CN116852665A - Injection molding process parameter intelligent adjusting method based on mixed model - Google Patents
Injection molding process parameter intelligent adjusting method based on mixed model Download PDFInfo
- Publication number
- CN116852665A CN116852665A CN202310862943.4A CN202310862943A CN116852665A CN 116852665 A CN116852665 A CN 116852665A CN 202310862943 A CN202310862943 A CN 202310862943A CN 116852665 A CN116852665 A CN 116852665A
- Authority
- CN
- China
- Prior art keywords
- injection molding
- model
- process parameter
- kriging
- target value
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 134
- 238000001746 injection moulding Methods 0.000 title claims abstract description 79
- 230000008569 process Effects 0.000 claims abstract description 85
- 230000007547 defect Effects 0.000 claims abstract description 25
- 238000005457 optimization Methods 0.000 claims abstract description 13
- 238000004088 simulation Methods 0.000 claims abstract description 13
- 239000003795 chemical substances by application Substances 0.000 claims abstract description 6
- 238000001514 detection method Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 23
- 238000005070 sampling Methods 0.000 claims description 20
- 238000003062 neural network model Methods 0.000 claims description 13
- 230000008859 change Effects 0.000 claims description 11
- 238000004519 manufacturing process Methods 0.000 claims description 11
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 238000002474 experimental method Methods 0.000 claims description 8
- 238000012549 training Methods 0.000 claims description 7
- 230000006872 improvement Effects 0.000 claims description 6
- 238000010276 construction Methods 0.000 claims description 5
- 230000004044 response Effects 0.000 claims description 5
- 230000001502 supplementing effect Effects 0.000 claims description 5
- 238000000540 analysis of variance Methods 0.000 claims description 4
- 238000002347 injection Methods 0.000 claims description 4
- 239000007924 injection Substances 0.000 claims description 4
- 238000012544 monitoring process Methods 0.000 claims description 4
- 238000012797 qualification Methods 0.000 claims description 4
- 238000013398 bayesian method Methods 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 2
- 238000013507 mapping Methods 0.000 abstract 1
- 238000005259 measurement Methods 0.000 abstract 1
- 230000035945 sensitivity Effects 0.000 abstract 1
- 230000000007 visual effect Effects 0.000 description 3
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000013401 experimental design Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000012778 molding material Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C45/00—Injection moulding, i.e. forcing the required volume of moulding material through a nozzle into a closed mould; Apparatus therefor
- B29C45/17—Component parts, details or accessories; Auxiliary operations
- B29C45/76—Measuring, controlling or regulating
- B29C45/78—Measuring, controlling or regulating of temperature
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C45/00—Injection moulding, i.e. forcing the required volume of moulding material through a nozzle into a closed mould; Apparatus therefor
- B29C45/17—Component parts, details or accessories; Auxiliary operations
- B29C45/76—Measuring, controlling or regulating
- B29C45/77—Measuring, controlling or regulating of velocity or pressure of moulding material
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C2945/00—Indexing scheme relating to injection moulding, i.e. forcing the required volume of moulding material through a nozzle into a closed mould
- B29C2945/76—Measuring, controlling or regulating
- B29C2945/76494—Controlled parameter
- B29C2945/76498—Pressure
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C2945/00—Indexing scheme relating to injection moulding, i.e. forcing the required volume of moulding material through a nozzle into a closed mould
- B29C2945/76—Measuring, controlling or regulating
- B29C2945/76494—Controlled parameter
- B29C2945/76531—Temperature
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B29—WORKING OF PLASTICS; WORKING OF SUBSTANCES IN A PLASTIC STATE IN GENERAL
- B29C—SHAPING OR JOINING OF PLASTICS; SHAPING OF MATERIAL IN A PLASTIC STATE, NOT OTHERWISE PROVIDED FOR; AFTER-TREATMENT OF THE SHAPED PRODUCTS, e.g. REPAIRING
- B29C2945/00—Indexing scheme relating to injection moulding, i.e. forcing the required volume of moulding material through a nozzle into a closed mould
- B29C2945/76—Measuring, controlling or regulating
- B29C2945/76494—Controlled parameter
- B29C2945/76595—Velocity
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/04—Constraint-based CAD
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/08—Probabilistic or stochastic CAD
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Mechanical Engineering (AREA)
- Biomedical Technology (AREA)
- Manufacturing & Machinery (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Injection Moulding Of Plastics Or The Like (AREA)
Abstract
The invention discloses an intelligent injection molding process parameter adjusting method based on a mixed model. Obtaining initial data of a Kriging proxy model through mapping simulation of defects and technological parameters of an injection molding product, taking time measurement data of an injection molding machine and Kriging fitting data as follow-up data, and defining a minimum signal-to-noise ratio function; defining an optimal process parameter combination fitness function; defining a process parameter contribution rate; determining optimal initial technological parameters; constructing a GA-BP model with high prediction accuracy; constructing a global optimization Kriging agent model based on EGO; dynamic adjustment of process parameters was performed using the Kriging model and the GA-BP model. The method can solve the problems of high calculation precision of the Kriging model and sensitivity to noise data, and combines a prediction model and a machine vision detection means to realize dynamic adjustment of injection molding process parameters and effectively improve the adjustment efficiency and the automation degree in the injection molding process.
Description
Technical Field
The invention belongs to the technical field of machine learning and industrial automation, and particularly relates to an intelligent injection molding process parameter adjusting method based on a hybrid model.
Technical Field
Injection molding is a complex process with multiple variables, distribution parameters, intermittent operation, large hysteresis, strong coupling, nonlinearity and strong dispersibility, and the factors influencing the quality of molded injection products are many and are roughly divided into injection molding machine parameters, injection molding material parameters, injection molding process parameters and disturbance. The process parameters are important regulating objects in the injection molding process and are determining factors of the quality of injection molding products.
Due to multi-parameter coupling of injection molding, the adjustment of the technological parameters of injection molding of the existing injection molding machine is mostly carried out by experience and expertise accumulated by skilled workers, so that the production quality of injection molding products is high in randomness and the adjustment efficiency is low.
Therefore, on the premise that the product mold and the machine are selected, how to effectively replace the physical model of the injection molding process with multiple independent variables and multiple uncontrollable factors by using a data-driven proxy model, and simultaneously set a dynamic regulation strategy to establish a dynamic regulation frame, monitor and feed back the product quality state in real time, and dynamically regulate the process parameters in real time through an injection molding process parameter regulation system is a key for improving the quality and efficiency of the injection molding process.
Disclosure of Invention
In order to solve the problems in the background art, the invention provides an intelligent injection molding process parameter adjusting method based on a mixed model
The technical scheme of the invention comprises the following steps:
step 1: randomly sampling values in an empirical range of technological parameters required in an injection molding process by using Latin hypercube sampling, and combining the randomly sampled values by using a field orthogonal method to obtain a plurality of groups of experimental schemes; simulating the experimental scheme by using Moldflow to obtain a target value, performing analysis of variance on a simulation result, and obtaining contribution rates of all process parameters so as to determine main process parameters affecting the quality of injection molding products;
step 2: constructing a BP-GA neural network model by using the main process parameters of the step 1;
step 3: constructing a global optimization Kriging agent model based on EGO;
step 4: starting injection molding production products, and periodically detecting the injection molding products in the production process to obtain a plurality of groups of process parameter combination data and corresponding target values;
step 5: selecting effective data from the multiple groups of process parameter combination data in the step 4 to calculate a signal-to-noise ratio, updating a GA-BP neural network model, and meanwhile, carrying out targeted dotting through a Latin hypercube sampling and EGO global optimization method, supplementing and updating training data of a Kriging model, and retraining the Kriging prediction model;
step 6: visual monitoring is carried out on a target value of an injection molding product in the production process in real time, whether the target value exceeds a set threshold value is checked, the process parameter corresponding to the target value exceeding the threshold value is corrected, the corrected process parameter is input into a GA-BP neural network model, the signal-to-noise ratio value is predicted, whether the signal-to-noise ratio accords with a set range is judged, and if the signal-to-noise ratio exceeds the threshold value, the process parameter correction is continued until the signal-to-noise ratio requirement is met;
step 7: inputting the technological parameter combination meeting the signal-to-noise ratio requirement obtained in the step 6 into a Kriging model, predicting a target value of the injection molding product and judging whether the target value meets a set threshold value or not:
if not, returning to the step 6 to continuously correct the technological parameters;
if so, obtaining a process parameter combination which meets both the signal-to-noise ratio requirement and the quality standard;
step 8: carrying out process parameter adjustment on the injection molding machine by using the process parameter combination finally obtained in the step 7, comparing the actual defect detection result of the injection molding machine after adjustment with the target value predicted in the step 7, and correcting after feeding back the error to the Kriging model;
and 6, forming a dynamic adjusting system of injection molding process parameters through the steps 6-8.
In the step 1), the process parameter contribution rate θ is calculated by the following formula:
in SS (x) t Is the sum of the squares, SS i SS is the sum of squares of the process parameters e Is the sum of squares of errors, y is a response value, is the sum of indexes of all experiments, T i The total experimental times are calculated for the experimental indexes of the same level of each technological parameter, N is the number of factors; DF (DF) i ＝a i -1，DF T ＝N-1，a i A horizontal number that is an ith factor; f is the statistic, MSB is the inter-group variance, MSE is the intra-group variance;
contribution rate θ=seqss i /∑SeqSS i ；
In the method, in the process of the invention,
wherein s is j The defect normalized average value is represented, n is the number of the process parameters, and i represents the ith process parameter;
the main process parameters in the step 1 are process parameters with a contribution rate of more than 5%.
The target value comprises other main quality judgment bases such as defects, glossiness and the like of the injection molding product.
The step 2 specifically comprises the following steps:
2.1 Main technological parameters (melt temperature x) 1 Die temperature x 2 Injection rate x 3 Dwell time x 4 Pressure maintaining pressure x 5 ) As the input of BP-GA neural network, train BP-GA neural network model with corresponding signal-to-noise ratio value as output;
2.2 Increasing the number of samples): randomly sampling in the range of a technological parameter interval by using a Latin hypercube sampling method to obtain a technological parameter combination as an added sample, and obtaining a corresponding target value by using simulation or actual experiments;
2.3 Optimizing BP-GA neural network structure: the model structure is simplified by reducing hidden layer nodes, and the number of the hidden layer nodes is determined specifically by the following formula:
h＝log 2 m
wherein h is the number of hidden layer nodes; m is the number of input layer nodes; l is the number of output layer nodes; alpha is a constant, and an integer of 1 to 10 is taken;
2.4 The weight and the threshold of the model are optimized by using a Bayesian method, so that the model fitting accuracy is ensured, the prediction accuracy meets the requirement, and the construction of the BP-GA neural network model is completed.
The step 3 specifically comprises the following steps:
3.1 Constructing a Kriging model using the initial samples:
the initial sample comprises main technological parameters determined in the step 1, technological parameters added by a Latin hypercube sampling method in the step 2, and target values corresponding to all the technological parameters;
3.2 For sample data X) * Carrying out normalization treatment;
3.3 Using latin hypercube sampling and EGO global optimization to obtain new sample points:
latin hypercube sampling is carried out within the technological parameter range to obtain M new sample points;
the new sample point is determined by minimizing the response surface and maximizing the desired improvement function, as follows:
wherein x is (i) Representing the i new sample point, including characteristics of process parameter combinations and target values, i.e {1,2, … M }, x (i) ∈X＝{x (1) ，...，x (i) ，...x (M) }；y(x (i) ) For sample x (i) Target values obtained after experiments or simulations;
the improvement amount is defined as i=max (y * -y(x (i) ) 0), y is the target value of the sample
It is desirable to improve the function EI (x (i) ) The method comprises the following steps:
wherein CDF and PDF are accumulated distribution functions and probability density functions; y (μ (x)) is the sample data of step 3.2 plus the sample point x (i) Is obtained using a kriging model, sigma (x) is the sample data of step 3.2) plus the sample point x) (i) Is a process parameter of (2)Is a variance of (2);
selecting the corresponding x when EI (x) is maximum (i) As a new sample point x * Will x * Adding X * Obtain a new set X * Using the updated sample data set X * Re-fitting the Kriging agent model to finish updating;
3.4 Prediction error term for updated Kriging model:
the Kriging model is optimized for minimum prediction error, and the objective function is as follows:
in the method, in the process of the invention,as an objective function, y is the target value of the sample, +.>A predicted value that is a sample target value; g j (x) Is an objective function->Is a constraint function of (2); />The process parameters x in the samples x are respectively i Upper and lower limits of X e X * ；N c Is the number of constraints;
3.5 When an objective functionRepeating 3.3) to 3.4) when the target value is not smaller than the set target value;
when the objective functionAnd (3) stopping updating the Kriging model when the set target value is smaller than the set target value, so that the construction of the global optimization Kriging model based on EGO is completed.
In the step 2 and the step 5, the signal-to-noise ratio corresponding to the technological parameter is obtained through the following signal-to-noise ratio function:
wherein, SNR is signal-to-noise ratio;average value of all sample target values; />For sample point x (i) Target value under the m-th repeated simulation experiment; t is a target value; s is S 2 Is the variance;
where N represents the number of experimental replicates.
The effective data in the step 5 is the technological parameters corresponding to the time period when the injection molding machine enters stable production, namely the sample data X'.
In the step 5:
the method for updating the GA-BP neural network model comprises the steps of 2.3) to 2.4);
the method for supplementing and updating the training data of the Kriging model is steps 3.3) to 3.5).
In the step 6, the process parameters are corrected according to the empirical rules, the dynamic rules and the summarized qualitative rules, and the specific rules are as follows:
setting each process parameter x i Conservation threshold of (2)Qualification threshold->And a variability threshold->Calculating a detection period T 1 Time series of process parameters in->Rate of change sequence->And overall rate of change->Setting an adjustment coefficient theta, wherein the theta is set according to experience, and theta epsilon (-1, 1);
a) When the target value to be corrected does not exceed the conservation threshold value, directly judging that the process parameter adjustment is not performed;
b) When the target value to be corrected is larger than the conservation threshold value and smaller than or equal to the qualification threshold value, judging the degree of change of the process parameters: when the change degree is smaller than the change degree threshold value, the injection molding process parameter adjustment is not performed; otherwise, the technological parameters are subjected to technological parameter fine adjustment, and the adjustment basis is as follows:
will beLast time length T 1 Multiplying a sequence of (2) by θ to obtain +.>And use->Taking the interval t of the time sequence as a basis, taking N intervals as a period, taking the corresponding technological parameter of each interval as an adjusting target, and gradually adjusting each technological parameter;
c) When the target value to be corrected is larger than the qualified threshold value, skipping the change degree judgment condition, and adjusting the process parameters according to the following conditions: and carrying out multi-objective optimization on points on the kriging prediction model to obtain a process parameter combination which optimizes the target value to be corrected.
The invention has the beneficial effects that:
according to the method, the manually driven injection molding process parameters can be adjusted and updated into the data driven injection molding process parameters, the product quality data is obtained through real-time monitoring, the dynamic process parameters of the injection molding product are adjusted, and the quality and efficiency of the injection molding process are effectively improved.
Drawings
FIG. 1 is a flow chart for determining initial process parameters of injection molding;
FIG. 2 is a flow chart for dynamic adjustment of injection molding process parameters.
Detailed Description
The invention is further described below with reference to the drawings and examples.
As shown in fig. 1:
step 1: setting a signal-to-noise ratio function:
wherein, the SNR is the signal-to-noise ratio,the target value (the defect value of the injection molding product corresponding to the process parameter) under the mth repeated simulation experiment is referred to, and t is the target value;
wherein N represents the number of repetitions;
step 2: defining an optimal process parameter combination fitness function:
s.t.
LS j ≤x j ≤US j ；j＝1,2,…n
wherein T (X) is an objective function; s represents the total number of defects; p (P) sni A signal-to-noise ratio predicted value for the ith defect; SN (SN) i The highest signal-to-noise value for the ith defect value; x is x j Is a technological parameter of the injection molding process; LS (least squares) j ，US j The lower limit and the upper limit of the technological parameters are respectively; n is the total number of process parameters;
step 3: defining a process parameter contribution rate:
in SS (x) t Is the sum of the squares, SS i SS is the sum of squares of the process parameters e Is the sum of squares of errors, y is a response value, T is the sum of indexes of all experiments, T i The total experimental times are calculated for the experimental indexes of the same level of each technological parameter, N is the number of factors; DF (DF) i ＝a i -1，DF T ＝N-1，a i A horizontal number that is an ith factor; f is the statistic, MSB is the inter-group variance, MSE is the intra-group variance；
Contribution θ=seq SS i /∑Seq SS i ；s j Represents a defect normalized average.
Step 4: determining optimal initial technological parameters based on a field orthogonal method and analysis of variance;
4.1: combining all the process parameters by a field orthogonal method to obtain a plurality of groups of experimental schemes, simulating the experimental schemes by using Moldflow, performing variance analysis on simulation results, and obtaining the contribution rate of all the process parameters and main process parameters affecting the quality of the product;
the contribution rate of the main technological parameters is more than 5%, and technological parameters with the contribution rate less than 5% are excluded;
4.2: putting the orthogonal test process parameter combination and corresponding signal-to-noise ratio (S/N) data into a BP-GA neural network for training, and establishing a prediction model, wherein the specific table is shown below;
process parameter calculation table
4.3: calculating the contribution rate of each technological parameter:
analysis of variance results table
The minimum signal to noise ratio is predicted by BP-GA, and the corresponding process parameter combination is obtained and used as the injection molding initial process parameter combination.
Step 5: constructing a GA-BP model and improving the prediction accuracy of the model;
the GA-BP network construction in the step 5 specifically comprises the following steps:
5.1: the main technological parameters (melt temperature x 1 Die temperature x 2 Injection rate x 3 Dwell time x 4 Pressure maintaining pressure x 5 ) As BP neural network input, taking the signal-to-noise ratio value as output, training BP-GA neural network model;
5.2: randomly sampling in the range of a technological parameter interval by using a Latin hypercube sampling method to obtain a technological parameter combination, and increasing the number of samples;
5.3: optimizing the neural network structure, reducing hidden layer joints and simplifying the model structure;
determining the number of hidden layer nodes:
h＝log 2 m
wherein: h is the number of hidden layer nodes, n is the number of input layer nodes, s is the number of output layer nodes, alpha is a constant, and an integer of 1-10;
5.4: the weight and the threshold of the model are optimized by using a Bayesian method, so that the model fitting accuracy is ensured, and the prediction accuracy meets the requirement
Step 6: the global optimization Kriging agent model based on EGO is constructed, which comprises the following steps:
6.1: constructing a Kriging model by using the initial discrete data samples;
the initial sample is all the technological parameter combinations of the step 4 and the step 5 and the corresponding defect values;
6.2: normalization processing is carried out on sample data X:
6.3: new sample points are obtained using latin hypercube sampling and EGO global optimization:
new sample points are obtained by minimizing the response surface and maximizing the desired improvement function (EI), as follows:
wherein: x is x (i) Represents the i new sample point, i ε {1,2, … M }, x (i) ∈X＝{x (1) ，...，x (i) ，...x (M) }；y(x (i) ) For sample x (i) Target values obtained after experiments or simulations;
the improvement amount is defined as i=max (y * -y(x (i) ) 0), y is the target value of the sample
The desired improvement function is:
wherein CDF and PDF refer to cumulative distribution functions and probability density functions;
selecting the corresponding x when EI (x) is maximum (i) As a new sample point x * Will x * Adding X * ；
6.4: predicting error terms for the updated Kriging model:
the Kriging model is optimized to minimize the prediction error, and the objective function is as follows:
in the method, in the process of the invention,is an objective function; g i (x) Is a constraint function; x is x u ，x l The upper and lower limits of the process parameter x, respectively; n (N) c Is the number of constraints;
6.5: repeating the steps of 6.3 to 6.4 until the objective functionAnd (3) if the set target value is smaller, stopping updating the Kriging model.
Step 7: dynamic adjustment of process parameters was performed using the Kriging model and the GA-BP model.
As shown in fig. 2, the dynamic adjustment method in step 7 specifically includes the following steps:
7.1, selecting an injection molding product, performing experimental design with multi-process parameters as variables and multi-product defect types as indexes through field orthogonal experiments, obtaining simulation data through Moldflow, determining main process parameters of various defects and significance of influence of the process parameters on various defects of the product, and determining significant influence factor combinations of the multi-defect type indexes according to results;
7.2, carrying out actual tests to obtain real data, and starting injection molding to produce products;
7.3, periodically detecting injection molding products in the production process to obtain a plurality of groups of technological parameter combination data and corresponding defect values, and selecting historical data corresponding to a time period when the injection molding machine enters stable production as effective data; calculating a signal-to-noise ratio through effective data, updating the GA-BP neural network at intervals, and meanwhile, carrying out targeted dotting through a Latin hypercube sampling and EGO optimizing method, periodically supplementing and updating training data of a Kriging model, and retraining the Kriging prediction model;
7.4, carrying out visual monitoring on the defect condition of the injection molding product in real time, if a certain defect or a plurality of defects exceed a set threshold value, obtaining current technological parameters and defect value data, correcting the technological parameters according to experience rules, dynamic rules and summarized qualitative rules, inputting the corrected technological parameters into a GA-BP signal-to-noise ratio model, predicting signal-to-noise ratio values and judging whether the signal-to-noise ratio accords with a set range, and if the defect or the plurality of defects exceed the threshold value, continuing to correct the technological parameters until the signal-to-noise ratio requirements are met;
7.5, bringing the obtained process parameter combination meeting the signal-to-noise ratio requirement into a Kriging model, predicting the defect value of the injection molding product and judging whether the defect value meets a set threshold value, and circulating the steps to obtain the process parameter combination meeting the signal-to-noise ratio requirement and the quality standard;
adjusting technological parameters of the injection molding machine, comparing the adjusted visual detection result with the prediction result, feeding back an error to the prediction model, and correcting the prediction model;
7.6 combining steps 7.4 and 7.5 forms a dynamic adjustment system for injection molding process parameters.
While the invention has been described with reference to preferred embodiments, it is not intended to be limiting. Those skilled in the art will appreciate that various modifications and adaptations can be made without departing from the spirit and scope of the present invention. Accordingly, the scope of the invention is defined by the appended claims.
Claims (9)
1. The intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of:
step 1: randomly sampling values in an empirical range of technological parameters required in an injection molding process by using Latin hypercube sampling, and combining the randomly sampled values by using a field orthogonal method to obtain a plurality of groups of experimental schemes; simulating the experimental scheme by using Moldflow to obtain a target value, performing analysis of variance on a simulation result, and obtaining contribution rates of all process parameters so as to determine main process parameters affecting the quality of injection molding products;
step 2: constructing a BP-GA neural network model by using the main process parameters of the step 1;
step 3: constructing a global optimization Kriging agent model based on EGO;
step 4: starting injection molding production products, and periodically detecting the injection molding products in the production process to obtain a plurality of groups of process parameter combination data and corresponding target values;
step 5: selecting effective data from the multiple groups of process parameter combination data in the step 4 to calculate a signal-to-noise ratio, updating a GA-BP neural network model, and meanwhile, carrying out targeted dotting through a Latin hypercube sampling and EGO global optimization method, supplementing and updating training data of a Kriging model, and retraining the Kriging prediction model;
step 6: monitoring a target value of an injection molding product in a production process in real time, checking whether the target value exceeds a set threshold value, correcting a process parameter corresponding to the target value exceeding the threshold value, inputting the corrected process parameter into a GA-BP neural network model, predicting a signal-to-noise ratio value, judging whether the signal-to-noise ratio accords with a set range, and if the signal-to-noise ratio exceeds the threshold value, continuing to correct the process parameter until the signal-to-noise ratio requirement is met;
step 7: inputting the technological parameter combination meeting the signal-to-noise ratio requirement obtained in the step 6 into a Kriging model, predicting a target value of the injection molding product and judging whether the target value meets a set threshold value or not:
if not, returning to the step 6 to continuously correct the technological parameters;
if so, obtaining a process parameter combination which meets both the signal-to-noise ratio requirement and the quality standard;
step 8: carrying out process parameter adjustment on the injection molding machine by using the process parameter combination finally obtained in the step 7, comparing the actual defect detection result of the injection molding machine after adjustment with the target value predicted in the step 7, and correcting after feeding back the error to the Kriging model;
and 6, forming a dynamic adjusting system of injection molding process parameters through the steps 6-8.
2. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 1, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of: in the step 1), the process parameter contribution rate θ is calculated by the following formula:
contribution θ=seq SS i /∑Seq SS i ；
In the method, in the process of the invention,
wherein s is j The defect normalized average value is represented, n is the number of the process parameters, and i represents the ith process parameter;
the main process parameters in the step 1 are process parameters with a contribution rate of more than 5%.
3. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 1, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of: the target values include defects, gloss of the injection molded article.
4. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 1, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of: the step 2 specifically comprises the following steps:
2.1 Main technological parameters are selected as the input of the BP-GA neural network, and the corresponding signal to noise ratio value is used as the output to train the BP-GA neural network model;
2.2 Increasing the number of samples): randomly sampling in the range of a technological parameter interval by using a Latin hypercube sampling method to obtain a technological parameter combination as an added sample, and obtaining a corresponding target value by using simulation or actual experiments;
2.3 Optimizing BP-GA neural network structure: the model structure is simplified by reducing hidden layer nodes, and the number of the hidden layer nodes is determined specifically by the following formula:
h＝log 2 m
wherein h is the number of hidden layer nodes; m is the number of input layer nodes; l is the number of output layer nodes; alpha is a constant, and an integer of 1 to 10 is taken;
2.4 The weight and the threshold of the model are optimized by using a Bayesian method, so that the model fitting accuracy is ensured, the prediction accuracy meets the requirement, and the construction of the BP-GA neural network model is completed.
5. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 1, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of: the step 3 specifically comprises the following steps:
3.1 Constructing a Kriging model using the initial samples:
the initial sample comprises main technological parameters determined in the step 1, technological parameters added by a Latin hypercube sampling method in the step 2, and target values corresponding to all the technological parameters;
3.2 For sample data X) * Carrying out normalization treatment;
3.3 Using latin hypercube sampling and EGO global optimization to obtain new sample points:
latin hypercube sampling is carried out within the technological parameter range to obtain M new sample points;
the new sample point is determined by minimizing the response surface and maximizing the desired improvement function, as follows:
wherein x is (i) Representing the i new sample point, including characteristics of the process parameter combination and target values,i∈{1，2，…M}，x (i) ∈X＝{x (1) ，...，x (i) ，...x (M) }；y(x (i) ) For sample x (i) Target values obtained after experiments or simulations;
it is desirable to improve the function EI (x (i) ) The method comprises the following steps:
wherein CDF and PDF are accumulated distribution functions and probability density functions; y (μ (x)) is the sample data of step 3.2 plus the sample point x (i) Is obtained using a kriging model, sigma (x) is the sample data of step 3.2) plus the sample point x) (i) Variance of the process parameters of (a);
selecting the corresponding x when EI (x) is maximum (i) As a new sample point x * Will x * Adding X * Obtain a new set X * Using the updated sample data set X * Re-fitting the Kriging agent model to finish updating;
3.4 Prediction error term for updated Kriging model:
the Kriging model is optimized for minimum prediction error, and the objective function is as follows:
in the method, in the process of the invention,as an objective function, y is the target value of the sample, +.>A predicted value that is a sample target value; g j (x) Is an objective function->Is a constraint function of (2); />The process parameters x in the samples x are respectively i Upper and lower limits of X e X * ；N c Is the number of constraints;
3.5 When an objective functionRepeating 3.3) to 3.4) when the target value is not smaller than the set target value;
when the objective functionAnd (3) stopping updating the Kriging model when the set target value is smaller than the set target value, so that the construction of the global optimization Kriging model based on EGO is completed.
6. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 4, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized in that: the signal-to-noise ratio corresponding to the process parameters is obtained through the following signal-to-noise ratio function:
wherein, SNR is signal-to-noise ratio;average value of all sample target values; />The target value of the sample point under the mth repeated simulation experiment is set; t is a target value; s is S 2 Is the variance;
where N represents the number of experimental replicates.
7. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 1, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of: the effective data in the step 5 is the technological parameters corresponding to the time period when the injection molding machine enters stable production, namely the sample data X'.
8. The intelligent injection molding process parameter adjusting method based on the mixed model according to any one of claims 4 and 5, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized in that: in the step 5:
the method for updating the GA-BP neural network model comprises the steps of 2.3) to 2.4);
the method for supplementing and updating the training data of the Kriging model is steps 3.3) to 3.5).
9. The intelligent injection molding process parameter adjusting method based on the mixed model as claimed in claim 1, wherein the intelligent injection molding process parameter adjusting method based on the mixed model is characterized by comprising the following steps of: in the step 6, the process parameters are corrected according to the empirical rules, the dynamic rules and the summarized qualitative rules, and the specific rules are as follows:
setting each process parameter x i Conservation threshold of (2)Qualification threshold->And a degree of change thresholdCalculating a detection period T 1 Time series of process parameters in->Rate of change sequence->And overall rate of changeSetting an adjustment coefficient theta, wherein the theta is set according to experience, and theta epsilon (-1, 1);
a) When the target value to be corrected does not exceed the conservation threshold value, directly judging that the process parameter adjustment is not performed;
b) When the target value to be corrected is larger than the conservation threshold value and smaller than or equal to the qualification threshold value, judging the degree of change of the process parameters: when the change degree is smaller than the change degree threshold value, the injection molding process parameter adjustment is not performed; otherwise, the technological parameters are subjected to technological parameter fine adjustment, and the adjustment basis is as follows:
will beLast time length T 1 Multiplying a sequence of (2) by θ to obtain +.>And use->Taking the interval t of the time sequence as a basis, taking N intervals as a period, taking the corresponding technological parameter of each interval as an adjusting target, and gradually adjusting each technological parameter;
c) When the target value to be corrected is larger than the qualified threshold value, skipping the change degree judgment condition, and adjusting the process parameters according to the following conditions: and carrying out multi-objective optimization on points on the kriging prediction model to obtain a process parameter combination which optimizes the target value to be corrected.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310862943.4A CN116852665A (en) | 2023-07-13 | 2023-07-13 | Injection molding process parameter intelligent adjusting method based on mixed model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310862943.4A CN116852665A (en) | 2023-07-13 | 2023-07-13 | Injection molding process parameter intelligent adjusting method based on mixed model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116852665A true CN116852665A (en) | 2023-10-10 |
Family
ID=88226545
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310862943.4A Pending CN116852665A (en) | 2023-07-13 | 2023-07-13 | Injection molding process parameter intelligent adjusting method based on mixed model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116852665A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117644625A (en) * | 2024-01-30 | 2024-03-05 | 陕西美伦包装有限公司 | Intelligent injection molding method based on machine vision |
| CN117681400A (en) * | 2024-01-31 | 2024-03-12 | 苏州宝富塑料制品有限公司 | Plastic corner protector forming treatment system based on ABS resin composite material |
-
2023
- 2023-07-13 CN CN202310862943.4A patent/CN116852665A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117644625A (en) * | 2024-01-30 | 2024-03-05 | 陕西美伦包装有限公司 | Intelligent injection molding method based on machine vision |
| CN117644625B (en) * | 2024-01-30 | 2024-04-05 | 陕西美伦包装有限公司 | Intelligent injection molding method based on machine vision |
| CN117681400A (en) * | 2024-01-31 | 2024-03-12 | 苏州宝富塑料制品有限公司 | Plastic corner protector forming treatment system based on ABS resin composite material |
| CN117681400B (en) * | 2024-01-31 | 2024-04-16 | 苏州宝富塑料制品有限公司 | Plastic corner protector forming treatment system based on ABS resin composite material |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116852665A (en) | Injection molding process parameter intelligent adjusting method based on mixed model | |
| CN107728477A (en) | A kind of industrialized aquiculture water quality dissolved oxygen prediction control method and system | |
| CN101863088A (en) | Method for forecasting Mooney viscosity in rubber mixing process | |
| CN111985825B (en) | Crystal face quality assessment method for roller mill orientation instrument | |
| CN112734128A (en) | 7-day power load peak value prediction method based on optimized RBF | |
| CN113095550A (en) | Air quality prediction method based on variational recursive network and self-attention mechanism | |
| CN110046377B (en) | Selective integration instant learning soft measurement modeling method based on heterogeneous similarity | |
| CN115983721B (en) | Cable production quality data management system based on big data of Internet of things | |
| US20230168643A1 (en) | Industrial internet of things based on abnormal identification, control method, and storage media thereof | |
| CN115049024A (en) | Training method and device of wind speed prediction model, electronic equipment and storage medium | |
| CN101285816A (en) | Copper matte air refining procedure parameter soft sensing instrument and its soft sensing method | |
| CN114707692A (en) | Wetland effluent ammonia nitrogen concentration prediction method and system based on hybrid neural network | |
| CN111898799B (en) | BFA-Elman-based power load prediction method | |
| CN111639680B (en) | Identity recognition method based on expert feedback mechanism | |
| CN112699229A (en) | Self-adaptive question-pushing method based on deep learning model | |
| CN115879750B (en) | Aquatic seedling environment monitoring management system and method | |
| CN110542748B (en) | Knowledge-based robust effluent ammonia nitrogen soft measurement method | |
| CN116880201A (en) | Water network channel state control system based on fuzzy neural network | |
| CN116432822A (en) | Carbon emission data prediction method, system, equipment and readable storage medium | |
| CN112651444B (en) | Self-learning-based non-stationary process anomaly detection method | |
| CN110648023A (en) | Method for establishing data prediction model based on quadratic exponential smoothing improved GM (1,1) | |
| CN115169453A (en) | Hot continuous rolling width prediction method based on density clustering and depth residual error network | |
| CN115438309A (en) | Unmanned aerial vehicle engine health assessment method based on LSTM-SVR | |
| CN112613909A (en) | Agricultural product short-term price prediction method and device based on improved LSTM model | |
| CN111626465A (en) | New energy power short-term interval prediction method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |