Precise structural part mold correction process optimization method based on extended Kalman filter

Abstract

The invention discloses a precision structural part mold correction process optimization method based on an extended Kalman filter, and the method comprises the steps: enabling the extended Kalman filter to serve as a process model of injection molding-prior estimation-mold modification-posteriori estimation of multiple iterations, and representing the change rule of the size of a product through a posteriori estimation value of each iteration; and thus, the functions of mold modification effect evaluation, product size measurement error analysis, molded product size prediction after mold repair and the like are realized. The mold modification efficiency and the size precision of a formed product can be improved, so that the production efficiency and the product quality are improved.

Description

Technical Field

[0001] The present invention relates to the field of precision structural parts injection molding, and in particular to a precision structural parts mold correction process optimization method based on an extended Kalman filter (Extended Kalman Filter, hereinafter referred to as EKF). Background Art

[0002] Precision structural injection molded parts are widely used in electronics, automobiles, medical and other fields due to their high precision and high quality. As users' requirements for product precision continue to increase, precision structural parts molding technology and precision mold design have also developed rapidly. However, various factors in the injection molding process, such as inconsistent product structure deformation and uneven cooling in different areas, lead to problems such as product shrinkage and warping, resulting in deviations between the final product size and the mold design size, which seriously affects the accuracy of precision injection molded structural parts.

[0003] In order to ensure that the precision of injection molded parts meets the tolerance requirements of users, injection molding companies usually need to go through multiple iterations of injection molding and mold modification processes. At present, the mold modification method commonly used by companies is mainly based on the empirical reverse compensation method, that is, according to the shrinkage characteristics of the material during the molding process, the size of the mold is determined by reverse calculation from the target product size. However, due to the diversity and complexity of the size of precision structural parts, this simplified method that relies on experience is often difficult to accurately determine the mold size, which affects the precision of the product, the accuracy and efficiency of mold modification. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention proposes a method for optimizing the correction process of a precision structural part mold based on an extended Kalman filter.

[0005] The specific technical solutions are as follows:

[0006] A method for optimizing a precision structural part mold correction process based on an extended Kalman filter comprises the following steps:

[0007] S1: Prepare a standard mold according to the standard drawing, and use it to perform injection molding to obtain an initial product; measure each characteristic dimension of the initial product to obtain a measurement value z0; perform correlation analysis on each characteristic dimension to obtain a dimension correlation matrix;

[0008] S2: Based on the injection molding characteristics, the state transfer function f(·) and process noise covariance matrix Q of the extended Kalman filter are obtained, and based on the size measurement characteristics, the measurement function h(·) and measurement noise covariance matrix R of the extended Kalman filter are obtained; the state vector of the extended Kalman filter is initialized and covariance matrix P0;

[0009] S3: Obtain the prior estimate value of the state vector in this iteration according to the estimation and correction equations of the extended Kalman filter Posterior estimate value Based on the error from the standard drawing size, implement die modification to obtain the die modification value u for this iteration k ; Inject using the modified die to obtain the new product size measurement value z k ;

[0010] S4: If the posterior estimate value in this iteration does not meet the set tolerance standard, execute S5; if it meets, conduct multiple measurements on the product obtained by injection molding in this iteration. If all measurement results z k all meet the tolerance standard, end the iteration. If there is a certain measurement result that does not meet the tolerance standard, execute S5;

[0011] S5: Repeat S3 - S4 to start the next iteration; the prior estimate value of the next iteration is obtained according to the estimation equation of the extended Kalman filter and the posterior estimate value of this iteration ;

[0012] If when the number of iterations reaches the preset upper limit, the posterior estimate value and the measured value after die repair still do not meet the tolerance standard, then change the parameters of the extended Kalman filter including the state transition function, and repeat S3 - S5 using the changed extended Kalman filter.

[0013] Furthermore, in S1, the dimension correlation matrix N is an m×m symmetric matrix, where m is the number of characteristic dimensions. If the i - th characteristic dimension and the j - th characteristic dimension are correlated with each other, then N (i，j) = N (j，i) = 1, and the remaining elements are all 0.

[0014] Furthermore, in S2, the expression of the measurement function h(·) is:

[0015] h(x k ) = x k

[0016] The expression of the state transition function f(·) is:

[0017]

[0018] where k is the number of iterations, k = 1, 2, 3, …, K, and K is the iteration upper limit; x k represents the true value of the product size obtained in the k - th iteration, and u kdenotes the die size modification value at the k-th iteration; α is the non-linear equivalent coefficient, used to adjust the uncertainty of the product shrinkage rate, obtained by a data-driven method, and its value range is (1, 3); x k + αu k is used to represent the die size value after reverse compensation; is the approximate shrinkage rate; β is a real number, used to adjust the uncertainty of the extended Kalman filter, and its value range is (-0.1, 0.1).

[0019] Furthermore, the non-linear equivalent coefficient α is calculated by a data-driven method, and the specific steps are as follows:

[0020] (1) Simulation data acquisition: Through injection molding simulation software, multiple groups of simulation data of products obtained by injection molding are acquired. The simulation data includes: geometric parameters of the products and process parameters during the injection molding process; the simulation data covers different die cavity adjustment schemes and process conditions;

[0021] (2) The input of the extended Kalman filter is the geometric parameters of the product. The extended Kalman filter uses an optimization algorithm to estimate the α parameter of the state transition function. The α parameter is used to reflect the influence of the die modification value u k on the iterative process;

[0022] (3) The simulation data is divided into a training set and a test set. The training set is used to train the extended Kalman filter, and the test set is used to evaluate the performance of the trained extended Kalman filter; through cross-validation, ensure that the performance of the extended Kalman filter model is consistent on different data sets. Specifically, calculate the error between the predicted value and the actual measured value of the extended Kalman filter, analyze the source and distribution characteristics of the error, and adjust the structure or parameters of the extended Kalman filter according to the error analysis results;

[0023] (4) Input the new product size measurement value into the trained extended Kalman filter to obtain the corresponding α parameter.

[0024] Furthermore, in step S3, the state equation expression of the extended Kalman filter is as follows:

[0025] x k = f(x k-1 , u k-1 ) + w k-1

[0026] In the formula, k is the iteration number, k = 1, 2, 3, …, K, and K is the iteration upper limit; x k represents the true value of the product size obtained at the k-th iteration, and x k-1 represents the true value of the product size obtained at the (k - 1)-th iteration; u k-1denotes the die modification value at the (k - 1)-th iteration; w k-1 denotes the process noise at the (k - 1)-th iteration, whose covariance matrix follows a normal distribution P(w k ) ~ N(0, Q);

[0027] The measurement equation of the extended Kalman filter is as follows:

[0028] z k = h(x k-1 ) + v k-1

[0029] where z k denotes the measured value of the product size obtained at the k-th iteration, and v k-1 denotes the measurement noise at the (k - 1)-th iteration, whose covariance matrix follows a normal distribution P(v k ) ~ N(0, R);

[0030] The estimation equation of the extended Kalman filter is as follows:

[0031]

[0032] where denotes the prior estimate value at the k-th iteration, denotes the posterior estimate value at the (k - 1)-th iteration; is the covariance matrix of denotes the difference between the true value and the prior estimate value at the k-th iteration, follows a Gaussian distribution, P k-1 is the covariance matrix of e k-1 k-1 denotes the difference between the true value and the posterior estimate value at the (k - 1)-th iteration, and e k-1 follows a Gaussian distribution, P(e k-1 ) ~ N(0, P k-1 ); F k-1 is the Jacobian matrix of the state transition function;

[0033] The correction equation of the extended Kalman filter is as follows:

[0034]

[0035] where K k is the Kalman gain at the k-th iteration, is the posterior estimate value at the k-th iteration, I is the identity matrix, and P k is the covariance matrix of the posterior estimate value , and H k is the Jacobian matrix of the measurement function.

[0036] Furthermore, the mold modification value u for this iteration k is specifically obtained according to the following sub-steps:

[0037] S3.1: Identify the current posterior estimate of the characteristic dimensions that exceed the tolerance standard set by the user;

[0038] S3.2: Conduct a correlation analysis on the characteristic dimensions that exceed the tolerance standard obtained in S3.1 to obtain a set of relevant dimensions;

[0039] S3.3: Based on the mold modification analysis and the set of relevant dimensions, determine the mold cavity adjustment plan and the expected product dimensions. According to the mold cavity adjustment plan, modify the mold. All the characteristic dimensions of the product obtained by injection molding using the modified mold should be within the tolerance range. Calculate the mold modification value u according to the mold cavity adjustment plan k .

[0040] Furthermore, in S2, the measurement noise covariance matrix R is obtained by multiplying the identity matrix by a coefficient, and this coefficient is related to the accuracy of the measuring instrument.

[0041] Furthermore, in S1, the standard drawing includes: the dimensions and tolerances of the product to be injection molded.

[0042] The beneficial effects of the present invention are:

[0043] (1) High-precision prediction: By introducing the extended Kalman filter, the present invention can accurately predict the variation law of the product dimensions after mold modification, ensuring that the final product dimensions meet the set tolerance standards;

[0044] (2) Measurement error identification: By introducing the extended Kalman filter, the present invention can effectively identify the errors in the measurement process, provide more reliable dimension measurement results, and reduce the repeated mold modification caused by measurement errors.

[0045] (3) Intelligent evaluation of mold modification effect: The present invention can realize the automation and intelligence of the mold modification process, reduce manual intervention, and improve production efficiency.

[0046] (4) Strong robustness: The extended Kalman filter used in the present invention is applicable to complex nonlinear systems, can handle the dimensional diversity and process complexity of different products, and has wide applicability. Description of the Drawings

[0047] Figure 1 is a flow chart of the optimized method for the mold correction process of precision structural parts based on the extended Kalman filter in the embodiments of the present invention.

[0048] Figure 2It is a key dimension change curve graph of the mold injection product after multiple iterations in the embodiment of the present invention.

[0049] Figure 3 It is a convergence situation graph of the extended Kalman filter in the embodiment of the present invention. Specific embodiments

[0050] The present invention will be described in detail below according to the accompanying drawings and preferred embodiments. The purpose and effect of the present invention will become more apparent. The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0051] As Figure 1 shown, an optimization method for the mold correction process of precision structural parts based on an extended Kalman filter specifically includes the following steps:

[0052] S1: Prepare a standard mold according to the standard drawing provided by the user (the standard drawing includes information such as product dimensions and their tolerances), and perform the first injection molding to obtain an initial product. The true values of the respective characteristic dimensions of the initial product are x0; measure multiple characteristic dimensions of the initial product to obtain measurement values z0, and perform a correlation analysis on the respective characteristic dimensions to obtain a dimension correlation matrix.

[0053] The dimension correlation matrix N is an m×m symmetric matrix, where m is the number of characteristic dimensions. If the i-th characteristic dimension and the j-th characteristic dimension are correlated with each other, then N (i，j) = N (j，i) = 1. Otherwise, the remaining elements of the matrix are all 0.

[0054] S2: Obtain the state transition function f(·) (non-linear) and the process noise covariance matrix Q of the extended Kalman filter according to the injection molding characteristics, and obtain the measurement function h(·) (non-linear) and the measurement noise covariance matrix R of the extended Kalman filter according to the dimension measurement characteristics; initialize the state vector and the covariance matrix P0 of the extended Kalman filter.

[0055] Furthermore, the measurement noise covariance matrix R is obtained by multiplying the identity matrix by a coefficient, and this coefficient is related to the accuracy of the measuring instrument.

[0056] The expression of the obtained measurement function is:

[0057] h(x k ) = x k

[0058] The expression of the obtained state transition function is:

[0059]

[0060] In the formula, k is the number of iterations, k = 1, 2, 3, …, K, where K is the upper limit of iterations; x k represents the true value of the product size obtained in the k-th iteration, and u k represents the mold size modification value in the k-th iteration; α is the non-linear equivalent coefficient, and its value range is (1, 3), which is used to adjust the uncertainty of the product shrinkage rate and is obtained through a data-driven method; x k + αu k is used to represent the mold size value after reverse compensation; is the approximate shrinkage rate obtained according to the true value x k of the product size obtained in the k-th iteration and the mold size modification value u k ; β is a real number, and its value range is (-0.1, 0.1), which is used to adjust the uncertainty of the extended Kalman filter.

[0061] Furthermore, the non-linear equivalent coefficient α is calculated through a data-driven method, and the specific implementation steps are as follows:

[0062] (1) Simulation data acquisition: By using injection molding simulation software such as Moldflow, multiple groups of simulation data of products obtained through injection molding are acquired. The simulation data includes: geometric parameters of the product, process parameters (such as temperature, pressure, etc.) during the injection molding process. The simulation data covers different mold cavity adjustment schemes and process conditions to improve the generalization ability of the extended Kalman filter.

[0063] (2) Data-driven modeling (the model is the extended Kalman filter): By substituting multiple groups of simulation data into the extended Kalman filter and using the least squares method or other optimization algorithms, the α parameter of the state transition function is estimated. The α parameter reflects the influence of the mold modification value u k on the iterative process.

[0064] (3) Model verification and adjustment: The simulation data is divided into a training set and a test set. The extended Kalman filter is trained using the training set, and the performance of the trained extended Kalman filter is evaluated using the test set.

[0065] Through cross-validation, ensure that the performance of the extended Kalman filter model is consistent on different data sets. Specifically, calculate the error between the predicted value and the actual measured value of the extended Kalman filter, analyze the source and distribution characteristics of the error, and adjust the structure or parameters of the extended Kalman filter according to the error analysis results to improve the prediction accuracy.

[0066] (4) Input the new product size measurement value into the trained extended Kalman filter to obtain the corresponding α parameter.

[0067] During the actual production process, new measurement data can be continuously collected and input into the extended Kalman filter for online update and optimization. Through continuous iteration, the adaptability and prediction ability of the extended Kalman filter can be gradually improved. According to the changes in actual production (such as material properties, environmental conditions, etc.), the parameters of the extended Kalman filter are dynamically adjusted to ensure that the extended Kalman filter can always accurately reflect the dynamic characteristics of the system.

[0068] S3: Obtain the prior estimate value of the state vector in this iteration according to the estimation and correction equations of the extended Kalman filter Posterior estimate value Based on The error from the standard drawing size, implement die modification to obtain the die modification value u for this iteration (i.e., the k-th iteration), k , and then use the modified die for injection molding and obtain the new product size measurement value z k .

[0069] The extended Kalman filter has stronger reliability and robustness for predicting the current injection mold modification process of precision structural parts. Its state equation is as follows:

[0070] x k = f(x k-1 , u k-1 ) + w k-1

[0071] In the formula, x k-1 represents the true value of the product size obtained in the (k - 1)-th iteration (i.e., the previous iteration); u k-1 represents the die modification value in the (k - 1)-th iteration; w k-1 represents the process noise in the (k - 1)-th iteration, and its covariance matrix follows a normal distribution P(w k ) ~ N(0, Q), and take Q I 10×10 .

[0072] Its measurement equation is as follows:

[0073] z k = h(x k-1 ) + v k-1

[0074] In the formula, z k represents the measurement value of the product size obtained in the k-th iteration, and v k-1 represents the measurement noise in the (k - 1)-th iteration, and its covariance matrix follows a normal distribution P(v k ) ~ N(0, R).

[0075] The estimation equation of the extended Kalman filter is as follows:

[0076]

[0077] In the formula, represents the prior estimate value of the k-th iteration, represents the posterior estimate value of the (k - 1)-th iteration; is 's covariance matrix, represents the difference between the true value and the prior estimate value of the k-th iteration. Assuming obeys a Gaussian distribution, P k-1 is e k-1 's covariance matrix, represents the difference between the true value and the posterior estimate value of the (k - 1)-th iteration. Assuming e k-1 obeys a Gaussian distribution, P(e k-1 ) ~ N(0, P k-1 ); F k-1 is the Jacobian matrix of the state transition function.

[0078] The correction equation of the extended Kalman filter is as follows:

[0079]

[0080]

[0081] In the formula, K k is the Kalman gain of the k-th iteration, is the posterior estimate value of the k-th iteration, I is the identity matrix, P k is the covariance matrix of the posterior estimate value , H k is the Jacobian matrix of the measurement function.

[0082] The mold modification value u of this iteration k is specifically obtained according to the following sub-steps:

[0083] S3.1: Identify the feature dimensions in the current posterior estimate value that exceed the tolerance standard set by the user.

[0084] S3.2: Conduct a correlation analysis on the above feature dimensions that exceed the tolerance standard to obtain a set of related dimensions. This step of operation can also obtain the feature dimensions related to them in the dimension correlation matrix according to the above feature dimensions that exceed the tolerance standard, and set them to obtain a set of related dimensions.

[0085] S3.3: Based on the mold repair analysis and the set of related dimensions, determine the mold cavity adjustment plan and the expected product dimensions. All the feature dimensions of the product obtained by injection molding the mold according to the mold cavity adjustment plan should be within the tolerance range, and calculate the mold modification value u according to the mold cavity adjustment plank 。

[0086] S4: If the posterior estimate value of the current iteration does not meet the set tolerance standard, then execute S5; if it meets, perform multiple measurements on the product obtained by injection molding in this iteration to reduce measurement errors. If all measurement results z k all meet the tolerance standard, then end the iteration and complete the optimization. If there is a certain measurement result that does not meet the tolerance standard, then execute S5.

[0087] S5: Repeat S3 - S4 to start the next iteration; among them, the prior estimate value of the next iteration is obtained according to the estimation equation of the extended Kalman filter and the posterior estimate value of this iteration , that is

[0088] If when the number of iterations reaches the preset upper limit, the posterior estimate value and the measured value after mold repair still do not meet the tolerance standard, then it is judged that there may be defects in the existing mold cavity adjustment plan, change the parameters of the extended Kalman filter including the state transition function, and then perform the operations of S3 - S5 using the optimized extended Kalman filter.

[0089] If there is still a situation where the iteration reaches the preset upper limit but the posterior estimate value still does not meet the tolerance standard after changing the mold repair strategy, then the mold repair personnel need to evaluate the operations involved in the mold repair process such as the mold repair method, mold repair strategy, and measurement method to find the problem.

[0090] The present invention will be specifically described below in conjunction with embodiments.

[0091] Embodiment

[0092] S1: Prepare a standard mold according to the standard drawing and perform injection molding. The obtained initial product has ten characteristic dimensions, and its measured value z0 is [0.01 0.69 0.44 0.15 0.06 0.10 0.01 0.08 0.07 0.53] T . Correspondingly, perform a correlation analysis on these ten dimensions, and the expression of the dimension correlation matrix N is:

[0093]

[0094] S2: Take the covariance matrix P0 and the process noise covariance matrix Q as the identity matrix, R = 0.1·I 10×10 , α = 2, β = 0.05. At this time, the measurement function h(x k ) = x k , and the state transition function

[0095] S3: The state equation of the Kalman filter is The measurement equation is z k = x k-1 + v k-1 . Take the Jacobian matrix F of the state transition function k-1 , the Jacobian matrix H of the measurement function k Both are identity matrices.

[0096] Substitute the initialized state vector and covariance matrix P0 obtained in S2 into the estimation equation to obtain the prior estimate value of the state vector in the k-th iteration. The expression is as follows:

[0097]

[0098] Then, according to the correction equation, obtain the posterior estimate value of the state vector in the k-th iteration. The expression is as follows:

[0099]

[0100] Subsequently, sequentially execute S4 - S5. After multiple iterations, finally obtain the corrected mold. Injecting with this mold, the change curve of the posterior estimate values of the ten characteristic dimensions of the product is as Figure 2 shown. According to Figure 2 , it can be known that there is a set of posterior estimate values within the tolerance range Then it indicates that the product obtained this time meets the tolerance standard requirements proposed by the user. Find the mold cavity size corresponding to the mold modification in the k-th time. After injecting the product with this cavity size and measuring the size of the product, if the measurement results of multiple times all meet the requirements, it indicates that this mold meets the production requirements.

[0101] The convergence situation of the extended Kalman filter during the iteration process is as Figure 3 shown. From Figure 3 , it can be known that by using the method of the present invention to optimize the correction process, the extended Kalman filter can achieve convergence, indicating that using the extended Kalman filter as the model for mold modification has good reliability and model generalization ability.

[0102] Those of ordinary skill in the art can understand that the above are only preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing examples, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention.

Claims

1. A precision structural parts mold correction process optimization method based on extended Kalman filter, characterized in that: The following steps are involved: S1: Prepare a standard mold according to the standard drawing, and use it to perform injection molding to obtain an initial product; measure each characteristic dimension of the initial product to obtain a measurement value z0; perform correlation analysis on each characteristic dimension to obtain a dimension correlation matrix; S2: Based on the injection molding characteristics, the state transfer function f(·) and process noise covariance matrix Q of the extended Kalman filter are obtained, and based on the size measurement characteristics, the measurement function h(·) and measurement noise covariance matrix R of the extended Kalman filter are obtained; the state vector of the extended Kalman filter is initialized and covariance matrix P0; S3: According to the estimation and correction equation of the extended Kalman filter, the prior estimate of the state vector in this iteration is obtained Posterior Estimates based on The error with the standard drawing size is used to implement mold correction and obtain the mold modification value u for this iteration k ; Use the modified mold for injection molding to obtain new product size measurements z k ; S4: If the posterior estimate of this iteration If the set tolerance standard is not met, S5 is executed; if it is met, the product obtained by this iterative injection molding is measured multiple times. If all the measurement results are z k If all the measurement results meet the tolerance standard, the iteration ends. If any measurement result does not meet the tolerance standard, S5 is executed. S5: Repeat S3-S4 to start the next iteration; the a priori estimate of the next iteration is based on the estimation equation of the extended Kalman filter and the a posteriori estimate of this iteration get; If the posterior estimation value and the measured value after modeling still do not meet the tolerance standard when the number of iterations reaches the preset upper limit, the parameters of the extended Kalman filter including the state transfer function are changed, and the changed extended Kalman filter is used to repeat S3-S5.

2. The method for optimizing the correction process of precision structural parts mold based on the extended Kalman filter according to claim 1 is characterized in that: In S1, the dimension correlation matrix N is a symmetric matrix of m×m, where m is the number of characteristic dimensions. If the i-th characteristic dimension and the j-th characteristic dimension are correlated with each other, then N (i，j) =N (j，i) =1, and the rest of the elements are 0.

3. The method for optimizing the correction process of precision structural parts mold based on extended Kalman filter according to claim 1, characterized in that: In S2, the expression of the measurement function h(·) is: h(x k )=x k The expression of the state transfer function f(·) is: Where k is the number of iterations, k = 1, 2, 3, ..., K, K is the upper limit of iterations; x k Indicates the true value of the product size obtained in the kth iteration, u k represents the mold size modification value of the kth iteration; α is the nonlinear equivalent coefficient, which is used to adjust the uncertainty of the product shrinkage rate. It is obtained by data-driven method and has a value range of (1,3); x k +αu k Used to indicate the mold size value after reverse compensation; is the approximate shrinkage rate; β is a real number used to adjust the uncertainty of the extended Kalman filter, and its value range is (-0.1, 0.1).

4. The method for optimizing the correction process of precision structural parts mold based on extended Kalman filter according to claim 3 is characterized in that: The nonlinear equivalent coefficient α is calculated in a data-driven way. The specific steps are as follows: (1) Acquisition of simulation data: using injection molding simulation software, multiple sets of simulation data of products obtained by injection molding are acquired. The simulation data include: geometric parameters of the products and process parameters during the injection molding process. The simulation data covers different mold cavity adjustment schemes and process conditions. (2) The input of the extended Kalman filter is the geometric parameters of the product. The extended Kalman filter uses an optimization algorithm to estimate the α parameter of the state transfer function. The α parameter is used to reflect the mold modification value u k Impact on the iterative process; (3) Dividing the simulation data into a training set and a test set, the training set is used to train the extended Kalman filter, and the test set is used to evaluate the performance of the trained extended Kalman filter; through cross-validation, the performance of the extended Kalman filter model on different data sets is ensured to be consistent. Specifically, the error between the predicted value of the extended Kalman filter and the actual measured value is calculated, the source and distribution characteristics of the error are analyzed, and the structure or parameters of the extended Kalman filter are adjusted according to the error analysis results; (4) Input the new product size measurement value into the trained extended Kalman filter to obtain the corresponding α parameter.

5. The method for optimizing the correction process of precision structural parts mold based on extended Kalman filter according to claim 3 is characterized in that: In S3, the state equation expression of the extended Kalman filter is as follows: x k =f(x k-1 ,u k-1 )+w k-1 Where k is the number of iterations, k = 1, 2, 3, ..., K, K is the upper limit of iterations; x k Indicates the true value of the product size obtained in the kth iteration, x k-1 Indicates the true value of the product size obtained in the k-1th iteration; u k-1 represents the mold modification value of the k-1th iteration; w k-1 represents the process noise of the k-1th iteration, whose covariance matrix follows the normal distribution P(w k )~N(0,Q); The measurement equation of the extended Kalman filter is as follows: z k =h(x k-1 )+v k-1 In the formula, z k Indicates the measured value of the product size obtained in the kth iteration, v k-1 represents the measurement noise of the k-1th iteration, whose covariance matrix follows the normal distribution P(v k )~N(0,R); The estimation equation of the extended Kalman filter is as follows: In the formula, represents the prior estimate of the kth iteration, represents the posterior estimate of the k-1th iteration; yes The covariance matrix of represents the difference between the true value and the prior estimate at the kth iteration, Obeying Gaussian distribution, P k-1 Yes k-1 The covariance matrix of represents the difference between the true value and the posterior estimate at the k-1th iteration, e k-1 Obeying Gaussian distribution, P(e k-1 )~N(0,P k-1 );F k-1 is the Jacobian matrix of the state transfer function; The correction equation of the extended Kalman filter is as follows: In the formula, K k is the Kalman gain of the kth iteration, is the posterior estimate of the kth iteration, I is the identity matrix, P k is the posterior estimate The covariance matrix, H k is the Jacobian matrix of the measurement function.

6. The method for optimizing the correction process of precision structural parts mold based on extended Kalman filter according to claim 1, characterized in that: The mold modification value u for this iteration k The specific steps are as follows: S3.1: Identify the current a posteriori estimate Feature dimensions that exceed the tolerance standards set by the user; S3.2: Perform correlation analysis on the characteristic dimensions exceeding the tolerance standard obtained in S3.1 to obtain a set of related dimensions; S3.3: Based on the mold modification analysis and related dimension sets, determine the mold cavity adjustment plan and the expected product size, modify the mold according to the mold cavity adjustment plan, and all characteristic dimensions of the product obtained by injection molding with the modified mold must be within the tolerance range. Calculate the mold modification value u according to the mold cavity adjustment plan k .

7. The method for optimizing the precision structural parts mold correction process based on the extended Kalman filter according to claim 1, characterized in that: In S2, the measurement noise covariance matrix R is obtained by multiplying the unit matrix by a coefficient, and the coefficient is related to the accuracy of the measuring instrument.

8. The method for optimizing the correction process of precision structural parts mold based on extended Kalman filter according to claim 1, characterized in that: In S1, the standard drawing includes: the dimensions and tolerances of the product to be injection molded.

Images