CN116913423A - Synthetic process optimization method and system for unsaturated polyester resin - Google Patents
Synthetic process optimization method and system for unsaturated polyester resin Download PDFInfo
- Publication number
- CN116913423A CN116913423A CN202310763583.2A CN202310763583A CN116913423A CN 116913423 A CN116913423 A CN 116913423A CN 202310763583 A CN202310763583 A CN 202310763583A CN 116913423 A CN116913423 A CN 116913423A
- Authority
- CN
- China
- Prior art keywords
- synthesis
- unsaturated polyester
- polyester resin
- performance
- result
- 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.)
- Pending
Links
- 229920006337 unsaturated polyester resin Polymers 0.000 title claims abstract description 144
- 238000000034 method Methods 0.000 title claims abstract description 108
- 238000005457 optimization Methods 0.000 title claims abstract description 77
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 210
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 210
- 239000002994 raw material Substances 0.000 claims abstract description 82
- 230000008569 process Effects 0.000 claims abstract description 74
- 238000007621 cluster analysis Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims description 53
- 125000000524 functional group Chemical group 0.000 claims description 47
- 238000011156 evaluation Methods 0.000 claims description 41
- 239000011159 matrix material Substances 0.000 claims description 28
- 238000004458 analytical method Methods 0.000 claims description 24
- 238000012512 characterization method Methods 0.000 claims description 15
- 238000004422 calculation algorithm Methods 0.000 claims description 14
- QQVIHTHCMHWDBS-UHFFFAOYSA-N isophthalic acid Chemical compound OC(=O)C1=CC=CC(C(O)=O)=C1 QQVIHTHCMHWDBS-UHFFFAOYSA-N 0.000 claims description 12
- XNGIFLGASWRNHJ-UHFFFAOYSA-N phthalic acid Chemical compound OC(=O)C1=CC=CC=C1C(O)=O XNGIFLGASWRNHJ-UHFFFAOYSA-N 0.000 claims description 12
- IISBACLAFKSPIT-UHFFFAOYSA-N bisphenol A Chemical compound C=1C=C(O)C=CC=1C(C)(C)C1=CC=C(O)C=C1 IISBACLAFKSPIT-UHFFFAOYSA-N 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 7
- 229920002994 synthetic fiber Polymers 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 229920001567 vinyl ester resin Polymers 0.000 claims description 6
- 238000004220 aggregation Methods 0.000 claims description 5
- 230000002776 aggregation Effects 0.000 claims description 5
- 238000007417 hierarchical cluster analysis Methods 0.000 claims description 4
- 230000008030 elimination Effects 0.000 claims description 3
- 238000003379 elimination reaction Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims 1
- 238000006243 chemical reaction Methods 0.000 abstract description 18
- 230000006870 function Effects 0.000 description 22
- 230000000694 effects Effects 0.000 description 5
- 239000000463 material Substances 0.000 description 5
- 239000000047 product Substances 0.000 description 5
- 230000035484 reaction time Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000000126 substance Substances 0.000 description 4
- 239000003054 catalyst Substances 0.000 description 3
- 238000010276 construction Methods 0.000 description 2
- 239000012467 final product Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 239000000178 monomer Substances 0.000 description 2
- 238000011056 performance test Methods 0.000 description 2
- 229920005989 resin Polymers 0.000 description 2
- 239000011347 resin Substances 0.000 description 2
- 239000002904 solvent Substances 0.000 description 2
- 229920006305 unsaturated polyester Polymers 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 239000000853 adhesive Substances 0.000 description 1
- 230000001070 adhesive effect Effects 0.000 description 1
- 239000004566 building material Substances 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000000576 coating method Methods 0.000 description 1
- -1 coatings Substances 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000004132 cross linking Methods 0.000 description 1
- 239000003431 cross linking reagent Substances 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 239000013078 crystal Substances 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013401 experimental design Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000000691 measurement method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000013386 optimize process Methods 0.000 description 1
- 150000002894 organic compounds Chemical class 0.000 description 1
- 229920000728 polyester Polymers 0.000 description 1
- 229920001225 polyester resin Polymers 0.000 description 1
- 239000004645 polyester resin Substances 0.000 description 1
- 238000006116 polymerization reaction Methods 0.000 description 1
- 239000000376 reactant Substances 0.000 description 1
- 230000009257 reactivity Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000001308 synthesis method Methods 0.000 description 1
- 230000002194 synthesizing effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C60/00—Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/20—Identification of molecular entities, parts thereof or of chemical compositions
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
-
- 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)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- Crystallography & Structural Chemistry (AREA)
- Chemical & Material Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Macromonomer-Based Addition Polymer (AREA)
Abstract
The invention provides a synthetic process optimization method and a system for unsaturated polyester resin, which relate to the technical field of process optimization and comprise the following steps: obtaining synthesis basic information of unsaturated polyester resin, including synthesis raw material information and synthesis control parameters, inputting a structure prediction model to output an unsaturated polyester resin prediction structure, performing cluster analysis to obtain a performance prediction result, judging whether a desired performance threshold is met, if not, optimizing to obtain a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the desired performance threshold of the unsaturated polyester resin, and controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result. The invention solves the technical problems of unstable reaction, low reaction rate and low intelligent degree caused by the fact that the existing synthesis process cannot accurately control synthesis conditions and accurately monitor key parameters and cannot adaptively adjust the synthesis process.
Description
Technical Field
The invention relates to the technical field of process optimization, in particular to a synthetic process optimization method and a synthetic process optimization system for unsaturated polyester resin.
Background
Unsaturated polyester resins (Unsaturated Polyester Resins, UPR) are an important class of polymeric materials with a wide range of applications including composites, building materials, coatings, adhesives, etc., which are composed of an unsaturated polyester resin matrix, a crosslinking agent and a solvent, and form a high molecular structure through chemical reaction, and as chemical science progresses, the synthesis method thereof is increasingly studied.
However, the existing synthesis process of unsaturated polyester resin has certain disadvantages, the existing synthesis process cannot accurately control synthesis conditions and accurately monitor key parameters, and the synthesis process cannot be adaptively adjusted, so that unstable reaction, low reaction rate and low intelligent degree are caused. Therefore, there is a certain space for the synthesis of unsaturated polyester resins to be improved.
Disclosure of Invention
The application provides a method and a system for optimizing a synthesis process of unsaturated polyester resin, which aim to solve the technical problems that the existing synthesis process cannot accurately control synthesis conditions and accurately monitor key parameters, and cannot adaptively adjust the synthesis process, so that the reaction is unstable, the reaction rate is low and the intelligent degree is low.
In view of the above, the present application provides a synthetic process optimization method and system for unsaturated polyester resins.
In a first aspect of the present disclosure, there is provided a synthetic process optimization method for an unsaturated polyester resin, the method comprising: obtaining synthesis basic information of unsaturated polyester resin, wherein the synthesis basic information comprises synthesis raw material information and synthesis control parameters; inputting the synthesis raw material information and the synthesis control parameters into a structure prediction model, and outputting an unsaturated polyester resin prediction structure; based on a k neighbor clustering algorithm, carrying out clustering analysis on the unsaturated polyester resin prediction structure to obtain a performance prediction result; judging whether the performance prediction result meets the expected performance threshold of the unsaturated polyester resin; if not, optimizing the synthesis raw material information and/or the synthesis control parameters to obtain a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the expected performance threshold of the unsaturated polyester resin; and controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
In another aspect of the present disclosure, there is provided a synthesis process optimization system for unsaturated polyester resins, the system being used in the above method, the system comprising: the base information acquisition module is used for acquiring synthesis base information of the unsaturated polyester resin, wherein the synthesis base information comprises synthesis raw material information and synthesis control parameters; the prediction structure output module is used for inputting the synthetic raw material information and the synthetic control parameters into a structure prediction model and outputting an unsaturated polyester resin prediction structure; the cluster analysis module is used for carrying out cluster analysis on the unsaturated polyester resin prediction structure based on a k-nearest neighbor clustering algorithm to obtain a performance prediction result; the prediction result judging module is used for judging whether the performance prediction result meets the expected performance threshold of the unsaturated polyester resin; the control parameter optimization module is used for optimizing the synthesis raw material information and/or the synthesis control parameters if the synthesis raw material information and/or the synthesis control parameters are not met, and obtaining a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the expected performance threshold of the unsaturated polyester resin; and the synthesis control module is used for controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
One or more technical schemes provided by the application have at least the following technical effects or advantages:
obtaining synthesis basic information of unsaturated polyester resin, including synthesis raw material information and synthesis control parameters, inputting a structure prediction model, outputting an unsaturated polyester resin prediction structure, performing cluster analysis on the unsaturated polyester resin prediction structure, obtaining a performance prediction result, judging whether a desired performance threshold is met, if not, optimizing, and obtaining a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the desired performance threshold of the unsaturated polyester resin, and controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result. The method solves the technical problems that the existing synthesis process cannot accurately control synthesis conditions and accurately monitor key parameters, and cannot adaptively adjust the synthesis process, so that the reaction is unstable, the reaction rate is low, and the intelligent degree is low, realizes the accurate control of synthesis conditions and monitoring of key parameters, and adaptively adjusts the synthesis process, thereby achieving the technical effects of improving the intelligent degree, the synthesis efficiency and the product quality.
The foregoing description is only an overview of the present application, and is intended to be implemented in accordance with the teachings of the present application in order that the same may be more clearly understood and to make the same and other objects, features and advantages of the present application more readily apparent.
Drawings
FIG. 1 is a schematic flow diagram of a synthetic process optimization method for unsaturated polyester resins provided in an embodiment of the application;
FIG. 2 is a schematic diagram of a possible flow chart of outputting a predicted structure of an unsaturated polyester resin in a synthetic process optimization method for the unsaturated polyester resin according to an embodiment of the application;
FIG. 3 is a schematic diagram of a possible process for obtaining a performance prediction result in a synthetic process optimization method for unsaturated polyester resin according to an embodiment of the application;
FIG. 4 is a schematic diagram of a possible configuration of a synthesis process optimization system for unsaturated polyester resins according to an embodiment of the application.
Reference numerals illustrate: the system comprises a basic information acquisition module 10, a prediction structure output module 20, a cluster analysis module 30, a prediction result judgment module 40, a control parameter optimization module 50 and a synthesis control module 60.
Detailed Description
The embodiment of the application solves the technical problems that the existing synthesis process cannot accurately control synthesis conditions and accurately monitor key parameters and cannot adaptively adjust the synthesis process, so that the reaction is unstable, the reaction rate is low and the intelligent degree is low, realizes the accurate control of synthesis conditions and monitor key parameters and adaptively adjusts the synthesis process, and further achieves the technical effects of improving the intelligent degree, the synthesis efficiency and the product quality.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Example 1
As shown in fig. 1, an embodiment of the present application provides a synthetic process optimization method for an unsaturated polyester resin, the method comprising:
step S100: obtaining synthesis basic information of unsaturated polyester resin, wherein the synthesis basic information comprises synthesis raw material information and synthesis control parameters;
specifically, the relevant knowledge about the synthesis of unsaturated polyester resin is obtained through experimental design or literature investigation, firstly, raw materials for synthesizing unsaturated polyester resin, including monomers, catalysts, solvents and the like, are determined, and the types and the proportions of the raw materials are determined to be very critical to the synthesis process; all relevant synthesis control parameters during the synthesis of the unsaturated polyester resin, including temperature, reaction time, pressure, etc., are then collected and recorded, the settings of which affect the properties of the resin and the quality of the final product. Obtaining synthesis basis information helps to understand the impact of different raw materials and control parameters on resin properties, and to determine the start of synthesis.
Step S200: inputting the synthesis raw material information and the synthesis control parameters into a structure prediction model, and outputting an unsaturated polyester resin prediction structure;
further, as shown in fig. 2, step S200 of the present application includes:
step S210: collecting unsaturated polyester resin based on a federal domain to generate record data, wherein the federal domain is a service information sharing platform constructed by combining with a blockchain;
step S220: obtaining an unsaturated polyester resin matrix type, wherein the unsaturated polyester resin matrix type comprises phthalic acid type, isophthalic acid type, bisphenol A type, vinyl ester type and halogenated unsaturated polyester resin;
step S230: clustering the unsaturated polyester resin generated record data based on the phthalic acid type, the isophthalic acid type, the bisphenol A type, the vinyl ester type and the halogenated unsaturated polyester resin to obtain a first record data group, a second record data group, a third record data group, a fourth record data group and a fifth record data group;
step S240: training the structure prediction model by the first record data set, the second record data set, the third record data set, the fourth record data set, and the fifth record data set;
Step S250: the synthetic raw material information comprises a synthetic material matrix type, and the synthetic material matrix type, the synthetic raw material information and the synthetic control parameters are input into the structure prediction model for training to obtain the unsaturated polyester resin prediction structure.
Specifically, a federal domain architecture is designed, which includes integration of participants, data sharing mechanisms, and blockchain technologies, wherein the participants may be manufacturers, research institutions, suppliers, and the like, related entities with unsaturated polyester resins. In the federal domain, determining the roles and permissions of each participant in the federal domain, and establishing a blockchain network for storing and verifying the unsaturated polyester resin-generated record data, blockchain techniques can ensure the reliability, decentralization, and protection against data tampering. The participants generate unsaturated polyester resins in their respective business environments and upload the generated record data to a blockchain network in the federal domain, including synthesis process parameters, reaction results, performance tests, and the like.
The general matrix type of unsaturated polyester resin is obtained by referring to the related scientific literature, patent database or industry standard, etc.
The obtained unsaturated polyester resin-produced record data are arranged into a format suitable for cluster analysis, so that the data contain enough characteristics, such as synthesis process parameters, performance test results and the like, and are subjected to appropriate data cleaning. The generated record data is clustered by adopting a K-means clustering algorithm, which is a common unsupervised learning algorithm for dividing the data set into K different clusters, and the algorithm realizes the clustering of the data set by distributing each data point to the cluster center nearest to the data point and iteratively updating the position of the cluster center. The algorithm divides the data into different clusters according to the characteristics and similarity measurement of the data, and clustering clusters such as a first record data set, a second record data set, a third record data set, a fourth record data set and the like are obtained according to the data distribution and the characteristics in the clustering result.
The subset of data from the first through fifth recorded data sets is sorted, which data includes features and labels related to the target variable, i.e. the structure to be predicted. The network structure of the structure prediction model is built based on a BP neural network, a data subset is divided into a training set and a verification set, exemplary, 70% of data is used for training the model, the remaining 30% is used for verifying the performance of the model, the training set is used for training the structure prediction model, model parameters are adjusted through optimizing an objective function so that the model parameters can be best fit with training data, the verification set is used for evaluating the performance of the structure prediction model obtained through training, when the target requirement is met, for example, the accuracy reaches 95%, the structure prediction model is output, and the model can predict the values of target variables based on the data subsets of the first record data set, the fifth record data set and the like and is used for carrying out structure prediction on the obtained data and providing support for further analysis and decision.
And taking the matrix type of the synthetic material, the synthetic raw material information, the synthetic control parameters and the like as input data, and selecting a corresponding model from the trained structure prediction models according to task requirements and data characteristics to predict the structure of the unsaturated polyester resin. The input data is input into a structure prediction model for prediction, the model predicts the structure of unsaturated polyester resin by learning the mode and the relevance of the previous training data, the unsaturated polyester resin prediction structure is obtained, the prediction structure can be a molecular structure, a crystal structure and the like, depending on the type of the researched material and the structural characteristics, and the prediction result is used for guiding the optimization of the synthesis process, the material design, the performance evaluation and the like in the actual scene.
Further, step S240 of the present application includes:
step S241: acquiring synthetic raw material record data, intra-cluster matrix type and molecular structure record data according to the first record data group or the second record data group or the third record data group or the fourth record data group or the fifth record data group;
step S242: constructing a training loss evaluation function:
wherein, loss 1 Characterizing a first Loss function, loss 2 Characterizing a second loss function, count () characterizing a Count function, d (A i ,A i0 ) Characterizing the distance deviation of the predicted position and the recorded position of the ith functional group, A i Characterization of the predicted position of the ith functional group, A i0 Characterization of the i-th functional group recording position, f (A k ,A k0 ) Characterization of the type of functional group at the kth position and the type of native functional group at the kth predicted positionIs the consistency comparison result of A k For the type of functional group predicted at the kth position, A k0 The functional group type of the k position record is characterized, the output is 1 when the functional group is consistent, the output is 0 when the functional group is inconsistent, and d 0 Default value = 0, d 0 More than or equal to 0, T is a loss evaluation period, and j is a loss evaluation period starting count value;
step S243: taking the synthetic raw material record data and the intra-cluster matrix type as input data, taking the molecular structure record data as output data, and constructing a first structure base predictor, a second structure base predictor, a third structure base predictor, a fourth structure base predictor and a fifth structure base predictor based on a BP neural network;
step S244: when the loss evaluation period is satisfied, processing based on the training loss evaluation function according to output prediction results of the first structure base predictor or the second structure base predictor or the third structure base predictor or the fourth structure base predictor or the fifth structure base predictor, and obtaining a first training loss and a second training loss;
Step S245: and when the first training loss is smaller than a first preset loss amount and the second training loss is smaller than a second preset loss amount, the first training loss is regarded as convergence of the first structural base predictor or the second structural base predictor or the third structural base predictor or the fourth structural base predictor or the fifth structural base predictor, and the structural prediction model is obtained.
Specifically, one data set is selected from the first to fifth record data sets as an input, and synthetic raw material record data in the data set is extracted according to the selected data set, including material information involved in the synthesis process, such as reactant names, concentrations, proportions, and the like; extracting matrix type information in clusters in the data set, namely a specific matrix used in the synthesis process, wherein the type plays an important role in subsequent structure prediction; molecular structure record data in the data set are extracted, and the data describe information about molecular structures or chemical bonds generated in the synthesis process, such as atom types, bond lengths, angles, dihedral angles and the like.
The training loss evaluation function is constructed, and the function is used for measuring the performance and the error of the structure prediction model in the training process. Wherein, the Functional group refers to a structural unit or group with certain specific chemical property and reactivity in the organic compound, is a key part with similar property and function in the compound, and plays an important role in chemical reaction.
The synthetic raw material record data and the matrix type in the cluster are taken as input data, the input data are appropriately processed and encoded so as to be used for training of the neural network, the molecular structure record data are taken as output data, and the output data are appropriately processed and encoded so as to be matched with the output of the neural network. The structure of the task is designed based on the architecture of the BP neural network, wherein the structure comprises an input layer, a hidden layer and an output layer, the quantity of neurons of each layer is determined according to the complexity of the task and the scale of a data set, and first to fifth structural base predictors are obtained, and can accurately predict corresponding molecular structure record data, namely information of specific structural bases according to input synthetic raw material record data and matrix types in clusters.
Judging whether the condition for carrying out loss evaluation is satisfied or not according to the set period, wherein the period can be a fixed period determined according to experimental requirements or a dynamic period automatically adjusted according to training progress of a model. When the model is satisfied, one of the first to fifth structure-based predictors is selected as a predictor currently used, a prediction result is obtained, the prediction result is input into the training loss evaluation function, the obtained prediction result is compared with standard data or real data, a loss value of the model is calculated by using the training loss evaluation function, and a first training loss and a second training loss are obtained and used for recording performance evaluation of the model in the training process.
The first preset loss amount is a preset value for the first training loss, the second preset loss amount is a preset value for the second training loss, the preset loss amounts are determined according to actual application scenes or experience and represent expected or required performances of the model, the accuracy or the accuracy of model convergence is controlled through different threshold settings, smaller preset loss amounts can lead to higher model requirements, and larger preset loss amounts have lower performance requirements on the model performances.
Checking whether the first training loss is smaller than a preset first loss amount, checking whether the second training loss is smaller than a preset second loss amount, if the first and second training losses are smaller than the corresponding preset loss amounts, judging that the selected structure-based predictor is converged, and storing and using the predictor as a structure prediction model, wherein the prediction model can accurately predict corresponding molecular structure information through the training and optimizing process.
Step S300: based on a k neighbor clustering algorithm, carrying out clustering analysis on the unsaturated polyester resin prediction structure to obtain a performance prediction result;
Specifically, in the k-nearest neighbor clustering algorithm, k is a designated number of neighbors, and is used for dividing samples, an optimal k value is determined through a method such as cross-validation, which affects the quality of a clustering result, and then a similarity measurement method, such as cosine similarity, is used for calculating the similarity between each sample, so as to determine whether the samples belong to the same category. The samples are divided into different clusters according to the calculated similarity matrix using a k-nearest neighbor clustering algorithm that divides each sample into its corresponding clusters according to k nearest neighbor samples around it.
Further, as shown in fig. 3, step S300 of the present application includes:
step S310: transmitting the unsaturated polyester resin prediction structure to a federal domain, setting a performance analysis task, and obtaining a performance retrieval result;
step S320: obtaining molecular structure record data of the performance retrieval result, and processing the molecular structure record data and the unsaturated polyester resin predicted structure based on a structure distance evaluation function to obtain a structure distance evaluation result;
step S330: and based on the structural distance evaluation result, Q performance retrieval results are extracted from near to far to be fused, and the performance prediction result is obtained, wherein Q is more than or equal to 5 and is a self-defined parameter.
Specifically, data comprising the predicted unsaturated polyester resin structure is transmitted to the federal domain, the data is uploaded to a cloud server or provided to other participants to share the data in a federal learning or federal computing environment, performance analysis tasks are set for the transmitted predicted unsaturated polyester resin structure, and in the federal domain, there are multiple data sources or model providers that can perform performance calculations and model training using the respective data and models, perform performance analysis using the predicted unsaturated polyester resin structure, and generate corresponding performance search results. And after performance calculation and model training, the performance search results of all the providers are polymerized to obtain a comprehensive result which represents the predicted structural performance of the unsaturated polyester resin.
Specifically, the structural distance evaluation function is:
D(a,b)＝|L a -L b | α +{Count[f(A k ,A k0 )＝0]} β ，
wherein D (a, b) represents the distance between the molecular structure of the performance search result and the predicted structure of the unsaturated polyester resin, L a Characterizing the length, L, of the predicted structure of the unsaturated polyester resin b Characterization of the length of the molecular Structure of the Performance search results, f (A k ,A k0 ) Characterization of the results of the identity comparison of the functional group type at the kth position and the original functional group type at the kth predicted position, A k For the type of functional group predicted at the kth position, A k0 The functional group type recorded at the kth position is characterized, the output is 1 when the functional group is consistent, the output is 0 when the functional group is inconsistent, and alpha and beta are normalized adjustment parameters.
And extracting corresponding molecular structure record data from the performance search result, wherein the data contains information about atoms, bond connection, environment and the like of molecules so as to describe the molecular structure of the unsaturated polyester resin. And comparing the obtained molecular structure record data of the unsaturated polyester resin predicted structure and the performance retrieval result, and calculating the structural distance between the unsaturated polyester resin predicted structure and the performance retrieval result based on the structural distance evaluation function to obtain a structural distance evaluation result, wherein the structural distance evaluation result represents the structural difference or the structural similarity between the unsaturated polyester resin predicted structure and the performance retrieval result.
According to the obtained structure distance evaluation results, sorting the performance retrieval results according to the order from large to small, selecting the first Q results from the sorted performance retrieval result list to be fused, wherein Q is a self-defined parameter which needs to be set according to specific conditions and is required to meet Q not less than 5, and fusing the selected Q performance retrieval results, for example, carrying out weighted average on the Q performance retrieval results to obtain a final performance prediction result, so that the result is more reliable.
Further, the step S300 of the present application further includes:
step S330-1: performing hierarchical clustering analysis on Q performance search results according to preset performance deviation to obtain search result clustering data, wherein the search result clustering data comprise intra-class support, and the intra-class support represents the ratio of the aggregation quantity in the class to Q;
step S330-2: and eliminating the performance search results with the support degree in the class smaller than a support degree threshold value, obtaining P performance search results, carrying out mean analysis, and obtaining the performance prediction result.
Specifically, a preset performance deviation value is defined according to actual requirements, the value represents a tolerance range for performance difference, for Q performance retrieval results, similarity between performance indexes is calculated through Euclidean distance, a Q×Q similarity matrix is constructed based on calculation results of the performance similarity, each element in the matrix represents a similarity value between two performance retrieval results, clustering analysis is carried out on the similarity matrix by using a hierarchical clustering algorithm, hierarchical clustering is divided into two strategies of bottom-up and top-down, and how samples are clustered into clusters is determined through different combining rules, wherein the common combining rules comprise minimum distance, maximum distance, average distance and the like.
In the clustering process, the internal aggregation degree of each cluster is evaluated by calculating the intra-class support degree of each cluster, wherein the intra-class support degree represents the ratio of the intra-class aggregation quantity to Q and is used for representing the reliability and stability of the cluster, and the specific calculation method is to count the number of samples belonging to the same cluster and divide the number by the total number of samples Q.
A support threshold is set, which is a criterion for determining the division points of the support level within the class, and clusters with support levels below this threshold within all classes are considered unreliable. According to the set support degree threshold value, cluster clusters with support degree smaller than the threshold value in the class are removed from the cluster data of the search results, and meanwhile, performance search results associated with the clusters are removed together, so that data which are not reliable and stable enough in statistics can be removed. After eliminating the performance search results with low support degree, a group of reliable performance search results are obtained, P results (P is less than or equal to Q) in the performance search results are selected to carry out subsequent average analysis, and P can be selected according to actual requirements, can be fixed in number or can be dynamically adjusted according to specific conditions.
And (3) performing average analysis operation on the selected P performance retrieval results, namely calculating the average value of the results to obtain a final performance prediction result, comprehensively utilizing the selected performance retrieval result through average analysis, and obtaining a prediction performance value with more representation and accuracy.
Step S400: judging whether the performance prediction result meets the expected performance threshold of the unsaturated polyester resin;
specifically, the desired performance threshold of the unsaturated polyester resin is well defined, which is a reference value, for determining whether the performance prediction results reach the desired level, depending on the specific application needs and requirements. Comparing the obtained performance prediction result with an expected performance threshold, and if the result meets or exceeds the expected performance threshold, judging that the performance prediction result meets the requirement, and considering that the performance prediction result meets the performance standard expected by the unsaturated polyester resin; conversely, if the desired performance threshold is not met or is below, then it is determined that the requirement is not met, and further optimization or adjustment of the material or process parameters may be required to achieve the desired performance requirement.
Step S500: if not, optimizing the synthesis raw material information and/or the synthesis control parameters to obtain a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the expected performance threshold of the unsaturated polyester resin;
further, step S500 of the present application includes:
step S510: setting synthesis raw material type constraint information, synthesis raw material proportion constraint information and synthesis control parameter constraint information;
Step S520: based on the synthesis raw material type constraint information, the synthesis raw material proportion constraint information and the synthesis control parameter constraint information, the synthesis raw material information and/or the synthesis control parameter are adjusted to obtain a synthesis raw material adjustment result and/or a synthesis control parameter adjustment result;
step S530: performing performance analysis based on the synthetic raw material adjustment result and/or the synthetic control parameter adjustment result to obtain an adjustment process performance prediction result;
step S540: when the predicted result of the process performance adjustment meets the expected performance threshold value of the unsaturated polyester resin, setting the predicted result of the process performance adjustment as the optimized result of the synthesis process;
step S550: when the predicted result of the process performance adjustment does not meet the expected performance threshold value of the unsaturated polyester resin, the predicted result of the process performance adjustment is added into an elimination data set, and optimization is repeated;
step S560: and when a preset optimization algebra is met and a synthesis process meeting the expected performance threshold of the unsaturated polyester resin still does not appear, carrying out optimal value screening on the eliminated data set, and setting the optimal value as the synthesis process optimization result.
Specifically, when the performance prediction result does not satisfy the desired performance threshold of the unsaturated polyester resin, the synthesis raw material information and/or the synthesis control parameters are optimized to satisfy the desired performance threshold of the unsaturated polyester resin.
Specifically, the kind of synthetic raw material that can be used is determined, and for example, it may be limited to a specific polyester monomer or prepolymer, or a specific functional group that can be employed, etc.; setting the proportion relation between the synthetic raw materials, limiting the mixing proportion of different chemical substances according to the characteristics of the required polymerization degree, crosslinking degree and the like, wherein the constraint information can ensure that the synthetic reaction reaches balance in proportion and the required performance is obtained; the parameters controlling the synthesis process are defined in ranges including reaction temperature, reaction time, catalyst concentration, etc., and are varied within safe and effective ranges by setting appropriate constraints to ensure final product quality and performance. By setting the synthesis raw material type constraint information, the synthesis raw material proportion constraint information and the synthesis control parameter constraint information, the selectable variable range in the system can be determined, and reasonable parameter selection guidance can be provided.
Determining available synthetic raw material types according to the synthetic raw material type constraint information, and adjusting the use proportion of different raw materials according to the synthetic raw material proportion constraint information, wherein the use proportion can be realized by changing a raw material formula or replacing certain raw materials, and the adjusted synthetic raw material information comprises the selected raw material types and the corresponding proportions; based on the set constraint information of the synthesis control parameters, the control parameters in the synthesis process are adjusted, including modification of parameters such as temperature, pressure, reaction time, catalyst concentration and the like, and the adjusted synthesis control parameters determine the conditions and operation modes of the synthesis process. By adjusting the synthesis raw material information and the synthesis control parameters, the synthesis scheme can be optimized on the premise of meeting constraint information, so that the optimal raw material combination and control parameter setting can be found, and the required polyester resin performance can be realized.
And performing performance analysis on the synthetic raw material adjustment result and/or the synthetic control parameter adjustment result by adopting the same method to obtain an adjustment process performance prediction result.
If the performance prediction of the tuning process is within acceptable performance thresholds, which means that the process has reached the desired performance requirements, the process may continue to advance to subsequent steps and set the performance prediction as an optimized result of the synthesis process.
Adding the adjusted process performance prediction results that do not meet the desired performance threshold to the obsolete data set, which is used for subsequent analysis and decision, and re-optimizing the process based on the unsatisfied performance prediction results in the obsolete data set, including adjusting the synthetic raw material ratio, synthetic control parameters, or other related variables, to attempt to achieve results that more meet the desired performance requirements. And according to the re-optimized process adjustment result, performing performance analysis to obtain a new performance prediction result, comparing the new performance prediction result with an expected performance threshold value, and judging whether the requirement is met. If the new performance prediction result does not still meet the desired performance threshold, the process is repeated for iterative optimization until the desired performance requirement is met or a certain number of optimizations or optimization time limit is reached.
Obtaining a preset optimization algebra, for example, setting 500 times, when the optimization algebra reaches 500 times and a synthesis process meeting the expected performance threshold of the unsaturated polyester resin is still not found, carrying out optimal value screening on the obsolete data set, specifically, carrying out performance evaluation and analysis on the unsatisfied performance prediction result recorded in the obsolete data set to determine the actual performance of each adjustment process, searching the adjustment process with the performance closest to the expected performance requirement in the obsolete data set by comparing the difference between the performance index and the expected performance threshold, and setting the screened adjustment process with the optimal performance as the final optimization result of the synthesis process, which means that whether the expected performance threshold is reached or not, the optimization result is accepted and applied to actual production.
Through optimal value screening, after the preset optimization iteration times are reached, an optimal synthesis process scheme is ensured to be found, and even if the scheme fails to completely meet the expected performance threshold, the scheme still has the performance closest to the expected performance requirement, so that the optimal synthesis process scheme can be selected for practical application under the condition that the optimization cannot be continued.
Step S600: and controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
Specifically, according to the synthesis process optimization results, the required synthesis conditions including synthesis temperature, reaction time, raw material ratio, etc. are determined, ensuring that experimental parameters are set near the optimal values so that the synthesis process can achieve the optimal performance. According to the raw material proportion appointed in the synthesis process optimization result, accurately feeding each raw material into the reactor according to the corresponding proportion, and ensuring that the quantity and sequence of the fed raw materials are consistent with the optimization result. After the synthesis process reaches the preset reaction time, stopping the reaction and preparing to collect the synthesized product to obtain the final unsaturated polyester resin product.
In summary, the method and the system for optimizing the synthesis process of the unsaturated polyester resin provided by the embodiment of the application have the following technical effects:
obtaining synthesis basic information of unsaturated polyester resin, including synthesis raw material information and synthesis control parameters, inputting a structure prediction model, outputting an unsaturated polyester resin prediction structure, performing cluster analysis on the unsaturated polyester resin prediction structure, obtaining a performance prediction result, judging whether a desired performance threshold is met, if not, optimizing, and obtaining a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the desired performance threshold of the unsaturated polyester resin, and controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
The method solves the technical problems that the existing synthesis process cannot accurately control synthesis conditions and accurately monitor key parameters, and cannot adaptively adjust the synthesis process, so that the reaction is unstable, the reaction rate is low, and the intelligent degree is low, realizes the accurate control of synthesis conditions and monitoring of key parameters, and adaptively adjusts the synthesis process, thereby achieving the technical effects of improving the intelligent degree, the synthesis efficiency and the product quality.
Example two
Based on the same inventive concept as the synthetic process optimization method for unsaturated polyester resin in the foregoing embodiments, as shown in fig. 4, the present application provides a synthetic process optimization system for unsaturated polyester resin, the system comprising:
a basic information acquisition module 10, wherein the basic information acquisition module 10 is configured to acquire synthesis basic information of unsaturated polyester resin, and the synthesis basic information includes synthesis raw material information and synthesis control parameters;
the predicted structure output module 20 is used for inputting the synthesis raw material information and the synthesis control parameters into a structure prediction model and outputting an unsaturated polyester resin predicted structure;
The cluster analysis module 30 is used for carrying out cluster analysis on the unsaturated polyester resin prediction structure based on a k-nearest neighbor cluster algorithm to obtain a performance prediction result;
a predicted result judging module 40, wherein the predicted result judging module 40 is used for judging whether the performance predicted result meets the expected performance threshold value of the unsaturated polyester resin;
the control parameter optimization module 50 is configured to optimize the synthesis raw material information and/or the synthesis control parameter if the performance prediction result of the synthesis process optimization result meets the desired performance threshold of the unsaturated polyester resin, so as to obtain a synthesis process optimization result;
and a synthesis control module 60, wherein the synthesis control module 60 is used for controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
Further, the system further comprises:
the record data acquisition module is used for acquiring unsaturated polyester resin based on a federal domain to generate record data, wherein the federal domain is a service information sharing platform constructed by combining with a block chain;
a type acquisition module for acquiring an unsaturated polyester resin matrix type, wherein the unsaturated polyester resin matrix type includes phthalic acid type, isophthalic acid type, bisphenol a type, vinyl ester type, and halogenated unsaturated polyester resin;
A clustering module, configured to cluster recording data generated by the unsaturated polyester resin based on the phthalic acid type, the isophthalic acid type, the bisphenol a type, the vinyl ester type and the halogenated unsaturated polyester resin, and obtain a first recording data set, a second recording data set, a third recording data set, a fourth recording data set and a fifth recording data set;
a model training module for training the structure prediction model by the first record data set, the second record data set, the third record data set, the fourth record data set, and the fifth record data set;
and the predicted structure acquisition module is used for inputting the synthetic material matrix type, the synthetic material information and the synthetic control parameters into the structure prediction model for training to acquire the unsaturated polyester resin predicted structure.
Further, the system further comprises:
the molecular structure record data acquisition module is used for acquiring synthetic raw material record data, intra-cluster matrix types and molecular structure record data according to the first record data group or the second record data group or the third record data group or the fourth record data group or the fifth record data group;
The loss evaluation function construction module is used for constructing a training loss evaluation function:
wherein, loss 1 Characterizing a first Loss function, loss 2 Characterizing a second loss function, count () characterizing a Count function, d (A i ,A i0 ) Characterizing the distance deviation of the predicted position and the recorded position of the ith functional group, A i Characterization of the predicted position of the ith functional group, A i0 Characterization of the i-th functional group recording position, f (A k ,A k0 ) Characterization of the results of the identity comparison of the functional group type at the kth position and the original functional group type at the kth predicted position, A k For the type of functional group predicted at the kth position, A k0 The functional group type of the k position record is characterized, the output is 1 when the functional group is consistent, the output is 0 when the functional group is inconsistent, and d 0 Default value = 0, d 0 More than or equal to 0, T is a loss evaluation period, and j is a loss evaluation period starting count value;
the predictor construction module is used for taking the synthetic raw material record data and the intra-cluster matrix type as input data, taking the molecular structure record data as output data and constructing a first structure base predictor, a second structure base predictor, a third structure base predictor, a fourth structure base predictor and a fifth structure base predictor based on a BP neural network;
the training loss acquisition module is used for processing according to the output prediction result of the first structure base predictor or the second structure base predictor or the third structure base predictor or the fourth structure base predictor or the fifth structure base predictor and based on the training loss evaluation function when the loss evaluation period is met, so as to acquire a first training loss and a second training loss;
And the structure prediction model acquisition module is used for considering the first structure base predictor or the second structure base predictor or the third structure base predictor or the fourth structure base predictor or the fifth structure base predictor to converge when the first training loss is smaller than a first preset loss amount and the second training loss is smaller than a second preset loss amount, and acquiring the structure prediction model.
Further, the system further comprises:
the performance retrieval result acquisition module is used for transmitting the unsaturated polyester resin prediction structure to a federal domain, setting a performance analysis task and acquiring a performance retrieval result;
the evaluation result acquisition module is used for acquiring the molecular structure record data of the performance retrieval result, predicting the structure with the unsaturated polyester resin, processing the molecular structure record data based on a structure distance evaluation function, and acquiring a structure distance evaluation result;
and the performance prediction result acquisition module is used for extracting Q performance retrieval results from near to far based on the structure distance evaluation result to fuse the performance retrieval results, and acquiring the performance prediction result, wherein Q is more than or equal to 5 and is a self-defined parameter.
Further, the structural distance evaluation function is:
D(a,b)＝|L a -L b | α +{Count[f(A k ,A k0 )＝0]} β ，
Wherein D (a, b) represents the distance between the molecular structure of the performance search result and the predicted structure of the unsaturated polyester resin, L a Characterizing the length, L, of the predicted structure of the unsaturated polyester resin b Characterization of the length of the molecular Structure of the Performance search results, f (A k ,A k0 ) Characterization of the results of the identity comparison of the functional group type at the kth position and the original functional group type at the kth predicted position, A k For the type of functional group predicted at the kth position, A k0 The functional group type recorded at the kth position is characterized, the output is 1 when the functional group is consistent, the output is 0 when the functional group is inconsistent, and alpha and beta are normalized adjustment parameters.
Further, the system further comprises:
the hierarchical clustering analysis module is used for performing hierarchical clustering analysis on Q performance search results according to preset performance deviation to obtain search result clustering data, wherein the search result clustering data comprises intra-class support, and the intra-class support represents the ratio of the number of intra-class aggregation to Q;
and the rejecting module is used for rejecting the performance retrieval results with the support degree in the class smaller than a support degree threshold value, acquiring P performance retrieval results, carrying out mean analysis, and acquiring the performance prediction results.
Further, the system further comprises:
The constraint information setting module is used for setting constraint information of synthetic raw material types, constraint information of synthetic raw material proportions and constraint information of synthetic control parameters;
the control parameter adjustment module is used for adjusting the synthesis raw material information and/or the synthesis control parameters based on the synthesis raw material type constraint information, the synthesis raw material proportion constraint information and the synthesis control parameter constraint information to obtain a synthesis raw material adjustment result and/or a synthesis control parameter adjustment result;
the performance analysis module is used for performing performance analysis based on the synthetic raw material adjustment result and/or the synthetic control parameter adjustment result to obtain an adjustment process performance prediction result;
the process optimization result acquisition module is used for setting the process adjustment performance prediction result as the synthesis process optimization result when the process adjustment performance prediction result meets the expected performance threshold of the unsaturated polyester resin;
the repeated optimization module is used for adding the predicted result of the process performance adjustment into the elimination data set when the predicted result of the process performance adjustment does not meet the expected performance threshold value of the unsaturated polyester resin, and repeating optimization;
And the optimal value screening module is used for screening the optimal value of the eliminated data set and setting the eliminated data set as the synthesis process optimization result when the synthesis process meeting the expected performance threshold of the unsaturated polyester resin still does not appear in the preset optimization algebra.
From the foregoing detailed description of the method for optimizing the synthesis process for the unsaturated polyester resin, those skilled in the art can clearly understand the method and the system for optimizing the synthesis process for the unsaturated polyester resin in this embodiment, and for the apparatus disclosed in the embodiments, since the apparatus corresponds to the method disclosed in the embodiments, the description is relatively simple, and the relevant points refer to the description of the method section.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (8)
1. A method for optimizing a synthetic process for an unsaturated polyester resin, the method comprising:
obtaining synthesis basic information of unsaturated polyester resin, wherein the synthesis basic information comprises synthesis raw material information and synthesis control parameters;
inputting the synthesis raw material information and the synthesis control parameters into a structure prediction model, and outputting an unsaturated polyester resin prediction structure;
based on a k neighbor clustering algorithm, carrying out clustering analysis on the unsaturated polyester resin prediction structure to obtain a performance prediction result;
judging whether the performance prediction result meets the expected performance threshold of the unsaturated polyester resin;
if not, optimizing the synthesis raw material information and/or the synthesis control parameters to obtain a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the expected performance threshold of the unsaturated polyester resin;
and controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
2. The method of claim 1, wherein inputting the synthesis feedstock information and the synthesis control parameters into a structure prediction model, outputting an unsaturated polyester resin prediction structure, comprises:
Collecting unsaturated polyester resin based on a federal domain to generate record data, wherein the federal domain is a service information sharing platform constructed by combining with a blockchain;
obtaining an unsaturated polyester resin matrix type, wherein the unsaturated polyester resin matrix type comprises phthalic acid type, isophthalic acid type, bisphenol A type, vinyl ester type and halogenated unsaturated polyester resin;
clustering the unsaturated polyester resin generated record data based on the phthalic acid type, the isophthalic acid type, the bisphenol A type, the vinyl ester type and the halogenated unsaturated polyester resin to obtain a first record data group, a second record data group, a third record data group, a fourth record data group and a fifth record data group;
training the structure prediction model by the first record data set, the second record data set, the third record data set, the fourth record data set, and the fifth record data set;
the synthetic raw material information comprises a synthetic material matrix type, and the synthetic material matrix type, the synthetic raw material information and the synthetic control parameters are input into the structure prediction model for training to obtain the unsaturated polyester resin prediction structure.
3. The method of claim 2, wherein training the structure prediction model with the first record data set, the second record data set, the third record data set, the fourth record data set, and the fifth record data set comprises:
acquiring synthetic raw material record data, intra-cluster matrix type and molecular structure record data according to the first record data group or the second record data group or the third record data group or the fourth record data group or the fifth record data group;
constructing a training loss evaluation function:
wherein, loss 1 Characterizing a first Loss function, loss 2 Characterizing a second loss function, count () characterizing a Count function, d (A i ,A i0 ) Characterizing the distance deviation of the predicted position and the recorded position of the ith functional group, A i Characterization of the predicted position of the ith functional group, A i0 Characterization of the i-th functional group recording position, f (A k ,A k0 ) Characterization of the results of the identity comparison of the functional group type at the kth position and the original functional group type at the kth predicted position, A k For the type of functional group predicted at the kth position, A k0 The functional group type of the k position record is characterized, the output is 1 when the functional group is consistent, the output is 0 when the functional group is inconsistent, and d 0 Default value = 0, d 0 More than or equal to 0, T is a loss evaluation period, and j is a loss evaluation period starting count value;
taking the synthetic raw material record data and the intra-cluster matrix type as input data, taking the molecular structure record data as output data, and constructing a first structure base predictor, a second structure base predictor, a third structure base predictor, a fourth structure base predictor and a fifth structure base predictor based on a BP neural network;
when the loss evaluation period is satisfied, processing based on the training loss evaluation function according to output prediction results of the first structure base predictor or the second structure base predictor or the third structure base predictor or the fourth structure base predictor or the fifth structure base predictor, and obtaining a first training loss and a second training loss;
and when the first training loss is smaller than a first preset loss amount and the second training loss is smaller than a second preset loss amount, the first training loss is regarded as convergence of the first structural base predictor or the second structural base predictor or the third structural base predictor or the fourth structural base predictor or the fifth structural base predictor, and the structural prediction model is obtained.
4. The method of claim 1, wherein performing cluster analysis on the unsaturated polyester resin prediction structure based on a k-nearest neighbor clustering algorithm to obtain a performance prediction result comprises:
transmitting the unsaturated polyester resin prediction structure to a federal domain, setting a performance analysis task, and obtaining a performance retrieval result;
obtaining molecular structure record data of the performance retrieval result, and processing the molecular structure record data and the unsaturated polyester resin predicted structure based on a structure distance evaluation function to obtain a structure distance evaluation result;
and based on the structural distance evaluation result, Q performance retrieval results are extracted from near to far to be fused, and the performance prediction result is obtained, wherein Q is more than or equal to 5 and is a self-defined parameter.
5. The method of claim 4, wherein the structural distance evaluation function is:
D(a,b)＝|L a -L b | α +{Count[f(A k ,A k0 )＝0]} β ，
wherein D (a, b) represents the distance between the molecular structure of the performance search result and the predicted structure of the unsaturated polyester resin, L a Characterizing the length, L, of the predicted structure of the unsaturated polyester resin b Characterization of the length of the molecular Structure of the Performance search results, f (A k ,A k0 ) Characterization of the results of the identity comparison of the functional group type at the kth position and the original functional group type at the kth predicted position, A k For the type of functional group predicted at the kth position, A k0 The functional group type recorded at the kth position is characterized, the output is 1 when the functional group is consistent, the output is 0 when the functional group is inconsistent, and alpha and beta are normalized adjustment parameters.
6. The method of claim 4, wherein extracting Q performance search results from near to far for performance fusion based on the structural distance evaluation result, obtaining the performance prediction result, further comprises:
performing hierarchical clustering analysis on Q performance search results according to preset performance deviation to obtain search result clustering data, wherein the search result clustering data comprise intra-class support, and the intra-class support represents the ratio of the aggregation quantity in the class to Q;
and eliminating the performance search results with the support degree in the class smaller than a support degree threshold value, obtaining P performance search results, carrying out mean analysis, and obtaining the performance prediction result.
7. The method of claim 1, wherein optimizing the synthesis feedstock information and/or the synthesis control parameters to obtain synthesis process optimization results comprises:
setting synthesis raw material type constraint information, synthesis raw material proportion constraint information and synthesis control parameter constraint information;
Based on the synthesis raw material type constraint information, the synthesis raw material proportion constraint information and the synthesis control parameter constraint information, the synthesis raw material information and/or the synthesis control parameter are adjusted to obtain a synthesis raw material adjustment result and/or a synthesis control parameter adjustment result;
performing performance analysis based on the synthetic raw material adjustment result and/or the synthetic control parameter adjustment result to obtain an adjustment process performance prediction result;
when the predicted result of the process performance adjustment meets the expected performance threshold value of the unsaturated polyester resin, setting the predicted result of the process performance adjustment as the optimized result of the synthesis process;
when the predicted result of the process performance adjustment does not meet the expected performance threshold value of the unsaturated polyester resin, the predicted result of the process performance adjustment is added into an elimination data set, and optimization is repeated;
and when a preset optimization algebra is met and a synthesis process meeting the expected performance threshold of the unsaturated polyester resin still does not appear, carrying out optimal value screening on the eliminated data set, and setting the optimal value as the synthesis process optimization result.
8. A synthetic process optimization system for unsaturated polyester resins, characterized by being adapted to implement the synthetic process optimization method for unsaturated polyester resins according to any one of claims 1 to 7, comprising:
The base information acquisition module is used for acquiring synthesis base information of the unsaturated polyester resin, wherein the synthesis base information comprises synthesis raw material information and synthesis control parameters;
the prediction structure output module is used for inputting the synthetic raw material information and the synthetic control parameters into a structure prediction model and outputting an unsaturated polyester resin prediction structure;
the cluster analysis module is used for carrying out cluster analysis on the unsaturated polyester resin prediction structure based on a k-nearest neighbor clustering algorithm to obtain a performance prediction result;
the prediction result judging module is used for judging whether the performance prediction result meets the expected performance threshold of the unsaturated polyester resin;
the control parameter optimization module is used for optimizing the synthesis raw material information and/or the synthesis control parameters if the synthesis raw material information and/or the synthesis control parameters are not met, and obtaining a synthesis process optimization result, wherein the performance prediction result of the synthesis process optimization result meets the expected performance threshold of the unsaturated polyester resin;
and the synthesis control module is used for controlling the synthesis of the unsaturated polyester resin according to the synthesis process optimization result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310763583.2A CN116913423A (en) | 2023-06-27 | 2023-06-27 | Synthetic process optimization method and system for unsaturated polyester resin |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310763583.2A CN116913423A (en) | 2023-06-27 | 2023-06-27 | Synthetic process optimization method and system for unsaturated polyester resin |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116913423A true CN116913423A (en) | 2023-10-20 |
Family
ID=88359264
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310763583.2A Pending CN116913423A (en) | 2023-06-27 | 2023-06-27 | Synthetic process optimization method and system for unsaturated polyester resin |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116913423A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113065143A (en) * | 2021-03-17 | 2021-07-02 | 四川大学 | Block chain based secure sharing of industrial data |
| CN113349267A (en) * | 2021-04-09 | 2021-09-07 | 三江侗族自治县仙池茶业有限公司 | Method and system for intelligently adjusting environment of black tea fermentation chamber |
| CN114255826A (en) * | 2021-12-08 | 2022-03-29 | 清华大学 | Electrolyte design method, device, equipment, medium and program product |
| US20230073816A1 (en) * | 2020-06-12 | 2023-03-09 | Petrochina Company Limited | Optimization Method and System for Whole Process of Molecular-level Oil Refinery Processing and Storage Medium |
| CN115964663A (en) * | 2022-09-16 | 2023-04-14 | 天津理工大学 | Cluster-based federal domain adaptation method |
-
2023
- 2023-06-27 CN CN202310763583.2A patent/CN116913423A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230073816A1 (en) * | 2020-06-12 | 2023-03-09 | Petrochina Company Limited | Optimization Method and System for Whole Process of Molecular-level Oil Refinery Processing and Storage Medium |
| CN113065143A (en) * | 2021-03-17 | 2021-07-02 | 四川大学 | Block chain based secure sharing of industrial data |
| CN113349267A (en) * | 2021-04-09 | 2021-09-07 | 三江侗族自治县仙池茶业有限公司 | Method and system for intelligently adjusting environment of black tea fermentation chamber |
| CN114255826A (en) * | 2021-12-08 | 2022-03-29 | 清华大学 | Electrolyte design method, device, equipment, medium and program product |
| CN115964663A (en) * | 2022-09-16 | 2023-04-14 | 天津理工大学 | Cluster-based federal domain adaptation method |
Non-Patent Citations (1)
| Title |
|---|
| "工艺参数优化研究", Retrieved from the Internet <URL:wenku.baidu.com> * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Evaluation of GO-based functional similarity measures using S. cerevisiae protein interaction and expression profile data | |
| CN107992976B (en) | Hot topic early development trend prediction system and prediction method | |
| CN113987033B (en) | Main transformer online monitoring data group deviation identification and calibration method | |
| CN108681742B (en) | Analysis method for analyzing sensitivity of driver driving behavior to vehicle energy consumption | |
| CN113609779A (en) | Modeling method, device and equipment for distributed machine learning | |
| AU2019403566A1 (en) | Method, apparatus, and computer-readable medium for efficiently optimizing a phenotype with a specialized prediction model | |
| CN112164427A (en) | Method and device for predicting activity of small drug molecule target based on deep learning | |
| CN115171807B (en) | Molecular coding model training method, molecular coding method and molecular coding system | |
| CN112634992A (en) | Molecular property prediction method, training method of model thereof, and related device and equipment | |
| CN113066525A (en) | Multi-target drug screening method based on ensemble learning and hybrid neural network | |
| CN114822698A (en) | Knowledge reasoning-based biological large sample data set analysis method and system | |
| CN116913423A (en) | Synthetic process optimization method and system for unsaturated polyester resin | |
| CN111666981A (en) | System data anomaly detection method based on genetic fuzzy clustering | |
| CN115312135B (en) | Chemical reaction condition prediction method, system, device and storage medium | |
| CN115661546A (en) | Multi-objective optimization classification method based on feature selection and classifier joint design | |
| CN113724197B (en) | Thread screwing property judging method based on meta-learning | |
| Zhu et al. | Developing robust nonlinear models through bootstrap aggregated deep belief networks | |
| CN117116370A (en) | Chemical reaction yield prediction method and electronic equipment | |
| KR102377545B1 (en) | Method of transfer learning using multi-layer ontology-based database platform | |
| US11862300B1 (en) | Formulation graph for machine learning of chemical products | |
| WO2023181936A1 (en) | Prediction device, prediction system, and prediction program | |
| Chin et al. | Optimized local protein structure with support vector machine to predict protein secondary structure | |
| Korobchynskyi et al. | Optimization of Data Preprocessing Procedure in the Systems of High Dimensional Data Clustering | |
| Lin et al. | Multiclass microarray data classification using GA/ANN method | |
| Li et al. | Research on Factor Importance Feature Recognition Algorithm Based on KPCA-GBDT-LightGBM Intelligent Coupling Algorithm |
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 |