process parameters is shown; will be provided with
Figure 81608DEST_PATH_IMAGE030
Normalization is carried out, and the vector of the normalized control working point is
Figure 544951DEST_PATH_IMAGE032
(ii) a Setting control operating point count。
(2) And (3) controlling the working point iteration based on a parallel perturbation method: the continuous control working point sequence of the production is set as
Figure 768349DEST_PATH_IMAGE036
(ii) a Wherein
Figure 539996DEST_PATH_IMAGE038
The focus estimation has already been carried out,
Figure 428318DEST_PATH_IMAGE040
is newThe control working point vector to be estimated is generated by a parallel perturbation method,
Figure 497774DEST_PATH_IMAGE040
the specific production method is as follows.
a. Order to
Figure 850258DEST_PATH_IMAGE042
，“
Figure 730489DEST_PATH_IMAGE044
"means the symbol is divided by the integer,
Figure 856839DEST_PATH_IMAGE046
is the remainder; if it is not
Figure 543035DEST_PATH_IMAGE046
=0, turn (f); if it is not
Figure 687709DEST_PATH_IMAGE046
=1, go (b); if it is not
Figure 988109DEST_PATH_IMAGE046
=2, go (e).
b. Updating the gain of the parallel perturbation method according to a formula
Figure 913340DEST_PATH_IMAGE048
,The gain is updated.
c. Generation of new ones by Monte CarloDimension random vector-perturbation vector
Figure 106664DEST_PATH_IMAGE052
Wherein each dimension of the vector is randomly generated by a Bernoulli distributionThe probabilities of generating +1 and-1 are all 0.5.
d. Generating a positive perturbation point: generating a forward shot point according to the perturbation vector:
Figure 456874DEST_PATH_IMAGE054
(ii) a Order to
Figure 484873DEST_PATH_IMAGE040
=
Figure 853406DEST_PATH_IMAGE056
And (4) turning to the step (3).
e. And (3) reverse perturbation point generation:
Figure 246342DEST_PATH_IMAGE058
(ii) a Order to
Figure 880585DEST_PATH_IMAGE040
=
Figure 135111DEST_PATH_IMAGE060
And (4) turning to the step (3).
f. Order to(ii) a If s =0, let
Figure 616DEST_PATH_IMAGE040
=
Figure 309107DEST_PATH_IMAGE032
Turning to the step (3); otherwise, estimating at
Figure 616591DEST_PATH_IMAGE064
Approximate gradient at point:
Figure 772766DEST_PATH_IMAGE066
(ii) a Wherein,
Figure 255307DEST_PATH_IMAGE068
is an estimate of the focal length of the last control operating point.
g. Search for the next iteration point along the approximate gradient direction:
Figure 801826DEST_PATH_IMAGE070
(ii) a Order to
Figure 709739DEST_PATH_IMAGE040
=
Figure 907371DEST_PATH_IMAGE072
And (4) turning to the step (3).
(3) Model-free estimation preprocessing: control operating point to be estimated
Figure 813010DEST_PATH_IMAGE040
Is a normalized process parameter combination vector; in the pretreatment stage, the
Figure 909142DEST_PATH_IMAGE040
The corresponding physical process parameters can be obtained by carrying out reverse normalization
Figure 309161DEST_PATH_IMAGE074
(ii) a After conversion, it is necessary to ensure
Figure 112032DEST_PATH_IMAGE074
Still meeting the physical constraints of the process parameters. Is provided with
Figure 250890DEST_PATH_IMAGE076
Is a feasible process parameter if
Figure 21268DEST_PATH_IMAGE078
Then, then
Figure 474247DEST_PATH_IMAGE080
(ii) a Otherwise, getWherein
Figure 189185DEST_PATH_IMAGE084
As Euclidean norm, i.e. distance in the feasible domain using process parameters
Figure 197593DEST_PATH_IMAGE074
Nearest point to replaceAs an alternative to controlling the operating point.
(4) Controlling a working point on-line test: according to the control operating point
Figure 277730DEST_PATH_IMAGE086
Modifying the corresponding process parameter value on the operation panel of the injection molding machine to enable the first step
Figure 696073DEST_PATH_IMAGE088
The set value of each process parameter is
Figure 254093DEST_PATH_IMAGE090
. Producing plastic lens by injection molding machine, measuring lens focal length by lensometer, and repeating measurement
Figure 534027DEST_PATH_IMAGE012
And recording the corresponding focal length measurement
Figure 165997DEST_PATH_IMAGE092
(ii) a Wherein
Figure 83137DEST_PATH_IMAGE094
The numbers are numbered for the times of the on-line tests,
Figure 49825DEST_PATH_IMAGE096
the number of measurements is numbered.
(5) Model-free estimation post-processing: in the model-free post-processing stage of the control working point, the number of repeated tests is initially set according to the model-free control method
Figure 382717DEST_PATH_IMAGE010
Calling (4) to control the working point to perform online test until the test times
Figure 197090DEST_PATH_IMAGE094
To achieve
Figure 705038DEST_PATH_IMAGE010
At this time, the working point can be calculated at the control working point
Figure 972071DEST_PATH_IMAGE074
Focal length estimate of (b):
Figure 108654DEST_PATH_IMAGE098
。
(6) and (3) verifying a control working point: if the operating point is controlled
Figure 964484DEST_PATH_IMAGE074
Estimate of the focal length of
Figure 957848DEST_PATH_IMAGE100
Satisfies the following conditions:
Figure 649860DEST_PATH_IMAGE102
，
Figure 75288DEST_PATH_IMAGE104
if the absolute value operation is represented, the focal length index is considered to be initially reached, otherwise, the step (7) is carried out; due to uncertain factors such as noise, additional measures are required
Figure 598673DEST_PATH_IMAGE014
The additional verification test is added, and the step (4) is called
Figure 966200DEST_PATH_IMAGE014
Next, the verification value was obtained by the following formula
Figure 394776DEST_PATH_IMAGE106
:
Figure 669900DEST_PATH_IMAGE108
(ii) a If the verification value satisfies
Figure 251054DEST_PATH_IMAGE110
Then the optimum process parameters control the operating point
Figure 271706DEST_PATH_IMAGE112
Turning to (8); otherwise, the step (7) is carried out.
(7) Controlling the working point to count: controlling operating point counting
Figure 628DEST_PATH_IMAGE114
If, if
Figure 17125DEST_PATH_IMAGE116
Turning to the step (2); otherwise, returning to the step (1) to readjust the optimization goals and parameters.
(8) And (3) implementation of an optimal control working point:
Figure 515103DEST_PATH_IMAGE118
namely the optimal control working point found by the method. According to
Figure 473700DEST_PATH_IMAGE120
Setting relevant technological parameters for controlling focal length on the control panel of the injection molding machine respectively to ensure that the technological parameters
Figure 362022DEST_PATH_IMAGE088
Is set to
Figure 244527DEST_PATH_IMAGE122
(ii) a And performing online implementation on the optimal working point, wherein the focal length of the produced injection molding lens meets the control target.
FIG. 3 is a schematic flow chart of the parallel perturbation method for generating control working points in the invention.
Examples
The following describes an implementation process of a model-free control method for the focal length of an injection lens by taking the focal length control of a plastic magnifier lens as an example.
Firstly, initializing the method through the step (1); focal length control target of the lens=90 mm, and the focal length target deviation tolerance is set according to the focal length noise fluctuation condition of the lens and the focal length tolerance of the lens
Figure DEST_PATH_IMAGE124
mm, weight fluctuation variance tolerance
Figure DEST_PATH_IMAGE126
mm, number of repeated testsNumber of repeated measurements
Figure DEST_PATH_IMAGE130
Number of verification tests
Figure DEST_PATH_IMAGE132
Maximum number of testsCoefficient of parallel perturbation algorithm
Figure DEST_PATH_IMAGE136
(ii) a Selecting the technological parameters to be optimized as follows: the injection pressure, the dwell pressure and the dwell time, wherein the injection section is divided into a first section and a second section for respective control, the segmentation point of which is also a key process parameter. Thus, in the lens focus control problem, a total of 5 process parameters are required to be adjusted, and the process parameters are selectedInitial values of numbers are:
Figure DEST_PATH_IMAGE138
105 bar (bar) for the first injection section, 105 bar for the second injection section, 45% of the injection stroke for the first and second injection sections, 80 bar for the holding pressure, and 8 second for the holding time; to pairNormalization is performed so that each dimension variable is [0,100 ]]Interval, normalized vector
Figure DEST_PATH_IMAGE140
。
Turning to the step (2), iteratively generating a control working point sequence by a parallel perturbation method, wherein the generated sequence is determined by a parallel perturbation method sequence generation flow related by the invention; assume that the current control operating point is generated as
Figure 476029DEST_PATH_IMAGE040
。
In step (3), forModel-free estimation preprocessing is carried out to generate corresponding control working points which can be implemented on line
Figure 57631DEST_PATH_IMAGE074
。
Executing the step (4) to control the working pointPerforming an on-line test according to
Figure 220945DEST_PATH_IMAGE074
Changing parameter setting of injection molding machine by control panel of injection molding machine, starting injection molding process, producing a plastic lens, and measuring focal length of the lens by a lensometerAnd recorded, measured three times in total, corresponding to the measured values:,,。
step (5) performing model-free post-processing on the control working point due to repeated test timesTherefore, it is required to call (4) for final production
Figure 67602DEST_PATH_IMAGE128
A plastic lens and record its focal length (
Figure DEST_PATH_IMAGE148
,
Figure DEST_PATH_IMAGE150
,
Figure DEST_PATH_IMAGE152
；
Figure DEST_PATH_IMAGE154
,
Figure DEST_PATH_IMAGE156
,
Figure DEST_PATH_IMAGE158
) (ii) a The estimated value at that point is
Figure DEST_PATH_IMAGE160
。
Step (6) for controlling the working pointAnd (4) carrying out verification: if it is not
Figure 639846DEST_PATH_IMAGE074
Focal length estimate of a point
Figure 667845DEST_PATH_IMAGE100
Satisfies the following conditions:
Figure DEST_PATH_IMAGE162
mm，
Figure 308534DEST_PATH_IMAGE104
representing absolute value operation, the focus control index is considered to be initially reached, at which time additional operations are requiredSecondary additional verification test; otherwise, turning to (7); and (5) calling the step (4) twice during verification test, and obtaining a verification value through the following formula:
Figure DEST_PATH_IMAGE164
(ii) a If the verification value satisfiesmm, optimal process parameter control set valueTurning to (8); otherwise, turning to the step (7).
In step (7), the test operating points are counted
Figure 777190DEST_PATH_IMAGE114
If, if
Figure DEST_PATH_IMAGE168
If yes, continuing to generate a control working point by using a parallel perturbation method in the step (2); otherwise, the optimization targets and parameters need to be adjusted again in the step (1), and a new control process needs to be executed again.
Step (8) of optimizingAnd controlling the working point to be implemented.
Figure 16541DEST_PATH_IMAGE118
=[109 120 41 77 5.9]T
Namely the optimal control working point found by the method. According toThe technological parameters related to the focus control are set on the control panel of the injection molding machine, wherein the pressure of the first injection section is 109 bar (bar), the pressure of the second injection section is 120 bar, the segmentation points of the first injection section and the second injection section are 41 percent of the injection stroke, the pressure maintaining pressure is 77 bar, and the pressure maintaining time is 5.9 second. The focal length of the injection-molded product produced under the parameter setting can meet the control target.
Fig. 4 (a) is a control effect diagram of the embodiment, and an optimal control operating point is finally found through 7 iterations; to further illustrate the performance of the control method, FIG. 4 (b) provides a graph showing the results from a different initial setting parameter(s) (b)
Figure DEST_PATH_IMAGE170
) And starting another control track, and under the control track, carrying out 6 iterations on model-free control to reach an optimal control working point. The result proves that the method can overcome the process noise to search the set focal length control target under a small number of test times and a small number of iteration batches, and ensure the stability of the optimal control test point.
As described above, the present invention utilizes the low cost and batch repeatable nature of the injection molded plastic lens manufacturing process to design a model-less control method for lens focal length. The control method can reduce the lens control overhead and shorten the control time; the practical application result shows that the invention has ideal application effect.

Claims (2)

1. A model-free control method for focal length of an injection molding lens is characterized by mainly comprising the following steps:
(1) the method is initialized: set of focus modeless control method parameters
Figure FDA00002431033400011
Wherein: f. oftgIs a focal length control index set value, mu is a focal length target deviation tolerance, sigma is a focal length fluctuation variance tolerance, tau is a repeated test frequency,
Figure FDA00002431033400012
for repeated measurement times, χ is the number of verification tests, Max is the maximum test point count, and { a, A, c, α, γ } is the coefficient of the parallel perturbation method; n process parameters are set, which can be respectively defined as x(1),…,x(n)(ii) a Order toCharacterizing a pth control operating point formed by the combination of process parameters; setting the initial control operating point by the operatorn is the number of process parameters; x is to be1Normalization is carried out, and the vector of the normalized control working point is
Figure FDA00002431033400015
Setting the control working point count s to be 0;
(2) and (3) controlling the working point iteration based on a parallel perturbation method: the continuous control working point sequence of the production is set as
Figure FDA00002431033400016
Wherein
Figure FDA00002431033400017
The focus estimation has already been carried out,
Figure FDA00002431033400018
for a new control working point vector to be estimated, a parallel perturbation method is used for generating:
(3) model-free estimation preprocessing: control operating point to be estimated
Figure FDA00002431033400019
Is a normalized process parameter combination vector; in the pretreatment stage, the
Figure FDA000024310334000110
The corresponding physical process parameter x can be obtained by carrying out reverse normalizations(ii) a After conversion, x needs to be ensuredsStill meet the physical constraint of the technological parameter; let xi be a feasible range of process parameters, if xsE.g. xi, xs=xs(ii) a Otherwise, get xs≡arg min{||x-xs| | x ∈ xi }, where | | | x-xsI | is the Euclidean norm, i.e. the distance x in the feasible domain using the process parameterssNearest point to replace xsAs an alternative control operating point;
(4) controlling a working point on-line test: according to the control operating point
Figure FDA000024310334000111
Modifying the corresponding process parameter value on the operation panel of the injection molding machine to ensure that the set value of the process parameter k is
Figure FDA000024310334000112
Producing plastic lens by injection molding machine, measuring lens focal length by lensometer, and repeating measurement
Figure FDA000024310334000113
And recording the corresponding focal length measurement
Figure FDA000024310334000114
Wherein i is an online test frequency number, and j is a measurement frequency number;
(5) model-free estimation post-processing: in the model-free post-processing stage of the control working point, the repeated test times tau are initially set according to the model-free control method, the (4) is called to carry out the on-line test of the control working point until the test times i reach tau, and at the moment, the control working point x can be calculatedsFocal length estimate of (b):
Figure FDA00002431033400021
(6) and (3) verifying a control working point: if the operating point x is controlledsAt an estimated value f of the focal lengthsSatisfies the following conditions: d (x)s)=abs(fs-ftg) Mu is less than or equal to mu, abs (·) represents absolute value operation, then the focal length index is considered to be initially reached, otherwise, step (7) is carried out; for the reason of uncertain factors, additional X times of additional verification tests are needed, the step (4) is called, and the verification value is obtained through the following formula
Figure FDA00002431033400023
If the verification value satisfies
Figure FDA00002431033400024
The optimal process parameter controls the operating point xopt=xsTurning to (8); otherwise, connecting to step (7);
(7) controlling the working point to count: controlling the working point count s to be s +1, and if s is less than Max, turning to the step (2); otherwise, returning to the step (1) to readjust the optimization targets and parameters;
(8) and (3) implementation of an optimal control working point:
Figure FDA00002431033400025
namely the optimal control working point found by the method; according to xoptRespectively setting the relevant process parameters of focal length control on the control panel of the injection molding machine to ensure that the set value of the process parameter k is
Figure FDA00002431033400026
And performing online implementation on the optimal working point, wherein the focal length of the produced injection molding lens meets the control target.
2. The method for mold-less control of focal length of injection molded lens according to claim 1, wherein in the step (2), the step (2) is performed
Figure FDA00002431033400027
The specific production method is as follows:
(a) let flag = s |3, "|" denote an integer division symbol, and flag is the remainder; if flag =0, turn (f); if flag =1, turn (b); if flag =2, turn (e);
(b) updating the gain of the parallel perturbation method according to the formula ak=a/(A+k)α,ck=c/kγUpdating the gain;
(c) generation of a new n-dimensional random vector-perturbation vector delta by Monte CarlokWherein each dimension of the vector is randomly generated by a Bernoulli distribution, wherein the probabilities of generating +1, -1 are all 0.5;
(d) generating a positive perturbation point: generating a forward shot point x from the perturbation vectork-phus=xk+ckΔk(ii) a Order to x ‾ s = x k - plus , Turning to the step (3);
(e) reverse perturbation Point Generation xk-minus=xk-ckΔk(ii) a Order to
Figure FDA00002431033400032
Turning to the step (3);
(f) order to
Figure FDA00002431033400033
If s =0, let
Figure FDA00002431033400034
Turning to the step (3); otherwise, estimate at x from perturbation pointskApproximate gradient at point gk=(fs-1-fs)/2ckΔk(ii) a Wherein f iss-1Is the last controlFocal length estimation of a working point is made;
(g) search for the next iteration point along the approximate gradient direction: x is the number ofk+1=xk-akgk(ii) a Order to
Figure FDA00002431033400035
And (4) turning to the step (3).
CN 201010599452 2010-12-22 2010-12-22 Model-free control method for focal distance of injection molded plastic lens Expired - Fee Related CN102079125B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201010599452 CN102079125B (en) 2010-12-22 2010-12-22 Model-free control method for focal distance of injection molded plastic lens

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201010599452 CN102079125B (en) 2010-12-22 2010-12-22 Model-free control method for focal distance of injection molded plastic lens

Publications (2)

Publication Number Publication Date
CN102079125A CN102079125A (en) 2011-06-01
CN102079125B true CN102079125B (en) 2013-04-17

Family

ID=44085416

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201010599452 Expired - Fee Related CN102079125B (en) 2010-12-22 2010-12-22 Model-free control method for focal distance of injection molded plastic lens

Country Status (1)

Country Link
CN (1) CN102079125B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108447737B (en) * 2018-05-18 2019-06-14 厦门理工学院 A kind of relay base quality optimization system based on simplex search

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5914884A (en) * 1997-01-02 1999-06-22 General Electric Company Method for evaluating moldability characteristics of a plastic resin in an injection molding process
CN1715036A (en) * 2004-06-29 2006-01-04 钰德科技股份有限公司 Jet forming method and its device without suture
CN101094763A (en) * 2004-12-30 2007-12-26 博士伦公司 Core locking assembly and method for orientation of asymmetric tooling

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
SG91321A1 (en) * 2000-03-27 2002-09-17 Sumitomo Heavy Industries Method for controlling an injection molding machine capable of reducing variations in weight of molded products
JP2006015538A (en) * 2004-06-30 2006-01-19 Canon Inc Injection molding method and production system control method using it

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5914884A (en) * 1997-01-02 1999-06-22 General Electric Company Method for evaluating moldability characteristics of a plastic resin in an injection molding process
CN1715036A (en) * 2004-06-29 2006-01-04 钰德科技股份有限公司 Jet forming method and its device without suture
CN101094763A (en) * 2004-12-30 2007-12-26 博士伦公司 Core locking assembly and method for orientation of asymmetric tooling

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
JP特开2006-15538A 2006.01.19

Also Published As

Publication number Publication date
CN102079125A (en) 2011-06-01

Similar Documents

Publication Publication Date Title
Chetverikov Testing regression monotonicity in econometric models
CN108287471B (en) Parameter self-tuning method of MIMO offset format model-free controller based on system error
CN113544599A (en) Method for executing a process and optimizing a control signal used in the process
CN102101352B (en) Model-free weight controlling method for injection molding product
US11860590B2 (en) Model-free optimization method of process parameters of injection molding
CN107609298A (en) A kind of Jiles Atherton model parameter identification methods and device
CN111784057A (en) Short-term power load prediction combination method based on snapshot feedback mechanism
CN110765704A (en) Novel automatic deep neural network modeling method applied to microwave device
CN114841106A (en) Integrated circuit optimization method and system based on rule-guided genetic algorithm
CN102079125B (en) Model-free control method for focal distance of injection molded plastic lens
CN104772878A (en) Product weight control method based on iteration modeling and optimization for injection molding process
CN113408622A (en) Non-invasive load identification method and system considering characteristic quantity information expression difference
Wei et al. A bayesian approach to robust inverse reinforcement learning
CN108447737B (en) A kind of relay base quality optimization system based on simplex search
CN111259340A (en) Saturated load prediction method based on logistic regression
CN107222328B (en) Method and device for predicting saturation inflection point
CN115932594A (en) Multi-innovation least square online parameter identification method for power battery
CN104298213A (en) Index time varying gain type iterative learning control algorithm based on reference batch
CN115167150A (en) Batch process two-dimensional off-orbit strategy staggered Q learning optimal tracking control method with unknown system dynamics
CN112861315A (en) One-dimensional descent search method for non-convex single-target optimal power flow global solution of power system
CN107947181B (en) Decoupling full-linear optimal power flow model based on similar hot start environment
Lu et al. Quality control of batch process using natural gradient based model-free optimization
Kek et al. First-Order Linear Ordinary Differential Equation for Regression Modelling
CN117094130B (en) Fractional order piezoelectric ceramic identification method based on maximum likelihood least square algorithm
CN114311574B (en) Injection speed optimization control method, system and device of injection molding machine

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20130417

Termination date: 20191222

CF01 Termination of patent right due to non-payment of annual fee