Prediction method and system of anti-impact polypropylene microstructure and storage medium

The neural network model predicts the process parameters of impact-resistant polypropylene, which solves the problem that traditional methods are difficult to capture complex polymerization processes, improves the strength and toughness of the product, and achieves more efficient process parameter adjustment.

CN118983036AActive Publication Date: 2024-11-19EAST CHINA UNIV OF SCI & TECH +1

Patent Information

Application Number
CN202411064309.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-19
Estimated Expiration
2044-08-05

AI Technical Summary

Technical Problem

Traditional methods are difficult to accurately predict the microstructure of impact-resistant polypropylene, especially when considering multi-component and complex phase structures, the model is complex and computational, making it difficult to capture the multiple interactions and molecular arrangements during the polymerization process, which affects the Product strength and toughness.

Method used

The prediction method based on neural network model is adopted, and the BP neural network model is trained through a large amount of historical data, and the process parameters are adjusted to improve the molecular weight characteristics of impact-resistant polypropylene and improve the adaptability of process parameters.

Benefits of technology

It achieves more accurate prediction and improvement of the quality and performance of impact polypropylene products, improves the adaptability of process parameters, and significantly improves the mechanical properties of the products.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118983036A_ABST
    Figure CN118983036A_ABST
Patent Text Reader

Abstract

The invention provides an anti-impact polypropylene microstructure prediction method and system and a storage medium. The method for predicting the microstructure of the impact-resistant polypropylene comprises the following steps: acquiring technological parameters for processing the impact-resistant polypropylene; the process parameters comprise a first raw material concentration and homopolymerization time of at least one first raw material in a homopolymerization stage, and a second raw material concentration and copolymerization time of at least one second raw material in a copolymerization stage; determining a predicted value of the molecular weight characteristic of the anti-impact polypropylene product according to the process parameters; and adjusting the process parameters according to the difference between the target value and the predicted value of the molecular weight characteristic so as to obtain the impact-resistant polypropylene product loaded with the target value. According to the method, training and prediction can be carried out based on the neural network model and by utilizing a large amount of historical data, the technological parameters for processing the impact-resistant polypropylene can be adjusted more accurately, the adaptability to various technological parameters is improved, and therefore the quality and performance of the impact-resistant polypropylene product are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of computer chemical technology, and in particular to a prediction method for the microstructure of impact-resistant polypropylene, a prediction system for the microstructure of impact-resistant polypropylene, and a computer-readable storage medium. Background Art

[0002] Impact polypropylene is a complex multi-component multi-phase polymer alloy system, whose matrix is ​​polypropylene. By introducing an appropriate amount of impact modifier into the polypropylene matrix, it can significantly improve the impact resistance and low-temperature toughness while maintaining the excellent properties of polypropylene. It has been widely used in the fields of automobiles, home appliances and home furnishings. With the continuous improvement of product quality requirements, high-performance impact-resistant polypropylene products have become the goal pursued by manufacturers. Researchers have continuously studied the process parameters and mechanical properties of impact polypropylene and found that different polymerization process parameters will form different molecular weight characteristics, which in turn affect the strength, toughness and other mechanical properties of impact polypropylene. However, traditional preparation methods, such as mathematical models or chemical process simulations using Aspen software, have limited adaptability. Especially for impact polypropylene, a polymer with multi-component and complex phase structure characteristics, it is challenging to use mechanism models to predict its performance. In addition, the various interactions and molecular arrangements involved in the polymerization process make it difficult for differential equation systems to fully capture these complexities. Especially when considering factors such as the orientation of polymer chains, copolymer composition, and the influence of additives, the complexity and computational complexity of the model are further increased.

[0003] In order to overcome the above-mentioned defects of the prior art, the art urgently needs an improved prediction method for the microstructure of impact-resistant polypropylene, which can be used to more accurately adjust the process parameters for processing impact-resistant polypropylene and improve the adaptability to a variety of process parameters, thereby improving the quality and performance of impact-resistant polypropylene products. Summary of the invention

[0004] A brief summary of one or more aspects is given below to provide a basic understanding of these aspects. This summary is not an exhaustive overview of all conceived aspects, and is neither intended to identify the key or decisive elements of all aspects nor to define the scope of any or all aspects. Its only purpose is to give some concepts of one or more aspects in a simplified form as a prelude to a more detailed description that will be given later.

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method for predicting the microstructure of impact-resistant polypropylene, a system for predicting the microstructure of impact-resistant polypropylene and a computer-readable storage medium, which can be used to more accurately adjust the process parameters for processing impact-resistant polypropylene and improve the adaptability to a variety of process parameters through training and prediction based on a neural network model and using a large amount of historical data, thereby improving the quality and performance of impact-resistant polypropylene products.

[0006] Specifically, the method for predicting the microstructure of impact polypropylene provided in accordance with the first aspect of the present invention comprises the following steps: obtaining process parameters for processing impact polypropylene. The process parameters include the first raw material concentration and homopolymerization time of at least one first raw material in the homopolymerization stage, and the second raw material concentration and copolymerization time of at least one second raw material in the copolymerization stage; determining the predicted value of the molecular weight characteristics of the impact polypropylene product according to the process parameters; and adjusting the process parameters according to the difference between the target value and the predicted value of the molecular weight characteristics to obtain an impact polypropylene product that meets the target value.

[0007] Furthermore, in some embodiments of the present invention, the at least one first raw material comprises propylene and hydrogen. The at least one second raw material comprises propylene, ethylene and hydrogen. The target value and predicted value of the molecular weight characteristic correspondingly relate to at least one of the first number average molecular weight Mn_PP, the first weight average molecular weight Mw_PP, the first molecular weight distribution index PDI_PP, and the first mass percentage PP_per of polypropylene, the second number average molecular weight Mn_iPP, the second weight average molecular weight Mw_iPP, and the second molecular weight distribution index PDI_iPP of isotactic polypropylene, the third number average molecular weight Mn_aPP, the third weight average molecular weight Mw_aPP, and the third molecular weight distribution index PDI_aPP of random polypropylene, the fourth number average molecular weight Mn_EPR, the fourth weight average molecular weight Mw_EPR, the fourth molecular weight distribution index PDI_EPR, and the fourth mass percentage EPR_per of ethylene-propylene random copolymer, the first ethylene molar content E_mole_EPR, the fifth number average molecular weight Mn_EPS, the fifth weight average molecular weight Mw_EPS, the fifth molecular weight distribution index PDI_EPS, the fifth mass percentage EPS_per, the second ethylene molar content E_mole_EPS, and the third ethylene molar content E_in_IPC in the polymer of ethylene-propylene multi-block copolymer.

[0008] Furthermore, in some embodiments of the present invention, the step of determining the predicted value of the molecular weight characteristics of the impact-resistant polypropylene product based on the process parameters includes: inputting the process parameters into a pre-trained BP neural network model to determine the predicted value of the molecular weight characteristics of the impact-resistant polypropylene product via the BP neural network model.

[0009] Furthermore, in some embodiments of the present invention, the step of training the BP neural network model includes: obtaining multiple groups of historical sample data for processing impact-resistant polypropylene to construct a historical sample data set. Each of the historical sample data includes the process parameters and the true value of the molecular weight characteristics of the corresponding historical sample; building a BP neural network model to be trained. The BP neural network model consists of an input layer, two hidden layers and an output layer; and inputting each group of training sample data in the historical sample data set into the BP neural network model in turn to obtain the corresponding molecular weight characteristic prediction value, and adjusting the learning parameters of the BP neural network model according to the difference between the predicted value and the true value of the molecular weight characteristic of the corresponding training sample, so as to train the BP neural network model.

[0010] Furthermore, in some embodiments of the present invention, before inputting each training sample in the historical sample data set into the BP neural network model in sequence, the prediction method further comprises the following steps: normalizing each historical sample data in the historical sample data set to normalize its value between -1 and 1:

[0011]

[0012] Among them, X reduced,p,i is the normalized variable of the pth historical sample data of the i-th neuron. p,i X is the original variable of the pth historical sample data of the i-th neuron. min,i is the minimum value of the variable of the ith neuron in the historical sample data set. max,i is the maximum value of the variable of the ith neuron in the historical sample data set.

[0013] Furthermore, in some embodiments of the present invention, the step of inputting each training sample data in the historical sample data set into the BP neural network model in sequence to obtain the corresponding molecular weight characteristic prediction value includes: randomly dividing the historical sample data set into a training set and a test set according to a preset ratio; defining the historical sample data in the training set as training sample data; and inputting each group of training sample data in the training set into the BP neural network model in sequence to obtain the corresponding molecular weight characteristic prediction value.

[0014] Furthermore, in some embodiments of the present invention, the following steps are also included: defining the historical sample data in the test set as test sample data; in response to completing the training of the BP neural network model, inputting each group of test sample data in the test set into the BP neural network model in turn to obtain the corresponding molecular weight characteristic test value; and verifying the validity of the BP neural network model based on the degree of fit between the regression curve graph of each test value and the true value of the corresponding test sample.

[0015] Furthermore, in some embodiments of the present invention, the number of neurons N in the input layer is i The number of neurons in the output layer is determined according to the output dimension of the predicted value of the molecular weight characteristic. The number of neurons in the hidden layer is determined according to the number of neurons in the input layer, the number of neurons in the output layer, and the number of groups of the training sample data, N. s Sure:

[0016]

[0017] The value of α depends on the type of molecular weight property to be predicted, and ranges from 4 to 10.

[0018] Further, in some embodiments of the present invention, the step of sequentially inputting each set of training sample data in the historical sample data set into the BP neural network model to obtain the corresponding molecular weight characteristic prediction value includes: inputting the weight w of the connection between each neuron i and j in the BP neural network model into the BP neural network model. j,i Initialize to a normal distribution:

[0019]

[0020] Among them, U is the function of normal distribution, n in is the number of rows in the training sample data matrix, n out is the number of columns in the training sample data matrix; the threshold value B of each neuron i in the BP neural network model is i Initialized to 0; based on the weights of the connections between the neurons and the thresholds of the neurons in each layer of the neural network, the net input variables of each layer of the neural network are calculated:

[0021]

[0022] in, is the net input variable of the pth data of the jth neuron of the k+1th layer of neural network, is the threshold of the jth neuron in the k+1th layer of the neural network, is the weight of the connection between the jth neuron in the k+1th layer of the neural network and the ith neuron in the kth layer of the neural network, is the output variable of the pth data of the ith neuron of the kth layer of the neural network. The output variable of the jth neuron of the k+1th layer of the neural network is calculated by transferring the net input variable of the jth neuron of the k+1th layer of the neural network according to the custom activation function:

[0023]

[0024] in, is the net input variable of the pth data of the jth neuron of the k+1th layer of neural network, is the output variable of the pth data of the jth neuron in the k+1th layer of the neural network, and the value of α1 is 1.5.

[0025] Further, in some embodiments of the present invention, the step of adjusting the learning parameters of the BP neural network model according to the difference between the predicted value and the true value of the molecular weight characteristic of the corresponding training sample to train the BP neural network model includes: calculating the mean square error between the predicted value and the true value of the molecular weight characteristic of each training sample:

[0026]

[0027] in, is the true value of the pth training sample of the jth neuron. is the predicted value of the pth training sample of the jth neuron; and using the Adam optimizer to perform back propagation based on the mean square error on the BP neural network model until the training of the BP neural network model is completed.

[0028] Further, in some embodiments of the present invention, the step of using the Adam optimizer to perform back propagation based on the mean square error on the BP neural network model until the training of the BP neural network model is completed includes: setting a maximum number of iterations t for training the BP neural network model max , indicating parameter θ1, and a patience parameter to prevent overfitting. The value of the patience parameter is calculated according to the formula patience θ =max(input_n, output_n) / 2. input_n is the number of input nodes. output_n is the number of output nodes. The initial value of the indication parameter θ1 is 0; in response to the current round number t not reaching the maximum iteration round number t max , and the indicator parameter θ1 is less than or equal to the patience parameter, the first-order moment estimate m of the Adam optimizer t and the second-order moment estimate v tCombined together, the weights of the connections between neurons in the BP neural network model and the thresholds of the neurons are updated:

[0029]

[0030] Among them, w t , b t are the weights and thresholds after the tth round update. t-1 , b t-1 are the weights and thresholds of the t-1th round. α2 is the learning rate. θ2 is the zero-proof constant. The first-order moment estimate m of the Adam optimizer t The expression is:

[0031] m t =β1m t-1 +(1-β1)h t

[0032] Among them, m t-1 and m t They represent the first-order moment estimates before and after iteration, and β1 is the exponentially weighted average parameter. t Represents the gradient of the weight ω or threshold b.

[0033] The second-order moment estimate v of the Adam optimizer t The expression is:

[0034] v t =β2v t-1 +(1-β2)h t 2

[0035] Among them, v t-1 and v t are the second-order moment estimates before and after iteration, and β2 is the exponentially weighted average parameter. t 2 represents the square of the gradient of the weight ω or the threshold b; determines whether the mean square error of the current round increases, if it increases, adds 1 to the indicator parameter θ1, otherwise sets the indicator parameter θ1 to zero; and in response to the current round number t reaching the maximum iteration round number t max , or the indication parameter θ1 is greater than the patience parameter, it is determined that the training of the BP neural network model is completed.

[0036] In addition, the above-mentioned prediction system of the microstructure of impact-resistant polypropylene provided according to the second aspect of the present invention includes a memory and a processor. The memory stores computer instructions. The processor is connected to the memory and is configured to execute the computer instructions stored in the memory to implement the prediction method of the microstructure of impact-resistant polypropylene provided according to the first aspect of the present invention.

[0037] In addition, the computer-readable storage medium provided in the third aspect of the present invention stores computer instructions, which, when executed by a processor, implement the method for predicting the microstructure of impact-resistant polypropylene provided in the first aspect of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The above features and advantages of the present invention can be better understood after reading the detailed description of the embodiments of the present disclosure in conjunction with the following drawings. In the drawings, the components are not necessarily drawn to scale, and components with similar related properties or features may have the same or similar reference numerals.

[0039] Figure 1 A schematic structural diagram of a BP neural network model provided according to some embodiments of the present invention is shown.

[0040] Figure 2A A regression curve diagram of the second number average molecular weight Mn iPP provided according to some embodiments of the present invention is shown.

[0041] Figure 2B A regression curve diagram of the second weight average molecular weight MwiPP provided according to some embodiments of the present invention is shown.

[0042] Figure 3A A regression curve diagram of the fourth number average molecular weight MnEPR of ethylene-propylene random copolymer provided according to some embodiments of the present invention is shown.

[0043] Figure 3B A regression curve chart of the fifth number average molecular weight Mn EPS of ethylene-propylene multi-block copolymers provided according to some embodiments of the present invention is shown.

[0044] Figure 4 A schematic flow chart of a method for predicting the microstructure of impact-resistant polypropylene provided according to some embodiments of the present invention is shown. DETAILED DESCRIPTION

[0045] The following specific embodiments illustrate the implementation of the present invention, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. Although the description of the present invention will be introduced in conjunction with the preferred embodiment, this does not mean that the features of this invention are limited to this implementation. On the contrary, the purpose of introducing the invention in conjunction with the implementation is to cover other options or modifications that may be extended based on the claims of the present invention. In order to provide a deep understanding of the present invention, the following description will include many specific details. The present invention can also be implemented without using these details. In addition, in order to avoid confusion or blurring the focus of the present invention, some specific details will be omitted in the description.

[0046] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0047] In addition, the terms "upper", "lower", "left", "right", "top", "bottom", "horizontal" and "vertical" used in the following description should be understood as the directions shown in the paragraph and the related drawings. Such relative terms are only used for the convenience of description and do not mean that the device described therein must be manufactured or operated in a specific direction, and therefore should not be understood as limiting the present invention.

[0048] It is understood that although the terms "first", "second", "third", etc. may be used herein to describe various components, regions, layers and / or parts, these components, regions, layers and / or parts should not be limited by these terms, and these terms are only used to distinguish different components, regions, layers and / or parts. Therefore, the first component, region, layer and / or part discussed below may be referred to as a second component, region, layer and / or part without departing from some embodiments of the present invention.

[0049] As mentioned above, traditional preparation methods, such as those based on mathematical models or using Aspen software for chemical process simulation, have limited adaptability. Especially for polymers such as impact polypropylene, which have multi-component and complex phase structure characteristics, it is challenging to use mechanism models to predict their performance. In addition, the various interactions and molecular arrangements involved in the polymerization process make it difficult for differential equation systems to fully capture these complexities. Especially when considering factors such as the orientation of polymer chains, copolymer composition, and the influence of additives, the complexity and computational complexity of the model are further increased.

[0050] In order to overcome the above-mentioned defects of the prior art, the present invention provides a prediction method for the microstructure of impact-resistant polypropylene, a prediction system for the microstructure of impact-resistant polypropylene and a computer-readable storage medium, which can be based on a neural network model and use a large amount of historical data for training and prediction, so as to more accurately adjust the process parameters for processing impact-resistant polypropylene and improve the adaptability to a variety of process parameters, thereby improving the quality and performance of impact-resistant polypropylene products.

[0051] In some non-limiting embodiments, the prediction method of the above-mentioned impact-resistant polypropylene microstructure provided by the first aspect of the present invention can be implemented based on the prediction system of the above-mentioned impact-resistant polypropylene microstructure provided by the second aspect of the present invention. Specifically, the prediction system of the impact-resistant polypropylene microstructure can be configured with a memory and a processor. The memory includes but is not limited to the above-mentioned computer-readable storage medium provided by the third aspect of the present invention, on which computer instructions are stored. The processor is connected to the memory and is configured to execute the computer instructions stored on the memory to implement the prediction method of the above-mentioned impact-resistant polypropylene microstructure provided by the first aspect of the present invention.

[0052] Specifically, the processor can pre-train the BP neural network model offline. The processor can first obtain multiple sets of historical sample data for processing impact polypropylene to construct a historical sample data set. Here, each historical sample data includes the process parameters and molecular weight characteristic true values ​​of the corresponding historical sample.

[0053] Please refer to Figure 1 . Figure 1 A schematic structural diagram of a BP neural network model provided according to some embodiments of the present invention is shown.

[0054] like Figure 1 As shown, the processor can build a BP neural network model to be trained. Here, the BP neural network model consists of an input layer 11, two hidden layers 12 and an output layer 13.

[0055] Afterwards, in some preferred embodiments, the processor may first perform normalization processing on each historical sample data in the historical sample data set to normalize its value between -1 and 1:

[0056]

[0057] Among them, X reduced,p,i is the normalized variable of the pth historical sample data of the ith neuron, X p,i is the original variable of the pth historical sample data of the ith neuron, X min,i is the minimum value of the variable of the i-th neuron in the historical sample data set, X max,i is the maximum value of the variable of the i-th neuron in the historical sample data set.

[0058] Afterwards, the processor can randomly divide the historical sample data set into a training set and a test set according to a preset ratio (e.g., 8:2). Afterwards, the processor can define the historical sample data in the training set as training sample data, and input each group of training sample data in the training set into the BP neural network model in sequence to obtain the corresponding molecular weight characteristic prediction value.

[0059] In addition, in the process of determining the structure of the above BP neural network model, the number of neurons in the input layer N i It is determined according to the input dimension of each group of training sample data, and the number of neurons in the output layer N o It is determined by the output dimension of the predicted value of the molecular weight characteristic, and the number of neurons in the hidden layer is determined by the number of neurons N in the input layer. i , the number of neurons in the output layer N o , and the number of groups of training sample data N s Sure:

[0060]

[0061] The value of α depends on the type of molecular weight characteristics to be predicted, and the value range is 4 to 10. Here, when predicting the molecular weight characteristics of homopolymers and copolymers at the same time, α can be set to 4, when predicting the molecular weight characteristics of homopolymers, α can be set to 8, and when predicting the molecular weight characteristics of copolymers, α can be set to 10.

[0062] Furthermore, after inputting each group of training sample data in the training set into the BP neural network model in sequence, the processor can set the weight w of the connection between each neuron i and j in the BP neural network model. j,i Initialize to a normal distribution:

[0063]

[0064] Among them, U is the normal distribution function, n in is the number of rows in the training sample data matrix, n out is the number of columns in the training sample data matrix.

[0065] Afterwards, the processor can first set the threshold value B of each neuron i in the BP neural network model. i Initialized to 0, and then calculated based on the weights of the connections between neurons and the thresholds of the neurons in each layer of the neural network:

[0066]

[0067] in, is the net input variable of the pth data of the jth neuron of the k+1th layer of neural network, is the threshold of the jth neuron in the k+1th layer of the neural network, is the weight of the connection between the jth neuron in the k+1th layer of the neural network and the ith neuron in the kth layer of the neural network, is the output variable of the pth data of the ith neuron of the kth layer of the neural network. The output variable of the jth neuron of the k+1th layer of the neural network is calculated by transferring the net input variable of the jth neuron of the k+1th layer of the neural network according to the custom activation function:

[0068]

[0069] in, is the net input variable of the pth data of the jth neuron of the k+1th layer of neural network, is the output variable of the pth data of the jth neuron in the k+1th layer of the neural network, and the value of α1 is 1.5.

[0070] In this way, the processor can further process the variables in the hidden layer and the output layer using the same above formula to obtain the output variables of the BP neural network model.

[0071] Afterwards, the processor may adjust the learning parameters of the BP neural network model according to the difference between the actual value of the molecular weight characteristic of the corresponding training sample and the actual value of the molecular weight characteristic of the corresponding training sample, so as to train the BP neural network model.

[0072] Specifically, the processor can calculate the mean square error between the predicted value and the true value of the molecular weight characteristic for each training sample:

[0073]

[0074] in, is the true value of the pth training sample of the jth neuron, is the predicted value of the p-th training sample of the j-th neuron.

[0075] Afterwards, the processor may use the Adam optimizer to perform back propagation based on the mean square error on the BP neural network model until the training of the BP neural network model is completed.

[0076] Specifically, during the back propagation of the BP neural network, the processor can first set the maximum number of iterations tm for training the BP neural network model. ax、 Indicative parameter θ1, and the patience parameter to prevent overfitting. Here, the value of the patience parameter is calculated according to the formula θ =max(input_n, output_n) / 2, input_n is the number of input nodes, output_n is the number of output nodes, and the initial value of the indicator parameter θ1 is 0. Here, in the process of predicting the molecular weight characteristics of the homopolymer, the patience parameter patience θ It can be set to 5. In the process of predicting the molecular weight characteristics of copolymers, the patience parameter patienceθ Can be set to 6.

[0077] Afterwards, in response to the current round number t not reaching the maximum iteration round number t max , and the indicator parameter θ1 is less than or equal to the patience parameter, the processor can use the first-order moment estimate m of the Adam optimizer t and the second-order moment estimate v t Combined together, the weights of the connections between neurons and the thresholds of each neuron in the BP neural network model are updated:

[0078]

[0079] Among them, w t , b t is the weight and threshold after the tth round update, w t-1 , b t-1 is the weight and threshold of the t-1th round. α2 is the learning rate, which is 0.0025. θ2 is the anti-zero constant, which is 10 -8 .

[0080] Here, the first-order moment estimate m of the Adam optimizer t The expression is:

[0081] m t =β1m t-1 +(1-β1)h t

[0082] Among them, m t-1 and m t They represent the first-order moment estimates before and after iteration, β1 is the exponentially weighted average parameter, and h t Represents the gradient of the weight ω or threshold b.

[0083] Here, the second-order moment estimate v of the Adam optimizer t The expression is:

[0084] v t =β2v t-1 +(1-β2)h t 2

[0085] Among them, v t-1 and v t are the second-order moment estimates before and after iteration, β2 is the exponentially weighted average parameter, and h t 2 Represents the squared gradient of the weight ω or threshold b.

[0086] Afterwards, the processor may determine whether the mean square error of the current round increases, and if so, add 1 to the indicator parameter θ1, otherwise, set the indicator parameter θ1 to zero.

[0087] Afterwards, in response to the current round number t reaching the maximum iteration round number t max , or the indication parameter θ1 is greater than the patience parameter, the processor can determine that the training of the BP neural network model is completed.

[0088] In addition, in some preferred embodiments, the processor can also define the historical sample data in the test set as the test sample data. In response to completing the training of the BP neural network model, the processor can sequentially input each group of test sample data in the test set into the BP neural network model to obtain the corresponding molecular weight characteristic test value. Afterwards, the processor can verify the validity of the BP neural network model based on the degree of fit between the regression curve graph of each test value and the true value of the corresponding test sample.

[0089] Specifically, the processor can use mean squared error (MSE), mean absolute error (MAE), goodness of fit (r 2 ) Three indicators are used to evaluate performance:

[0090]

[0091] Among them, y i represents the true value of the test sample, represents the average value of the true value of the test sample, Represents the true value of the test sample.

[0092] Here, the target value and predicted value of the above-mentioned molecular weight characteristics correspondingly involve at least one of the first number average molecular weight Mn_PP, the first weight average molecular weight Mw_PP, the first molecular weight distribution index PDI_PP, and the first mass percentage PP_per of polypropylene, the second number average molecular weight Mn_iPP, the second weight average molecular weight Mw_iPP, and the second molecular weight distribution index PDI_iPP of isotactic polypropylene, the third number average molecular weight Mn_aPP, the third weight average molecular weight Mw_aPP, and the third molecular weight distribution index PDI_aPP of random polypropylene, the fourth number average molecular weight Mn_EPR, the fourth weight average molecular weight Mw_EPR, the fourth molecular weight distribution index PDI_EPR, and the fourth mass percentage EPR_per of ethylene-propylene random copolymer, the first ethylene molar content E_mole_EPR, the fifth number average molecular weight Mn_EPS, the fifth weight average molecular weight Mw_EPS, the fifth molecular weight distribution index PDI_EPS, the fifth mass percentage EPS_per, the second ethylene molar content E_mole_EPS, and the third ethylene molar content E_in_IPC in the polymer of ethylene-propylene multi-block copolymer.

[0093] Please refer to Table 1 for details. Figure 2A and Figure 2B Table 1 shows a performance index table of homopolymer molecular weight characteristics provided according to some embodiments of the present invention. Figure 2A A regression curve diagram of the second number average molecular weight Mn_iPP provided according to some embodiments of the present invention is shown. Figure 2B A regression curve diagram of the second weight average molecular weight Mw_iPP provided according to some embodiments of the present invention is shown.

[0094] Table 1 Performance index table of homopolymer molecular weight characteristics

[0095] MSE MAE <![CDATA[R 2 ]]> 0.00034036 0.01100553 0.9966

[0096] As shown in Table 1, when the value of α is 8, the BP neural network model predicts the molecular weight characteristics of the homopolymer. The mean square error of the above BP neural network model is 0.00034036, the mean absolute error is 0.01100553, and the goodness of fit is 0.9966. Figure 2A and Figure 2B As shown, the BP neural network model has a lower error and higher accuracy for the molecular weight characteristics of homopolymers.

[0097] In addition, please refer to Table 2. Figure 3A and Figure 3B Table 2 shows a performance index table of molecular weight characteristics of copolymers provided according to some embodiments of the present invention. Figure 3A A regression curve diagram of the fourth number average molecular weight Mn_EPR of ethylene-propylene random copolymer provided according to some embodiments of the present invention is shown. Figure 3B A regression curve chart of the fifth number average molecular weight Mn_EPS of ethylene-propylene multi-block copolymers provided according to some embodiments of the present invention is shown.

[0098] Table 2 Performance index table of copolymer molecular weight characteristics

[0099] MSE MAE <![CDATA[R 2 ]]> 0.00064456 0.01680668 0.9601

[0100] As shown in Table 2, when the value of α is 10, the BP neural network model predicts the molecular weight characteristics of the copolymer. The mean square error of the above BP neural network model is 0.00064456, the mean absolute error is 0.01680668, and the goodness of fit is 0.9601. Figure 3A and Figure 3B As shown, the BP neural network model has lower error and higher accuracy for the molecular weight characteristics of copolymers.

[0101] Please refer to Figure 4 . Figure 4 A schematic flow chart of a method for predicting the microstructure of impact-resistant polypropylene provided according to some embodiments of the present invention is shown.

[0102] like Figure 4 As shown, during the online processing of impact polypropylene, the processor can first obtain the process parameters for processing impact polypropylene. Here, the process parameters include the first raw material concentration and homopolymerization time of at least one first raw material in the homopolymerization stage, and the second raw material concentration and copolymerization time of at least one second raw material in the copolymerization stage.

[0103] Furthermore, in some embodiments, the at least one first raw material includes propylene and hydrogen, and the at least one second raw material includes propylene, ethylene, and hydrogen.

[0104] The processor may then determine a predicted value for the molecular weight characteristic of the impact polypropylene product based on the process parameters.

[0105] Specifically, the processor can input the process parameters into the pre-trained BP neural network model to determine the predicted value of the molecular weight characteristic of the impact polypropylene product via the BP neural network model.

[0106] Thereafter, the processor can adjust the process parameters according to the difference between the target value and the predicted value of the molecular weight characteristic to obtain an impact-resistant polypropylene product with a load target value.

[0107] In summary, the above-mentioned impact-resistant polypropylene microstructure prediction method, impact-resistant polypropylene microstructure prediction system and computer-readable storage medium provided by the present invention can be used to more accurately adjust the process parameters for processing impact-resistant polypropylene and improve the adaptability to various process parameters through training and prediction based on a neural network model and using a large amount of historical data, thereby improving the quality and performance of impact-resistant polypropylene products.

[0108] Although the above methods are illustrated and described as a series of actions for simplicity of explanation, it should be understood and appreciated that these methods are not limited by the order of the actions, because according to one or more embodiments, some actions may occur in a different order and / or concurrently with other actions from those illustrated and described herein or not illustrated and described herein but understandable to those skilled in the art.

[0109] The previous description of the disclosure is provided to enable any person skilled in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the spirit or scope of the disclosure. Thus, the disclosure is not intended to be limited to the examples and designs described herein, but should be granted the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for predicting the microstructure of impact-resistant polypropylene, characterized in that: The following steps are involved: Obtaining process parameters for processing impact-resistant polypropylene, wherein the process parameters include a first raw material concentration and a homopolymerization time of at least one first raw material in a homopolymerization stage, and a second raw material concentration and a copolymerization time of at least one second raw material in a copolymerization stage; Determining a predicted value of a molecular weight characteristic of an impact polypropylene product based on the process parameters; and According to the difference between the target value and the predicted value of the molecular weight characteristic, the process parameters are adjusted to obtain an impact-resistant polypropylene product that meets the target value.

2. The prediction method according to claim 1, characterized in that: The at least one first raw material comprises propylene and hydrogen, the at least one second raw material comprises propylene, ethylene and hydrogen, The target value and predicted value of the molecular weight characteristic correspondingly relate to at least one of the first number average molecular weight Mn_PP, the first weight average molecular weight Mw_PP, the first molecular weight distribution index PDI_PP, and the first mass percentage PP_per of polypropylene, the second number average molecular weight Mn_iPP, the second weight average molecular weight Mw_iPP, and the second molecular weight distribution index PDI_iPP of isotactic polypropylene, the third number average molecular weight Mn_aPP, the third weight average molecular weight Mw_aPP, and the third molecular weight distribution index PDI_aPP of random polypropylene, the fourth number average molecular weight Mn_EPR, the fourth weight average molecular weight Mw_EPR, the fourth molecular weight distribution index PDI_EPR, and the fourth mass percentage EPR_per of ethylene-propylene random copolymer, the first ethylene molar content E_mole_EPR, the fifth number average molecular weight Mn_EPS, the fifth weight average molecular weight Mw_EPS, the fifth molecular weight distribution index PDI_EPS, the fifth mass percentage EPS_per, the second ethylene molar content E_mole_EPS, and the third ethylene molar content E_in_IPC in the polymer of ethylene-propylene multi-block copolymer.

3. The prediction method according to claim 1 or 2, characterized in that: The step of determining the predicted value of the molecular weight characteristic of the impact-resistant polypropylene product according to the process parameters comprises: The process parameters are input into a pre-trained BP neural network model to determine the predicted value of the molecular weight characteristics of the impact-resistant polypropylene product via the BP neural network model.

4. The prediction method according to claim 3, characterized in that: The steps of training the BP neural network model include: Acquire multiple groups of historical sample data of processing impact-resistant polypropylene to construct a historical sample data set, wherein each of the historical sample data includes a process parameter and a true value of a molecular weight characteristic of a corresponding historical sample; Building a BP neural network model to be trained, wherein the BP neural network model consists of an input layer, two hidden layers and an output layer; and Each group of training sample data in the historical sample data set is input into the BP neural network model in turn to obtain the corresponding molecular weight characteristic prediction value, and according to the difference between the predicted value and the true value of the molecular weight characteristic of the corresponding training sample, the learning parameters of the BP neural network model are adjusted to train the BP neural network model.

5. The prediction method according to claim 4, characterized in that: Before inputting each training sample in the historical sample data set into the BP neural network model in sequence, the prediction method further includes the following steps: Each historical sample data in the historical sample data set is normalized to normalize its value between -1 and 1: Among them, X reduced,p,i is the normalized variable of the pth historical sample data of the ith neuron, X p,i is the original variable of the pth historical sample data of the ith neuron, X min,i is the minimum value of the variable of the ith neuron in the historical sample data set, X max,i is the maximum value of the variable of the ith neuron in the historical sample data set.

6. The prediction method according to claim 4, characterized in that: The step of sequentially inputting each training sample data in the historical sample data set into the BP neural network model to obtain the corresponding molecular weight characteristic prediction value comprises: The historical sample data set is randomly divided into a training set and a test set according to a preset ratio; defining the historical sample data in the training set as training sample data; and Each group of training sample data in the training set is sequentially input into the BP neural network model to obtain the corresponding molecular weight characteristic prediction value.

7. The prediction method according to claim 6, characterized in that: The following steps are also included: The historical sample data in the test set is defined as test sample data; In response to completing the training of the BP neural network model, inputting each group of test sample data in the test set into the BP neural network model in sequence to obtain corresponding molecular weight characteristic test values; and The validity of the BP neural network model is verified based on the degree of agreement between the regression curve of each test value and the true value of the corresponding test sample.

8. The prediction method according to claim 4, characterized in that: The number of neurons in the input layer is N i The number of neurons N in the output layer is determined according to the input dimension of each group of training sample data. o is determined according to the output dimension of the molecular weight characteristic prediction value, and the number of neurons in the hidden layer is determined according to the number of neurons N in the input layer. i , the number of neurons in the output layer N o , and the number of groups N of the training sample data s Sure: The value of α depends on the type of molecular weight property to be predicted, and ranges from 4 to 10.

9. The prediction method according to claim 4, characterized in that: The step of sequentially inputting each group of training sample data in the historical sample data set into the BP neural network model to obtain the corresponding molecular weight characteristic prediction value comprises: The weight w of the connection between neurons i and j in the BP neural network model j,i Initialize to a normal distribution: Among them, U is the normal distribution function, n in is the number of rows in the training sample data matrix, n out is the number of columns in the training sample data matrix; The threshold value B of each neuron i in the BP neural network model i Initialized to 0; According to the weights of the connections between the neurons and the thresholds of the neurons in each layer of the neural network, the net input variables of each layer of the neural network are calculated: in, is the net input variable of the pth data of the jth neuron of the k+1th layer of neural network, is the threshold of the jth neuron in the k+1th layer of the neural network, is the weight of the connection between the jth neuron in the k+1th layer of the neural network and the ith neuron in the kth layer of the neural network, is the output variable of the pth data of the ith neuron of the kth layer of the neural network. The output variable of the jth neuron of the k+1th layer of the neural network is calculated by transferring the net input variable of the jth neuron of the k+1th layer of the neural network according to the custom activation function: in, is the net input variable of the pth data of the jth neuron of the k+1th layer of neural network, is the output variable of the pth data of the jth neuron in the k+1th layer of the neural network, and the value of α1 is 1.

5.

10. The prediction method according to claim 9, characterized in that: The step of adjusting the learning parameters of the BP neural network model according to the difference between the actual value of the molecular weight characteristic of the corresponding training sample to train the BP neural network model comprises: Calculate the mean square error between the predicted value and the true value of the molecular weight characteristic of each training sample: in, is the true value of the pth training sample of the jth neuron, is the predicted value of the jth neuron for the pth training sample; and The Adam optimizer is used to perform back propagation based on the mean square error on the BP neural network model until the training of the BP neural network model is completed.

11. The prediction method according to claim 10, characterized in that: The step of using the Adam optimizer to perform back propagation based on the mean square error on the BP neural network model until the training of the BP neural network model is completed includes: Set the maximum number of iterations t for training the BP neural network model max , an indicator parameter θ1, and a patience parameter to prevent overfitting in training, wherein the value of the patience parameter is calculated according to the formula θ =max(input_n,output_n) / 2, input_n is the number of input nodes, output_n is the number of output nodes, and the initial value of the indicator parameter θ1 is 0; In response to the current round number t not reaching the maximum iteration round number t max , and the indicator parameter θ1 is less than or equal to the patience parameter, the first-order moment estimate m of the Adam optimizer t and the second-order moment estimate v t Combined together, the weights of the connections between neurons in the BP neural network model and the thresholds of the neurons are updated: Among them, w t , b t is the weight and threshold after the tth round update, w t-1 , b t-1 is the weight and threshold of the t-1th round, α2 is the learning rate, θ2 is the anti-zero constant, and the first-order moment estimate m of the Adam optimizer t The expression is: m t =β1m t-1 +(1-β1)h t Among them, m t-1 and m t They represent the first-order moment estimates before and after iteration, β1 is the exponentially weighted average parameter, and h t represents the gradient of the weight ω or threshold b, The second-order moment estimate v of the Adam optimizer t The expression is: v t =β2v t-1 +(1-β2)h t 2 Among them, v t-1 and v t are the second-order moment estimates before and after iteration, β2 is the exponentially weighted average parameter, and h t 2 Represents the square of the gradient of the weight ω or the threshold b; Determine whether the mean square error of the current round increases, if so, add 1 to the indicator parameter θ1, otherwise, set the indicator parameter θ1 to zero; and In response to the current round number t reaching the maximum iteration round number t max , or the indication parameter θ1 is greater than the patience parameter, it is determined that the training of the BP neural network model is completed.

12. A prediction system for the microstructure of impact-resistant polypropylene, characterized in that: include: a memory having computer instructions stored thereon; and A processor is connected to the memory and is configured to execute computer instructions stored in the memory to implement the method for predicting the microstructure of impact-resistant polypropylene according to any one of claims 1 to 11.

13. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the computer instructions are executed by a processor, the method for predicting the microstructure of impact-resistant polypropylene according to any one of claims 1 to 11 is implemented.

Citation Information

Patent Citations

  • On-line estimation system and method for gas-phase copolymerization impact-resistant polypropylene quality indexes

    CN104102182A

  • Rapid modelling method for polyolefin molecular weight distribution in industrial production conditions

    CN104112082A

  • High polymer material service life prediction method based on environmental big data and machine learning

    CN112331281A

  • method of predicting properties of polymer product

    KR100829706B1

  • Method and system for polypropylene and polypropylene article production modeling using artificial intelligence algorithms

    US20210110891A1

Cited By

  • Impact polypropylene microstructure prediction method and system, and storage medium

    WO2026031777A1