Vacuum glue filling control method and system based on machine learning

Abstract

The invention provides a vacuum glue filling control method and system based on machine learning, and relates to the technical field of intelligent control, and the method comprises the steps: defining the geometric shape, material attributes, boundary conditions and initial conditions of a finite element model to simulate a real glue filling environment, and the physical characteristics comprise the rheological property and thermal conductivity of glue; the glue filling process under different glue filling speeds, glue temperatures and vacuum degrees is simulated through the three-dimensional geometric model of the glue filling equipment, so that simulation results are obtained, and the simulation results comprise glue flowing behaviors, temperature distribution and stress distribution; training the simulation result to construct a prediction model for predicting the glue pouring result; real-time parameters in the glue filling process are obtained, wherein the real-time parameters comprise the real-time glue filling speed, the glue temperature, the vacuum degree and the glue filling time; according to the method, the flowing and filling conditions of the glue can be accurately predicted and controlled in the glue filling process, so that the glue filling precision and consistency are improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent control technology, and in particular to a vacuum glue filling control method and system based on machine learning. Background Art

[0002] The traditional vacuum glue filling control method, although it will refer to the operator's experience and skills to a certain extent, still mainly relies on preset parameters and fixed control logic. In this method, the operator will set a relatively fixed set of key parameters such as glue filling speed, glue temperature, vacuum degree, etc. according to the standard working procedures of the glue filling equipment, the recommended use conditions of the glue, and common environmental factors. However, due to the complexity and variability of the actual glue filling environment, this control method based on preset parameters and fixed logic often cannot guarantee the consistency and stability of glue filling.

[0003] Specifically, when faced with different glue filling requirements, changes in glue properties, or fluctuations in environmental factors, the preset parameters and fixed control logic may not be able to adapt to these changes, resulting in glue overflow, uneven filling, or bubble residue during the filling process. Although operators will try to make adjustments based on actual conditions, it is still difficult to completely avoid the above problems due to the limitations of accuracy and timeliness of adjustments. These problems not only affect the quality and performance of the product, but may also cause production line shutdowns and waste of resources. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a vacuum glue pouring control method and system based on machine learning, which helps to accurately predict and control the flow and filling of glue during the glue pouring process, thereby improving the accuracy and consistency of glue pouring.

[0005] In order to solve the above technical problems, the technical solution of the present invention is as follows: In a first aspect, a vacuum glue filling control method based on machine learning is provided, the method comprising: According to the physical characteristics of the vacuum glue filling process, a three-dimensional geometric model of the glue filling equipment is constructed and the three-dimensional geometric model is meshed; the geometric shape, material properties, boundary conditions and initial conditions of the finite element model are defined to simulate the real glue filling environment. The physical characteristics include the rheology and thermal conductivity of the glue; The three-dimensional geometric model of the glue filling equipment is used to simulate the glue filling process under different glue filling speeds, glue temperatures, and vacuum degrees to obtain simulation results, including glue flow behavior, temperature distribution, and stress distribution; The simulation results are trained to construct a prediction model for predicting the glue filling results; Obtain real-time parameters during the glue filling process, including real-time glue filling speed, glue temperature, vacuum degree and glue filling time; According to the real-time glue filling speed, glue temperature, vacuum degree and glue filling time and prediction model, the predicted glue filling result can be obtained; Use the three-dimensional geometric model of the glue filling equipment to simulate and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time to obtain the analysis results; Evaluate the key indicators of glue flow behavior and temperature distribution under current glue filling conditions based on the analysis results; Dynamically adjust the control parameters in the glue pouring process based on the predicted glue pouring results and key indicators.

[0006] Furthermore, a finite element model is constructed according to the physical characteristics of the vacuum filling process, including: Determine the physical phenomena to be simulated, the geometric dimensions, and the simulation time. The physical phenomena include glue flow behavior, temperature distribution, and stress distribution. Determine the physical parameters and material properties to be simulated. Material properties include density, viscosity, thermal conductivity and specific heat capacity of the glue; physical parameters include vacuum degree, ambient temperature, pressure and the geometry of the glue filling equipment; According to physical phenomena, geometric dimensions, simulation time, physical parameters and material properties, a three-dimensional geometric model of the glue potting equipment is constructed, wherein the three-dimensional geometric model of the glue potting equipment is a finite element model.

[0007] Furthermore, the simulation results are trained to construct a prediction model for predicting the glue filling results, including: Preprocessing the simulation result data to obtain a preprocessed simulation data set; Divide the simulated data set into training set and test set; According to the characteristics of the simulated data set, determine the neural network structure, activation function and loss function. The neural network structure includes input layer, hidden layer and output layer; Initialize the weights and biases of the neural network structure, set the learning rate, batch size, and number of training rounds during training; According to the simulated data set and the neural network structure, the output data is calculated through forward propagation; Calculate the loss value of the neural network structure according to the loss function, and update the weights and biases of the neural network structure through the back propagation algorithm, and repeat the steps until the preset number of training rounds is reached to obtain the neural network model; Use the test set to evaluate the neural network model and evaluate the performance indicators of the neural network model on the test set to obtain the evaluation results; The neural network model is optimized according to the evaluation results to obtain a trained neural network model, wherein the trained neural network model is a prediction model.

[0008] Furthermore, the simulated data set is divided into a training set and a test set, including: Set the size of the particle swarm, where each particle represents a data set partitioning scheme, including the ratio or specific index of the training set and the test set; initialize the position and velocity of each particle; Determine a fitness function for evaluating the quality of the data partitioning scheme represented by each particle; For each particle, update its position according to its current position and velocity to generate a new data partitioning scheme; Update the individual position and global position of each particle, and adjust the speed of each particle according to the individual position and the global position until the preset number of iterations is reached to obtain the final division scheme; According to the final partitioning scheme, the simulated data set is divided into a training set and a test set.

[0009] Furthermore, according to the simulated data set and the neural network structure, the output data is calculated through forward propagation, including: Pass the input data from the simulated dataset to the input layer of the neural network; For each neuron in the input layer, the input data is passed to the first hidden layer. During this process, the weights between the input data and the neurons in the first hidden layer are weighted and summed, and a bias term is added. For each hidden layer, the output data of the neurons in the previous layer is received as input signal; The input signal is weighted and summed and a bias term is added to obtain a processing result, and the processing result is processed by an activation function to generate output data of the current hidden layer; The output data of the current hidden layer is passed to the next layer until the last hidden layer is reached; The output data of the last hidden layer is passed to the output layer; In the output layer, the output data undergoes weighted summation and bias adjustment again, and the final prediction result of the neural network is calculated through the activation function. The final prediction result is the prediction result of the simulated data set.

[0010] Furthermore, the control parameters include glue filling speed, glue temperature and vacuum degree.

[0011] Furthermore, the calculation formula of the fitness function is: ; in, It is The weighted accuracy of each particle; represents the number of different categories in the simulated dataset; It is The weight of each category; For the particles, the number of true positives, indicating that the correct prediction is The number of samples in the class; For the particles, the number of false positives, indicating that the The sample prediction of class The number of classes, where = ; Is the first Classification Cost of class; Indicates that for particles, the number of false negatives, indicating that the The number of sample predictions for the class; Represents the index of the particle; Indicates the index of the category.

[0012] In a second aspect, a vacuum glue filling control system based on machine learning is applied to the method described, comprising: The partitioning module is used to build a 3D geometric model of the glue filling equipment according to the physical characteristics of the vacuum glue filling process and to mesh the 3D geometric model; define the geometry, material properties, boundary conditions and initial conditions of the finite element model to simulate the real glue filling environment. The physical characteristics include the rheology and thermal conductivity of the glue; The simulation module is used to simulate the glue filling process under different glue filling speeds, glue temperatures, and vacuum degrees through the three-dimensional geometric model of the glue filling equipment to obtain simulation results, including glue flow behavior, temperature distribution, and stress distribution; A construction module is used to train the simulation results to build a prediction model for predicting the glue filling results; The acquisition module is used to obtain the real-time parameters in the glue filling process, including the real-time glue filling speed, glue temperature, vacuum degree and glue filling time; The prediction module is used to obtain the predicted glue filling result according to the real-time glue filling speed, glue temperature, vacuum degree and glue filling time and the prediction model; The analysis module is used to simulate and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time by using the three-dimensional geometric model of the glue filling equipment to obtain the analysis results; Evaluation module, used to evaluate the key indicators of glue flow behavior and temperature distribution under the current glue filling conditions based on the analysis results; The adjustment module is used to dynamically adjust the control parameters in the glue filling process according to the predicted glue filling results and key indicators.

[0013] According to a third aspect, a computing device includes: one or more processors; The storage device is used to store one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors implement the method described.

[0014] In a fourth aspect, a computer-readable storage medium stores a program, and when the program is executed by a processor, the method described is implemented.

[0015] The above solution of the present invention includes at least the following beneficial effects: By constructing a three-dimensional geometric model of the glue filling equipment and combining it with machine learning technology, the present invention can accurately simulate the glue flow behavior, temperature distribution and stress distribution under different glue filling conditions. This helps to accurately predict and control the flow and filling of glue during the glue filling process, thereby improving the accuracy and consistency of glue filling and reducing the occurrence of problems such as glue overflow and uneven filling.

[0016] The present invention uses a prediction model to predict and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time, which can timely discover potential glue filling problems and dynamically adjust control parameters according to the analysis results. This optimization method can not only improve glue filling efficiency, but also effectively avoid risks such as waste of resources and production line shutdown.

[0017] The present invention reduces the reliance on operator experience through machine learning and simulation analysis technology, so that even inexperienced operators can complete high-quality glue pouring work under the guidance of the system. Since the present invention can accurately control various parameters in the glue pouring process to ensure that the glue is filled and cured in the best state, the quality and performance of the product can be significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of a vacuum glue filling control method based on machine learning provided in an embodiment of the present invention.

[0019] Figure 2 It is a schematic diagram of a vacuum glue filling control system based on machine learning provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in a form and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0021] like Figure 1As shown, an embodiment of the present invention proposes a vacuum glue filling control method based on machine learning, and the method comprises the following steps: Step 11, construct a three-dimensional geometric model of the glue filling equipment according to the physical characteristics of the vacuum glue filling process, and mesh the three-dimensional geometric model; define the geometric shape, material properties, boundary conditions and initial conditions of the finite element model to simulate the real glue filling environment, and the physical properties include the rheology and thermal conductivity of the glue; Step 12, simulating the glue filling process under different glue filling speeds, glue temperatures, and vacuum degrees through a three-dimensional geometric model of the glue filling equipment to obtain simulation results, including glue flow behavior, temperature distribution, and stress distribution; Step 13, training the simulation results to construct a prediction model for predicting the glue filling results; Step 14, obtaining real-time parameters during the glue filling process, the real-time parameters including real-time glue filling speed, glue temperature, vacuum degree and glue filling time; Step 15, obtaining a predicted glue filling result according to the real-time glue filling speed, glue temperature, vacuum degree and glue filling time and the prediction model; Step 16, using the three-dimensional geometric model of the glue filling equipment to simulate and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time to obtain analysis results; Step 17, evaluating the key indicators of glue flow behavior and temperature distribution under the current glue filling conditions according to the analysis results; Step 18, dynamically adjusting the control parameters in the glue pouring process according to the predicted glue pouring results and key indicators; the control parameters include glue pouring speed, glue temperature and vacuum degree.

[0022] In an embodiment of the present invention, step 11, by constructing an accurate three-dimensional geometric model and meshing it, the real glue filling environment can be simulated more accurately. At the same time, considering the physical properties of glue such as rheology and thermal conductivity, the simulation results are closer to the actual situation, and the accuracy and credibility of the simulation are improved. Step 12, by simulating the glue filling process under different glue filling conditions, a rich simulation result can be obtained, including glue flow behavior, temperature distribution and stress distribution. Step 13, by training the simulation results, a prediction model that can predict the glue filling results is constructed. This step enables the glue filling results to be quickly predicted according to the current parameters in the subsequent real-time glue filling process, so as to adjust the control parameters in time to ensure the quality and efficiency of glue filling. Step 14, real-time acquisition of key parameters in the glue filling process, including glue filling speed, glue temperature, vacuum degree and glue filling time, etc., provides real-time data support for subsequent prediction and analysis. This helps to timely discover abnormal conditions in the glue filling process and take corresponding measures to intervene. Step 15, combined with real-time parameters and prediction models, the predicted glue filling results can be quickly obtained. This enables operators to promptly understand the possible glue pouring results during the glue pouring process, so that they can make corresponding adjustments and improve the accuracy and stability of glue pouring.

[0023] Step 16, using a three-dimensional geometric model to simulate and analyze the real-time parameters, more detailed analysis results can be obtained, which helps to have a deeper understanding of the characteristics of the glue flow behavior and temperature distribution under the current glue filling conditions. Step 17, evaluating key indicators based on the analysis results, can quantitatively evaluate the characteristics of the glue flow behavior and temperature distribution under the current glue filling conditions, which provides operators with clear evaluation criteria, allowing them to judge whether the glue filling process meets the expected requirements based on these indicators, and take corresponding adjustment measures in time. Step 18, dynamically adjusting the control parameters based on the predicted glue filling results and key indicators can ensure that the glue filling process is always carried out in the best state. This not only improves the quality and efficiency of glue filling, but also reduces the dependence on the operator's experience, making the entire glue filling process more intelligent and automated. At the same time, by adjusting the control parameters in real time, it can also effectively respond to various emergencies and environmental changes to ensure the stability and reliability of the glue filling process.

[0024] In a preferred embodiment of the present invention, the above step 11, constructing a finite element model according to the physical characteristics of the vacuum casting process, may include: Step 111, determining the physical phenomena to be simulated, geometric dimensions and simulation time, the physical phenomena including glue flow behavior, temperature distribution and stress distribution; Step 112, determining the physical parameters and material properties to be simulated, the material properties include density, viscosity, thermal conductivity and specific heat capacity of the glue; the physical parameters include vacuum degree, ambient temperature, pressure and geometric shape of the glue filling equipment; Step 113, constructing a three-dimensional geometric model of the glue potting equipment according to physical phenomena, geometric dimensions, simulation time, physical parameters and material properties, wherein the three-dimensional geometric model of the glue potting equipment is a finite element model.

[0025] In an embodiment of the present invention, step 111 provides a clear direction and scope for building an accurate finite element model by determining the physical phenomena to be simulated (such as glue flow behavior, temperature distribution and stress distribution), geometric dimensions and simulation time. This helps to ensure the pertinence and accuracy of the simulation, so that the simulation results are more consistent with the physical characteristics of the actual glue filling process. Step 112, determining the physical parameters and material properties to be simulated is a key step in building a finite element model. By considering in detail the material properties of the glue such as density, viscosity, thermal conductivity and specific heat capacity, as well as physical parameters such as vacuum degree, ambient temperature, pressure and the geometry of the glue filling equipment, a more realistic and comprehensive glue filling environment model can be constructed. This helps to improve the accuracy and credibility of the simulation, so as to better predict and control the actual glue filling process. Step 113, based on the physical phenomena, geometric dimensions, simulation time, physical parameters and material properties determined in the previous steps, a three-dimensional geometric model of the glue filling equipment is constructed. This step is the basis of the entire simulation process. By accurately constructing a three-dimensional geometric model, the actual structure and working environment of the glue filling equipment can be more realistically reflected, which helps to more deeply understand the dynamic behavior of the glue filling process.

[0026] The following is a detailed description of the implementation process of step 111 to step 113: Step 111, analyze the main physical phenomena involved in the vacuum glue filling process, which have a direct impact on the quality and efficiency of glue filling; select glue flow behavior, temperature distribution and stress distribution as key physical phenomena for simulation, because these phenomena determine the filling uniformity, curing effect and structural strength of the glue. Measure or obtain detailed geometric dimension data of glue filling equipment (such as glue filling chamber, glue hose, nozzle, etc.); set the corresponding geometric parameters in the simulation software according to the actual size of the equipment to ensure that the simulated geometric environment is consistent with the actual. Evaluate the duration of the glue filling process, which is based on the fluidity of the glue, the amount of glue filling and the working efficiency of the equipment; set the total time range of the simulation to cover the entire process from the beginning to the end of glue filling to ensure that all key dynamic changes can be captured.

[0027] Step 112, through experimental measurement or reference to relevant literature, obtain the key material properties of the glue, such as density, viscosity, thermal conductivity and specific heat capacity. These properties will directly affect the flow properties and heat transfer characteristics of the glue and must be accurately set in the simulation; evaluate the external physical conditions such as vacuum degree, ambient temperature, pressure, etc. involved in the glue filling process; according to the actual process requirements, set the specific values ​​or ranges of these physical parameters in the simulation; the geometric shape of the glue filling equipment will also be input into the simulation model as an important physical parameter.

[0028] Step 113, collect detailed design information of the glue filling equipment, including CAD drawings, equipment specifications, etc.; select appropriate 3D modeling software (such as SolidWorks, AutoCAD, etc.) to build the model; in the modeling software, gradually build the various components of the glue filling equipment according to the geometric dimensions, physical parameters and material properties determined in steps 111 and 112; ensure the accuracy and detail integrity of the model, especially for key parts that affect the flow of glue and temperature distribution (such as the corners of the glue filling cavity, the structure of the nozzle, etc.); after completing the preliminary modeling, verify the model to check whether there are errors or unreasonable designs, and optimize and adjust the model according to the verification results until the simulation requirements are met.

[0029] For example, in the glue pouring process of an electronic device, the physical phenomena to be simulated are the flow behavior of glue in a vacuum environment, the temperature distribution change of glue during the glue pouring process, and the stress distribution generated after the glue is cured. The geometric dimensions of the device include the length, width and height of the glue pouring cavity, the diameter and length of the nozzle, etc. The simulation time is set to the time required from the start of glue pouring to the complete curing of the glue, such as 30 minutes.

[0030] Through experimental measurements, the density of the glue is 1.2g / cm³, the viscosity is 500cP, the thermal conductivity is 0.2W / (m·K), and the specific heat capacity is 1.5J / (g·K). At the same time, according to the process requirements, the vacuum degree in the simulation is set to -0.1MPa, the ambient temperature is 25℃, and the pressure in the glue filling cavity is normal pressure. The geometric shape of the equipment is imported into the simulation software through precise CAD drawings. SolidWorks software is used to build a three-dimensional geometric model of the glue filling equipment according to the CAD drawings and equipment specifications. The model presents key components such as the glue filling cavity, nozzle, vacuum system, and temperature control system in detail. After verification and optimization, the model was used in subsequent simulation analysis.

[0031] In another preferred embodiment of the present invention, the above step 11 meshes the three-dimensional geometric model; defines the geometric shape, material properties, boundary conditions and initial conditions of the finite element model to simulate the real glue pouring environment. The physical properties include the rheology and thermal conductivity of the glue, which may include: Open the mesh generation software, such as ANSYS Meshing or HyperMesh; import the previously constructed 3D geometry model file and ensure that the file format is compatible with the software. Analyze the complexity of the model and determine the areas that need to be finely divided, such as the glue flow path, temperature change sensitive areas, etc.; according to the simulation accuracy requirements, set the global mesh size and perform local mesh refinement in key areas to improve the accuracy of the simulation; select the mesh shape and type, such as tetrahedron, hexahedron, etc., to adapt to the geometric characteristics of the model; start the meshing program and monitor the meshing process to ensure that no errors or warning messages appear; after completion, check the quality of the generated mesh to ensure that there is no deformed mesh or excessive mesh aspect ratio; save and export the finite element mesh model for subsequent finite element analysis.

[0032] Define the geometry, material properties, boundary conditions, and initial conditions of the finite element model: Open finite element analysis software, such as ANSYS Workbench or Abaqus; import the previously generated finite element mesh model to ensure the accuracy of the geometric shape; obtain the key material properties of the glue, such as density, viscosity, thermal conductivity, specific heat capacity, etc. according to experimental data or literature; create a new material property set for the glue material in the software, and enter the corresponding property values ​​to ensure that the material properties are correctly applied to the corresponding mesh units. According to the actual situation of the glue filling equipment, set fixed constraints, such as fixed supports for the equipment housing, etc.; define the inlet and outlet conditions of the glue, including flow, velocity distribution or pressure distribution, etc.; set the boundary conditions of vacuum to simulate the real vacuum glue filling environment, and specify external boundary conditions such as ambient temperature and pressure.

[0033] Determine the initial glue temperature at the beginning of the simulation and apply it to the corresponding mesh elements, set the initial flow velocity of the glue to reflect the state at the beginning of the glue filling process, and check and confirm that all initial conditions and boundary conditions are set correctly. After completing the above steps, you can get a complete and accurate finite element model to simulate the real glue filling environment and its physical properties, including the rheology and thermal conductivity of the glue. Next, you can perform simulation calculations to analyze the dynamic behavior and performance of the glue filling process.

[0034] In another preferred embodiment of the present invention, in the above step 12, the three-dimensional geometric model of the glue filling equipment is used to simulate the glue filling process under different glue filling speeds, glue temperatures, and vacuum degrees to obtain simulation results. The simulation results include glue flow behavior, temperature distribution, and stress distribution, and may include: Set different glue pouring speeds according to experimental requirements or actual production scenarios. This can be achieved by adjusting the speed value or flow value in the inlet boundary condition. Modify the initial temperature value of the glue to simulate the glue pouring process at different temperatures. Make sure this temperature value is within the material's tolerance range. Adjust the vacuum setting in the simulation environment. This involves modifying the ambient pressure boundary condition to reflect the effect of different vacuum levels on the glue pouring process.

[0035] According to the physical characteristics of the glue filling process, select a solver dedicated to fluid dynamics simulation. For example, if the simulation involves the viscous flow of glue, select a solver that can handle non-Newtonian fluids. Confirm whether the selected solver supports all physical effects involved in the simulation, such as heat conduction, convection, etc. Enter the parameter setting interface of the solver, and set an appropriate time step according to the accuracy requirements and computing resources of the simulation. A smaller time step can improve the accuracy of the simulation. Set the number of iterations. For complex nonlinear problems, multiple iterations are required to obtain a converged solution. According to the complexity of the problem and the recommended value of the solver, set an upper limit on the number of iterations, check and adjust other related parameters, such as convergence criteria, relaxation factors, etc., to ensure the stability and convergence of the solution process.

[0036] After confirming that all solver parameters have been set and are correct, click "Start" or a similar button to start the simulation program. During the simulation, pay close attention to the output information of the software, including calculation progress, residual convergence, etc. This information can help determine whether the simulation is running normally and whether the expected accuracy requirements are met. If a warning or error message appears during the simulation, the simulation should be paused immediately and detailed information should be viewed. According to the type and content of the warning or error, take appropriate measures to troubleshoot and repair. For example, it may be necessary to adjust the meshing, modify the boundary conditions, or reset the solver parameters. Make sure to keep monitoring the software running status throughout the simulation process until the simulation is completed and the final result data is output.

[0037] After the simulation is completed, extract the simulation result data. These data include the flow behavior of the glue at different time points (velocity field, pressure field, etc.), temperature distribution map, and stress distribution map; use the post-processing tools provided by the software to visualize and quantitatively analyze the simulation results. For example, you can create a streamline diagram to show the flow path of the glue, or generate a temperature cloud map to show the temperature distribution inside the device; based on the analysis results, evaluate the effect of the glue filling process under different parameter combinations. Pay special attention to whether the glue is evenly filled, whether there are potential problems such as excessive temperature gradient or stress concentration.

[0038] According to the problems or deficiencies found in the simulation results, adjust the design parameters or process conditions of the glue filling equipment to optimize the glue filling process. Compare and verify the simulation results with the experimental results to ensure the accuracy and reliability of the simulation. This can be done by designing corresponding experimental plans and conducting glue filling experiments in actual environments. Based on the comparison results, further adjust and improve the simulation model and parameter settings to improve the prediction ability of the simulation.

[0039] In another preferred embodiment of the present invention, the above step 13, training the simulation results to construct a prediction model for predicting the glue filling results, may include: Step 131, preprocessing the simulation result data to obtain a preprocessed simulation data set; Step 132, dividing the simulated data set into a training set and a test set; Step 133, determining a neural network structure, an activation function, and a loss function according to the characteristics of the simulated data set, wherein the neural network structure includes an input layer, a hidden layer, and an output layer; Step 134, initializing the weights and biases of the neural network structure, and setting the learning rate, batch size, and number of training rounds during the training process; Step 135, calculating output data through forward propagation according to the simulated data set and the neural network structure; Step 136, calculating the loss value of the neural network structure according to the loss function, and updating the weight and bias of the neural network structure through the back propagation algorithm, and repeating the steps until a preset number of training rounds is reached to obtain a neural network model; Step 137, using the test set to evaluate the neural network model, evaluating the performance index of the neural network model on the test set to obtain an evaluation result; Step 138, optimizing the neural network model according to the evaluation result to obtain a trained neural network model, wherein the trained neural network model is a prediction model.

[0040] In an embodiment of the present invention, step 131, data preprocessing can clean and organize the original data, eliminate outliers, missing values ​​and duplicate values, and improve the quality and consistency of the data. Step 132, by dividing the training set and the test set, it can be ensured that the model has independent verification data during the training process, thereby avoiding overfitting and improving the generalization ability of the model. Step 133, designing a suitable neural network structure, activation function and loss function for the characteristics of the data set can enable the model to better capture the complex relationships in the data and improve the prediction accuracy and efficiency of the model. Step 134, initializing weights and biases and setting training parameters can accelerate the convergence speed of the model, reduce training time, and avoid problems such as gradient disappearance or gradient explosion in the model during training. Step 135, forward propagation is the basic step of neural network training, and outputs data through calculation. Step 136, by continuously optimizing the weights and biases of the neural network through the back propagation algorithm, the prediction error of the model can be gradually reduced and the prediction performance of the model can be improved. At the same time, setting the number of training rounds can ensure that the model stops training after reaching a certain performance, avoiding performance degradation caused by overtraining. Step 137, using an independent test set to evaluate the model can objectively measure the performance of the model and ensure the reliability and effectiveness of the model in practical applications. Step 138, optimizing the model based on the evaluation results can further improve the performance of the model and make it better meet the needs of practical application scenarios. At the same time, the optimization process can also help discover potential problems in the model.

[0041] When applied specifically, the above steps include the following processes: Step 131, remove invalid values, missing values ​​and outliers in the simulation result data; convert the simulation result data to the same dimension, use z-score standardization or min-max normalization; select features with high correlation with the glue filling results according to the characteristics of the simulation data; transform nonlinear features, such as logarithmic transformation, polynomial transformation, etc., to improve the fitting ability of the model; encode categorical features, such as One-Hot Encoding.

[0042] Step 132, randomly divide the simulated data set into a training set and a test set; Step 133, analyze the simulated data set, including the type (such as numerical type, classification type, etc.), range, and distribution of the data; identify the key features in the data set; determine the number of neurons in the input layer based on the number of features of the simulated data set, and each neuron corresponds to a feature value; the number of hidden layers and the number of neurons in each layer are parameters when designing the neural network, and these parameters are determined by experience (such as gradually increasing the number of layers or the number of neurons until the performance no longer improves); the number of neurons in the output layer is related to the complexity of the prediction task. In the prediction of the glue filling result, it may be one or more continuous values ​​(such as glue filling amount, position, etc.), so the output layer can be set to the corresponding number of neurons. Select an activation function, for example, the ReLU function. Determine the loss function. For regression tasks (such as glue filling amount prediction), the loss function includes the mean squared error (MSE).

[0043] Step 134, use a random initialization method (such as He initialization, Xavier initialization) to assign values ​​to the weights and biases of the neural network; determine the learning rate (such as 0.01, 0.001, etc.), batch size (such as 32, 64, 128, etc.) and number of training rounds (such as 100, 200 rounds, etc.).

[0044] Step 135, input the data in the training set into the neural network in batches; according to the hierarchical structure of the neural network, calculate the output value of each neuron layer by layer until the result of the final output layer is obtained.

[0045] Step 136, measure the difference between the network prediction and the true label through a loss function (such as mean square error). For the current batch of data, compare the network output with the true label and calculate the loss value through the loss function. This loss value reflects the degree of mismatch between the model prediction and the true data under the current network parameters. Use the back propagation algorithm to calculate the gradient of the loss value with respect to each weight and bias: Backpropagation is an efficient way to calculate the gradients of weights and biases in a neural network. In this process, the gradients of the output layer with respect to the loss function are first calculated, and then these gradients are propagated forward layer by layer until the input layer of the network is reached. For each layer, the gradient of the loss function with respect to the weights and biases of that layer is calculated according to the chain rule. These gradients indicate how the parameters of the network should be adjusted to reduce the loss.

[0046] Update the weights and biases of the neural network based on the learning rate and the calculated gradient: With the gradient information, you can use gradient descent or its variants (such as Adam, RMSprop, etc.) to update the weights and biases of the network. The learning rate is an important hyperparameter that controls the step size of the parameter update. A larger learning rate may lead to unstable training, while a smaller learning rate may lead to slow training. During the update process, each weight and bias is subtracted from the product of its corresponding gradient and the learning rate. In this way, the parameters of the network will be adjusted in the direction of reducing the loss. Repeat the above steps until the preset number of training rounds is reached. Training a neural network requires multiple rounds of iterations (epochs). Each round will traverse the entire dataset once (or more times, depending on the batch size); by repeating the above steps, the parameters of the network will gradually be optimized, so that the performance of the model on the training data will be improved; when the preset number of training rounds is reached, the training process will stop to obtain the trained neural network model. At this point, the performance of the model on the validation set or test set can be evaluated to determine its generalization ability.

[0047] Step 137, input the data in the test set into the trained neural network; calculate the performance indicators (such as accuracy, recall, F1 score, etc.) based on the true labels of the test set and the prediction results of the neural network.

[0048] Step 138, analyze the evaluation results of the test set to find out the problems of the neural network model (such as overfitting, underfitting, etc.); adjust the structure of the neural network (such as increasing or decreasing the number of hidden layers, the number of neurons, etc.), change the activation function or loss function, adjust the training parameters, etc. according to the analysis results to optimize the performance of the model, and use the optimized model to repeat the training and evaluation steps until satisfactory performance is achieved.

[0049] In another preferred embodiment of the present invention, the above step 132, dividing the simulated data set into a training set and a test set, may include: Step 1321, setting the size of the particle swarm, where each particle represents a data set partitioning scheme, including the ratio or specific index of the training set and the test set; initializing the position and velocity of each particle; Step 1322, determining a fitness function for evaluating the quality of the data partitioning scheme represented by each particle; Step 1323, for each particle, update its position according to its current position and velocity to generate a new data partitioning scheme; Step 1324, updating the individual position and global position of each particle, and adjusting the speed of each particle according to the individual position and the global position until a preset number of iterations is reached to obtain a final division scheme; Step 1325, according to the final division scheme, divide the simulated data set into a training set and a test set.

[0050] In an embodiment of the present invention, the particle swarm optimization algorithm can be used to systematically search for the best solution for data set partitioning. The design of the fitness function can ensure that the partitioning scheme not only meets the basic training-testing ratio requirements, but also takes into account more complex factors such as data distribution and category balance, thereby optimizing the effect of model training and verification. The particles in the particle swarm algorithm represent different partitioning schemes, which means that the algorithm can flexibly explore a variety of possible partitioning methods. By adjusting the initial position, speed and update strategy of the particles, the algorithm can easily adapt to different data set characteristics and partitioning requirements. The particle swarm optimization algorithm has a powerful global search capability. Through the update mechanism of individual position and global position, the algorithm can find the optimal solution in the entire search space, avoid falling into the local optimum, and thus obtain a more reliable data set partitioning scheme.

[0051] When applied specifically, the above step 1321 may include: Determine the number of particles in the particle swarm, which is a preset parameter. Too many particles may increase the computational burden, while too few particles may lead to insufficient search space and affect the optimization effect. Each particle represents a data set partitioning scheme, which can be an array representing the index of the training set and the test set, or a numerical value representing the partition ratio. For example, if the data set has 1000 samples, a particle may be represented as a binary array containing 1000 elements, where 1 indicates that the sample with the corresponding index is in the training set and 0 indicates that it is in the test set. Initialize the particle position and velocity. The position initialization can be random to ensure the diversity of the initial partitioning scheme.

[0052] Step 1322, defining a fitness function, where the fitness function is used to evaluate the quality of the data partitioning scheme represented by each particle.

[0053] Step 1323: Calculate the new position of the particle using the particle swarm optimization update formula according to the current position and velocity of the particle, and convert the updated particle position into a new data set partitioning scheme.

[0054] Step 1324, use the fitness function to evaluate the quality of the newly generated partitioning scheme for each particle. If the fitness of the new scheme is higher than the historical best fitness of the individual particle, the individual optimal position of the particle is updated. If the fitness of the new scheme is higher than the historical best fitness of the entire particle swarm, the global optimal position is updated. According to the individual optimal position and the global optimal position, the speed update formula of the particle swarm optimization is used to adjust the speed of each particle. Repeat steps 1323 and 1324 until the preset number of iterations is reached.

[0055] Step 1325, after the iteration is completed, the particle corresponding to the global optimal position is selected as the final partitioning scheme, and the simulated data set is divided into a training set and a test set according to the scheme.

[0056] In another preferred embodiment of the present invention, the calculation formula of the fitness function is: ; in, It is The weighted accuracy of each particle; represents the number of different categories in the simulated dataset; It is The weight of each category; For the particles, the number of true positives, indicating that the correct prediction is The number of samples in the class; For the particles, the number of false positives, indicating that the The sample prediction of class The number of classes, where = ; Is the first Classification Cost of class; Indicates that for particles, the number of false negatives, indicating that the The number of sample predictions for the class; Represents the index of the particle; Indicates the index of the category.

[0057] In the embodiment of the present invention, by weight , which can handle unbalanced datasets and give higher importance to minority categories, thereby increasing the model's attention to these categories. , which takes into account not only the accuracy of the prediction, but also the cost of different types of wrong predictions. This is very useful in many practical applications. ), false positive examples ( ) and false negatives ( ), which can more comprehensively evaluate the performance of the model. This fitness function is particularly suitable for the particle swarm optimization (PSO) algorithm, in which each particle represents a potential solution. By calculating the weighted accuracy of each particle, the PSO algorithm can effectively search for the optimal solution in the solution space. By considering various types of prediction errors, this function helps promote the generalization ability of the model, so that the model can perform well not only on the training data.

[0058] For particles, whose speed exist The update formula in the dimensional space is: ; in, and They are Particles in In dimensional space, time and speed. is the inertia weight, which is used to balance the capabilities of global search and local search. and is the learning factor, represents the individual learning factor, Represents the social learning factor. and is a random number between [0, 1], used to increase the randomness of the search. It is Particles in The optimal position of an individual in dimensional space. All particles in The global optimal position in dimensional space. It is Particles in In dimensional space, time location.

[0059] The position update formula is: For particles, whose positions exist The update in the dimensional space is calculated based on the velocity, specifically: ; in, and They are Particles in In dimensional space, time and location.

[0060] In another preferred embodiment of the present invention, the above step 135, according to the simulated data set and the neural network structure, outputs data through forward propagation calculation, including: Step 1351, passing the input data in the simulated data set to the input layer of the neural network, specifically includes: reading the simulated data set, which is a structured data matrix, in which each row represents a sample and each column represents a feature; initializing the input layer of the neural network, ensuring that the number of neurons in the input layer matches the number of features in the simulated data set; passing each sample in the simulated data set (i.e., each row in the data matrix) as input data to the corresponding neuron in the input layer; Step 1352, for each neuron in the input layer, the input data is passed to the first hidden layer. In this process, the weights between the input data and the neurons in the first hidden layer are weighted and summed, and a bias term is added. Specifically, for each neuron in the input layer, the input data received by the neuron is weighted and summed with the weights connected to the neurons in the first hidden layer; a bias term (bias) is added to the result of the weighted summation, which is a learnable parameter used to adjust the output of the neuron, and the weighted summation and bias term addition operation from all neurons in the input layer to the neurons in the first hidden layer is completed; Step 1353, for each hidden layer, receiving the output data of the neurons in the previous layer as input signals, specifically including: for each hidden layer in the neural network (starting from the first hidden layer), its input signal is the output data of the neurons in the previous layer (i.e., the previous layer); each neuron in the current hidden layer receives the output data from all neurons in the previous layer (or neurons in specific connections, depending on the network structure); Step 1354, performing weighted summation on the input signal and adding a bias term to obtain a processing result, and processing the processing result through an activation function to generate output data of the current hidden layer, specifically including: performing a weighted summation operation on the received input signal (i.e., the output data of the neurons in the previous layer) in the current hidden layer, using the connection weights between the neurons in the current layer and the neurons in the previous layer; adding the bias term of the neurons in the current layer to the result of the weighted summation; passing the result of the weighted summation and the addition of the bias term to an activation function (such as ReLU, Sigmoid, etc.), and the activation function performs a nonlinear transformation on the result to generate output data of the neurons in the current hidden layer; Step 1355, the output data of the current hidden layer is passed to the next layer until the last hidden layer is reached, specifically including: after the current hidden layer processes the input signal and generates output data, the output data will be passed to the next layer (i.e., a deeper hidden layer or output layer). This transmission process is a continuation of forward propagation, and the data flows forward layer by layer in the neural network; Step 1356, the output data of the last hidden layer is passed to the output layer, specifically including: after the data is passed to the last hidden layer and processed, the output data of the layer will be passed to the output layer of the neural network, and the output layer is responsible for receiving the output of the last hidden layer and preparing for the final prediction or decision; Step 1357, in the output layer, the output data is again subjected to weighted summation and adjustment of the bias term, and the final prediction result of the neural network is calculated through the activation function. The final prediction result is the prediction result of the simulated data set, which specifically includes: in the output layer, a weighted summation operation is performed on the output data of the last hidden layer received, and the bias term of the output layer neuron is added to the weighted summation result using the connection weights between the output layer neurons and the last hidden layer neurons; the result of the weighted summation and the addition of the bias term is passed to the activation function of the output layer (in some cases, such as regression tasks, the activation function may not be used or a linear activation function may be used); the result after the activation function processing is the final prediction result of the neural network, which is output and compared with the true label in the simulated data set to evaluate the performance of the network.

[0061] In an embodiment of the present invention, step 1351, which is the starting point for the neural network to process data, ensures that the original data can be received by the network and further processed. Step 1352, this step realizes the preliminary conversion and feature extraction of the data. The introduction of weighted summation and bias terms enables the network to learn the complex patterns of the data. Step 1353, the design of the hidden layer enables the network to learn the deep-level features of the data, and each layer abstracts and extracts higher-level information. Step 1354, the introduction of the activation function brings nonlinear characteristics to the network, enabling the network to learn and represent complex nonlinear relationships. Step 1355, through multi-layer transmission, the network can gradually extract and process the features in the data to achieve the recognition and learning of complex patterns. Step 1356, this step marks the end of the feature extraction stage, and the output layer is ready to receive the final feature representation for decision or prediction. Step 1357, the processing of the output layer enables the network to make a final prediction based on the learned feature representation. The activation function may be used here for tasks such as classification and regression to achieve the functional goals of the neural network.

[0062] When it comes to specific applications, a manufacturing company decided to use a neural network model to predict and control the vacuum glue filling process in order to improve the automation level and product quality of the vacuum glue filling process. First, a series of simulation data was collected to train and test the neural network model.

[0063] The simulation data set contains 1000 samples, each of which has 5 characteristic parameters: glue temperature (℃), vacuum pressure (kPa), glue flow rate ( ml / s), glue pouring time (s) and glue viscosity (cP). These parameters are organized into a 1000×5 data matrix, where each row represents a sample and each column represents a feature.

[0064] For prediction, a neural network with two hidden layers, each with 8 neurons, was designed. The input layer had 5 neurons (matching the number of features) and the output layer had 1 neuron, which was used to predict the quality index of the glue filling (a value between 0 and 1, with the closer to 1 indicating higher quality).

[0065] Specific steps of forward propagation: Step 1351, take the first sample from the simulated data set (i.e., the first row of the data matrix), which is a vector containing 5 eigenvalues: [glue temperature, vacuum pressure, glue flow rate, glue filling time, glue viscosity]. This vector is passed to the input layer of the neural network, and each eigenvalue corresponds to an input neuron.

[0066] In step 1352, each neuron in the input layer performs a weighted summation of the received feature value and the weight connected to the first hidden layer, and adds a bias term. For example, the first neuron in the first hidden layer will calculate: w1×glue temperature+w2×vacuum pressure+w3×glue flow rate+w4×glue filling time+w5×glue viscosity+b, where w1-w5 are weights and b is a bias term. All neurons in the first hidden layer will perform similar calculations.

[0067] From step 1353 to step 1354, the first hidden layer passes the processed data to the second hidden layer. Each neuron in the second hidden layer is again weighted summed, bias adjusted, and activated. Activation functions (such as ReLU) are used to introduce nonlinearity, allowing the neural network to learn complex patterns.

[0068] From step 1355 to step 1356, after the second hidden layer processes the data, it passes the data to the output layer. In this example, the output layer has only one neuron, which is responsible for predicting the quality index of the glue filling.

[0069] Step 1357, the neurons in the output layer perform the final weighted summation, bias adjustment and activation function processing on the received data. The processed value is the prediction result of the neural network for the glue filling quality index of the first sample. Repeat the above steps until all samples in the simulated data set have been processed. Each sample will get a predicted value of the glue filling quality index.

[0070] In another preferred embodiment of the present invention, step 14 is to obtain real-time parameters during the glue pouring process. The specific implementation process of this step is as follows: Sensors are installed on the glue dispensing equipment to monitor the glue dispensing speed, glue temperature, vacuum degree and glue dispensing time in real time. Real-time data of the glue dispensing process is continuously collected through sensors. These data can be transmitted to the data processing system via data cables or wirelessly. The collected real-time data is preprocessed, including filtering, denoising, unit conversion, etc., to ensure the quality and consistency of the data. The preprocessed real-time data is stored in the database for subsequent analysis and use. The real-time parameters are displayed through the user interface (such as dashboard, curve chart, etc.) so that the operator can monitor the status of the glue dispensing process in real time.

[0071] Step 15 is to obtain the predicted glue filling result based on the real-time parameters and the prediction model. The following is the specific implementation process of this step: The real-time parameters (glue filling speed, glue temperature, vacuum degree and glue filling time) obtained in step 14 are used as input data and prepared to be input into the prediction model. Make sure that the format of the input data is consistent with the data format used when training the prediction model. This includes data normalization, encoding and other processing. Input the formatted real-time parameters into the prediction model, perform forward propagation calculations, and obtain the predicted glue filling results. The prediction results may include key indicators such as glue filling amount and glue filling position. Output the prediction results to the user interface or store them in the database for subsequent analysis and use.

[0072] Step 16 is to use the 3D geometric model to simulate and analyze the real-time parameters. The following is the specific implementation process of this step: According to the actual structure and size of the glue filling equipment, establish an accurate three-dimensional geometric model. This can be done using professional modeling software (such as SolidWorks, AutoCAD, etc.). Set simulation parameters corresponding to real-time parameters in the three-dimensional geometric model, including glue filling speed, glue temperature, vacuum degree, and glue filling time. Use fluid dynamics simulation software (such as ANSYS Fluent, COMSOL Multiphysics, etc.) to simulate and analyze the three-dimensional geometric model with simulation parameters set. During the analysis, pay attention to key indicators such as the flow behavior of the glue and temperature distribution. After the simulation analysis is completed, extract and save the analysis results. These results include key data such as glue flow velocity field, temperature field, and pressure field.

[0073] Step 17 is to evaluate key indicators based on the analysis results. The following is the specific implementation process of this step: Visualize the analysis results obtained in step 16, such as displaying key indicators such as glue flow behavior and temperature distribution through cloud maps, vector diagrams, curve diagrams, etc. Evaluate key indicators such as glue flow behavior and temperature distribution based on the visualized analysis results. During the evaluation process, you can refer to preset standards or historical data to determine whether the current glue filling conditions meet the requirements. If problems are found in key indicators during the evaluation process (such as poor glue flow, uneven temperature distribution, etc.), the specific cause and location of the problem need to be identified. Record the evaluation results and identified problems for subsequent parameter adjustment and optimization.

[0074] Step 18 is to dynamically adjust the control parameters based on the predicted glue filling results and key indicators. The following is the specific implementation process of this step: According to the predicted glue pouring results obtained in step 15 and the key indicators evaluated in step 17, determine the control parameters that need to be adjusted and their adjustment directions (such as increasing the glue pouring speed, reducing the glue temperature, etc.). Through the control system of the glue pouring equipment, make real-time adjustments to the control parameters that need to be adjusted. During the adjustment process, attention should be paid to the response speed and stability of the equipment to ensure that the adjustment effect meets expectations. After the adjustment is completed, perform the glue pouring operation again, and monitor the key indicators of the glue pouring process and the final glue pouring results in real time. Verify the effect of the parameter adjustment by comparing the data before and after the adjustment. If the verification result shows that the parameter adjustment is effective, you can continue to iterate and optimize to further improve the efficiency and stability of the glue pouring process. If the verification effect is not good, you need to re-analyze the cause and adjust the strategy.

[0075] like Figure 2 As shown, an embodiment of the present invention further provides a vacuum glue filling control system 20 based on machine learning, comprising: The partitioning module 21 is used to construct a three-dimensional geometric model of the glue filling equipment according to the physical characteristics of the vacuum glue filling process, and to mesh the three-dimensional geometric model; define the geometric shape, material properties, boundary conditions and initial conditions of the finite element model to simulate the real glue filling environment, and the physical characteristics include the rheology and thermal conductivity of the glue; The simulation module 22 is used to simulate the glue filling process under different glue filling speeds, glue temperatures, and vacuum degrees through a three-dimensional geometric model of the glue filling equipment to obtain simulation results, including glue flow behavior, temperature distribution, and stress distribution; A construction module 23 is used to train the simulation results to construct a prediction model for predicting the glue filling results; The acquisition module 24 is used to acquire real-time parameters during the glue filling process, including real-time glue filling speed, glue temperature, vacuum degree and glue filling time; Prediction module 25, used to obtain predicted glue filling results according to real-time glue filling speed, glue temperature, vacuum degree and glue filling time and prediction model; The analysis module 26 is used to simulate and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time by using the three-dimensional geometric model of the glue filling equipment to obtain the analysis result; Evaluation module 27, used to evaluate the key indicators of glue flow behavior and temperature distribution under the current glue filling conditions according to the analysis results; The adjustment module 28 is used to dynamically adjust the control parameters in the glue pouring process according to the predicted glue pouring results and key indicators.

[0076] It should be noted that the system is a system corresponding to the above method, and all implementation methods in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0077] The embodiment of the present invention further provides a computing device, comprising: a processor, a memory storing a computer program, wherein when the computer program is executed by the processor, the method described above is executed. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0078] The embodiment of the present invention also provides a computer-readable storage medium storing instructions, which, when executed on a computer, enable the computer to execute the method described above. All implementations in the above method embodiment are applicable to this embodiment and can achieve the same technical effect.

[0079] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A vacuum glue filling control method based on machine learning, characterized in that: The method comprises: According to the physical characteristics of the vacuum glue filling process, a 3D geometric model of the glue filling equipment is constructed and meshed. The geometric shape, material properties, boundary conditions and initial conditions of the finite element model are defined to simulate the real glue filling environment. The physical characteristics include the rheology and thermal conductivity of the glue. The three-dimensional geometric model of the glue dispensing equipment is used to simulate the glue dispensing process under different glue dispensing speeds, glue temperatures, and vacuum degrees to obtain simulation results, including glue flow behavior, temperature distribution, and stress distribution; The simulation results are trained to construct a prediction model for predicting the glue filling results; Obtain real-time parameters during the glue filling process, including real-time glue filling speed, glue temperature, vacuum degree and glue filling time; According to the real-time glue filling speed, glue temperature, vacuum degree and glue filling time and prediction model, the predicted glue filling result can be obtained; Use the three-dimensional geometric model of the glue filling equipment to simulate and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time to obtain the analysis results; Evaluate the key indicators of glue flow behavior and temperature distribution under current glue filling conditions based on the analysis results; Dynamically adjust the control parameters in the glue pouring process based on the predicted glue pouring results and key indicators.

2. The vacuum glue filling control method based on machine learning according to claim 1 is characterized in that: The finite element model is constructed based on the physical characteristics of the vacuum potting process, including: Determine the physical phenomena to be simulated, the geometric dimensions, and the simulation time. The physical phenomena include glue flow behavior, temperature distribution, and stress distribution. Determine the physical parameters and material properties to be simulated. Material properties include density, viscosity, thermal conductivity and specific heat capacity of the glue; physical parameters include vacuum degree, ambient temperature, pressure and the geometry of the glue filling equipment; According to physical phenomena, geometric dimensions, simulation time, physical parameters and material properties, a three-dimensional geometric model of the glue potting equipment is constructed, wherein the three-dimensional geometric model of the glue potting equipment is a finite element model.

3. The vacuum glue filling control method based on machine learning according to claim 2 is characterized in that: The simulation results are trained to build a prediction model for predicting the glue filling results, including: Preprocessing the simulation result data to obtain a preprocessed simulation data set; Divide the simulated data set into training set and test set; According to the characteristics of the simulated data set, determine the neural network structure, activation function and loss function. The neural network structure includes input layer, hidden layer and output layer; Initialize the weights and biases of the neural network structure, set the learning rate, batch size, and number of training rounds during training; According to the simulated data set and the neural network structure, the output data is calculated through forward propagation; Calculate the loss value of the neural network structure according to the loss function, and update the weights and biases of the neural network structure through the back propagation algorithm, and repeat the steps until the preset number of training rounds is reached to obtain the neural network model; Use the test set to evaluate the neural network model and evaluate the performance indicators of the neural network model on the test set to obtain the evaluation results; The neural network model is optimized according to the evaluation results to obtain a trained neural network model, wherein the trained neural network model is a prediction model.

4. The vacuum glue filling control method based on machine learning according to claim 3 is characterized in that: Divide the simulated data set into training and testing sets, including: Set the size of the particle swarm, where each particle represents a data set partitioning scheme, including the ratio or specific index of the training set and the test set; initialize the position and velocity of each particle; Determine a fitness function for evaluating the quality of the data partitioning scheme represented by each particle; For each particle, update its position according to its current position and velocity to generate a new data partitioning scheme; Update the individual position and global position of each particle, and adjust the speed of each particle according to the individual position and the global position until the preset number of iterations is reached to obtain the final division scheme; According to the final partitioning scheme, the simulated data set is divided into a training set and a test set.

5. The vacuum glue filling control method based on machine learning according to claim 4 is characterized in that: According to the simulated data set and the neural network structure, the output data is calculated through forward propagation, including: Pass the input data from the simulated dataset to the input layer of the neural network; For each neuron in the input layer, the input data is passed to the first hidden layer. During this process, the weights between the input data and the neurons in the first hidden layer are weighted and summed, and a bias term is added. For each hidden layer, the output data of the neurons in the previous layer is received as input signal; The input signal is weighted and summed and a bias term is added to obtain a processing result, and the processing result is processed by an activation function to generate output data of the current hidden layer; The output data of the current hidden layer is passed to the next layer until the last hidden layer is reached; The output data of the last hidden layer is passed to the output layer; In the output layer, the output data undergoes weighted summation and bias adjustment again, and the final prediction result of the neural network is calculated through the activation function. The final prediction result is the prediction result of the simulated data set.

6. The vacuum glue filling control method based on machine learning according to claim 5 is characterized in that: Control parameters include glue filling speed, glue temperature and vacuum degree.

7. The vacuum glue filling control method based on machine learning according to claim 6 is characterized in that: The calculation formula of the fitness function is: ； in, It is The weighted accuracy of each particle; represents the number of different categories in the simulated dataset; It is The weight of each category; For the particles, the number of true positives, indicating that the correct prediction is The number of samples in the class; For the particles, the number of false positives, indicating that the The sample prediction of class The number of classes, where = ; Is the first Classification Cost of class; Indicates that for particles, the number of false negatives, indicating that the The number of sample predictions for the class; Represents the index of the particle; Indicates the index of the category.

8. A vacuum glue filling control system based on machine learning, characterized in that: Applied to the method according to any one of claims 1 to 7, comprising: The partitioning module is used to build a 3D geometric model of the glue filling equipment according to the physical characteristics of the vacuum glue filling process and to mesh the 3D geometric model; define the geometry, material properties, boundary conditions and initial conditions of the finite element model to simulate the real glue filling environment. The physical characteristics include the rheology and thermal conductivity of the glue; The simulation module is used to simulate the glue filling process under different glue filling speeds, glue temperatures, and vacuum degrees through the three-dimensional geometric model of the glue filling equipment to obtain simulation results, including glue flow behavior, temperature distribution, and stress distribution; A construction module is used to train the simulation results to build a prediction model for predicting the glue filling results; The acquisition module is used to obtain the real-time parameters in the glue filling process, including the real-time glue filling speed, glue temperature, vacuum degree and glue filling time; The prediction module is used to obtain the predicted glue filling result according to the real-time glue filling speed, glue temperature, vacuum degree and glue filling time and the prediction model; The analysis module is used to simulate and analyze the real-time glue filling speed, glue temperature, vacuum degree and glue filling time by using the three-dimensional geometric model of the glue filling equipment to obtain the analysis results; Evaluation module, used to evaluate the key indicators of glue flow behavior and temperature distribution under the current glue filling conditions based on the analysis results; The adjustment module is used to dynamically adjust the control parameters in the glue filling process according to the predicted glue filling results and key indicators.

9. A computing device, characterized in that include: one or more processors; A storage device for storing one or more programs, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as claimed in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images