A pressure control system for a precision injection molding numerical control device

Abstract

The present invention relates to a pressure control system for precision injection molding numerical control equipment. Using time series pressure data, with time as the independent variable and pressure as the dependent variable, a pressure smoothing curve is plotted; the pressure smoothing curves of products with different sizes and shapes are obtained, and a deep learning model between the pressure smoothing curve and the size and shape of the product is constructed; according to the sizes and shapes of two or more products that need to cooperate with each other, the corresponding pressure smoothing curve is obtained by using the deep learning model to control the pressure during the injection molding process. The present invention quantifies the complex non-linear relationship between time and injection molding pressure by using a generalized additive model, improves the pressure control accuracy, and uses a convolutional neural network model to determine that different pressure smoothing curves can obtain products with different sizes and shapes, realizing the cooperation of different products and improving the processing quality of injection molded products.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent manufacturing, and relates to a pressure control system for precision injection molding numerical control equipment. Background Art

[0002] Precision injection molding numerical control equipment is a high-end equipment in the modern plastic processing industry, integrating advanced numerical control technology, precision mechanical design and plastic injection molding technology. It mainly precisely controls various parameters in the injection molding process, such as temperature, pressure, speed, etc., to achieve precise molding of plastic raw materials.

[0003] In the production process of plastic injection molded products such as toys, the quality of the injection molding process directly affects the quality of the products. Traditional injection molding equipment has great defects in pressure control. For example, the pressure fluctuation is large, and it is difficult to achieve precise pressure regulation. And the pressure regulation during injection molding is the key to ensuring the stable and uniform pressure of the plastic melt during the injection molding process, thereby avoiding molding defects.

[0004] Due to the difficulty in achieving precise pressure regulation, the produced plastic products have poor dimensional accuracy and a large tolerance range, and it is difficult to meet the requirements of high-precision products. Injection molding defects such as flash, sink marks, and material shortage are likely to appear on the product surface, which greatly affects the product quality and the yield rate. At the same time, due to the differences in size and shape between different products, the fitting accuracy between products is relatively low. During the assembly process by players, problems such as loose splicing and unstable structure are likely to occur, seriously affecting the user experience. Summary of the Invention

[0005] To solve the above problems existing in the prior art, the present invention provides a pressure control system for precision injection molding numerical control equipment, aiming to collect pressure data samples of multiple injection molding processes, establish a pressure smoothing curve at different times to solve the problem of unstable injection molding pressure. At the same time, establish the relationship between the pressure smoothing curve of different products and their size and shape to improve the fitting accuracy between products.

[0006] The object of the present invention can be achieved by the following technical solutions:

[0007] The present application provides a pressure control system for precision injection molding numerical control equipment, including a data acquisition module, a pressure curve module, a curve application module and a pressure control module;

[0008] The data acquisition module is used to collect pressure data of multiple complete injection molding cycles, and the pressure data is divided at a specific time step to obtain time series pressure data;

[0009] The pressure curve module is used to use the time series pressure data, with time as the independent variable and pressure as the dependent variable, to draw a pressure smoothing curve;

[0010] The curve application module is used to obtain the pressure smoothing curves of products with different sizes and shapes, and construct a deep learning model between the pressure smoothing curves and the sizes and shapes of the products;

[0011] The pressure control module is used to control the pressure of the injection molding process by using the corresponding pressure smoothing curves obtained from the deep learning model according to the sizes and shapes of two or more products that need to cooperate with each other.

[0012] Further, the data acquisition module consists of multiple high-precision pressure sensors, and the sensors are respectively installed at the barrel feed inlet, the front end of the screw, and the mold cavity of the injection molding machine.

[0013] Further, the pressure smoothing curve is fitted by a generalized additive model and then plotted by R language software.

[0014] Further, the generalized additive model includes the following construction steps:

[0015] S1. Determine the model structure: The model takes pressure as the response variable, and the corresponding explanatory variable is time;

[0016] S2. Select the smoothing function: Select the cubic spline function as the smoothing function of the model, and divide the time into multiple small segments. Within each small segment, use a cubic polynomial to describe the pressure change;

[0017] S3. Parameter estimation: By continuously adjusting the model parameters, minimize the gap between the predicted value and the actual value of the pressure, so as to determine the final parameters;

[0018] S4. Model testing and evaluation: Use the model with adjusted parameters to predict the pressure value, and use the coefficient of determination to test the accuracy of the predicted value;

[0019] S5. Plot the pressure smoothing curve: According to the adjusted model and parameters, calculate the predicted pressure values at different time points; and use time as the horizontal axis and the predicted pressure values as the vertical axis, and draw the pressure smoothing curve with the help of a drawing tool.

[0020] Further, the expression of the generalized additive model is:

[0021] ,

[0022] In the formula, g represents the injection pressure; x represents time; s represents the smoothing function; k represents the number of knots of the smoothing function; μ represents the intercept term.

[0023] Further, the deep learning model is a convolutional neural network model configured with the pressure smoothing curve as the input feature and the size and shape of the product as the output target.

[0024] Further, the convolutional neural network model includes the following construction steps:

[0025] T1. Data preprocessing: Convert the pressure smoothing curve data into a standardized form with zero mean and unit variance, and at the same time encode the size and shape information of the product.

[0026] T2. Model architecture design: Refer to the dimension of the pressure smoothing curve data to determine the number of input layer nodes; construct a convolutional layer to capture the features of the pressure smoothing curve, and determine the number, size, and stride of the convolutional kernels; add a pooling layer to perform dimensionality reduction on the data; set a fully connected layer to achieve feature integration and mapping from the output of the pooling layer to the final output layer; determine the number of output layer nodes according to the feature dimension of the product size and shape.

[0027] T3. Model training: Select a loss function and an optimizer according to the output target, and adjust the hyperparameters; divide the data into a training set, a validation set, and a test set, input the training set data into the model, obtain the output through forward propagation and calculate the error, and then update the model parameters through backpropagation, continuously iterating the training process.

[0028] T4. Model evaluation and adjustment: Use the test set to evaluate the trained model, and judge the model performance by calculating evaluation metrics.

[0029] Further, the pressure control module calculates the pressure deviation according to the preset pressure smoothing curve and the real-time collected pressure data, and performs data processing through the PID control algorithm to generate accurate control instructions.

[0030] Further, the pressure control module further includes a human-machine interaction module, configures a pressure display screen, and the operator can monitor the injection pressure situation in real time according to the pressure display screen.

[0031] Advantages of the present invention:

[0032] (1) Utilize the time-series pressure data, with time as the independent variable and pressure as the dependent variable, to plot the pressure smoothing curve; obtain the pressure smoothing curves of products with different sizes and shapes, construct a deep learning model between the pressure smoothing curve and the size and shape of the product; according to the sizes and shapes of two or more products that need to cooperate with each other, use the deep learning model to obtain the corresponding pressure smoothing curve to control the pressure during the injection molding process. This solves the problems of large fluctuations in injection pressure in the prior art and poor matching accuracy of products with different sizes and shapes.

[0033] (2) By using a generalized additive model to fit the pressure smoothing curve, the complex non-linear relationship between time and injection pressure in an injection molding cycle is quantified, improving the pressure control accuracy and enhancing the processing quality of injection molded products.

[0034] (3) By using a convolutional neural network model, it is determined that different pressure smoothing curves can obtain products of different sizes and shapes, realizing the matching of different products and enhancing the processing quality of injection molded products. Description of the Drawings

[0035] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the drawings.

[0036] Figure 1 It is a structural diagram of a pressure control system for a precision injection molding numerical control device in the present invention.

[0037] Figure 2 It is the construction steps of the generalized additive model in an embodiment of the present invention.

[0038] Figure 3 It is the construction steps of the convolutional neural model in an embodiment of the present invention. Detailed Embodiments

[0039] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following will, in conjunction with the drawings and preferred embodiments, detail the specific embodiments, structures, features and their effects of the present invention as follows.

[0040] Please refer to Figures 1 - 3 , this application provides a pressure control system for a precision injection molding numerical control device, including a data acquisition module, a pressure curve module, a curve application module and a pressure control module;

[0041] The data acquisition module is used to collect pressure data of multiple complete injection molding cycles, and the pressure data is divided at a specific time step to obtain time series pressure data;

[0042] In this embodiment, the main work of the data acquisition module is to collect pressure data for multiple complete injection molding cycles. The reason for focusing on multiple cycles is that the injection molding production process is not constant, and each injection may be affected by many factors, such as the subtle characteristic differences of different batches of plastic raw materials, and the performance changes caused by minor wear during the continuous operation of the equipment. These factors will cause fluctuations in the injection pressure. Only by collecting data of multiple complete injection molding cycles can the actual law of pressure change in the injection molding process be comprehensively and truly reflected, and thus rich and reliable data support can be provided for the accurate analysis of each subsequent module of the system.

[0043] After collecting the pressure data, the data acquisition module will carefully divide these data according to a specific time step, and finally obtain time series pressure data. The setting of this specific time step is not fixed, but needs to be flexibly determined according to the specific characteristics of the injection molding process and the actual production requirements. If the pressure changes rapidly during the injection molding process and the pressure value changes significantly in a short time, then a smaller time step, such as 0.01 seconds, needs to be set to ensure that every instantaneous change in pressure can be accurately captured; conversely, if the pressure changes relatively smoothly and the fluctuation range is small, the time step can be appropriately increased. After the time step division, the originally chaotic pressure data will be arranged in sequence according to the time sequence, forming ordered time series pressure data. This data form provides convenient conditions for subsequent drawing of the pressure smoothing curve and in-depth exploration of the pressure change law, enabling the system to conduct in-depth analysis of the dynamic change of pressure over time.

[0044] Furthermore, the data acquisition module is composed of multiple high-precision pressure sensors, which are respectively installed at the barrel feed inlet, the front end of the screw, and the mold cavity of the injection molding machine.

[0045] In this embodiment, the core of the data acquisition module is multiple high-precision pressure sensors, which accurately detect the pressure changes during the injection molding process. The reason for choosing high-precision pressure sensors is that the injection molding process has extremely high requirements for the accuracy of pressure data, and even extremely small pressure fluctuations may have a significant impact on product quality.

[0046] In the specific implementation process, Kistler 6183A / 6185A series diaphragm pressure sensors can be selected. This type of pressure sensor is designed specifically for injection mold pressure monitoring, with a measuring range covering 0–2000 bar; the accuracy can reach ±0.5% FS, and the dynamic response frequency is 5 kHz. It can monitor the mold cavity pressure changes in real time and optimize the process parameters during the holding pressure stage. Or TE Connectivity M5200 series piezoresistive sensors, which have small size and high sensitivity, and are suitable for injection molding pressure monitoring of small injection molding machines in scenarios with limited space. These types of sensors are placed at key positions of the injection molding machine. At the barrel feed inlet, the sensor can monitor the initial pressure of the plastic raw material entering the barrel in real time. Due to the differences in the entering speed, flow rate, and its own characteristics of the plastic raw material, the pressure here will change, and these changes will directly affect the subsequent injection molding process. Therefore, accurate monitoring of the feed inlet pressure is crucial.

[0047] The sensor at the front end of the screw is responsible for monitoring the pressure change when the screw propels the plastic melt. The screw plays a crucial role in the injection molding process. It pushes the plastic melt into the mold cavity, and the magnitude of its propelling force is directly reflected in the pressure value. By collecting the pressure at the front end of the screw in real time, the system can timely grasp the working state of the screw and then precisely control the injection pressure.

[0048] The mold cavity is the place where the plastic particles finally take shape. The sensor installed here can sensitively sense the pressure condition when the plastic fills the cavity. The pressure distribution in the mold cavity directly determines the molding quality of the product. Problems such as uneven filling, air bubbles or shrink marks are all closely related to the pressure in the cavity. By precisely monitoring the pressure in the mold cavity, the system can adjust the injection molding parameters in a timely manner according to the feedback information to ensure that the produced plastic products have precise dimensions, smooth surfaces, and meet the high-quality production requirements.

[0049] The pressure curve module is used to draw a pressure smoothing curve using time series pressure data, with time as the independent variable and pressure as the dependent variable.

[0050] Furthermore, the pressure smoothing curve is fitted through a generalized additive model and then drawn using R language software.

[0051] In the pressure control system of precision injection molding numerical control equipment, the generation and drawing of the pressure smoothing curve have a rigorous process. First, use the generalized additive model to fit the time series pressure data obtained from the data acquisition module. The generalized additive model can effectively handle the complex non-linear relationship between pressure and time. It breaks through the limitations of traditional linear models and can flexibly capture the pressure change trend.

[0052] During the fitting process, time is used as the independent variable and pressure as the dependent variable. The model analyzes the data and uses smoothing techniques such as spline functions to estimate the influence of the independent variable on the dependent variable, thereby constructing a relationship model that can accurately reflect the change of pressure with time.

[0053] After the fitting is completed, use R language software to draw the pressure smoothing curve. R language is a powerful data analysis and visualization tool with rich plotting functions and extension packages. When drawing the pressure smoothing curve, plotting packages such as ggplot2 can be used. Input the data obtained by fitting through the generalized additive model into the corresponding function, set the horizontal axis as time and the vertical axis as pressure, and adjust the appearance attributes such as the color and line type of the curve to generate an intuitive, clear and beautiful pressure smoothing curve. This method of combining fitting through the generalized additive model and drawing using R language provides a reliable and accurate reference for the pressure change trend for the subsequent curve application module to construct a deep learning model and the pressure control module to achieve precise pressure control.

[0054] In this embodiment, the pressure curve module is responsible for drawing the pressure smoothing curve, and using the Generalized Additive Model (GAM) for fitting and drawing has unique advantages and principles.

[0055] The Generalized Additive Model is a semi-parametric regression model. It relaxes the strict assumption that the independent variable and the dependent variable have a linear relationship in the traditional linear model, allowing for a more flexible and complex non-linear relationship between the independent variable and the dependent variable. In the processing of injection pressure data, the actual pressure change is not always describable by a simple linear equation. It is comprehensively affected by various complex factors such as raw material characteristics, the mechanical performance of the injection molding machine, and mold heat transfer, presenting complex non-linear characteristics.

[0056] When using the Generalized Additive Model to fit and draw the pressure smoothing curve, first, take time as the independent variable and pressure as the dependent variable, and input the time series pressure data obtained by the data acquisition module into the model. GAM analyzes these data and automatically estimates the non-linear influence of each independent variable (time) on the dependent variable (pressure). It will use smoothing techniques such as spline functions according to the actual distribution of the data to fit the data points, thereby constructing a smoothing curve that can accurately reflect the trend of pressure change over time.

[0057] Compared with other simple fitting methods, the pressure smoothing curve drawn by the Generalized Additive Model can better capture the subtle features and complex trends of pressure changes. This enables us to more accurately grasp the dynamic changes of pressure when analyzing the injection molding process, providing a more reliable and accurate reference for the pressure change trend for the subsequent curve application module to construct a deep learning model and the pressure control module to achieve precise pressure regulation, and helping to further improve the quality and precision of injection molded products.

[0058] Furthermore, the Generalized Additive Model includes the following construction steps:

[0059] S1. Determine the model structure: The Generalized Additive Model is used to explore the complex relationship between pressure and time, rather than a simple linear association. The model includes a base value, a part that varies with time, and some random errors. Given that the main focus is on the change of pressure over time, the core is to clarify this special relationship.

[0060] S2. Select the smoothing function: The role of the smoothing function is to make the pressure change curve smoother. Commonly used is the spline function, such as the cubic spline function. It divides time into multiple small segments, and within each small segment, a cubic polynomial is used to describe the pressure change. To make the description accurate, it is necessary to reasonably select the segmentation positions and, through trial and verification, find appropriate parameters so that the function can fit the data well without being overly complex.

[0061] S3. Parameter Estimation: To find the parameters that can make the model accurately reflect the relationship between pressure and time is like adjusting the machine knobs to make it run precisely. Some methods can be adopted, such as determining a set of parameters to minimize the difference between the pressure values predicted by the model and the actual measured values. By continuously adjusting the parameters to minimize the difference, the required parameters can be determined.

[0062] S4. Model Testing and Evaluation: Use the model with adjusted parameters to predict the pressure value and test the prediction accuracy. Some metrics are used to measure the model prediction effect, such as checking the average error between the predicted value and the actual value, or evaluating the overall interpretability of the model to the data. If the model prediction is not good, return to the previous steps to adjust the model structure, smoothing function or parameters until the prediction effect is good.

[0063] S5. Plotting the Pressure Smoothing Curve: According to the adjusted model and parameters, calculate the predicted pressure values at different time points. Use time as the horizontal axis and the predicted pressure values as the vertical axis, and draw a curve with the help of a plotting tool. This curve is the pressure smoothing curve, which can clearly show the change of pressure with time during the injection molding process.

[0064] The curve application module is used to obtain the pressure smoothing curves of products with different sizes and shapes, and construct a deep learning model between the pressure smoothing curves and the sizes and shapes of the products;

[0065] Furthermore, the deep learning model is configured as a convolutional neural network model with the pressure smoothing curve as the input feature and the size and shape of the product as the output target.

[0066] In this embodiment, the deep learning model adopts a convolutional neural network (CNN) architecture, and this choice is closely related to the data characteristics of the pressure smoothing curve. CNN performs excellently in processing data with grid structure or sequential characteristics, and the pressure smoothing curve can be regarded as a kind of sequential data changing with time, which exactly fits the processing advantages of CNN.

[0067] Taking the pressure smoothing curve as the input feature means inputting the information of the change of pressure with time contained in the curve into the model in a specific format. This information includes the pressure values at different time points, the rate of pressure change, the fluctuation amplitude, etc., which are organized in a sequential form to form the initial input of the model. The model sets the size and shape of the product as the output target. The size of the product, such as length, width, height, etc., and the shape features, such as whether it is regular, whether there are special concave and convex structures, etc., are encoded into the output format recognizable by the model.

[0068] Inside the CNN model, the convolutional layer is one of the core components. The convolutional layer slides a convolutional kernel over the input pressure smoothing curve data to perform a convolution operation and extract local features from the data. For example, the convolutional kernel can capture the characteristic pattern of a rapid rise or fall in the pressure curve within a certain time period. Stacking multiple convolutional layers can gradually extract features from simple to complex.

[0069] The pooling layer follows the convolutional layer immediately. Its role is to reduce the dimensionality of the data, reducing the amount of data while retaining key features. For example, using the max pooling or average pooling method, the maximum or average value in the local area is selected to represent the features of that area, reducing the computational complexity and preventing the model from overfitting.

[0070] After being processed multiple times by the convolutional layer and the pooling layer, the data is flattened and input into the fully connected layer. The fully connected layer synthesizes the features extracted previously and, through the operation of the weight matrix, maps them to the output space corresponding to the product size and shape, and finally outputs the prediction result of the product size and shape.

[0071] During the model training process, with a large amount of pressure smoothing curve data with annotations (i.e., known sizes and shapes), the parameters inside the model (such as the weights of the convolutional kernel, the weights of the fully connected layer, etc.) are continuously adjusted to gradually reduce the error between the prediction result of the model and the actual annotation until the model reaches a high prediction accuracy, thus accurately establishing the mapping relationship between the pressure smoothing curve and the product size and shape.

[0072] Furthermore, the convolutional neural network model includes the following construction steps:

[0073] T1. Data preprocessing: After obtaining the existing product injection molding pressure smoothing curve and corresponding size and shape data, comprehensively screen for missing values and outliers among them, reasonably fill in the missing parts, and correct or remove the abnormal data. Subsequently, convert the pressure smoothing curve data into a standardized form with zero mean and unit variance, and at the same time encode the size and shape information of the product into digital features that are easy for the model to recognize, laying a solid foundation for subsequent model training.

[0074] T2. Model Architecture Design: Referring to the dimension of the pressure smoothing curve data, accurately determine the number of input layer nodes to build an entrance for data input. Sequentially construct the convolutional layer, carefully select the number, size, and stride of the convolutional kernels so that they can effectively capture the features of different levels of the pressure curve. After the convolutional layer, reasonably add the pooling layer, and through scientifically setting the pooling window size and stride, perform appropriate dimensionality reduction on the data to reduce the computational burden. According to the complexity of the task and actual requirements, set the fully connected layer as appropriate to achieve the feature integration and mapping from the output of the pooling layer to the final output layer. Finally, determine the number of output layer nodes based on the feature dimensions of the product size and shape to ensure that the model can accurately output the prediction results.

[0075] T3. Model Training: According to the specific nature of the output target, select the loss function. For example, select the mean squared error loss function for regression tasks and the cross-entropy loss function for classification tasks. Select a suitable optimizer from stochastic gradient descent and its derivative optimizers, and finely adjust hyperparameters such as the learning rate. Carefully divide the data into training set, validation set, and test set. Input the training set data into the model, obtain the output through forward propagation and calculate the error, and then rely on backpropagation to update the model parameters, continuously iterating the training process. At the same time, use the validation set to dynamically adjust the hyperparameters to effectively prevent the model from overfitting.

[0076] T4. Model Evaluation and Adjustment: According to the task type, reasonably select the evaluation indicators. For example, for regression tasks, focus on mean squared error and mean absolute error, and for classification tasks, focus on indicators such as accuracy and recall. Use the test set to comprehensively evaluate the trained model. By calculating the values of various evaluation indicators, objectively judge the model performance. If the model performance fails to meet the expected standard, deeply analyze the reasons behind it, and adjust the model architecture, hyperparameter settings, or data preprocessing methods accordingly. Then restart the training and evaluation process until the model performance meets the requirements.

[0077] T5. Model Deployment and Application: Properly save the trained model with qualified performance for ready-to-use. Subsequently, successfully deploy the model into the actual injection molding production environment and deeply integrate it with the production system. In this way, the model can quickly and accurately predict the size and shape of the product based on the real-time input pressure smoothing curve, providing strong support for injection molding production.

[0078] The said pressure control module is used to control the pressure of the injection molding process by using a deep learning model to obtain the corresponding pressure smoothing curve according to the sizes and shapes of two or more products that need to cooperate with each other.

[0079] Further, the pressure control module calculates the pressure deviation based on a preset pressure smoothing curve and real-time acquired pressure data, and processes the data through a PID control algorithm to generate precise control instructions.

[0080] In this embodiment, the pressure control module is a key part of the pressure control system of precision injection molding numerical control equipment, and its operation process closely revolves around the preset pressure smoothing curve and real-time pressure data. This module first compares the pressure data real-time acquired by the injection molding machine with the pressure value at the corresponding time point of the preset pressure smoothing curve obtained through the analysis of the deep learning model according to the sizes and shapes of different products. By subtracting the pressure value of the preset curve at the corresponding time point from the real-time pressure value, the pressure deviation is calculated, and this deviation intuitively reflects the gap between the current injection pressure and the desired pressure.

[0081] Immediately afterwards, the pressure control module inputs the calculated pressure deviation into the PID control algorithm for processing. The PID control algorithm consists of three key links: proportional (P), integral (I), and derivative (D). The proportional link outputs a control action according to a fixed ratio based on the magnitude of the pressure deviation. The larger the deviation, the stronger the control action. The integral link performs an integral operation on the pressure deviation, accumulating the deviations over a period of time in the past to eliminate the steady-state error of the system. The derivative link outputs a control action in advance according to the change rate of the pressure deviation to effectively suppress the overshoot of the system. These three links cooperate with each other to comprehensively and meticulously process the pressure deviation data.

[0082] After being processed by the PID control algorithm, the pressure control module finally generates precise control instructions. These instructions will be transmitted to the relevant actuating mechanisms of the injection molding machine, such as adjusting the screw propulsion speed of the injection molding machine to control the pushing amount of the plastic melt by changing the screw rotation speed, or adjusting the pressure of the hydraulic system to change the clamping force and injection pressure of the injection molding machine. Through the adjustment of these parameters, the actual injection pressure can quickly and accurately follow the change of the preset pressure smoothing curve, ensuring that the injection molding process is always in an ideal state, strongly guaranteeing the production quality and precision of the product, and meeting the requirements of high-quality injection molding production.

[0083] Further, the pressure control module further includes a human-machine interaction module, which is configured with a pressure display screen, and the operator can monitor the injection pressure situation in real time according to the pressure display screen.

[0084] In this embodiment, the human-machine interaction module in the pressure control module greatly improves the convenience and real-time performance of the operator's monitoring of the injection pressure by configuring a pressure display screen. The pressure display screen presents the pressure data during the injection process in an intuitive graphical or numerical form. Without the need to rely on complex data analysis tools, the operator can directly observe the screen to know key information such as the current injection pressure value and the pressure change trend.

[0085] For example, when the pressure value approaches or exceeds the preset reasonable range, the display screen can issue a warning through color changes, flashing, etc. Based on this, the operator can quickly judge whether the injection process is normal. Once abnormal pressure is found, corresponding measures can be taken in a timely manner. If the pressure is too high, the parameters of the PID control algorithm can be adjusted through the human-machine interaction module to reduce the pressure; if the pressure is too low, the reverse operation can be performed to increase the pressure.

[0086] In addition, the human-machine interaction module can also design an operation interface that allows the operator to flexibly adjust the preset pressure smoothing curve according to the production requirements of different products, thereby optimizing the injection pressure control. This function enables the operator to more accurately control the injection pressure when facing diverse production tasks, ensuring that the products produced meet the quality standards and improving production efficiency and product quality.

[0087] The above is only a preferred embodiment of the present invention and does not impose any form of limitation on the present invention. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or decorations equivalent to changes within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modifications, equivalent changes, and decorations made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A pressure control system for precision injection molding CNC equipment, characterized in that: It includes data acquisition module, pressure curve module, curve application module and pressure control module; The data acquisition module is used to collect pressure data of multiple complete injection molding cycles, and the pressure data is divided into specific time steps to obtain time series pressure data; The pressure curve module is used to draw a pressure smooth curve using time series pressure data, with time as the independent variable and pressure as the dependent variable; The curve application module is used to obtain the pressure smoothing curves of products of different sizes and shapes, and to construct a deep learning model between the pressure smoothing curve and the size and shape of the product; The pressure control module is used to control the pressure of the injection molding process by using a deep learning model to obtain a corresponding pressure smooth curve according to the size and shape of two or more products that need to cooperate with each other; The deep learning model is configured as a convolutional neural network model with a pressure smooth curve as an input feature and a product size and shape as an output target; The convolutional neural network model includes the following construction steps: T1. Data preprocessing: convert the pressure smoothing curve data into a standardized form with zero mean and unit variance, and encode the size and shape information of the product; T2. Model architecture design: Determine the number of input layer nodes based on the dimension of the pressure smooth curve data; construct a convolution layer to capture the characteristics of the pressure smooth curve, and determine the number, size, and step size of the convolution kernels; add a pooling layer to reduce the dimensionality of the data; set up a fully connected layer to achieve feature integration and mapping from the output of the pooling layer to the final output layer; Determine the number of output layer nodes based on the characteristic dimensions of product size and shape; T3, model training: select the loss function and optimizer according to the output target, and adjust the hyperparameters; divide the data into training set, validation set and test set, input the training set data into the model, obtain the output and calculate the error with the help of forward propagation, and then update the model parameters through back propagation, and continuously iterate the training process; T4. Model evaluation and adjustment: Use the test set to evaluate the trained model and determine the model performance by calculating evaluation indicators.

2. A pressure control system for precision injection molding CNC equipment according to claim 1, characterized in that: The data acquisition module is composed of a plurality of high-precision pressure sensors, and the sensors are respectively installed at the barrel feed port, the front end of the screw and the mold cavity of the injection molding machine.

3. A pressure control system for precision injection molding CNC equipment according to claim 1, characterized in that: The pressure smoothing curve is fitted by a generalized additive model and then drawn by R language software.

4. A pressure control system for precision injection molding CNC equipment according to claim 3, characterized in that: The generalized additive model includes the following construction steps: S1. Determine the model structure: The model uses pressure as the response variable and the corresponding explanatory variable is time; S2. Select smoothing function: select cubic spline function as the smoothing function of the model, and divide the time into multiple small segments. In each small segment, use cubic polynomial to describe the pressure change; S3. Parameter estimation: By continuously adjusting the model parameters, the gap between the predicted value and the actual value of the pressure is minimized, thereby determining the final parameters; S4, model verification and evaluation: use the model with adjusted parameters to predict the pressure value, and use the determination coefficient to verify the accuracy of the predicted value; S5. Draw a smooth pressure curve: Calculate the predicted pressure values ​​at different time points based on the adjusted model and parameters; and draw a smooth pressure curve with the help of drawing tools, with time as the horizontal axis and the predicted pressure value as the vertical axis.

5. The pressure control system of a precision injection molding CNC equipment according to claim 3 is characterized in that: The generalized additive model is expressed as: ， In the formula, g Indicates injection pressure; x Indicates time; s represents a smooth function; k The number of nodes representing the smoothing function; μ represents the intercept term.

6. The pressure control system of a precision injection molding CNC equipment according to claim 1, characterized in that: The pressure control module calculates the pressure deviation according to the preset pressure smoothing curve and the pressure data collected in real time, and performs data processing through the PID control algorithm to generate accurate control instructions.

7. The pressure control system of precision injection molding CNC equipment according to claim 1, characterized in that: The pressure control module also includes a human-computer interaction module, which is equipped with a pressure display screen, and the operator monitors the injection pressure in real time according to the pressure display screen.

Images