Wire stoving varnish control method and system

Abstract

The invention provides a wire stoving varnish control method and system, and relates to the technical field of control, and the key points of the technical scheme are that basic parameters of a wire to be stoved with varnish are obtained, and a database is called for matching to obtain stoving varnish process parameters; the basic parameters comprise a wire material, a wire diameter, a paint film preset thickness and a paint type, and the baking varnish process parameters comprise the temperature of each partition of a baking varnish furnace, a hot air circulation speed, a wire conveying speed and a baking varnish duration; and according to the baking finish process parameters, wire baking finish is implemented, and dynamic parameters in the baking finish process are collected in real time. The wire baking varnish control method and system provided by the invention have the advantages that the initial process is accurately adapted to reduce quality fluctuation, the process regulation and control efficiency is improved through graded risk management and control and graded correction, and the quality integrity of a paint film is guaranteed by means of staged cooling and stress regulation and control.

Description

Technical Field

[0001] This application relates to the field of control technology, and more specifically, to a method and system for controlling wire baking paint. Background Technology

[0002] The processing of enameled wire involves several steps: drawing thick wire to thinner thickness, applying enameled paint to the thinner wire, baking the paint, applying a coating solution, and winding up the wire. Baking the paint is a crucial step in the production of wires, cables, and metal components, as the adhesion, thickness uniformity, and corrosion resistance of the paint film directly determine the wire's performance and lifespan. Currently, the industry's methods for controlling wire baking have the following shortcomings: 1. Crude matching of process parameters: The selection of baking paint process parameters relies heavily on manual experience and fails to accurately match them with basic parameters such as wire material and diameter. This results in a mismatch between the process and the characteristics of the wire, which can easily lead to fluctuations in paint film quality. 2. Lack of targeted process modification: Parameter adjustment often adopts a single threshold strategy, without quantifying the impact of dynamic parameters on quality deviations, which easily leads to over-correction or under-correction. 3. Inadequate cooling process control: The use of a single cooling medium and a fixed cooling rate can easily lead to stress concentration and cracking of the paint film due to improper cooling, reducing adhesion and integrity.

[0003] There is currently no effective technical solution to the above problems. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for controlling wire coating, which has the advantages of accurately adapting to the initial process to reduce quality fluctuations, improving process control efficiency through graded risk management and graded correction, and ensuring the integrity of the coating quality by relying on staged cooling and stress control.

[0005] In a first aspect, this application provides a method for controlling the coating process of wires, the method comprising the following steps: The basic parameters of the wire to be painted are obtained, and the painting process parameters are obtained by matching the database. The basic parameters include wire material, wire diameter, preset paint film thickness and paint type. The painting process parameters include temperature of each zone of the painting oven, hot air circulation speed, wire conveying speed and painting time. The wire is baked with paint according to the baking process parameters, and dynamic parameters during the baking process are collected in real time. The dynamic parameters include the real-time temperature of each zone in the baking oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the curing progress of the paint film, and the real-time speed of the wire conveying. The basic and dynamic parameters are input into the paint quality prediction model, and the prediction results of the paint quality are output; the prediction results include the adhesion level prediction value, the thickness uniformity prediction value, and the corrosion resistance prediction value. The deviation between the predicted results and the preset quality standards is analyzed, and the baking paint process parameters are dynamically corrected through an improved fuzzy PID algorithm; the preset quality standards include adhesion level standard values, thickness uniformity standard values, and corrosion resistance standard values; The surface cooling temperature of the cured wire is collected in real time, and the cooling parameters are dynamically adjusted according to a preset cooling curve; the cooling parameters include the flow rate and temperature of the cooling medium. The system detects the pass rate of the paint quality of the finished product after cooling. When the pass rate is 100%, the system associates and stores the data and updates the database. When the pass rate is <100%, the system generates process optimization suggestions and iterates the paint quality prediction model.

[0006] Furthermore, in this application, after the step of obtaining the basic parameters of the wire to be coated and before the step of performing the wire coating according to the coating process parameters, the following is also included: Detect the uniformity of the paint coating on the surface of the wire before it enters the oven and determine whether it reaches the preset uniformity threshold; When the paint uniformity does not reach the preset uniformity threshold, a paint correction signal is generated and fed back to the previous paint process, and the paint feeding is suspended. When the paint uniformity reaches the preset uniformity threshold, the subsequent baking process is started.

[0007] Furthermore, in this application, the steps of inputting basic parameters and dynamic parameters into the paint quality prediction model and outputting the paint quality prediction result also include: The detection predicts whether the prediction result exceeds the prediction anomaly warning threshold of the paint quality prediction model. The prediction anomaly warning threshold is set according to the basic parameters and dynamic parameters. When the deviation between five consecutive sets of predicted results and actual sampling results exceeds a preset threshold, the paint quality prediction model retraining process is automatically triggered, and an early warning signal is simultaneously output to prompt staff to check the data collection link.

[0008] Furthermore, in this application, the step of automatically triggering the retraining process of the paint quality prediction model when the deviation between five consecutive sets of predicted results and actual sampling results exceeds a preset threshold also includes: Historical basic parameters and historical dynamic parameters are collected as input samples, and corresponding finished paint quality inspection results are collected as label samples. A CNN-LSTM hybrid neural network is used to train the paint quality prediction model. During training, cross-validation was used to optimize the parameters of the paint quality prediction model to ensure its accuracy.

[0009] Furthermore, in this application, the step of analyzing the deviation between the predicted results and the preset quality standards, and dynamically correcting the baking paint process parameters using an improved fuzzy PID algorithm, also includes: S401. Based on the prediction results and dynamic parameter analysis, calculate the influence factor of each dynamic parameter on the deviation of different quality indicators. The formula is: ; Where F i,j Let ΔQ be the influence factor of the j-th dynamic parameter on the deviation of the i-th quality index; j Let D be the predicted deviation value of the i-th quality indicator, that is, the deviation between the predicted result and the preset quality standard; j The j-th dynamic parameter is the real-time acquired value; the quality indicators are adhesion level, thickness uniformity, and corrosion resistance. S402. Establish a risk grading standard for paint baking quality based on influencing factors: "Single quality indicator deviation value ≤ 1% and no high influencing factors (F)" i,j "Dynamic parameters ≥0.6" are classified as low risk; "Single quality indicator deviation value 1%-3% or the existence of one high-impact factor dynamic parameter" are classified as medium risk; "Any quality indicator deviation value >3% or the existence of two or more high-impact factor dynamic parameters" are classified as high risk. S403. Output graded response instructions according to the paint quality risk grading standard: For low risk, output regular correction instructions and a simplified analysis report including only the deviation value; for medium risk, output enhanced correction instructions and a complete analysis report including the deviation value and the ranking of influencing factors; for high risk, output emergency correction instructions, a complete analysis report and suggestions for investigating abnormal collection of dynamic parameters of high influencing factors, suspend material feeding and review.

[0010] Furthermore, in this application, the step of outputting a graded response instruction based on the paint quality risk grading standard also includes: Based on the graded response instructions and the corresponding influencing factors, the deviation and deviation change rate between the corresponding prediction results and the preset quality standards are used as the core input of the fuzzy controller. At the same time, the correction intensity is adapted in combination with the graded response instructions: the normal correction step size is used for the regular correction instructions, the correction step size is increased by 1.2-1.5 times for the enhanced correction instructions, and the maximum safe step size is used for the emergency correction instructions. By using fuzzy rule reasoning to obtain the correction values ​​of the proportional coefficient, integral coefficient, and derivative coefficient of the PID controller, the initial PID parameters are dynamically adjusted to achieve precise correction of the baking paint process parameters.

[0011] Furthermore, in this application, the step of real-time acquisition of the surface cooling temperature of the cured wire and dynamic adjustment of the cooling parameters according to a preset cooling curve also includes: Based on the basic parameters of the wire to be painted, a graded adaptation rule for the preset cooling curve is set, and a preset cooling curve is constructed according to the graded adaptation rule: in the initial stage of cooling, a rapid cooling mode is adopted, and the cooling rate is controlled at 5-8℃ / s; a switching threshold between the rapid cooling mode and the slow cooling mode is set, and when the surface temperature of the wire drops to 150-180℃, it automatically switches to the slow cooling mode, and the cooling rate is controlled at 1-2℃ / s. Set cooling stop criteria and collect wire surface temperature in real time: when the wire surface temperature drops to room temperature ±5℃, stop cooling and record all parameters of the complete cooling process.

[0012] Furthermore, in this application, after the step of setting a graded adaptation rule for the preset cooling curve based on the basic parameters of the wire to be coated, and before the step of setting the cooling stop judgment condition and collecting the wire surface temperature in real time, the following steps are also included: Based on the basic parameters of the wire to be painted, two cooling media adapted to different cooling modes are preset, including air cooling medium adapted to rapid cooling mode and inert gas cooling medium adapted to slow cooling mode. At the same time, the suitable temperature range and switching judgment conditions of each cooling medium are preset. The surface temperature of the wire during the cooling process is acquired in real time and compared with the suitable temperature range and switching conditions of each cooling medium. The inlet and outlet temperatures of the cooling medium, the temperature and humidity of the cooling environment and the real-time speed of the wire are acquired simultaneously to form a multi-dimensional dynamic parameter set for the cooling process. When the surface temperature of the wire drops from the rapid cooling mode to the threshold of the slow cooling mode, the cooling medium switching process is automatically triggered, and the cooling medium purging and replacement program is started simultaneously to ensure that there is no residual air-cooled medium in the cooling channel. After the replacement is completed, the inert gas cooling medium supply is turned on. After the cooling medium is switched, based on the multi-dimensional dynamic parameter set of the cooling process, the supply pressure, temperature and flow rate of the inert gas cooling medium are adjusted synchronously through the improved PID collaborative control algorithm, so that the cooling rate of the wire is kept stable within the target range, and the surface stress value of the paint film is monitored in real time. When the stress value exceeds the preset safety threshold, the temperature of the inert gas cooling medium is dynamically adjusted by ±2℃ until the stress value is lower than the preset safety threshold.

[0013] Furthermore, in this application, when the pass rate is <100%, the steps of generating process optimization suggestions and iterating the paint baking quality prediction model also include: Obtain the specific parameters of the defective products, analyze the factors causing the defects, and generate the adjustment direction and adjustment range for the specific parameters; When the non-compliance factor is insufficient paint film adhesion, the proposed process optimization is to increase the temperature of the curing zone of the paint oven by 5-10℃ and extend the curing time by 10-20 seconds. When the non-compliance factor is uneven paint film thickness, the proposed process optimization is to adjust the hot air circulation speed by ±0.5m / s and correct the wire conveying speed by ±0.2m / min.

[0014] Secondly, this application also provides a wire coating control system, the system comprising: The first acquisition module is used to acquire the basic parameters of the wire to be painted and call the database to match and obtain the painting process parameters. The basic parameters include wire material, wire diameter, preset paint film thickness and paint type. The painting process parameters include the temperature of each zone of the painting oven, hot air circulation speed, wire conveying speed and painting time. The first control module is used to perform wire baking paint according to the baking paint process parameters and to collect dynamic parameters in real time during the baking paint process; the dynamic parameters include the real-time temperature of each zone in the baking paint oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the paint film curing progress, and the real-time speed of wire conveying. The second control module is used to input basic parameters and dynamic parameters into the paint quality prediction model and output the paint quality prediction results; the prediction results include adhesion level prediction value, thickness uniformity prediction value and corrosion resistance prediction value; The third control module is used to analyze the deviation between the predicted results and the preset quality standards, and dynamically correct the baking paint process parameters through an improved fuzzy PID algorithm; the preset quality standards include adhesion level standard values, thickness uniformity standard values, and corrosion resistance standard values. The fourth control module is used to collect the surface cooling temperature of the cured wire in real time and dynamically adjust the cooling parameters according to the preset cooling curve; the cooling parameters include the flow rate and temperature of the cooling medium. The fifth control module is used to detect the pass rate of the paint quality of the finished product after cooling. When the pass rate is 100%, it associates and stores data and updates the database; when the pass rate is <100%, it generates process optimization suggestions and iterates the paint quality prediction model.

[0015] Other features and advantages of this application will be set forth in the following description and will be apparent in part from the description or may be learned by practicing embodiments of this application. The objectives and other advantages of this application may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0016] Beneficial effects: 1) Precisely adapt to the initial process to reduce quality fluctuations: By calling the database to match the baking paint process parameters based on basic parameters, the manual experience matching is replaced. This ensures that the baking paint process parameters (oven temperature, conveyor speed, etc.) are precisely matched with the wire characteristics (material, diameter, etc.), thereby reducing quality fluctuations caused by improper initial processes from the source.

[0017] 2) Pre-inspection and process monitoring to reduce waste in ineffective processes: By conducting pre-inspection of coating uniformity, unqualified wires can be intercepted in advance, preventing them from entering the baking process and causing resource waste; by collecting dynamic parameters in real time and predicting abnormalities, the quality of the baking process can be monitored in real time, risks can be identified in advance, and batches of unqualified products can be avoided.

[0018] 3) Improve the accuracy of quality prediction and realize model self-iteration: Adopt a CNN-LSTM hybrid neural network, extract the spatial correlation and temporal change features of parameters, and optimize the parameters by combining cross-validation to improve the accuracy of the quality prediction model; the model retraining operation is triggered by 5 consecutive sets of deviations to realize the automatic iterative update of the model and maintain the prediction accuracy in the long term without manual intervention.

[0019] 4) Graded risk control and precise correction to improve process control efficiency: By calculating influencing factors, the correlation between dynamic parameters and quality deviations is quantified, and the quality risk is accurately determined by combining risk grading standards; by using fuzzy PID graded correction, the correction intensity is adapted according to the risk level, and the PID parameters are dynamically adjusted to avoid process fluctuations caused by over-correction and ensure that quality deviations converge quickly.

[0020] 5) Staged cooling and stress control to ensure the integrity of the coating quality: Through staged cooling curves (rapid cooling and slow cooling) and dual cooling medium adaptation, the cooling requirements of different temperature ranges of the wire are matched; through PID collaborative control and stress fine-tuning, the cooling rate is stabilized and stress concentration of the coating is avoided, reducing defects such as coating cracking and insufficient adhesion.

[0021] 6) Closed-loop data iteration to continuously optimize production quality: After the pass rate reaches the target, the database is updated to accumulate high-quality correlation data of wire basic parameters, process parameters, and qualified quality, thereby improving the accuracy of subsequent process matching; quantitative optimization suggestions are generated for different non-conforming factors (insufficient adhesion, uneven thickness) to achieve targeted process improvement, forming a closed loop of collection, prediction, correction, and iteration, thereby improving production quality and efficiency in a long-term development manner. Attached Figure Description

[0022] Figure 1 A flowchart of the wire coating control method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the wire coating control system provided in an embodiment of this application.

[0023] Labeling explanation: 201, First acquisition module; 202, First control module; 203, Second control module; 204, Third control module; 205, Fourth control module; 206, Fifth control module. Detailed Implementation

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0025] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0026] The following discloses and provides many different implementation methods or examples to achieve the purpose of the present invention and to solve the problems existing in the prior art.

[0027] Please refer to Figure 1 As shown in the figure, this application provides a method for controlling wire coating, which includes the following steps: S1. Obtain the basic parameters of the wire to be painted, and call the database to match and obtain the painting process parameters; the basic parameters include wire material, wire diameter, preset paint film thickness and paint type, and the painting process parameters include the temperature of each zone of the painting oven, hot air circulation speed, wire conveying speed and painting time. S2. Implement wire baking paint according to the baking paint process parameters, and collect dynamic parameters in real time during the baking paint process; the dynamic parameters include the real-time temperature of each zone in the baking paint oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the curing progress of the paint film, and the real-time speed of the wire conveying. S3. Input the basic parameters and dynamic parameters into the paint quality prediction model and output the paint quality prediction results; the prediction results include the adhesion level prediction value, the thickness uniformity prediction value, and the corrosion resistance prediction value. S4. Analyze the deviation between the predicted results and the preset quality standards, and dynamically correct the baking paint process parameters through an improved fuzzy PID algorithm; the preset quality standards include adhesion level standard values, thickness uniformity standard values, and corrosion resistance standard values. S5. Real-time acquisition of the surface cooling temperature of the cured wire, and dynamic adjustment of cooling parameters according to a preset cooling curve; the cooling parameters include the flow rate and temperature of the cooling medium; S6. Detect the paint quality pass rate of the finished product after cooling. When the pass rate is 100%, link and store the data and update the database. When the pass rate is <100%, generate process optimization suggestions and iterate the paint quality prediction model.

[0028] In step S1, obtaining the basic parameters of the wire to be coated and matching them with the coating process parameters from the database serves the following purpose: Based on the inherent characteristics (basic parameters) of the wire, retrieving suitable process parameters from the database is a preparatory step before the coating process, ensuring that the process parameters match the wire characteristics and avoiding quality fluctuations caused by generic parameters. The basic parameters include wire material, wire diameter, preset coating thickness, and coating type. The coating process parameters include the temperature of each zone of the coating oven, hot air circulation speed, wire conveying speed, and coating duration. Listing the basic parameters clarifies the specific content of the basic parameters and coating process parameters. Listing the basic parameters defines the core characteristic indicators of the wire itself, ensuring the targeted matching of subsequent process parameters; listing the coating process parameters defines the core operational indicators that need to be controlled during the coating process, giving the retrieval and adjustment of the coating process parameters a clear implementation target.

[0029] Specifically, the following steps are included after step S1 and before step S2: S101. Detect the uniformity of the coating on the surface of the wire before it enters the baking oven and determine whether it reaches the preset uniformity threshold: Before the baking process starts, the core quality indicator of the wire's pre-coating (coating uniformity) is detected and qualified. This is a pre-quality inspection step before baking to prevent wires with substandard coating uniformity from entering the baking process, thereby reducing the risk of quality defects such as uneven paint film thickness and insufficient adhesion after subsequent baking.

[0030] S102. When the coating uniformity fails to reach the preset uniformity threshold, a coating correction signal is generated and fed back to the preceding coating process. Simultaneously, the baking paint feed is suspended. For cases of pre-inspection failure, inter-process anomaly linkage is initiated. Generating the coating correction signal and feeding it back to the preceding process guides the preceding coating process to adjust process parameters, addressing the coating uniformity defect at its root. Suspending the baking paint feed prevents substandard wire from entering the baking paint process, reducing resource waste from ineffective baking paint operations and enabling timely interception of quality anomalies.

[0031] S103. When the coating uniformity reaches the preset uniformity threshold, the subsequent baking process is started: As the qualified release exit of the pre-inspection stage, only wires with coating uniformity that meet the standard are allowed to enter the baking process, ensuring that the quality of the input materials for the baking process meets the requirements, and providing a prerequisite guarantee for the stable execution of the subsequent baking process and the final quality standard.

[0032] In step S2, wire baking is performed according to the baking process parameters, and dynamic parameters during the baking process are collected in real time. Its purpose is to connect the process parameters from step S1, execute the baking operation, and simultaneously collect process data. This serves as both the execution link of the baking process and the data source support link for subsequent quality prediction and parameter correction, achieving synchronization between process execution and process monitoring. The dynamic parameters include the real-time temperature of each zone in the baking oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the curing progress of the paint film, and the real-time speed of the wire conveyor. These parameters are used to identify the core process indicators that need to be monitored during the baking process, ensuring the comprehensiveness and relevance of data collection, and providing complete real-time input data for the subsequent baking quality prediction model.

[0033] In step S3, basic and dynamic parameters are input into the paint quality prediction model, which outputs a prediction result for the paint quality. Its purpose is to predict the paint quality in advance through analysis of the paint process data, thereby achieving quality control during the paint process. This avoids discovering problems only after the product is finished and tested, reducing ineffective processes and lowering costs. The prediction results include predicted values ​​for adhesion level, thickness uniformity, and corrosion resistance, which are used to clarify the core evaluation dimensions of paint quality, providing specific and quantifiable quality indicators to avoid ambiguity in the prediction.

[0034] Specifically, step S3 also includes the following sub-steps: S301. Detect whether the predicted result exceeds the predicted anomaly warning threshold of the paint baking quality prediction model. The predicted anomaly warning threshold is set based on basic parameters and dynamic parameters. After the paint baking quality prediction model outputs the predicted result, an anomaly monitoring step is added. The validity of the predicted result is verified through a standardized predicted anomaly warning threshold. Its core function is to identify predicted results exceeding a reasonable range (abnormal prediction) in advance, avoiding the execution of subsequent process parameter corrections in step S4 based on erroneous predicted results, and reducing paint baking quality fluctuations caused by ineffective corrections. The purpose of setting the predicted anomaly warning threshold is to clarify that the "predicted anomaly warning threshold" is not a conventional fixed value, but is dynamically set in combination with the basic parameters of the wire to be painted (material, diameter, etc.) and the dynamic parameters of the painting process (furnace temperature, humidity, etc.), ensuring that the predicted anomaly warning threshold adapts to different working conditions and improves the targeting and accuracy of anomaly monitoring.

[0035] S302. When the deviation between five consecutive sets of predicted results and actual sampling results exceeds a preset threshold, the paint quality prediction model retraining process is automatically triggered, and an early warning signal is simultaneously output to prompt staff to check the data acquisition link. This establishes a closed-loop monitoring mechanism for paint quality prediction model accuracy calibration and anomaly tracing, ensuring the long-term stability and reliability of the paint quality prediction model. The specific functions of step S302 are as follows: Triggering paint quality prediction model retraining: Using five consecutive sets of deviations exceeding the preset threshold as a quantitative judgment condition (avoiding accidental triggering by random deviations), the paint quality prediction model retraining is automatically started. By updating the paint quality prediction model parameters, the prediction accuracy decay problem is corrected, ensuring that the paint quality prediction model adapts to the dynamic changes in process conditions. Early warning and tracing guidance: An early warning signal is simultaneously output and a clear prompt is made to check the data acquisition link, accurately pointing to the core potential cause of the deviation exceeding the standard (such as abnormal sensor acquisition, data transmission failure, etc.), guiding staff to quickly troubleshoot the problem, ensuring the reliability of subsequent model training data from the data source, and simultaneously realizing human-machine collaborative handling of abnormal situations.

[0036] Specifically, step S302 further includes the following sub-steps: S3021. Collect historical basic parameters and historical dynamic parameters as input samples, and collect corresponding finished paint quality inspection results as label samples. Use a CNN-LSTM hybrid neural network to train the paint quality prediction model: Clarifying the sample source and core network architecture for model retraining and solving the technical problem of "how to train the model" is the core implementation step of the paint quality prediction model retraining. At the sample design level, historical basic parameters and historical dynamic parameters are used as input samples, corresponding to the initial conditions and process states of paint baking; finished paint quality inspection results are used as label samples to achieve accurate correspondence between input and output and ensure the effectiveness of training data. At the network architecture level, a CNN-LSTM hybrid neural network is used. CNN extracts the spatial correlation features of multi-dimensional parameters, and LSTM captures the temporal change features of the paint baking process, taking into account information mining in both spatial and temporal dimensions. Compared with a single neural network, this significantly improves the model's prediction accuracy of paint quality. At the technical value level, this provides a specific technical implementation path for triggering the paint quality prediction model retraining in step S302, transforming the paint quality prediction model retraining from a conceptual requirement into a feasible operational process.

[0037] More specifically, step S3021 achieves the retraining and accuracy assurance of the paint quality prediction model through the process of "sample construction → CNN-LSTM hybrid network training → cross-validation parameter optimization". The specific process is as follows: I. Implementation logic of step S3021 (CNN-LSTM hybrid neural network training).

[0038] Core objective: To train a model that can accurately predict paint quality using historical data. The steps involved are: "sample preparation → spatial feature extraction (CNN) → temporal feature modeling (LSTM)".

[0039] 1. Sample preparation.

[0040] Input samples: Collect time-series data of historical basic parameters (wire material, diameter, etc.) and historical dynamic parameters (real-time temperature and humidity of the paint oven, etc.) to form the input sample matrix X; Label samples: Collect the finished paint quality test results (adhesion level, thickness uniformity, etc.) of the corresponding batches and use them as labels for model training (matching the model output results).

[0041] 2. Spatial feature extraction (CNN module).

[0042] The spatial feature extraction formula for CNN convolutional layers is as follows: This formula extracts spatial correlation features of multi-dimensional parameters from input samples (such as the coordinated variation features of "paint oven temperature-humidity"). O c =σ(W c X+b c ); Among them O c σ represents the spatial feature sequence output by the CNN convolutional layer (used as input to the LSTM module), σ represents the ReLU activation function (adapted for feature extraction from industrial time-series data, enhancing the non-linear expressive power of features), and W represents the spatial feature sequence output by the CNN convolutional layer (used as input to the LSTM module). c Represents the convolution kernel weight matrix. This represents a binary convolution operation (to extract spatial features), where X represents the temporal data matrix of the input samples (composed of a temporal sequence of historical basic parameters and historical dynamic parameters), and b. c Indicates the convolutional layer bias term; 3. Temporal feature modeling (LSTM module).

[0043] By capturing the temporal correlation features of the paint baking process using LSTM (such as the temporal dimension correlation of "temperature change - paint film curing progress"), LSTM achieves the memory and update of temporal features through forget gate, input gate, and output gate.

[0044] The forget gate formula is used to control the retention of historical states, and the formula is as follows: f t =σ(W f ·[h t-1 O c,t ]+b f ); The input gate formula is used to control the input of new features, and the formula is as follows: it =σ(W i ·[h t-1 O c,t ]+b i ); The candidate cell state formula is used to generate new state features. The formula is as follows: ; The cell state update formula is used to fuse historical and new features. The formula is as follows: ; The output gate formula is used to generate the model output at the current time step, and the formula is: o t =σ(W o ·[h t-1 O c,t ]+b o ); The final output formula of LTSM is used to output the prediction result of paint quality. The formula is: h t =o t ⊙tanh(C t ); Where f t i t o t These represent the outputs of the forget gate, input gate, and output gate at time t, respectively; h t-1 Indicates the hidden state of the LSTM layer at time t-1; O c,t W represents the spatial features of the CNN output at time t. f W i W C W o These represent the weight matrices corresponding to the gating; b f b i b C b o These represent the bias terms of the corresponding gating; Represents the candidate cell state at time t; ⊙ represents the element-wise product of matrices; C t Indicates the cell state at time t; h t This represents the model output at time t (i.e., the paint quality prediction feature, matched with the label sample).

[0045] S3022. During training, the parameters of the paint quality prediction model are optimized using cross-validation to ensure its prediction accuracy. This involves clarifying the parameter optimization methods and objectives for training the paint quality prediction model, ensuring that the model's performance meets standards after retraining. Specifically, at the optimization level, cross-validation is introduced. This involves iterative training by dividing the sample set into training and validation subsets to avoid overfitting and improve the model's generalization ability to new working conditions. This method is also used to fine-tune the core parameters of the CNN-LSTM hybrid neural network (such as CNN kernel size and LSTM hidden layer dimensions). At the objective level, ensuring prediction accuracy is the core goal. Parameter optimization is directly linked to the performance of the paint quality prediction model, ensuring that the retrained model can stably output reliable quality prediction results, providing accurate data support for subsequent steps S4, including bias analysis and parameter correction. More specifically, the purpose of step S3022 is to avoid model overfitting, improve the model's generalization ability to new data, and ensure prediction accuracy. The steps are "sample splitting → multi-fold training → accuracy evaluation → parameter optimization," and the specific process is as follows: 1. Sample division.

[0046] Divide the input samples and label samples into K mutually exclusive subsets (e.g., 50% or 100% subsets, depending on the sample size).

[0047] 2. Accuracy calculation.

[0048] Model performance is evaluated using single-fold accuracy and average accuracy. The formula for single-fold cross-validation prediction accuracy is: ; The formula for the average accuracy of K-fold cross-validation is: ; Among them Acc k N represents the prediction accuracy on the k-th fold validation set; correct,k N represents the number of samples in the k-th validation set where the model's predicted results are consistent with the actual finished product quality inspection results; total,k This represents the total number of samples in the k-th fold validation set; This represents the average accuracy of K-fold cross-validation (the core objective of model parameter optimization).

[0049] 3. Parameter optimization.

[0050] With the goal of maximizing accuracy, we adjusted model parameters such as the convolution kernel size of the CNN and the hidden layer dimension of the LSTM to finally obtain a paint quality prediction model with satisfactory accuracy.

[0051] In step S4, the deviation between the predicted result and the preset quality standard is analyzed, and the baking paint process parameters are dynamically corrected using an improved fuzzy PID algorithm. Its purpose is to precisely adjust the process parameters based on the deviation between the predicted result and the preset quality standard, ensuring that the subsequent baking paint process meets quality standards through correction using the improved fuzzy PID algorithm. The preset quality standards include adhesion level standard values, thickness uniformity standard values, and corrosion resistance standard values, which are used to clearly define the benchmark for quality compliance, providing a clear reference for the deviation analysis process and ensuring the targeted nature of the correction of the baking paint process parameters.

[0052] Specifically, step S4 also includes the following sub-steps: S401. Based on the prediction results and dynamic parameter analysis, calculate the influence factor of each dynamic parameter on the deviation of different quality indicators. The formula is: ; Where F i,j Let ΔQ be the influence factor of the j-th dynamic parameter on the deviation of the i-th quality index; j Let D be the predicted deviation value of the i-th quality indicator, that is, the deviation between the predicted result and the preset quality standard; j The j-th dynamic parameter is the real-time acquired value; the quality indicators are adhesion level, thickness uniformity, and corrosion resistance. It should be noted that by introducing quantitative analysis methods and using partial derivative formulas to accurately calculate the correlation (i.e., the influencing factor) between dynamic parameters and quality indicator deviations, the deviation analysis in step S4 is upgraded from qualitative judgment to quantitative calculation, providing objective and quantifiable data support for subsequent risk grading. By standardizing the technical terms corresponding to each symbol in the formula, the specific meanings of the influencing factor, predicted deviation value, and dynamic parameter are clarified, while the specific dimensions of the "quality indicator" are further defined, ensuring the understandability and feasibility of this quantitative analysis step.

[0053] S402. Establish a risk grading standard for paint baking quality based on influencing factors: "Single quality indicator deviation value ≤ 1% and no high influencing factors (F)" i,j "Dynamic parameters with a deviation of ≥0.6" are classified as low risk; "a single quality indicator deviation of 1%-3% or the presence of one high-impact factor dynamic parameter" is classified as medium risk; "any quality indicator deviation >3% or the presence of two or more high-impact factor dynamic parameters" is classified as high risk. Based on the quantitative impact factors in step S401, standardized risk classification rules are established, combining the degree of quality deviation with the number of key influencing factors to achieve accurate and measurable classification of paint quality risks, avoiding subjectivity in risk assessment, and providing a clear basis for subsequent differentiated handling.

[0054] S403. Based on the paint quality risk grading standard, output graded response instructions: For low risk, output routine correction instructions and a simplified analysis report including only deviation values; for medium risk, output enhanced correction instructions and a complete analysis report including deviation values ​​and impact factor ranking; for high risk, output emergency correction instructions, a complete analysis report, and suggestions for investigating anomalies in the collection of dynamic parameters of high impact factors, suspend material intake and review: Achieve precise matching between risk grading and handling strategies, outputting differentiated response instructions for different risk levels: For low risk, output routine correction instructions and a simplified analysis report including only deviation values, balancing efficiency and quality; for medium risk, output enhanced correction instructions and a complete analysis report including deviation values ​​and impact factor ranking, accurately locating impact factors; for high risk, output emergency correction instructions, a complete analysis report, and suggestions for investigating anomalies in the collection of dynamic parameters of high impact factors, suspend material intake and review, maximizing the reduction of batch quality defect risks; at the same time, achieve information transparency through analysis reports, and achieve a closed loop of traceability from "deviation to data to process" through suggestions for investigating anomalies, ensuring the stability of subsequent processes.

[0055] Specifically, step S403 further includes the following sub-steps: S4031. Based on the graded response instructions and corresponding influencing factors, the deviation and rate of change of the corresponding predicted results from the preset quality standards are used as the core inputs of the fuzzy controller. Simultaneously, the correction intensity is adapted in conjunction with the graded response instructions: a standard correction step size is used for regular correction instructions, the correction step size is increased by 1.2-1.5 times for enhanced correction instructions, and preliminary correction is performed using the maximum safe step size for emergency correction instructions. This achieves the connection between the graded response instructions and the control logic, transforming the risk graded instructions of step S403 into specific inputs and correction strategies for the fuzzy controller. Specifically, at the input correlation level: the predicted deviation and rate of change of deviation serve as the core inputs of the fuzzy controller, combined with influencing factors and graded response instructions, ensuring that the control input simultaneously covers the degree of deviation, the trend of deviation, and key influencing factors, guaranteeing the comprehensiveness of the input information. At the correction intensity adaptation level, differentiated correction step sizes (regular / enhanced / emergency) are matched for graded response instructions corresponding to low / medium / high risks, avoiding a "one-size-fits-all" correction approach. This prevents excessive correction in low-risk scenarios from causing process fluctuations and ensures that the correction intensity is sufficient to quickly suppress deviations in high-risk scenarios, improving the accuracy and adaptability of the correction.

[0056] More specifically, the core action of step S4031 is to determine the input variables of the fuzzy controller and match the correction intensity corresponding to the grading instruction. The specific implementation logic is as follows: 1. Construct the core inputs of the fuzzy controller.

[0057] The "deviation between the predicted result and the preset quality standard" and the "rate of change of deviation" are used as the core inputs of the fuzzy controller. Deviation (e): the predicted deviation value △Q of the i-th quality indicator in steps S401~S403. i The formula is: e=△Q i =Predicted value of quality indicator - Preset standard value of quality indicator; Deviation change rate (ec): The rate of change of deviation over time, reflecting the trend of deviation development. The formula is: ; Where △t represents the time interval between two consecutive predictions, △(△Q) i ) indicates the change in the deviation between two consecutive predictions.

[0058] 2. Adaptive correction strength.

[0059] Based on the graded response instructions in steps S401~S403, match the corresponding process parameter correction step size: When the instruction is "Standard Correction" (low risk): Use a standard correction step size Δu normal ; When the instruction is "Enhanced Correction" (medium risk): the correction step size is adjusted to 1.2~1.5△u. normal ; When the instruction is "urgent correction" (high risk): first use the maximum safe step size △u normal Make initial corrections (to prevent the deviation from expanding rapidly).

[0060] S4032. Obtain the correction amounts of the proportional, integral, and derivative coefficients of the PID controller through fuzzy rule reasoning, and dynamically adjust the initial PID parameters to achieve precise correction of the baking paint process parameters: This completes the core execution step of the improved fuzzy PID algorithm, transforming the previous input and correction strategies into actual process parameter adjustments. Specifically, at the algorithm execution level, through fuzzy rule reasoning, information such as deviation, deviation change rate, and correction intensity are transformed into correction amounts of the proportional (P), integral (I), and derivative (D) coefficients of the PID controller, achieving dynamic optimization of the initial PID parameters. At the technical value level, the abstract requirement of dynamically correcting the baking paint process parameters through the improved fuzzy PID algorithm in step S4 is translated into specific parameter adjustment operations, while simultaneously accepting graded response instructions, ultimately achieving precise and dynamic correction of the baking paint process parameters, ensuring that the subsequent baking paint quality converges to the preset standard.

[0061] More specifically, the core action of step S4032 is to obtain the PID parameter correction amount through fuzzy rule reasoning and dynamically adjust the process parameters. The workflow of the fuzzy PID controller is "fuzzification → rule reasoning → defuzzification → PID parameter update", as follows: 1. Blurring process.

[0062] Convert precise input variables (e, ec) into fuzzy sets (e.g., "negative large (NB), negative medium (NM), zero (Z), positive medium (PM), positive large (PB)"), for example: If e = 2% (within the 1-3% deviation range of steps S401-403), then it is fuzzified to "PM"; If ec = 0.5% / s, then it is blurred to "positive small (PS)".

[0063] 2. Fuzzy rule reasoning.

[0064] Based on preset fuzzy rules (matching empirical logic with the baking paint process), the fuzzy correction amount of the PID controller parameters is inferred.

[0065] Example rule: "If the deviation e is 'positive middle (PM)' and the deviation change rate ec is 'positive small (PS)', then the proportional coefficient correction is 'positive small (PS)', the integral coefficient correction is 'zero (Z)', and the differential coefficient correction is 'negative small (NS)'."

[0066] 3. Defuzzing process.

[0067] The "fuzzy correction amount" obtained from fuzzy rule reasoning is converted into a precise correction value (△K). p , △K i , △K d The commonly used method is the center of gravity method.

[0068] 4. Dynamic adjustment of PID parameters.

[0069] The correction value is added to the initial PID parameters to obtain the adjusted PID parameters: ; Where K p0 / K i0 / K d0 These represent the initial proportional / integral / derivative coefficients of the PID controller parameters, respectively; ΔK p / △K i / △K d K represents the parameter correction amount obtained from fuzzy rule inference; p / K i / K d This indicates the adjusted PID controller parameters.

[0070] 5. Process parameter correction.

[0071] The adjusted PID controller outputs correction values ​​for baking process parameters (such as oven temperature and hot air circulation speed) based on the input e and ec, thereby achieving precise dynamic adjustment of process parameters and ultimately bringing quality deviations closer to the preset standard.

[0072] In step S5, the surface cooling temperature of the cured wire is collected in real time, and the cooling parameters are dynamically adjusted according to a preset cooling curve. This process controls the cooling after baking, preventing problems such as paint film cracking and decreased adhesion due to improper cooling rates. It is a crucial post-processing control step to ensure the quality of the final product. The cooling parameters include the flow rate and temperature of the cooling medium, clearly defining the core adjustable operational indicators during the cooling process. This allows for specific and targeted implementation of cooling control, ensuring the accuracy of cooling adjustments.

[0073] Specifically, step S5 includes the following sub-steps: S501. Based on the basic parameters of the wire to be painted, set the hierarchical adaptation rules of the preset cooling curve, and construct the preset cooling curve according to the hierarchical adaptation rules: In the initial stage of cooling, a rapid cooling mode is adopted, and the cooling rate is controlled at 5-8℃ / s; set the switching threshold between the rapid cooling mode and the slow cooling mode. When the surface temperature of the wire drops to 150-180℃, it automatically switches to the slow cooling mode, and the cooling rate is controlled at 1-2℃ / s. Its function is to realize the personalization and precise control of the preset cooling curve in stages, and solve the problem of adapting different cooling strategies to different wire characteristics. At the hierarchical adaptation level, adaptation rules are set based on the basic parameters of the wire to be coated (such as material and diameter) to match the preset cooling curve with the characteristics of the wire, avoiding insufficient compatibility caused by universal cooling curves. At the staged control level, a dual-mode cooling curve of rapid cooling and slow cooling is constructed, and the cooling rate and switching threshold of each stage are clearly defined. Specifically, rapid cooling in the initial stage: quickly reduces the surface temperature of the wire to prevent the coating film from staying in the high temperature range for too long and causing performance degradation; after triggering the switching threshold, it switches to slow cooling: avoids thermal stress between the coating film and the wire caused by a sudden drop in temperature, and prevents defects such as coating film cracking and reduced adhesion.

[0074] Specifically, the following steps are included after step S501 and before step S502: S5011. Based on the basic parameters of the wire to be coated, two cooling media adapted to different cooling modes are preset: an air-cooled medium adapted to the rapid cooling mode and an inert gas cooling medium adapted to the slow cooling mode. The suitable temperature range and switching conditions for each cooling medium are also preset. This aims to achieve precise matching between the cooling medium and the cooling mode, solving the problem that a single cooling medium cannot simultaneously achieve both "rapid cooling efficiency" and "slow cooling stability." Specifically, at the medium matching level, the air-cooled medium (high heat dissipation efficiency) is adapted to the rapid cooling mode, while the inert gas cooling medium (gentle heat dissipation and strong stability) is adapted to the slow cooling mode, ensuring a high degree of compatibility between the characteristics of the cooling medium and the requirements of the cooling mode. At the condition preset level, the suitable temperature range and switching conditions for each medium are clearly defined, providing a clear execution basis for subsequent medium switching during the cooling process and avoiding blind switching.

[0075] More specifically, the implementation logic of step S5011 is to match the cooling medium based on the basic parameters of the wire and preset the compatibility conditions of the medium. The specific process is as follows: 1. First, obtain the basic parameters of the wire to be coated (such as material M, diameter D, preset coating thickness T). film Based on the thermal characteristics of the basic parameters (such as rapid heat dissipation of metal wires and the need for gradual cooling when the coating film is thicker), two cooling media are preset: Air-cooled medium: compatible with "rapid cooling mode" (high heat dissipation efficiency, matching the rapid cooling requirements of the high-temperature section after paint baking). Inert gas cooling medium: adapted to "slow cooling mode" (gradual heat dissipation, avoiding stress on the paint film due to sudden cooling).

[0076] 2. Preset suitable temperature range: Air-cooled medium compatibility range: T wire >T th (T) wire T is the surface temperature of the wire. th =150~180℃, which is the threshold for switching between rapid and slow cooling modes. Inert gas medium compatibility range: T wire ≤T th .

[0077] 3. Preset switching judgment condition: When the surface temperature T of the wire wire Down to T th At that time, the cooling medium switching is triggered.

[0078] S5012: Real-time acquisition of wire surface temperature during the cooling process, comparing it with the suitable temperature range and switching criteria of each cooling medium, and simultaneously acquiring the inlet and outlet temperatures of the cooling medium, the temperature and humidity of the cooling environment, and the real-time speed of wire transmission to form a multi-dimensional dynamic parameter set for the cooling process. Its function is to construct a full-dimensional data monitoring system for the cooling process. Specifically, this manifests as: temperature comparison: providing real-time triggering basis for medium switching, ensuring that the switching timing accurately matches the cooling mode requirements; multi-dimensional parameter acquisition: incorporating the state of the cooling medium, environmental conditions, and wire transmission status into the monitoring scope to form a "multi-dimensional dynamic parameter set for the cooling process," providing comprehensive and real-time data support for subsequent control and adjustment after cooling medium switching, and ensuring the accuracy of the control strategy.

[0079] More specifically, the implementation logic of step S5012 lies in real-time acquisition of cooling process parameters to form a multi-dimensional dynamic parameter set. The specific process is as follows: 1. Real-time acquisition of wire surface temperature T using a temperature sensor. wire (t), and compare it with the "cooling medium matching temperature range" and "switching judgment condition" preset in step S5011 to determine the cooling medium that should be used at present. 2. The following parameters are collected synchronously through the corresponding sensors: Cooling medium inlet / outlet temperature: T in (t), T out (t); Cooling environment parameters: Ambient temperature T env (t), ambient humidity H env (t); Wire conveying parameters: Real-time wire conveying speed V wire (t); 3. Integrate the above parameters into a multi-dimensional dynamic parameter set for the cooling process: P(t)={T wire (t),T in (t),T out (t)}.

[0080] S5013. When the surface temperature of the wire drops from the rapid cooling mode to the threshold of the slow cooling mode, the cooling medium switching process is automatically triggered, and the cooling medium purging and replacement procedure is started simultaneously to ensure that there is no residual air-cooled medium in the cooling channel. After the replacement is completed, the inert gas cooling medium supply is turned on. Its function is to realize the automated and clean switching of the cooling medium. Specifically, it is manifested in the following ways: Automatic switching: Connecting to the cooling mode switching threshold in step S501, ensuring that the cooling medium switching is synchronized with the temperature stage; Purging and replacement procedure: Removing residual air-cooled medium in the cooling channel to avoid the mixing of different media affecting the cooling effect of the slow cooling mode (such as the inert gas being contaminated by air in the air-cooled medium, causing changes in heat dissipation characteristics), ensuring the stability of the slow cooling mode and the consistency of the cooling effect.

[0081] More specifically, the implementation logic of step S5013 is that temperature triggers the switching of the cooling medium to ensure that there is no residue of the cooling medium. The specific process is as follows: 1. Real-time monitoring of T wire (t), when T wire (t)≤T th When this occurs, the cooling medium switching process is automatically triggered, where T here... th Indicates the threshold for the slow cooling mode; 2. Start the cooling medium purging and replacement procedure: introduce inert gas to purge the cooling channel for a preset time (e.g., 30s), or detect "residual air-cooling medium concentration Cair ≤ 0.1%" through the gas concentration sensor to determine that there is no residual air-cooling medium in the channel; 3. After the replacement is completed, close the air-cooled medium supply valve and open the inert gas cooling medium supply valve.

[0082] S5014. After the cooling medium is switched, based on the multi-dimensional dynamic parameter set of the cooling process, the supply pressure, temperature, and flow rate of the inert gas cooling medium are synchronously adjusted through an improved PID collaborative control algorithm to keep the wire cooling rate stable within the target range. Simultaneously, the surface stress value of the coating film is monitored in real time. Its function is to achieve precise and coordinated control in the slow cooling mode. Specifically, this is manifested in: coordinated adjustment: through the improved PID collaborative control algorithm, the supply pressure, temperature, and flow rate of the inert gas are synchronously adjusted to avoid fluctuations in the cooling rate caused by single parameter adjustment, ensuring that the cooling rate stably matches the target range of the slow cooling mode in step S501; stress monitoring: the surface stress value of the coating film is monitored synchronously to detect the risk of stress accumulation caused by improper cooling in advance, providing an early warning basis for subsequent dynamic correction.

[0083] More specifically, the implementation logic of step S5014 is as follows: through improved PID collaborative control, the cooling rate is stabilized and stress is monitored, with the control objective of "the wire cooling rate stabilizing within the target range (1~2℃ / s)". Based on the multi-dimensional dynamic parameter set P(t) of the cooling process, the supply pressure, temperature, and flow rate of the inert gas are synchronously adjusted through the improved PID collaborative control algorithm. 1. Define control variables and objectives.

[0084] Control objective: Cooling rate r target ∈[1,2]℃ / s (target interval); Controlled variable: Supply pressure P of inert gas gas Temperature T gas Traffic Q gas ; Deviation: Cooling rate deviation e r =r target -r cool ( (actual cooling rate); Deviation change rate: .

[0085] 2. Improved PID collaborative control formula.

[0086] For each control variable (with supply pressure P) gas For example, the correction formula is: ; Similarly, temperature T gas Traffic Q gas The formula for the correction amount is: ; ; Where e r This represents the deviation in cooling rate, i.e., the target cooling rate r. target Compared with the actual cooling rate r cool The difference; The cooling rate e is represented by r The integral term over time; K p,P / K i,P / K d,P These represent the proportional, integral, and derivative coefficients of the PID controller corresponding to pressure control; K p,T / K i,T / K d,T These represent the PID proportional / integral / derivative coefficients corresponding to temperature control; K p,Q / K i,Q / K d,Q These represent the PID proportional / integral / derivative coefficients corresponding to flow control; ΔPgas / △T gas / △Q gas These represent the correction amounts for the corresponding control variables.

[0087] 4. Control variable updates and detection.

[0088] Adding the correction amount to the initial value yields the adjusted control variable: P gas,new =P gas,init +△P gas ; T gas,new =T gas,init +△T gas ; Q gas,new =Q gas,init +△Q gas ; Simultaneously, the stress value σ on the paint film surface is monitored in real time using a stress sensor. film .

[0089] Where P gas,new This indicates the adjusted supply pressure of the inert gas cooling medium; P gas,init Indicates the initial supply pressure of the inert gas cooling medium (the initial setpoint when the control process starts); △P gas The correction amount representing the inert gas supply pressure (calculated using the improved PID collaborative control formula); T gas,new Indicates the temperature of the inert gas cooling medium after adjustment; T gas,init Indicates the initial temperature of the inert gas cooling medium (the initial setpoint when the control process starts); △T gas The correction amount representing the inert gas temperature (calculated using the improved PID collaborative control formula); Q gas,new Indicates the adjusted flow rate of the inert gas cooling medium; Q gas,init Indicates the initial flow rate of the inert gas cooling medium (the initial setpoint when the control process starts); △Q gas This represents the correction amount for the inert gas flow rate (calculated using the improved PID collaborative control formula).

[0090] S5015. When the stress value exceeds the preset safety threshold, the temperature of the inert gas cooling medium is dynamically fine-tuned by ±2℃ until the stress value is lower than the preset safety threshold. Its function is to construct a closed-loop correction mechanism for paint film stress. Specifically, it is manifested in the following ways: dynamic fine-tuning: to address the risk of stress exceeding the standard, the cooling rate is corrected by fine-tuning the inert gas temperature by a small amount (±2℃), avoiding large adjustments (such as ≥5℃) that could lead to new process fluctuations; risk elimination: by continuously fine-tuning until the stress returns to the safety threshold, the quality defects such as cracking and peeling of the paint film due to excessive stress are effectively avoided, ensuring the quality stability of the final baked paint product.

[0091] More specifically, the implementation logic of step S5015 is that when the stress exceeds the threshold, the temperature of the cooling medium is dynamically fine-tuned. The specific process is as follows: 1. Real-time monitoring of the surface stress value σ of the paint film film , and the preset safety threshold σ safe Comparison; 2. When σ film >σ safe At that time, the temperature of the inert gas cooling gas is dynamically fine-tuned; the fine-tuning range is ±2℃ (if the stress is too high, Tgas is appropriately increased to slow down the cooling rate and reduce the stress on the paint film). 3. Continuously fine-tune and monitor σ film until σ film ≤σ safe .

[0092] S502. Set cooling stop criteria and collect wire surface temperature in real time: When the wire surface temperature drops to room temperature ±5℃, stop cooling and record all parameters of the complete cooling process. This serves to standardize the termination criteria and data accumulation for the cooling process. Specifically, cooling stop criteria: use "wire surface temperature dropping to room temperature ±5℃" as a clear quantitative condition to avoid insufficient or excessive cooling and ensure the stability of the wire after cooling; parameter recording: record all parameters of the complete cooling process, providing data support for subsequent optimization of the preset cooling curve and connecting with the "associative data storage" step in S6 to enrich the process data dimensions of the database and facilitate iterative optimization of subsequent process parameters.

[0093] In step S6, the pass rate of the baked paint quality of the cooled finished product is detected. When the pass rate is 100%, the associated data is stored and the database is updated. This serves to accumulate high-quality process data, storing the associated data of wire characteristics, process parameters, and qualified quality into the database to optimize the accuracy of subsequent process parameter matching and achieve iterative optimization of the method. When the pass rate is <100%, process optimization suggestions are generated and the baked paint quality prediction model is iterated. This involves two-way optimization of quality issues, providing process adjustment suggestions to address current quality defects and iterating the prediction model to improve the accuracy of subsequent quality predictions, thus achieving continuous improvement of the method in long-term use.

[0094] Specifically, step S6 also includes the following sub-steps: S601. Obtain the specific parameters of non-conforming products, analyze the non-conforming factors, and generate the adjustment direction and range of specific parameters: Its function is to establish the correlation logic between non-conforming products and process adjustments, serving as the "analysis and decision-making link" for process optimization suggestions. Specifically, by obtaining the specific parameters of non-conforming products, the specific manifestations of quality defects can be accurately located; after analyzing the non-conforming factors, the corresponding process parameter adjustment direction and range are clarified, avoiding ambiguity in optimization suggestions, providing a clear basis for subsequent targeted adjustments, and ensuring the accuracy of process optimization.

[0095] S602. When the non-compliance factor is insufficient paint film adhesion, the generated process optimization suggestion is to increase the temperature of the curing zone of the paint oven by 5-10℃ and extend the curing time by 10-20s: For the specific non-compliance factor of "insufficient paint film adhesion", a quantitative process optimization solution is provided. Specifically, insufficient adhesion is usually related to incomplete paint film curing, and increasing the temperature of the curing zone of the paint oven and extending the curing time are targeted solutions; the quantitative adjustment range of 5-10℃ and 10-20s is clearly defined, so that the process optimization suggestion is transformed from "qualitative guidance" into "directly executable operating parameters", ensuring the feasibility and effectiveness of the optimization measures.

[0096] S603. When the non-conforming factor is uneven paint film thickness, the generated process optimization suggestion is to adjust the hot air circulation speed by ±0.5m / s and correct the wire conveying speed by ±0.2m / min: For the specific non-conforming factor of "uneven paint film thickness", a quantitative process optimization solution is provided. Specifically, uneven paint film thickness is directly related to the uniformity of hot air circulation and the stability of wire conveying during the baking process. Adjusting the hot air circulation speed and wire conveying speed are targeted solutions. The quantitative adjustment range of ±0.5m / s and ±0.2m / min is clearly defined, making the process optimization suggestion in this scenario equally executable, accurately solving the quality defect of uneven thickness, and improving the quality consistency of subsequent finished products.

[0097] Please refer to Figure 2This application provides a wire coating control system, which corresponds to the method in the above embodiments. Specifically, the system includes: The first acquisition module 201 is used to acquire the basic parameters of the wire to be painted and call the database to match and obtain the painting process parameters. The basic parameters include wire material, wire diameter, preset paint film thickness and paint type. The painting process parameters include the temperature of each zone of the painting oven, hot air circulation speed, wire conveying speed and painting time. The first control module 202 is used to perform wire baking paint according to the baking paint process parameters and to collect dynamic parameters in real time during the baking paint process; the dynamic parameters include the real-time temperature of each zone in the baking paint oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the curing progress of the paint film and the real-time speed of the wire conveying. The second control module 203 is used to input basic parameters and dynamic parameters into the paint quality prediction model and output the paint quality prediction result; the prediction result includes the adhesion level prediction value, the thickness uniformity prediction value and the corrosion resistance prediction value. The third control module 204 is used to analyze the deviation between the prediction result and the preset quality standard, and dynamically correct the baking paint process parameters through an improved fuzzy PID algorithm; the preset quality standard includes the adhesion level standard value, the thickness uniformity standard value, and the corrosion resistance standard value. The fourth control module 205 is used to collect the surface cooling temperature of the cured wire in real time and dynamically adjust the cooling parameters according to the preset cooling curve; the cooling parameters include the flow rate and temperature of the cooling medium; The fifth control module 206 is used to detect the paint quality pass rate of the finished product after cooling. When the pass rate is 100%, it associates and stores data and updates the database; when the pass rate is <100%, it generates process optimization suggestions and iterates the paint quality prediction model.

[0098] As can be seen from the above, the wire coating control method and system provided in this application have the following advantages: 1) Precisely adapt to the initial process to reduce quality fluctuations: By calling the database to match the baking paint process parameters based on basic parameters, the manual experience matching is replaced. This ensures that the baking paint process parameters (oven temperature, conveyor speed, etc.) are precisely matched with the wire characteristics (material, diameter, etc.), thereby reducing quality fluctuations caused by improper initial processes from the source.

[0099] 2) Pre-inspection and process monitoring to reduce waste in ineffective processes: By conducting pre-inspection of coating uniformity, unqualified wires can be intercepted in advance, preventing them from entering the baking process and causing resource waste; by collecting dynamic parameters in real time and predicting abnormalities, the quality of the baking process can be monitored in real time, risks can be identified in advance, and batches of unqualified products can be avoided.

[0100] 3) Improve the accuracy of quality prediction and realize model self-iteration: Adopt a CNN-LSTM hybrid neural network, extract the spatial correlation and temporal change features of parameters, and optimize the parameters by combining cross-validation to improve the accuracy of the quality prediction model; the model retraining operation is triggered by 5 consecutive sets of deviations to realize the automatic iterative update of the model and maintain the prediction accuracy in the long term without manual intervention.

[0101] 4) Graded risk control and precise correction to improve process control efficiency: By calculating influencing factors, the correlation between dynamic parameters and quality deviations is quantified, and the quality risk is accurately determined by combining risk grading standards; by using fuzzy PID graded correction, the correction intensity is adapted according to the risk level, and the PID parameters are dynamically adjusted to avoid process fluctuations caused by over-correction and ensure that quality deviations converge quickly.

[0102] 5) Staged cooling and stress control to ensure the integrity of the coating quality: Through staged cooling curves (rapid cooling and slow cooling) and dual cooling medium adaptation, the cooling requirements of different temperature ranges of the wire are matched; through PID collaborative control and stress fine-tuning, the cooling rate is stabilized and stress concentration of the coating is avoided, reducing defects such as coating cracking and insufficient adhesion.

[0103] 6) Closed-loop data iteration to continuously optimize production quality: After the pass rate reaches the target, the database is updated to accumulate high-quality correlation data of wire basic parameters, process parameters, and qualified quality, thereby improving the accuracy of subsequent process matching; quantitative optimization suggestions are generated for different non-conforming factors (insufficient adhesion, uneven thickness) to achieve targeted process improvement, forming a closed loop of collection, prediction, correction, and iteration, thereby improving production quality and efficiency in a long-term development manner.

[0104] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for controlling the baking paint application of wires, characterized in that, The method includes the following steps: The basic parameters of the wire to be painted are obtained, and the painting process parameters are obtained by matching the database. The basic parameters include wire material, wire diameter, preset paint film thickness and paint type. The painting process parameters include temperature of each zone of the painting oven, hot air circulation speed, wire conveying speed and painting time. The wire is baked with paint according to the baking process parameters, and dynamic parameters during the baking process are collected in real time. The dynamic parameters include the real-time temperature of each zone in the baking oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the curing progress of the paint film, and the real-time speed of the wire conveying. The basic and dynamic parameters are input into the paint quality prediction model, and the prediction results of the paint quality are output; the prediction results include the adhesion level prediction value, the thickness uniformity prediction value, and the corrosion resistance prediction value. The deviation between the predicted results and the preset quality standards is analyzed, and the baking paint process parameters are dynamically corrected through an improved fuzzy PID algorithm; the preset quality standards include adhesion level standard values, thickness uniformity standard values, and corrosion resistance standard values; The surface cooling temperature of the cured wire is collected in real time, and the cooling parameters are dynamically adjusted according to a preset cooling curve; the cooling parameters include the flow rate and temperature of the cooling medium. The system detects the pass rate of the paint quality of the finished product after cooling. When the pass rate is 100%, the system associates and stores the data and updates the database. When the pass rate is <100%, the system generates process optimization suggestions and iterates the paint quality prediction model.

2. The wire coating control method according to claim 1, characterized in that, After obtaining the basic parameters of the wire to be coated, and before implementing the coating process according to the coating parameters, the following steps are also included: Detect the uniformity of the paint coating on the surface of the wire before it enters the oven and determine whether it reaches the preset uniformity threshold; When the paint uniformity does not reach the preset uniformity threshold, a paint correction signal is generated and fed back to the previous paint process, and the paint feeding is suspended. S103. When the paint uniformity reaches the preset uniformity threshold, start the subsequent baking process.

3. The wire coating control method according to claim 2, characterized in that, The steps of inputting basic and dynamic parameters into the paint quality prediction model and outputting the paint quality prediction results also include: The detection predicts whether the prediction result exceeds the prediction anomaly warning threshold of the paint quality prediction model. The prediction anomaly warning threshold is set according to the basic parameters and dynamic parameters. When the deviation between five consecutive sets of predicted results and actual sampling results exceeds a preset threshold, the paint quality prediction model retraining process is automatically triggered, and an early warning signal is simultaneously output to prompt staff to check the data collection link.

4. The wire coating control method according to claim 3, characterized in that, When the deviation between five consecutive sets of predicted results and actual sampling results exceeds a preset threshold, the steps that automatically trigger the retraining process of the paint quality prediction model also include: Historical basic parameters and historical dynamic parameters are collected as input samples, and corresponding finished paint quality inspection results are collected as label samples. A CNN-LSTM hybrid neural network is used to train the paint quality prediction model. During training, cross-validation was used to optimize the parameters of the paint quality prediction model to ensure its accuracy.

5. The wire coating control method according to claim 4, characterized in that, The steps of analyzing the deviation between the predicted results and the preset quality standards, and dynamically correcting the baking paint process parameters using an improved fuzzy PID algorithm, also include: Based on the predicted results and dynamic parameter analysis, the influence factors of each dynamic parameter on the deviation of different quality indicators are calculated. The formula is as follows: ； Where F i,j Let ΔQ be the influence factor of the j-th dynamic parameter on the deviation of the i-th quality index; j Let D be the predicted deviation value of the i-th quality indicator, that is, the deviation between the predicted result and the preset quality standard; j The j-th dynamic parameter is the real-time acquired value; the quality indicators are adhesion level, thickness uniformity, and corrosion resistance. Establish a risk grading standard for paint coating quality based on influencing factors: "Single quality indicator deviation value ≤ 1% and no high influencing factors (F)" i,j "≥0.6) dynamic parameters" are classified as low risk; "a single quality indicator deviation value of 1%-3% or the existence of one high-impact factor dynamic parameter" is classified as medium risk; "any quality indicator deviation value >3% or the existence of two or more high-impact factor dynamic parameters" is classified as high risk. Based on the paint quality risk grading standard, the following graded response instructions are output: For low risk, a routine correction instruction and a simplified analysis report including only the deviation value are output; for medium risk, an enhanced correction instruction and a complete analysis report including the deviation value and the ranking of influencing factors are output; for high risk, an emergency correction instruction, a complete analysis report, and suggestions for investigating abnormalities in the collection of dynamic parameters of high-impact factors are output, and material feeding is suspended and reviewed.

6. The wire coating control method according to claim 5, characterized in that, The steps for issuing graded response instructions based on the paint quality risk grading standard also include: Based on the graded response instructions and the corresponding influencing factors, the deviation and deviation change rate between the corresponding prediction results and the preset quality standards are used as the core input of the fuzzy controller. At the same time, the correction intensity is adapted in combination with the graded response instructions: the normal correction step size is used for the regular correction instructions, the correction step size is increased by 1.2-1.5 times for the enhanced correction instructions, and the maximum safe step size is used for the emergency correction instructions. By using fuzzy rule reasoning to obtain the correction values ​​of the proportional coefficient, integral coefficient, and derivative coefficient of the PID controller, the initial PID parameters are dynamically adjusted to achieve precise correction of the baking paint process parameters.

7. The wire coating control method according to claim 6, characterized in that, The steps of real-time acquisition of the surface cooling temperature of the cured wire and dynamic adjustment of cooling parameters according to a preset cooling curve also include: Based on the basic parameters of the wire to be painted, a graded adaptation rule for the preset cooling curve is set, and a preset cooling curve is constructed according to the graded adaptation rule: in the initial stage of cooling, a rapid cooling mode is adopted, and the cooling rate is controlled at 5-8℃ / s; a switching threshold between the rapid cooling mode and the slow cooling mode is set, and when the surface temperature of the wire drops to 150-180℃, it automatically switches to the slow cooling mode, and the cooling rate is controlled at 1-2℃ / s. Set cooling stop criteria and collect wire surface temperature in real time: when the wire surface temperature drops to room temperature ±5℃, stop cooling and record all parameters of the complete cooling process.

8. The wire coating control method according to claim 7, characterized in that, After setting the hierarchical adaptation rules for the preset cooling curve based on the basic parameters of the wire to be painted, and constructing the preset cooling curve according to the hierarchical adaptation rules, and before setting the cooling stop judgment condition and collecting the wire surface temperature in real time, the following steps are also included: Based on the basic parameters of the wire to be painted, two cooling media adapted to different cooling modes are preset, including air cooling medium adapted to rapid cooling mode and inert gas cooling medium adapted to slow cooling mode. At the same time, the suitable temperature range and switching judgment conditions of each cooling medium are preset. The surface temperature of the wire during the cooling process is acquired in real time and compared with the suitable temperature range and switching conditions of each cooling medium. The inlet and outlet temperatures of the cooling medium, the temperature and humidity of the cooling environment and the real-time speed of the wire are acquired simultaneously to form a multi-dimensional dynamic parameter set for the cooling process. When the surface temperature of the wire drops from the rapid cooling mode to the threshold of the slow cooling mode, the cooling medium switching process is automatically triggered, and the cooling medium purging and replacement program is started simultaneously to ensure that there is no residual air-cooled medium in the cooling channel. After the replacement is completed, the inert gas cooling medium supply is turned on. After the cooling medium is switched, based on the multi-dimensional dynamic parameter set of the cooling process, the supply pressure, temperature and flow rate of the inert gas cooling medium are adjusted synchronously through the improved PID collaborative control algorithm, so that the cooling rate of the wire is kept stable within the target range, and the surface stress value of the paint film is monitored in real time. When the stress value exceeds the preset safety threshold, the temperature of the inert gas cooling medium is dynamically adjusted by ±2℃ until the stress value is lower than the preset safety threshold.

9. The wire coating control method according to claim 8, characterized in that, When the pass rate is <100%, the steps for generating process optimization suggestions and iterating the paint baking quality prediction model also include: Obtain the specific parameters of the defective products, analyze the factors causing the defects, and generate the adjustment direction and adjustment range for the specific parameters; When the non-compliance factor is insufficient paint film adhesion, the proposed process optimization is to increase the temperature of the curing zone of the paint oven by 5-10℃ and extend the curing time by 10-20 seconds. When the non-compliance factor is uneven paint film thickness, the proposed process optimization is to adjust the hot air circulation speed by ±0.5m / s and correct the wire conveying speed by ±0.2m / min.

10. A wire coating control system, characterized in that, The system includes: The first acquisition module (201) is used to acquire the basic parameters of the wire to be painted and call the database to match and obtain the painting process parameters. The basic parameters include wire material, wire diameter, preset paint film thickness and paint type. The painting process parameters include the temperature of each zone of the painting oven, hot air circulation speed, wire conveying speed and painting time. The first control module (202) is used to implement wire baking paint according to the baking paint process parameters and to collect dynamic parameters in real time during the baking paint process; the dynamic parameters include the real-time temperature of each zone in the baking paint oven, the ambient humidity of each zone, the real-time temperature of the wire surface, the curing progress of the paint film and the real-time speed of the wire conveying. The second control module (203) is used to input the basic parameters and dynamic parameters into the paint quality prediction model and output the paint quality prediction results; the prediction results include the adhesion level prediction value, the thickness uniformity prediction value and the corrosion resistance prediction value. The third control module (204) is used to analyze the deviation between the prediction result and the preset quality standard, and dynamically correct the baking paint process parameters through an improved fuzzy PID algorithm; the preset quality standard includes the adhesion level standard value, the thickness uniformity standard value and the corrosion resistance standard value. The fourth control module (205) is used to collect the surface cooling temperature of the cured wire in real time and dynamically adjust the cooling parameters according to the preset cooling curve; the cooling parameters include the flow rate and temperature of the cooling medium; The fifth control module (206) is used to detect the paint quality pass rate of the finished product after cooling. When the pass rate is 100%, it associates and stores data and updates the database; when the pass rate is <100%, it generates process optimization suggestions and iterates the paint quality prediction model.

Images