Sixteen-line four-roller plating line production control method

Abstract

The invention discloses a sixteen-line four-roller electroplating line production control method, and relates to the technical field of electroplating production line control, and the method comprises the steps: deploying edge nodes on all production lines, and collecting electroplating liquid state data in real time through a sensor; according to the state data of the electroplating liquid, identifying the state data abnormity of the electroplating liquid by utilizing a machine learning algorithm; collecting visual image data and infrared image data of the plating line by utilizing a visual sensor and infrared sensing equipment; based on the visual image data and the infrared image data, carrying out quality evaluation on the plating line through an image processing algorithm; according to the abnormal state of the electroplating liquid and the quality evaluation result of the electroplating line, the state parameters of the electroplating liquid are adjusted through a self-adaptive algorithm; according to the method, the electroplating liquid state and the electroplating line quality evaluation result are comprehensively considered, so that the consistency and stability of the electroplating quality are remarkably improved.

Description

Technical Field

[0001] The invention relates to the technical field of electroplating production line control, in particular to a production control method for a sixteen-line four-roller electroplating line. Background Art

[0002] Traditional electroplating production lines usually use a single plating tank and a small number of rollers. This configuration is more efficient in small-scale production, but in large-scale production, especially in electroplating lines with multiple production lines working in parallel, such as sixteen lines and four rollers, there are significant technical bottlenecks.

[0003] In a sixteen-line four-drum electroplating production line, ensuring the optimal electroplating quality of each line and being able to adjust the state of the plating solution in real time is a major challenge facing the current electroplating technology field. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a sixteen-line four-roller electroplating line production control method to solve the problem of real-time adjustment of the state of the electroplating solution.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides a sixteen-line four-roller electroplating line production control method, which comprises:

[0008] Deploy edge nodes on all production lines and use sensors to collect real-time data on the status of the electroplating solution;

[0009] Based on the status data of the electroplating solution, the machine learning algorithm is used to identify the abnormal status data of the electroplating solution;

[0010] Use visual sensors and infrared sensing equipment to collect visual image data and infrared image data of the electroplating line;

[0011] Based on visual image data and infrared image data, the quality of the electroplating line is evaluated through image processing algorithms;

[0012] According to the abnormal state of the plating solution and the quality assessment results of the plating line, the state parameters of the plating solution are adjusted through an adaptive algorithm.

[0013] As a preferred solution of the sixteen-line four-roller electroplating line production control method of the present invention, wherein: the edge nodes are deployed on all production lines, and sensors are used to collect electroplating liquid status data in real time. The specific steps are as follows:

[0014] Using sensors on edge nodes, real-time data on the status of electroplating solutions on all production lines is collected;

[0015] The real-time collected electroplating liquid status data is smoothed and standardized on the edge node.

[0016] As a preferred solution of the sixteen-line four-roller electroplating line production control method of the present invention, wherein: according to the electroplating liquid state data, the state data abnormality of the electroplating liquid is identified by using a machine learning algorithm, and the specific steps are as follows:

[0017] Constructing a state data set according to the state data of the electroplating solution;

[0018] According to the range of the status data set for the electroplating solution on the production line, the data that is not within the range of the status data set is labeled as abnormal, and the data that is within the range of the status data set is labeled as normal;

[0019] Mix state data with abnormal labels and normal labels and divide them into training set and test set;

[0020] Based on nonlinear support vector machine (SVM), a plating solution state judgment model is built on the edge node;

[0021] Input the training set into the electroplating solution state judgment model and calculate the target loss function;

[0022] Divide the training set into k subsets, use k-fold cross validation, and use one of the subsets as the validation set and the rest as the training set for training and validation;

[0023] According to the range setting of the status data of the electroplating solution on the production line, the accuracy, precision and recall thresholds are preset;

[0024] After each training and verification, the accuracy, precision and recall are calculated. When the accuracy, precision and recall are all greater than the accuracy, precision and recall thresholds, the training ends;

[0025] The test set is input into the electroplating liquid state judgment model to calculate the accuracy, precision and recall. When the accuracy, precision and recall are all greater than the accuracy, precision and recall thresholds, the electroplating liquid state judgment model training is completed.

[0026] Integrate the incremental learning mechanism into the electroplating solution state judgment model to update the electroplating solution state judgment model in real time;

[0027] The status data of the plating liquid is collected in real time, and the status of the plating liquid on the production line is judged by the plating liquid status judgment model on the edge node.

[0028] As a preferred solution of the sixteen-line four-roller electroplating line production control method of the present invention, wherein: the incremental learning mechanism is integrated into the electroplating liquid state judgment model to update the electroplating liquid state judgment model in real time. The specific steps are as follows:

[0029] By calculating the distance between the newly collected state data and the hyperplane of the electroplating solution state judgment model, the influence degree of the newly collected state data on the electroplating solution state judgment model is judged;

[0030] The newly collected state data are added to the support vector set of the electroplating solution state judgment model, and the hyperplane and decision function are updated according to the new support vector set.

[0031] As a preferred solution of the sixteen-line four-roller electroplating line production control method of the present invention, wherein: the visual image data and infrared image data of the electroplating line are collected by using a visual sensor and an infrared sensing device, and the specific steps are as follows:

[0032] Use visual sensors to collect visual image data of the electroplating line;

[0033] Use infrared sensing equipment to collect infrared image data of the electroplating line;

[0034] Processing of visual image data and infrared image data.

[0035] As a preferred solution of the production control method of the sixteen-line four-roller electroplating line of the present invention, the specific steps of processing the visual image data and the infrared image data are as follows:

[0036] Use bilateral filtering to denoise visual image data;

[0037] The histogram equalization method is used to improve the contrast of the image by adjusting the brightness distribution of the image;

[0038] The non-local means method is used to perform weighted averaging of similar pixels in the image, remove noise while maintaining details, and adjust the brightness and contrast of the infrared image.

[0039] As a preferred solution of the sixteen-line four-roller electroplating line production control method of the present invention, wherein: the quality of the electroplating line is evaluated by an image processing algorithm based on visual image data and infrared image data, and the specific steps are as follows:

[0040] Calculate the thickness of the electroplating layer based on the relationship between the infrared radiation intensity of the infrared sensing device and the thickness of the electroplating layer;

[0041] The electroplating layer area of ​​the electroplating line is extracted by image segmentation method, and the standard deviation of the electroplating layer thickness in each area is calculated as a uniformity evaluation index;

[0042] Classify and annotate known defects in visual image data (including normal, cracks, bubbles, and peeling);

[0043] Each image is annotated with multiple defect categories and a bounding box is provided for each defect area;

[0044] Based on convolutional neural network (CNN), a defect recognition model is constructed;

[0045] By dividing the visual image data with known defects classified and labeled into a training set and a validation set, and inputting the training set into the defect recognition model;

[0046] The input defect recognition model is trained by calculating the cross entropy loss function of the visual images classified and annotated with known defects in the training set;

[0047] Use the gradient descent method to minimize the loss function and iteratively optimize the network weights. After each training, evaluate the performance of the model through the validation set and adjust the hyperparameters.

[0048] Evaluate the performance of the model by calculating the accuracy and recall of the defect recognition model;

[0049] Use the trained defect recognition model to perform defect detection on the newly collected visual image data;

[0050] The overall quality of the electroplating layer is evaluated by comprehensively considering the thickness, uniformity, and surface defects of the electroplating layer.

[0051] As a preferred solution of the sixteen-line four-roller electroplating line production control method of the present invention, wherein: according to the abnormal state of the electroplating solution and the quality evaluation result of the electroplating line, the state parameters of the electroplating solution are adjusted by an adaptive algorithm, and the specific steps are as follows:

[0052] Normalize the abnormal data of the plating solution and the quality assessment results of the plating line;

[0053] Adopting an adaptive control algorithm based on reinforcement learning, the reward function is used to guide the adjustment of the electroplating solution state data;

[0054] During the reinforcement learning training process, the adjustment process of the electroplating solution state data is optimized by the policy gradient method;

[0055] The control parameters of the plating solution are gradually adjusted through the back-propagation algorithm;

[0056] By adjusting the plating liquid state data and recalculating the reward function according to the current state data and quality evaluation results at each iteration;

[0057] When the reward function reaches the maximum value, it means that the electroplating liquid state data is optimal and the electroplating quality of the electroplating line is optimal.

[0058] In a second aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the sixteen-line four-roller electroplating line production control method as described in the first aspect of the present invention is implemented.

[0059] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the sixteen-line four-roller electroplating line production control method as described in the first aspect of the present invention is implemented.

[0060] The beneficial effects of the present invention are as follows: by deploying edge nodes on the production line, the present invention collects the status data of the electroplating solution in real time and performs smoothing and standardization processing, thereby improving the reliability of the data and providing high-quality basic data for subsequent abnormal identification and quality control. By using machine learning algorithms, the abnormal status of the electroplating solution can be identified in real time and adaptively adjusted to ensure the stability of the electroplating solution. The collection and processing of visual and infrared image data effectively improves the detection accuracy of the thickness, uniformity and defects of the electroplating layer, reduces the limitations of manual inspection, and by comprehensively considering the status of the electroplating solution and the quality assessment results of the electroplating line, the intelligently controlled production process significantly improves the consistency and stability of the electroplating quality, reduces production fluctuations and defective products, improves production efficiency, and ensures high standards of product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0062] Figure 1 This is a flow chart of the production control method of the sixteen-line four-roller electroplating line in Example 1. DETAILED DESCRIPTION

[0063] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0064] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0065] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0066] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides a sixteen-line four-roller electroplating line production control method, comprising the following steps: S1. Deploy edge nodes on all production lines and use sensors to collect plating liquid status data in real time.

[0068] Using sensors on edge nodes (including but not limited to temperature sensors, flow meters, current and voltage sensors, pH sensors, and chemical composition concentration sensors) to collect status data of plating solutions on all production lines in real time (including but not limited to temperature, flow rate, current, voltage, pH value, and plating ion concentration);

[0069] Smoothing and standardizing the real-time collected electroplating solution status data on the edge node;

[0070] Specifically, the missing value is calculated and filled by interpolating the data of two adjacent points using linear interpolation;

[0071] Use the sliding window method to reduce the impact of short-term fluctuations by calculating the mean or median within a certain window range around the data point;

[0072] The data were transformed into a standard normal distribution by subtracting the mean and dividing by the standard deviation using the Z-score normalization method.

[0073] S2. Based on the status data of the electroplating solution, use a machine learning algorithm to identify abnormalities in the status data of the electroplating solution.

[0074] Constructing a state data set according to the state data of the electroplating solution;

[0075] According to the range of the status data set for the electroplating solution on the production line, the data that is not within the range of the status data set is labeled as abnormal, and the data that is within the range of the status data set is labeled as normal;

[0076] Mix state data with abnormal labels and normal labels and divide them into training set and test set;

[0077] Based on nonlinear support vector machine (SVM), a plating solution state judgment model is built on the edge node;

[0078] The training set is input into the electroplating liquid state judgment model, and the target loss function is calculated. The expression is as follows:

[0079]

[0080] Among them, L is the target loss function value, n is the number of training samples, i is the index of the number of training samples, α i ,α i+1 are the i-th and i+1-th Lagrange multipliers, y i ,y i+1 is the label of the i-th and i+1-th training samples, normal is 1, abnormal is -1, b is the bias term of the hyperplane, σ is α i and α i+1 The standard deviation of K(x i ,x i+1 ) is the kernel function;

[0081] Divide the training set into k subsets, use k-fold cross validation, and use one of the subsets as the validation set and the rest as the training set for training and validation;

[0082] According to the range setting of the status data of the electroplating solution on the production line, the accuracy, precision and recall thresholds are preset;

[0083] After each training and verification, the accuracy, precision and recall are calculated. When the accuracy, precision and recall are all greater than the accuracy, precision and recall thresholds, the training ends;

[0084] The test set is input into the electroplating liquid state judgment model to calculate the accuracy, precision and recall. When the accuracy, precision and recall are all greater than the accuracy, precision and recall thresholds, the electroplating liquid state judgment model training is completed.

[0085] Integrate the incremental learning mechanism into the electroplating solution state judgment model to update the electroplating solution state judgment model in real time;

[0086] By calculating the distance between the newly collected state data and the hyperplane of the electroplating liquid state judgment model, the influence of the newly collected state data on the electroplating liquid state judgment model is judged. The expression is as follows:

[0087] |w×x new +b|<∈;

[0088] Among them, w is the weight vector of the hyperplane, xnew is the newly collected state data, b is the bias term of the hyperplane, ∈ is the distance threshold between the state data and the hyperplane;

[0089] When the above formula is satisfied, it means that the newly collected state data has a great impact on the current electroplating solution state judgment model, and the electroplating solution state judgment model needs to be updated;

[0090] The newly collected state data is added to the support vector set of the electroplating solution state judgment model, and the hyperplane and decision function are updated according to the new support vector set;

[0091] Specifically, by recalculating the weighted average of the support vectors and updating the hyperplane, the updated support vector set will affect the weight vector and bias term of the SVM, and further affect the classification decision function;

[0092] Regular incremental validation with new datasets ensures the model can make good predictions about new patterns.

[0093] The status data of the plating liquid is collected in real time, and the status of the plating liquid on the production line is judged by the plating liquid status judgment model on the edge node.

[0094] S3. Use visual sensors and infrared sensing equipment to collect visual image data and infrared image data of the electroplating line.

[0095] Use visual sensors to collect visual image data of the electroplating line and capture surface defects of the electroplating line during the electroplating process;

[0096] Use infrared sensing equipment to collect infrared image data of the electroplating line, and accurately measure the thickness and uniformity of the electroplating layer by emitting infrared radiation and receiving reflected signals;

[0097] Processing visual image data and infrared image data;

[0098] Use bilateral filtering to denoise visual image data;

[0099] By using the histogram equalization method, the image contrast is improved by adjusting the brightness distribution of the image, making the details of the electroplating layer clearer;

[0100] The non-local means method is used to perform weighted averaging on similar pixels in the image, remove noise while maintaining details, and adjust the brightness and contrast of the infrared image to make the temperature difference of the electroplating layer more obvious, which is convenient for subsequent thickness measurement.

[0101] S4. Based on the visual image data and infrared image data, the quality of the electroplating line is evaluated through image processing algorithms.

[0102] According to the relationship between the infrared radiation intensity of the infrared sensing device and the thickness of the electroplating layer, the thickness of the electroplating layer is calculated. The expression is as follows:

[0103]

[0104] Wherein, d is the thickness of the electroplating layer, T1 is the reference infrared reflection intensity before electroplating, T2 is the infrared reflection intensity after electroplating, and β is the infrared conductivity coefficient;

[0105] The electroplating layer area of ​​the electroplating line is extracted by image segmentation method, and the standard deviation of the electroplating layer thickness in each area is calculated as the uniformity evaluation index. The smaller the uniformity index, the more uniform the electroplating layer. The expression is as follows:

[0106]

[0107] Among them, γ is the uniformity evaluation index, δ is the thickness standard deviation of each area, and μ is the average thickness of each area;

[0108] Classify and annotate known defects in visual image data (including normal, cracks, bubbles, and peeling);

[0109] Each image is annotated with multiple defect categories and a bounding box is provided for each defect area;

[0110] Based on convolutional neural network (CNN), a defect recognition model is constructed;

[0111] By dividing the visual image data with known defects classified and labeled into a training set and a validation set, and inputting the training set into the defect recognition model;

[0112] The input defect recognition model is trained by calculating the cross entropy loss function of the visual images classified and annotated with known defects in the training set, which is expressed as:

[0113]

[0114] Among them, Loss is the cross entropy loss value, N is the number of training data, m is the index of the number of training data, z m To identify the defect, Identify defects for prediction;

[0115] Use gradient descent method (such as Adam optimizer) to minimize the loss function and iteratively optimize the network weights. After each training, evaluate the performance of the model through the validation set and adjust the hyperparameters (such as learning rate, batch size, etc.);

[0116] Evaluate the performance of the model by calculating the accuracy and recall of the defect recognition model;

[0117] Use the trained defect recognition model to perform defect detection on the newly collected visual image data;

[0118] The overall quality of the electroplating layer is evaluated by comprehensively considering the thickness, uniformity, and surface defects of the electroplating layer. The expression is as follows:

[0119] Q = 0.3d + 0.3γ + 0.4D;

[0120] Among them, Q is the quality assessment value of the electroplating layer, and D is the number of defects.

[0121] S5. According to the abnormal state of the electroplating solution and the quality assessment result of the electroplating line, the state parameters of the electroplating solution are adjusted through an adaptive algorithm.

[0122] Normalize the abnormal data of the plating solution and the quality assessment results of the plating line to ensure that all data are in the same dimension;

[0123] Adopting an adaptive control algorithm based on reinforcement learning, the reward function is used to guide the adjustment of the electroplating solution state data;

[0124] Among them, the reward function expression is:

[0125] R = 0.5ΔQ + 0.5ΔX;

[0126] Where R represents the reward function value, ΔQ is the deviation between the quality evaluation result of the electroplating line and the target quality, and ΔX is the deviation between the state parameter of the electroplating solution and its target value;

[0127] During the reinforcement learning training process, the adjustment process of the electroplating solution state data is optimized by the policy gradient method;

[0128] The control parameters of the plating solution are gradually adjusted through the back propagation algorithm to ensure that the state data change of the plating solution is as close to the target value as possible;

[0129] By adjusting the plating liquid state data and recalculating the reward function according to the current state data and quality evaluation results at each iteration;

[0130] When the reward function reaches the maximum value, it means that the electroplating liquid state data is optimal and the electroplating quality of the electroplating line is optimal.

[0131] This embodiment also provides a computer device, which is suitable for the case of a sixteen-line four-roller electroplating line production control method, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the sixteen-line four-roller electroplating line production control method proposed in the above embodiment.

[0132] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0133] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the production control method of a sixteen-line four-roller electroplating line proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0134] In summary, the present invention improves the reliability of data by: deploying edge nodes on the production line, collecting plating liquid state data in real time and performing smoothing and standardization processing, providing high-quality basic data for subsequent abnormal identification and quality control, and using machine learning algorithms to identify the abnormal state of the plating liquid in real time and perform adaptive adjustments to ensure the stability of the plating liquid. The collection and processing of visual and infrared image data effectively improves the detection accuracy of the thickness, uniformity and defects of the electroplating layer, reduces the limitations of manual inspection, and through comprehensive consideration of the plating liquid state and the plating line quality assessment results, through intelligent production process control, significantly improves the consistency and stability of the electroplating quality, reduces production fluctuations and defective products, improves production efficiency, and ensures high standards of product quality.

[0135] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A sixteen-line four-roller electroplating line production control method, characterized in that: include, Deploy edge nodes on all production lines and use sensors to collect plating solution status data in real time; Based on the status data of the electroplating solution, the machine learning algorithm is used to identify the abnormal status data of the electroplating solution; Use visual sensors and infrared sensing equipment to collect visual image data and infrared image data of the electroplating line; Based on visual image data and infrared image data, the quality of the electroplating line is evaluated through image processing algorithms; According to the abnormal state of the plating solution and the quality assessment results of the plating line, the state parameters of the plating solution are adjusted through an adaptive algorithm.

2. The sixteen-line four-roller electroplating line production control method according to claim 1, characterized in that: The edge nodes are deployed on all production lines, and sensors are used to collect the status data of the electroplating liquid in real time. The specific steps are as follows: Using sensors on edge nodes, real-time data on the status of electroplating liquids on all production lines is collected; The real-time collected electroplating liquid status data is smoothed and standardized on the edge node.

3. The sixteen-line four-roller electroplating line production control method according to claim 2, characterized in that: The method of identifying abnormalities in the state data of the electroplating solution by using a machine learning algorithm according to the state data of the electroplating solution is as follows: Constructing a state data set according to the state data of the electroplating solution; According to the range of the status data set for the electroplating solution on the production line, the data that is not within the range of the status data set is labeled as abnormal, and the data that is within the range of the status data set is labeled as normal; Mix state data with abnormal labels and normal labels and divide them into training set and test set; Based on nonlinear support vector machine (SVM), a plating solution state judgment model is built on the edge node; Input the training set into the electroplating solution state judgment model and calculate the target loss function; Divide the training set into k subsets, use k-fold cross validation, and use one of the subsets as the validation set and the rest as the training set for training and validation; According to the range setting of the status data of the electroplating solution on the production line, the accuracy, precision and recall thresholds are preset; After each training and verification, the accuracy, precision and recall are calculated. When the accuracy, precision and recall are all greater than the accuracy, precision and recall thresholds, the training ends; The test set is input into the electroplating liquid state judgment model to calculate the accuracy, precision and recall. When the accuracy, precision and recall are all greater than the accuracy, precision and recall thresholds, the electroplating liquid state judgment model training is completed. Integrate the incremental learning mechanism into the electroplating solution state judgment model to update the electroplating solution state judgment model in real time; The status data of the plating liquid is collected in real time, and the status of the plating liquid on the production line is judged by the plating liquid status judgment model on the edge node.

4. The sixteen-line four-roller electroplating line production control method according to claim 3, characterized in that: The incremental learning mechanism is incorporated into the electroplating liquid state judgment model to update the electroplating liquid state judgment model in real time. The specific steps are as follows: By calculating the distance between the newly collected state data and the hyperplane of the electroplating solution state judgment model, the influence degree of the newly collected state data on the electroplating solution state judgment model is judged; The newly collected state data are added to the support vector set of the electroplating solution state judgment model, and the hyperplane and decision function are updated according to the new support vector set.

5. The sixteen-line four-roller electroplating line production control method according to claim 1, characterized in that: The visual image data and infrared image data of the electroplating line are collected by using a visual sensor and an infrared sensing device. The specific steps are as follows: Use visual sensors to collect visual image data of the electroplating line; Use infrared sensing equipment to collect infrared image data of the electroplating line; Processing of visual image data and infrared image data.

6. The sixteen-line four-roller electroplating line production control method according to claim 5, characterized in that: The specific steps of processing the visual image data and the infrared image data are as follows: Use bilateral filtering to denoise visual image data; The histogram equalization method is used to improve the contrast of the image by adjusting the brightness distribution of the image; The non-local means method is used to perform weighted averaging of similar pixels in the image, remove noise while maintaining details, and adjust the brightness and contrast of the infrared image.

7. The sixteen-line four-roller electroplating line production control method according to claim 6, characterized in that: Based on the visual image data and infrared image data, the quality of the electroplating line is evaluated by image processing algorithm. The specific steps are as follows: Calculate the thickness of the electroplating layer based on the relationship between the infrared radiation intensity of the infrared sensing device and the thickness of the electroplating layer; The electroplating layer area of ​​the electroplating line is extracted by image segmentation method, and the standard deviation of the electroplating layer thickness in each area is calculated as a uniformity evaluation index; Classify and annotate known defects in visual image data (including normal, cracks, bubbles, and peeling); Each image is annotated with multiple defect categories and a bounding box is provided for each defect area; Based on convolutional neural network (CNN), a defect recognition model is constructed; By dividing the visual image data with known defects classified and labeled into a training set and a validation set, and inputting the training set into the defect recognition model; The input defect recognition model is trained by calculating the cross entropy loss function of visual images classified and annotated with known defects in the training set; Use the gradient descent method to minimize the loss function and iteratively optimize the network weights. After each training, evaluate the performance of the model through the validation set and adjust the hyperparameters. Evaluate the performance of the model by calculating the accuracy and recall of the defect recognition model; Use the trained defect recognition model to perform defect detection on the newly collected visual image data; The overall quality of the electroplating layer is evaluated by comprehensively considering the thickness, uniformity, and surface defects of the electroplating layer.

8. The production control method of a sixteen-line four-roller electroplating line according to claim 7, characterized in that: According to the abnormal state of the electroplating solution and the quality evaluation result of the electroplating line, the state parameters of the electroplating solution are adjusted by an adaptive algorithm. The specific steps are as follows: Normalize the abnormal data of the plating solution and the quality assessment results of the plating line; Adopting an adaptive control algorithm based on reinforcement learning, the reward function guides the adjustment of the electroplating solution state data; During the reinforcement learning training process, the adjustment process of the electroplating solution state data is optimized by the policy gradient method; The control parameters of the plating solution are gradually adjusted through the back-propagation algorithm; By adjusting the plating liquid state data and recalculating the reward function according to the current state data and quality evaluation results at each iteration; When the reward function reaches the maximum value, it means that the electroplating liquid state data is optimal and the electroplating quality of the electroplating line is optimal.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the sixteen-line four-roller electroplating line production control method described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the sixteen-line four-roller electroplating line production control method described in any one of claims 1 to 8 are implemented.

Images