Injection molding machine debugging parameter recommendation method combining multi-model prediction and PSO (Particle Swarm Optimization)

Through the recommended method of injection molding machine parameter combination with multi-model prediction and particle swarm algorithm, the problem of cumbersome process parameter adjustment in injection molding products is solved, intelligent calculation and efficiency improvement of parameters are achieved, and production efficiency and success rate are improved.

CN119004942APending Publication Date: 2024-11-22ZHANGZHOU HONGXINGTAI ELECTRONICS CO LTD

Patent Information

Application Number
CN202410890063.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-04
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

In the production process of injection molded products, due to the complex environment, the process parameters are cumbersome to adjust, which affects production efficiency. The difference in manual experience leads to irregular machine adjustment, making it difficult to quickly determine the appropriate process parameters value.

Method used

Using a method of combining multi-model prediction and particle swarm algorithm (PSO), a data set is constructed, a workpiece size combination prediction model is built, and the process parameters are adjusted through the particle swarm optimization algorithm, and the appropriate process parameter values ​​are automatically calculated instead of manual empirical adjustment.

Benefits of technology

The intelligent calculation process of injection molding machine parameters is realized, the efficiency of process parameter determination is improved, the situation of irregular machine regulation is reduced, and the production efficiency and success rate are improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004942A_ABST
    Figure CN119004942A_ABST
Patent Text Reader

Abstract

The invention discloses an injection molding debugging parameter recommendation method combining multi-model prediction and PSO, and relates to the field of injection molding process parameter adjustment. According to the multi-model prediction and PSO combined injection molding machine adjustment parameter recommendation method, a data set DataSet including forming machine state data, machine process setting parameters and workpiece inspection size is constructed; building a workpiece size combination prediction model, selecting n kinds of mutually independent single workpiece size combination prediction models, respectively predicting corresponding workpiece sizes, and recording the workpiece sizes as Li (i = 1, 2,..., m), m being the number of size edges of the workpiece; corresponding weights w are given to the selected n prediction models respectively, optimization is carried out based on the particle swarm optimization until an optimal workpiece size prediction result is obtained, a set of complete calculation process is provided for determination of injection molding adjustment parameters through production data, and parameter adjustment intelligentization is achieved to a certain extent.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the field of injection molding process parameter adjustment, and in particular to an injection molding machine adjustment parameter recommendation method combining multi-model prediction and PSO. Background Art

[0002] The production of injection molded products is usually a continuous process, and currently relies mainly on manual labor to control product quality. During the production process, workers will manually adjust process parameters according to the defects of injection molded products to ensure the normal operation of production.

[0003] However, during the production process, when there are defects in the product, workers are often urgently needed to determine the appropriate process parameter values ​​in time to ensure the production quality of the product. However, the production environment of injection molding products is complex, and it is often impossible to quickly determine the reasonable values ​​of process parameters, which seriously affects the production efficiency of the company.

[0004] Therefore, how to quickly determine the appropriate values ​​of injection molding process parameters, reduce the situation of non-standard machine adjustment and blind trial and error caused by experience differences of on-site machine adjustment personnel, and thus improve the efficiency and success rate of machine adjustment has become an urgent problem to be solved.

[0005] The Chinese patent application number is: CN202311044811.7, and the name is: A method for progressive dynamic adjustment of injection molding process parameters based on product quality prediction. The invention defines a defect function based on and technology, and constructs and trains a hybrid model that integrates and. Based on this model, combined with the acquired key state data of injection molding, the defect changes of the product are dynamically predicted, achieving the purpose of early warning to adjust the injection molding process parameters. However, the invention focuses on the early warning of product defects, and lacks description on how to quickly adjust the numerical values ​​of process parameters.

[0006] The Chinese patent application number is: CN202311421564.8, and its name is: A self-adjusting control method for process parameters of an injection molding machine. The invention first inputs the melt temperature, mold temperature, and other parameters related to the product raw material name and the raw materials into the control terminal of the injection molding machine, and then inputs the calculation formulas for the mold opening and closing time parameters, injection time parameters, holding time parameters, and cooling time parameters into the control terminal. Finally, the raw material name, product wall thickness, and product weight of the product are determined on the control terminal of the injection molding machine, and the relevant parameters are retrieved at the control terminal. According to the above corresponding formulas, the above parameters are automatically calculated. Although this invention effectively solves the problem of cumbersome manual setting of process parameters of products using different raw materials in the prior art to a certain extent. However, the invention ignores the impact of the internal environment of the molding machine on product quality during the production process. Summary of the invention

[0007] In view of the shortcomings of the prior art, the present invention provides a method for recommending injection molding machine parameters by combining multi-model prediction and PSO, which solves the problem of cumbersome operation caused by manually setting parameters when process parameters need to be adjusted urgently during the production process.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a method for recommending injection molding machine parameters combining multi-model prediction and PSO, comprising the following steps: constructing a data set Data_Set containing molding machine status data, machine process setting parameters, and workpiece inspection dimensions; building a workpiece dimension combination prediction model, and selecting n independent single workpiece dimension combination prediction models to predict the corresponding workpiece dimensions respectively, denoted as L i (i=1,2,...,m), where m is the number of edges of the workpiece size; for the selected n prediction models, the corresponding weights w are assigned respectively, and they are optimized based on the particle swarm algorithm until the optimal workpiece size prediction result is obtained; when the currently set process parameters do not meet the requirements, the machine process setting parameters suitable for the current working conditions are solved by combining the prediction model and the particle swarm algorithm, and the adjusted parameter determination values ​​are input into the injection molding machine console to start a new round of production.

[0009] Furthermore, the data set Data_Set is constructed by horizontally merging the molding machine status data, machine process setting parameters, and workpiece inspection dimension data according to the same production batch, and retaining the relevant row data of the workpiece inspection dimension that is qualified.

[0010] Furthermore, among the n selected prediction models and the corresponding weights w, the size after combined prediction is obtained through linear superposition: Among them, L i is the predicted workpiece size result of the i-th prediction algorithm.

[0011] Furthermore, the particle swarm algorithm is used for optimization until the optimal workpiece size prediction result is obtained, which is specifically the weight w of each prediction model. i Optimize and finally get the combined prediction workpiece size result after optimizing the weights in, It is the weight corresponding to the prediction result of each single prediction algorithm after particle swarm optimization.

[0012] Furthermore, the specific steps of particle swarm optimization to optimize the weights of each prediction model are as follows: Establish the objective function Where l is the design size of the workpiece; the weight w i (i=1,2,...,n) as the particle swarm to be optimized, set the particle swarm number M, the maximum number of iterations D, the particle position boundary condition X, and initialize the particle swarm population extreme value g best; Based on the particle swarm prediction model weight optimization, and update the position and velocity information of the particle swarm, and at the same time update the target optimization function corresponding to each iteration, select the individual extreme value p of the optimal particle best , and compare the group extreme value g best , when p best <g best , the corresponding particle position is taken as the current optimal solution; determine whether the number of optimizations meets the maximum number of iterations, and if so, obtain the corresponding weights of each single prediction model after optimization, and use the obtained weight value w i (i=1,2,...,n)Substitute Get the optimal workpiece size prediction result.

[0013] Furthermore, when the currently set process parameters do not meet the requirements, the specific steps of solving the machine process setting parameters suitable for the current working conditions by combining the prediction model and the particle swarm algorithm are as follows: determine the optimization variables and variable boundary conditions, and record the variable boundary conditions as [lb i ,ub i ], where lb i To find the lower bound of the optimal variable, ub i is the upper boundary of the optimization variable, i represents the i-th variable, and the optimization variable is specifically the machine adjustment parameter variables to be optimized, namely, holding time x1, holding pressure x2 and mold temperature x3; establish the particle swarm objective optimization function Among them, L i is the size of each edge predicted based on the LSTM model, l i Design the size of the workpiece, n represents the geometric number of the workpiece; the weight w i (i=1,2,...,n) as the particle swarm to be optimized, set the particle swarm number M, the maximum number of iterations D, the particle position boundary condition X, and initialize the particle swarm population extreme value g best ; The particle swarm prediction model weight optimization is performed, and the position and velocity information of the particle swarm is updated. At the same time, the target optimization function ΔL corresponding to each iteration is updated, and the individual extreme value p of the optimal particle is selected. best , compare the group extreme value g best , when p best <g best , the corresponding particle position is taken as the current optimal solution; determine whether the number of optimizations meets the maximum number of iterations D, and obtain the machine adjustment parameter x when the condition is met. i (i=1,2,3).

[0014] The present invention has the following beneficial effects:

[0015] (1) This method of recommending injection molding machine parameters that combines multi-model prediction and PSO uses production data to provide a complete calculation process for determining injection molding machine parameters, and to a certain extent realizes the intelligent adjustment of parameters.

[0016] (2) The injection molding machine parameter recommendation method that combines multi-model prediction and PSO replaces the experience of manual adjustment of process parameters through a multi-model combination prediction method. The particle swarm algorithm can more quickly calculate the corresponding values ​​of process parameters suitable for the current working conditions, thereby improving the efficiency of process parameter determination.

[0017] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flowchart of the multi-model workpiece size combination prediction type described in the present invention.

[0019] Figure 2 The present invention is a flowchart for solving the machine process setting parameters suitable for the current working conditions by combining the algorithm described in the present invention. DETAILED DESCRIPTION

[0020] The embodiment of the present application solves the problem of cumbersome operation caused by manually setting parameters when process parameters need to be adjusted urgently during the production process by using a method for recommending injection molding machine parameters that combines multi-model prediction and PSO.

[0021] An embodiment of the present invention provides a technical solution: a method for recommending injection molding machine parameters combining multi-model prediction and PSO, which is specifically implemented according to the following steps: constructing a data set Data_Set including molding machine status data, machine process setting parameters, and relay housing plastic part inspection dimensions, wherein the height l1, upper surface diameter l2, long side wall thickness l3, and long side wall thickness l4 of the relay housing plastic part are key production dimensions, and the four are recorded as l=[l1, l2, l3, l4].

[0022] The data set Data_Set is constructed by horizontally merging the molding machine status data, machine process setting parameters, and relay housing plastic parts inspection dimension data according to the same production batch, and retaining the relevant row data of the workpiece inspection dimension that meets the requirements.

[0023] A combination prediction model for the size of relay housing plastic parts is built, and two independent single workpiece size combination prediction models are selected to predict the corresponding workpiece sizes respectively, and two prediction models are selected: XGBoost and SVR.

[0024] For the two selected prediction models, the corresponding weights w are assigned respectively, and the combined prediction model of relay housing plastic parts size is established through linear superposition: L = wXGBoost L XGBoost +w SVR L SVR .

[0025] It should be explained that in order to improve the prediction accuracy, the particle swarm algorithm (PSO) is used to optimize the weights w of each prediction model. i Optimize, and finally get the combined prediction brightness result after optimizing weights. The process of particle swarm workpiece size combination prediction algorithm is as follows: Figure 1 As shown, the steps are as follows:

[0026] Establish the objective function. The fitness function is the target optimization function. In order to make the prediction accuracy of the combined model higher, the error index of the combined prediction result should be lower. Therefore, the objective function is: ΔL = min|w XGBoost L XGBoost +w SVR L SVR -l|, where l is the design size of the relay housing plastic part.

[0027] The weight w XGBoost 、w SVR As the particle swarm to be optimized, set the particle population number M = 200, the maximum number of iterations D = 1000, and initialize the group extreme value g best = 100, the particle position boundary condition X constraint:

[0028]

[0029] The particle swarm prediction model weight optimization is used to update the position and velocity information of the particle swarm, update the target optimization function corresponding to each iteration, and select the individual extreme value p of the optimal particle. best , compare the group extreme value g best , when p best <g best , the corresponding particle position is taken as the current optimal solution.

[0030] Determine whether the number of optimizations meets the maximum number of iterations D = 1000. After optimization, the model weight w is obtained. XGBoost =0.9,w SVR =0.1. Therefore, the predicted size of the relay housing plastic part is: L = 0.9 × L XGBoost +0.1×L SVR .

[0031] It should be explained that as the internal state data of the molding machine changes, the relay shell size of the current kth batch is L k=[99.8,99.8,499.8,299.8], and the corresponding design size of the drawing is, l = [100.15,100.15,500,300], the allowable error range Δ = [±0.15,±0.15,±0.2,±0.2]. Therefore, the current batch of relay housing plastic parts exceeds the allowable error range of the design size, and the value needs to be further adjusted. The combined prediction model and the PSO algorithm will be used to solve the machine process setting parameters suitable for the current working conditions. The process is shown in Figure 2 , the specific steps are as follows:

[0032] Determine the optimization variables and boundary conditions. The machine parameters to be optimized include holding time x1, holding pressure x2, and mold temperature x3. The corresponding variable boundary X values ​​are [5,65], [16,25], and [16,115], respectively.

[0033] The particle swarm objective optimization function is established. The objective function is to minimize the error between the relay housing plastic size predicted by the combined prediction model and the design size. Therefore, the objective function is: ΔL = min | 0.9 × L XGBoost +0.1×L SVR -l|.

[0034] The machine parameters holding time x1, holding pressure x2 and mold temperature x3 are used as the particle swarm to be optimized, and the particle population number M = 500, the maximum number of iterations D = 600, the particle position boundary condition X, and the group extreme value g of the initialized particle swarm are set. best .

[0035] The particle swarm prediction model weighted optimization is used to input the parameters to be optimized and other state data into the combined prediction model. After particle swarm solution, new machine adjustment parameter values ​​are obtained: holding time x1=24.60, holding pressure x2=23.20 and mold temperature x3=63.26. The solution time for this round of new machine adjustment parameter values ​​is t=2.62 seconds.

[0036] The machine receives the new adjustment parameter values ​​and obtains a new batch of relay housing plastic parts with dimensions L k+1 =[100.08, 100.06, 499.90, 300.12]. After inspection, the error of the new workpiece production size after adjusting the value is Δ1 = [-0.07, -0.09, 0.1, 0.12], which is within the allowable error range, indicating that the process parameter adjustment has achieved the effect.

[0037] In order to better verify the effectiveness of the method for determining the machine adjustment value, the present invention conducted 600 machine adjustment experiments, and finally calculated the corresponding machine adjustment success rate α=92.7%, and the average machine adjustment parameter value determination time t=1.96 seconds. The corresponding partial experimental results are shown in Table 1.

[0038] Table 1 Example of machine adjustment test data

[0039]

[0040] In summary, this application has at least the following effects:

[0041] The use of production data provides a complete calculation process for determining the injection molding machine adjustment parameters, which realizes the intelligent adjustment to a certain extent.

[0042] The multi-model combination prediction method replaces the experience of manually adjusting process parameters. The particle swarm algorithm can more quickly calculate the corresponding values ​​of process parameters suitable for the current working conditions, thereby improving the efficiency of process parameter determination.

[0043] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0044] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for recommending injection molding machine parameters combining multi-model prediction and PSO, characterized in that: The following steps are involved: Construct a data set Data_Set that includes molding machine status data, machine process setting parameters, and workpiece inspection dimensions; Build a workpiece size combination prediction model, and select n independent single workpiece size combination prediction models to predict the corresponding workpiece sizes, denoted as L i (i=1,2,...,m), where m is the number of edges of the workpiece; For the selected n prediction models, the corresponding weights w are assigned respectively, and the optimization is performed based on the particle swarm algorithm until the optimal workpiece size prediction result is obtained; When the currently set process parameters do not meet the requirements, the machine process setting parameters suitable for the current working conditions are solved by combining the prediction model and the particle swarm algorithm, and the adjusted parameter values ​​are input into the injection molding machine console to start a new round of production.

2. The method for recommending injection molding machine parameters combining multi-model prediction and PSO according to claim 1 is characterized in that: The data set Data_Set is constructed by horizontally merging the molding machine status data, machine process setting parameters, and workpiece inspection dimension data according to the same production batch, and retaining the relevant row data of the workpiece inspection dimension that meets the requirements.

3. The method for recommending injection molding machine parameters combining multi-model prediction and PSO according to claim 1 is characterized in that: Among the n selected prediction models and the corresponding weights w, the size of the combined prediction is obtained through linear superposition: Among them, L i is the predicted workpiece size result of the i-th prediction algorithm.

4. The method for recommending injection molding machine parameters combining multi-model prediction and PSO according to claim 1, characterized in that: And based on the particle swarm algorithm, optimization is performed until the optimal workpiece size prediction result is obtained, which is specifically the weight w of each prediction model. i Optimize and finally get the combined prediction workpiece size result after optimizing the weights in, It is the weight corresponding to the prediction result of each single prediction algorithm after particle swarm optimization.

5. The method for recommending injection molding machine parameters combining multi-model prediction and PSO according to claim 4 is characterized in that: The specific steps of particle swarm optimization to optimize the weights of each prediction model are as follows: Establishing the objective function Among them, l is the design size of the workpiece; The weight w i (i=1,2,...,n) as the particle swarm to be optimized, set the particle swarm number M, the maximum number of iterations D, the particle position boundary condition X, and initialize the particle swarm population extreme value g best ; Based on the particle swarm prediction model weight optimization, the position and velocity information of the particle swarm is updated, and the target optimization function corresponding to each iteration is updated to select the individual extreme value p of the optimal particle. best , and compare the group extreme value g best , when p best <g best When , the corresponding particle position is taken as the current optimal solution; Determine whether the number of optimizations meets the maximum number of iterations. If the condition is met, obtain the weights corresponding to the optimization of each single prediction model and calculate the weight value w i (i=1,2,...,n)Substitute Get the optimal workpiece size prediction result.

6. The method for recommending injection molding machine parameters combining multi-model prediction and PSO according to claim 1, characterized in that: When the currently set process parameters do not meet the requirements, the specific steps for solving the machine process setting parameters suitable for the current working conditions by combining the prediction model and the particle swarm algorithm are as follows: Determine the optimization variables and variable boundary conditions, and record the variable boundary conditions as [lb i ,ub i ], where lb i To find the lower bound of the optimal variable, ub i is the upper boundary of the optimization variable, i represents the i-th variable, and the optimization variable is specifically the machine adjustment parameter variables to be optimized, namely, holding time x1, holding pressure x2, and mold temperature x3; Establish particle swarm objective optimization function Among them, L i is the size of each edge predicted based on the LSTM model, l i is the design size of the workpiece, n represents the geometric number of the workpiece; The weight w i (i=1,2,...,n) as the particle swarm to be optimized, set the particle swarm number M, the maximum number of iterations D, the particle position boundary condition X, and initialize the particle swarm population extreme value g best ; The particle swarm prediction model weights are optimized, and the position and velocity information of the particle swarm is updated. At the same time, the target optimization function ΔL corresponding to each iteration is updated, and the individual extreme value p of the optimal particle is selected. best , compare the group extreme value g best , when p best <g best When , the corresponding particle position is taken as the current optimal solution; Determine whether the number of optimizations meets the maximum number of iterations D. If the condition is met, obtain the machine tuning parameter x i (i=1,2,3).

Citation Information

Patent Citations

  • Product quality pre-judgment-based progressive dynamic adjustment method for injection molding process parameters

    CN117067541A

  • Process parameter self-adjustment control method of injection molding machine

    CN117754835A

Cited By

  • Parameter adjustment method of injection molding process, electronic device, and readable storage medium

    CN122683977A