CN107944147A - Shooting Technique optimization method and Shooting Technique based on GRNN neutral nets - Google Patents
Shooting Technique optimization method and Shooting Technique based on GRNN neutral nets Download PDFInfo
- Publication number
- CN107944147A CN107944147A CN201711211044.9A CN201711211044A CN107944147A CN 107944147 A CN107944147 A CN 107944147A CN 201711211044 A CN201711211044 A CN 201711211044A CN 107944147 A CN107944147 A CN 107944147A
- Authority
- CN
- China
- Prior art keywords
- neutral nets
- optimization
- parameters
- grnn
- species
- 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
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/18—Manufacturability analysis or optimisation for manufacturability
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Aviation & Aerospace Engineering (AREA)
- Automation & Control Theory (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computational Mathematics (AREA)
- Pure & Applied Mathematics (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Injection Moulding Of Plastics Or The Like (AREA)
Abstract
A kind of Shooting Technique optimization method based on GRNN neutral nets provided by the invention, comprises the following steps:Establish the CAE analysis model of injecting products;The molding proces s parameters of the CAE analysis model are emulated using CAE software, determine to influence the molding proces s parameters species that the injecting products produce injection defect;The molding proces s parameters species is converted into control parameter species;Based on the control parameter species, network training is carried out using GRNN neutral nets, obtains final optimization pass control parameter.The above-mentioned Shooting Technique optimization method based on GRNN neutral nets, network training is carried out using GRNN neutral nets, compared to other neutral nets, especially compared to BP neural network, the parameter of calling is few, network training speed, convergence improve, it is possible to increase the accuracy of molding proces s parameters optimization.
Description
Technical field
The present invention relates to Shooting Technique field, more particularly to a kind of Shooting Technique optimization side based on GRNN neutral nets
Method and Shooting Technique.
Background technology
In automotive trim plastic part, elongated heteromorpha product is relatively more, such as inner decoration strip, air louver grid, handrail bar product,
The problem of injection molding of these products is maximum is deformation, wherein in the majority with buckling deformation.The origin cause of formation of buckling deformation mainly has 4 kinds
Factor:Inhomogeneous cooling, contraction is uneven, molecularly oriented is inconsistent and wall and corner effect, and the influence ratio of 4 kinds of factors respectively has not again
Together, it is difficult to find out its clear and definite correspondence between injection molding machine injection controlling element, such a relation is nonlinear Control relation,
Analysis and control thus need to be optimized to this by means of neutral net, in the hope of as far as possible requiring Deformation control in product quality
It is interior.In addition, in product injection, the control of injection is actual to be determined by the rotating speed of screw in injection molding machine, and screw rod control actually appears
Control and screw rod reverse push feed flow path increment to the relative rotation speed of screw rod.The change in shape of combination product, ordinary circumstance
Under, according to the charging needs of stream each section on product, when electric injection molding machine is molded, and correspondingly control the phase of screw rod
To rotating speed and stroke is promoted, to reach the target of control product injection molding quality.Each section of screw speed and stroke promote with most
Also without clear and definite control relationship between whole warpage results control, traditional way is constantly repetition test, and efficiency is very
It is low and random larger, and after CAE sunykatuib analyses, rational control mould will be established between warpage results and screw speed control
Type relation, also need to be by means of the forecast function of neutral net.
Although 106055787 A of Chinese patent application CN disclose a kind of automotive trim panel based on BP neural network
Injection molding process, but based on BP (error backpropagation algorithm, Error Back Propagation Training) nerves
The initial trim panel injection molding process tune ginseng of network is more, and network training is slower, and convergence is poor.
The content of the invention
Based on this, it is necessary to join more, network for the automotive trim panel injection molding process tune based on BP neural network
The problem of training is slower, and convergence is poor, there is provided a kind of Shooting Technique optimization method and injection work based on GRNN neutral nets
Skill.
A kind of Shooting Technique optimization method based on GRNN neutral nets provided by the invention, comprises the following steps:
Establish the CAE analysis model of injecting products;
The molding proces s parameters of the CAE analysis model are emulated using CAE software, determine to influence the injection production
Product produce the molding proces s parameters species of injection defect;
The molding proces s parameters species is converted into control parameter species;
Based on the control parameter species, network training is carried out using GRNN neutral nets, obtains final optimization pass control ginseng
Number.
In one of which embodiment, the molding proces s parameters of the CAE analysis model are imitated using CAE software
Very, determine to influence the molding proces s parameters species that the injecting products produce injection defect, comprise the following steps:
The gate system of the injecting products is optimized using the CAE software.
In one of which embodiment, the gate system of the injecting products is optimized using the CAE software, including
Following steps:
Design the cast gate quantity of the injecting products and corresponding gate location;
The cast gate quantity of the injecting products and corresponding position are optimized using the CAE software;
Gate size and channel size are optimized using the CAE software, obtain gate system optimum results.
In one of which embodiment, the gate system Optimization Steps of the injecting products are being carried out using CAE software
Afterwards, it is further comprising the steps of:
Shaping window analysis and optimization are carried out to the result that the gate system optimizes using the CAE software;
The result after the shaping window analysis and optimization is filled using the CAE software, pressurize, warpage point
Analysis, determines to influence the molding proces s parameters species that the injecting products produce injection defect.
In one of which embodiment, the gate system Optimization Steps of the injecting products are being carried out using CAE software
Afterwards, it is further comprising the steps of:
Analysis, the optimization of shaping window are carried out to the result that the gate system optimizes using the CAE software;
Using the CAE software result after the shaping window analysis, optimization is filled, pressurize and warpage
Analysis, optimization, obtain the first optimization molding proces s parameters.
In one of which embodiment, the molding proces s parameters species include gate location, cold runner inlet temperature,
When mode temperature, injection pressure, injection time, dwell pressure, dwell time, cooling flow velocity, cooling inlet temperature and cooling
Between in any one or a few.
In one of which embodiment, the molding proces s parameters species is converted into control parameter species, is by institute
State cold runner inlet temperature and be converted into material temperature, be mould temperature by the mode temperature inversion, by the injection pressure, injection time,
Dwell pressure and dwell time are separately converted to the screw rod relative rotation speed and screw position of corresponding section.
In one of which embodiment, by the molding proces s parameters species be converted into control parameter species step it
Afterwards, it is further comprising the steps of:
Described first optimization molding proces s parameters are corresponded to and are converted into the first optimal control parameter.
In one of which embodiment, based on the control parameter species, network instruction is carried out using GRNN neutral nets
Practice, obtain final optimization pass control parameter, comprise the following steps:
Orthogonal Experiment and Design is carried out according to the control parameter species and the first optimal control parameter, by it is described just
Hand over experimental design result be filled using CAE software, pressurize, warping Analysis, acquisition be based on amount of warpage and weld bond quantity
Sample;
Some samples are chosen as training sample, net is carried out to the training sample using the GRNN neutral nets
Network training, obtains trained first GRNN neutral nets;
Some remaining samples are chosen as test samples, sample is examined using described in the first GRNN neutral nets
Originally it is predicted, obtains the first GRNN neural network prediction values;
The first GRNN neural network predictions value is compared with the CAE simulation results of corresponding test samples, is chosen opposite
Optimization control parameter;
Network instruction is carried out to the opposite optimization control parameter after secondary densification using the first GRNN neutral nets
Practice optimizing again, obtain optimizing result;
The optimizing result is handled using renormalization, obtains optimal molding proces s parameters.
Present invention also offers a kind of Shooting Technique, the Shooting Technique passes through the above-mentioned note based on GRNN neutral nets
Technique optimization method is moulded to obtain.
The above-mentioned Shooting Technique optimization method based on GRNN neutral nets, network training is carried out using GRNN neutral nets,
Compared to other neutral nets, especially compared to BP neural network, the parameter of calling is few, network training speed, convergence
Property improve, it is possible to increase molding proces s parameters optimization accuracy.Further, molding proces s parameters conversion is joined in order to control
Number, so that the final optimization pass control parameter obtained can be with direct applied Shooting Technique, without further according to molding proces s parameters
Corresponding adjustment control parameter, improves optimization efficiency.
Brief description of the drawings
Fig. 1 is the flow chart of the Shooting Technique optimization method based on GRNN neutral nets of one embodiment of the invention;
Fig. 2 is GRNN neural networks principles figure of the present invention;
Fig. 3 is the utilization GRNN neutral nets of the Shooting Technique optimization method based on GRNN neutral nets shown in Fig. 1
The training result figure of network training is carried out to the training sample;
Fig. 4 is utilization the first GRNN nerves of the Shooting Technique optimization method based on GRNN neutral nets shown in Fig. 1
The prediction result figure that test samples described in network are predicted.
Embodiment
Shown in please referring to Fig.1 to Fig.4, the Shooting Technique optimization side based on GRNN neutral nets of one embodiment of the invention
Method, comprises the following steps:
S100, establishes the CAE analysis model of injecting products;
S200, emulates the molding proces s parameters of CAE analysis model using CAE software, determines to influence injecting products
Produce the molding proces s parameters species of injection defect;
S300, control parameter species is converted into by molding proces s parameters species;
S400, based on control parameter species, network training is carried out using GRNN neutral nets, obtains final optimization pass control ginseng
Number.
The above-mentioned Shooting Technique optimization method based on GRNN neutral nets, utilizes GRNN neutral net (general regression neurals
Network, Genera-lized Regression Neural Network) network training is carried out, compared to other neutral nets,
Especially compared to BP neural network, the parameter of calling is few, network training speed, and convergence improves, it is possible to increase injection
The accuracy of process parameter optimizing, improves efficiency and effect that warpage improves.
Further, molding proces s parameters are converted into control parameter, so that the final optimization pass control parameter obtained can
With direct applied Shooting Technique, without corresponding to adjustment control parameter further according to molding proces s parameters, optimization efficiency is improved.
In the present embodiment, Shooting Technique optimization is carried out by taking car multimedia panel construction as an example.The automobile of the present embodiment
Multimedia panel shape is wing product, i.e., middle is main body panel, and upper and lower ends are extension empennage, and basic appearance and size is
275.4mm × 170mm × 116.1mm, small product size 20.27cm3, plastics selection ABS+PC, average wall thickness 2.5mm.Product
Mainly potential quality problems have during injection:1) in irregular shape, gate location, which opens up, to be needed to optimize;2) Injection Molding changes of section
Greatly, the generation of injection imbalance and warpage issues is caused;3) screw column position and buckle position are more, and stowing pressure is high and pressure is distributed
It is uneven;4) product appearance quality requires high, it is impossible to has the problems such as obvious weld mark and stomata.Thus, to ensure the note of product
Polymer amount, the present invention provides the method that combination CAE emulation technologies and GRNN neutral nets optimize molding proces s parameters, improves
Die trial efficiency, and ensure the stable quality requirement of product batch production.
In the present invention, the CAD stereochemical structures of injecting products are primarily based on, the CAD stereochemical structures of injecting products are carried out
After wall thickness analysis, CAE software is imported to establish the CAE analysis model of injecting products, and CAE emulation point is carried out easy to injecting products
Analysis.
Effect is preferably molded to obtain, CAE emulation is carried out before mold design is carried out to injecting products and optimizes analysis,
It mainly solves problems with:1) Design of Runner System problem, to obtain preferable cast effect, while to obtain preferable point
Type designing scheme;2) potential injection quality problems, such as warpage, stomata, weld bond, potential crack are observed;3) injection stablized
Technological parameter, to ensure the stability of product batch production quality.
Alternatively, it is of the invention using CAE software to the molding proces s parameters of CAE analysis model carry out emulation include it is following
Step:Determine CAE analysis mould;Gate system optimizes;Filling quality controls the optimization of molding proces s parameters;Electric injection molding machine is molded
When screw rod rotate control technological parameter optimization.
Still optionally further, can also comprise the following steps:Molding proces s parameters further optimize after injection molding machine die trial
Adjustment.
Alternatively, CAE software of the invention is Moldflow2015 softwares, using the software to CAE analysis model meshes
Emulated after division, grid selects two-sided layer triangular mesh.In the CAE analysis models of the present embodiment, triangular mesh
Sum is 112 624, number of nodes 56 312, grid total surface area 251.32cm2, small product size 20.3cm3, product throwing
Shadow area is 61.5cm2, grid aspect ratio maximum 9.6, minimum 1.15, average 1.64, mesh fitting rate 92.4%, can meet to stick up
Song analysis etc. requires.
Still optionally further, the injected plastics material that the present embodiment injecting products are selected is ABS+PC, trade mark HAC8244H
(Kumho Chemicals Inc).Injected plastics material recommend molding proces s parameters be:55 DEG C of mold surface temperature, melt temperature
250 DEG C, 40~70 DEG C of mold temperature range, 230~280 DEG C of melt temperature range, 320 DEG C of absolute maximum melt temperature, ejection
116 DEG C, the maximum shear stress 0.25MPa of temperature, maximum shear speed 10000 (1/s).Alternatively, select in the present embodiment
The injection molding machine of model HTF86X.
As a kind of optional embodiment, step S200 comprises the following steps:Cast gate using CAE software to injecting products
System optimization.
In the present invention, before carrying out network training using GRNN neutral nets, preliminary molding proces s parameters have been carried out
Optimization, for example, the selection of injected plastics material, the molding proces s parameters of species initial option according to injected plastics material, gate system
Optimization etc., can reduce the data processing amount using GRNN neutral nets, and summation improves optimization efficiency.
Still optionally further, comprised the following steps using gate system optimization of the CAE software to injecting products:
Design the cast gate quantity of injecting products and corresponding gate location;
The cast gate quantity of injecting products and corresponding position are optimized using CAE software;
The size and channel size of cast gate are optimized using CAE software, obtain gate system optimum results.
Cast gate quantity and the corresponding gate location of design injecting products in view of injecting products volume, parting, it is necessary to set
The convenience and injection cost of meter.In the present embodiment, cast gate is using single cast gate, side feeding manner.
It is corresponding to pour after further being optimized using CAE software to the cast gate quantity of injecting products and corresponding position
Mouth position is set for the eccentric.
Further the size and channel size of cast gate are optimized using CAE software, obtain gate system optimization knot
Fruit.
In the present embodiment, it is contemplated that injecting products small volume, its runner waste material should not be excessive, and runner is short favourable
In factors such as raising injection quality, plastics materials costs, therefore use cold runner+hot flow path mode.
Optimized by gate system, obtained including cast gate quantity, corresponding gate location, gate size and runner ruler
Very little gate system optimum results.
As a kind of optional embodiment, after the gate system Optimization Steps of injecting products are carried out using CAE software,
It is further comprising the steps of:
Shaping window analysis and optimization are carried out to the result that gate system optimizes using CAE software;
Using CAE software to shaping window analysis and optimization after result be filled, pressurize, warping Analysis, determine
Influence the molding proces s parameters species that injecting products produce injection defect.
Shaping window analysis and optimization is carried out to the result that gate system optimizes using CAE software, it is analyzed and excellent
For the result of change to be 46.7 DEG C in mould temperature, material temperature is 239.4 DEG C, when injection time is 0.9468s, shaping first water point
Number is 90%.Required pressure is 60.57MPa in corresponding die cavity, and melt front flowing temperature is cut between 229~230.1 DEG C
Cutting speed rate is 7027.1 (1/s), shear stress between 0.171MPa, cooling time 11.36s.So after parameter rounding,
47 DEG C of initial technological parameter modulus temperature, 239 DEG C of material temperature, injection time 0.95s.
Further, using CAE software to shaping window analysis and optimization after result be filled, pressurize, warpage
Analysis, determines to influence the molding proces s parameters species that injecting products produce injection defect.Alternatively, the first optimization is further obtained
Molding proces s parameters.
Alternatively, molding proces s parameters species include cold runner inlet temperature, mode temperature, injection pressure, injection time,
Dwell pressure, dwell time, cooling flow velocity, cooling inlet temperature and any one or a few in cooling time.
Required according to the control of the injecting products quality of production, the product after injection needs to ensure appearance zero defect, shape distortion
Deform it is small in favor of assembling, using CAE software after considering【Filling+pressurize+warpage】Scheme, based on foregoing shaping window
The analysis of mouth and optimum results, molding proces s parameters are specifically configured to:47 DEG C of mould temperature, 239 DEG C of material temperature, injection time
0.95s;Pressure switch points are 98.5%;Initial pressurize is arranged to two sections of pressurizes, is respectively 60MPa-8s, 30MPa-2s.Operation
After analysis, product stomata, flow direction, filling full weight, fuse, freeze situation etc. there is no it is potential the defects of ask
Topic, but potential injection problem is mainly reflected in that weld bond is more, warpage two main problem bigger than normal.
From the point of view of the distribution of weld bond, weld bond is concentrated and appears in injecting products everywhere, predominantly weld line.Wherein go out
It is located at the first existing weld bond, at the second weld bond, at the 3rd dissolving line at product cavity surface side, position at the 4th dissolving line
Weld bond quantity is more at core side, the second weld bond and at the 3rd dissolving line, and at the first weld bond and the 4th is molten
Then quantity is relatively fewer for connecting part, but lines is relatively long.For combination product requirement, these weld bonds not only influence to make
The presentation quality of product, and weld bond has an impact product strength, and when product japanning is when post processing, weld bond is difficult to locate
Reason.
The solution for shortening welding line length has two kinds:A kind of is the welding line position based on CAE software prediction, is changed
Product design or gate design, be partially moved to intensity and the presentation quality for making generation weld bond as far as possible are not important positions
Put;Another kind allows for weld bond and produces the place that molten resin collaborates in injection molding, thus improves injection pressure, melts
Running gate system effectively can will inject pressurize pressure to the flow resistance of melt when material temperature degree and mould-cavity temperature help to overcome injection
Power is delivered to stream front end, stream meet is merged at elevated pressures, so that density increases at weld mark, intensity carries
It is high;Therefore melt merges preferably, so as to reduce weld mark quantity and length.At the same time, it is contemplated that the requirement of article construction in itself makes
Weld bond/fusion trace is inevitable, but by controlling the two bursts of difference in melt temperature met (being no more than 10 DEG C), improving injection pressure
Can improve the intensity at melt fusion with technological parameter conditions such as mould-cavity temperatures.
From the point of view of the analysis result of warpage, in two tail ends of mould filling, the warpage of product compare it is larger, from entirety
From the point of view of deflection, both ends have respectively reached 3.94mm and 2.27mm.After element paritng, inhomogeneous cooling for 0.082~0.125mm
Between, shrink inequality between 3.21~3.86mm, be orientated inequality between 0.081~0.12mm, corner effect is
Between 0.05~0.08mm.From the point of view of the result of element paritng, the main cause that inequality is buckling deformation is shunk.
It is mainly related to the temperature in molding proces s parameters and the gate location in mold design that product produces weld bond.
With reference to production practices, the measure for welding line defect has:1) mobility of plastics is improved;2) injection moulding speed, increase note are accelerated
Blow pressure;3) heated near weld bond;4) mold temperature is improved;5) cast gate is increased;6) exhaust is increased on mould.In this reality
Apply in example, in the case that cast gate has been set, temperature optimization adjustment is preferred.
It is mainly (cold runner inlet temperature, mode temperature, cold with the injection temperature in molding proces s parameters that product produces warpage
But flow velocity, cooling inlet temperature and cooling time), injection moulding speed (injection time), injection pressure (injection pressure) and protect
Press (dwell pressure, dwell time) related.With reference to production practices, improved measure includes improving plastics performance, plastic component structure, mould
Lamps structure, molding parameter, demoulding subsequent treatment etc., the preferential means combination CAE analysis result that improves is from molding parameter in the present invention
Set about.
Above-mentioned influence injecting products produce the molding proces s parameters of injection defect and injection defect-weld bond and warpage
Without clear and definite mathematical model between evaluation quantizating index, it can be assumed that being non-linear relation, GRNN god will be passed through in the present invention
Network training is carried out through network, to obtain final optimization pass control parameter.
Determined for foregoing CAE simulation results and analysis, injecting products are influenced in the present embodiment and produce injection defect
Molding proces s parameters species has:
1) cold runner inlet temperature Tθ;
2) mode temperature Ts;
3) injection pressure PI;
4) injection time ti;
5) dwell pressure Ph;
6) dwell time th;
7) flow velocity V is cooled downt;
8) inlet temperature T is cooled downc;
9) t cooling timec。
Wherein, it is contemplated that actual influence acts on, cooling flow velocity Vt, cooling inlet temperature Tc, t cooling timecThree kinds cooling because
Element can be attributed to mode temperature TsInfluence.Therefore, top-priority technological parameter species is cold runner inlet temperature Tθ、
Mode temperature Ts, injection pressure PI, injection time ti, dwell pressure Ph, dwell time th.In addition, in pressurize related process parameters
Generally multistage pressurize, with reference to foregoing CAE as a result, two sections of pressurizes of the product are difficult to reach preferable pressure holding effect, thus at this
Pressurize is carried out in embodiment in a manner of the pressurize of three sections of pressurizes.
However, during actual injection,【Filling+pressurize+warpage】In filling control be arranged to [with respect to screw speed+screw rod
Position] after, Shooting Technique species injection pressure PI, injection time ti, dwell pressure Ph, dwell time thSpecially injection screw
Rotate with promoting stroke to control, thus need to be by injection pressure PI, injection time ti, dwell pressure Ph, dwell time thTransformation
For screw rod control parameter.
Screw rod control parameter is generally 8 sections of controls.First 5 sections are filling control, and latter 3 sections are holding pressure control.Wherein, screw rod control
The paragraph 1 of system is runner establishment stage, and the 2nd~5 section is injection filling section;6th~8 section is pressurize section.Injection pressure PI, injection
Time ti, change then depend on spiral shell before 5 sections filling control.Dwell pressure Ph, dwell time thChange then rely on after 3 sections
Holding pressure control.Therefore, correspondingly by injection pressure PI, injection time ti, dwell pressure Ph, dwell time thIt is converted into following
Screw rod control parameter:First screw rod relative rotation speed N1, the second screw rod relative rotation speed N2, the 3rd screw rod relative rotation speed N3, the 4th spiral shell
Bar relative rotation speed N4, substrate screw rod relative rotation speed N5, pressurize first segment pressure P1, pressurize first segment time t1, pressurize second segment
The 3rd section of pressure P2, pressurize second segment time t2, the 3rd section of P3 pressure of pressurize and pressurize time t3.Thus, by four kinds of injections
Technological parameter is converted into 11 kinds of control parameters.
Further, after molding proces s parameters species is converted into control parameter species step, further include following
Step:
First optimization molding proces s parameters are corresponded to and are converted into the first optimal control parameter.
As a kind of optional embodiment, based on control parameter species, network training is carried out using GRNN neutral nets, is obtained
Final optimization pass control parameter is taken, is comprised the following steps:
Orthogonal Experiment and Design is carried out according to control parameter species and the first optimal control parameter, by Orthogonal Experiment and Design knot
Fruit is filled using CAE software, pressurize, warping Analysis, obtain the sample based on amount of warpage and weld bond quantity;
Some samples are chosen as training sample, network training is carried out to training sample using GRNN neutral nets, is obtained
Trained first GRNN neutral nets;
Some remaining samples are chosen as test samples, are predicted using the first GRNN neutral net test samples,
Obtain the first GRNN neural network prediction values;
First GRNN neural network predictions value is compared with the CAE simulation results of corresponding test samples, is chosen relatively optimal
Control parameter;
Network training is carried out using the first GRNN neutral nets to the opposite optimization control parameter after secondary densification to seek again
It is excellent, obtain optimizing result;
Optimizing result is handled using renormalization, obtains optimal molding proces s parameters.
Based on CAE software【Filling+pressurize+warpage】Analysis result, filling control mode use【Screw rod relative rotation speed+
Screw position】Control mode, the level of each governing factor are divided into three levels, and level control is as shown in table 1.
1 governing factor parameter level of table
Orthogonal test conceptual design, in the case of without considering factor interaction, structure are carried out using Mintab softwares
The factor level neural network sample training built is with examining orthogonal array L27 (313).Each Orthogonal Composite parameter setting is inputted
CAE analysis scheme【Filling+pressurize+warpage】In, simulation analysis are carried out, acquisition is as shown in Table 2 to be based on amount of warpage and welding
The sample of line number amount.
Alternatively, choose sample 1 to 21 and be used as training sample, choose remaining sample 22 to 27 and be used as test samples.
GRNN neural networks principles that the present invention uses are as shown in figure 3, wherein, P- input vectors;Q- sample numbers;R- samples
Element;KW- weight matrixs;L- linear convergent rate layer weight matrixs;‖ dist ‖-offset distance function;B1- hidden layer critical values;n1、n2-
Output vector;A1, a2- linear transfer function;Nprod- output layer weight functions.
In the present invention, input vector P=[Tr Tθ N1 N2 N3 N4 N5 P1 t1 P2 t2 P3 t3];Output vector y
=[M G].Neuron number R=13, hidden layer neuron number Q=21, its weight function ‖ dist ‖ for Euclidean distance function such as
Shown in formula (1).
In formula:N1=‖ dist ‖ b1.
Intermediate layer transmission function is using Gaussian function shown in formula (2).
Output layer is specific linear layer, contains 21 neurons in interior, its node function is purely linear function, and LW is it
Corresponding weight matrix.GNRR networks are exported as shown in formula (3).
a2=purelin (LW × a1/suna1) (3)
It is the following spy based on GNRR networks when the amount of warpage and weld bond of the GRNN neural network predictions that the present invention is built
Point:Network structure is simple, adjusts ginseng few, and network training is than very fast, opposite good, the optimal extension constant with BP neural network convergence
Spread values, optimal output and input easily are to obtain.Using Matlab2015 implementation model functions, GNRR neural network predictions
Process is:
1) cross-training is carried out to 21 training samples in table 2, finds spread parameter optimal values, obtain training sample
Optimal input, output valve, obtain the GNRR neutral nets of the trained first GRNN neutral nets present invention in network training
When training sample is subjected to cross validation, first by the scope for the optimal spread values that institute's optimizing goes out be arranged on 0.005~1 it
Between, the optimal spread values that final optimizing obtains are 0.01, and calculate out optimal input and the output valve of training sample accordingly.
Then with it is optimal input, output valve and with optimal spread values construct warpage M, welding line number G trained first
GRNN neutral nets.
2) use 6 test samples in table 2 to be predicted trained GRNN neutral nets, obtain the first GRNN god
Through neural network forecast value.
Refer to shown in Fig. 3 to Fig. 4, the prediction result based on GRNN neutral nets to training sample and test samples,
Predicted value and the test samples value goodness of fit are higher, illustrate the network application in control technological parameter and warpage, weld bond result
Accuracy it is high.
3) by the first GNRR neural network predictions value and 6 test samples CAE analysis results contrasts, phase is chosen from table 1
Optimization control parameter is carried out to carry out secondary densification using L25 (56).
4) the relation control parameter optimizing again after secondary densification is obtained using trained first GNRR neutral nets
Optimizing result;
5) renormalization processing, which corresponds to optimizing, is reduced to corresponding technological parameter.
Alternatively, and will technological parameter input CAE analysis scheme in verified, with obtain less welding line number and compared with
Small amount of warpage.
Alternatively, by the GNRR neutral nets obtained by above-mentioned optimizing result, after secondary densification parameter densification is carried out
L25 (56) scheme carries out secondary optimization, and the warpage minimum value for predicting acquisition is 0.68mm, weld bond quantity 10.To input to
After measuring renormalization, the optimal procedure parameters of acquisition vector for P=[245,50,9,22,47.5,72,38,65,4,55,4,
38th, 2], optimizing acquisition technological factor carry out in input Moldflow【Filling+pressurize+warpage】After analysis, emulated
Prediction.
From the point of view of the result after Optimizing Process Parameters analysis, product weld bond sum is 7, is all melt run, to product
Intensity there is no influencing, also do not influence the appearance of product.The total deformation of warpage is less than 0.93mm, and maximum distortion is cast gate pair
The injection tail end of side, the total deformation of cast gate side tail end are less than 0.642mm.Buckling deformation situation in all directions is:1) in X-direction
Deflection be 0.576~(- 0.663) mm;Deflection in Y-direction is 0.771~(- 0.543) mm;Deformation in Z-direction
Measure as 0.215~(- 0.776) mm;Corner effect deflection is 0.121~(- 0.113) mm, and each upward deflection has been less than
Below quality judgment criteria 0.95mm, can meet production requirement.Accordingly the corresponding control parameter value of each influence factor is:
Tr (245 DEG C), T θ (50), N1 (9%), N2 (22%), N3 (47.5%), N4 (72%), N5 (38%), 65MPa-4s,
55MPa-4s、38MPa-2s。
Present invention also offers a kind of Shooting Technique, Shooting Technique passes through the above-mentioned injection work based on GRNN neutral nets
Skill optimization method obtains.In the present embodiment, the control parameter of Shooting Technique is Tr (245 DEG C), T θ (50), N1 (9%), N2
(22%), N3 (47.5%), N4 (72%), N5 (38%), 65MPa-4 s, 55MPa-4s, 38MPa-2s.
The molding proces s parameters obtained by above-mentioned optimization method are arranged on production injection molding machine, carry out die trial, product
On welding heading line off, be molded after 72h and carry out warpage measurement and be respectively less than 0.8mm, product quality is effectively improved.
Pass through the above-mentioned detailed description to embodiment, it is known that the Shooting Technique of the invention based on GRNN neutral nets is excellent
Change method has the characteristics that:
1) requirement for reducing die trial cost and shortening die manufacturing cycle is combined, with CAE simulation means to product mold
Such as gate location before structure design, running gate system, exhaust open up equiprobable potential problems and are examined, CAE analysis
The result shows that the molding technique parameter obtained by carrying out shaping window optimization to running gate system can't fully meet product
Actual injection needs, and particularly potential weld bond is more and larger two problems of buckling deformation.
2) it is difficult to realize comprehensive solve that weld bond is more and larger two synthesis of buckling deformation in orthogonal preferably injection parameters
In the case of problems, preferable orthogonal experiments data are carried out to technological parameter with GNRR neutral nets and are handled,
Construct targetedly neural network prediction model, after Sample, to the molding technique parameter after secondary densification into
Simulation and forecast is gone, preferable combination of process parameters can realize the zero defect injection production of product very well after prediction.
3) GNRR neutral nets are combined with mould stream CAE optimization analyses, by being preferably molded in being produced to injection practice
Molding technique parameter has preferable prediction directive function to solve injection quality multidimensional defect problem, is worthy to be popularized.
Embodiment described above only expresses the several embodiments of the present invention, its description is more specific and detailed, but simultaneously
Therefore the limitation to the scope of the claims of the present invention cannot be interpreted as.It should be pointed out that for those of ordinary skill in the art
For, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the guarantor of the present invention
Protect scope.Therefore, the protection domain of patent of the present invention should be determined by the appended claims.
Claims (10)
1. a kind of Shooting Technique optimization method based on GRNN neutral nets, it is characterised in that comprise the following steps:
Establish the CAE analysis model of injecting products;
The molding proces s parameters of the CAE analysis model are emulated using CAE software, determine to influence the injecting products production
The molding proces s parameters species of raw injection defect;
The molding proces s parameters species is converted into control parameter species;
Based on the control parameter species, network training is carried out using GRNN neutral nets, obtains final optimization pass control parameter.
2. the Shooting Technique optimization method according to claim 1 based on GRNN neutral nets, it is characterised in that utilize
CAE software emulates the molding proces s parameters of the CAE analysis model, and determining, which influences the injecting products generation injection, lacks
Sunken molding proces s parameters species, comprises the following steps:
The gate system of the injecting products is optimized using the CAE software.
3. the Shooting Technique optimization method according to claim 2 based on GRNN neutral nets, it is characterised in that utilize institute
State CAE software to optimize the gate system of the injecting products, comprise the following steps:
Design the cast gate quantity of the injecting products and corresponding gate location;
The cast gate quantity of the injecting products and corresponding position are optimized using the CAE software;
Gate size and channel size are optimized using the CAE software, obtain gate system optimum results.
4. the Shooting Technique optimization method according to claim 3 based on GRNN neutral nets, it is characterised in that utilizing
It is further comprising the steps of after CAE software carries out the gate system Optimization Steps of the injecting products:
Shaping window analysis and optimization are carried out to the result that the gate system optimizes using the CAE software;
Using the CAE software to it is described shaping window analysis and optimization after result be filled, pressurize, warping Analysis,
Determine to influence the molding proces s parameters species that the injecting products produce injection defect.
5. the Shooting Technique optimization method according to claim 4 based on GRNN neutral nets, it is characterised in that utilizing
It is further comprising the steps of after CAE software carries out the gate system Optimization Steps of the injecting products:
Analysis, the optimization of shaping window are carried out to the result that the gate system optimizes using the CAE software;
The result after the shaping window analysis, optimization is filled using the CAE software, point of pressurize and warpage
Analysis, optimization, obtain the first optimization molding proces s parameters.
6. the Shooting Technique optimization method according to claim 5 based on GRNN neutral nets, it is characterised in that the note
Moulding technological parameter species includes gate location, cold runner inlet temperature, mode temperature, injection pressure, injection time, pressurize pressure
Power, dwell time, cooling flow velocity, cooling inlet temperature and any one or a few in cooling time.
7. the Shooting Technique optimization method according to claim 6 based on GRNN neutral nets, it is characterised in that by described in
Molding proces s parameters species is converted into control parameter species, is that the cold runner inlet temperature is converted into material temperature, by the mould
Benevolence temperature inversion is mould temperature, and the injection pressure, injection time, dwell pressure and dwell time are separately converted to corresponding section
Screw rod relative rotation speed and screw position.
8. the Shooting Technique optimization method according to claim 7 based on GRNN neutral nets, it is characterised in that
It is further comprising the steps of after the molding proces s parameters species is converted into control parameter species step:
Described first optimization molding proces s parameters are corresponded to and are converted into the first optimal control parameter.
9. the Shooting Technique optimization method according to claim 8 based on GRNN neutral nets, it is characterised in that
Based on the control parameter species, network training is carried out using GRNN neutral nets, obtains final optimization pass control parameter, bag
Include following steps:
Orthogonal Experiment and Design is carried out according to the control parameter species and the first optimal control parameter, by the orthogonal examination
Test design result be filled using CAE software, pressurize, warping Analysis, obtain the sample based on amount of warpage and weld bond quantity
This;
Some samples are chosen as training sample, network instruction is carried out to the training sample using the GRNN neutral nets
Practice, obtain trained first GRNN neutral nets;
Some remaining samples are chosen as test samples, using test samples described in the first GRNN neutral nets into
Row prediction, obtains the first GRNN neural network prediction values;
The first GRNN neural network predictions value is compared with the CAE simulation results of corresponding test samples, is chosen relatively optimal
Control parameter;
Network training is carried out again to the opposite optimization control parameter after secondary densification using the first GRNN neutral nets
Secondary optimizing, obtains optimizing result;
The optimizing result is handled using renormalization, obtains optimal molding proces s parameters.
A kind of 10. Shooting Technique, it is characterised in that the Shooting Technique by described in claim 1 to 9 any one based on
The Shooting Technique optimization method of GRNN neutral nets obtains.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711211044.9A CN107944147A (en) | 2017-11-28 | 2017-11-28 | Shooting Technique optimization method and Shooting Technique based on GRNN neutral nets |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711211044.9A CN107944147A (en) | 2017-11-28 | 2017-11-28 | Shooting Technique optimization method and Shooting Technique based on GRNN neutral nets |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107944147A true CN107944147A (en) | 2018-04-20 |
Family
ID=61949229
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711211044.9A Pending CN107944147A (en) | 2017-11-28 | 2017-11-28 | Shooting Technique optimization method and Shooting Technique based on GRNN neutral nets |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107944147A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109002615A (en) * | 2018-07-20 | 2018-12-14 | 西南科技大学 | The optimum design method of family's molding jig polyisocyanate cavity structure |
CN109866401A (en) * | 2018-12-05 | 2019-06-11 | 青岛再特模具有限公司 | A kind of improvement technique of injection molding process |
CN111319206A (en) * | 2018-12-13 | 2020-06-23 | 杭州电子科技大学 | Parameter optimization method and device in injection molding system |
CN111563347A (en) * | 2020-04-03 | 2020-08-21 | 江苏师范大学 | Injection molding process parameter optimization method of fiber reinforced composite material |
CN111823530A (en) * | 2020-06-29 | 2020-10-27 | 广东海洋大学 | Injection molding process of beacon light shell based on orthogonal test and BPNN-GA algorithm |
CN112380760A (en) * | 2020-10-13 | 2021-02-19 | 重庆大学 | Multi-algorithm fusion based multi-target process parameter intelligent optimization method |
CN112659498A (en) * | 2020-12-14 | 2021-04-16 | 广东工业大学 | Real-time optimal control method for injection molding machine deep neural network |
CN115619050A (en) * | 2022-12-19 | 2023-01-17 | 绵阳丰瑞尚雅木业有限公司 | Method and system for predicting processing quality of fireproof aluminum-wood veneer and storage medium |
WO2024077944A1 (en) * | 2022-10-12 | 2024-04-18 | 成都航天模塑股份有限公司 | Low-pressure injection molding aided design method and low-pressure injection molding method for automotive interior part |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103914581A (en) * | 2013-12-27 | 2014-07-09 | 华中科技大学 | Optimization method for plastic injection molding technological parameter |
CN106055787A (en) * | 2016-05-30 | 2016-10-26 | 广西科技大学 | Automotive trim panel injection moulding technology based on BP (Back Propagation) neural network |
-
2017
- 2017-11-28 CN CN201711211044.9A patent/CN107944147A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103914581A (en) * | 2013-12-27 | 2014-07-09 | 华中科技大学 | Optimization method for plastic injection molding technological parameter |
CN106055787A (en) * | 2016-05-30 | 2016-10-26 | 广西科技大学 | Automotive trim panel injection moulding technology based on BP (Back Propagation) neural network |
Non-Patent Citations (4)
Title |
---|
孔轶艳 等: "基于CAE和RBF神经网络的注塑工艺优化分析", 《塑料》 * |
林幼文 等: "基于CAE和GRNN神经网络的汽车进气格栅注塑成型优化", 《塑料工业》 * |
林幼文 等: "基于CAE和神经网络的汽车进气格栅注塑成型优化", 《塑料工业》 * |
陈英 等: "基于BP神经网络的显示仪框注塑CAE优化分析", 《塑料》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109002615B (en) * | 2018-07-20 | 2022-01-28 | 西南科技大学 | Optimized design method for multiple special-shaped cavity structure of family mold |
CN109002615A (en) * | 2018-07-20 | 2018-12-14 | 西南科技大学 | The optimum design method of family's molding jig polyisocyanate cavity structure |
CN109866401A (en) * | 2018-12-05 | 2019-06-11 | 青岛再特模具有限公司 | A kind of improvement technique of injection molding process |
CN111319206A (en) * | 2018-12-13 | 2020-06-23 | 杭州电子科技大学 | Parameter optimization method and device in injection molding system |
CN111563347A (en) * | 2020-04-03 | 2020-08-21 | 江苏师范大学 | Injection molding process parameter optimization method of fiber reinforced composite material |
CN111563347B (en) * | 2020-04-03 | 2024-02-09 | 揭阳市恭发塑胶有限公司 | Injection molding process parameter optimization method for fiber reinforced composite material |
CN111823530A (en) * | 2020-06-29 | 2020-10-27 | 广东海洋大学 | Injection molding process of beacon light shell based on orthogonal test and BPNN-GA algorithm |
CN112380760B (en) * | 2020-10-13 | 2023-01-31 | 重庆大学 | Multi-algorithm fusion based multi-target process parameter intelligent optimization method |
CN112380760A (en) * | 2020-10-13 | 2021-02-19 | 重庆大学 | Multi-algorithm fusion based multi-target process parameter intelligent optimization method |
CN112659498A (en) * | 2020-12-14 | 2021-04-16 | 广东工业大学 | Real-time optimal control method for injection molding machine deep neural network |
WO2024077944A1 (en) * | 2022-10-12 | 2024-04-18 | 成都航天模塑股份有限公司 | Low-pressure injection molding aided design method and low-pressure injection molding method for automotive interior part |
CN115619050A (en) * | 2022-12-19 | 2023-01-17 | 绵阳丰瑞尚雅木业有限公司 | Method and system for predicting processing quality of fireproof aluminum-wood veneer and storage medium |
CN115619050B (en) * | 2022-12-19 | 2023-03-28 | 绵阳丰瑞尚雅木业有限公司 | Method and system for predicting processing quality of fireproof aluminum-wood veneer and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107944147A (en) | Shooting Technique optimization method and Shooting Technique based on GRNN neutral nets | |
CN108515678A (en) | Shooting Technique optimization method based on RBF neural | |
CN111563347B (en) | Injection molding process parameter optimization method for fiber reinforced composite material | |
Spina | Injection moulding of automotive components: comparison between hot runner systems for a case study | |
CN107972243A (en) | Shooting Technique optimization method and Shooting Technique based on SOM neutral nets | |
CN106055787A (en) | Automotive trim panel injection moulding technology based on BP (Back Propagation) neural network | |
JP6479771B2 (en) | How to run a machine that processes plastic | |
CN100471644C (en) | Design method of metal mould, mfg. method of injection moulded products, program and injection moulding appts. | |
Kovács et al. | Experimental validation of simulated weld line formation in injection moulded parts | |
Sin et al. | Computer aided injection moulding process analysis of polyvinyl alcohol–starch green biodegradable polymer compound | |
Nasir et al. | Strength of the weld line and warpage defects on the molded parts in injection molding process | |
Zheng et al. | Influence of processing parameters on warpage according to the Taguchi experiment | |
CN101028740A (en) | Injection-machine parameter setting method with automatically-resetting graded injection speed and position | |
CN201538022U (en) | Device for module chamber temperature controlling and pressure maintaining process | |
Huang et al. | Investigation on the machine calibration effect on the optimization through design of experiments (DOE) in injection molding parts | |
CN104772878A (en) | Product weight control method based on iteration modeling and optimization for injection molding process | |
CN109130062A (en) | A kind of improved washing machine cover board injection molding process | |
CN105936105A (en) | Optimization method for flow balance of slender rod multicavity mold | |
CN109109259A (en) | A kind of modified air-conditioner housing injection molding process | |
Walkare et al. | Design and analysis of extrusion-compression molding setup for processing of glass fiber reinforced polypropylene composites | |
US20030047825A1 (en) | Method and apparatus for producing an optimally reinforced structural component | |
CN101746028A (en) | Needle-valve order hot runner and gas-assisted injection molding device | |
Liu et al. | Warpage optimization of the GFRP injection molding process parameters | |
JP2014124827A (en) | Injection pattern simulation method of molten resin material, and resin molded articles-manufacturing method using this method | |
CN114953381A (en) | Modulus optimization method for fiber reinforced resin injection molding part |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180420 |
|
RJ01 | Rejection of invention patent application after publication |