CN111444619B - Online analysis method and equipment for injection mold cooling system - Google Patents

Abstract

The invention belongs to the field of mold design and evaluation, and discloses an online analysis method and equipment for an injection mold cooling system. The method comprises the following steps: performing model cooling analysis on the mold and the plastic part without considering a cooling system to obtain internal factors representing the influence of the internal factors on the cooling effect of the plastic part; introducing a finite-length continuous-line heat source model of a constant-temperature boundary, and calculating an external factor representing the influence of the external factor on the cooling effect of the plastic part; simulating the whole system model to obtain the cooling time of the plastic part, namely the comprehensive heat influence effect; establishing a neural network model of an internal factor, an external factor and a comprehensive heat influence effect and training; and during on-line analysis, recalculating the external factors according to the size and position data of the modified cooling pipeline, and predicting the comprehensive heat influence effect by adopting the trained neural network model without recalculating the internal factors. The method is simple and easy to implement, and can realize the on-line analysis of the cooling system.

Description

An online analysis method and equipment for an injection mold cooling system

technical field

The invention belongs to the field of mold design and evaluation, and more specifically relates to an online analysis method and equipment for an injection mold cooling system.

Background technique

Injection molding is widely used because of its high production efficiency, good product quality, low material consumption, and low production cost. At present, the output of injection molds accounts for more than one-third of the entire plastic mold design and production. With the development of the trend of substituting plastic for steel and plastic for wood, plastic products have been used more and more in daily life and industry, and industries such as automobiles, energy, machinery, electronics, information, aerospace, etc. requirements are also increasing.

The injection molding process is mainly composed of clamping, plasticizing, injection, pressure holding, cooling, mold opening, ejection and other stages, in which the cooling time accounts for about 50% to 80% of the molding cycle. Injection mold cooling system is an important factor to determine the quality and production efficiency of plastic parts. It not only determines the molding performance, dimensional accuracy and mechanical properties of plastic parts, but also avoids uneven wall thickness, warping deformation, dimensional changes and residual stress of plastic parts. Defects, and affect the length of the molding cycle by determining the cooling process. A reasonable cooling system can adjust the temperature environment inside the injection mold to achieve uniform and rapid cooling of the plastic parts, thereby ensuring the quality of the plastic parts and shortening the injection molding cycle and improving production efficiency. Therefore, it is of great significance to quickly analyze the cooling system in the mold design stage, and then use it as the basis for the adjustment and optimization of the cooling system, to shorten the design cycle of the entire mold and the molding cycle in the actual production of the mold.

Traditional injection mold cooling system design and analysis mainly rely on the designer's experience and intuition, which puts forward high requirements on the designer's level, which is not conducive to the work of junior designers. In addition, design based on experience lacks theoretical basis and scientific calculations, and it is necessary to judge whether the design is reasonable through continuous mold testing. When there are defects in plastic parts, it is mainly solved by repairing molds, which leads to low production efficiency, high cost and high quality of plastic parts. Difficult to guarantee. With the development of computer application technology, the application of computer-aided engineering (CAE) in the field of injection molds has greatly improved the design efficiency and quality of injection mold cooling systems. CAE mainly uses numerical analysis methods such as finite element (FEM), finite difference (FDM) and boundary element (BEM), and analyzes the cooling process through iterative methods. The cooling time in the analysis results can reflect the comprehensive thermal influence effect of plastic parts. Although the results of CAE are relatively accurate, it is necessary to reconstruct the CAD model into a CAE model of the corresponding format. After the cooling system is modified, the mold needs to be re-meshed. It takes a long time and it is difficult to meet the requirements of online analysis.

Contents of the invention

Aiming at the above defects or improvement needs of the prior art, the present invention provides an online analysis method for the cooling system of injection molds, the purpose of which is to divide the factors affecting the cooling of plastic products into two types: internal factors and external factors, which are calculated according to the two types of factors respectively The internal factor and external factor are combined to calculate the comprehensive thermal influence effect reflecting the actual cooling condition of the product, so as to realize the online analysis of the cooling system, thereby solving the technical problems of long analysis time and low efficiency of the current cooling system.

In order to achieve the above object, according to one aspect of the present invention, an online analysis method for an injection mold cooling system is provided, comprising the following steps:

Offline training phase:

Step 1: Regardless of the influence of the cooling system, the model of the mold and the plastic part is discretized into triangular units; among them, the triangular units containing the same node are adjacent to each other, and any triangular unit is selected as the central unit, and the central unit and its The cooling time of adjacent units is jointly used as an internal factor that characterizes the influence of internal factors on the cooling effect of plastic parts. At this time, the internal factor is a matrix, which is recorded as:

分别表示第1圈第1～n₁个三角单元的冷却时间，

分别表示第2圈第1～n₂个三角单元的冷却时间，

和

分别表示第3圈第1～n₃个三角单元的冷却时间，

和

分别表示第m圈第1～n_m个三角单元的冷却时间；Among them, θ _in represents the internal factor, t represents the cooling time of the currently selected triangular unit,

Respectively represent the cooling time of the _1st to n1 triangular units in the first cycle,

Respectively represent the cooling time of the 1st to n ₂ triangular units in the second cycle,

with

Respectively represent the cooling time of the 1st to n ₃ triangular units in the third circle,

with

Respectively represent the cooling time of the 1st to n _m triangular units in the mth circle;

Step 2: Introduce a finite-length continuous line heat source model with a constant temperature boundary into the model of step 1 to cool the model of step 1, so as to calculate the external factors that represent the influence of external factors on the cooling effect of plastic parts. The formula is as follows:

Among them, θ _ex represents the external factor, i=1...q represents the different boundary surfaces of the mold, q is the number of the boundary surfaces of the mold,

和

分别表示中心单元质心相对于源线热源轴线方向的坐标和源线热源关于边界面i的镜像线热源轴线方向的坐标，

and

represent the coordinates of the centroid of the central unit relative to the axis direction of the source-line heat source and the coordinates of the source-line heat source with respect to the axis direction of the mirror image line of the boundary surface i, respectively,

和

分别表示源线热源的两个端点在轴线方向的坐标，

and

respectively represent the coordinates of the two end points of the source line heat source in the axis direction,

和

分别表示源线热源关于边界面i的镜像线热源的两个端点在轴线方向的坐标，

with

Respectively represent the coordinates of the two end points of the source line heat source on the mirror line heat source of the boundary surface i in the axial direction,

r _i ¹ and r _i ² represent the distance from the centroid of the central unit to the heat source on the source line and the distance from the centroid of the central unit to the heat source on the image line of the heat source on the boundary surface i;

Step 3: Take the model of the mold and plastic part in step 1 and the initial cooling system model in step 2 as a whole, and obtain the cooling time of the plastic part through simulation, that is, the comprehensive thermal influence effect of the plastic part;

Step 4: Establish a neural network model, take the internal factor θ _in of step 1 and the external factor θ _ex of step 2 as input features, and the comprehensive thermal impact effect of step 3 as the expected output, train the neural network model, and obtain the comprehensive heat Impact effect prediction model;

Online analysis stage:

Step 5: During the adjustment and optimization process of the injection mold cooling system, according to the size and position data of the adjusted cooling pipe, recalculate the external factors according to the method of step 2, and the internal factors directly follow the results of step 1, and then use the results of step 4 The integrated thermal impact effect prediction model obtained by training is used to predict the integrated thermal impact effect.

Further, the value of m in step 1 is 1-5.

Further, the neural network model in step 4 is a BP neural network model, and the BP neural network model includes four layers: an input layer, a hidden layer 1, a hidden layer 2 and an output layer; the number of nodes _in the input layer m is between 1 and between 60, the number is equal to the total number of selected central units and adjacent units; the number of nodes in the hidden layer 1 h ₁ ≥ 2m _into +1, the number of nodes in the hidden layer 2 is between 2 and 20 , the number of nodes in the output layer is 1, and the node transfer function selects the Sigmoid function.

Further, the BP neural network model in step 4 is iteratively optimized by means of variable step size adjustment and batch processing.

Further, in step 4, before training, internal factors and external factors are respectively used as original data, and the maximum and minimum normalization processes are performed according to the following formulas:

Among them, x ₀ , x _min , x _max and x _norm represent the original data, the minimum value in the original data sequence, the maximum value in the original data sequence, and the maximum and minimum normalized data, respectively.

In order to achieve the above object, according to another aspect of the present invention, a computer-readable storage medium is provided, and a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the above-mentioned Methods.

In order to achieve the above object, according to another aspect of the present invention, an online analysis device for the cooling system of an injection mold is provided, which is characterized in that it includes the computer-readable storage medium and a processor as described above, and the processor is used to call and process A computer program stored on a computer readable storage medium.

Generally speaking, compared with the prior art, the above technical solutions conceived by the present invention can achieve the following beneficial effects:

1. The analysis method of the present invention can be run directly in the CAE system, wherein the offline training stage only needs CAE analysis twice at most, that is, the analysis of internal factors and comprehensive thermal influence effects; and the online analysis stage can directly follow the internal analysis of the offline training stage. factors, and the external factors can be directly recalculated without remeshing the finite element mesh. Therefore, by adopting the method of the present invention, the number of iterative calculations is greatly reduced, the process is simple, the amount of calculation is small, and time is saved. The online analysis of the cooling system can be realized, and fast or even real-time analysis can be realized under the condition that the cooling analysis results meet the accuracy of use. calculate.

2. The number of adjacent cells selected in step 1 is preferably the number of grid cells included in 1 to 5 circles from the center of the calculation unit, which can further reduce the calculation amount and calculation time under the premise of ensuring calculation accuracy.

3. By selecting the BP neural network model and designing the specific structure and number of nodes of the BP neural network model, the accuracy can be guaranteed and the training efficiency can be improved.

4. The BP neural network model adopts the method of variable step size adjustment and batch processing to improve the algorithm, which can overcome the disadvantages of long training time and slow convergence of traditional algorithms.

5. Before training, perform maximum and minimum normalization on internal factors and external factors, which can further improve the accuracy and convergence speed of the model.

Description of drawings

Fig. 1 is a schematic diagram of an algorithm flow chart of the present invention;

Fig. 2 is a schematic diagram of grid unit heat transfer;

Fig. 3 is a schematic diagram of a finite-length continuous line heat source at a constant temperature boundary;

Fig. 4 is the BP neural network schematic diagram of the present invention;

(a) (b) of Fig. 5 is a schematic diagram of the comparison between the predicted result and the actual result of the implementation case.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described here are only used to explain the present invention, not to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not constitute a conflict with each other.

The basic principle of the present invention is: take the unit cooling time analyzed by CAE when the cooling system is not designed as the internal factor affecting the cooling of the plastic part, and use the simplified constant temperature boundary finite-length continuous line heat source model to calculate and characterize the external factor affecting the cooling of the plastic part , through the internal factors and external factors to obtain the comprehensive thermal influence effect on plastic parts.

This method divides the factors that determine the cooling result into two types: internal factors and external factors. Internal factors include the material, shape, size, and mold temperature of plastic parts and molds. External factors include the size and location of cooling pipes and the temperature of cooling medium. , flow rate, etc. For the same pair of molds, the internal factors are constant. Here, the cooling time obtained from CAE analysis when the cooling system is not designed is used to characterize the influence of internal factors on the cooling process, also known as internal factors. To modify and optimize the cooling system, what changes is the size and position of the cooling pipe. Generally speaking, the length of the cooling pipe is 1 to 2 orders of magnitude larger than the diameter, so the cooling pipe can be approximated as a line heat source. By solving the Fourier heat conduction differential Equations for cooling channel influence effects, also known as external factors. This method only needs two CAE analyzes to build a prediction model through the BP neural network, and use the cooling time (internal factor) of each grid unit of the plastic part obtained from the CAE analysis and the calculated external factor when the cooling system is not designed as the training model. Input, including the cooling time of each grid unit of the plastic part obtained from the CAE analysis of the initial cooling system as the output of the training model. In the cooling system adjustment and optimization stage, it is only necessary to recalculate the external factors, and then use the BP neural network prediction model to obtain the comprehensive thermal impact effect, and realize the online analysis of the cooling system.

The present invention adopts following technical scheme:

Step 1: Use CAE software to conduct cooling analysis on the model including only the mold and plastic parts, and obtain the cooling time of each grid unit of the plastic parts. Since the temperature of each unit of the plastic part is different, there is heat transfer caused by the temperature difference, which in turn affects the cooling process of the unit. After the surface of the plastic part is discretized into triangular units, there is a geometric topological relationship between the units. The triangular units containing the same node are adjacent to each other. Therefore, the cooling time of this unit and the adjacent units is selected as the internal factor of the unit. At this point the internal factor is a matrix.

Preferably, the cooling process of each unit in step 1 is affected by all other units, increasing the number of adjacent units can better characterize the internal factors, thereby improving the accuracy of the comprehensive thermal influence effect, but increasing the calculation cost and time, generally Select 1 to 5 circles of adjacent units.

Step 2: Design the initial cooling system, calculate the coordinates of the centroid of all grid cells and the coordinates of the end points of the cooling pipe, and calculate the external factor through the finite-length continuous line heat source model of the constant temperature boundary. The formula used is:

和

分别表示单元质心相对于源线热源轴线方向的坐标和源线热源关于边界面i的镜像线热源轴线方向的坐标，

和

表示源线热源的两个端点在轴线方向的坐标，

和

表示源线热源关于边界面i的镜像线热源的两个端点在轴线方向的坐标，r_i ¹和r_i ²分别表示单元质心相对于源线热源距离和单元质心到源线热源关于边界面i的镜像线热源的距离。The θ _ex represents the external factor, i=1...6 represents the six boundary surfaces of the mould,

with

represent the coordinates of the unit centroid relative to the axis direction of the source line heat source and the coordinates of the source line heat source with respect to the axis direction of the mirror line heat source of the boundary surface i, respectively,

with

Indicates the coordinates of the two end points of the source line heat source in the axial direction,

and

Indicates the coordinates of the two end points of the source line heat source on the boundary surface i of the mirror line heat source in the axis direction, r _i ¹ and r _i ² represent the distance between the unit centroid and the source line heat source and the unit centroid to the source line heat source on the boundary surface i The distance of the mirror line to the heat source.

Step 3: Use CAE software to analyze the scheme including the initial cooling system design, and obtain the cooling time of the grid unit, that is, the comprehensive thermal influence effect.

Step 4: Based on the data obtained in Step 1, Step 2 and Step 3, establish a neural network model of internal factors, external factors and comprehensive thermal impact effects, internal factors and external factors as input features, and comprehensive thermal impact effects as expected output.

Preferably, the neural network model in step 4 adopts BP neural network, and the model includes four layers: input layer, hidden layer 1, hidden layer 2 and output layer. The number of nodes _in the input layer m is between 1 and 60, and its number depends on the number of adjacent units selected; the number of nodes in the hidden layer 1 is h1≥2m _in +1, and the number of nodes in the hidden layer 2 is 2 ~20, the number of nodes in the output layer is 1. The node transfer function selects the Sigmoid function, and the formula is:

The training method selects the Levenberg-Marquardt algorithm, which is the fastest for medium-scale network training. For the training of the neural network, an appropriate maximum number of training times, accuracy requirements, learning rate and minimum gradient requirements are selected according to the amount of data obtained in step one.

Preferably, the maximum-minimum normalization processing is performed on the data obtained in step 1 and step 2, and each row of the internal factor matrix and the external factors need to be normalized separately.

The max-min normalization method is:

The x ₀ , x _min , x _max and x _norm represent the original data, the minimum value in the original data sequence, the maximum value in the original data sequence and the normalized data, respectively.

Preferably, the BP neural network model adopts variable step size adjustment and batch processing methods for algorithm improvement, so as to overcome the shortcomings of traditional algorithms such as long training time and slow convergence.

Step 5: During the adjustment and optimization process of the injection mold cooling system, the external factors are recalculated according to the size and position data of the modified cooling pipe, and the internal factors do not need to be recalculated. The BP neural network model trained in step 4 is used to predict the comprehensive thermal impact effect .

The following takes plastic products as an example to describe the above method in more detail:

Step 1: Figure 2 shows the part of the mesh of the plastic product in this case. The units with common nodes are adjacent to each other. Unit 1 in the figure (that is, the unit marked with "①" in the center of Figure 2) is from near to far The cells with two different shadings represent the adjacent cells of the first ring and the adjacent cells of the second ring respectively. Since there is a temperature difference between unit 1 (that is, the unit marked with "①" in the center of Figure 2) and adjacent units, there is heat transfer (as shown by the arrow in Figure 2), so each unit cannot be calculated in isolation when considering internal factors .

The cooling of plastic products is closely related to its shape and size. Specifically, each unit is affected by all other units, but including all units in the calculation of internal factors will cause a lot of redundancy, which will greatly increase the amount of calculation and calculation time. Therefore In this embodiment, preferably 1 to 5 layers of adjacent units are sufficient. The expression of the internal factor is as follows:

和

分别表示第1圈第1个邻接单元和第n₁个单元的冷却时间，

和

分别表示第2圈第1个邻接单元和第n₂个单元的冷却时间，

和

分别表示第3圈第1个邻接单元和第n₃个单元的冷却时间，

和

分别表示第m圈第1个邻接单元和第n_m个单元的冷却时间，以上冷却时间均是在未设计冷却系统时计算所得。未设计冷却系统的本质其实是不考虑冷却系统的影响，反映在实际的仿真模拟计算过程中，先不进行冷却系统的建模，处理起来会更为简单。在其他实施例中，也可以在仿真过程中设计冷却系统后将其对单元冷却的影响调为0，但是模型调试过程会稍微复杂一些。The θ _in represents an internal factor, and t represents the cooling time of the current unit (i.e. the currently selected center unit),

and

Respectively represent the cooling time of the _1st adjacent unit and the n1th unit in the 1st circle,

and

Respectively represent the cooling time of the first adjacent unit and the n2th unit in the _second circle,

and

Respectively represent the cooling time of the first adjacent unit and the nth _3rd unit in the third circle,

with

Respectively represent the cooling time of the first adjacent unit of the mth circle and the n _mth unit, and the above cooling times are calculated when the cooling system is not designed. The essence of not designing the cooling system is that it does not consider the influence of the cooling system. It is reflected in the actual simulation calculation process that it will be easier to deal with without modeling the cooling system first. In other embodiments, after the cooling system is designed during the simulation, its influence on the cooling of the unit can be adjusted to 0, but the model debugging process will be slightly more complicated.

Step 2: Figure 3 shows a schematic diagram of a finite-length continuous line heat source with a constant temperature boundary. According to the Fourier heat conduction differential equation, the analytical solution of the heat transfer of an instantaneous point heat source can be obtained, and the finite-length continuous line can be obtained by integrating in time and space Analytical solution for a linear heat source.

Due to the constant temperature of the boundary, a mirror image heat source is added at the position where the source heat source is symmetrical to the boundary, and the subtraction of the two is an external factor. In this embodiment, the mold has 6 surfaces, namely upper, lower, left, right, front and rear. Set up 6 mirrored heat sources. Therefore, the calculation formula of the external factor is as follows:

和

分别表示中心单元质心相对于源线热源轴线方向的坐标和源线热源关于边界面i的镜像线热源轴线方向的坐标，

和

表示源线热源的两个端点在轴线方向的坐标，

和

表示源线热源关于边界面i的镜像线热源的两个端点在轴线方向的坐标，r_i ¹和r_i ²分别表示中心单元质心相对于源线热源距离和单元质心到源线热源关于边界面i的镜像线热源的距离。上述公式适用于具有不同数量边界面的模具。The θ _ex represents the external factor, i=1...6 represents the six boundary surfaces of the mould,

and

represent the coordinates of the centroid of the central unit relative to the axis direction of the source-line heat source and the coordinates of the source-line heat source with respect to the axis direction of the mirror image line of the boundary surface i, respectively,

with

Indicates the coordinates of the two end points of the source line heat source in the axial direction,

with

Indicates the coordinates of the two end points of the source line heat source on the boundary surface i of the mirror image line heat source in the axis direction, r _i ¹ and r _i ² represent the distance from the center unit centroid to the source line heat source and the unit centroid to the source line heat source on the boundary surface The distance of the heat source to the image line of i. The above formula is valid for dies with different numbers of boundary surfaces.

Step 3: After obtaining the internal factors and external factors, conduct CAE analysis on the cooling system scheme including the initial design to obtain the comprehensive thermal impact effect;

Step 4: Then establish a neural network model of internal factors, external factors and comprehensive thermal impact effects, use internal factors and external factors as the input features of the neural network model, and comprehensive thermal impact effects as the expected output of the neural network model.

Preferably, the BP neural network model is used in this case, the maximum number of iteration steps for model training is 1000, the learning rate is 0.1, and the iteration error is allowed to be 0.0001. As a simple example, the BP neural network model designed in this case contains four layers: input layer, hidden layer 1, hidden layer 2, and output layer; the input layer represents internal factors and external factors, and contains 57 nodes; Hidden layer 1 and hidden layer 2 contain 120 and 10 nodes, respectively; the output layer represents the comprehensive thermal effect and contains 1 node. The node transfer function selects the Sigmoid function, and the formula is:

Among them, x represents the input data of the node, and S(x) represents the output data of the node;

Preferably, the Levenberg-Marquardt algorithm is selected as the training method, and the training process adopts the method of variable step size adjustment and batch processing to improve the algorithm, so as to overcome the shortcomings of traditional algorithms such as long training time and slow convergence.

Preferably, before training, the internal factors and external factors are subjected to maximum and minimum normalization processing. The normalization method is:

The x ₀ , x _min , x _max and x _norm represent the original data, the minimum value in the original data sequence, the maximum value in the original data sequence, and the maximum and minimum normalized data, respectively.

For the trained BP model, the method of using it for online detection is as follows:

Step 5: During the adjustment and optimization process of the injection mold cooling system, according to the size and position data of the modified cooling pipe, recalculate the external factors according to the method of step 2, and the internal factors do not need to be recalculated, and the results of step 1 are directly used, and then Use the BP neural network model trained in step 4 to predict the comprehensive thermal impact effect. The prediction results and actual measurement results of this case are shown in Figure 5. From Figure 5, it can be seen that the variation rules of the prediction results and the actual results are highly consistent, which can accurately reflect the differences in the real influence of the cooling system on various parts of plastic products due to their different positions and structures. , can accurately predict defects such as hot spots of products.

It is easy for those skilled in the art to understand that the above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present invention, All should be included within the protection scope of the present invention.

Claims (7)

1. an injection mold cooling system online analysis method, is characterized in that, comprises the following steps: Offline training phase: Step 1: Regardless of the influence of the cooling system, the model of the mold and the plastic part is discretized into triangular units; among them, the triangular units containing the same node are adjacent to each other, and any triangular unit is selected as the central unit, and the central unit and its The cooling time of adjacent units is jointly used as an internal factor that characterizes the influence of internal factors on the cooling effect of plastic parts. At this time, the internal factor is a matrix, which is recorded as:

Among them, θ _in represents the internal factor, t represents the cooling time of the currently selected triangular unit,

Respectively represent the cooling time of the _1st to n1 triangular units in the first cycle,

Respectively represent the cooling time of the 1st to n ₂ triangular units in the second cycle,

with

Respectively represent the cooling time of the 1st to n ₃ triangular units in the third circle,

with

Respectively represent the cooling time of the 1st to n _m triangular units in the mth circle; Step 2: Introduce a finite-length continuous line heat source model with a constant temperature boundary into the model of step 1 to cool the model of step 1, so as to calculate the external factors that represent the influence of external factors on the cooling effect of plastic parts. The formula is as follows:

Among them, θ _ex represents the external factor, i=1...q represents the different boundary surfaces of the mold, q is the number of the boundary surfaces of the mold,

and

represent the coordinates of the centroid of the central unit relative to the axis direction of the source-line heat source and the coordinates of the source-line heat source with respect to the axis direction of the mirror image line of the boundary surface i, respectively,

and

respectively represent the coordinates of the two end points of the source line heat source in the axis direction,

with

Respectively represent the coordinates of the two end points of the source line heat source on the mirror line heat source of the boundary surface i in the axial direction, r _i ¹ and r _i ² represent the distance from the centroid of the central unit to the heat source on the source line and the distance from the centroid of the central unit to the heat source on the image line of the heat source on the boundary surface i; Step 3: Take the model of the mold and plastic part in step 1 and the initial cooling system model in step 2 as a whole, and obtain the cooling time of the plastic part through simulation, that is, the comprehensive thermal influence effect of the plastic part; Step 4: Establish a neural network model, take the internal factor θ _in of step 1 and the external factor θ _ex of step 2 as input features, and the comprehensive thermal impact effect of step 3 as the expected output, train the neural network model, and obtain the comprehensive heat Impact effect prediction model; Online analysis stage: Step 5: During the adjustment and optimization process of the injection mold cooling system, according to the size and position data of the adjusted cooling pipe, recalculate the external factors according to the method of step 2, and the internal factors directly follow the results of step 1, and then use the results of step 4 The integrated thermal impact effect prediction model obtained by training is used to predict the integrated thermal impact effect. 2 . The online analysis method of the injection mold cooling system according to claim 1 , wherein the value of m in step 1 is 1-5. 3 . 3. a kind of injection mold cooling system online analysis method according to claim 1, is characterized in that, the neural network model in step 4 is BP neural network model, and BP neural network model comprises four layers: input layer, hidden layer 1 , the hidden layer 2 and the output layer; the number of nodes _in the input layer m is between 1 and 60, which is equal to the total number of selected central units and adjacent units; the number of nodes in the hidden layer 1 h ₁ ≥ 2m _into +1, the number of nodes in the hidden layer 2 is between 2 and 20, the number of nodes in the output layer is 1, and the transfer function of the nodes is a Sigmoid function. 4. An online analysis method for an injection mold cooling system according to any one of claims 1 to 3, characterized in that the BP neural network model in step 4 is iteratively optimized by means of variable step size adjustment and batch processing. 5. An online analysis method for an injection mold cooling system according to any one of claims 1 to 3, characterized in that, in step 4, before training, internal factors and external factors are used as raw data respectively, according to the following formula Perform maximum and minimum normalization processing respectively:

Among them, x ₀ , x _min , x _max and x _norm represent the original data, the minimum value in the original data sequence, the maximum value in the original data sequence, and the maximum and minimum normalized data, respectively. 6. A computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the method according to any one of claims 1-5 is implemented. 7. An online analysis device for an injection mold cooling system, characterized in that it comprises the computer-readable storage medium as claimed in claim 6 and a processor, the processor is used to call and process the computer program stored in the computer-readable storage medium.

Images