Preparation method and system of biconvex lens, terminal equipment and storage medium
By optimizing the process parameters of biconvex lenses using molding optimization algorithms and predictive models, the problem of process parameter deviations caused by manual calculations was solved, thereby improving the molding quality and optical performance of biconvex lenses.
Patent Information
- Application Number
- CN202510945999.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-09
- Publication Date
- 2025-12-02
AI Technical Summary
In the prior art, the process parameters of the biconvex lens forming mold are calculated manually, which leads to deviations between the process parameters and the actual parameters, affecting the forming quality and optical performance of the biconvex lens.
A molding optimization algorithm is used to optimize the set of process parameters. The target process parameters are obtained through the molding prediction model and evaluation function, and the working parameters of the biconvex lens molding die are adjusted.
It reduces the subjective interference of human calculation, improves the forming quality and optical performance of biconvex lenses, and enhances forming accuracy and lens optical performance.
Smart Images
Figure CN121044795A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of optical manufacturing technology. More specifically, this application relates to a method, system, terminal device, and storage medium for fabricating a biconvex lens. Background Technology
[0002] Biconvex lenses are formed by hot-pressing a glass preform using a biconvex lens forming mold. However, in existing technologies, the process parameters of the biconvex lens forming mold are usually calculated manually. Manual calculation is subjective, leading to deviations between the process parameters and actual conditions. This results in low-quality biconvex lenses formed by hot pressing using the mold, thus affecting their optical performance. Summary of the Invention
[0003] The purpose of this application is to provide a method, system, terminal device, and storage medium for manufacturing a biconvex lens, which can improve the forming quality of the biconvex lens. This application is mainly achieved through the following technical solutions:
[0004] A first aspect of this application provides a method for fabricating a biconvex lens, comprising:
[0005] Obtain a set of process parameters, wherein each process parameter in the set includes molding temperature, molding speed, and cooling rate;
[0006] The set of process parameters is optimized using a molding optimization algorithm to obtain the target process parameters;
[0007] The working parameters of the biconvex lens forming mold are adjusted based on the target process parameters so that the biconvex lens forming mold can produce a biconvex lens according to the working parameters.
[0008] According to one embodiment of this application, the step of optimizing the set of process parameters using a molding optimization algorithm to obtain the target process parameters includes:
[0009] The molding prediction model using the molding optimization algorithm is used to predict each process parameter in the set of process parameters to obtain the target molding data corresponding to each process parameter.
[0010] An evaluation function is used to evaluate the target molding data corresponding to each process parameter, and the evaluation result of each process parameter is obtained.
[0011] All evaluation results are filtered and processed according to preset rules to obtain the target result;
[0012] The process parameters corresponding to the target result are taken as the target process parameters.
[0013] According to one embodiment of this application, the step of using the molding prediction model of the molding optimization algorithm to predict each process parameter in the set of process parameters and obtaining the target molding data corresponding to each process parameter includes:
[0014] The first hidden layer of the molding prediction model is used to perform feature extraction processing on each process parameter in the process parameter set to obtain the first feature vector corresponding to each process parameter;
[0015] The first objective activation function of the compression molding prediction model is used to perform nonlinear transformation on the first feature vector corresponding to each process parameter to obtain the second feature vector corresponding to each process parameter.
[0016] The second hidden layer of the compression molding prediction model is used to perform feature extraction processing on the second feature vector corresponding to each process parameter to obtain the third feature vector corresponding to each process parameter.
[0017] The second objective activation function of the compression molding prediction model is used to perform nonlinear transformation on the third feature vector corresponding to each process parameter to obtain the fourth feature vector corresponding to each process parameter.
[0018] The third hidden layer of the molding prediction model is used to extract features from the fourth feature vector corresponding to each process parameter to obtain the fifth feature vector corresponding to each process parameter.
[0019] The fifth feature vector corresponding to each process parameter is nonlinearly transformed using the third objective activation function of the compression molding prediction model to obtain the target molding data corresponding to each process parameter.
[0020] According to one embodiment of this application, the method for preparing the biconvex lens further includes a training step for the molding prediction model, wherein the training step for the molding prediction model includes:
[0021] Obtain a training dataset and a real label set. Each training data in the training dataset includes molding training temperature, molding training speed, and cooling training speed. Each training data has a one-to-one correspondence with one of the real labels in the real label set.
[0022] The original molding prediction model is used to perform prediction processing on each training data in the training dataset to obtain the prediction result corresponding to each training data.
[0023] The loss function is calculated using the prediction result corresponding to each training data point and the true label corresponding to each training data point.
[0024] The first model parameters of the original molding prediction model are adjusted based on the loss function to form the molding prediction model.
[0025] According to one embodiment of this application, after the step of adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model, the training step of the molding prediction model further includes:
[0026] The compression molding prediction model is evaluated using a model evaluation method to obtain the evaluation results;
[0027] The second model parameters of the compression molding prediction model are adjusted based on the evaluation results.
[0028] According to one embodiment of this application, after the step of adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model, the training step of the molding prediction model further includes:
[0029] The third model parameters of the compression molding prediction model are optimized using a first machine learning algorithm.
[0030] According to one embodiment of this application, the training step of the molding prediction model further includes a step of obtaining the real label set, wherein the step of obtaining the real label set includes:
[0031] Obtain a set of molding data, wherein each set of molding data includes a first residual stress, a first angular deviation, and a first shape deviation;
[0032] The established dataset is optimized using an orthogonal experimental method to obtain the true label set.
[0033] A second aspect of this application provides a system for fabricating a biconvex lens, comprising: a terminal device and a biconvex lens forming mold. The terminal device includes a processor and a memory. The memory is used to store a computer program. The processor is used to call and run the computer program stored in the memory to perform the steps of the biconvex lens fabrication method according to any one of claims 1 to 7. The processor is also used to adjust the operating parameters of the biconvex lens forming mold.
[0034] A third aspect of this application provides a terminal device, including a processor and a memory, the memory being used to store a computer program, and the processor being used to call and run the computer program stored in the memory to execute the steps of the biconvex lens fabrication method provided in the first aspect of this application.
[0035] A fourth aspect of this application provides a computer-readable storage medium for storing a computer program that causes a computer to perform the steps of the method for preparing a biconvex lens provided in the first aspect of this application.
[0036] The beneficial effects of the embodiments of this application include:
[0037] This application employs a molding optimization algorithm to replace manual calculation. Specifically, this application obtains a set of process parameters, each of which includes molding temperature, molding speed, and cooling rate. The molding optimization algorithm is then used to optimize this set of process parameters to obtain target process parameters. Based on these target process parameters, the operating parameters of the biconvex lens molding die are adjusted so that the die produces the biconvex lens according to these operating parameters. Compared to existing technologies that use manual calculation, this application uses a molding optimization algorithm to obtain the final process parameters, effectively reducing subjective interference caused by manual calculation, minimizing deviations between process parameters and actual conditions, thereby improving the molding quality of the biconvex lens and ultimately enhancing its optical performance. Attached Figure Description
[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the conventional technology, the drawings used in the description of the embodiments or the conventional technology will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0039] Figure 1 The flowcharts for some embodiments of the method for fabricating the biconvex lens of this application are shown below;
[0040] Figure 2 This is a schematic diagram of the molding prediction model of this application;
[0041] Figure 3 This is a flowchart illustrating the training process of the compression molding prediction model in this application.
[0042] Figure 4 This is a block diagram illustrating the principle of the biconvex lens fabrication system of this application in some embodiments;
[0043] Figure 5 This is a schematic diagram of the structure of the biconvex lens forming mold of this application in some embodiments;
[0044] Figure 6 This is a cross-sectional view of the biconvex lens forming mold of this application in some embodiments;
[0045] Figure 7 This is a schematic diagram of the structure of a biconvex lens forming mold in a heated and heat-preserving state.
[0046] Figure 8 A schematic diagram of a biconvex lens forming mold in a hot-pressing and pressure-holding state;
[0047] Figure 9 This is a schematic diagram of the structure of a biconvex lens forming mold in the cooling and demolding state.
[0048] Figure 10 This is a schematic block diagram of the terminal device of this application in some embodiments. Detailed Implementation
[0049] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application are described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.
[0050] It should be noted that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0051] The terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.
[0052] The terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, system, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units that are expressly listed, but may include other steps or units that are not expressly listed or that are inherent to such process, method, product, or apparatus.
[0053] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used in this application includes any and all combinations of one or more of the associated listed items.
[0054] The specific embodiments of this application will be further described below with reference to the accompanying drawings.
[0055] refer to Figure 1 The diagram shown is a flowchart of a method for fabricating a biconvex lens according to a first aspect of an embodiment of this application. Figure 1 The method for preparing the biconvex lens includes:
[0056] S1. Obtain a set of process parameters, wherein each process parameter in the set includes molding temperature, molding speed and cooling speed.
[0057] In other embodiments, each process parameter may also include molding pressure or other data, and the specific content of each process parameter can be set by those skilled in the art according to actual needs.
[0058] S2. The set of process parameters is optimized using a molding optimization algorithm to obtain the target process parameters.
[0059] Further, step S2 includes: using the molding prediction model of the molding optimization algorithm to predict each process parameter in the set of process parameters, and obtaining the target molding data corresponding to each process parameter; using an evaluation function to evaluate the target molding data corresponding to each process parameter, and obtaining the evaluation result of each process parameter; filtering all the evaluation results according to preset rules to obtain the target result; and using the process parameter corresponding to the target result as the target process parameter.
[0060] The compression molding prediction model includes multiple hidden layers and multiple target activation functions. The multiple hidden layers can be referenced... Figure 2 As shown, in Figure 2 In this context, the multi-layered hidden layer is m layers, and those skilled in the art can set the specific value of m according to actual needs; Figure 2 In this context, S2 represents the number of neurons in the first hidden layer, S3 represents the number of neurons in the second hidden layer, and S4 represents the number of neurons in the m-th hidden layer. The specific values of S2, S3, and S4 can be set by those skilled in the art according to actual needs.
[0061] In this embodiment, the multiple hidden layers are a first hidden layer, a second hidden layer, and a third hidden layer. The plurality of target activation functions include a first target activation function, a second target activation function, and a third target activation function. In other embodiments, the number of the multiple hidden layers and the number of the plurality of target activation functions can be set by those skilled in the art according to actual needs.
[0062] Further, the step of using the molding prediction model of the molding optimization algorithm to predict each process parameter in the process parameter set and obtain the target molding data corresponding to each process parameter includes: using the first hidden layer of the molding prediction model to perform feature extraction processing on each process parameter in the process parameter set to obtain a first feature vector corresponding to each process parameter; using the first target activation function of the molding prediction model to perform nonlinear transformation processing on the first feature vector corresponding to each process parameter to obtain a second feature vector corresponding to each process parameter; using the second hidden layer of the molding prediction model to perform feature extraction processing on the second feature vector corresponding to each process parameter to obtain a third feature vector corresponding to each process parameter; using the second target activation function of the molding prediction model to perform nonlinear transformation processing on the third feature vector corresponding to each process parameter to obtain a fourth feature vector corresponding to each process parameter; using the third hidden layer of the molding prediction model to perform feature extraction processing on the fourth feature vector corresponding to each process parameter to obtain a fifth feature vector corresponding to each process parameter; and using the third target activation function of the molding prediction model to perform nonlinear transformation processing on the fifth feature vector corresponding to each process parameter to obtain the target molding data corresponding to each process parameter.
[0063] The number of neurons in the first hidden layer is set to 64. In other embodiments, the number of neurons in the first hidden layer can be other values, which can be set by those skilled in the art according to actual needs.
[0064] For example, the first hidden layer of the molding prediction model is used to perform feature extraction processing on each process parameter in the process parameter set. The calculation formula for obtaining the first feature vector corresponding to each process parameter can be: Z1 = W1X + b1; Z1 is the first feature vector corresponding to each process parameter; W1 is the first weight matrix; X is each process parameter in the process parameter set; b1 is the first bias.
[0065] In other embodiments, the first hidden layer of the molding prediction model is used to perform feature extraction processing on each process parameter in the set of process parameters. The calculation formula for obtaining the first feature vector corresponding to each process parameter can also be other formulas, which can be set by those skilled in the art according to actual needs.
[0066] The first target activation function is the ReLU (Rectified Linear Unit) activation function. In other embodiments, the first target activation function may also be a Sigmoid (i.e., sigmoid function), a Tanh (i.e., hyperbolic tangent function), or a Leaky ReLU (Leaky Rectified Linear Unit) function, which can be specifically set by those skilled in the art according to actual needs.
[0067] The number of neurons in the second hidden layer is set to 32. In other embodiments, the number of neurons in the second hidden layer can be other values, which can be set by those skilled in the art according to actual needs.
[0068] For example, the second hidden layer of the compression molding prediction model is used to perform feature extraction processing on the second feature vector corresponding to each process parameter. The calculation formula for obtaining the third feature vector corresponding to each process parameter can be: Z2 = W2H1 + b2; Z2 is the third feature vector corresponding to each process parameter; W2 is the second weight matrix; H1 is the second feature vector corresponding to each process parameter; b2 is the second bias.
[0069] In other embodiments, the second hidden layer of the molding prediction model is used to perform feature extraction processing on the second feature vector corresponding to each process parameter. The calculation formula for obtaining the third feature vector corresponding to each process parameter can also be other formulas, which can be set by those skilled in the art according to actual needs.
[0070] The second target activation function is the ReLU activation function. In other embodiments, the second target activation function may also be a Sigmoid function, a Tanh function, or a Leaky ReLU function, which can be set by those skilled in the art according to actual needs.
[0071] The number of neurons in the third hidden layer is set to 16. In other embodiments, the number of neurons in the third hidden layer can be other values, which can be set by those skilled in the art according to actual needs.
[0072] For example, the third hidden layer of the molding prediction model is used to perform feature extraction processing on the fourth feature vector corresponding to each process parameter, and the calculation formula for obtaining the fifth feature vector corresponding to each process parameter can be: Z3 = W3H2 + b3; Z3 is the fifth feature vector corresponding to each process parameter; W3 is the third weight matrix; H2 is the fourth feature vector corresponding to each process parameter; b3 is the third bias.
[0073] In other embodiments, the third hidden layer of the molding prediction model is used to perform feature extraction processing on the fourth feature vector corresponding to each process parameter, and the calculation formula for obtaining the fifth feature vector corresponding to each process parameter can also be other formulas, which can be set by those skilled in the art according to actual needs.
[0074] The third objective activation function is a ReLU activation function. In other embodiments, the third objective activation function may also be a Sigmoid function, a Tanh function, or a Leaky ReLU function, which can be set by those skilled in the art according to actual needs.
[0075] In other embodiments, refer to Figure 2 As shown, the molding prediction model further includes an input layer and an output layer, with the input layer located before the first hidden layer and the output layer located after the third hidden layer. Figure 2 In this design, the number of neurons in the input layer is S1, and those skilled in the art can set the specific value of S1 according to actual needs. The number of neurons in the output layer is S5, and those skilled in the art can set the specific value of S5 according to actual needs.
[0076] During the forward propagation of the molding prediction model, the data from the input layer is processed by weighted summaries and activation functions of each layer of neurons before being progressively transmitted to the output layer. It should also be understood that the output layer typically does not use activation functions and directly outputs continuous predicted values. In this embodiment, the output layer directly outputs the target molding data, which includes indicators such as second residual stress, second angular deviation, and second shape deviation. In other embodiments, the target molding data also includes thickness deviation.
[0077] Furthermore, the evaluation function is used to evaluate the target molding data corresponding to each process parameter. The calculation formula for obtaining the evaluation result of each process parameter is: f(x)=A1×RS+A2×AD+A3×SD; A1 is the first weighting coefficient; RS is the second residual stress of the target molding data corresponding to each process parameter; A2 is the second weighting coefficient; AD is the second angular deviation of the target molding data corresponding to each process parameter; A3 is the third weighting coefficient; SD is the second shape deviation of the target molding data corresponding to each process parameter.
[0078] It should be understood that the smaller the second angle deviation and thickness deviation, the higher the forming accuracy of the biconvex lens; the smaller the second residual stress, the better the optical performance and thermal stability of the lens, and the higher the quality.
[0079] Furthermore, the first weighting coefficient, the second weighting coefficient, and the third weighting coefficient satisfy a preset condition. The formula for calculating the preset condition is: A1 + A2 + A3 = 1.
[0080] Furthermore, the steps to filter all evaluation results according to preset rules and obtain the target result include:
[0081] All evaluation results are sorted in descending or ascending order, and the minimum value among all sorted results is selected as the target result. The process parameters corresponding to the target result are the optimal optimization scheme.
[0082] S3. Adjust the working parameters of the biconvex lens forming mold based on the target process parameters, so that the biconvex lens forming mold can produce a biconvex lens according to the working parameters.
[0083] In the above embodiments, this application uses a molding optimization algorithm to replace manual calculation. Compared with the prior art that uses manual calculation, this application uses a molding optimization algorithm to obtain the final process parameters, which can effectively reduce the subjective interference caused by manual calculation, reduce the deviation between process parameters and reality, thereby improving the molding quality of the biconvex lens and thus enhancing the optical performance of the biconvex lens.
[0084] The embodiments of this application can quickly predict the lens forming process based on process parameters and obtain the most ideal processing parameters for fabricating biconvex lenses. These embodiments also achieve lower residual stress, resulting in higher forming accuracy and optical performance of the biconvex lens. The biconvex lens fabrication method proposed in these embodiments has promising engineering application prospects.
[0085] In some embodiments, the method for preparing the biconvex lens further includes a training step for the molding prediction model. The training step includes: acquiring a training dataset and a set of real labels; each training data point in the training dataset includes molding training temperature, molding training speed, and cooling training speed; each training data point has a one-to-one correspondence with one of the real labels in the set of real labels; performing prediction processing on each training data point in the training dataset using the original molding prediction model to obtain a prediction result corresponding to each training data point; calculating a loss function using the prediction result corresponding to each training data point and the real label corresponding to each training data point; and adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model.
[0086] The data structure and data type of the molding training temperature are the same as those of the molding temperature. The data structure and data type of the molding training speed are the same as those of the molding speed. The data structure and data type of the cooling training speed are the same as those of the cooling speed.
[0087] The original molding prediction model can be a multilayer perceptron or a long short-term memory network. In other embodiments, the original molding prediction model can also be other neural network algorithms, which can be set by those skilled in the art according to actual needs.
[0088] Further, the step of using the original molding prediction model to predict each training data in the training dataset and obtaining the prediction result corresponding to each training data includes: using the fourth hidden layer of the original molding prediction model to perform feature extraction processing on each training data in the training dataset to obtain the sixth feature vector corresponding to each training data; using the fourth objective activation function of the original molding prediction model to perform nonlinear transformation processing on the sixth feature vector corresponding to each training data to obtain the seventh feature vector corresponding to each training data; using the fifth hidden layer of the original molding prediction model to perform feature extraction processing on the seventh feature vector corresponding to each training data to obtain the eighth feature vector corresponding to each training data; using the fifth objective activation function of the original molding prediction model to perform nonlinear transformation processing on the eighth feature vector corresponding to each training data to obtain the ninth feature vector corresponding to each training data; using the sixth hidden layer of the original molding prediction model to perform feature extraction processing on the ninth feature vector corresponding to each training data to obtain the tenth feature vector corresponding to each training data; and using the sixth objective activation function of the original molding prediction model to perform nonlinear transformation processing on the tenth feature vector corresponding to each training data to obtain the prediction result corresponding to each training data.
[0089] The number of neurons in the fourth hidden layer is the same as the number of neurons in the first hidden layer. The number of neurons in the fifth hidden layer is the same as the number of neurons in the second hidden layer. The number of neurons in the sixth hidden layer is the same as the number of neurons in the third hidden layer. The fourth target activation function is the same as the first target activation function. The fifth target activation function is the same as the second target activation function. The sixth target activation function is the same as the third target activation function.
[0090] The only difference between the original molding prediction model and the molding prediction model is the difference in model parameters.
[0091] Furthermore, the loss function is the mean squared error loss function. The formula for calculating the loss function is: Where n is the length of the training dataset; y i It is the true label corresponding to the i-th training data; g i It is the prediction result corresponding to the i-th training data.
[0092] In other implementations, the loss function may be the mean absolute error loss function, the cross-entropy loss function, or other loss functions, which can be set by those skilled in the art according to actual needs.
[0093] In some implementations, after adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model, the training step of the molding prediction model further includes: evaluating the molding prediction model using a model evaluation method to obtain an evaluation result; and adjusting the second model parameters of the molding prediction model based on the evaluation result.
[0094] The model evaluation method can be R2 scoring or cross-validation. In other embodiments, the model evaluation method can also be other methods, which can be set by those skilled in the art according to actual needs.
[0095] The use of the model evaluation method can ensure the accuracy of the compression molding prediction model under different process parameters.
[0096] In some implementations, after the step of adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model, the training step of the molding prediction model further includes: optimizing the third model parameters of the molding prediction model using a first machine learning algorithm.
[0097] The first machine learning algorithm is a particle swarm optimization algorithm. The global search capability of the particle swarm optimization algorithm can effectively avoid local optima problems and improve the stability and generalization ability of the model.
[0098] In other embodiments, the first machine learning algorithm may be other machine learning algorithms, which can be set by those skilled in the art according to actual needs.
[0099] In some implementations, before the step of using the original molding prediction model to predict each training data point in the training dataset and obtaining the prediction result corresponding to each training data point, the training step of the molding prediction model further includes: adjusting the initial weights of the original molding prediction model using a second machine learning algorithm. The adjustment of the initial weights of the original molding prediction model using the second machine learning algorithm can be understood as... Figure 3 The "Initialization of Original Compression Molding Prediction Model" step in the process.
[0100] The second machine learning algorithm is a genetic algorithm. In other embodiments, the second machine learning algorithm may also be other machine learning algorithms, which can be set by those skilled in the art according to actual needs.
[0101] This embodiment of the application, by setting the fitness function of the genetic algorithm, can continuously evolve and select the optimal solution, making the initial weights of the original molding prediction model more reasonable, thereby accelerating the model training process. The application of the first machine learning algorithm and the second machine learning algorithm can further improve the performance of the molding prediction model.
[0102] In some embodiments, the training steps of the molding prediction model further include: avoiding overfitting of the molding prediction model by using an early stopping method.
[0103] In some embodiments, the present application employs a backpropagation algorithm to train the molding prediction model. The step of training the molding prediction model using a backpropagation algorithm can be understood as follows: Figure 3 The step in the text is "calculate the backpropagation of the error and update the model parameters".
[0104] The main function of backpropagation is to calculate the gradient of each parameter based on the loss function and update the weights and biases using these gradients, thereby minimizing the prediction error. Traditional backpropagation uses gradient descent to update parameters, but it can also be combined with various optimization algorithms to improve training speed and accuracy. Common gradient descent methods include traditional batch gradient descent, stochastic gradient descent, and mini-batch gradient descent. Among these, stochastic gradient descent is the most widely used optimization method, which accelerates convergence and reduces computational cost by randomly sampling training data for each parameter update. To improve the model's generalization ability, commonly used regularization methods include L2 regularization, Dropout (random deactivation), and early stopping. Dropout randomly discards some neurons during each training cycle, forcing the model to rely on more features and paths, thus avoiding overfitting.
[0105] In some embodiments, the training steps of the molding prediction model further include the step of obtaining the true label set, which includes: obtaining a molding data set, wherein each molding data in the molding data set includes a first residual stress, a first angle deviation, and a first shape deviation; and optimizing the molding data set using an orthogonal experimental method to obtain the true label set.
[0106] For example, the steps of optimizing the established dataset using an orthogonal experiment method to obtain the real label set include: setting an orthogonal table based on the established dataset and obtaining the average index of each factor in the orthogonal table at different levels; extracting the maximum value from all average indices of each factor as the target index corresponding to each factor; and taking the value of the factor corresponding to each target index in the orthogonal table as one element of the real label in the real label set.
[0107] All the factors in the orthogonal array are the first residual stress, the first angular deviation, and the first shape deviation.
[0108] The average index of each factor at different levels refers to the average index corresponding to each row of values of each factor in the orthogonal table.
[0109] In other embodiments, the specific steps of optimizing the formed data set using orthogonal experimental methods to obtain the real label set can also be implemented by other methods or by using existing technologies, and can be set by those skilled in the art according to actual needs.
[0110] By using the orthogonal experimental method to obtain data and train the model, it is possible to achieve effective fitting of complex thermo-mechanical coupling relationships.
[0111] The embodiments of this application can improve the systematization level of the molding process and the lens quality, making them suitable for industrial application.
[0112] Furthermore, in this embodiment of the application, the average index of each factor is analyzed using the range analysis method to obtain the range value corresponding to each factor; the range values corresponding to all factors are sorted to obtain the order of importance of all factors.
[0113] The sorting process refers to a sorting operation from largest to smallest or smallest to largest. In other embodiments, the sorting process can also be other sorting methods, which can be set by those skilled in the art according to actual needs.
[0114] The priority order is used to identify the key process factors affecting lens forming performance as process optimization targets. For example, the factor with the highest range value is considered the most important process factor affecting lens forming performance.
[0115] Furthermore, the steps of analyzing all average indicators of each factor using range analysis to obtain the range value corresponding to each factor include: calculating the difference between the largest and smallest average indicators among all average indicators of each factor to obtain the range value corresponding to each factor.
[0116] In some implementations, before the step of using the original molding prediction model to predict each training data point in the training dataset and obtaining the prediction result corresponding to each training data point, the training step of the molding prediction model further includes: performing data normalization processing on the training dataset and the real label set. This step can be referred to... Figure 3 The "data normalization" step in the process.
[0117] In some implementations, after the step of using the original molding prediction model to predict each training data in the training dataset and obtaining the prediction result corresponding to each training data, the training step of the molding prediction model further includes: calculating the target error between the prediction result and the true label corresponding to each training data; comparing the target error with a first preset expected value; if the target error is less than the first preset expected value, determining whether the test of the original molding prediction model has reached a second preset expected value; if the target error is greater than or equal to the first preset expected value, performing backpropagation and updating the model parameters of the original molding prediction model; if the test of the original molding prediction model reaches the second preset expected value, the training step of the original molding prediction model ends; if the test of the original molding prediction model does not reach the second preset expected value, performing backpropagation and updating the model parameters of the original molding prediction model. This process can be referenced. Figure 3 The steps include "prediction processing", "whether the target error is less than the first preset expected value", "whether the model test reaches the second preset expected value" and "calculating the backpropagation of error and updating the model parameters".
[0118] The first preset expected value and the second preset expected value can be set by those skilled in the art according to actual needs.
[0119] refer to Figure 4 The diagram shown is a schematic block diagram of a biconvex lens fabrication system provided in the second aspect of an embodiment of this application. Figure 4 The biconvex lens fabrication system 100 includes a terminal device 101 and a biconvex lens molding die 102. The terminal device 101 includes a processor and a memory. The memory stores computer programs, and the processor calls and runs the computer programs stored in the memory to execute the steps of the biconvex lens fabrication method provided in the first aspect of this application. The processor is also used to adjust the operating parameters of the biconvex lens molding die 102. The biconvex lens molding die 102 is used to mold a spherical glass preform to form a biconvex lens. A schematic diagram of the biconvex lens molding die 102 can be found in [reference needed]. Figure 5 As shown.
[0120] In some implementations, reference Figure 6 As shown, the biconvex lens forming mold 102 includes an upper mold 1021, a lower mold 1022, and a sleeve 1023. When the biconvex lens forming mold 102 is in operation, the spherical glass preform 200 is disposed within the accommodating space formed by the upper mold 1021, the lower mold 1022, and the sleeve 1023.
[0121] In practical applications, the glass molding process includes the basic steps of heating and heat preservation, hot pressing and pressure preservation, and cooling and demolding. The lens forming effect is controlled by adjusting process parameters such as molding temperature, molding speed, and cooling rate. A schematic diagram of the biconvex lens forming mold 102 in the heating and heat preservation state can be found here. Figure 7 As shown, the structural schematic diagram of the biconvex lens forming mold 102 in the hot-pressing and pressure-holding state can be referred to... Figure 8 As shown, the structural schematic diagram of the biconvex lens forming mold 102 in the cooling and demolding state can be referred to... Figure 9 As shown.
[0122] A third aspect of this application provides a terminal device, the schematic diagram of which is as follows: Figure 10 As shown. The terminal device includes a processor, memory, network interface, display screen, and temperature sensor connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the terminal device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for fabricating a biconvex lens. The display screen can be a liquid crystal display (LCD) or an electronic ink display (EI). The temperature sensor is pre-installed inside the terminal device to detect the operating temperature of the internal components.
[0123] Those skilled in the art will understand that Figure 10 The schematic diagram shown is only a partial structural diagram related to the present invention and does not constitute a limitation on the terminal device to which the present invention is applied. The specific terminal device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.
[0124] In some embodiments, this application provides a terminal device including a processor and a memory. The memory stores a computer program, and the processor calls and runs the computer program stored in the memory to perform the steps of the biconvex lens fabrication method provided in the first aspect of this application. In a fourth aspect, this application provides a computer-readable storage medium storing a computer program that causes a computer to perform the steps of the biconvex lens fabrication method provided in the first aspect of this application.
[0125] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.
[0126] The technical features of the above embodiments can be combined without changing the basic principles of this application. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0127] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the patent protection scope of this application should be determined by the appended claims.
Claims
1. A method for manufacturing a biconvex lens, characterized in that, include: Obtain a set of process parameters, wherein each process parameter in the set includes molding temperature, molding speed, and cooling rate; The set of process parameters is optimized using a molding optimization algorithm to obtain the target process parameters; The working parameters of the biconvex lens forming mold are adjusted based on the target process parameters so that the biconvex lens forming mold can produce a biconvex lens according to the working parameters.
2. The method for preparing a biconvex lens according to claim 1, characterized in that, The steps of optimizing the set of process parameters using a molding optimization algorithm to obtain the target process parameters include: The molding prediction model using the molding optimization algorithm is used to predict each process parameter in the set of process parameters to obtain the target molding data corresponding to each process parameter. An evaluation function is used to evaluate the target molding data corresponding to each process parameter, and the evaluation result of each process parameter is obtained. All evaluation results are filtered and processed according to preset rules to obtain the target result; The process parameters corresponding to the target result are taken as the target process parameters.
3. The method for preparing a biconvex lens according to claim 2, characterized in that, The steps of using the compression molding optimization algorithm to predict each process parameter in the set of process parameters and obtain the target molding data corresponding to each process parameter include: The first hidden layer of the molding prediction model is used to perform feature extraction processing on each process parameter in the process parameter set to obtain the first feature vector corresponding to each process parameter; The first objective activation function of the compression molding prediction model is used to perform nonlinear transformation on the first feature vector corresponding to each process parameter to obtain the second feature vector corresponding to each process parameter. The second hidden layer of the compression molding prediction model is used to perform feature extraction processing on the second feature vector corresponding to each process parameter to obtain the third feature vector corresponding to each process parameter. The second objective activation function of the compression molding prediction model is used to perform nonlinear transformation on the third feature vector corresponding to each process parameter to obtain the fourth feature vector corresponding to each process parameter. The third hidden layer of the molding prediction model is used to extract features from the fourth feature vector corresponding to each process parameter to obtain the fifth feature vector corresponding to each process parameter. The fifth feature vector corresponding to each process parameter is nonlinearly transformed using the third objective activation function of the compression molding prediction model to obtain the target molding data corresponding to each process parameter.
4. The method for preparing a biconvex lens according to claim 2, characterized in that, The method for preparing the biconvex lens further includes a training step for the molding prediction model, which includes: Obtain a training dataset and a real label set. Each training data in the training dataset includes molding training temperature, molding training speed, and cooling training speed. Each training data has a one-to-one correspondence with one of the real labels in the real label set. The original molding prediction model is used to perform prediction processing on each training data in the training dataset to obtain the prediction result corresponding to each training data. The loss function is calculated using the prediction result corresponding to each training data point and the true label corresponding to each training data point. The first model parameters of the original molding prediction model are adjusted based on the loss function to form the molding prediction model.
5. The method for preparing a biconvex lens according to claim 4, characterized in that, After adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model, the training step of the molding prediction model further includes: The compression molding prediction model is evaluated using a model evaluation method to obtain the evaluation results; The second model parameters of the compression molding prediction model are adjusted based on the evaluation results.
6. The method for preparing a biconvex lens according to claim 4, characterized in that, After adjusting the first model parameters of the original molding prediction model based on the loss function to form the molding prediction model, the training step of the molding prediction model further includes: The third model parameters of the compression molding prediction model are optimized using a first machine learning algorithm.
7. The method for preparing a biconvex lens according to claim 4, characterized in that, The training steps of the compression molding prediction model also include the step of obtaining the real label set, which includes: Obtain a set of molding data, wherein each set of molding data includes a first residual stress, a first angular deviation, and a first shape deviation; The established dataset is optimized using an orthogonal experimental method to obtain the true label set.
8. A system for fabricating a biconvex lens, characterized in that, include: A terminal device and a biconvex lens forming mold, wherein the terminal device includes a processor and a memory, the memory is used to store a computer program, the processor is used to call and run the computer program stored in the memory to perform the steps of the biconvex lens preparation method according to any one of claims 1 to 7, and the processor is also used to adjust the working parameters of the biconvex lens forming mold.
9. A terminal device, characterized in that, include: A processor and a memory, the memory for storing a computer program, the processor for calling and running the computer program stored in the memory to perform the steps of the method for preparing a biconvex lens according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, Used to store a computer program that causes a computer to perform the steps of the method for preparing a biconvex lens according to any one of claims 1 to 7.