A method, apparatus, storage medium, and terminal for predicting membrane performance based on a joint model.

Abstract

This invention discloses a method, apparatus, storage medium, and terminal for predicting membrane performance based on a joint model. The method includes: obtaining the actual condition parameters required to generate a PVDF-Ni membrane under orthogonal array conditions; inputting the actual condition parameters into a pre-trained membrane performance prediction model; wherein the membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model; and outputting the predicted flux and predicted rejection rate corresponding to the actual condition parameters. Because this application uses an RSM model and an ANN model to construct an RSM-ANN joint model, and uses existing data to train the RSM-ANN joint model to obtain a pre-trained membrane performance prediction model, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a method, apparatus, storage medium and terminal for predicting membrane performance based on a joint model. Background Technology

[0002] Membrane treatment plays an increasingly important role in water treatment. However, membrane fouling hinders its application. Modified membranes offer advantages such as antifouling properties, high rejection rates, and high flux. Electroless nickel-plated modified membranes, in particular, have been extensively studied recently. However, a suitable method for selecting the optimal modification conditions for electroless nickel-plated modified membranes is currently lacking. Furthermore, it is impossible to predict the membrane performance under different modification conditions.

[0003] Currently, the prediction of membrane performance still relies on electron microscopy to study membrane morphology or on extensive experimental testing to determine membrane performance. Optimal performance obtained under limited experimental conditions is not accurate enough, and extensive experiments not only waste human and material resources but also cause environmental pollution. Summary of the Invention

[0004] This application provides a method, apparatus, storage medium, and terminal for predicting membrane performance based on a joint model. To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments. Its sole purpose is to present some concepts in a simple form as a prelude to the detailed description that follows.

[0005] In a first aspect, embodiments of this application provide a membrane performance prediction method based on a joint model, the method comprising:

[0006] Obtain the actual condition parameters required to generate PVDF-Ni films under orthogonal array conditions;

[0007] The actual condition parameters are input into the pre-trained membrane performance prediction model; the membrane performance prediction model is a joint model constructed based on the RSM model and the ANN model.

[0008] Output the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

[0009] Optionally, a pre-trained membrane performance prediction model is generated by following these steps:

[0010] Identify and select multiple parameters that affect the coating process;

[0011] Raw data were collected based on the orthogonal experimental table;

[0012] The selected parameters and the collected raw data are divided into a first number of training sets and a second number of test sets; wherein the first number is greater than the second number.

[0013] A membrane performance prediction model was constructed using RSM and ANN models.

[0014] Input the first set of training data into the membrane performance prediction model and output the model loss value;

[0015] A pre-trained membrane performance prediction model is generated based on a second set of test data and the model loss value.

[0016] Optionally, a pre-trained membrane performance prediction model is generated based on a second set of test data and the model loss values, including:

[0017] The second set of test data is input into the pre-trained membrane performance prediction model to evaluate the training effect, and the model evaluation results are output.

[0018] When the model evaluation results meet the preset evaluation values, a pre-trained membrane performance prediction model is generated based on the model loss value.

[0019] Optionally, a pre-trained membrane performance prediction model is generated based on the model loss value, including:

[0020] When the model loss value reaches the preset loss value and the number of training iterations reaches the preset number of training iterations, a pre-trained membrane performance prediction model is generated; or,

[0021] When the model loss value does not reach the preset loss value, the model loss value is backpropagated to update the model parameters, and the step of inputting the first number of training sets into the membrane performance prediction model continues.

[0022] Optionally, before obtaining the actual condition parameters required to generate PVDF-Ni films under orthogonal array conditions, the following steps are also included:

[0023] Commercial PVDF membranes were pretreated to obtain PVDF-Ni membranes;

[0024] Calculate the actual flux and actual rejection rate of the PVDF-Ni membrane; where,

[0025] The formula for calculating the actual rejection rate is: C0 represents the absorbance of the solution before filtration, and C1 represents the absorbance of the solution after filtration.

[0026] Optionally, the method also includes:

[0027] Calculate the difference between the actual flux and the predicted flux;

[0028] Calculate the difference between the actual rejection rate and the predicted rejection rate;

[0029] When the difference is within the preset difference range, it is determined that the pre-trained membrane performance prediction model conforms to the actual application scenario.

[0030] Optionally, the RSM model can analyze the interaction between parameters; the ANN model is used to accurately predict membrane performance; the ANN model has a BP neural network structure, and the step size of the ANN model is dynamically adjusted using the LM algorithm and the SCG algorithm.

[0031] Secondly, embodiments of this application provide a membrane performance prediction device based on a joint model, the device comprising:

[0032] The parameter acquisition module is used to acquire the actual condition parameters required to generate PVDF-Ni films under orthogonal array conditions.

[0033] The parameter input module is used to input actual condition parameters into a pre-trained membrane performance prediction model; the membrane performance prediction model is a joint model constructed based on the RSM model and the ANN model.

[0034] The performance output module is used to output the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

[0035] Thirdly, embodiments of this application provide a computer storage medium storing multiple instructions adapted for loading and execution of the above-described method steps by a processor.

[0036] Fourthly, embodiments of this application provide a terminal that may include: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and executed by the above-described method steps.

[0037] The technical solutions provided in this application embodiment may include the following beneficial effects:

[0038] In this embodiment, the membrane performance prediction device based on a joint model first obtains the actual condition parameters required to generate a PVDF-Ni membrane under orthogonal array conditions, and then inputs the actual condition parameters into a pre-trained membrane performance prediction model. The membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model. Finally, it outputs the predicted flux and predicted rejection rate corresponding to the actual condition parameters. Because this application uses an RSM model and an ANN model to construct an RSM-ANN joint model, and uses existing data to train the RSM-ANN joint model to obtain a pre-trained membrane performance prediction model, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources.

[0039] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0040] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0041] Figure 1 This is a schematic flowchart of a membrane performance prediction method based on a joint model provided in an embodiment of this application;

[0042] Figure 2 This is a comparison chart of the predicted flux of modified membrane by RSM and the actual measured value provided in the embodiments of this application;

[0043] Figure 3 This is a comparison chart of the predicted and actual measured values ​​of the RSM for the retention rate of the modified membrane provided in the embodiments of this application;

[0044] Figure 4 This is a schematic block diagram of a membrane performance prediction process provided in an embodiment of this application;

[0045] Figure 5 This is a flowchart illustrating a membrane performance prediction model training method provided in an embodiment of this application.

[0046] Figure 6 This is a structural diagram of a membrane performance prediction function provided in an embodiment of this application;

[0047] Figure 7 This is a schematic diagram of the structure of a membrane performance prediction device based on a joint model provided in an embodiment of this application;

[0048] Figure 8 This is a schematic diagram of the structure of a terminal provided in an embodiment of this application. Detailed Implementation

[0049] The following description and accompanying drawings fully illustrate specific embodiments of the invention to enable those skilled in the art to practice them.

[0050] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0051] In the following description, when referring to the accompanying drawings, the same numbers in different drawings denote the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0052] In the description of this invention, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of these terms in this invention based on the specific circumstances. Furthermore, in the description of this invention, unless otherwise stated, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0053] This application provides a method, apparatus, storage medium, and terminal for predicting membrane performance based on a joint model, to address the problems existing in the aforementioned related technologies. In the technical solution provided by this application, because an RSM-ANN joint model is constructed using an RSM model and an ANN model, and a pre-trained membrane performance prediction model is obtained by training the RSM-ANN joint model using existing data, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources. The following detailed description uses exemplary embodiments.

[0054] The following will be combined with the appendix Figure 1 -Appendix Figure 6 This application provides a detailed description of the membrane performance prediction method based on a joint model, as described in its embodiments. This method can be implemented using a computer program and can run on a joint model-based membrane performance prediction device based on the von Neumann architecture. The computer program can be integrated into the application or run as a standalone utility application.

[0055] Please see Figure 1 This document provides a flowchart illustrating a membrane performance prediction method based on a joint model, as described in an embodiment of this application. Figure 1 As shown, the method in this application embodiment may include the following steps:

[0056] S101, Obtain the actual condition parameters required to generate PVDF-Ni film under orthogonal array conditions;

[0057] In this embodiment of the application, before obtaining the actual condition parameters required to generate the PVDF-Ni membrane under orthogonal array conditions, it is necessary to preprocess the commercial PVDF membrane to obtain the PVDF-Ni membrane, and then calculate the actual flux and actual rejection rate of the PVDF-Ni membrane; wherein,

[0058] The formula for calculating the actual rejection rate is: C0 represents the absorbance of the solution before filtration, and C1 represents the absorbance of the solution after filtration.

[0059] It should be noted that the membrane rejection rate was determined by analyzing the absorbance measured by a UV spectrophotometer before and after filtration.

[0060] Specifically, during the pretreatment of commercial PVDF membranes, the membranes were cleaned and then ultrasonically cleaned in pure water. After ultrasonic cleaning, the membranes were first ultrasonically cleaned in AgNO3 solution, then allowed to stand for 20 minutes to ensure sufficient Ag+ ions on each PVDF membrane. For the electroplating solution pretreatment, a solution containing NiSO4·6H2O, Na4P2O7·10H2O, and C2H10BN was ultrasonically treated to promote thorough mixing of the components. The amount of NH3·H2O added was determined according to the orthogonal experimental table and added 5 minutes before the end of ultrasonication. After ultrasonication, the solution was placed in a constant temperature water bath and allowed to stand until the temperature reached the orthogonal table's temperature. A thermometer was used to measure whether the solution actually reached the experimental temperature. For the preparation of PVDF-Ni, the PVDF membranes were removed from the AgNO3 solution and placed in the electroplating solution at the required temperature. The PVDF-Ni membranes were then removed sequentially according to different nickel plating times.

[0061] In one possible implementation, after obtaining the actual flux and actual rejection rate of the PVDF-Ni membrane, it is necessary to use a pre-trained membrane performance prediction model for prediction. The model prediction results are then compared and analyzed with the actual flux and actual rejection rate to determine whether the model can accurately predict in actual application scenarios.

[0062] Furthermore, when making predictions using a pre-trained membrane performance prediction model, it is first necessary to obtain the actual condition parameters required to generate PVDF-Ni membranes under the orthogonal array conditions in the above-mentioned actual measurement experiments.

[0063] S102, Input the actual condition parameters into the pre-trained membrane performance prediction model;

[0064] Among them, the membrane performance prediction model is a joint model constructed based on the RSM model and the ANN model;

[0065] Typically, RSM (Research-Based Modeling) is a tool that relies on statistical analysis models to analyze experimental parameters. RSM can build models to analyze the individual effects of factors or the interactions between factors within a limited number of trials. ANN (Analogous Neural Networks) is a mathematical model that simulates the activity of human neurons and has good predictive power for nonlinear data with unclear connections. Using ANN in conjunction with RSM will improve the accuracy of predictions.

[0066] In this embodiment, a pre-trained membrane performance prediction model can be generated according to the following steps: First, multiple parameters affecting the coating process are determined and selected. Then, raw data is collected according to the orthogonal experimental table. The selected multiple parameters and the collected raw data are divided into a first number of training sets and a second number of test sets, wherein the first number is greater than the second number. Then, the membrane performance prediction model is constructed using the RSM model and the ANN model. Finally, the first number of training sets is input into the membrane performance prediction model, the model loss value is output, and the pre-trained membrane performance prediction model is generated according to the second number of test sets and the model loss value.

[0067] It should be noted that the RSM model can analyze the interactions between various parameters; the ANN model is used to accurately predict membrane performance; the ANN model has a BP neural network structure, and the step size of the ANN model is dynamically adjusted using the LM algorithm and the SCG algorithm. Both the LM and SCG algorithms are iterative methods for finding the extrema of a function. Compared to the original BP-ANN, their advantage is that they do not require setting the step size. When the step size decreases too quickly, the LM-ANN uses a small step size to make the result closer to Newton's method; when the step size decreases too slowly, it uses a large step size to make the result closer to gradient descent. The step size of the SCG is also dynamically adjusted. ANNs with different hidden layers are used to predict membrane performance.

[0068] In one possible implementation, after obtaining the actual condition parameters required to generate the PVDF-Ni membrane under the orthogonal array conditions according to step S101, these parameters can be input into a pre-trained membrane performance prediction model for model processing to predict the membrane performance. The membrane performance includes predicted flux and predicted rejection rate.

[0069] S103 outputs the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

[0070] In one possible implementation, after the model processing is complete, the model can output the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

[0071] Furthermore, a comparative analysis can be performed between the actual and predicted results. First, the difference between the actual flux and the predicted flux is calculated. Then, the difference between the actual rejection rate and the predicted rejection rate is calculated. Finally, when the difference is within a preset range, it is determined that the pre-trained membrane performance prediction model conforms to the actual application scenario. For example... Figure 2 The graph showing the comparison between the RSM's predicted flux of the modified membrane and the actual measured values, and... Figure 3 The graph shows a comparison between the RSM prediction and the actual measured value of the modified membrane retention rate.

[0072] For example Figure 4 As shown, Figure 4 This application provides a schematic flowchart of a membrane performance prediction process. First, training data is acquired and imported. Then, a joint model constructed based on the RSM model and the ANN model is used for model training. Specifically, this includes RSM analysis and prediction, as well as ANN training and analysis and prediction. Finally, after training, the trained model can be combined with data from real-world scenarios to predict membrane performance.

[0073] In this embodiment, the membrane performance prediction device based on a joint model first obtains the actual condition parameters required to generate a PVDF-Ni membrane under orthogonal array conditions, and then inputs the actual condition parameters into a pre-trained membrane performance prediction model. The membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model. Finally, it outputs the predicted flux and predicted rejection rate corresponding to the actual condition parameters. Because this application uses an RSM model and an ANN model to construct an RSM-ANN joint model, and uses existing data to train the RSM-ANN joint model to obtain a pre-trained membrane performance prediction model, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources.

[0074] Please see Figure 5 This is a flowchart illustrating a method for training a membrane performance prediction model, as provided in an embodiment of this application. Figure 5 As shown, the method in this application embodiment may include the following steps:

[0075] S201, Identify and select multiple parameters that affect the coating process;

[0076] Several parameters, including temperature, time, and ammonia concentration, are involved. The growth of the active layer on the membrane surface is affected by temperature, time, and ammonia concentration.

[0077] S202, Collect raw data according to the orthogonal experimental table;

[0078] The raw data includes, for example, the modification conditions and the properties of the modified membrane.

[0079] S203, the selected parameters and the collected raw data are divided into a first number of training sets and a second number of test sets; wherein the first number is greater than the second number;

[0080] Of these, the first-largest training set accounts for 70%, and the second-largest test set accounts for 30%.

[0081] S204, a membrane performance prediction model was constructed using the RSM model and the ANN model;

[0082] S205, input the first number of training sets into the membrane performance prediction model and output the model loss value;

[0083] S206, Generate a pre-trained membrane performance prediction model based on a second set of test data and the model loss value.

[0084] In one possible implementation, a second set of test data is first input into a pre-trained membrane performance prediction model to evaluate the training effect, and the model evaluation result is output. Then, when the model evaluation result meets the preset evaluation value, the pre-trained membrane performance prediction model is generated based on the model loss value.

[0085] Specifically, when generating a pre-trained membrane performance prediction model based on the model loss value, the pre-trained membrane performance prediction model is generated first when the model loss value reaches the preset loss value and the number of training iterations reaches the preset number of training iterations; or, when the model loss value does not reach the preset loss value, the model loss value is backpropagated to update the model parameters, and the step of inputting the first number of training sets into the membrane performance prediction model continues.

[0086] For example Figure 6 As shown, a functional structure diagram for membrane performance prediction is provided. The RSM-ANN joint system is used for membrane performance prediction. The specific structure of the RSM-ANN joint system includes a data import / export unit, an RSM-ANN training unit, and a training result evaluation unit. The data import / export unit can import experimental conditions affecting the modified membrane and their corresponding membrane properties, and export the predicted modified membrane properties. The RSM-ANN training unit is further subdivided into an RSM analysis unit and an ANN analysis unit. RSM dynamically studies the interactions between various factors, while ANN is used to accurately predict membrane performance. The training result evaluation unit is further subdivided into a data analysis unit and a prediction unit, which analyze the model's stability and prediction accuracy. Finally, the predicted data can be exported and saved in an Excel spreadsheet.

[0087] In this embodiment, the ANN used to predict membrane performance employs a backpropagation (BP) neural network, specifically the LM and SCG algorithms. Both the LM and SCG algorithms are iterative methods for finding function extrema. Compared to the original BP-ANN, their advantage is that they do not require setting a step size. The LM-ANN uses a small step size when the step size decreases too quickly, making the result closer to Newton's method; when the step size decreases too slowly, it uses a large step size, making the result closer to gradient descent. The step size of the SCG algorithm is also dynamically adjusted. ANNs with different hidden layers are used to predict membrane performance.

[0088] In this embodiment, the membrane performance prediction device based on a joint model first obtains the actual condition parameters required to generate a PVDF-Ni membrane under orthogonal array conditions, and then inputs the actual condition parameters into a pre-trained membrane performance prediction model. The membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model. Finally, it outputs the predicted flux and predicted rejection rate corresponding to the actual condition parameters. Because this application uses an RSM model and an ANN model to construct an RSM-ANN joint model, and uses existing data to train the RSM-ANN joint model to obtain a pre-trained membrane performance prediction model, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources.

[0089] The following are embodiments of the apparatus of the present invention, which can be used to execute embodiments of the method of the present invention. For details not disclosed in the embodiments of the apparatus of the present invention, please refer to the embodiments of the method of the present invention.

[0090] Please see Figure 7 This diagram illustrates a structural schematic of a membrane performance prediction device based on a joint model, provided by an exemplary embodiment of the present invention. This joint model-based membrane performance prediction device can be implemented as all or part of a terminal through software, hardware, or a combination of both. The device 1 includes a parameter acquisition module 10, a parameter input module 20, and a performance output module 30.

[0091] Parameter acquisition module 10 is used to acquire the actual condition parameters required to generate PVDF-Ni film under orthogonal array conditions;

[0092] The parameter input module 20 is used to input actual condition parameters into a pre-trained membrane performance prediction model; wherein, the membrane performance prediction model is a joint model constructed based on the RSM model and the ANN model;

[0093] The performance output module 30 is used to output the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

[0094] It should be noted that the membrane performance prediction device based on the joint model provided in the above embodiments is only illustrated by the division of the above functional modules when executing the membrane performance prediction method based on the joint model. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the membrane performance prediction device based on the joint model provided in the above embodiments and the membrane performance prediction method based on the joint model are based on the same concept, and the implementation process is detailed in the method embodiments, which will not be repeated here.

[0095] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0096] In this embodiment, the membrane performance prediction device based on a joint model first obtains the actual condition parameters required to generate a PVDF-Ni membrane under orthogonal array conditions, and then inputs the actual condition parameters into a pre-trained membrane performance prediction model. The membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model. Finally, it outputs the predicted flux and predicted rejection rate corresponding to the actual condition parameters. Because this application uses an RSM model and an ANN model to construct an RSM-ANN joint model, and uses existing data to train the RSM-ANN joint model to obtain a pre-trained membrane performance prediction model, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources.

[0097] The present invention also provides a computer-readable medium having program instructions stored thereon, which, when executed by a processor, implement the membrane performance prediction method based on the joint model provided in the above-described method embodiments. The present invention also provides a computer program product containing instructions that, when run on a computer, causes the computer to execute the membrane performance prediction method based on the joint model of the above-described method embodiments.

[0098] Please see Figure 8 This is a schematic diagram of the structure of a terminal provided in an embodiment of this application. Figure 8 As shown, terminal 1000 may include: at least one processor 1001, at least one network interface 1004, user interface 1003, memory 1005, and at least one communication bus 1002.

[0099] The communication bus 1002 is used to realize the connection and communication between these components.

[0100] The user interface 1003 may include a display screen and a camera. Optionally, the user interface 1003 may also include a standard wired interface and a wireless interface.

[0101] The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0102] The processor 1001 may include one or more processing cores. The processor 1001 connects to various parts within the electronic device 1000 using various interfaces and lines. It executes various functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in the memory 1005, and by calling data stored in the memory 1005. Optionally, the processor 1001 may be implemented using at least one hardware form selected from Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 1001 may integrate one or more of the following: a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), and a modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content to be displayed on the screen; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip, without being integrated into the processor 1001.

[0103] The memory 1005 may include random access memory (RAM) or read-only memory. Optionally, the memory 1005 may include a non-transitory computer-readable storage medium. The memory 1005 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 1005 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 1005 may also be at least one storage device located remotely from the aforementioned processor 1001. Figure 8 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a membrane performance prediction application based on a joint model.

[0104] exist Figure 8 In the terminal 1000 shown, the user interface 1003 is mainly used to provide an input interface for the user and to obtain the user's input data; while the processor 1001 can be used to call the membrane performance prediction application based on the joint model stored in the memory 1005, and specifically perform the following operations:

[0105] Obtain the actual condition parameters required to generate PVDF-Ni films under orthogonal array conditions;

[0106] The actual condition parameters are input into the pre-trained membrane performance prediction model; the membrane performance prediction model is a joint model constructed based on the RSM model and the ANN model.

[0107] Output the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

[0108] In one embodiment, when generating a pre-trained membrane performance prediction model, the processor 1001 specifically performs the following operations:

[0109] Identify and select multiple parameters that affect the coating process;

[0110] Raw data were collected based on the orthogonal experimental table;

[0111] The selected parameters and the collected raw data are divided into a first number of training sets and a second number of test sets; wherein the first number is greater than the second number.

[0112] A membrane performance prediction model was constructed using RSM and ANN models.

[0113] Input the first set of training data into the membrane performance prediction model and output the model loss value;

[0114] A pre-trained membrane performance prediction model is generated based on a second set of test data and the model loss value.

[0115] In one embodiment, when the processor 1001 generates a pre-trained membrane performance prediction model based on a second number of test sets and model loss values, it specifically performs the following operations:

[0116] The second set of test data is input into the pre-trained membrane performance prediction model to evaluate the training effect, and the model evaluation results are output.

[0117] When the model evaluation results meet the preset evaluation values, a pre-trained membrane performance prediction model is generated based on the model loss value.

[0118] In one embodiment, when the processor 1001 executes the generation of a pre-trained membrane performance prediction model based on the model loss value, it specifically performs the following operations:

[0119] When the model loss value reaches the preset loss value and the number of training iterations reaches the preset number of training iterations, a pre-trained membrane performance prediction model is generated; or,

[0120] When the model loss value does not reach the preset loss value, the model loss value is backpropagated to update the model parameters, and the step of inputting the first number of training sets into the membrane performance prediction model continues.

[0121] In one embodiment, before the processor 1001 performs the actual condition parameters required to generate the PVDF-Ni film under orthogonal array conditions, it also performs the following operations:

[0122] Commercial PVDF membranes were pretreated to obtain PVDF-Ni membranes;

[0123] Calculate the actual flux and actual rejection rate of the PVDF-Ni membrane; where,

[0124] The formula for calculating the actual rejection rate is: C0 represents the absorbance of the solution before filtration, and C1 represents the absorbance of the solution after filtration.

[0125] In one embodiment, the processor 1001 also performs the following operations:

[0126] Calculate the difference between the actual flux and the predicted flux;

[0127] Calculate the difference between the actual rejection rate and the predicted rejection rate;

[0128] When the difference is within the preset difference range, it is determined that the pre-trained membrane performance prediction model conforms to the actual application scenario.

[0129] In this embodiment, the membrane performance prediction device based on a joint model first obtains the actual condition parameters required to generate a PVDF-Ni membrane under orthogonal array conditions, and then inputs the actual condition parameters into a pre-trained membrane performance prediction model. The membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model. Finally, it outputs the predicted flux and predicted rejection rate corresponding to the actual condition parameters. Because this application uses an RSM model and an ANN model to construct an RSM-ANN joint model, and uses existing data to train the RSM-ANN joint model to obtain a pre-trained membrane performance prediction model, the pre-trained membrane performance prediction model can predict membrane performance with fewer experimental parameters, thereby improving the efficiency of membrane performance determination and saving significant manpower and resources.

[0130] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program for predicting membrane performance based on the joint model can be stored in a computer-readable storage medium. When executed, the program can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory, or random access memory, etc.

[0131] The above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application shall still fall within the scope of this application.

Claims

1. A membrane performance prediction method based on a joint model, characterized in that, The method includes: Commercial PVDF membranes were pretreated to obtain PVDF-Ni membranes; Calculate the actual flux and actual rejection rate of the PVDF-Ni membrane; Obtain the actual condition parameters required to generate PVDF-Ni films under orthogonal array conditions; The actual condition parameters are input into a pre-trained membrane performance prediction model; wherein the membrane performance prediction model is a joint model constructed based on an RSM model and an ANN model; wherein the RSM model can analyze the interaction between the parameters; and the ANN model is used to accurately predict membrane performance. Generate a pre-trained membrane performance prediction model according to the following steps: Multiple parameters affecting the coating process are identified and selected; raw data are collected according to an orthogonal experimental table; the selected parameters and the collected raw data are divided into a first number of training sets and a second number of test sets, wherein the first number is greater than the second number; a membrane performance prediction model is constructed using an RSM model and an ANN model; the first number of training sets is input into the membrane performance prediction model, and the model loss value is output; a pre-trained membrane performance prediction model is generated based on the second number of test sets and the model loss value. Output the predicted flux and predicted rejection rate corresponding to the actual condition parameters; By comparing and analyzing the model's predictions with the actual flux and rejection rate, we can determine whether the model can accurately predict in real-world application scenarios.

2. The method according to claim 1, characterized in that, The step of generating a pre-trained membrane performance prediction model based on the second number of test sets and the model loss value includes: The second number of test sets are input into the pre-trained membrane performance prediction model for training effect evaluation, and the model evaluation results are output. When the model evaluation result meets the preset evaluation value, a pre-trained membrane performance prediction model is generated based on the model loss value.

3. The method according to claim 2, characterized in that, The step of generating a pre-trained membrane performance prediction model based on the model loss value includes: When the model loss value reaches a preset loss value and the number of training iterations reaches a preset number of training iterations, a pre-trained membrane performance prediction model is generated; or, When the model loss value does not reach the preset loss value, the model loss value is backpropagated to update the model parameters, and the step of inputting the first number of training sets into the membrane performance prediction model continues.

4. The method according to claim 1, characterized in that, The formula for calculating the actual retention rate is as follows: Retention rate = The C 0 represents the absorbance of the solution before filtration. C 1 represents the absorbance of the solution after filtration.

5. The method according to claim 1, characterized in that, Whether the judgment model can accurately predict in real-world application scenarios includes: Calculate the difference between the actual flux and the predicted flux; Calculate the difference between the actual rejection rate and the predicted rejection rate; When the difference is within a preset range, the pre-trained membrane performance prediction model is determined to be consistent with the actual application scenario.

6. The method according to claim 1, characterized in that, The ANN model has a BP neural network structure, and the step size of the ANN model is dynamically adjusted using the LM algorithm and the SCG algorithm.

7. A membrane performance prediction device based on a joint model implemented using the method described in any one of claims 1-6, characterized in that, The device includes: The parameter acquisition module is used to acquire the actual condition parameters required to generate PVDF-Ni films under orthogonal array conditions. The parameter input module is used to input the actual condition parameters into a pre-trained membrane performance prediction model; wherein the membrane performance prediction model is a joint model constructed based on the RSM model and the ANN model; The performance output module is used to output the predicted flux and predicted rejection rate corresponding to the actual condition parameters.

8. A computer storage medium, characterized in that, The computer storage medium stores a plurality of instructions adapted for loading by a processor and executing the method steps as claimed in any one of claims 1-6.

9. A terminal, characterized in that, include: A processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to execute the method steps as claimed in any one of claims 1-6.

Images