Feedforward intelligent coating thickness closed-loop control method and system

Abstract

The present invention relates to the field of coating thickness control. Disclosed are a feedforward intelligent coating thickness closed-loop control method and system. The method comprises: collecting data to train a model; constructing time-series features of the data and multi-dimensional correlation features between the data; building and training a neural network that uses the time-series features and the multi-dimensional correlation features between the data as input, and outputs the classification of coating thickness trend adjustment directions and the adjustment amounts for each execution mechanism, so as to obtain a trained neural network model; and executing coating thickness closed-loop control on the basis of the trained neural network model. By means of predicting a coating thickness trend direction and the adjustment amount of each control mechanism, the present invention reduces the hysteresis of closed-loop control, thereby improving the uniformity of coating thickness.

Description

A feedforward intelligent coating thickness closed-loop control method and system Technical Field

[0001] This invention relates to the field of coating thickness control, and more particularly to a feedforward intelligent coating thickness closed-loop control method and system. Background Technology

[0002] Electrode sheets are the core of lithium-ion batteries, and coating is a crucial process in their production. During coating, the quality of the coating amount, or coating thickness, directly affects the battery's consistency and safety. Currently, the mainstream production method involves operators observing and manually adjusting the coating amount. This can affect the timeliness of adjustments and lead to human error, resulting in poor coating consistency and material loss exceeding the process limits.

[0003] During the coating process, the coating thickness is uneven due to factors such as slurry temperature or viscosity, requiring adjustment. Current technology primarily relies on manual adjustment of the coating thickness gauge to observe fluctuations in the data and determine if they deviate from the target value. This adjustment mainly involves adjusting the slurry pump mechanism, the distance between the coating head lip and the substrate, and the size of the coating lip slit. However, manual adjustment is susceptible to human error, leading to significant errors, inadequate or excessive adjustment, resulting in uneven coating thickness and poor coating quality. Furthermore, manual adjustment is inefficient and detrimental to automated production.

[0004] Summary of the Invention

[0005] The main objective of this invention is to solve the technical problems of uneven coating thickness and poor yield in existing technologies. A feedforward intelligent coating thickness closed-loop control method includes the following steps:

[0006] The collected data is used to train the model. The data includes coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, and adjustment amount.

[0007] Construct time series features of data and multi-dimensional correlation features between data points;

[0008] Build and train a neural network, input the above time series features and multi-dimensional correlation features between data, and output a neural network model that classifies the coating thickness trend adjustment direction and the adjustment amount of each actuator, thus obtaining a trained neural network model.

[0009] If the coating thickness does not meet expectations, then calculate △P61. The formula for calculating △P61 is as follows:

[0010] Where σ represents the surface tension of the slurry, H represents the coating height, θ is the static contact angle between the slurry and the substrate, φ is the static contact angle between the slurry and the coating head die, l1 is the length upstream of the coating bead, l2 is the length downstream of the coating bead, Uc is the coating speed, W is the slit width of the coating head, and h m The predicted coating thickness is given by h, which represents the stable coating thickness, b, which represents the coating width, and μ, which represents the density of the slurry.

[0011] Based on the calculation results of △P61, adjust the coating speed, coating gap, and lip flow rate.

[0012] A second aspect of the present invention provides a feedforward intelligent coating thickness closed-loop control system, comprising:

[0013] The data acquisition unit is used to collect data for training the model. The data includes coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, and adjustment amount.

[0014] Feature building unit, used to construct time series features of data and multi-dimensional correlation features between data;

[0015] The model building and training unit is used to build and train the neural network. It takes the above time series features and multi-dimensional correlation features between data as input, and outputs a neural network model that classifies the coating thickness trend adjustment direction and the adjustment amount of each actuator, thus obtaining a trained neural network model.

[0016] The control execution unit is used to perform closed-loop control of coating thickness based on the trained neural network model.

[0017] A third aspect of the present invention provides an electronic device, comprising: a memory and at least one processor, wherein the memory stores instructions, and the memory and the at least one processor are interconnected via a circuit; the at least one processor invokes the instructions in the memory to cause the electronic device to execute the feedforward intelligent coating thickness closed-loop control method described above.

[0018] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the feedforward intelligent coating thickness closed-loop control method described above.

[0019] The present invention has the following beneficial effects:

[0020] This invention predicts the coating thickness trend and the adjustment amount of each control mechanism, and inputs this information in advance to the corresponding adjustment mechanisms, thereby reducing the lag of closed-loop control and improving the uniformity of coating thickness.

[0021] To address the lag in measurement feedback adjustment, this invention constructs a self-learning multi-feature fusion network model in advance based on internal factors and external trend characteristics affecting coating thickness. This model predicts the subsequent trend changes in coating weight and thickness, thereby controlling the slurry adjustment structure in advance, further improving coating thickness uniformity and product quality. Attached Figure Description

[0022] Figure 1 is a flowchart of the feedforward intelligent coating thickness closed-loop control method provided in an embodiment of the present invention;

[0023] Figure 2 is an architecture diagram of the interactive feature extraction module of the present invention.

[0024] Figure 3 is a schematic diagram of the initial flow field during slit coating.

[0025] Figure 4 shows the morphology of the upstream of the coated beads under four critical conditions. Detailed Implementation

[0026] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] The coating machine consists of multiple fully automatic dies. To ensure optimal operation, these dies need to be adjusted according to the machine's parameters. Current methods typically measure the coating thickness at the wet or dry film application point using methods such as laser, ultrasound, X-ray, or beta-ray. This lag-prone measurement is used as feedback, and the die's slurry dispensing control mechanism automatically adjusts the coating thickness to achieve closed-loop control. While this method achieves basic automation, it relies on readjustment based on measurement feedback, resulting in significant lag and impacting coating thickness uniformity and yield.

[0028] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to Figure 1. The first embodiment of the feedforward intelligent coating thickness closed-loop control method in the present invention includes:

[0029] The collected data is used to train the model. The data includes coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, and adjustment amount.

[0030] Specifically, it includes:

[0031] Continuous data collection: Using sensors and measuring equipment, various parameters during the coating process are collected in real time, including coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, etc.

[0032] Adjustment mechanism data: Simultaneously record the adjustment label and adjustment amount of the adjustment mechanism. This data is crucial for subsequent analysis of the relationship between the adjustment process and the coating thickness.

[0033] Construct time series features of data and multi-dimensional correlation features between data points;

[0034] Areal density thickness prediction using time-series data from areal density thickness gauges is based on the premise that "the fluctuations in time-series data imply underlying patterns." Therefore, the main task of the model is to learn these fluctuation patterns. Manually constructing combined features is time-consuming and laborious. This invention employs a neural network model based on factorization machines to enhance the mining of input data and automatically learn the combined effects of different time-series features on areal density thickness trends. Specifically, it includes:

[0035] Time series characteristics:

[0036] Timestamp: Records the timestamp of each data point, which is used for subsequent analysis of the data's trend over time.

[0037] Time interval: Calculates the time interval between adjacent data points to analyze the frequency and rate of data change.

[0038] Time series statistics: Calculate statistical data for each time period (such as per minute, per hour), such as average, maximum, minimum, standard deviation, etc., to capture the dynamic changes in the coating process.

[0039] Multi-dimensional correlation features:

[0040] Correlation analysis: Calculate the correlation coefficients between different parameters to understand their linear relationship.

[0041] Lag and Lead Features: Considering the time delay effect between different parameters, lag and lead features are constructed to capture the time dependencies between them.

[0042] Feature combination: Based on process knowledge and experience, different parameters are combined to create new features to capture the nonlinear relationships between them.

[0043] Build and train a neural network, input the above time series features and multi-dimensional correlation features between data, and output a neural network model that classifies the coating thickness trend adjustment direction and the adjustment amount of each actuator, thus obtaining a trained neural network model.

[0044] Model selection: Choose an appropriate neural network structure based on the characteristics of the problem, such as a combination of recurrent neural networks (RNN), long short-term memory networks (LSTM), or convolutional neural networks (CNN) and fully connected layers (DNN), to process time series data and capture multi-dimensional correlation features.

[0045] Input layer design: The preprocessed time series features and correlation features are used as input to the neural network to ensure that the format and dimension of the input data match the model structure.

[0046] Hidden layer design: Based on the complexity of the problem and the characteristics of the data, design an appropriate number of hidden layers and neurons, as well as activation functions and regularization strategies.

[0047] Output layer design: The output layer should include neurons for classifying the coating thickness trend adjustment direction (e.g., using the softmax function for multi-class classification) and neurons for adjusting the magnitude of each actuator (e.g., using a linear output layer for regression).

[0048] Model training: The neural network is trained using the collected dataset, and the parameters of the model are optimized through the backpropagation algorithm to minimize the prediction error.

[0049] Model evaluation: The trained model is evaluated using a validation set or a test set to verify its generalization ability and prediction performance.

[0050] Based on the trained neural network model, closed-loop control of coating thickness is executed.

[0051] Real-time data input: The real-time collected data is preprocessed and then input into the trained neural network model.

[0052] Predicted output: The model outputs predicted values ​​based on the input data, classifying the direction of coating thickness trend adjustment and the magnitude of adjustment of each actuator.

[0053] Actuator control: Based on the model's predicted output, the actuators of the coating equipment are controlled to make corresponding adjustments, such as adjusting the pump speed and lip opening, in order to achieve closed-loop control of the coating thickness.

[0054] Feedback and Adjustment: Continuously collect new data and monitor the actual value of coating thickness. Compare the actual value with the predicted value and adjust the model parameters or control strategy according to the error to optimize the control effect of coating thickness.

[0055] If the coating thickness does not meet expectations, then calculate △P61. The formula for calculating △P61 is as follows:

[0056] Where σ represents the surface tension of the slurry, H represents the coating height, θ is the static contact angle between the slurry and the substrate, φ is the static contact angle between the slurry and the coating head die, l1 is the length upstream of the coating bead, l2 is the length downstream of the coating bead, Uc is the coating speed, W is the slit width of the coating head, and h m The predicted coating thickness is given by h, which represents the stable coating thickness, b, which represents the coating width, and μ, which represents the density of the slurry.

[0057] Based on the calculation results of △P61, adjust the coating speed, coating gap, and lip flow rate.

[0058] For example, the negative electrode slurry of a power lithium battery is fed into the homogenizing chamber through the die inlet channel via the feeding system. After filling the homogenizing chamber, the slurry flows out uniformly from the slit channel, generating relative flow between the die and the substrate and gradually stabilizing to form an external flow field. When the slurry contacts the substrate, due to the relative velocity between the coating head and the substrate, the slurry briefly accumulates at the lip of the coating head before stabilizing, forming a crescent shape. The crescent-shaped liquid in the slit forms a liquid bridge due to the relative movement of the coating head and the substrate. The liquid portion contained between the lip and the substrate is called a coating bead, and its flow field diagram is shown in Figure 3. After the coating stabilizes, the shape of the coating bead remains constant, and its flow process can be regarded as the steady flow of a viscous incompressible fluid. By analyzing the coating bead, the calculation formula for ΔP61 is obtained.

[0059] As shown in Figure 4, during the slot coating process, there are four possible morphologies upstream of the coating beads, each corresponding to a critical pressure drop value. The morphologies of each condition are shown in Figure 4-1. By comparing the value of ΔP61 with the critical pressure drop values ​​for the four conditions, the morphology and location upstream of the coating beads can be determined, thus predicting the coating stability under these conditions. The specific coating results for each condition are as follows:

[0060] Based on the calculation results of △P61, adjust the coating speed, coating gap, and lip flow rate as follows:

[0061] When the value of △P61 is greater than 200,000:

[0062] At this point, first adjust the coating speed, then adjust the coating gap, without adjusting the lip flow rate;

[0063] ② When the value of △P61 is 150,000 to 200,000:

[0064] At this point, first adjust the coating speed, then adjust the coating gap. If the desired result is not achieved, fine-tune the lip flow rate.

[0065] ③ When the value of △P61 is 100,000 to 150,000:

[0066] At this point, the adjustment weights of coating gap, coating speed, and lip flow rate are equal, and they can be adjusted simultaneously without regard to the order.

[0067] ④ When the value of △P61 is 50,000 to 100,000:

[0068] At this point, prioritize adjusting the lip flow rate, then adjust the coating gap and coating speed.

[0069] ⑤ When the value of △P61 is less than 50,000:

[0070] At this point, prioritize adjusting the lip flow rate, then adjust the coating gap, and finally adjust the coating speed.

[0071] As a preferred implementation, the construction of data time series features and multi-dimensional correlation features between data includes the following steps:

[0072] Employing a dynamic time-bending algorithm:

[0073] Input: Two time series of thickness gauge readings X = {x0, x1, ... x2} m-1} and Y = {y0, y1, ... y n-1 Their lengths are m and n respectively, where m may not be equal to n.

[0074] Output: The optimal matching path between time series X and Y is W = {w0, w1, ... w}. k}, and the cumulative distance DTW(X,Y) of path W;

[0075] Specifically, it includes:

[0076] Step 1: Construct an m x n cost matrix, where the matrix elements D mn Where matrix d i,j x represents i and y i Euclidean distance between two points Path W in w k Represents the position in the matrix and satisfies the following three constraints:

[0077] (1) Boundary conditions: w1 = D(0,0) and w k = D(m-1, n-1), the two sequences can be of different lengths, as long as the start and end positions are aligned;

[0078] (2) Continuity, if the path element w has been determined k-1 =D(a', b'), then for the next path point w k =D(a,b) needs to satisfy (a-a')≤1 and (b-b')≤1 to ensure that the two time series are continuous during matching and will not skip or miss any points;

[0079] (3) Monotonicity, if the path element w has been determined k-1 =D(a', b'), then for the next path point w k =D(a,b) needs to satisfy (a-a')≥0 and (b-b')≥0 to ensure that the points on the path are monotonically increasing on the time axis, that is, only the right, top, and upper right of the current point can be selected, ensuring that the path will not turn back;

[0080] Step 2: Initialize the cumulative distance: γ(0,0)=d(x0,y0)

[0081] Step 3: Calculate the minimum cumulative distance γ(m-1,n-1) according to the recursive formula:

[0082] γ(i,j)=d(x i ,y j )+min{γ(i-1,j-1),γ(i-1,j),γ(i,j-1)}

[0083] Step 4: Backtrack to find all matching point pairs in path W: Based on the best local path of the previous step, backtrack from the matching point pair (m-1, n-1) all the way to the point (0,0).

[0084] As a preferred embodiment, the construction and training of the neural network includes:

[0085] The feature vector F is obtained by combining low-order features of multi-source time series data using a factorization model. M ;

[0086] The feature vector F is obtained by combining high-order features using a deep neural network. DNN ,

[0087] The feature vector F M and F DNN The features are concatenated together to obtain the concatenated feature vector;

[0088] The concatenated feature vector is used as input and passed to the input layer of the neural network.

[0089] We choose the cross-entropy loss function as the loss function and use the gradient descent optimization algorithm for model training.

[0090] Specifically, as shown in Figure 2, this invention employs an interactive feature extraction module. Firstly, this invention utilizes a factorization model to perform low-order (first-order and second-order) feature combination on the multi-source time-series data features to obtain F. M Secondly, F is obtained by combining high-order features using a deep neural network. DNN Finally, the two results are added together and activated to obtain the final result. As the result of extracting combined features. As shown in the formula:

[0091] The feature vector F is obtained by combining low-order features of multi-source time series data using a factorization model. M ,include:

[0092] F M The formula represents the calculation process of the factorization machine:

[0093] In the formula f i The state vector T in the input data t or I t The i-th dimension feature, f i ,w i ∈R, Part 1 As a first-order combination of features, w i The feature weights to be learned; where V represents j1 and F j1 Multiplying the elements at corresponding positions in the matrix, by The eigenvectors formed by these vectors are called embeddings.

[0094] The feature vector F is obtained by using a deep neural network to perform high-order feature combination. DNN ,include:

[0095] The embedding is used as input to the DNN part, denoted as a. 0 :

[0096] a 0 =[e0,e1…,e m ]

[0097] Let the output of the l-th hidden layer be denoted as a. l+1 :

[0098] a l+1 =σ(W la l +b l )

[0099] W l b l Given the parameters of the l-th layer, the final output of the DNN is:

[0100] F DNN =σ(W H+1 a H +b H+1 )

[0101] Where H represents the number of hidden layers.

[0102] The input layer of the neural network consists of three layers, namely:

[0103] A. First fully connected layer: Assume the input feature vector is x∈R d The weight matrix of the first layer is W. (1) ∈R h×d Where h is the number of neurons in the first layer, and the output of the first layer is Z. (1) =W (1) x+b (1) ;

[0104] Among them, b (1) It is a bias vector, which is then passed through a non-linear activation function f. (1) (·), the activated output is obtained as a (1) =f (1) (z (1) );

[0105] B. Second Fully Connected Layer: Assume the weight matrix of the second layer is W. (2) ∈R h’×h Where h' is the number of neurons in the second layer, and the output of the second layer is Z. (2) =W (2) a (1) +b (2) Then, it is activated by a non-linear activation function f. (2) (·) Obtain the activated output a (2) =f (2) (z (2) );

[0106] C. Third Fully Connected Layer: Assume the weight matrix of the third layer is W. (3) ∈R k×h’ Where k is the number of output categories, and the output of the third layer is Z. (3) =W (3) a (2) +b (3) .

[0107] The output layer of the neural network is:

[0108] Multi-class prediction is performed using the softmax function.

[0109] The Softmax function can be expressed as:

[0110] Where p = (y = iz) (3) Z (3) Let y be the input vector; y represents the category, and i represents the predicted probability.

[0111] Loss function and optimization algorithm:

[0112] We choose the cross-entropy loss function as the loss function and use gradient descent optimization algorithm for model training. Cross-entropy loss function: Assuming the true label is y,

[0113] The cross-entropy loss, with a predicted probability of p, can be expressed as:

[0114] During training, the dataset is divided into a training set, a validation set, and a test set. The training set is used to train the model, the validation set is used to tune the model parameters, and finally the test set is used to evaluate the model performance.

[0115] The above describes the feedforward intelligent coating thickness closed-loop control method in the embodiments of the present invention. The following describes the feedforward intelligent coating thickness closed-loop control device in the embodiments of the present invention. The first embodiment of the feedforward intelligent coating thickness closed-loop control device in the embodiments of the present invention includes:

[0116] The data acquisition unit is used to collect data for training the model. The data includes coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, and adjustment amount.

[0117] Feature building unit, used to construct time series features of data and multi-dimensional correlation features between data;

[0118] The model building and training unit is used to build and train the neural network. It takes the above time series features and multi-dimensional correlation features between data as input, and outputs a neural network model that classifies the coating thickness trend adjustment direction and the adjustment amount of each actuator, thus obtaining a trained neural network model.

[0119] The control execution unit is used to perform closed-loop control of coating thickness based on the trained neural network model.

[0120] Specifically, the feedforward intelligent coating thickness closed-loop control device performs the following steps:

[0121] Step 1: First, collect data to train the model. The data includes coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, adjustment amount, etc.

[0122] Step 2: Preparations before training, including data cleaning, preprocessing, and constructing time series features and multi-dimensional correlation features between data.

[0123] Step 3: Build a classification and regression network. Input the above time series features and multi-dimensional correlation features between data, and output the model as the classification of coating thickness trend adjustment direction and the adjustment amount of each actuator.

[0124] Step 4: When deploying the coating, data such as coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, etc., including adjustment labels and adjustment amount data of the adjustment mechanism.

[0125] Step 5: Preprocess the data collected in Step 4, construct features according to the processing method of the training dataset in Step 2, and input them into the model trained in Step 4;

[0126] Step 6: Output the adjustment direction label and the adjustment amount of each actuator, and input the actuator controller for pump speed, lip opening, distance between coating head and back roller, etc.

[0127] Step 7: The actuator performs closed-loop online adjustment according to the adjustment direction and adjustment amount.

[0128] This invention provides an electronic device structure that can vary significantly depending on configuration or performance. It may include one or more central processing units (CPUs) and memory, and one or more storage media (e.g., one or more mass storage devices) for storing applications or data. The memory and storage media can be temporary or persistent storage. The program stored in the storage media may include one or more modules (not shown in the figures), each module including a series of instruction operations on the electronic device. Furthermore, the processor may be configured to communicate with the storage media and execute the series of instruction operations stored in the storage media on the electronic device.

[0129] The electronic device may also include one or more power supplies, one or more wired or wireless network interfaces, one or more input / output interfaces, and / or one or more operating systems, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that the structure of the electronic device does not constitute a limitation on the electronic device itself, and may include more or fewer components than described above, or combine certain components, or have different component arrangements.

[0130] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the steps of a feedforward intelligent coating thickness closed-loop control method.

[0131] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system, device, or unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0132] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0133] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A feedforward intelligent coating thickness closed-loop control method, characterized in that, Includes the following steps: The collected data is used to train the model. The data includes coating thickness, pump speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, and adjustment amount. Construct time series features of data and multi-dimensional correlation features between data points; Build and train a neural network, input the above time series features and multi-dimensional correlation features between data, and output a neural network model that classifies the coating thickness trend adjustment direction and the adjustment amount of each actuator, thus obtaining a trained neural network model. The coating thickness is predicted based on the trained neural network model.

2. A feedforward intelligent coating thickness closed-loop control method according to claim 1, characterized in that, If the predicted coating thickness does not meet expectations, then △P61 is calculated. The formula for calculating △P61 is as follows: Where σ represents the surface tension of the slurry, H represents the coating height, θ is the static contact angle between the slurry and the substrate, φ is the static contact angle between the slurry and the coating head die, l1 is the length upstream of the coating bead, l2 is the length downstream of the coating bead, Uc is the coating speed, W is the slit width of the coating head, and h m The predicted coating thickness is given by h, which represents the stable coating thickness, b, which represents the coating width, and μ, which represents the density of the slurry. Based on the calculation results of △P61, adjust the coating speed, coating gap, and lip flow rate.

3. A feedforward intelligent coating thickness closed-loop control method according to claim 1, characterized in that, The adjustment of coating speed, coating gap, and lip flow rate based on the calculation results of △P61 is as follows: When the value of △P61 is greater than 200,000: At this point, first adjust the coating speed, then adjust the coating gap, without adjusting the lip flow rate; ② When the value of △P61 is 150,000 to 200,000: At this point, first adjust the coating speed, then adjust the coating gap. If the desired result is not achieved, fine-tune the lip flow rate. ③ When the value of △P61 is 100,000 to 150,000: At this point, the adjustment weights of coating gap, coating speed, and lip flow rate are equal, and they can be adjusted simultaneously without regard to the order. ④ When the value of △P61 is 50,000 to 100,000: At this point, prioritize adjusting the lip flow rate, then adjust the coating gap and coating speed. ⑤ When the value of △P61 is less than 50,000: At this point, prioritize adjusting the lip flow rate, then adjust the coating gap, and finally adjust the coating speed.

4. A feedforward intelligent coating thickness closed-loop control method according to claim 1, characterized in that, The construction of time series features of the data and multi-dimensional correlation features between data includes the following steps: Employing a dynamic time-bending algorithm: Input: Two time series of thickness gauge readings X = {x0, x1, ..., x...} m-1 } and Y = {y0, y1, ... y n-1 Their lengths are m and n respectively, where m may not be equal to n. Output: The optimal matching path between time series X and Y is W = {w0, w1, ..., w...} k }, and the cumulative distance DTW(X,Y) of path W; Specifically, it includes: Step 1: Construct an m x n cost matrix, where the matrix elements D mn Where matrix d i,j x represents i and y i Euclidean distance between two points Path W in w k Represents the position in the matrix and satisfies the following three constraints: (1) Boundary conditions: w1 = D(0,0) and w k = D(m-1, n-1), the two sequences can be of different lengths, as long as the start and end positions are aligned; (2) Continuity, if the path element w has been determined k-1 =D(a', b'), then for the next path point w k =D(a,b) needs to satisfy (a-a')≤1 and (b-b')≤1 to ensure that the two time series are continuous during matching and will not skip or miss any points; (3) Monotonicity, if the path element w has been determined k-1 =D(a', b'), then for the next path point w k =D(a,b) needs to satisfy (a-a')≥0 and (b-b')≥0 to ensure that the points on the path are monotonically increasing on the time axis, that is, only the right, top, and upper right of the current point can be selected, ensuring that the path will not turn back; Step 2: Initialize the cumulative distance: γ(0,0)=d(x0,y0) Step 3: Calculate the minimum cumulative distance γ(m-1,n-1) according to the recursive formula: γ(i,j)=d(x i ,y j )+min{γ(i-1,j-1),γ(i-1,j),γ(i,j-1)} Step 4: Backtrack to find all matching point pairs in path W: Based on the best local path of the previous step, backtrack from the matching point pair (m-1, n-1) all the way to the point (0,0).

5. A feedforward intelligent coating thickness closed-loop control method according to claim 1, characterized in that, The construction and training of the neural network includes: The feature vector F is obtained by combining low-order features of multi-source time series data using a factorization model. M ; The feature vector F is obtained by combining high-order features using a deep neural network. DNN , The feature vector F M and F DNN The features are concatenated together to obtain the concatenated feature vector; The concatenated feature vector is used as input and passed to the input layer of the neural network. We choose the cross-entropy loss function as the loss function and use the gradient descent optimization algorithm for model training.

6. A feedforward intelligent coating thickness closed-loop control method according to claim 5, characterized in that, The feature vector F is obtained by using a factorization model to perform low-order feature combination on the features of multi-source time series data. M ,include: F M The formula represents the calculation process of the factorization machine: In the formula f i The state vector T in the input data t Or I t The i-th dimension feature, f i ,w i ∈R, Part 1 As a first-order combination of features, w i The feature weights to be learned; where V represents j1 and F j1 Multiplying the elements at corresponding positions in the matrix, by The eigenvectors constructed by these vectors are called embeddings. The feature vector F is obtained by using a deep neural network to perform high-order feature combination. DNN ,include: The embedding is used as input to the DNN part, denoted as a. 0 : a 0 =[e0,e1···,e m ] Let the output of the l-th hidden layer be denoted as a. l+1 : a l+1 =σ(W l a l +b l ) W l b l Given the parameters of the l-th layer, the final output of the DNN is: F DNN =σ(W H+1 a H +b H+1 ) Where H represents the number of hidden layers.

7. A feedforward intelligent coating thickness closed-loop control method according to claim 5, characterized in that, The input layer of the neural network consists of three layers, namely: A. First fully connected layer: Assume the input feature vector is x∈R d The weight matrix of the first layer is W. (1) ∈R h×d Where h is the number of neurons in the first layer, and the output of the first layer is Z. (1) =W (1) x+b (1) ; Among them, b (1) It is a bias vector, which is then passed through a non-linear activation function f. (1) (·), the activated output is obtained as a (1) =f (1) (z (1) ); B. Second Fully Connected Layer: Assume the weight matrix of the second layer is W. (2) ∈R h’×h Where h' is the number of neurons in the second layer, and the output of the second layer is Z. (2) =W (2) a (1) +b (2) Then, it is activated by a non-linear activation function f. (2) (·) Obtain the activated output a (2) =f (2) (z (2) ); C. Third Fully Connected Layer: Assume the weight matrix of the third layer is W. (3) ∈R k×h’ Where k is the number of output categories, and the output of the third layer is Z. (3) =W (3) a (2) +b (3) .

8. A feedforward intelligent coating thickness closed-loop control system, characterized in that, The system includes: The data acquisition unit is used to collect data for training the model. The data includes coating thickness, pump... Speed, lip opening, distance between coating head and back roller, machine speed, mold cavity pressure, feeding pressure, slurry temperature, viscosity, adjustment mechanism adjustment label, adjustment amount; Feature building unit, used to construct time series features of data and multi-dimensional correlation features between data; The model building and training unit is used to build and train the neural network. It takes the above time series features and multi-dimensional correlation features between data as input, and outputs a neural network model that classifies the coating thickness trend adjustment direction and the adjustment amount of each actuator, thus obtaining a trained neural network model. The control execution unit is used to perform closed-loop control of coating thickness based on the trained neural network model.

9. An electronic device comprising a memory and at least one processor, wherein the memory stores instructions; The at least one processor invokes the instructions in the memory to cause the electronic device to perform the steps of the feedforward intelligent coating thickness closed-loop control method as described in any one of claims 1-7.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the various steps of the feedforward intelligent coating thickness closed-loop control method as described in any one of claims 1-7.

Images