Performance prediction method of waterborne polyurethane emulsion for glass fibers

Abstract

The invention relates to the technical field of organic matter preparation, in particular to a glass fiber water-based polyurethane emulsion performance prediction method which comprises the following steps: S1, constructing a glass fiber water-based polyurethane emulsion database; s2, constructing a water-based polyurethane emulsion multi-channel platform for the glass fibers; s3, constructing and training an intelligent model for predicting the performance of the water-based polyurethane emulsion for the glass fiber based on the database; performing targeted optimization training and upgrading iteration according to the performance index parameters of the waterborne polyurethane emulsion in the database to respectively obtain intelligent models for predicting different performance index parameters, and combining the intelligent models to form an intelligent model network; s4, predicting the performance index parameters of the product by using the intelligent model in the intelligent model network according to the formula and process of the waterborne polyurethane emulsion randomly generated by the multi-channel platform in the restricted domain; and if the prediction result is satisfied, entering a multi-channel platform for experimental verification. According to the method, the concerned performance parameters of glass fiber application can be quickly predicted, and the development efficiency of the special PUD is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of organic matter preparation, and in particular relates to a performance prediction method for water-based polyurethane emulsion for glass fibers. BACKGROUND

[0002] Water-based polyurethane (WPU) has many advantages and is widely used in leather processing, coatings, adhesives and other fields. However, the use of water-based polyurethane for surface treatment of glass fibers is a new field of application of WPU. When WPU is applied to glass fiber manufacturing, it not only affects the formation of glass fiber, but also affects the performance of downstream composite materials. Therefore, it is difficult, time-consuming and inefficient to successfully prepare and screen WPU suitable for glass fibers.

[0003] At present, technical personnel have constructed various methods to accelerate the experimental process based on the concept of high-throughput experiments, but these methods have certain defects or do not meet the requirements of PUD performance screening for glass fibers. Patent CN111128311A discloses a catalytic material screening method and system based on high-throughput experiments and calculations, which uses high-throughput methods to explore the structure-activity relationship of catalysts, thereby improving the screening accuracy and speed of catalytic materials. However, the structure-activity relationship studied in this patent is mainly for inorganic materials, and more physical structure logic is involved, which does not meet the prediction requirements of PUD type high molecular organic matter.

[0004] Patent CN117558380A discloses a high-throughput preparation method and system for magnetic micro-nano materials based on intelligent algorithms, forming an innovative development system for magnetic micro-nano materials. However, this patent obtains relevant data from internet big data, which is not suitable for the field of PUD for glass fibers with less public data. Meanwhile, in the implementation process, AI is difficult to judge the authenticity from opposite data, which can easily cause data distortion and lead to prediction failure.

[0005] Therefore, there is an urgent need for a performance prediction method for water-based polyurethane emulsion for glass fibers, which can quickly predict the emulsion particle size, emulsion viscosity, emulsion film mechanical properties, emulsion film heat yellowing resistance and other properties that are concerned in the application of glass fibers, thereby effectively improving the efficiency of special PUD development. SUMMARY

[0006] The purpose of the present application is to provide a performance prediction method for water-based polyurethane emulsion for glass fibers, which can quickly predict the emulsion particle size, emulsion viscosity, emulsion film mechanical properties, emulsion film heat yellowing resistance and other properties that are concerned in the application of glass fibers, thereby effectively improving the efficiency of special PUD development.

[0007] In order to achieve the above purpose, a performance prediction method for water-based polyurethane emulsion for glass fibers is provided, which comprises the following steps:

[0008] S1, constructing a database Cspu-Room of waterborne polyurethane emulsion for glass fiber; the database Cspu-Room includes common raw materials for preparing the polyurethane emulsion, basic physical and chemical property data of the common raw materials, and existing polyurethane emulsion formulations, processes and corresponding performance index data for glass fiber;

[0009] S2, constructing a multi-channel platform Cspu-Muler of waterborne polyurethane emulsion for glass fiber, which is used for random formulation synthesis and performance index evaluation of the waterborne polyurethane emulsion in a limited domain and random process synthesis and performance index evaluation of the waterborne polyurethane emulsion in the limited domain based on the database Cspu-Room; returning the polyurethane emulsion formulation, process and corresponding performance index evaluation data of the waterborne polyurethane emulsion for glass fiber verified by the multi-channel platform Cspu-Muler to the database Cspu-Room;

[0010] S3, constructing and training an intelligent model Cspu-Brain for predicting the performance of the waterborne polyurethane emulsion for glass fiber based on the database Cspu-Room; according to the performance index parameters A1-A n , respectively, and combining to form an intelligent model network Cspu-Brain(N); n

[0011] S4, using the intelligent model Cspu-Brain(A n ) in the intelligent model network Cspu-Brain(N) to predict the performance index parameters A n of the product; the formulation and process meeting the target demand of the prediction result enter the multi-channel platform Cspu-Muler for experimental verification.

[0012] Further, the performance related index parameters A1-A n of the waterborne polyurethane emulsion are characteristic performances of the waterborne polyurethane emulsion for glass fiber, including but not limited to emulsion particle size, emulsion viscosity, emulsion centrifugal sedimentation content, film strength, film elongation, film thermal yellowing and film adhesion.

[0013] Further, the step S3 specifically includes the following steps:

[0014] S301, constructing an initial version of the intelligent model Cspu-Brain for predicting the performance of the waterborne polyurethane emulsion for glass fiber based on the database Cspu-Room;

[0015] ​S302, using the glass fiber with water-based polyurethane emulsion database Cspu-Room in A n Performance-related data, feeding training to the intelligent model Cspu-Brain of the glass fiber with water-based polyurethane emulsion, and obtaining the initial version of the intelligent model Cspu-Brain(A n );

[0016] S303, repeating step S302, respectively obtaining intelligent models Cspu-Brain(A n );

[0017] S304, combining the intelligent models Cspu-Brain(A n ) to form an intelligent model network Cspu-Brain(N);

[0018] S305, according to the performance-related index parameters A n of the water-based polyurethane emulsion in the continuously updated database Cspu-Room, respectively optimizing and upgrading the intelligent models Cspu-Brain(A n ), and completing the self-iterative upgrade of the intelligent model network Cspu-Brain(N).

[0019] Further, the step S305 includes the following steps:

[0020] S3051, making the intelligent model Cspu-Brain(A1) read the experimental data related to the performance of A1 through the database Cspu-Room, and completing the self-iterative upgrade;

[0021] S3052, making the intelligent model Cspu-Brain(A2) read the experimental data related to the performance of A2 through the database Cspu-Room, and completing the self-iterative upgrade;

[0022] S3053, making the intelligent model Cspu-Brain(A3) read the experimental data related to the performance of A3 through the database Cspu-Room, and completing the self-iterative upgrade;

[0023] S3054, repeating the same operation process as above, completing the self-iterative upgrade of the intelligent models Cspu-Brain(A1) to Cspu-Brain(A n ), and realizing the iterative upgrade of the intelligent model network Cspu-Brain(N).

[0024] Further, the intelligent models Cspu-Brain(A n)In the process of continuous iteration and upgrading, and before the formation of the intelligent model network Cspu-Brain(N), set up each intelligent model Cspu-Brain(A n )The accuracy of a single performance model is greater than 75%, preferably greater than 80%.

[0025] Further, the formula and process group GN that meets the comprehensive performance of the waterborne polyurethane emulsion for glass fibers has a data entry number range of 0-n. If G N The data entry number is 0, and when multiple formulas and processes are randomly generated and screened, G N The data entry number is still 0, then generate a prompt, the prompt is: The probability of having a waterborne polyurethane emulsion formula and process that meets the comprehensive performance in the current limit domain is close to 0, the limit domain should be adjusted.

[0026] Further, in step S404, at the stage of obtaining the formula and process group G n )that meets the comprehensive performance(A1-A N )of the waterborne polyurethane emulsion for glass fibers, the order of the intelligent model Cspu-Brain(A1-A n )is adjusted according to the demand.

[0027] Further, the multi-channel platform Cspu-Muler is built by combining a multi-channel PUD synthesis kettle, an A1 performance test platform, an A2 performance test platform, and an A n performance test platform, which completes the whole chain experiment of PUD from raw materials to synthesis, and then to performance test.

[0028] Further, the writing and iteration method of Cspu-Brain(A n )includes but is not limited to machine learning and random forest algorithm.

[0029] Principle and advantages:

[0030] 1. This scheme establishes a prediction, verification and iteration method for waterborne polyurethane emulsion for glass fibers based on high-throughput experiment, which can quickly screen the formula and process and improve the efficiency of the development of PUD for glass fibers. The multi-channel platform Cspu-Muler can randomly generate and verify the formula, process and performance index evaluation, so the data volume can be continuously enriched through iteration, so that the database Cspu-Room can also be continuously improved and enriched. The database Cspu-Room, the multi-channel platform Cspu-Muler and the intelligent model Cspu-Brain(A n )are based on A nThe performance forms a three-dimensional coupling closed loop, the database Cspu-Room, the multi-channel platform Cspu-Muler and the intelligent model net Cspu-Brain(N) are based on the comprehensive performance (A1-A n ) of the glass fiber waterborne polyurethane emulsion to form a three-dimensional coupling closed loop.

[0031] 2, the scheme is through the intelligent model Cspu-Brain (A1), the intelligent model Cspu-Brain (A2), the intelligent model Cspu-Brain (A3) …… Cspu-Brain (A n ) and so on are organized into intelligent model net Cspu-Brain (N), compared with the scheme of single intelligent model screening comprehensive performance, the workload of intelligent model iteration is reduced, and the performance prediction accuracy is higher.

[0032] 3, the intelligent model Cspu-Brain (A1) of the scheme, the intelligent model Cspu-Brain (A2), the intelligent model Cspu-Brain (A3) …… Cspu-Brain (A n ) and so on select different data points of the same experiment to carry out independent iteration, which overcomes the problem that different performances are affected each other in single model iteration, so that the intelligent model net Cspu-Brain (N) formed by the scheme has higher precision.

[0033] 4, the scheme can give early warning to unreasonable waterborne polyurethane emulsion limit domain, and has guiding significance to the overall research and development direction. DETAILED DESCRIPTION

[0034] Figure 1 It is a flowchart of the performance prediction method of the glass fiber waterborne polyurethane emulsion of the embodiment of the application;

[0035] Figure 2 It is a schematic diagram of predicting, verifying and iterating the glass fiber waterborne polyurethane emulsion formula process in the performance prediction method of the glass fiber waterborne polyurethane emulsion of the embodiment of the application;

[0036] Figure 3 It is a precision verification data analysis schematic diagram of Cspu-Brain (particle size) intelligent model;

[0037] Figure 4 It is a precision verification data analysis schematic diagram of Cspu-Brain (tensile strength) intelligent model

[0038] Figure 5 It is a precision verification data analysis schematic diagram of Cspu-Brain (YI) intelligent model;

[0039] Figure 6The data table shows the required formulation, process and test results for the experimental data. DETAILED DESCRIPTION

[0040] The following is further described in detail by specific embodiments:

[0041] EMBODIMENT

[0042] A performance prediction method for water-based polyurethane emulsion for glass fibers, substantially as shown in the accompanying drawings, comprising the following steps: Figure 1

[0043] S1, constructing a water-based polyurethane emulsion database Cspu-Room for glass fibers; the database Cspu-Room includes the commonly used raw materials for preparing polyurethane emulsion and their physical and chemical basic property data, and the existing polyurethane emulsion formula, process and corresponding performance index data for glass fibers;

[0044] S2, constructing a multi-channel platform Cspu-Muler for water-based polyurethane emulsion for glass fibers, which is used for random formula synthesis and performance index evaluation of water-based polyurethane emulsion within the limit domain and random process synthesis and performance index evaluation of water-based polyurethane emulsion within the limit domain based on the database Cspu-Room; the water-based polyurethane emulsion formula, process and corresponding performance index evaluation data verified by the multi-channel platform Cspu-Muler for glass fibers are returned to the Cspu-Room database; the multi-channel platform Cspu-Muler is composed of a multi-channel PUD synthesis kettle, an A1 performance test platform, an A2 performance test platform and an A3 performance test platform, which are combined to complete the whole chain experiment from raw materials to synthesis, and then to performance test of PUD. n

[0045] S3, constructing and training an intelligent model Cspu-Brain for predicting the performance of water-based polyurethane emulsion for glass fibers based on the database Cspu-Room; according to the performance index parameters A1-A n , the intelligent model Cspu-Brain is optimized and iterated, respectively obtaining intelligent models Cspu-Brain(A1-A n ), and combining to form an intelligent model network Cspu-Brain(N);

[0046] The step S3 specifically includes the following steps:

[0047] ​​S301, constructing an initial version of an intelligent model Cspu-Brain for predicting the performance of a waterborne polyurethane emulsion for glass fibers based on a database Cspu-Room; the technologies used in the writing and iteration methods of the intelligent model Cspu-Brain include but are not limited to Python, machine learning, random forest algorithm, and data cleaning method.

[0048] S302, using the A n performance-related data to feed and train the intelligent model Cspu-Brain for waterborne polyurethane emulsion for glass fibers, and obtain an initial version of the intelligent model Cspu-Brain(A n ); the performance-related index parameters A1-A n of the waterborne polyurethane emulsion are characteristic performances of the waterborne polyurethane emulsion for glass fibers, including but not limited to emulsion particle size, emulsion viscosity, emulsion centrifugal sediment content, film strength, film elongation, film thermal yellowing, and film adhesion.

[0049] S303, repeating step S302 to obtain intelligent models Cspu-Brain(A1-A n );

[0050] S304, combining the intelligent models Cspu-Brain(A1-A n ) to form an intelligent model network Cspu-Brain(N);

[0051] S305, according to the performance-related index parameters A n of the waterborne polyurethane emulsion in the continuously updated database Cspu-Room, respectively optimizing and upgrading the intelligent models Cspu-Brain(A n ), and completing the self-iteration upgrade of the intelligent model network Cspu-Brain(N). The step S305 includes the following steps:

[0052] S3051, making the intelligent model Cspu-Brain(A1) read the experimental data related to the performance of A1 through the database Cspu-Room, and complete the self-iteration upgrade;

[0053] S3052, making the intelligent model Cspu-Brain(A2) read the experimental data related to the performance of A2 through the database Cspu-Room, and complete the self-iteration upgrade;

[0054] S3053, making the intelligent model Cspu-Brain(A3) read the experimental data related to the performance of A3 through the database Cspu-Room, and complete the self-iteration upgrade;

[0055] S3054, repeat the same operation process as above to complete the intelligent model Cspu-Brain (A1) to the intelligent model Cspu-Brain (A n ) to realize the iterative upgrade of the intelligent model network Cspu-Brain(N).

[0056] The intelligent model Cspu-Brain (A n ) After continuous iteration and upgrading, and before forming the intelligent model network Cspu-Brain (N), set each intelligent model Cspu-Brain (A n ) The accuracy of the single performance model is greater than 75%, preferably greater than 80%.

[0057] S4. The formula and process of randomly synthesized waterborne polyurethane emulsion in the restricted domain are based on the intelligent model Cspu-Brain (A) in the intelligent model network Cspu-Brain (N). n ) Performance index parameter A of the product n The prediction is performed; the formula and process that meet the target requirements are entered into the multi-channel platform Cspu-Muler for experimental verification. The step S4 also includes the following steps:

[0058] S401, randomly generating a waterborne polyurethane emulsion formula and process group G0 within a restricted domain, using the Cspu-Brain(A1) intelligent model in Cspu-Brain(N) to predict the product performance index parameter A1, and obtaining a formula process group G1 that meets the performance index parameter A1 requirements;

[0059] S402, importing the data of the recipe process group G1 into Cspu-Brain (A2), and predicting the performance index parameter A2 of the product to obtain the recipe and process group G2 that meet the performance index parameter A1 and the performance index parameter A2;

[0060] S403, importing the data of the recipe process group G2 into Cspu-Brain (A3), and predicting the performance index parameter A3 of the product to obtain a recipe and process group G3 that meets the performance requirements of A1, A2 and A3;

[0061] S404, repeat the same operation process as above to obtain a water-based polyurethane emulsion that meets the comprehensive performance of glass fiber (A1~A n ) of the formula and process group G N In other embodiments, in step S404, after obtaining the comprehensive performance of the water-based polyurethane emulsion for glass fiber (A1~A n ) of the formula and process group G N In the stage, the intelligent model Cspu-Brain (A1~An ) is adjusted. For example, the selection order of the above-mentioned model for each performance index parameter is adjusted from Cspu-Brain (A1), Cspu-Brain (A2), Cspu-Brain (A3) to Cspu-Brain (A1), Cspu-Brain (A3), Cspu-Brain (A2).

[0062] S405, using the formula and process group G N into the multi-channel platform Cspu-Muler for experimental verification; the experimental data obtained by the multi-channel platform Cspu-Muler are stored into the Cspu-Room database, and the experimental data include the formula and process group and the corresponding performance index parameter.

[0063] The number of data entries in the formula and process group GN that meets the comprehensive performance of the waterborne polyurethane emulsion for glass fibers ranges from 0 to n. If the number of data entries in G N is 0, and when the formula and process are randomly generated and screened multiple times, the number of data entries in G N is still 0, a prompt is generated, which is: the probability of having a waterborne polyurethane emulsion formula and process that meets the comprehensive performance in the current limit domain is close to 0, and the limit domain should be adjusted.

[0064] The implementation case is as follows:

[0065] Take the preparation of a waterborne polyurethane emulsion as an example (emulsion particle size = 200-600 nm, tensile strength = 40-55 Mpa, thermal yellowing yellowness index YI ≤ 60), and the implementation case is as follows.

[0066] 1. Based on the existing polyurethane emulsion formula, process and corresponding performance index data for glass fibers in the laboratory, a database Cspu-Room is established.

[0067] 2. The initial version of the intelligent model Cspu-Brain is fed and iteratively upgraded using the emulsion particle size (A1), tensile strength (A2), thermal yellowing (A3) and the corresponding formula and process data in the database Cspu-Room, to obtain three independent intelligent models Cspu-Brain (particle size A1), Cspu-Brain (tensile strength A2) and Cspu-Brain (thermal yellowing A3).

[0068] 3. The intelligent models Cspu-Brain (particle size A1), Cspu-Brain (tensile strength A2) and Cspu-Brain (thermal yellowing A3) are iteratively upgraded.

[0069] (1) Randomly generate waterborne polyurethane emulsion formula and process within the limit domain (isocyanate index = 1.2-2.2, mass fraction of hydrophilic monomer ≤ 15%, mass fraction of aliphatic isocyanate = 10%-50%, amine chain extender = 70%-100%), use the multi-channel platform Cspu-Muler to prepare waterborne polyurethane emulsion and complete the emulsion particle size A1, emulsion film tensile strength A2, and thermal yellowing YI A3 tests;

[0070] (2) Use the emulsion particle size and corresponding formula process data to feed and iteratively upgrade Cspu-Brain (particle size A1). After feeding and iteration of 43 groups of data, the prediction value of the Cspu-Brain (particle size A1) intelligent model is consistent with the actual value trend based on particle size. Randomly generate 5 more formulas, use the multi-channel platform Cspu-Muler to prepare waterborne polyurethane emulsion and complete the emulsion particle size test, and the prediction accuracy reaches 94.04%, see Figure 3 .

[0071] (3) Use the tensile strength and corresponding formula process data to feed and iteratively upgrade Cspu-Brain (tensile strength A2). After feeding and iteration of 67 groups of data, the prediction value of the Cspu-Brain (tensile strength A2) intelligent model is similar to the actual value trend based on tensile strength. Randomly generate 5 more formulas, use the multi-channel platform Cspu-Muler to prepare waterborne polyurethane emulsion and complete the emulsion film tensile strength test, and the prediction accuracy reaches 84.81%, see Figure 4 .

[0072] (4) Use the thermal yellowing YI and corresponding formula process data to feed and iteratively upgrade Cspu-Brain (thermal yellowing A3). After feeding and iteration of 56 groups of data, the prediction value of the Cspu-Brain (thermal yellowing A3) intelligent model is similar to the actual value trend based on thermal yellowing YI. Randomly generate 5 more formulas, use the multi-channel platform Cspu-Muler to prepare waterborne polyurethane emulsion and complete the emulsion film tensile strength test, and the prediction accuracy reaches 86.43%, see Figure 5 .

[0073] 4, Network the intelligent models Cspu-Brain (particle size A1), Cspu-Brain (tensile strength A2), and Cspu-Brain (thermal yellowing A3) to form the intelligent model network Cspu-Brain (N).

[0074] 5, Randomly generate waterborne polyurethane emulsion formulas and processes within the limit domain (isocyanate index = 1.2-2.2, mass fraction of hydrophilic monomer ≤ 15%, mass fraction of aliphatic isocyanate = 10%-50%, amine chain extender = 70%-100%), a total of 10,000 groups.

[0075] 6、Using the intelligent model Cspu-Brain (Particle Size A1), based on the condition of emulsion particle size = 200-600 nm, 10,000 groups of data are screened to obtain a total of 2,371 groups of data entries meeting the requirements.

[0076] 7、Using the intelligent model Cspu-Brain (Thermal Yellowing A3), based on the condition of thermal yellowing YI≤60, the 2,371 groups of data screened by the intelligent model Cspu-Brain (Particle Size A1) are screened to obtain a total of 369 groups of data entries meeting the requirements.

[0077] 8、Using the intelligent model Cspu-Brain (Tensile Strength A2), based on the condition of tensile strength = 40-55 Mpa, the 369 groups of data screened by the intelligent model Cspu-Brain (Particle Size A1) and Cspu-Brain (Thermal Yellowing A3) are screened to obtain 17 groups of data entries meeting the requirements.

[0078] 9、Using the multi-channel platform Cspu-Muler, according to the 17 groups of waterborne polyurethane emulsion formula and process information screened by the intelligent model Cspu-Brain (Particle Size A1), Cspu-Brain (Thermal Yellowing A3) and Cspu-Brain (Tensile Strength A2), the waterborne polyurethane emulsion is prepared, and the emulsion particle size, emulsion film tensile strength and thermal yellowing YI test are completed. The test results show that 5 groups of experimental formula and process test results meet the requirements of emulsion particle size, tensile strength and thermal yellowing YI, and the formula, process and test results are shown in Figure 6 .

[0079] 10、The intelligent model Cspu-Brain (Particle Size A1) reads the measured emulsion particle size data and completes iterative upgrading; Cspu-Brain (Thermal Yellowing A3) reads the measured thermal yellowing YI data and completes iterative upgrading; Cspu-Brain (Tensile Strength A2) reads the measured emulsion film tensile strength data and completes iterative upgrading.

[0080] The test methods used include:

[0081] The particle size of the waterborne polyurethane is tested according to GB / T 19077-2016 using a 90Plus laser particle size analyzer (Brookhaven Corporation, USA).

[0082] The mechanical properties of the waterborne polyurethane adhesive film are tested according to GB / T 1040-2018 using an Instron 5982 universal material testing machine (Instron Corporation, USA).

[0083] The waterborne polyurethane coating film was tested for film yellowness index (YI) after heating in an oven at 210°C for 30 min according to ASTM E313 using a CS-820N benchtop spectrophotometer (Hangzhou Caipu Technology Co., Ltd.).

[0084] The above-mentioned are only embodiments of the present application, and the common knowledge of specific structures and properties in the scheme and the like will not be described too much. The ordinary skilled person in the art knows all the ordinary technical knowledge in the technical field of the present application before the filing date or the priority date, can know all the prior art in the field, and has the ability to apply conventional experimental means before that date. The ordinary skilled person in the art can perfect and implement the present scheme under the guidance of the present application, combined with their own ability. Some typical known structures or known methods should not be an obstacle for the ordinary skilled person in the art to implement the present application. It should be noted that, for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can be made, which should also be considered within the scope of protection of the present application. These will not affect the effect of the implementation of the present application and the practicality of the patent. The scope of protection claimed in the present application should be subject to the content of its claims, and the specific implementation mode and the like in the specification can be used to explain the content of the claims.

Claims

1. A method for predicting the performance of water-based polyurethane emulsion for glass fiber, characterized in that: The following steps are involved: S1. Constructing a database Cspu-Room for water-based polyurethane emulsions for glass fiber; the database Cspu-Room includes data on commonly used raw materials for preparing polyurethane emulsions and their basic physical and chemical properties, as well as data on existing polyurethane emulsion formulas, processes, and corresponding performance indicators for glass fiber; S2. Construct a multi-channel platform Cspu-Muler for water-based polyurethane emulsion for glass fiber, wherein the multi-channel platform Cspu-Muler is used to perform random formula synthesis and performance index evaluation of water-based polyurethane emulsions within a restricted domain and random process synthesis and performance index evaluation of water-based polyurethane emulsions within a restricted domain based on the database Cspu-Room; return the formula, process and corresponding performance index evaluation data of water-based polyurethane emulsion for glass fiber verified by the multi-channel platform Cspu-Muler to the Cspu-Room database; S3, based on the database Cspu-Room, build and train the intelligent model Cspu-Brain for predicting the performance of water-based polyurethane emulsion for glass fiber; according to the performance index parameters A1~A1 of water-based polyurethane emulsion in the database Cspu-Room, n , and conduct targeted optimization training and upgrade iteration on the intelligent model Cspu-Brain, respectively obtaining the intelligent model Cspu-Brain (A1~A n ) and combine to form the intelligent model network Cspu-Brain(N); S4. The formula and process of randomly synthesized waterborne polyurethane emulsion in the restricted domain are based on the intelligent model Cspu-Brain (A) in the intelligent model network Cspu-Brain (N). n ) Performance index parameter A of the product n Make predictions; the formula and process that meet the target requirements are entered into the multi-channel platform Cspu-Muler for experimental verification.

2. The method for predicting properties of an aqueous polyurethane emulsion for glass fiber according to claim 1, wherein: The performance-related index parameters A1 to A1 of the aqueous polyurethane emulsion n The characteristic properties of water-based polyurethane emulsion for glass fiber are of concern, including but not limited to emulsion particle size, emulsion viscosity, emulsion centrifugal sedimentation content, film strength, film elongation, film thermal yellowing and film adhesion.

3. The method for predicting properties of a water-based polyurethane emulsion for glass fiber according to claim 1, wherein: The step S3 specifically includes the following steps: S301. Build the initial version of the intelligent model Cspu-Brain for predicting the properties of glass fiber water-based polyurethane emulsion based on the database Cspu-Room; S302, using the glass fiber water-based polyurethane emulsion database Cspu-Room A n The performance data were fed to the glass fiber water-based polyurethane emulsion intelligent model Cspu-Brain to train the initial version of the intelligent model Cspu-Brain (A n ); S303, loop step S302, and obtain the intelligent model Cspu-Brain (A1~A n ); S304, the intelligent model Cspu-Brain (A1~A n ) are combined to form the intelligent model network Cspu-Brain(N); S305, according to the continuously updated database Cspu-Room waterborne polyurethane emulsion performance related index parameters A n , respectively for the intelligent model Cspu-Brain (A n ) to conduct targeted optimization training and upgrade iterations, and complete the self-iteration upgrade of the intelligent model network Cspu-Brain(N).

4. The method for predicting properties of an aqueous polyurethane emulsion for glass fiber according to claim 3, wherein: The step S305 includes the following steps: S3051. The intelligent model Cspu-Brain (A1) reads experimental data related to A1's performance through the database Cspu-Room to complete self-iteration and upgrade. S3052: The intelligent model Cspu-Brain (A2) reads experimental data related to A2 performance through the database Cspu-Room to complete self-iteration and upgrade. S3053. The intelligent model Cspu-Brain (A3) reads experimental data related to A3 performance through the database Cspu-Room to complete self-iteration and upgrade. S3054, repeat the same operation process as above to complete the intelligent model Cspu-Brain (A1) to the intelligent model Cspu-Brain (A n ) to realize the iterative upgrade of the intelligent model network Cspu-Brain(N).

5. The method for predicting properties of aqueous polyurethane emulsion for glass fiber according to claim 4, wherein: The step S4 further includes the following steps: S401, randomly generating a waterborne polyurethane emulsion formula and process group G0 within a restricted domain, using the Cspu-Brain(A1) intelligent model in Cspu-Brain(N) to predict the product performance index parameter A1, and obtaining a formula process group G1 that meets the performance index parameter A1 requirements; S402, importing the data of the recipe process group G1 into Cspu-Brain (A2), and predicting the performance index parameter A2 of the product to obtain the recipe and process group G2 that meet the performance index parameter A1 and the performance index parameter A2; S403, importing the data of the recipe process group G2 into Cspu-Brain (A3), and predicting the performance index parameter A3 of the product to obtain a recipe and process group G3 that meets the performance requirements of A1, A2 and A3; S404, repeat the same operation process as above to obtain a water-based polyurethane emulsion that meets the comprehensive performance of glass fiber (A1~A n ) of the formula and process group G N ; S405, use formula and process group G N Enter the multi-channel platform Cspu-Muler for experimental verification; store the experimental data obtained by the multi-channel platform Cspu-Muler into the Cspu-Room database, wherein the experimental data includes the formula and process group and the corresponding performance index parameters.

6. The method for predicting properties of aqueous polyurethane emulsion for glass fiber according to claim 4, wherein: The intelligent model Cspu-Brain (A n ) After continuous iteration and upgrading, and before forming the intelligent model network Cspu-Brain (N), set each intelligent model Cspu-Brain (A n ) The accuracy of the single performance model is greater than 75%, preferably greater than 80%.

7. The method for predicting properties of aqueous polyurethane emulsion for glass fiber according to claim 5, wherein: The number of data entries in the formula and process group GN that meet the comprehensive performance of the glass fiber water-based polyurethane emulsion ranges from 0 to n. N The number of data entries in G is 0. At the same time, when multiple recipes and processes are randomly generated and screened, G N When the number of data entries is still 0, a prompt is generated, which is: the probability of having a water-based polyurethane emulsion formula and process that meets the comprehensive performance within the current restriction domain is close to 0, and the restriction domain should be adjusted.

8. The method for predicting properties of aqueous polyurethane emulsion for glass fiber according to claim 5, wherein: In the step S404, after obtaining the comprehensive performance of the water-based polyurethane emulsion for glass fiber (A1~A n ) of the formula and process group G N In the stage, the intelligent model Cspu-Brain (A1~A n ) in the order of .

9. The method for predicting properties of aqueous polyurethane emulsion for glass fiber according to claim 1, wherein: The multi-channel platform Cspu-Muler consists of a multi-channel PUD synthesis reactor, an A1 performance test platform, an A2 performance test platform and an A n The performance testing platform is built in combination to complete the full-chain experiment of PUD from raw materials to synthesis and then to performance testing.

10. The method for predicting properties of aqueous polyurethane emulsion for glass fiber according to claim 1, wherein: The Cspu-Brain (A n ) writing and iterative methods include but are not limited to machine learning and random forest algorithms.

Images