CN115329554B - Intelligent optimization method and system for process for preparing hollow nanostructure by emulsion soft template method - Google Patents
Intelligent optimization method and system for process for preparing hollow nanostructure by emulsion soft template method Download PDFInfo
- Publication number
- CN115329554B CN115329554B CN202210899774.7A CN202210899774A CN115329554B CN 115329554 B CN115329554 B CN 115329554B CN 202210899774 A CN202210899774 A CN 202210899774A CN 115329554 B CN115329554 B CN 115329554B
- Authority
- CN
- China
- Prior art keywords
- model
- hollow
- data
- solid
- soft template
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/26—Composites
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Evolutionary Computation (AREA)
- Geometry (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Polymerisation Methods In General (AREA)
Abstract
The invention relates to an intelligent optimization method and system for a process for preparing a hollow nano structure by an emulsion soft template method, which belong to the technical field of inorganic material preparation, can accurately and quickly predict the appearance and uniformity of a micro/nano product of emulsion polymerization reaction, and solve the problem of process optimization of a hollow nanosphere with fine granularity, narrow distribution and no solid impurities; the method comprises the steps of carrying out data classification according to reactant data, synthesis parameters and product morphology when a hollow nano structure is prepared by an emulsion soft template method, and constructing a hollow and solid prediction model and a size uniformity prediction model by using the classified data for predicting the product; and obtaining the influence degree of each reactant and each synthesis parameter on the hollow appearance and the size uniformity of the product through SHAP characteristic analysis, and using the influence degree to optimize and regulate the process of preparing the hollow nano structure by the emulsion soft template method.
Description
Technical Field
The invention relates to the technical field of inorganic material preparation, in particular to a process optimization method for preparing a hollow nano structure by an emulsion soft template method based on machine learning.
Background
The hollow structure has the advantages of large specific surface area, small density, capacity of carrying other substances and the like, and has wide application prospect in the fields of drug transportation, catalysis, adsorption and the like. The traditional preparation method of the hollow structure comprises a hard mask and a soft template. The hard mask method relates to a plurality of steps of preparation of a template agent, surface treatment, coating, removal of the template agent and the like, and not only wastes materials, but also has a complicated process. In the soft template method, some soft substances such as micelles, emulsion droplets and the like are used as templates which are easy to prepare and remove, have the characteristics of convenience, simplicity and easiness for large-batch preparation, and are potential large-scale preparation methods. However, the method for preparing the hollow material by using the soft template such as emulsion liquid drop and the like has the biggest problems that the template agent is easy to have uneven size and deformation, so that the product often contains solid impurities, uneven size, large size and the like. Therefore, how to finely regulate and control the technological parameters so as to obtain hollow spheres with fine particle size, narrow distribution and no solid impurities is an important problem which restricts the preparation method.
Usually, scientists rely on expert experience to design orthogonal experiments of reactants and synthesis parameters by trial and error. However, the number of possible reactants and synthesis parameters is infinite and optimization of the synthesis parameters is time consuming and laborious. Therefore, rational design of reactants and parameters for soft template synthesis of hollow micro/nanostructures remains a challenge.
Accordingly, there is a need to develop a process optimization method for preparing hollow nanostructures based on machine learning emulsion soft template method to address the deficiencies of the prior art, so as to solve or alleviate one or more of the above problems.
Disclosure of Invention
In view of the above, the invention provides an intelligent optimization method and system for a process for preparing a hollow nano structure by using an emulsion soft template method, which can accurately and rapidly predict the morphology and uniformity of a micro/nano product of emulsion polymerization reaction, and solve the problem of process optimization of hollow nanospheres with fine granularity, narrow distribution and no solid impurities.
On one hand, the invention provides an intelligent optimization method for a process for preparing a hollow nano structure by an emulsion soft template method, the method classifies data according to reactant data, synthesis parameters and product morphology when the hollow nano structure is prepared by the emulsion soft template method, and constructs a hollow and solid prediction model and a size uniformity prediction model by using the classified data for predicting the product;
and (3) obtaining the influence degree of each reactant and each synthesis parameter on the hollow appearance and the size uniformity of the product through SHAP characteristic analysis, and optimizing and regulating the process of preparing the hollow nano structure by the emulsion soft template method.
As to the above-mentioned aspects and any possible implementation manner, there is further provided an implementation manner, where the specific steps of the method include:
s1, collecting data, preprocessing the data into a model, and constructing a used data set;
collecting reactant data, synthesis parameters and product morphology data which influence the morphology of a product in the process of preparing the hollow nano structure by an emulsion soft template method; the product morphology data comprises whether the product is a hollow particle and size uniformity data;
s2, constructing an empty and solid prediction model;
dividing the data set obtained in the step S1 into a training set and a testing set, training an original model of the hollow and solid prediction model by using the training set, and then testing and evaluating the trained model by using the testing set to obtain the hollow and solid prediction model meeting the requirements;
s3, constructing a size uniformity prediction model;
removing the data example of the pure solid sphere in the data set obtained in the step S1, and then dividing the data example into a training set and a testing set; training an original model of the size uniformity prediction model by using the training set, and then testing and evaluating the trained model by using the testing set to obtain a size uniformity prediction model meeting the requirement;
s4, optimizing the process;
calculating the SHAP value of each feature in the data set, measuring the contribution degree of each feature to the prediction result of the empty and solid prediction model and/or the size uniformity prediction model according to the SHAP value, and determining the optimal process window according to the contribution degree.
As to the above-mentioned aspect and any possible implementation manner, there is further provided an implementation manner, where the content preprocessed in step S1 includes: and removing data with more repeated and missing values to form a data set used for model construction.
The above-described aspects and any possible implementation manners further provide an implementation manner, where the original model of the empty and solid prediction model is any one of a support vector classification model, a random forest classification model, a gradient boost classification model, and an Adaboost classification model;
the original model of the size uniformity prediction model is any one of a support vector classification model, a random forest classification model, a gradient lifting classification model and an Adaboost classification model.
The above-described aspects and any possible implementation further provide an implementation in which the training of the empty-solid prediction model or the original model of the size uniformity prediction model with the training set includes:
a grid search is performed on the original model and the parameters on the training set are adjusted with 5-fold cross validation to avoid overfitting.
The above-described aspects and any possible implementations further provide an implementation where the condition that the optimal process window needs to satisfy includes:
the result of the corresponding data in the process window after being predicted by the empty and solid prediction model is empty, an
The result after the prediction of the size uniformity prediction model is uniform.
The above-mentioned aspects and any possible implementation manner further provide an implementation manner, where the original model of the empty and solid prediction model is a gradient lifting classification model, and the setting of parameters of the gradient lifting classification model includes:
the learning rate is set to 0.01, the maximum depth is set to 4, and the number of learners is set to 100.
The above-described aspect and any possible implementation manner further provide an implementation manner, where the original model of the size uniformity prediction model is a gradient boosting classification model, and the setting of parameters of the gradient boosting classification model includes:
the learning rate is set to 0.5, the maximum depth is set to 5, and the number of learners is set to 30.
The above aspects and any possible implementations further provide an implementation in which the reactant data includes: the types and the amounts of the organic monomer, the oil phase, the initiator and the water phase;
the synthesis parameters include: number of emulsification, time of emulsification, mode of emulsification, reaction time and number of washes.
In another aspect, the present invention provides an intelligent optimization system for a process for preparing hollow nanostructures by an emulsion soft template method, wherein the system comprises:
the data acquisition input module is used for acquiring data to be predicted and inputting the data into the system;
the empty and solid prediction module is used for predicting whether the product is hollow particles or solid particles according to input data;
the size uniformity prediction module is used for predicting whether the size of the product meets the uniformity requirement or not according to the input data;
and the process optimization analysis module is used for calculating a SHAP value corresponding to the input data, measuring the contribution degree of the SHAP value to the prediction result of the empty and solid prediction module and/or the size uniformity prediction module according to the SHAP value, and determining the optimal process window according to the contribution degree.
Compared with the prior art, one of the technical schemes has the following advantages or beneficial effects: the invention overcomes the obstacles of narrow process window, large parameter space, complex dependency relationship between the front and the rear processes and difficult regulation and control in the process of preparing the hollow nano structure by a soft template method, improves the reactant consumption and the efficiency of optimizing the synthesis parameters, and has important guiding significance for the rational preparation of nano materials with high structural complexity.
Of course, it is not necessary for any one product in which the invention is practiced to achieve all of the above-described technical effects simultaneously.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart of a method for preparing hollow-structured particles by machine learning-guided emulsion interfacial polymerization according to an embodiment of the present invention.
Detailed Description
In order to better understand the technical scheme of the invention, the following detailed description of the embodiments of the invention is made with reference to the accompanying drawings.
It should be understood that the described embodiments are only some embodiments of the invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Aiming at the defects of the prior art, the invention provides a process optimization method for preparing a hollow nano structure by a soft template method based on machine learning. Meanwhile, by combining with a Shapley Additive exPlations (SHAP) analysis technology, the importance of each reactant and synthesis parameter on influencing the hollow/solid morphology and the size uniformity of a product is analyzed, the rational optimization and regulation of the reactant dosage and the synthesis parameter are realized, and the method has important guiding significance on the rational preparation of the nano material with high structural complexity.
The optimization method comprises the following steps:
step 1, data acquisition and preparation:
collecting and recording reactant and synthesis parameter information which may affect the product appearance in the experimental process, including: organic monomer, oil phase, initiator, water phase and corresponding dosage, and emulsification times, emulsification time, emulsification mode, reaction time and washing times. And observing the appearance of the product by using an electron microscope, grading and marking according to whether the product contains solid particles or not, and further marking the uniformity grade of the product if the product is hollow particles.
Preprocessing the original acquired data, removing data with more repeated and missing values, and forming a data set for modeling.
Step 2, establishing a hollow/solid morphology classification model:
the data set is divided into a training set and a testing set according to the ratio of 8. SVC obtains support vectors to determine decision functions by solving hollow/solid classification hyperplanes which can correctly divide training data sets and have maximum geometric intervals, and realizes hollow/solid morphology classification. The influence of reactants and synthesis parameters on the morphology of a product is a nonlinear problem, SVC needs to realize nonlinear separability through a kernel function, but for high-dimensional and huge unknown spaces of the reactants and the synthesis parameters, the difficulty in selecting a proper kernel function is very high, so that great uncertainty is brought to classification accuracy. RFC, GBC and AdaBC are all methods based on ensemble learning, thereby obtain better classification effect RFC based on bagging-off method through training a plurality of weak learners and integrating the learning result, and a plurality of base classifiers are trained in parallel, adopt the mode of voting to obtain the classification result. GBC and AdaBC train each base classifier in series based on a lifting method, and dependence exists between each base classifier. The basic idea is to stack the base classifiers layer by layer, each layer gives higher weight to the sample which is wrongly divided by the base classifier of the previous layer during training, and the final result is obtained according to the weighting of the structure of each layer of classifier during testing. RFC based on a bagging method can reduce the hollow/solid morphology classification variance, and GBC and AdaBC based on a lifting method can reduce the hollow/solid morphology classification deviation. Adaboost is to locate the deficiencies of the model by raising the weights of the misclassified samples, while GBC is to locate the deficiencies of the model by negative gradients, so GBC can use a wider variety of penalty functions. Meanwhile, the GBC adopting the decision tree as the weak classifier has better interpretability and robustness, and can automatically discover the high-order relation among the features.
A grid search was performed on the machine learning model, adjusting the parameters on the training set with 5-fold cross-validation to avoid overfitting. The grid searching method is mainly used for model parameter adjustment, namely, the grid searching method helps people to find a group of most appropriate model setting parameters so that the model prediction achieves a better effect, the group of parameters are different from the parameters learned in the model training process, the parameters need to be preset before training, and the parameters are called as hyper-parameters. The grid search method randomly combines values in a given parameter list and evaluates the model effect of each combination respectively, thereby finding a group of optimal hyper-parameters. In the invention, the learning rate learning _ rate of the GBC model is set to be 0.01, the maximum depth max _ depth is set to be 4, and the number n _ estimators of the learners is set to be 100, so that the accuracy and uncertainty of the GBC model are optimal. The number of n _ estimators as the largest weak learners is too small or too large, which easily causes classification under-fitting or over-fitting, and the weight reduction coefficient learning _ rate of each weak learner and the maximum depth max _ depth of the decision tree of the weak learner are matched according to the influence relationship of reactants and synthesis parameters on hollow and solid morphologies. Here we compared the product morphology classification effect under 5-fold cross validation of the training set under different combinations of n _ estimators, learning _ rate and max _ depth, as shown in Table 1. The best combination of accuracy and uncertainty of the GBC model is model 8.
TABLE 1
Using this set of parameters, the GBC hollow/solid classification model is retrained on the entire training set and evaluated on the retention test set.
Step 3, establishing a size uniformity classification model:
before training the size uniformity classification model, the manual removal of the data instances where the product was a pure solid sphere. And further constructing a size uniformity classification model in the same process as the hollow/solid model training process. The size uniformity classification model is implemented using a gradient lifting classification (GBC) model. The GBC model has a learning rate of 0.5 for learning, a maximum depth of 5 for max depth, and a number of learners n _ estimators of 30, with the best accuracy and uncertainty of the GBC model. The settings of n _ estimators, learning _ rate and max _ depth depend on the influence of the reactants and synthesis parameters on the size uniformity. Here we compare the size uniformity classification effect for different combinations of n _ estimators, learning _ rate and max _ depth under 5-fold cross validation of the training set, as shown in Table 2. The best combination of accuracy and uncertainty of the GBC model is model 11.
TABLE 2
The GBC model is then retrained using the optimal parameters over the entire training set and evaluated on the retention test set.
Step 4, regulating and controlling a process window: SHAP values for features across the entire data set are calculated, measuring the contribution of each feature to the two model predictions.
SHAP is a "model interpretation" package developed by Python that can interpret the output of any machine learning model. The name of the method is derived from Shapley Additive exPlanation, SHAP constructs an Additive interpretation model under the inspiration of cooperative game theory, and all characteristics are regarded as 'contributors'. For each prediction sample, the model produces a prediction value, and SHAP value is the value assigned to each feature in the sample.
The specific way to calculate the SHAP value is as follows:
based on the idea of game theory, all reactants and synthesis parameters are regarded as contributors, and an additive interpretation model is constructed
y i ＝y base +f(x i1 )+f(x i2 )+…+f(x ik )
Wherein the ith sample is x i The jth feature of the ith sample is x ij The predicted value of the model for the sample is y i The model base line is y base ，f(x ij ) Is x ij The SHAP value of (i), i.e. the j-th feature pair in the i-th sample is the final predicted value y i The contribution value of (c). For each prediction sample, the model generates a prediction value, and the SHAP value is the assigned value of each feature in the sampleTo the value of (c). When f (x) ij ) When the ratio is more than 0, the characteristic improves the predicted value and acts positively; conversely, the characteristic is shown to reduce the predicted value, and has adverse effect. According to the traditional importance analysis methods such as feature import and simulation import, the comprehensive importance of each reactant and synthesis parameter on influencing the classification of hollow/solid morphology and size uniformity can be obtained through calculation, but how the characteristics influence the prediction result cannot be calculated. SHAP value reflects the influence of each reactant and synthesis parameter in each sample, and shows the positive and negative of the influence.
Comprehensively designing the reactants and the process parameters by comprehensively analyzing the contribution of the reactant consumption and the synthesis parameters to the product appearance and size, and determining the optimal process window, wherein the setting principle of the process window meets the following requirements: the prediction result of the hollow/solid morphology classification model is hollow, and the prediction result of the size uniformity classification model is uniform.
Example 1:
the preparation of dopamine hollow nanosphere particles is exemplified by interfacial polymerization of a water/Trioctylamine (TOA) oil-in-water emulsion.
First, a certain amount of dopamine hydrochloride, deionized water and ethanol were added to a 100ml beaker and dissolved into a mixed solution under magnetic stirring. Then, a certain amount of TOA is dropped into the mixed solution to form a uniform oil film on the surface of the solution. The mixed solution was put into an ultrasonic wave machine with a power of 300W to perform first ultrasonic emulsification. Thereafter, magnetic stirring was performed at 1000rpm to emulsify the mixture. Then, a second ultrasonic emulsification is performed to obtain an emulsion. Then, a certain amount of ammonia water was added to the emulsion under magnetic stirring to conduct polymerization. And (3) repeatedly washing the product after reaction by using alcohol, centrifuging to obtain black precipitate, and drying at 60 ℃ to obtain the polydopamine.
The reactants and synthesis parameters were recorded, including the amounts of dopamine, TOA, ammonia, water and alcohol added, as well as the first and second phacoemulsification times, stir emulsification times, reaction times and wash times. The morphology of the products was characterized by field emission scanning electron microscopy (SEM, hitachi, SU 8100), which marks them as whether the hollow spheres comprise solid spheres and whether the size of the hollow spheres is uniform. We define the morphology of the product as a binary problem, i.e. one is that the product is a pure hollow sphere, which is denoted by the classification number "1", and the other is that the product is a mixture of hollow sphere and solid sphere (i.e. the hollow sphere product contains solid spheres) or the product is a pure solid sphere, which is denoted by the classification number "2". For product size uniformity, high and low uniformity are indicated by classification numbers of "1" and "2".
The data set is divided into a training set and a testing set according to 8, and a machine learning algorithm is selected from Support Vector Classification (SVC), random Forest Classification (RFC), gradient Boosting Classification (GBC) and Adaboost classification (AdaBC). A grid search was performed for each machine learning model, adjusting the parameters on the training set with 5-fold cross-validation to avoid overfitting. The GBC model has a learning rate learning _ rate set to 0.01, a maximum depth max _ depth set to 4, and an estimator number n _ estimators set to 100, with the best accuracy and uncertainty of the GBC model. Using this parameter, the GBC hollow/solid classification model is retrained on the entire training set and evaluated on the retention test set.
Before training the size uniformity classification model, the manual removal of the data instances where the product was a pure solid sphere. And further constructing a size uniformity classification model in the same process as the hollow/solid model training process. The GBC model learning rate is set to 0.5, the maximum depth max depth is set to 5, the number of estimators n _ estimators is set to 30, and the accuracy and uncertainty of the GBC model is optimal. The GBC model is then retrained using the optimal parameters over the entire training set and evaluated on the retention test set.
Then, SHAP values for the features are calculated over the entire data set, measuring the contribution of each feature to the two model predictions. Comprehensively designing the reactants and the process parameters by comprehensively analyzing the contribution of the reactant consumption and the synthesis parameters to the product appearance and size, and determining the optimal process window, wherein the setting principle of the process window meets the following requirements: the prediction result of the hollow/solid morphology classification model is hollow, and the prediction result of the size uniformity classification model is uniform.
The invention also provides a process optimization system for preparing the hollow nano structure by the emulsion soft template method based on machine learning, which comprises the following steps:
the data acquisition input module is used for acquiring data to be predicted and inputting the data into the system;
the empty and solid prediction module is used for predicting whether the product is hollow particles or solid particles according to input data;
the size uniformity prediction module is used for predicting whether the size of the product meets the uniformity requirement or not according to the input data;
and the process optimization analysis module is used for calculating a SHAP value corresponding to the input data, measuring the contribution degree of the SHAP value to the prediction result of the space-solid prediction module and/or the size uniformity prediction module, and determining the optimal process window according to the contribution degree.
The method and the system for optimizing the process for preparing the hollow nano structure by the emulsion soft template method based on machine learning provided by the embodiment of the application are introduced in detail. The above description of the embodiments is only for the purpose of helping to understand the method of the present application and its core idea; meanwhile, for a person skilled in the art, according to the idea of the present application, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present application.
Claims (9)
1. An intelligent optimization method for a process for preparing a hollow nano structure by an emulsion soft template method is characterized in that the method classifies data according to reactant data, synthesis parameters and product morphology when the hollow nano structure is prepared by the emulsion soft template method, and constructs a hollow and solid prediction model and a size uniformity prediction model by using the classified data for predicting the product;
through SHAP characteristic analysis, the influence degree of each reactant and each synthesis parameter on the hollow appearance and the size uniformity of the product is obtained, and the method is used for optimizing and regulating the process of preparing the hollow nano structure by the emulsion soft template method;
the method comprises the following specific steps:
s1, collecting data, and preprocessing the data into a data set required by constructing a corresponding model;
collecting reactant data, synthesis parameters and product morphology data which influence the morphology of a product in the process of preparing the hollow nano structure by an emulsion soft template method; the product morphology data comprises whether the product is a hollow particle and size uniformity data;
s2, constructing an empty and solid prediction model;
dividing the data set obtained in the step S1 into a training set and a test set, training an original model of the hollow and solid prediction model by using the training set, and performing test evaluation on the trained model by using the test set to obtain the hollow and solid prediction model meeting the requirements;
s3, constructing a size uniformity prediction model;
removing the data example of the pure solid sphere in the data set obtained in the step S1, and then dividing the data example into a training set and a testing set; training an original model of the size uniformity prediction model by using the training set, and then testing and evaluating the trained model by using the testing set to obtain a size uniformity prediction model meeting the requirement;
s4, optimizing the process;
and calculating a SHAP value of each feature in the data set, measuring the contribution degree of each feature to the prediction result of the empty and solid prediction model and/or the size uniformity prediction model according to the SHAP value, and determining the optimal process window according to the contribution degree.
2. The intelligent optimization method for preparing the hollow nanostructure by the emulsion soft template method according to claim 1, wherein the pretreatment in step S1 comprises: and removing data with more repeated and missing values to form a data set used for model construction.
3. The intelligent optimization method for the process for preparing the hollow nanostructure by the emulsion soft template method according to claim 1, wherein the original model of the hollow solid prediction model is any one of a support vector classification model, a random forest classification model, a gradient lifting classification model and an Adaboost classification model;
the original model of the size uniformity prediction model is any one of a support vector classification model, a random forest classification model, a gradient lifting classification model and an Adaboost classification model.
4. The intelligent optimization method for preparing hollow nano-structure process by emulsion soft template method according to claim 1, wherein the content of training the original model of the hollow solid prediction model or the size uniformity prediction model by using the training set comprises:
a grid search is performed on the original model and the parameters on the training set are adjusted with 5-fold cross validation to avoid overfitting.
5. The intelligent optimization method for the process of preparing the hollow nano structure by the emulsion soft template method according to claim 1, wherein the optimal process window needs to meet the conditions comprising:
the result of the corresponding data in the process window after being predicted by the empty and solid prediction model is empty, an
The result after the prediction of the size uniformity prediction model is uniform.
6. The intelligent optimization method for the process for preparing the hollow nano structure by the emulsion soft template method according to claim 3, wherein the original model of the hollow solid prediction model is a gradient lifting classification model, and the parameter setting of the gradient lifting classification model comprises the following steps:
the learning rate is set to 0.01, the maximum depth is set to 4, and the number of learners is set to 100.
7. The intelligent optimization method for the process of preparing the hollow nano structure by the emulsion soft template method according to claim 3, wherein the original model of the size uniformity prediction model is a gradient lifting classification model, and the parameter setting of the gradient lifting classification model comprises:
the learning rate is set to 0.5, the maximum depth is set to 5, and the number of learners is set to 30.
8. The intelligent optimization method for preparing hollow nano-structure process by emulsion soft template method according to claim 1, wherein the reactant data comprises: the types and the amounts of the organic monomer, the oil phase, the initiator and the water phase;
the synthesis parameters include: number of emulsification times, emulsification time, emulsification mode, reaction time and washing times.
9. An intelligent optimization system for a process for preparing hollow nanostructures by an emulsion soft template method, which is used for realizing the intelligent optimization method for the process for preparing the hollow nanostructures by the emulsion soft template method according to any one of claims 1 to 8, and comprises:
the data acquisition input module is used for acquiring data to be predicted and inputting the data into the system;
the empty and solid prediction module is used for predicting whether the product is hollow particles or solid particles according to input data;
the size uniformity prediction module is used for predicting whether the size of the product meets the uniformity requirement or not according to the input data;
and the process optimization analysis module is used for calculating a SHAP value corresponding to the input data, measuring the contribution degree of the SHAP value to the prediction result of the space-solid prediction module and/or the size uniformity prediction module, and determining the optimal process window according to the contribution degree.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210899774.7A CN115329554B (en) | 2022-07-28 | 2022-07-28 | Intelligent optimization method and system for process for preparing hollow nanostructure by emulsion soft template method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210899774.7A CN115329554B (en) | 2022-07-28 | 2022-07-28 | Intelligent optimization method and system for process for preparing hollow nanostructure by emulsion soft template method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115329554A CN115329554A (en) | 2022-11-11 |
| CN115329554B true CN115329554B (en) | 2023-03-10 |
Family
ID=83920031
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210899774.7A Active CN115329554B (en) | 2022-07-28 | 2022-07-28 | Intelligent optimization method and system for process for preparing hollow nanostructure by emulsion soft template method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115329554B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765412A (en) * | 2018-06-08 | 2018-11-06 | 湖北工业大学 | A kind of steel strip surface defect sorting technique |
| CN109658396A (en) * | 2018-12-06 | 2019-04-19 | 华中科技大学 | A kind of casting x-ray image defect identification method based on selective search and CNN |
| CN114334011A (en) * | 2021-12-03 | 2022-04-12 | 福建技术师范学院 | Bioactive peptide prediction method based on deep convolutional neural network |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018236373A1 (en) * | 2017-06-21 | 2018-12-27 | Halliburton Energy Services, Inc. | Prediction of cutting size and shape generated by a drill bit |
-
2022
- 2022-07-28 CN CN202210899774.7A patent/CN115329554B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108765412A (en) * | 2018-06-08 | 2018-11-06 | 湖北工业大学 | A kind of steel strip surface defect sorting technique |
| CN109658396A (en) * | 2018-12-06 | 2019-04-19 | 华中科技大学 | A kind of casting x-ray image defect identification method based on selective search and CNN |
| CN114334011A (en) * | 2021-12-03 | 2022-04-12 | 福建技术师范学院 | Bioactive peptide prediction method based on deep convolutional neural network |
Non-Patent Citations (1)
| Title |
|---|
| 纳米流体粘度的实验测量及预测建模研究进展;贾小权等;《热能动力工程》(第02期);9-18页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115329554A (en) | 2022-11-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Schulz et al. | Investigating convergence of restricted boltzmann machine learning | |
| Swersky et al. | A tutorial on stochastic approximation algorithms for training restricted Boltzmann machines and deep belief nets | |
| Tanielian et al. | Learning disconnected manifolds: a no gan’s land | |
| Refinetti et al. | Neural networks trained with SGD learn distributions of increasing complexity | |
| CN110222953A (en) | A kind of power quality hybrid perturbation analysis method based on deep learning | |
| CN105930846A (en) | Neighborhood information and SVGDL (support vector guide dictionary learning)-based polarimetric SAR image classification method | |
| CN110287985B (en) | Depth neural network image identification method based on variable topology structure with variation particle swarm optimization | |
| CN103208007A (en) | Face recognition method based on support vector machine and genetic algorithm | |
| Pezeshki et al. | Multi-scale feature learning dynamics: Insights for double descent | |
| CN104050680B (en) | Based on iteration self-organizing and the image partition method of multi-agent genetic clustering algorithm | |
| CN115329554B (en) | Intelligent optimization method and system for process for preparing hollow nanostructure by emulsion soft template method | |
| CN102722717B (en) | Cell division identification method | |
| Chen et al. | Ising granularity image analysis on VAE–GAN | |
| Elflein | Master's Thesis: Out-of-distribution Detection with Energy-based Models | |
| Lu et al. | Machine Learning for Analyses and Automation of Structural Characterization of Polymer Materials | |
| CN104537660B (en) | Image partition method based on Multiobjective Intelligent body evolution clustering algorithm | |
| Xie et al. | Deeply supervised residual network for HEp-2 cell classification | |
| CN116665039A (en) | Small sample target identification method based on two-stage causal intervention | |
| CN105975902A (en) | SAMME.RCW algorithm based face recognition optimization method | |
| Wang et al. | Kernel-based deep learning for intelligent data analysis | |
| CN112686881B (en) | Particle material mixing uniformity detection method based on image statistical characteristics and LSTM composite network | |
| CN115859115A (en) | Intelligent resampling technology based on Gaussian distribution | |
| Liu et al. | Facial expression recognition approach based on least squares support vector machine with improved particle swarm optimization algorithm | |
| Chen et al. | Kernel k-means based framework for aggregate outputs classification | |
| Zheng et al. | Fuzzy pre-extracting method for support vector machine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |