Control system and method of intelligent polaroid processing equipment
Through deep learning technology, the thickness and flow data are processed, and the flow rate of the coating liquid is automatically adjusted, which solves the problem of unstable coating quality during traditional polarizer coating, achieving higher consistency and stability.
Patent Information
- Application Number
- CN202411488240.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-24
- Publication Date
- 2024-12-03
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Manual operation and adjustment of parameters are required during the coating process of traditional polarizers, resulting in unstable coating quality and poor coating uniformity, which cannot meet the precise coating thickness and performance requirements.
The thickness measurement instrument is used to collect coating thickness data and the flowmeter to collect coating liquid flow value. Through deep learning technology, the flow rate is automatically adjusted to achieve more accurate coating liquid control.
Improves coating quality and production consistency and stability, reduces waste during manual intervention and coating process, and ensures that the flow rate of the coating liquid is best suited at different locations and in time points.
Smart Images

Figure CN119056699A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of intelligent control, and more specifically, to a control system and method for intelligent processing equipment for polarizers. Background Art
[0002] Polarizers are a polymer material, usually made of transparent crystalline materials. They are specially designed to allow certain wavelengths of light to pass through while blocking other wavelengths of light. Polarizers are widely used in LCD screens, 3D glasses and other fields.
[0003] During the processing of polarizers, it is necessary to coat the polarizers through coating processing equipment to improve their performance and protect their structure. Specifically, the surface coating of polarizers can effectively reduce the reflection and interference of external light, help polarizers better control the propagation of light, thereby improving the display effect, and protect the structure of polarizers from damage by external environmental factors such as moisture and air. Therefore, coating polarizers is an effective way to improve the performance of polarizers, protect their structure, and thus improve the overall performance of display devices.
[0004] However, the traditional method requires manual operation and parameter adjustment when coating polarizers. Most of the manual work is based on experience and personal level, which will lead to unstable coating quality and poor coating uniformity, thus affecting the optical performance of polarizers. In addition, for some special coating requirements, traditional methods may not be able to meet the precise coating thickness and performance requirements.
[0005] Therefore, an optimized control system for polarizer intelligent processing equipment is desired. Summary of the invention
[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a control system and method for intelligent processing equipment of polarizers, which collects the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions in real time by a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period in real time by a flow meter, and uses deep learning technology to perform data processing, analysis and correlation on the thickness data and the flow value, so as to determine whether the current flow rate should increase, decrease or remain unchanged. In this way, more accurate control of the coating liquid can be achieved, manual intervention can be reduced, waste in the coating process can be reduced, and the flow rate of the coating liquid at different positions and time points can be ensured to be the most suitable, thereby improving the coating quality and the consistency and stability of production.
[0007] According to one aspect of the present application, a control system for a polarizer intelligent processing device is provided, which includes:
[0008] A coating processing parameter acquisition module, which is used to acquire the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions through a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period acquired by a flow meter;
[0009] A thickness arrangement module, used for arranging the thickness data of a plurality of predetermined time points within a predetermined time period at the different positions into a thickness time series input matrix;
[0010] A flow value time series arrangement module, used to arrange the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension into a flow time series input vector and then transpose and multiply it with itself to obtain a flow time series input matrix;
[0011] A feature association module is used to perform perspective mapping matching on a thickness time series input matrix and a flow time series input matrix based on eigenvalue granularity association to obtain a thickness and flow time series association input matrix;
[0012] A multi-scale time series association module, used for performing multi-scale time series association perception on the thickness-flow time series association input matrix to obtain a thickness-flow time series association multi-scale feature matrix;
[0013] An attention enhancement module, used for performing attention enhancement on the thickness-flow time series correlation multi-scale feature matrix to obtain a thickness-flow time series correlation multi-scale enhanced feature vector;
[0014] A feature optimization module, used for performing vector optimization based on probability weight on the thickness-flow time series correlation multi-scale enhancement feature vector to obtain an optimized thickness-flow time series correlation multi-scale enhancement feature vector;
[0015] The flow control module is used to determine whether the current flow should be increased, decreased or kept unchanged based on the optimized thickness flow time series associated multi-scale enhanced feature vector.
[0016] In the control system of the above-mentioned polarizer intelligent processing equipment, the flow value timing arrangement module includes: a timing sample arrangement unit, which is used to arrange the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension as the flow timing input vector; and a feature multiplication unit, which is used to multiply the flow timing input vector with the transpose of the flow timing input vector to obtain the flow timing input matrix.
[0017] In the control system of the above-mentioned polarizer intelligent processing equipment, the feature association module includes: a thickness-flow timing feature expansion unit, which is used to sequentially expand the thickness timing input matrix and the flow timing input matrix to obtain a thickness timing input vector and a flow timing input vector; a thickness-flow timing characteristic value granularity association unit, which is used to perform perspective mapping matching based on characteristic value granularity association on the thickness timing input vector and the flow timing input vector to obtain a thickness-flow timing association input vector; a thickness-flow timing association feature reconstruction unit, which is used to dimensionally reconstruct the thickness-flow timing association input vector to obtain the thickness-flow timing association input matrix; wherein the thickness-flow timing characteristic value granularity association unit is used to: construct the thickness timing input vector and the flow timing input vector A thickness timing-flow timing distance topological matrix between the thickness timing input vector and the flow timing input vector is constructed; a thickness timing-flow timing characteristic value granularity association matrix is constructed between the thickness timing input vector and the flow timing input vector; based on the thickness timing-flow timing distance topological matrix, the thickness timing-flow timing characteristic value granularity association matrix is topologically associated modulated to obtain a thickness timing-flow timing topological modulation association matrix; using the thickness timing-flow timing topological modulation association matrix as a perspective co-projection space, the thickness timing input vector and the flow timing input vector are projected into the perspective co-projection space to obtain a perspective modulated thickness timing input vector and a perspective modulated flow timing input vector; and the perspective modulated thickness timing input vector and the perspective modulated flow timing input vector are fused to obtain the thickness flow timing association input vector.
[0018] In the control system of the above-mentioned polarizer intelligent processing equipment, the multi-scale time series association module includes: a multi-scale feature extraction unit, which is used to input the thickness flow time series association input matrix into a multi-scale time series association sensor including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow time series association feature matrix and a second-scale thickness flow time series association feature matrix; and a multi-scale cascade unit, which is used to cascade the first-scale thickness flow time series association feature matrix and the second-scale thickness flow time series association feature matrix to obtain the thickness flow time series association multi-scale feature matrix.
[0019] In the control system of the above-mentioned polarizer intelligent processing equipment, the attention enhancement module is used to: input the thickness flow time series correlation multi-scale feature matrix into the correlation feature extractor based on the bidirectional attention mechanism to obtain the thickness flow time series correlation multi-scale enhanced feature matrix, and expand the thickness flow time series correlation multi-scale enhanced feature matrix into the thickness flow time series correlation multi-scale enhanced feature vector.
[0020] In the control system of the above-mentioned polarizer intelligent processing equipment, the feature optimization module includes: a feature frequency calculation unit, which is used to calculate the frequency of the feature values of each position in the thickness flow time series associated multi-scale enhanced feature vector in the thickness flow time series associated multi-scale enhanced feature vector to obtain a classification feature frequency statistical vector; a pre-classification unit, which is used to pass the thickness flow time series associated multi-scale enhanced feature vector through a pre-classifier to obtain a category probability vector; a probability weight measurement unit, which is used to calculate the conditional probability of the classification feature frequency statistical vector relative to the category probability vector to obtain a probability weight measurement vector; and a position point multiplication unit, which is used to calculate the position point multiplication between the thickness flow time series associated multi-scale enhanced feature vector and the probability weight measurement vector to obtain the optimized thickness flow time series associated multi-scale enhanced feature vector.
[0021] In the control system of the above-mentioned polarizer intelligent processing equipment, the flow control module is used to: input the optimized thickness flow time series associated multi-scale enhanced feature vector into the classifier to obtain a classification result, and the classification result is used to indicate whether the current flow should increase, decrease or remain unchanged.
[0022] According to another aspect of the present application, a control method for a polarizer intelligent processing device is provided, which includes:
[0023] The thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions collected by the thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period collected by the flow meter;
[0024] Arranging the thickness data of a plurality of predetermined time points within the predetermined time periods at the different positions into a thickness time series input matrix;
[0025] Arranging the flow values of the coating liquid at a plurality of predetermined time points within the predetermined time period into a flow time series input vector according to the time dimension and then transposing and multiplying the vector with itself to obtain a flow time series input matrix;
[0026] Performing perspective mapping matching based on eigenvalue granularity association on the thickness time series input matrix and the flow time series input matrix to obtain a thickness-flow time series association input matrix;
[0027] Perform multi-scale time series correlation perception on the thickness-flow time series correlation input matrix to obtain a thickness-flow time series correlation multi-scale feature matrix;
[0028] Performing attention enhancement on the thickness-flow time series correlation multi-scale feature matrix to obtain a thickness-flow time series correlation multi-scale enhanced feature vector;
[0029] Performing vector optimization based on probability weight on the thickness-flow time series associated multi-scale enhanced feature vector to obtain an optimized thickness-flow time series associated multi-scale enhanced feature vector;
[0030] Based on the optimized thickness-flow time series associated multi-scale enhanced feature vector, it is determined whether the current flow should be increased, decreased, or kept unchanged.
[0031] In the control method of the above-mentioned polarizer intelligent processing equipment, the flow values of the coating liquid at multiple predetermined time points within the predetermined time period are arranged according to the time dimension as a flow timing input vector and then multiplied by its own transpose to obtain a flow timing input matrix, including: arranging the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension as the flow timing input vector; and, multiplying the flow timing input vector by the transpose of the flow timing input vector to obtain the flow timing input matrix.
[0032] In the control method of the above-mentioned polarizer intelligent processing equipment, the thickness flow timing correlation input matrix is subjected to multi-scale timing correlation perception to obtain a thickness flow timing correlation multi-scale feature matrix, including: inputting the thickness flow timing correlation input matrix into a multi-scale timing correlation sensor including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow timing correlation feature matrix and a second-scale thickness flow timing correlation feature matrix; and, cascading the first-scale thickness flow timing correlation feature matrix and the second-scale thickness flow timing correlation feature matrix to obtain the thickness flow timing correlation multi-scale feature matrix.
[0033] Compared with the prior art, the control system and method of a polarizer intelligent processing equipment provided by the present application uses the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions collected in real time by a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period collected in real time by a flow meter, and uses deep learning technology to perform data processing, analysis and correlation on the thickness data and the flow value, so as to determine whether the current flow rate should increase, decrease or remain unchanged. In this way, more accurate coating liquid control can be achieved, manual intervention can be reduced, waste in the coating process can be reduced, and the flow rate of the coating liquid at different positions and time points can be ensured to be the most suitable, thereby improving the coating quality and the consistency and stability of production. BRIEF DESCRIPTION OF THE DRAWINGS
[0034] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.
[0035] Figure 1 This is a block diagram of a control system of a polarizer intelligent processing device according to an embodiment of the present application.
[0036] Figure 2 Schematic diagram of the architecture of the control system of the polarizer intelligent processing equipment according to the embodiment of the present application.
[0037] Figure 3 It is a block diagram of a flow value timing arrangement module in a control system of a polarizer intelligent processing device according to an embodiment of the present application.
[0038] Figure 4 It is a block diagram of a multi-scale timing association module in a control system of a polarizer intelligent processing device according to an embodiment of the present application.
[0039] Figure 5 This is a flow chart of a control method for a polarizer intelligent processing device according to an embodiment of the present application. DETAILED DESCRIPTION
[0040] The embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein, but rather these embodiments are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes and are not intended to limit the scope of protection of the present disclosure.
[0041] It should be understood that the various steps described in the method embodiments of the present disclosure may be performed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present disclosure is not limited in this respect.
[0042] It should be noted that the terms "first\second\third" involved in the embodiments of the present application are only used to distinguish similar objects, and do not represent a specific order for the objects. It is understandable that the specific order or sequence of "first\second\third" can be interchanged where permitted. It should be understood that the objects distinguished by "first\second\third" can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.
[0043] As shown in the present disclosure and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.
[0044] Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.
[0045] Polaroid is a polymer material, mainly made of polyvinyl alcohol (PVA) film and triacetyl cellulose (TAC) film through stretching, compounding, coating and other processes. Its function is to control light waves, allowing light vibrating in a specific direction to pass through, while blocking light vibrating in other directions. The core raw material of polaroid is PVA film, whose transmission axis is perpendicular to the stretching direction, and its transmittance, polarization, hue and other indicators are controlled by its manufacturing difficulty and quality. All liquid crystals have two polarizing films in front and behind, which are close to the liquid crystal glass, forming a liquid crystal film with a total thickness of about 1mm. If any polarizing film is missing, the liquid crystal film cannot display images.
[0046] During the processing of polarizers, it is necessary to coat the polarizers through coating processing equipment to improve their performance and protect their structure. Specifically, the surface coating of polarizers can effectively reduce the reflection and interference of external light, help polarizers better control the propagation of light, thereby improving the display effect, and protect the structure of polarizers from damage by external environmental factors such as moisture and air. Therefore, coating polarizers is an effective way to improve the performance of polarizers, protect their structure, and thus improve the overall performance of display devices.
[0047] However, the traditional method requires manual operation and parameter adjustment when coating polarizers. Most of the manual work is based on experience and personal level, which will lead to unstable coating quality and poor coating uniformity, thus affecting the optical performance of polarizers. In addition, for some special coating requirements, traditional methods may not be able to meet the precise coating thickness and performance requirements.
[0048] Therefore, an optimized control system for polarizer intelligent processing equipment is desired, which collects the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions in real time by a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period in real time by a flow meter, and uses deep learning technology to process, analyze and associate the thickness data and the flow value, so as to determine whether the current flow should increase, decrease or remain unchanged. In this way, more accurate coating liquid control can be achieved, manual intervention can be reduced, waste in the coating process can be reduced, and the flow rate of the coating liquid at different positions and time points can be ensured to be the most suitable, thereby improving the coating quality and the consistency and stability of production.
[0049] Figure 1 This is a block diagram of a control system of a polarizer intelligent processing device according to an embodiment of the present application. Figure 2 FIG. 1 is a schematic diagram of the control system of the polarizer intelligent processing equipment according to the embodiment of the present application. Figure 1 and Figure 2 As shown, according to the control system 100 of the polarizer intelligent processing equipment of the embodiment of the present application, it includes: a coating processing parameter acquisition module 110, which is used to collect the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions through a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period collected by a flow meter; a thickness arrangement module 120, which is used to arrange the thickness data at multiple predetermined time points in the predetermined time period at different positions into a thickness time series input matrix; a flow value time series arrangement module 130, which is used to arrange the flow values of the coating liquid at multiple predetermined time points in the predetermined time period according to the time dimension into a flow time series input vector and then multiply it with its own transposition to obtain a flow time series input matrix; a feature association module 140, which is used to perform feature-based correlation on the thickness time series input matrix and the flow time series input matrix. The perspective mapping of the eigenvalue granularity association is matched to obtain a thickness flow time series association input matrix; a multi-scale time series association module 150 is used to perform multi-scale time series association perception on the thickness flow time series association input matrix to obtain a thickness flow time series association multi-scale feature matrix; an attention enhancement module 160 is used to perform attention enhancement on the thickness flow time series association multi-scale feature matrix to obtain a thickness flow time series association multi-scale enhanced feature vector; a feature optimization module 170 is used to perform vector optimization based on probability weights on the thickness flow time series association multi-scale enhanced feature vector to obtain an optimized thickness flow time series association multi-scale enhanced feature vector; and a flow control module 180 is used to determine whether the current flow should increase, decrease, or remain unchanged based on the optimized thickness flow time series association multi-scale enhanced feature vector.
[0050] In an embodiment of the present application, the coating processing parameter acquisition module 110 is used to collect the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions collected by a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period collected by a flow meter. It should be understood that the thickness data and the flow value play a key role in the polarizer intelligent processing equipment. Specifically, the thickness data reflects the thickness of the coating and can be used to evaluate the quality and uniformity of the coating. In the manufacture of polarizers, the thickness of the coating has an important influence on the optical properties. Therefore, by real-time monitoring of the thickness data, it can be ensured that the quality of the coating meets the requirements. The flow value reflects the flow rate of the coating liquid in a predetermined time period and can be used to control the supply speed of the coating liquid. By real-time monitoring of the flow value, the flow rate of the coating liquid can be adjusted to ensure that the thickness and quality of the coating meet the requirements. Based on this, in the technical solution of the present application, the thickness data of the coating at multiple predetermined time points within a predetermined time period at different positions collected by the thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points within a predetermined time period collected by the flow meter, that is, by obtaining the thickness data and flow value, real-time monitoring and adjustment of the coating quality and coating liquid control can be achieved, thereby ensuring that the product quality meets the standard requirements, optimizing the coating process, and improving production efficiency and product consistency.
[0051] In an embodiment of the present application, the thickness arrangement module 120 is used to arrange the thickness data of multiple predetermined time points within the predetermined time period at the different positions into a thickness time series input matrix. Accordingly, considering that the thickness data at the different positions have characteristic information in the time dimension, and the thickness data at different positions have an important influence and effect on the consistency of the layers in the intelligent processing process of the polarizer. Based on this, in the technical solution of the present application, the thickness data of multiple predetermined time points within the predetermined time period at the different positions. It should be understood that arranging the thickness data at different positions in chronological order can retain the time characteristic information of the thickness data, that is, the thickness data at different time points are arranged in order in the matrix, which can better capture the changing trend of the thickness data over time, help analyze the dynamic changes in the coating process, and thus better understand and utilize the thickness data in the coating process.
[0052] In the embodiment of the present application, the flow value timing arrangement module 130 is used to arrange the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension into a flow timing input vector and then transpose and multiply it with itself to obtain a flow timing input matrix. Figure 3 is a block diagram of a flow value timing arrangement module in a control system of a polarizer intelligent processing device according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 3As shown, the flow value timing arrangement module 130 includes: a timing sample arrangement unit 131, which is used to arrange the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension as the flow timing input vector; and a feature multiplication unit 132, which is used to multiply the flow timing input vector with the transpose of the flow timing input vector to obtain the flow timing input matrix.
[0053] Specifically, in an embodiment of the present application, the time-series sample arrangement unit 131 is used to arrange the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension as the flow time-series input vector. Similarly, considering that the flow value is constantly changing with the change of time, the flow value has characteristic information in the time dimension. Therefore, in the technical solution of the present application, the flow values of the coating liquid at multiple predetermined time points within the predetermined time period are arranged as the flow time-series input vector according to the time dimension, so that the dynamic trend and fluctuation of the flow value over time can be retained, which helps to capture the change pattern of the coating liquid flow over time, discover the periodicity and trend changes of the flow value, and possible abnormal conditions, so as to help analyze and predict future flow conditions, so as to ensure the stability of the coating process.
[0054] Specifically, in an embodiment of the present application, the feature multiplication unit 132 is used to multiply the flow time series input vector by the transpose of the flow time series input vector to obtain the flow time series input matrix. It should be understood that, considering the mutual correlation between different time points in the flow time series input vector about the characteristic information of the flow value, such as the changing trend and fluctuation of the flow between different time points. Therefore, in the technical solution of the present application, the flow time series input vector is multiplied by the transpose of the flow time series input vector to obtain the flow time series input matrix. That is, by multiplying the flow time series input vector with its transpose, a matrix can be obtained, in which each element represents the correlation between the flow data at the corresponding position, which can analyze the degree of correlation between the flow data at different time points, show the relationship between the flow data at different time points, find out the potential correlation between the flow data, and then reveal the changing trend and correlation between the flow data, which provides a basis for further analysis and processing of the flow data characteristics.
[0055] In an embodiment of the present application, the feature association module 140 is used to perform perspective mapping matching based on eigenvalue granularity association on the thickness time series input matrix and the flow time series input matrix to obtain the thickness flow time series association input matrix. It should be understood that in the technical solution of the present application, the thickness time series input vector and the flow time series input vector have relatively significant feature heterogeneity, wherein feature heterogeneity refers to the existence of different properties or statistical characteristics between different features (such as thickness and flow), which leads to the mismatch between the feature manifold and the feature space caused by heterogeneity when performing feature fusion of the thickness time series input vector and the flow time series input vector, which will cause the thickness flow time series association input vector to fail to have feature manifold smoothness, that is, due to the different properties of the data themselves, their distribution in high-dimensional space is also different, and this mismatch will make the feature manifolds not correspond well to each other, thereby affecting the certainty of their feature expression. Specifically, due to the non-smooth feature manifold, the model may find it difficult to capture effective patterns from the data, resulting in uncertainty in feature expression and decreased prediction performance. Based on this, in the technical solution of the present application, the thickness timing input matrix and the flow timing input matrix are matched by perspective mapping based on eigenvalue granularity association to obtain a thickness-flow timing association input matrix.
[0056] Specifically, in the embodiment of the present application, the feature association module includes: a thickness-flow time series feature expansion unit, which is used to sequentially expand the thickness time series input matrix and the flow time series input matrix to obtain a thickness time series input vector and a flow time series input vector; a thickness-flow time series eigenvalue granularity association unit, which is used to perform perspective mapping matching based on eigenvalue granularity association on the thickness time series input vector and the flow time series input vector to obtain a thickness-flow time series association input vector; a thickness-flow time series association feature reconstruction unit, which is used to perform dimension reconstruction on the thickness-flow time series association input vector to obtain the thickness-flow time series association input matrix;
[0057] More specifically, in an embodiment of the present application, the thickness-flow timing characteristic value granularity association unit is used to: construct a thickness timing-flow timing distance topological matrix between the thickness timing input vector and the flow timing input vector; construct a thickness timing-flow timing characteristic value granularity association matrix between the thickness timing input vector and the flow timing input vector; perform topological association modulation on the thickness timing-flow timing characteristic value granularity association matrix based on the thickness timing-flow timing distance topological matrix to obtain a thickness timing-flow timing topological modulation association matrix; use the thickness timing-flow timing topological modulation association matrix as a perspective co-projection space, project the thickness timing input vector and the flow timing input vector to the perspective co-projection space to obtain a perspective modulated thickness timing input vector and a perspective modulated flow timing input vector; and fuse the perspective modulated thickness timing input vector and the perspective modulated flow timing input vector to obtain the thickness-flow timing association input vector.
[0058] In particular, in the technical solution of the present application, the thickness time series input vector and the flow time series input vector are matched by perspective mapping based on eigenvalue granularity association to obtain a thickness flow time series associated input vector according to the following formula; wherein the formula is:
[0059] M t =D(V1,V2)
[0060]
[0061]
[0062] V c =α·V1′+β·V2′
[0063] Wherein, V1 represents the thickness time series input vector, V2 represents the flow time series input vector, and D(V1, V2) is the distance topological matrix between the thickness time series input vector and the flow time series input vector, that is, d i,j =d(v 1i ,v 2j ), d i,j is the eigenvalue of the position (i, j) in the distance topology matrix, d(v 1i ,v 2j ) represents the distance between the characteristic value of the ith position of the thickness time series input vector and the characteristic value of the jth position of the flow time series input vector, and V1 and V2 are both column vectors, M t represents the distance topology matrix, T represents the transpose of the vector, Represents matrix multiplication, M v represents the eigenvalue granularity association matrix, M crepresents the topological modulation correlation matrix, V1' represents the perspective modulation thickness time series input vector, V2' represents the perspective modulation flow time series input vector, α and β represent weighted hyperparameters, V c Represents the fused feature vector.
[0064] Accordingly, in the technical solution of the present application, the thickness time series input vector and the flow time series input vector are matched by perspective mapping based on the eigenvalue granularity association, which first constructs a thickness time series-flow time series distance topological matrix between the thickness time series input vector and the flow time series input vector. This step utilizes the idea of graph theory, takes the eigenvalues of each position in the thickness time series input vector and the flow time series input vector as nodes, and uses the distance metric function to determine the low-dimensional embedding expression of the edges between nodes, so as to construct a feature expression that can quantify the topological association relationship between the thickness time series input vector and the flow time series input vector. This step is crucial to understanding the spatial distribution of data points and their mutual connections, and also provides basic structural information for subsequent steps.
[0065] Next, a thickness timing-flow timing characteristic value granularity association matrix is constructed between the thickness timing input vector and the flow timing input vector. This step focuses on exploring the correlation between the internal attributes of the characteristic vectors. In a specific example, the product between the transposed vectors of the thickness timing input vector and the flow timing input vector is calculated to obtain the thickness timing-flow timing characteristic value granularity association matrix. Furthermore, the thickness timing-flow timing characteristic value granularity association matrix is topologically associated modulated based on the thickness timing-flow timing distance topological matrix to obtain the thickness timing-flow timing topological modulation association matrix. That is, the characteristic value granularity association information of the thickness timing input vector and the flow timing input vector is driven to wander on the distance topological map to construct a low-dimensional modulation perspective co-projection space domain for mapping the thickness timing input vector and the flow timing input vector to a continuous high-dimensional regression space property.
[0066] Next, the thickness timing-flow timing topological modulation association matrix is used as the perspective co-projection space, and the thickness timing input vector and the flow timing input vector are projected into the perspective co-projection space to obtain the perspective modulated thickness timing input vector and the perspective modulated flow timing input vector. That is, the key to this technical solution is to find an implicit third space with modulation capability, in which as much information of the original high-dimensional data as possible is retained, so that the converted new feature vector not only reduces the dimension, but also maintains the relative position relationship between the original features.
[0067] Finally, the perspective modulated thickness timing input vector and the perspective modulated flow timing input vector are fused to obtain the thickness flow timing correlation input vector. Fusion can be achieved in a variety of forms, ranging from simple arithmetic average to complex ensemble learning algorithms, depending on the application scenario and the goals pursued. The thickness flow timing correlation input vector finally obtained combines the main features of the two sets of input data and provides a more comprehensive data representation, which is not only conducive to improving the performance of machine learning models, but also promotes the ability of cross-modal or multi-source data analysis, thereby significantly enhancing the understanding and utilization efficiency of complex data structures.
[0068] In the embodiment of the present application, the multi-scale time series association module 150 is used to perform multi-scale time series association perception on the thickness-flow time series association input matrix to obtain a thickness-flow time series association multi-scale feature matrix. Figure 4 is a block diagram of a multi-scale time series association module in a control system of a polarizer intelligent processing device according to an embodiment of the present application. Specifically, in the embodiment of the present application, Figure 4 As shown, the multi-scale time series association module 150 includes: a multi-scale feature extraction unit 151, which is used to input the thickness flow time series association input matrix into a multi-scale time series association sensor including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow time series association feature matrix and a second-scale thickness flow time series association feature matrix; and a multi-scale cascade unit 152, which is used to cascade the first-scale thickness flow time series association feature matrix and the second-scale thickness flow time series association feature matrix to obtain the thickness flow time series association multi-scale feature matrix.
[0069] Specifically, in an embodiment of the present application, the multi-scale feature extraction unit 151 is used to input the thickness flow temporal correlation input matrix into a multi-scale temporal correlation perceptron including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow temporal correlation feature matrix and a second-scale thickness flow temporal correlation feature matrix. In particular, the multi-scale temporal correlation perceptron including the first convolutional neural network model and the second convolutional neural network model includes: a first convolutional layer, a second convolutional layer in parallel with the first convolutional layer, wherein the first convolutional layer has a first one-dimensional convolutional kernel of a first length, and the second convolutional layer has a second one-dimensional convolutional kernel of a second length, and the first length is different from the second length. Accordingly, considering that the relationship between thickness and flow may be affected by multiple factors during the coating process, the thickness flow temporal correlation input matrix has feature information at different scales, and the information at different time scales may be of great significance for understanding the correlation between thickness and flow. Based on this, in the technical solution of the present application, the thickness flow temporal correlation input matrix is input into a multi-scale temporal correlation perceptron including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow temporal correlation feature matrix and a second-scale thickness flow temporal correlation feature matrix. It is worth mentioning that the multi-scale feature extraction helps to improve the performance and generalization ability of the model. Different convolutional neural network models can learn features at different time scales respectively. Integrating these features together can better understand the complex relationship between thickness and flow, thereby better capturing the correlation characteristics between thickness and flow.
[0070] Specifically, in an embodiment of the present application, the multi-scale cascade unit 152 is used to cascade the first-scale thickness flow time series correlation feature matrix and the second-scale thickness flow time series correlation feature matrix to obtain the thickness flow time series correlation multi-scale feature matrix. It should be understood that, considering that the first-scale thickness flow time series correlation feature matrix and the second-scale thickness flow time series correlation feature matrix respectively represent the feature information of different scales between flow and thickness data, therefore, in order to more comprehensively represent the comprehensive correlation information between flow value and thickness data, in the technical solution of the present application, the first-scale thickness flow time series correlation feature matrix and the second-scale thickness flow time series correlation feature matrix are cascaded to obtain the thickness flow time series correlation multi-scale feature matrix, which can integrate the feature information extracted at different time scales, thereby more comprehensively capturing the correlation characteristics between thickness and flow, thereby increasing the diversity and richness of the thickness flow time series correlation multi-scale features, thereby improving the expression ability of the thickness flow time series correlation multi-scale features.
[0071] In an embodiment of the present application, the attention enhancement module 160 is used to perform attention enhancement on the thickness flow time series correlation multi-scale feature matrix to obtain a thickness flow time series correlation multi-scale enhanced feature vector. Specifically, in an embodiment of the present application, the attention enhancement module is used to: input the thickness flow time series correlation multi-scale feature matrix into a correlation feature extractor based on a bidirectional attention mechanism to obtain a thickness flow time series correlation multi-scale enhanced feature matrix, and expand the thickness flow time series correlation multi-scale enhanced feature matrix into the thickness flow time series correlation multi-scale enhanced feature vector. Accordingly, in order to enhance the expressive power of the thickness flow time series correlation multi-scale feature matrix and the importance of key features, in the technical solution of the present application, the thickness flow time series correlation multi-scale feature matrix is input into a correlation feature extractor based on a bidirectional attention mechanism to obtain the thickness flow time series correlation multi-scale enhanced feature matrix. In particular, the bidirectional attention mechanism can make the model more flexible in the process of correlation feature extraction, and can dynamically adjust the importance of features according to the input of different time steps, so as to better capture the correlation information between thickness and flow. That is, the bidirectional attention mechanism can capture and extract the key important feature information about the thickness-flow timing correlation in the thickness-flow timing correlation multi-scale feature matrix, helping the model to better focus on the key features of the thickness-flow timing correlation, thereby enhancing the feature characterization capability and improving the modeling effect of the thickness and flow correlation during the coating process.
[0072] In an embodiment of the present application, the feature optimization module 170 is used to perform vector optimization based on probability weights on the thickness flow time series associated multi-scale enhanced feature vector to obtain an optimized thickness flow time series associated multi-scale enhanced feature vector. In particular, in the technical solution of the present application, considering that in the process of multi-scale feature extraction, there may be information loss or noise introduction, resulting in the model learning unnecessary features or noise, increasing the risk of overfitting. If the correlation between the thickness data and the coating liquid flow data in the training data is not strong, or the amount of data is not enough to cover various situations, the model may overfit the noise in the training data instead of the real pattern. The thickness flow time series associated multi-scale enhanced feature vector may have a high-dimensional feature representation, which may cause the model to overfit on the training set and lose generalization ability. If the data distribution of the training set and the test set is inconsistent, the model performs well on the training set but poorly on the test set, indicating that the model overfits the specific features of the training set. In order to avoid the problem of overfitting, in the technical solution of the present application, the thickness flow time series associated multi-scale enhanced feature vector is vector optimized based on probability weights to obtain an optimized thickness flow time series associated multi-scale enhanced feature vector.
[0073] Specifically, in an embodiment of the present application, the feature optimization module includes: a feature frequency calculation unit, which is used to calculate the frequency of the feature values of each position in the thickness flow time series associated multi-scale enhanced feature vector in the thickness flow time series associated multi-scale enhanced feature vector to obtain a classification feature frequency statistical vector; a pre-classification unit, which is used to pass the thickness flow time series associated multi-scale enhanced feature vector through a pre-classifier to obtain a category probability vector; a probability weight measurement unit, which is used to calculate the conditional probability of the classification feature frequency statistical vector relative to the category probability vector to obtain a probability weight measurement vector; and a position point multiplication unit, which is used to calculate the position point multiplication between the thickness flow time series associated multi-scale enhanced feature vector and the probability weight measurement vector to obtain the optimized thickness flow time series associated multi-scale enhanced feature vector. It should be understood that, in response to the above technical problems, the thickness flow time series associated multi-scale enhanced feature vector is optimized based on probability weights to obtain an optimized thickness flow time series associated multi-scale enhanced feature vector, which helps evaluate the degree of influence of each feature on the target variable by calculating the probability weight of the thickness flow time series associated multi-scale enhanced feature vector, thereby identifying those features that are important for prediction, and identifying those features that contribute less to model prediction, and then identifying and retaining the most informative features while performing feature selection and dimensionality reduction to reduce data dimension and complexity, so as to improve the performance of the model, reduce the risk of overfitting, and improve the generalization ability of the model. In addition, by calculating the probability weight of the thickness flow time series associated multi-scale enhanced feature vector, the degree of dependence of the model on different features can be explained, thereby enhancing the interpretability of the model.
[0074] In an embodiment of the present application, the flow control module 180 is used to determine whether the current flow rate should increase, decrease, or remain unchanged based on the optimized thickness flow time series associated multi-scale enhanced feature vector. Specifically, in an embodiment of the present application, the flow control module is used to: input the optimized thickness flow time series associated multi-scale enhanced feature vector into a classifier to obtain a classification result, and the classification result is used to indicate whether the current flow rate should increase, decrease, or remain unchanged. That is, by performing classification processing through the optimized thickness flow time series associated multi-scale enhanced feature vector, it is determined whether the current flow rate should increase, decrease, or remain unchanged. In this way, more accurate coating liquid control can be achieved, manual intervention can be reduced, waste in the coating process can be reduced, and the flow rate of the coating liquid at different locations and time points can be ensured to be the most suitable, thereby improving the coating quality and the consistency and stability of production.
[0075] In summary, the control system 100 of the polarizer intelligent processing equipment based on the embodiment of the present application is explained, which collects the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions in real time by the thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period in real time by the flow meter, and uses deep learning technology to perform data processing, analysis and correlation on the thickness data and the flow value, so as to determine whether the current flow should increase, decrease or remain unchanged. In this way, more accurate coating liquid control can be achieved, manual intervention can be reduced, waste in the coating process can be reduced, and the flow rate of the coating liquid at different positions and time points can be ensured to be the most suitable, thereby improving the coating quality and the consistency and stability of production.
[0076] Figure 5 Flow chart of the control method of the polarizer intelligent processing equipment according to the embodiment of the present application. Figure 5 As shown, the control method of the polarizer intelligent processing equipment according to the embodiment of the present application includes: S110, collecting thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions by a thickness measuring instrument and flow values of the coating liquid at multiple predetermined time points in a predetermined time period by a flow meter; S120, arranging the thickness data at multiple predetermined time points in the predetermined time period at different positions into a thickness time series input matrix; S130, arranging the flow values of the coating liquid at multiple predetermined time points in the predetermined time period according to the time dimension into a flow time series input vector and then transposing and multiplying it with itself to obtain a flow time series input matrix; S140, performing eigenvalue particle-based calculation on the thickness time series input matrix and the flow time series input matrix. The perspective mapping of the thickness flow correlation is matched to obtain a thickness flow time series correlation input matrix; S150, the thickness flow time series correlation input matrix is subjected to multi-scale time series correlation perception to obtain a thickness flow time series correlation multi-scale feature matrix; S160, the thickness flow time series correlation multi-scale feature matrix is subjected to attention enhancement to obtain a thickness flow time series correlation multi-scale enhanced feature vector; S170, the thickness flow time series correlation multi-scale enhanced feature vector is subjected to vector optimization based on probability weight to obtain an optimized thickness flow time series correlation multi-scale enhanced feature vector; and, S180, based on the optimized thickness flow time series correlation multi-scale enhanced feature vector, it is determined whether the current flow should be increased, decreased, or kept unchanged.
[0077] Here, those skilled in the art can understand that the specific operations of each step in the control method of the above-mentioned polarizer intelligent processing equipment have been referred to above. Figures 1 to 4 The control system of the polarizer intelligent processing equipment has been introduced in detail, and therefore, its repeated description will be omitted.
[0078] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.
Claims
1. A control system for polarizer intelligent processing equipment, characterized in that: include: A coating processing parameter acquisition module, which is used to acquire the thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions through a thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period acquired by a flow meter; A thickness arrangement module, used for arranging the thickness data of a plurality of predetermined time points within a predetermined time period at the different positions into a thickness time series input matrix; A flow value time series arrangement module, used to arrange the flow values of the coating liquid at multiple predetermined time points within the predetermined time period according to the time dimension into a flow time series input vector and then transpose and multiply it with itself to obtain a flow time series input matrix; A feature association module is used to perform perspective mapping matching based on eigenvalue granularity association on a thickness time series input matrix and a flow time series input matrix to obtain a thickness flow time series association input matrix; A multi-scale time series association module, used for performing multi-scale time series association perception on the thickness-flow time series association input matrix to obtain a thickness-flow time series association multi-scale feature matrix; An attention enhancement module, used for performing attention enhancement on the thickness-flow time series correlation multi-scale feature matrix to obtain a thickness-flow time series correlation multi-scale enhanced feature vector; A feature optimization module, used for performing vector optimization based on probability weight on the thickness-flow time series correlation multi-scale enhancement feature vector to obtain an optimized thickness-flow time series correlation multi-scale enhancement feature vector; The flow control module is used to determine whether the current flow should be increased, decreased or kept unchanged based on the optimized thickness flow time series associated multi-scale enhanced feature vector.
2. The control system of the polarizer intelligent processing equipment according to claim 1, characterized in that: The flow value timing arrangement module includes: A time series sample arrangement unit, used for arranging the flow values of the coating liquid at a plurality of predetermined time points within the predetermined time period according to the time dimension into the flow time series input vector; A feature multiplication unit is used to multiply the flow time series input vector by the transpose of the flow time series input vector to obtain the flow time series input matrix.
3. The control system of the polarizer intelligent processing equipment according to claim 2, characterized in that: The feature association module includes: A thickness flow time series feature expansion unit is used to sequentially expand the thickness time series input matrix and the flow time series input matrix to obtain a thickness time series input vector and a flow time series input vector; A thickness-flow time series characteristic value granularity association unit, used for performing perspective mapping matching based on characteristic value granularity association on the thickness time series input vector and the flow time series input vector to obtain a thickness-flow time series associated input vector; A thickness-flow time series correlation feature reconstruction unit, used for reconstructing the dimension of the thickness-flow time series correlation input vector to obtain the thickness-flow time series correlation input matrix; Wherein, the thickness flow time series characteristic value granularity association unit is used to: Constructing a thickness timing-flow timing distance topological matrix between the thickness timing input vector and the flow timing input vector; Constructing a thickness time series-flow time series characteristic value granularity association matrix between the thickness time series input vector and the flow time series input vector; Based on the thickness time series-flow time series distance topological matrix, the thickness time series-flow time series characteristic value granularity association matrix is topologically modulated to obtain a thickness time series-flow time series topological modulation association matrix; Taking the thickness timing-flow timing topological modulation association matrix as the perspective co-projection space, projecting the thickness timing input vector and the flow timing input vector into the perspective co-projection space to obtain the perspective modulation thickness timing input vector and the perspective modulation flow timing input vector; The viewing angle modulated thickness timing input vector and the viewing angle modulated flow timing input vector are fused to obtain the thickness-flow timing associated input vector.
4. The control system of the polarizer intelligent processing equipment according to claim 3, characterized in that: The multi-scale temporal association module comprises: A multi-scale feature extraction unit, used for inputting the thickness flow temporal correlation input matrix into a multi-scale temporal correlation perceptron including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow temporal correlation feature matrix and a second-scale thickness flow temporal correlation feature matrix; The multi-scale cascading unit is used to cascade the first-scale thickness flow time series correlation feature matrix and the second-scale thickness flow time series correlation feature matrix to obtain the thickness flow time series correlation multi-scale feature matrix.
5. The control system of the polarizer intelligent processing equipment according to claim 4, characterized in that: The attention enhancement module is used to: input the thickness flow time series correlation multi-scale feature matrix into the correlation feature extractor based on the bidirectional attention mechanism to obtain the thickness flow time series correlation multi-scale enhanced feature matrix, and expand the thickness flow time series correlation multi-scale enhanced feature matrix into the thickness flow time series correlation multi-scale enhanced feature vector.
6. The control system of the polarizer intelligent processing equipment according to claim 5, characterized in that: The feature optimization module comprises: A characteristic frequency calculation unit, used for calculating the frequency of the characteristic value of each position in the thickness flow time series associated multi-scale enhanced characteristic vector in the thickness flow time series associated multi-scale enhanced characteristic vector to obtain a classification characteristic frequency statistical vector; A pre-classification unit, used for passing the thickness-flow time series associated multi-scale enhanced feature vector through a pre-classifier to obtain a class probability vector; A probability weight measurement unit, used to calculate the conditional probability of the classification feature frequency statistical vector relative to the category probability vector to obtain a probability weight measurement vector; The position point multiplication unit is used to calculate the position point multiplication between the thickness flow time series associated multi-scale enhanced feature vector and the probability weight metric vector to obtain the optimized thickness flow time series associated multi-scale enhanced feature vector.
7. The control system of the polarizer intelligent processing equipment according to claim 6, characterized in that: The flow control module is used for: The optimized thickness-flow time series associated multi-scale enhanced feature vector is input into a classifier to obtain a classification result, and the classification result is used to indicate whether the current flow should be increased, decreased, or kept unchanged.
8. A control method for polarizer intelligent processing equipment, characterized in that: include: The thickness data of the coating at multiple predetermined time points in a predetermined time period at different positions collected by the thickness measuring instrument and the flow value of the coating liquid at multiple predetermined time points in a predetermined time period collected by the flow meter; Arranging the thickness data of a plurality of predetermined time points within a predetermined time period at the different positions into a thickness time series input matrix; Arranging the flow values of the coating liquid at a plurality of predetermined time points within the predetermined time period into a flow time series input vector according to the time dimension and then transposing and multiplying the vector with itself to obtain a flow time series input matrix; Performing perspective mapping matching based on eigenvalue granularity association on the thickness time series input matrix and the flow time series input matrix to obtain a thickness and flow time series association input matrix; Perform multi-scale time series correlation perception on the thickness-flow time series correlation input matrix to obtain a thickness-flow time series correlation multi-scale feature matrix; Performing attention enhancement on the thickness-flow time series correlation multi-scale feature matrix to obtain a thickness-flow time series correlation multi-scale enhanced feature vector; Performing vector optimization based on probability weight on the thickness-flow time series associated multi-scale enhanced feature vector to obtain an optimized thickness-flow time series associated multi-scale enhanced feature vector; Based on the optimized thickness-flow time series associated multi-scale enhanced feature vector, it is determined whether the current flow should be increased, decreased, or kept unchanged.
9. The control method of the polarizer intelligent processing equipment according to claim 8, characterized in that: Arranging the flow values of the coating liquid at a plurality of predetermined time points within the predetermined time period into a flow time series input vector according to the time dimension and then transposing and multiplying the vector with itself to obtain a flow time series input matrix, including: Arranging the flow values of the coating liquid at a plurality of predetermined time points within the predetermined time period according to the time dimension as the flow time series input vector; The flow timing input vector is multiplied by the transpose of the flow timing input vector to obtain the flow timing input matrix.
10. The control method of the polarizer intelligent processing equipment according to claim 9, characterized in that: The thickness-flow time series correlation input matrix is subjected to multi-scale time series correlation perception to obtain a thickness-flow time series correlation multi-scale feature matrix, including: Inputting the thickness flow temporal correlation input matrix into a multi-scale temporal correlation perceptron including a first convolutional neural network model and a second convolutional neural network model to obtain a first-scale thickness flow temporal correlation feature matrix and a second-scale thickness flow temporal correlation feature matrix; The first-scale thickness flow time series correlation feature matrix and the second-scale thickness flow time series correlation feature matrix are cascaded to obtain the thickness flow time series correlation multi-scale feature matrix.
