Data processing method for assisting in improving yield of multi-step synthesized PVB (polyvinyl butyral)

Abstract

The invention discloses a data processing method for assisting in improving the yield of multi-step synthesized PVB (polyvinyl butyral). The method comprises the following steps: acquiring PVB production process parameter information; pre-processing the PVB production process parameter information to obtain target production process parameter information; and performing identification processing on the target production parameter information by using the target parameter identification model to obtain target identification production process parameter information.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a data processing method for assisting in improving the yield of multi-step synthesis of PVB. Background Art

[0002] Polyvinyl butyral (PVB) is an important polymer material, which is widely used in fields such as automotive safety glass and architectural glass. Its synthesis process is complex, involving multiple reaction steps and the interaction of various chemical substances. The traditional PVB synthesis process mainly relies on experienced operators to manually control reaction conditions such as temperature, pressure, and reaction time. However, this method has many limitations: First, the accuracy of manual control is limited, and it is difficult to precisely adjust the optimal conditions for each reaction step, resulting in large fluctuations in product quality and making it difficult to meet the increasingly strict performance requirements; Second, for the complex reaction mechanism in the multi-step synthesis process, it is difficult for humans to accurately identify and analyze the interrelationships between various key parameters, which limits the optimization and improvement of the synthesis process; In addition, with the expansion of production scale, the efficiency of manual control is low, making it difficult to meet the needs of large-scale production, increasing production costs and time. However, existing data processing methods mostly focus on single reaction steps or simple reaction systems, and there are obvious deficiencies in their applicability to complex reaction processes such as multi-step synthesis of PVB. On the one hand, the amount of data generated during the PVB synthesis process is huge, including one-dimensional data (such as sequence data of temperature, pressure, etc. changing with time) and two-dimensional data (such as matrix data of the concentrations of various chemical substances in different reaction steps). Traditional data processing methods are difficult to effectively integrate and analyze these multi-dimensional data; On the other hand, there are many key parameters in the PVB synthesis process and they are interrelated. Traditional data processing methods are difficult to accurately identify and extract the key parameters that have the greatest impact on the synthesis quality, resulting in low accuracy and reliability of data processing results. Therefore, providing a data processing method for assisting in improving the yield of multi-step synthesis of PVB to accurately identify and analyze the key parameters in the PVB synthesis process and improve the recognition accuracy and efficiency of key process parameters for multi-step synthesis of PVB has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a data processing method for assisting in improving the yield of multi-step synthesis of PVB, which is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0004] To solve the above technical problems, a first aspect of an embodiment of the present invention discloses a data processing method, the method comprising: Obtain PVB production process parameter information; Perform preprocessing on the PVB production process parameter information to obtain target production process parameter information; Use a target parameter recognition model to perform recognition processing on the target production parameter information to obtain target recognized production process parameter information.

[0005] A second aspect of an embodiment of the present invention discloses a data processing device, the device comprising: An acquisition module, configured to obtain PVB production process parameter information; A first processing module, configured to perform preprocessing on the PVB production process parameter information to obtain target production process parameter information; A second processing module, configured to use a target parameter recognition model to perform recognition processing on the target production parameter information to obtain target recognized production process parameter information.

[0006] A third aspect of the present invention discloses another data processing device, the device comprising: A memory storing executable program code; A processor coupled to the memory; The processor calls the executable program code stored in the memory and executes some or all of the steps in the data processing method disclosed in the first aspect of an embodiment of the present invention.

[0007] A fourth aspect of the present invention discloses a computer-readable storage medium, the computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute some or all of the steps in the data processing method disclosed in the first aspect of an embodiment of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention, and for those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0009] Figure 1 is a schematic diagram of the scenario of the data processing system provided by an embodiment of the present invention; Figure 2 is a flowchart of a data processing method disclosed by an embodiment of the present invention; Figure 3 is a schematic structural diagram of a data processing device disclosed by an embodiment of the present invention; Figure 4 It is a schematic structural diagram of another data processing device disclosed in an embodiment of the present invention; Figure 5 It is a schematic structural diagram of a target parameter recognition model disclosed in an embodiment of the present invention; Figure 6 It is a schematic structural diagram of a feature analysis module disclosed in an embodiment of the present invention; Figure 7 It is a schematic structural diagram of a sequence data processing module disclosed in an embodiment of the present invention. Detailed implementation manners

[0010] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0011] The terms "first", "second", etc. in the specification and claims of the present invention and the above drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or equipment.

[0012] Referring to "embodiment" herein means that a specific feature, structure or characteristic described in connection with the embodiment can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0013] In this application, the term "exemplary" is used to mean "serving as an example, illustration, or instance". Any embodiment described as "exemplary" in this application is not necessarily to be construed as more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use this application. In the following description, details are set forth for purposes of explanation. It should be understood that those of ordinary skill in the art can recognize that this application can be implemented without these specific details. In other instances, well-known structures and processes are not elaborated in detail to avoid obscuring the description of this application with unnecessary details. Therefore, this application is not intended to be limited to the embodiments shown, but rather to be consistent with the broadest scope that conforms to the principles and features disclosed in this application.

[0014] It should be noted that since the method of the embodiment of this application is executed in a computer device, the processing objects of each computer device exist in the form of data or information. For example, time, which is actually time information. It can be understood that in subsequent embodiments, if dimensions, quantities, positions, etc. are mentioned, they are all corresponding data existences for the computer device to process, and specific details are not elaborated here.

[0015] A brief introduction to the artificial intelligence-related technologies that may be involved in this application is provided. Artificial Intelligence (AI) is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology in computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can respond in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines to enable the machines to have the functions of perception, reasoning, and decision-making.

[0016] Artificial intelligence technology is an interdisciplinary subject with a wide range of fields, including both hardware-level technologies and software-level technologies. The basic technologies of artificial intelligence generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, and mechatronics. The software technologies of artificial intelligence mainly include several major directions such as computer vision technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0017] Computer Vision Technology (CV) Computer vision is a science that studies how to enable machines to "see". More specifically, it refers to using cameras and computers to replace human eyes for tasks such as object recognition and measurement in machine vision, and further performing image processing to make the computer-processed images more suitable for human eyes to observe or be transmitted to instruments for detection. As a scientific discipline, computer vision studies related theories and technologies, and attempts to build artificial intelligence systems that can obtain information from images or multi-dimensional data. Computer vision technology usually includes image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, 3D object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous localization and mapping, etc. technologies, and also includes common biometric recognition technologies such as face recognition and fingerprint recognition.

[0018] Single-modal information is data of only one type, such as one of the data information types like text, image, audio, video, electromagnetic signal, etc. Multi-modal information is data information that includes at least two single-modal information types. Further, multi-modal information is applicable to complex tasks that require integrating multiple information sources, such as sentiment analysis, robot interaction, autonomous driving, etc. By integrating information of multiple modalities, higher performance and accuracy can usually be achieved in tasks.

[0019] A large model refers to an artificial neural network model with a very large number of parameters. In the field of artificial intelligence, a large model usually refers to a model with hundreds of millions to trillions of parameters. The model usually needs to be trained on a large-scale dataset and requires a large amount of computing resources for optimization and adjustment. Large models are usually used to solve complex natural language processing, computer vision, and speech recognition tasks. Generative AI is a type of AI that can create new content and ideas, including conversations, stories, images, videos, and music. In the embodiments of this application, the large model can be large language models such as ChatGPT, BERT, XLNet, Zhipu Model, Claude, Moonshot AI Model, ChatGLM Model, Tongwen Qianyi Model, MiniMax Model, Spark Model, Llama Model, 360GPT Model, Qwen Model, Baichuan Model, Lark Model, vivoLM Model, and Wenxin Yiyan, and the embodiments of this application do not make any limitations.

[0020] The embodiments of this application provide a data processing method, device, computer device, and computer-readable storage medium, which will be described in detail below respectively.

[0021] Please refer to Figure 1 , Figure 1A schematic diagram of the scenario of the data processing system provided by the embodiments of the present application. The data processing system may include a computer device 100, and a data processing device is integrated in the computer device 100, such as Figure 1 the computer device in

[0022] In the embodiments of the present application, the computer device 100 is mainly used to obtain PVB production process parameter information; Perform preprocessing on the PVB production process parameter information to obtain target production process parameter information; Use the target parameter recognition model to perform recognition processing on the target production parameter information to obtain target recognized production process parameter information.

[0023] It can accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters of multi-step PVB synthesis, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0024] In the embodiments of the present application, the computer device 100 may be an independent server, or a server network or server cluster composed of servers. For example, the computer device 100 described in the embodiments of the present application includes, but is not limited to, a computer, a network host, a single network server, a set of multiple network servers, or a cloud server composed of multiple servers. Among them, the cloud server is composed of a large number of computers or network servers based on cloud computing (Cloud Computing).

[0025] It can be understood that the computer device 100 used in the embodiments of the present application may be a device that includes both receiving and transmitting hardware, that is, a device having receiving and transmitting hardware capable of performing two-way communication on a two-way communication link. Such devices may include: cellular or other communication devices, which have a single-line display or a multi-line display or a cellular or other communication device without a multi-line display. Specifically, the computer device 100 may specifically be a desktop terminal or a mobile terminal, and the computer device 100 may specifically also be one of a mobile phone, a tablet computer, a notebook computer, etc.

[0026] Those skilled in the art can understand that Figure 1 the application environment shown in Figure 1 is only one application scenario of the solution of the present application, and does not constitute a limitation on the application scenario of the solution of the present application. Other application environments may also include more or fewer computer devices than those shown in Figure 1 Only 1 computer device is shown in . It can be understood that the data processing system may further include one or more other services, which are not specifically limited here.

[0027] In addition, asFigure 1 As shown, the data processing system may further include a memory 200 for storing data, such as image data, location information, etc.

[0028] It should be noted that Figure 1 The scenario schematic diagram of the data processing system shown is only an example. The data processing system and scenarios described in the embodiments of the present application are for more clearly explaining the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art know that with the evolution of the data processing system and the emergence of new business scenarios, the technical solutions provided by the embodiments of the present application are equally applicable to similar technical problems.

[0029] The present invention discloses a data processing method, which is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process. The following will be described in detail respectively.

[0030] Embodiment 1 Please refer to Figure 2 , Figure 2 which is a schematic flow chart of a data processing method disclosed in an embodiment of the present invention. Among them, Figure 2 The described data processing method is applied to a management system, such as a local server or a cloud server for management, etc., and the embodiments of the present invention do not make limitations. As Figure 2 shown, the data processing method may include the following operations: 101. Obtain PVB production process parameter information.

[0031] 102. Perform preprocessing on the PVB production process parameter information to obtain target production process parameter information.

[0032] 103. Use the target parameter recognition model to perform recognition processing on the target production parameter information to obtain target recognized production process parameter information.

[0033] It should be noted that the above-mentioned PVB production process parameter information represents real-time data in the production process with the help of DCS industrial Internet collection technology, which is not limited by the embodiment of the present invention. Further, the above-mentioned PVB production process parameter information includes one-dimensional production process parameter information and two-dimensional production process parameter information, which is not limited by the embodiment of the present invention. Further, the above-mentioned one-dimensional production process parameter information represents that in intermittent production, constant parameters such as feed amount, ambient temperature, annual expected production quality, etc. can be maintained, and time series recording is not required, which is not limited by the embodiment of the present invention. Further, the above-mentioned two-dimensional production process parameter information represents the non-fixed-length time series data recorded by real-time sensors in production (related to the current reaction time stage, such as temperature, pressure and other parameters that change with time during the PVB synthesis process), which is not limited by the embodiment of the present invention.

[0034] It should be noted that the above-mentioned target production process parameter information includes first target production process parameter information (corresponding to one-dimensional production process parameter information) and second target production process parameter information (corresponding to two-dimensional production process parameter information), which is not limited in the embodiment of the present invention.

[0035] It should be noted that the above two-dimensional production process parameter information includes two-dimensional parameter information distributed according to sampling time, which is not limited in the embodiment of the present invention. Further, the representation form of the two-dimensional parameter information is (sampling time, process parameter value), (if the process parameter value collected at the sampling time point is vacant, - is used to represent the vacancy), which is not limited in the embodiment of the present invention.

[0036] It should be noted that the above-mentioned one-dimensional production process parameter information includes one-dimensional parameter values ​​distributed according to sampling time (if the one-dimensional parameter value collected at the sampling time point is vacant, - is used to represent the vacancy), which is not limited in the embodiment of the present invention.

[0037] In this optional embodiment, as an optional implementation, the above-mentioned pre-processing of the PVB production process parameter information to obtain the target production process parameter information includes: Sequentially counting the non-missing values ​​of the one-dimensional production process parameter information and the two-dimensional production process parameter information to obtain a first statistical value and a second statistical value; Determine whether the first statistical value is greater than or equal to the parameter threshold, and obtain a first statistical determination result; When the first statistical judgment result is yes, the one-dimensional parameter values ​​corresponding to the blank representations in the one-dimensional production process parameter information are removed, and the one-dimensional parameter values ​​are reordered according to the sampling time to obtain the first target production process parameter information; When the first statistical judgment result is negative, the first gap-filling model is used to fill the gaps to obtain the first target production process parameter information; Among them, the first gap filling model is: BQ1=(max(z11,z12,z13)+max(y11,y12,y13)) / 2; Wherein, BQ1 represents the completed one-dimensional parameter value; z11, z12, z13 are vacant and represent the three one-dimensional parameter values ​​adjacent to the left; y11, y12, y13 are vacant and represent the three one-dimensional parameter values ​​adjacent to the right; Determine whether the second statistical value is greater than or equal to the parameter threshold, and obtain a second statistical determination result; When the second statistical judgment result is yes, the two-dimensional parameter information corresponding to the blank representation in the two-dimensional production process parameter information is eliminated, and the two-dimensional parameter information is reordered according to the sampling time to obtain the second target production process parameter information; When the second statistical judgment result is no, the first gap filling model is used to fill in the missing values ​​to obtain the first target production process parameter information; Among them, the second gap filling model is: BQ2=1.5*max(z21,z22,z23,z24)- min(z21,z22,z23,z24)+ 1.5*max(y21,y22,y23,y24) -min(y21,y22,y23,y24); In the formula, BQ2 represents the completed process parameter value; the vacancies of z21, z22, z23, and z24 represent the 4 process parameter values ​​adjacent to the left; the vacancies of y21, y22, y23, and y24 represent the 4 process parameter values ​​adjacent to the right.

[0038] It should be noted that the above-mentioned use of the first completion model to complete the one-dimensional data takes into account the time series representation of the one-dimensional data, which is independent of time. It uses the maximum value of the left and right adjacent data to take the average processing to obtain effective completion. The use of the second completion model to complete the two-dimensional data takes into account the influence of time factors. Its process parameter values ​​are closely related to time. Therefore, it is necessary to use more left and right adjacent data at the same time, and use the maximum and minimum values ​​to highlight the parameter changes caused by the timing changes. The embodiments of the present invention are not limited to this.

[0039] Further, when the left and right adjacent data are also missing, the missing value is skipped and the value is searched to the left or right. Further, when the left or right data is insufficient, 0 is used for filling, which is not limited in the embodiment of the present invention.

[0040] It should be noted that the above parameter thresholds can be set by the user or inferred by the large model based on historical parameter thresholds. The embodiments of the present invention do not make any limitations in this regard. Further, the parameter thresholds are not less than 100, and the embodiments of the present invention do not make any limitations in this regard. Further, the number of acquisition parameters is analyzed through the parameter thresholds to determine whether to infer and generate missing values. When the total data volume is large and the missing values of some indicators are small, the missing data is selected for deletion processing. When the total data volume is small and the missing values of some indicators are large, the filling process is selected for this indicator, so as to improve the efficiency of data preprocessing on the premise of ensuring sufficient total data. The embodiments of the present invention do not make any limitations in this regard.

[0041] After determining the target recognition production process parameter information, it is also necessary to control the PVB synthesis process, considering that the reactor temperature is maintained within the allowable error of ±0.15 of the set value during the feeding stage and the holding stage.

[0042] For the PVB production condensation reaction, considering the uncertainties and disturbances in the process: 1) The raw material PVA is random and fixed in the same batch or different batches, which will affect the condensation reaction rate Rp.

[0043] 2) External disturbances affect the reaction process, such as relatively fixed disturbances like monomer feeding rate, environmental heat loss, and catalyst activity.

[0044] 3) Suddenly changing disturbances, such as changes in the pressure and temperature of hot steam and coolant, seriously affect the reaction process.

[0045] The original PID control accuracy cannot meet the precise temperature control requirements in actual production. Manual precise temperature control is adopted in production. To realize the online optimization of important parameters in the whole production process of the project, it is necessary to automatically and precisely control the reaction temperature of PVB production. A control model composed of a tracking differentiator (TD), an extended state observer (ESO), and a nonlinear state error feedback (NLSEF) is adopted. Among them, TD can obtain the transition process v from the reference signal v of each order and the difference signal v −v n ; ESO can estimate the system state z according to the input-output data and the total disturbance z 1 ; n+1 −z 1 n . According to the state error e of the system 1 -e n T , the control signal u is calculated by NLSEF. If both the extended state observer and the error feedback are linear functions, the active disturbance rejection controller can be simplified to a linear active disturbance rejection controller. It is more convenient for parameter setting and theoretical analysis.

[0046] ​A controlled object of uncertain nth order (1) Wherein, is the system output, is the control signal, is the external disturbance of the system, gives the unknown dynamics of the system. System (1) can be expressed as (2) Wherein, , expand the total disturbance of the system to a new state variable, let , . Thus, the original system (2) is expanded into a new (n + 1) - order system (3), and the linear ESO is designed as follows: (3) Wherein, is the observer gain parameter. By introducing the observer bandwidth to reduce the parameter, as follows (4) Then the characteristic equation of the linear ESO can be configured as (5) The above configuration can not only ensure the stability of the system, but also provide a better transient process.

[0047] The control law is selected as follows: (6) Then the original system can be approximated as an integral series form (7) The approximate system (7) can be easily controlled by a PD controller: (8) Wherein is the reference signal r and the differential signal of r, and , i = 1, 2,..., n - 1 are the controller gain parameters.

[0048] It can be seen that implementing the data processing method described in the embodiments of the present invention is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters of multi - step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0049] In an alternative embodiment, as Figure 5As shown in the figure, the above-mentioned target parameter recognition model includes a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a sixth feature extraction module, a seventh feature extraction module, an eighth feature extraction module, a first fusion module, a second fusion module, a third fusion module, a feature analysis module, a sequence data processing module, a first convolution module, a first normalization module, a first activation module, a dimension conversion module, and a multi-layer perceptron; among them, The input end of the first feature extraction module and the input end of the first fusion module are configured to receive the first model input of the target parameter recognition model; the first feature extraction module, the second feature extraction module, the third feature extraction module, the fourth feature extraction module, and the first fusion module are sequentially connected; the output end of the first fusion module is respectively connected to the input end of the fifth feature extraction module and the input end of the second fusion module; the fifth feature extraction module, the sixth feature extraction module, the seventh feature extraction module, the eighth feature extraction module, and the second fusion module are sequentially connected; the output end of the second fusion module is connected to the input end of the feature analysis module; the output end of the feature analysis module is connected to the input end of the third fusion module; the input end of the sequence data processing module is configured to receive the second model input of the target parameter recognition model; the output end of the sequence data processing module is connected to the input end of the third fusion module; the third fusion module, the first convolution module, the first normalization module, the first activation module, the dimension conversion module, and the multi-layer perceptron are sequentially connected; the output end of the multi-layer perceptron is configured to output the model output of the target parameter recognition model.

[0050] It should be noted that the above-mentioned second model input represents the first target production process parameter information, the first model input represents the second target production process parameter information, and the model output represents the target recognition production process parameter information. The embodiments of the present invention are not limited.

[0051] It should be noted that the model architectures of the above-mentioned first feature extraction module, second feature extraction module, third feature extraction module, fourth feature extraction module, fifth feature extraction module, sixth feature extraction module, seventh feature extraction module, and eighth feature extraction module are the same. The embodiments of the present invention are not limited.

[0052] It should be noted that the above-mentioned first fusion module, second fusion module, and third fusion module are constructed based on element-wise addition operations. The embodiments of the present invention are not limited.

[0053] It should be noted that the convolution kernel size of the above-mentioned first convolution module is 1×1, and the stride is 1. The embodiments of the present invention are not limited.

[0054] It should be noted that the above-mentioned first normalization module is constructed based on a batch normalization layer. The embodiments of the present invention are not limited.

[0055] It should be noted that the above first activation module is constructed based on a non-linear activation function, which is not limited in the embodiments of the present invention. Further, the representation form of the above non-linear activation function can be: f(xx)=max(0,wb T ·xx+bb); In the formula, f(xx) represents the output of the first activation module, xx represents the input of the first activation module, bb represents the bias term (a value between 0 and 1), and wb represents the weight array. The values of the weight elements in the weight array are between 0 and 1, which are not limited in the embodiments of the present invention.

[0056] It should be noted that the above dimension conversion module is constructed based on the flatten operation to flatten the data representation into one-dimensional data, so as to facilitate the mapping analysis and processing of the multi-layer perceptron, which is not limited in the embodiments of the present invention.

[0057] It should be noted that the above multi-layer perceptron includes an input layer, at least one hidden layer, an output layer, and a mapping activation function. The number of neurons in the hidden layer is not less than 20, which is not limited in the embodiments of the present invention. Further, the one-dimensional data representation output by the dimension conversion module is input into the multi-layer perceptron, and the neurons of multiple hidden layers of the multi-layer perceptron perform weighted calculation processing on the one-dimensional data representation output by the dimension conversion module, so as to perform multiple combinations and transformations on the target feature information, and then more accurately capture the key parameters affecting the PVB production process.

[0058] Further, the mapping activation function of the multi-layer perceptron can be expressed as: , represents the output data representation of the output layer, represents the model output.

[0059] It should be noted that the above first model input performs feature extraction in the module unit composed of the residual structure composed of multiple feature extraction modules and fusion modules to form a multi-dimensional time-series feature representation, and then the feature analysis model (which can significantly increase the prediction time distance, so as to facilitate the analysis and processing of long-time-series data features) performs encoding and decoding processing on it, continuously performs encoding and decoding processing of process parameters under the time-series input data, and splices the output data representation with the data representation processed by the sequence data processing module, and then performs final mapping analysis processing by the multi-layer perceptron after re-depth feature extraction of convolution-normalization-activation to output the parameter identification result of the PVB production process, so as to realize the identification of the key process parameters of PVB, which is not limited in the embodiments of the present invention.

[0060] It should be noted that the residual structure composed of the above-mentioned 4 feature extraction modules and 1 fusion module constitutes a network architecture for time series data modeling and analysis, which can effectively extract time series features through a multi-layer temporal convolutional network, thereby improving the model's ability to capture temporal dynamics. The embodiments of the present invention are not limited thereto.

[0061] It can be seen that implementing the data processing method described in the embodiments of the present invention is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0062] In another optional embodiment, as Figure 5 shown, the first feature extraction module includes a first convolutional unit, a first activation unit, a second convolutional unit, and a first fusion unit; wherein, the input ends of the first convolutional unit and the first fusion unit are both configured as the input end of the first feature extraction module; the first convolutional unit, the first activation unit, the second convolutional unit, and the first fusion unit are sequentially connected in series; the output end of the first fusion unit is configured as the output end of the first feature extraction module.

[0063] It should be noted that the above-mentioned first convolutional unit is constructed based on spatio-temporal convolution, with a convolution kernel size of 3×3 and a stride of 1. The embodiments of the present invention are not limited thereto.

[0064] It should be noted that the above-mentioned first activation unit is constructed based on the RELU activation function. The embodiments of the present invention are not limited thereto.

[0065] It should be noted that the above-mentioned second convolutional unit is constructed based on spatio-temporal convolution, with a convolution kernel size of 1×1 and a stride of 1. The embodiments of the present invention are not limited thereto.

[0066] It should be noted that the above-mentioned first fusion unit is constructed based on element-wise addition operation. The embodiments of the present invention are not limited thereto.

[0067] It should be noted that the above-mentioned first feature extraction module uses a one-way feature extraction branch constructed by 2 spatio-temporal convolutions to achieve the one-way transmission ability based on spatio-temporal convolution, realize feature extraction based on time series, and at the same time use the fusion unit to fuse with the original time series features, so as to realize the fusion representation of two different scale features based on the residual structure, further enriching the form of feature representation. The embodiments of the present invention are not limited thereto.

[0068] It can be seen that implementing the data processing method described in the embodiments of the present invention is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step PVB synthesis, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0069] In yet another alternative embodiment, as Figure 6 shown, the feature analysis module includes a third convolutional unit, a fourth convolutional unit, a fifth convolutional unit, a third fusion unit, a first attention unit, a second attention unit, a third attention unit, a fourth attention unit, a fifth attention unit, a first pooling unit, a second pooling unit, and a first encoding unit; wherein, The input ends of the third convolutional unit, the first encoding unit, and the fifth attention unit are all connected to the output end of the second fusion module; the output ends of the third convolutional unit and the first encoding unit are both connected to the input end of the third fusion unit; the third fusion unit, the first attention unit, the fourth convolutional unit, the first pooling unit, the second attention unit, the fifth convolutional unit, the second pooling unit, and the third attention unit are connected in sequence; the output ends of the third attention unit and the fifth attention unit are both connected to the input end of the fourth attention unit; the output end of the fourth attention unit is connected to the input end of the third fusion module.

[0070] It should be noted that the above first attention unit, second attention unit, third attention unit, fourth attention unit, and fifth attention unit can be constructed based on the multi-head attention mechanism, and the embodiments of the present invention do not make limitations. Further, the representation form of the above multi-head attention mechanism can be: , where, , , , represents the input data, , and represent learnable weight matrices, represents , and corresponding data dimensions.

[0071] It should be noted that the convolutional kernels of the above third convolutional unit, fourth convolutional unit, and fifth convolutional unit can be one of 1×1, 3×1, 3×3, 5×1, 5×3, 5×5, 7×5, and 7×7, and the stride is 1 or 2, and the embodiments of the present invention do not make limitations.

[0072] It should be noted that the above third fusion unit is constructed based on element-wise addition operation, and the embodiments of the present invention do not make limitations.

[0073] It should be noted that the above-mentioned first pooling unit and second pooling unit are constructed based on the max pooling layer, and the embodiments of the present invention are not limited thereto.

[0074] It should be noted that the above-mentioned first encoding unit may be constructed based on an embedding module. When dimensionality is increased for low-dimensional data, some other features may be amplified, or general features may be separated. The embodiments of the present invention are not limited thereto.

[0075] It should be noted that the above-mentioned feature analysis module highlights the dominant attention by cascading multiple attention units to halve the input of the layer, increases the robustness of the model, effectively processes ultra-long input sequences, and embeds convolutional-pooling module units between attention units to extract the dominant attention and reduce the network computation amount. Further, 1 parallel attention unit is used for a forward operation at the same time, greatly improving the inference speed of long-sequence data and further enhancing the processing efficiency of the model. The embodiments of the present invention are not limited thereto.

[0076] It can be seen that implementing the data processing method described in the embodiments of the present invention is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0077] In yet another alternative embodiment, as Figure 7 shown, the sequence data processing module includes a first normalization unit, a second normalization unit, a sixth attention unit, a fourth fusion unit, a fifth fusion unit, and a feedforward neural network; wherein, The input ends of the first normalization unit and the sixth attention unit are both configured to receive the second model input; the output end of the first normalization unit and the output end of the sixth attention unit are both connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is respectively connected to the input end of the feedforward neural network and the input end of the fifth fusion unit; the output end of the feedforward neural network is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the third fusion module.

[0078] It should be noted that the above-mentioned first normalization unit and second normalization unit are constructed based on the batch normalization layer, and the embodiments of the present invention are not limited thereto.

[0079] It should be noted that the above-mentioned sixth attention unit is constructed based on the multi-head attention mechanism, and the embodiments of the present invention are not limited thereto.

[0080] It should be noted that the above fourth fusion unit and fifth fusion unit are constructed based on element-wise addition operations, which are not limited in the embodiments of the present invention.

[0081] It should be noted that the above feedforward neural network includes an input layer, several hidden layers, and an output layer, which are not limited in the embodiments of the present invention. Further, in the feedforward neural network, the forward flow of data, that is, the data of the input layer is directly transmitted to the output layer through the hidden layer, and each node (or neuron) is only connected to the nodes of the previous layer to achieve a complex mapping from the input space to the output space through multiple compositions of simple non-linear functions, which are not limited in the embodiments of the present invention.

[0082] It should be noted that the above sequence data processing module extracts global features from the input data through an attention unit, and at the same time performs pooling processing on the input data to convert the data dimension and retain the original data features, and then fuses them with the extracted global features. Then, the data is split into two paths. One path is that the feedforward neural network and the pooling unit fuse the features through non-linear mapping of the fused features, and then the fusion module performs deep fusion of the features to obtain the processed sequence data. This sequence data contains both the globally deeply extracted features and retains the original features, forming a rich feature representation, which is convenient for subsequent identification of key parameters according to the feature information, which are not limited in the embodiments of the present invention.

[0083] It can be seen that implementing the data processing method described in the embodiments of the present invention is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0084] In an alternative embodiment, the above target parameter recognition model is obtained based on the following training steps: Obtain basic parameter training sample information; the basic parameter training sample information includes M basic parameter training samples; M is a positive integer not less than 500; Use the basic parameter training sample information to train the basic parameter recognition model to obtain the target parameter recognition model.

[0085] It should be noted that the above 500 basic parameter training samples can ensure the richness of the samples and avoid the problem of falling into local optimum or unable to converge during the model training process, which are not limited in the embodiments of the present invention.

[0086] It should be noted that the above basic parameter training samples can collect the successful production process data and failed production process data in the production process by means of DCS industrial Internet collection technology, and then label them to form positive and negative samples, which is more conducive to the rapid and efficient training of the model. The embodiments of the present invention do not make limitations. Further, after the data is collected, a fusion sliding window attention mechanism can be used to generate new sample data. The specific operation steps are as follows: Suppose the size of the window is w, then at the moment, the input data sequence read by the window is , and the multi-head attention is used to capture the degree of mutual influence of the input data. The formula is as follows: ; ; ; 、 and are all weight matrices, is a hyperparameter, is the local feature extracted at the moment, and then the features obtained at each moment are concatenated to form the newly generated sample data.

[0087] It can be seen that implementing the data processing method described in the embodiments of the present invention is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0088] In another optional embodiment, the basic parameter recognition model is trained using the basic parameter training sample information to obtain a target parameter recognition model, including: Based on the basic parameter training samples in the basic parameter training sample information, determine the target parameter training sample information; the target parameter training sample information includes N target parameter training samples; N is a positive integer between [20, 50]; Use the target parameter training sample information to train the basic parameter recognition model to obtain training result information and a training parameter recognition model; Use the loss function to calculate and process the training result information to obtain loss function value information; Judge whether the loss function values in the loss function value information are all within the loss fluctuation range to obtain the first training judgment result; the loss fluctuation range is [a, b], where 3% ≤ (b - a) / a ≤ 5%; When the first training judgment result is yes, determine the training parameter recognition model as the target parameter recognition model; When the first training judgment result is no, judge whether the number of training times in the training result information is equal to the number threshold to obtain a second training judgment result; When the second training judgment result is no, determine the training parameter recognition model as the new basic parameter recognition model, and trigger the execution of determining a candidate sample from the candidate sample set as the target sample; When the second training judgment result is yes, determine the training parameter recognition model as the target parameter recognition model.

[0089] It should be noted that the model architectures of the above basic parameter recognition model, training parameter recognition model, and target parameter recognition model are the same. The difference can be that the model parameters are different, and the embodiments of the present invention do not make any limitations.

[0090] It should be noted that the loss function during the training of the above basic parameter recognition model can be a cross-entropy loss function, and the embodiments of the present invention do not make any limitations.

[0091] It should be noted that the number threshold during the training of the above basic parameter recognition model is not less than 100 times, and the embodiments of the present invention do not make any limitations.

[0092] It should be noted that the training of the above basic parameter recognition model can be carried out on an NVIDIA 3090 graphics card based on Python 3.7.10 and above versions, and the embodiments of the present invention do not make any limitations.

[0093] It should be noted that during the training of the above basic parameter recognition model, accuracy, precision, and recall can be used to evaluate the trained model, and the embodiments of the present invention do not make any limitations.

[0094] It should be noted that the model architectures of the above basic parameter recognition model, training parameter recognition model, and target parameter recognition model are the same. The difference can be that the parameters of the model are different, and the embodiments of the present invention do not make any limitations.

[0095] It should be noted that the above-mentioned determination of the target parameter training sample information from the basic parameter training samples in the basic parameter training sample information is to randomly select N from the basic parameter training samples as the target parameter training samples, which is not limited in the embodiments of the present invention. Further, setting N to a value between 20 and 50 can ensure that there are sufficient sample numbers for each batch of model training, so as to have sufficient training data guarantee when evaluating the trained model later, improve the accuracy of model training, and at the same time, it will not make the number of training samples too large, lengthen the training time of each batch, thereby reducing the efficiency of the model training process. Moreover, when determining the training samples, they are selected in a random form, which can ensure the randomness of the samples. With a batch training sample size of 20 - 50, it realizes small-batch and fast-iteration model training, which can not only ensure the model training speed but also ensure the model training accuracy, and the embodiments of the present invention are not limited thereto.

[0096] It should be noted that the above-mentioned loss fluctuation range is [a, b], where 3% ≤ (b - a) / a ≤ 5% (the values of a and b are greater than 0 and less than 1). This strictly sets the fluctuation range of the loss, that is, it provides sufficient guarantee for the model training accuracy. Further, setting the amplitude of the loss fluctuation range between 3% and 5% means considering that there may be certain fluctuations in the accidental loss function values of random samples during the model training process. Considering the needs of engineering practice more fully, the highest-level fluctuation range is limited between the two most strict values, which can greatly improve the accuracy of model training, and the embodiments of the present invention are not limited thereto.

[0097] It can be seen that implementing the data processing method described in the embodiments of the present invention is beneficial to accurately identify and analyze the key parameters in the PVB synthesis process, improve the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0098] Embodiment 2 Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of a data processing device disclosed in the embodiments of the present invention. Among them, Figure 3 the described device can be applied to a management system, such as a local server or a cloud server for management, etc., which is not limited in the embodiments of the present invention. As Figure 3 shown, the device may include: An acquisition module 201, configured to acquire PVB production process parameter information; A first processing module 202, configured to perform pre-processing on the PVB production process parameter information to obtain target production process parameter information; The second processing module 203 is configured to perform recognition processing on the target production parameter information by using the target parameter recognition model to obtain the target recognized production process parameter information.

[0099] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately recognizing and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0100] In another optional embodiment, as Figure 3 shown, the target parameter recognition model includes a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a sixth feature extraction module, a seventh feature extraction module, an eighth feature extraction module, a first fusion module, a second fusion module, a third fusion module, a feature analysis module, a sequence data processing module, a first convolution module, a first normalization module, a first activation module, a dimension conversion module, and a multi-layer perceptron; wherein, the input end of the first feature extraction module and the input end of the first fusion module are configured to receive the first model input of the target parameter recognition model; the first feature extraction module, the second feature extraction module, the third feature extraction module, the fourth feature extraction module, and the first fusion module are sequentially connected in order; the output end of the first fusion module is respectively connected to the input end of the fifth feature extraction module and the input end of the second fusion module; the fifth feature extraction module, the sixth feature extraction module, the seventh feature extraction module, the eighth feature extraction module, and the second fusion module are sequentially connected in order; the output end of the second fusion module is connected to the input end of the feature analysis module; the output end of the feature analysis module is connected to the input end of the third fusion module; the input end of the sequence data processing module is configured to receive the second model input of the target parameter recognition model; the output end of the sequence data processing module is connected to the input end of the third fusion module; the third fusion module, the first convolution module, the first normalization module, the first activation module, the dimension conversion module, and the multi-layer perceptron are sequentially connected in order; the output end of the multi-layer perceptron is configured to output the model output of the target parameter recognition model.

[0101] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately recognizing and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0102] In yet another optional embodiment, as Figure 3As shown, the first feature extraction module includes a first convolutional unit, a first activation unit, a second convolutional unit, and a first fusion unit; among them, The input ends of the first convolutional unit and the first fusion unit are both configured as the input end of the first feature extraction module; the first convolutional unit, the first activation unit, the second convolutional unit, and the first fusion unit are sequentially connected in series; the output end of the first fusion unit is configured as the output end of the first feature extraction module.

[0103] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0104] In yet another alternative embodiment, as Figure 3 shown, the feature analysis module includes a third convolutional unit, a fourth convolutional unit, a fifth convolutional unit, a third fusion unit, a first attention unit, a second attention unit, a third attention unit, a fourth attention unit, a fifth attention unit, a first pooling unit, a second pooling unit, and a first encoding unit; among them, The input ends of the third convolutional unit, the first encoding unit, and the fifth attention unit are all connected to the output end of the second fusion module; the output ends of the third convolutional unit and the first encoding unit are both connected to the input end of the third fusion unit; the third fusion unit, the first attention unit, the fourth convolutional unit, the first pooling unit, the second attention unit, the fifth convolutional unit, the second pooling unit, and the third attention unit are sequentially connected in series; the output ends of the third attention unit and the fifth attention unit are both connected to the input end of the fourth attention unit; the output end of the fourth attention unit is connected to the input end of the third fusion module.

[0105] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0106] In yet another alternative embodiment, as Figure 3 shown, the sequence data processing module includes a first normalization unit, a second normalization unit, a sixth attention unit, a fourth fusion unit, a fifth fusion unit, and a feed-forward neural network; among them, The input ends of the first normalization unit and the sixth attention unit are both configured to receive the second model input; the output ends of the first normalization unit and the sixth attention unit are both connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is respectively connected to the input end of the feedforward neural network and the input end of the fifth fusion unit; the output end of the feedforward neural network is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the third fusion module.

[0107] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0108] In another optional embodiment, as Figure 3 shown, the target parameter recognition model is obtained based on the following training steps: Obtain the basic parameter training sample information; the basic parameter training sample information includes M basic parameter training samples; M is a positive integer not less than 500; Use the basic parameter training sample information to train the basic parameter recognition model to obtain the target parameter recognition model.

[0109] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0110] In another optional embodiment, as Figure 3 shown, using the basic parameter training sample information to train the basic parameter recognition model to obtain the target parameter recognition model includes: Based on the basic parameter training samples in the basic parameter training sample information, determine the target parameter training sample information; the target parameter training sample information includes N target parameter training samples; N is a positive integer between [20, 50]; Use the target parameter training sample information to train the basic parameter recognition model to obtain the training result information and the training parameter recognition model; Use the loss function to calculate and process the training result information to obtain the loss function value information; Determine whether all the loss function values in the loss function value information are within the loss fluctuation range to obtain a first training judgment result; the loss fluctuation range is [a, b], where 3% ≤ (b - a) / a ≤ 5%; When the first training judgment result is yes, determine that the training parameter recognition model is the target parameter recognition model; When the first training judgment result is no, determine whether the number of training times in the training result information is equal to the number threshold to obtain a second training judgment result; When the second training judgment result is no, determine the training parameter recognition model as the new basic parameter recognition model, and trigger the execution of determining a candidate sample from the candidate sample set as the target sample; When the second training judgment result is yes, determine that the training parameter recognition model is the target parameter recognition model.

[0111] It can be seen that implementing Figure 3 the described data processing device is conducive to accurately identifying and analyzing the key parameters in the PVB synthesis process, improving the recognition accuracy and efficiency of the key process parameters for multi-step synthesis of PVB, and has important practical significance and broad application prospects for improving the automation level and product quality of the PVB synthesis process.

[0112] Example III Please refer to Figure 4 , Figure 4 which is a schematic structural diagram of another data processing device disclosed in the embodiments of the present invention. Among them, Figure 4 the described device can be applied to a management system, such as a local server or a cloud server for management, etc., which is not limited in the embodiments of the present invention. As Figure 4 shown, the device may include: A memory 301 storing executable program code; A processor 302 coupled to the memory 301; The processor 302 calls the executable program code stored in the memory 301 to execute the steps in the data processing method described in Example I.

[0113] Example IV The embodiments of the present invention disclose a computer-readable storage medium that stores a computer program for electronic data exchange, wherein the computer program causes a computer to execute the steps in the data processing method described in Example I.

[0114] Example V The embodiments of the present invention disclose a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to execute the steps in the data processing method described in Example I.

[0115] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed to multiple network modules. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0116] Through the specific descriptions of the above embodiments, those skilled in the art can clearly understand that each implementation mode can be realized by means of software plus a necessary general hardware platform, and of course, it can also be realized by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, and the storage medium includes read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc memories, magnetic disk memories, tape memories, or any other computer-readable medium that can be used to carry or store data.

[0117] Finally, it should be noted that: what is disclosed in an embodiment of a data processing method of the present invention is only a preferred embodiment of the present invention, which is only used to illustrate the technical solution of the present invention and is not intended to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data processing method, characterized in that: The method comprises: Obtain PVB production process parameter information; Pre-processing the PVB production process parameter information to obtain target production process parameter information; The target parameter identification model is used to identify the target production parameter information to obtain target identification production process parameter information.

2. The data processing method according to claim 1, characterized in that: The target parameter recognition model includes a first feature extraction module, a second feature extraction module, a third feature extraction module, a fourth feature extraction module, a fifth feature extraction module, a sixth feature extraction module, a seventh feature extraction module, an eighth feature extraction module, a first fusion module, a second fusion module, a third fusion module, a feature analysis module, a sequence data processing module, a first convolution module, a first normalization module, a first activation module, a dimension conversion module and a multi-layer perceptron; wherein, The input end of the first feature extraction module and the input end of the first fusion module are configured to receive the first model input of the target parameter recognition model; the first feature extraction module, the second feature extraction module, the third feature extraction module, the fourth feature extraction module and the first fusion module are connected in sequence; the output end of the first fusion module is connected to the input end of the fifth feature extraction module and the input end of the second fusion module respectively; the fifth feature extraction module, the sixth feature extraction module, the seventh feature extraction module, the eighth feature extraction module and the second fusion module are connected in sequence; the output end of the second fusion module is connected to the input end of the feature analysis module; the output end of the feature analysis module is connected to the input end of the third fusion module; the input end of the sequence data processing module is configured to receive the second model input of the target parameter recognition model; the output end of the sequence data processing module is connected to the input end of the third fusion module; the third fusion module, the first convolution module, the first normalization module, the first activation module, the dimension conversion module and the multi-layer perceptron are connected in sequence; the output end of the multi-layer perceptron is configured to output the model output of the target parameter recognition model.

3. The data processing method according to claim 2, characterized in that: The first feature extraction module includes a first convolution unit, a first activation unit, a second convolution unit, and a first fusion unit; wherein, The input end of the first convolution unit and the input end of the first fusion unit are both configured as the input end of the first feature extraction module; the first convolution unit, the first activation unit, the second convolution unit, and the first fusion unit are connected in sequence; the output end of the first fusion unit is configured as the output end of the first feature extraction module.

4. The data processing method according to claim 2, characterized in that: The feature analysis module includes a third convolution unit, a fourth convolution unit, a fifth convolution unit, a third fusion unit, a first attention unit, a second attention unit, a third attention unit, a fourth attention unit, a fifth attention unit, a first pooling unit, a second pooling unit and a first encoding unit; wherein, The input end of the third convolution unit, the input end of the first encoding unit and the input end of the fifth attention unit are all connected to the output end of the second fusion module; the output end of the third convolution unit and the output end of the first encoding unit are both connected to the input end of the third fusion unit; the third fusion unit, the first attention unit, the fourth convolution unit, the first pooling unit, the second attention unit, the fifth convolution unit, the second pooling unit, and the third attention unit are connected in sequence; the output end of the third attention unit and the output end of the fifth attention unit are both connected to the input end of the fourth attention unit; the output end of the fourth attention unit is connected to the input end of the third fusion module.

5. The data processing method according to claim 2, characterized in that: The sequence data processing module includes a first normalization unit, a second normalization unit, a sixth attention unit, a fourth fusion unit, a fifth fusion unit and a feedforward neural network; wherein, The input end of the first normalization unit and the input end of the sixth attention unit are both configured to receive the second model input; the output end of the first normalization unit and the output end of the sixth attention unit are both connected to the input end of the fourth fusion unit; the output end of the fourth fusion unit is respectively connected to the input end of the feedforward neural network and the input end of the fifth fusion unit; the output end of the feedforward neural network is connected to the input end of the second normalization unit; the output end of the second normalization unit is connected to the input end of the fifth fusion unit; the output end of the fifth fusion unit is connected to the input end of the third fusion module.

6. The data processing method according to claim 2, characterized in that: The target parameter recognition model is obtained based on the following training steps: Obtain basic parameter training sample information; the basic parameter training sample information includes M basic parameter training samples; M is a positive integer not less than 500; The basic parameter recognition model is trained using the basic parameter training sample information to obtain the target parameter recognition model.

7. The data processing method according to claim 6, characterized in that: The step of training the basic parameter recognition model using the basic parameter training sample information to obtain the target parameter recognition model includes: Based on the basic parameter training samples in the basic parameter training sample information, target parameter training sample information is determined; the target parameter training sample information includes N target parameter training samples; N is a positive integer between [20, 50]; Using the target parameter training sample information to train the basic parameter recognition model, and obtaining training result information and a training parameter recognition model; Using the loss function to calculate the training result information to obtain loss function value information; Determine whether the loss function values ​​in the loss function value information are all within the loss fluctuation range, and obtain a first training judgment result; the loss fluctuation range is [a, b], where 3%≤(ba) / a≤5%; When the first training judgment result is yes, determining the training parameter identification model as the target parameter identification model; When the first training judgment result is no, determining whether the number of training times in the training result information is equal to a number threshold, and obtaining a second training judgment result; When the second training judgment result is no, determining the training parameter recognition model as a new basic parameter recognition model, and triggering the execution of determining one of the candidate samples from the candidate sample set as a target sample; When the second training judgment result is yes, the training parameter identification model is determined to be the target parameter identification model.

8. A data processing device, characterized in that: The device comprises: An acquisition module is used to obtain PVB production process parameter information; A first processing module is used to pre-process the PVB production process parameter information to obtain target production process parameter information; The second processing module is used to use the target parameter recognition model to identify the target production parameter information to obtain target recognition production process parameter information.

9. A data processing device, characterized in that: The device comprises: A memory storing executable program code; a processor coupled to the memory; The processor calls the executable program code stored in the memory to execute the data processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and when the computer instructions are called, they are used to execute the data processing method according to any one of claims 1 to 7.

Images