Ceramic film online thickness closed-loop control system and method based on machine learning

Abstract

The invention discloses a ceramic film online thickness closed-loop control system and method based on machine learning, and the method comprises the steps: in the production process of a ceramic film, a film thickness sensor and image recognition equipment monitor the film thickness, the film weight and the film density in real time, and transmit data to an AI diagnosis module; after being preprocessed, the data are input into a machine learning model to analyze the film thickness change trend, and abnormality is automatically diagnosed; according to the AI diagnosis result and the film thickness deviation, the system calculates the adjustment amount, generates a control signal, and transmits the control signal to the execution module through the feedback module to adjust production parameters; after adjustment is completed, online detection continues to be carried out; in combination with the relation of film weight, film thickness and film density, a dynamic coupling control strategy is adopted, film thickness control is automatically adjusted, and the stability of the production process is ensured; according to the ceramic film online thickness closed-loop control system and method based on machine learning, the technical target is achieved through real-time detection, feedback and closed-loop control execution.

Description

Technical Field

[0001] This invention belongs to the field of online thickness closed-loop control of ceramic thin films based on machine learning, and specifically relates to an online thickness closed-loop control system and method for ceramic thin films based on machine learning. Background Technology

[0002] A closed-loop online thickness control system for ceramic thin films based on machine learning primarily utilizes machine learning algorithms to monitor and adjust thickness variations during ceramic thin film production, thereby achieving automated and efficient production control. The system acquires real-time thickness data of the ceramic thin film through sensors, including the actual thickness and relevant parameters affecting thickness changes. Machine learning algorithms are then used to train and model the collected data. These models can identify key factors influencing film thickness and predict thickness variations under different operating conditions. Based on the predictions of the machine learning models, the system can automatically adjust parameters during production to maintain the film thickness within a set target range. This control method is "closed-loop" because the system monitors and corrects in real time, ensuring thickness stability. The machine learning algorithms can be continuously optimized based on historical data, improving prediction accuracy and control effectiveness. The system can learn and adapt to different production conditions, and its control precision continuously improves with data accumulation.

[0003] However, although machine learning-based online thickness closed-loop control systems for ceramic thin films have a high level of automation and optimization potential, they may also have some defects or challenges in practical applications. Although there are film thickness measurement technologies based on optical and mechanical means, they still face problems such as real-time feedback lag and low control accuracy during dynamic adjustment. Compensating for film thickness fluctuations through manual control often cannot make an accurate response in a short time, resulting in large film thickness deviations and failing to meet the high-precision control requirements. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a machine learning-based online thickness closed-loop control system and method for ceramic thin films. Through real-time detection, feedback, and execution of closed-loop control, it achieves self-correction of thickness deviation (target range ≤ 0.5 μm), utilizes artificial intelligence (AI) technology to diagnose anomalies in production, and realizes the technical goal of dynamic coupling control between film weight, film thickness, and density.

[0005] The technical solution adopted by this invention to solve its technical problem is:

[0006] A machine learning-based online thickness closed-loop control system for ceramic thin films includes:

[0007] The online detection module is used to monitor production parameters such as membrane thickness, membrane weight, and membrane density in real time. This module includes a membrane thickness sensor, a surface image recognition device, and a gas flow sensor monitoring device, which collects and feeds back data on parameters during the production process in real time.

[0008] The data processing and AI diagnostic module is used to process and analyze the real-time data acquired by the online detection module. This module is based on machine learning algorithms and uses trained models to detect anomalies in the data, automatically identifying anomalies that will occur in the production process.

[0009] The control decision and feedback module is used to calculate the adjustment amount based on the analysis results of the AI ​​diagnostic module and the deviation information of the film thickness, and generate the corresponding control signal and transmit the signal to the execution module.

[0010] The execution module is used to control the adjustment of production equipment and optimize production conditions in real time according to the instructions provided by the control decision module.

[0011] A machine learning-based online thickness closed-loop control method for ceramic thin films includes the following steps:

[0012] In the production process of ceramic thin films, online monitoring equipment using film thickness sensors and image recognition devices is used to obtain film thickness, film weight, and film density parameters in real time, and the detected data is transmitted to the data processing and AI diagnostic module for analysis.

[0013] After preprocessing the real-time detection data, it is input into the machine learning model for analysis. The trained model is used to analyze the trend of film thickness change and automatically diagnose anomalies that occur during the production process, such as vibration marks, lines, and bubbles.

[0014] Based on the AI ​​diagnostic results and film thickness deviation, the system calculates the adjustment amount, calculates the control parameters according to the film thickness change trend and the real-time status of the production equipment, and generates control signals.

[0015] The control signal is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal. After the adjustment is completed, online monitoring is continuously performed.

[0016] During the membrane thickness control process, a dynamic coupling control strategy is adopted by combining the relationship between membrane weight, membrane thickness, and membrane density. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy.

[0017] As a preferred method, during the production process of ceramic thin films, the film thickness, film weight, and film density parameters are acquired in real time using online monitoring equipment with film thickness sensors and image recognition devices, and the detected data is transmitted to a data processing and AI diagnostic module for analysis.

[0018] In the production process of ceramic thin films, film thickness sensors and image recognition equipment are used for online monitoring to detect and obtain production parameters such as film thickness, film weight, and film density in real time.

[0019] The above detection data is transmitted in real time to the data processing and AI diagnosis module via the communication interface for data analysis and anomaly diagnosis.

[0020] In the data processing and AI diagnostic module, a preset machine learning algorithm is used to analyze the collected detection data in real time, determine the quality of the film layer, identify impending film thickness deviations and surface defect anomalies, and generate control signals based on the diagnostic results.

[0021] As a preferred method, after preprocessing the real-time detection data, it is input into a machine learning model for analysis. The trained model is then used to analyze the trend of film thickness changes and automatically diagnose anomalies occurring during the production process, such as vibration marks, lines, and bubbles.

[0022] The real-time detection data obtained from the online detection equipment is preprocessed. The preprocessing steps include noise reduction, normalization, missing value imputation, and outlier removal.

[0023] The preprocessed data is input into the machine learning model, and the time series of film thickness, rate of change, and surface defect image features are extracted based on the film thickness change trend.

[0024] The input data is analyzed using a trained machine learning model. This model learns from historical data to capture the patterns of membrane thickness changes and predicts the membrane thickness trend in the current production process.

[0025] Based on the analysis results of the machine learning model, the system automatically diagnoses abnormalities that occur during the production process. If the film thickness changes abnormally, the model identifies defects such as vibration marks, lines, and bubbles, and prompts production personnel to make adjustments by outputting corresponding diagnostic information.

[0026] Based on the abnormal information from the automatic diagnosis, the system further provides adjustment suggestions or automatically generates control signals, which are then fed back to the production equipment for real-time adjustments.

[0027] As a preferred approach, based on AI diagnostic results and film thickness deviation, the system calculates adjustment amounts, calculates control parameters according to the film thickness change trend and the real-time status of the production equipment, and generates control signals using the following method:

[0028] The film thickness deviation Δh(t) is the difference between the real-time measured film thickness h(t) and the target film thickness h. target The difference between them:

[0029] Δh(t)=h(t)-h target

[0030] Where h(t) is the film thickness at time t; h target The target film thickness is set.

[0031] Film thickness variation trend The rate of change of film thickness with time is expressed as the derivative of the change in film thickness within the time window:

[0032]

[0033] Where Δt is the time step; h(t-Δt) is the film thickness at the previous time step t-Δt;

[0034] The AI ​​diagnostic model analyzes the trend of film thickness variation and other parameters to generate an adjustment factor A(t), which is used to correct film thickness deviations and optimize the adjustment of production equipment.

[0035]

[0036] Where f is an AI diagnostic function based on machine learning, and the output adjustment factor A(t) is a parameter related to the real-time status of the production equipment;

[0037] The system calculates the film thickness adjustment amount ΔH based on the film thickness deviation Δh(t) and the AI ​​diagnostic adjustment factor A(t). adjust (t), used to control the adjustment of production equipment:

[0038] ΔH adjust (t)=α·Δh(t)+β·A(t)

[0039] Where α and β are adjustment coefficients;

[0040] Adjustment amount ΔH based on membrane thickness adjust Given E(t) and the real-time status of the production equipment, the system calculates the adjusted control parameter C(t):

[0041] C(t) = γ·ΔH adjust (t)+·E(t)

[0042] Where γ and E(t) are adjustment coefficients; E(t) is the real-time state of the equipment.

[0043] The control signal S(t) is generated based on the control parameter C(t) and is used to adjust the production equipment.

[0044] S(t)=κ·C(t)

[0045] Where κ is a proportionality constant.

[0046] Preferably, the control signal is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal. After the adjustment is completed, the online monitoring is continuously performed as follows:

[0047] The control signal S(t) is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal, and the adjustment amount is ΔP. adjust (t) represents the adjustment of the production parameter P(t) by the execution module:

[0048] ΔP adjust (t)=η·S(t)

[0049] Where η is a proportional coefficient; S(t) is the control signal output from the control decision module;

[0050] The execution module adjusts the production parameter P(t) according to the control signal S(t) to obtain the new production parameter P. new (t):

[0051] P new (t)=P(t)+ΔP adjust (t)=P(t)+η·S(t)

[0052] Adjusted production parameters P new (t) Production parameters affecting membrane thickness: The online detection module will continue to monitor membrane thickness, membrane weight, and membrane density parameters, and acquire new detection data D. new (t):

[0053] D new (t) = OnlineDetection(P new (t))

[0054] Among them, D new (t) is based on the new production parameter P new (t) New detection data obtained after online detection;

[0055] New test data D new (t) will be transmitted to the data processing and AI diagnostic module for real-time analysis and anomaly diagnosis. This module will analyze the new data and output new adjustment suggestions and control signals.

[0056] After the system completes the adjustment through the execution module, the new production parameter P new (t) affects the membrane thickness parameter, and new data is obtained through the online detection system. The data is input to the control decision module to form a new feedback loop.

[0057] Preferably, during the membrane thickness control process, a dynamic coupling control strategy is adopted, taking into account the relationship between membrane weight, membrane thickness, and membrane density. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy as follows:

[0058] The following fundamental physical relationships exist between membrane weight (m), membrane thickness (h), and membrane density (ρ):

[0059] m=h·A·ρ

[0060] Where m is the membrane weight; h is the membrane thickness; A is the membrane surface area; and ρ is the membrane density.

[0061] During the membrane thickness control process, when the membrane density ρ or membrane weight m changes, the membrane thickness h is adjusted to maintain product quality. The adjustment amount is dynamically calculated using the following formula.

[0062] Define the change in membrane weight or membrane density as Δm(t) or Δρ(t). Based on the relationship between membrane weight and membrane thickness, calculate the adjustment amount for membrane thickness. The formula for the change in membrane weight is:

[0063] Δm(t)=m(t)-m target

[0064] Where, m target Target membrane weight;

[0065] Calculate the adjustment amount Δh(t) for membrane thickness based on the change in membrane weight:

[0066]

[0067] Change in membrane density:

[0068] Δρ(t)=ρ(t)-ρ target

[0069] Where, ρ target Target film density;

[0070] Calculate the adjustment amount Δh(t) of the membrane thickness based on the change in membrane density:

[0071]

[0072] When both membrane weight and membrane density change simultaneously, the combined adjustment amount for membrane thickness, Δh, is calculated by combining these two factors. total The formula for calculating (t) is:

[0073]

[0074] Based on the changing trend of membrane thickness, as well as the changes in membrane weight and membrane density, the control strategy is dynamically adjusted, and the control system generates a new control signal S(t) based on these changes.

[0075] The control signal S(t) is based on the adjustment amount Δh of the film thickness. total (t) and other relevant factors are used to generate a control signal, and a proportional relationship is established between the generated control signal and the film thickness adjustment amount. The control signal is expressed as:

[0076] S(t)=κ·Δh total (t)+λ·E(t)

[0077] Where κ is a proportionality constant; E(t) is the real-time state of the production equipment; and λ is the influence coefficient of the equipment state on the control signal.

[0078] The system transmits the control signal S(t) to the execution module. The execution module adjusts the production parameters according to the control signal. The adjusted production parameters affect the membrane thickness, membrane weight, and membrane density. The online detection module continuously monitors the membrane thickness parameter and feeds back the new detection data to the system.

[0079] Another technical problem to be solved by the present invention is to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements a machine learning-based online thickness closed-loop control system and method for ceramic thin films as described above.

[0080] Another technical problem to be solved by the present invention is to provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a machine learning-based online thickness closed-loop control system and method for ceramic thin films.

[0081] The beneficial effects of this invention are:

[0082] By training and adaptively adjusting machine learning models, precise control of membrane thickness is achieved, with a deviation range controlled within ≤0.5μm, ensuring stable product quality. AI technology is used to detect production anomalies in real time, reducing human intervention, shortening response time, and ensuring stable production processes. Dynamic coupling control between membrane weight, membrane thickness, and membrane density can improve the overall stability of the production process, reduce fluctuations, and increase production efficiency. Through machine learning and closed-loop control systems, automation and intelligence of the production process are achieved, reducing human intervention and improving production efficiency and product consistency. Attached Figure Description

[0083] Figure 1 This is a schematic diagram of the online thickness closed-loop control system for ceramic thin films based on machine learning, according to the present invention. Detailed Implementation

[0084] The principles and features of the present invention are described below. The examples given are for illustrative purposes only and are not intended to limit the scope of the invention. The invention is described more specifically by way of example in the following paragraphs. The advantages and features of the invention will become clearer from the following description and claims.

[0085] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0086] Example

[0087] The technical solution adopted by this invention to solve its technical problem is:

[0088] A machine learning-based online thickness closed-loop control system for ceramic thin films includes:

[0089] The online detection module is used to monitor production parameters such as membrane thickness, membrane weight, and membrane density in real time. This module includes a membrane thickness sensor, a surface image recognition device, and a gas flow sensor monitoring device, which collects and feeds back data on parameters during the production process in real time.

[0090] The data processing and AI diagnostic module is used to process and analyze the real-time data acquired by the online detection module. This module is based on machine learning algorithms and uses trained models to detect anomalies in the data, automatically identifying anomalies that will occur in the production process.

[0091] The control decision and feedback module is used to calculate the adjustment amount based on the analysis results of the AI ​​diagnostic module and the deviation information of the film thickness, and generate the corresponding control signal and transmit the signal to the execution module.

[0092] The execution module is used to control the adjustment of production equipment and optimize production conditions in real time according to the instructions provided by the control decision module.

[0093] By monitoring membrane thickness, weight, and density in real time, and combining AI algorithms for prediction and adjustment, the system can precisely control membrane thickness, reducing fluctuations and deviations during production. Automatic detection and adjustment minimize human error and intervention, ensuring a more stable and efficient production process. Machine learning algorithms can identify anomalies in a timely manner during production, providing early warnings of potential problems, reducing defective products, and lowering the scrap rate. Through continuous feedback on equipment status and production parameters, the system can self-regulate and optimize production conditions, thereby improving production efficiency and saving costs. The system can automatically adjust strategies according to different production needs and equipment status, exhibiting strong adaptability and accommodating different production environments. By precisely controlling membrane thickness and correcting deviations in a timely manner, the produced ceramic films are of more uniform and stable quality, meeting higher quality standards.

[0094] A machine learning-based online thickness closed-loop control method for ceramic thin films includes the following steps:

[0095] In the production process of ceramic thin films, online monitoring equipment using film thickness sensors and image recognition devices is used to obtain film thickness, film weight, and film density parameters in real time, and the detected data is transmitted to the data processing and AI diagnostic module for analysis.

[0096] After preprocessing the real-time detection data, it is input into the machine learning model for analysis. The trained model is used to analyze the trend of film thickness change and automatically diagnose anomalies that occur during the production process, such as vibration marks, lines, and bubbles.

[0097] Based on the AI ​​diagnostic results and film thickness deviation, the system calculates the adjustment amount, calculates the control parameters according to the film thickness change trend and the real-time status of the production equipment, and generates control signals.

[0098] The control signal is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal. After the adjustment is completed, online monitoring is continuously performed.

[0099] During the membrane thickness control process, a dynamic coupling control strategy is adopted by combining the relationship between membrane weight, membrane thickness, and membrane density. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy.

[0100] Through intelligent analysis of real-time data using machine learning models, the system can identify and predict potential production problems in real time, such as film thickness fluctuations, bubbles, and vibration patterns. This helps the production line respond and adjust quickly, maintaining the stability of the production process and the consistency of film thickness. Traditional film thickness control methods may rely on manual experience and intuitive adjustments, making them susceptible to operator errors. By using machine learning to detect anomalies in the production process, the system can provide early warnings of potential production anomalies, adjust production parameters in a timely manner, prevent defective products, and reduce scrap rates. The system can dynamically adjust the film thickness control strategy based on the interrelationships between parameters such as film weight, film thickness, and film density. The system integrates multiple modules, including online detection, data analysis, control decision-making, and execution feedback, achieving full automation from data acquisition to production adjustment. Through training the machine learning model, the system can gradually improve the accuracy of predicting film thickness change trends as production process data accumulates, making the control strategy more precise and personalized, adapting to different production environments and process requirements.

[0101] In the production process of ceramic thin films, the method of acquiring real-time parameters such as film thickness, film weight, and film density through online monitoring equipment using film thickness sensors and image recognition devices, and then transmitting the detected data to a data processing and AI diagnostic module for analysis, is as follows:

[0102] In the production process of ceramic thin films, film thickness sensors and image recognition equipment are used for online monitoring to detect and obtain production parameters such as film thickness, film weight, and film density in real time.

[0103] The above detection data is transmitted in real time to the data processing and AI diagnosis module via the communication interface for data analysis and anomaly diagnosis.

[0104] In the data processing and AI diagnostic module, a preset machine learning algorithm is used to analyze the collected detection data in real time, determine the quality of the film layer, identify impending film thickness deviations and surface defect anomalies, and generate control signals based on the diagnostic results.

[0105] Through real-time data acquisition and AI diagnostics, the system can accurately identify quality problems and thickness deviations in the membrane layer, enabling timely detection and adjustment, thereby improving the accuracy of membrane quality and reducing the possibility of human error. The system can provide early warnings before problems arise, such as predicting thickness deviation trends and surface defects, and automatically issuing adjustment signals to prevent the production of substandard products. This method reduces the need for manual operation through automated detection and analysis, making the production process more autonomous and intelligent. The system can provide real-time feedback and adjust production parameters to ensure consistent membrane quality throughout the production process and reduce fluctuations. Automated monitoring and control help improve production efficiency and avoid production stoppages and rework due to substandard membrane quality. The system can adapt to different production environments and conditions, optimizing control strategies under different operating modes. When membrane weight or density changes, the system can automatically adjust the membrane thickness control strategy to ensure product quality is maintained under various circumstances.

[0106] After preprocessing the real-time detection data, it is input into a machine learning model for analysis. The trained model is used to analyze the trend of film thickness change and automatically diagnose anomalies occurring during the production process, such as vibration marks, lines, and bubbles.

[0107] The real-time detection data obtained from the online detection equipment is preprocessed. The preprocessing steps include noise reduction, normalization, missing value imputation, and outlier removal.

[0108] The preprocessed data is input into the machine learning model, and the time series of film thickness, rate of change, and surface defect image features are extracted based on the film thickness change trend.

[0109] The input data is analyzed using a trained machine learning model. This model learns from historical data to capture the patterns of membrane thickness changes and predicts the membrane thickness trend in the current production process.

[0110] Based on the analysis results of the machine learning model, the system automatically diagnoses abnormalities that occur during the production process. If the film thickness changes abnormally, the model identifies defects such as vibration marks, lines, and bubbles, and prompts production personnel to make adjustments by outputting corresponding diagnostic information.

[0111] Based on the abnormal information from the automatic diagnosis, the system further provides adjustment suggestions or automatically generates control signals, which are then fed back to the production equipment for real-time adjustments.

[0112] By monitoring film thickness changes and surface defects in real time, the system can promptly detect film quality issues, preventing substandard products from entering the next stage. This solution can diagnose film thickness changes and defects without human intervention, significantly reducing errors caused by manual operation. The machine learning model can predict film thickness trends based on historical data and promptly identify anomalies in the production process. Based on model analysis results, the system not only provides accurate diagnostic information but also generates automatically adjusting control signals, directly feeding back to the production equipment for optimization. Automated defect identification and adjustment can quickly resolve problems related to film thickness changes, avoiding costs such as downtime and rework due to quality issues. Because it uses a machine learning model, the system can adapt to film thickness change trends under different production conditions and propose optimization solutions based on different defect types. This allows it to cope with various production environments and flexibly adjust control strategies. As production data accumulates, the machine learning model can continuously optimize itself, gradually improving the accuracy of predicting film thickness trends and the ability to identify surface defects, making the system more intelligent in long-term use.

[0113] Based on the AI ​​diagnostic results and film thickness deviation, the system calculates the adjustment amount. According to the film thickness change trend and the real-time status of the production equipment, the system calculates the control parameters and generates control signals using the following method:

[0114] The film thickness deviation Δh(t) is the difference between the real-time measured film thickness h(t) and the target film thickness h. target The difference between them:

[0115] Δh(t)=h(t)-h target

[0116] Where h(t) is the film thickness at time t; h target The target film thickness is set.

[0117] Film thickness variation trend The rate of change of film thickness with time is expressed as the derivative of the change in film thickness within the time window:

[0118]

[0119] Where Δt is the time step; h(t-Δt) is the film thickness at the previous time step t-Δt;

[0120] The AI ​​diagnostic model analyzes the trend of film thickness variation and other parameters to generate an adjustment factor A(t), which is used to correct film thickness deviations and optimize the adjustment of production equipment.

[0121]

[0122] Where f is an AI diagnostic function based on machine learning, and outputs an adjustment factor A(t); D(t) is a parameter related to the real-time status of the production equipment;

[0123] The system calculates the film thickness adjustment amount ΔH based on the film thickness deviation Δh(t) and the AI ​​diagnostic adjustment factor A(t). adjust (t), used to control the adjustment of production equipment:

[0124] ΔH adjust (t)=α·Δh(t)+β·A(t)

[0125] Where α and β are adjustment coefficients;

[0126] Adjustment amount ΔH based on membrane thickness adjust Given E(t) and the real-time status of the production equipment, the system calculates the adjusted control parameter C(t):

[0127] C(t) = γ·ΔH adjust (t)+·E(t)

[0128] Where γ and E(t) are adjustment coefficients; E(t) is the real-time state of the equipment.

[0129] The control signal S(t) is generated based on the control parameter C(t) and is used to adjust the production equipment.

[0130] S(t)=κ·C(t)

[0131] Where κ is a proportionality constant.

[0132] By monitoring membrane thickness, its variation trend, and equipment status in real time, the system can precisely control membrane thickness, avoiding excessive deviations and ensuring the stability of membrane quality. This solution can calculate membrane thickness adjustments in real time and automatically generate control signals based on a machine learning model, feeding them back to the production equipment for adjustment. The AI ​​diagnostic model can learn from historical data the patterns of membrane thickness changes and the relationship between equipment status and membrane thickness, thus intelligently providing dynamic adjustment factors for the production process, achieving more personalized and efficient adjustments. Precise membrane thickness control and timely equipment adjustment reduce unnecessary loads on the equipment, preventing over-operation and extending the lifespan of the production equipment. By reducing membrane thickness deviations and defects, product consistency can be improved, reducing the production of defective products and lowering costs associated with rework and scrap. The system makes dynamic adjustments based on real-time monitoring data and changes in production status, enabling rapid response to any changes during production and enhancing the flexibility and responsiveness of the production line. This solution forms a closed-loop control system. The system continuously analyzes and optimizes adjustment factors in real time through AI. As production data accumulates, the model's predictive accuracy gradually improves, thereby continuously optimizing the production process and membrane quality.

[0133] The control signal is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal. After the adjustment is completed, the method for continuous online monitoring is as follows:

[0134] The control signal S(t) is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal, and the adjustment amount is ΔP. adjust (t) represents the adjustment of the production parameter P(t) by the execution module:

[0135] ΔP adjust (t)=η·S(t)

[0136] Where η is a proportional coefficient; S(t) is the control signal output from the control decision module;

[0137] The execution module adjusts the production parameter P(t) according to the control signal S(t) to obtain the new production parameter P. new (t):

[0138] P new (t)=P(t)+ΔP adjust (t)=P(t)+η·S(t)

[0139] Adjusted production parameters P new (t) Production parameters affecting membrane thickness: The online detection module will continue to monitor membrane thickness, membrane weight, and membrane density parameters, and acquire new detection data P. new (t):

[0140] D new (t) = OnlineDetection(P new ((t))

[0141] Among them, D new (t) is based on the new production parameter P new (t) New detection data obtained after online detection;

[0142] New test data D new (t) will be transmitted to the data processing and AI diagnostic module for real-time analysis and anomaly diagnosis. This module will analyze the new data and output new adjustment suggestions and control signals.

[0143] After the system completes the adjustment through the execution module, the new production parameter P new (t) affects the membrane thickness parameter, and new data is obtained through the online detection system. The data is input to the control decision module to form a new feedback loop.

[0144] This solution ensures that key parameters such as membrane thickness remain within ideal ranges by real-time monitoring and dynamic adjustment of production parameters. The system employs closed-loop control, transmitting online monitoring data as feedback to the control decision module, creating a dynamic adjustment process. The AI ​​diagnostic module intelligently analyzes extensive historical and real-time monitoring data to provide precise adjustment suggestions. By monitoring key parameters such as membrane thickness, weight, and density online, the system effectively monitors every stage of the production process, ensuring the final product quality meets requirements. The automated adjustment process reduces the need for manual intervention, allowing the production line to self-optimize without human intervention. Through continuous monitoring and automatic adjustment of production parameters, the production system can quickly adapt to different production environments and demands, enhancing the flexibility and controllability of the production process. Because production parameters are continuously optimized under real-time monitoring and control, the system maximizes production efficiency, reduces resource waste, and lowers equipment maintenance costs. The AI ​​diagnostic module not only performs real-time analysis but also identifies potential anomalies by analyzing data trends, enabling preventative maintenance.

[0145] In the process of membrane thickness control, a dynamic coupling control strategy is adopted, taking into account the relationship between membrane weight, membrane thickness, and membrane density. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy as follows:

[0146] The following fundamental physical relationships exist between membrane weight (m), membrane thickness (h), and membrane density (ρ):

[0147] m=h·A·ρ

[0148] Where m is the membrane weight; h is the membrane thickness; A is the membrane surface area; and ρ is the membrane density.

[0149] During the membrane thickness control process, when the membrane density ρ or membrane weight m changes, the membrane thickness h is adjusted to maintain product quality. The adjustment amount is dynamically calculated using the following formula.

[0150] Define the change in membrane weight or membrane density as Δm(t) or Δρ(t). Based on the relationship between membrane weight and membrane thickness, calculate the adjustment amount for membrane thickness. The formula for the change in membrane weight is:

[0151] Δm(t)=m(t)-m target

[0152] Where, m target Target membrane weight;

[0153] Calculate the adjustment amount Δh(t) for membrane thickness based on the change in membrane weight:

[0154]

[0155] Change in membrane density:

[0156] Δρ(t)=ρ(t)-ρ target

[0157] Where, p target Target film density;

[0158] Calculate the adjustment amount Δh(t) of the membrane thickness based on the change in membrane density:

[0159]

[0160] When both membrane weight and membrane density change simultaneously, the combined adjustment amount for membrane thickness, Δh, is calculated by combining these two factors. total The formula for calculating (t) is:

[0161]

[0162] Based on the changing trend of membrane thickness, as well as the changes in membrane weight and membrane density, the control strategy is dynamically adjusted, and the control system generates a new control signal S(t) based on these changes.

[0163] The control signal S(t) is based on the adjustment amount Δh of the film thickness. total (t) and other relevant factors are used to generate a control signal, and a proportional relationship is established between the generated control signal and the film thickness adjustment amount. The control signal is expressed as:

[0164] S(t)=κ·Δh total (t)+λ·E(t)

[0165] Where κ is a proportionality constant; E(t) is the real-time state of the production equipment; and λ is the influence coefficient of the equipment state on the control signal.

[0166] The system transmits the control signal S(t) to the execution module. The execution module adjusts the production parameters according to the control signal. The adjusted production parameters affect the membrane thickness, membrane weight, and membrane density. The online detection module continuously monitors the membrane thickness parameter and feeds back the new detection data to the system.

[0167] This solution can respond in real time to changes in membrane weight or density, automatically adjusting membrane thickness to maintain membrane quality and consistency throughout the production process. By combining the physical relationship between membrane weight and density, the system achieves more precise membrane thickness control. Dynamically adjusting membrane thickness effectively optimizes membrane quality, ensuring that parameters such as thickness, weight, and density are always at their optimal levels. This directly impacts membrane lifespan and performance. The system employs closed-loop feedback control, continuously optimizing production parameters through real-time online detection and dynamic adjustment of control signals, ensuring production process stability and high-quality membrane products. This solution integrates automatic adjustment and intelligent control strategies, reducing manual intervention and improving the automation level of the production process. Through automated adjustment, it can quickly respond to changes in production, reducing production fluctuations and downtime, and improving overall production efficiency. This solution can optimize key parameters such as membrane thickness, weight, and density in real time, reducing waste in the production process, lowering energy consumption, and thus saving production costs. Precise membrane thickness control can reduce equipment wear caused by improper production parameters and extend equipment lifespan.

[0168] This embodiment also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a machine learning-based online thickness closed-loop control system and method for ceramic thin films as described above.

[0169] This embodiment also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the above-described machine learning-based online thickness closed-loop control system and method for ceramic thin films.

[0170] In the implementation of this invention, the machine learning model used can be a regression analysis model, a neural network model, or a support vector machine. By training a large amount of historical data, the model can predict the trend of membrane thickness change. Specific model training methods include, but are not limited to, supervised learning and unsupervised learning. The training process is based on different input features, such as membrane thickness, membrane weight, membrane density, and state parameters of production equipment.

[0171] To improve the prediction accuracy of machine learning models, real-time detection data undergoes preprocessing. This preprocessing includes denoising, standardization, normalization, and outlier detection. The purpose of preprocessing is to eliminate noisy data and ensure higher quality and consistency of the training data input to the machine learning model. Specifically, membrane thickness, membrane weight, and membrane density parameters are smoothed before being input into the model, and historical data is filled in using specific algorithms.

[0172] The dynamic coupling control strategy in this invention refers to the system calculating the amount of membrane thickness adjustment in real time based on the relationship between membrane weight, membrane thickness, and membrane density during the membrane thickness control process. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy according to the change law of membrane density and membrane weight. This strategy involves a cross-coupling algorithm of membrane weight, membrane thickness, and membrane density. Specifically, the membrane thickness is adjusted in real time through fuzzy control or other adaptive algorithms to ensure the stability of the production process.

[0173] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0174] Those skilled in the art will understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0175] The above embodiments of the present invention are not intended to limit the scope of protection of the present invention. The implementation of the present invention is not limited thereto. All other modifications, substitutions or alterations made to the above structure of the present invention based on the above content of the present invention, in accordance with ordinary technical knowledge and common practice in the field, without departing from the basic technical idea of ​​the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A closed-loop control system for online thickness of ceramic thin films based on machine learning, characterized in that, Including: The online detection module is used to monitor production parameters such as membrane thickness, membrane weight, and membrane density in real time. This module includes a membrane thickness sensor, a surface image recognition device, and a gas flow sensor monitoring device, which collects and feeds back data on parameters during the production process in real time. The data processing and AI diagnostic module is used to process and analyze the real-time data acquired by the online detection module. This module is based on machine learning algorithms and uses trained models to detect anomalies in the data, automatically identifying anomalies that will occur in the production process. The control decision and feedback module is used to calculate the adjustment amount based on the analysis results of the AI ​​diagnostic module and the deviation information of the film thickness, and generate the corresponding control signal and transmit the signal to the execution module. The execution module is used to control the adjustment of production equipment and optimize production conditions in real time according to the instructions provided by the control decision module.

2. A closed-loop online thickness control method for ceramic thin films based on machine learning, characterized in that, Includes the following steps: In the production process of ceramic thin films, online monitoring equipment using film thickness sensors and image recognition devices is used to obtain film thickness, film weight, and film density parameters in real time, and the detected data is transmitted to the data processing and AI diagnostic module for analysis. After preprocessing the real-time detection data, it is input into the machine learning model for analysis. The trained model is used to analyze the trend of film thickness change and automatically diagnose anomalies that occur during the production process, such as vibration marks, lines, and bubbles. Based on the AI ​​diagnostic results and film thickness deviation, the system calculates the adjustment amount, calculates the control parameters according to the film thickness change trend and the real-time status of the production equipment, and generates control signals. The control signal is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal. After the adjustment is completed, online monitoring is continuously performed. During the membrane thickness control process, a dynamic coupling control strategy is adopted by combining the relationship between membrane weight, membrane thickness, and membrane density. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy.

3. The online thickness closed-loop control method for ceramic thin films based on machine learning according to claim 2, characterized in that, In the production process of ceramic thin films, the method of acquiring real-time parameters such as film thickness, film weight, and film density through online monitoring equipment using film thickness sensors and image recognition devices, and then transmitting the detected data to a data processing and AI diagnostic module for analysis, is as follows: In the production process of ceramic thin films, film thickness sensors and image recognition equipment are used for online monitoring to detect and obtain production parameters such as film thickness, film weight, and film density in real time. The above detection data is transmitted in real time to the data processing and AI diagnosis module via the communication interface for data analysis and anomaly diagnosis. In the data processing and AI diagnostic module, a preset machine learning algorithm is used to analyze the collected detection data in real time, determine the quality of the film layer, identify impending film thickness deviations and surface defect anomalies, and generate control signals based on the diagnostic results.

4. The online thickness closed-loop control method for ceramic thin films based on machine learning according to claim 3, characterized in that, After preprocessing the real-time detection data, it is input into a machine learning model for analysis. The trained model is used to analyze the trend of film thickness change and automatically diagnose anomalies occurring during the production process, such as vibration marks, lines, and bubbles. The real-time detection data obtained from the online detection equipment is preprocessed. The preprocessing steps include noise reduction, normalization, missing value imputation, and outlier removal. The preprocessed data is input into the machine learning model, and the time series of film thickness, rate of change, and surface defect image features are extracted based on the film thickness change trend. The input data is analyzed using a trained machine learning model. This model learns from historical data to capture the patterns of membrane thickness changes and predicts the membrane thickness trend in the current production process. Based on the analysis results of the machine learning model, the system automatically diagnoses abnormalities that occur during the production process. If the film thickness changes abnormally, the model identifies defects such as vibration marks, lines, and bubbles, and prompts production personnel to make adjustments by outputting corresponding diagnostic information. Based on the abnormal information from the automatic diagnosis, the system further provides adjustment suggestions or automatically generates control signals, which are then fed back to the production equipment for real-time adjustments.

5. The online thickness closed-loop control method for ceramic thin films based on machine learning according to claim 4, characterized in that, Based on the AI ​​diagnostic results and film thickness deviation, the system calculates the adjustment amount. According to the film thickness change trend and the real-time status of the production equipment, the system calculates the control parameters and generates control signals using the following method: The film thickness deviation Δh(t) is the difference between the real-time measured film thickness h(t) and the target film thickness h. target The difference between them: Δh(t)=h(t)-h target Where h(t) is the film thickness at time t; h target The target film thickness is set. Film thickness variation trend The rate of change of film thickness with time is expressed as the derivative of the change in film thickness within the time window: Where Δt is the time step; h(t-Δt) is the film thickness at the previous time step t-Δt; The AI ​​diagnostic model analyzes the trend of film thickness variation and other parameters to generate an adjustment factor A(t), which is used to correct film thickness deviations and optimize the adjustment of production equipment. Where f is an AI diagnostic function based on machine learning, and outputs an adjustment factor A(t); D(t) is a parameter related to the real-time status of the production equipment; The system calculates the film thickness adjustment amount ΔH based on the film thickness deviation Δh(t) and the AI ​​diagnostic adjustment factor A(t). adjust (t), used to control the adjustment of production equipment: ΔH adjust (t)=α·Δh(t)+β·A(t) Where α and β are adjustment coefficients; Adjustment amount ΔH based on membrane thickness adjust Given E(t) and the real-time status of the production equipment, the system calculates the adjusted control parameter C(t): C(t)=γ·ΔH adjust (t)+·E(t) Where γ and E(t) are adjustment coefficients; E(t) is the real-time state of the equipment. The control signal S(t) is generated based on the control parameter C(t) and is used to adjust the production equipment. S(t)=κ·C(t) Where κ is a proportionality constant.

6. The online thickness closed-loop control method for ceramic thin films based on machine learning according to claim 5, characterized in that, The control signal is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal. After the adjustment is completed, the method for continuous online monitoring is as follows: The control signal S(t) is transmitted to the execution module through the feedback module. The execution module adjusts the production parameters according to the control signal, and the adjustment amount is ΔP. adjust (t) represents the adjustment of the production parameter P(t) by the execution module: ΔP adjust (t)=η·S(t) Where η is a proportional coefficient; S(t) is the control signal output from the control decision module; The execution module adjusts the production parameter P(t) according to the control signal S(t) to obtain the new production parameter P. new (t): P new (t)=P(t)+ΔP adjust (t)=P(t)+η·S(t) Adjusted production parameters P new (t) Production parameters affecting membrane thickness: The online detection module will continue to monitor membrane thickness, membrane weight, and membrane density parameters, and acquire new detection data D. new (t): D new (t)=OnlineDetection(P new (t)) Among them, D new (t) is based on the new production parameter P new (t) New detection data obtained after online detection; New test data D new (t) will be transmitted to the data processing and AI diagnostic module for real-time analysis and anomaly diagnosis. This module will analyze the new data and output new adjustment suggestions and control signals. After the system completes the adjustment through the execution module, the new production parameter P new (t) affects the membrane thickness parameter, and new data is obtained through the online detection system. The data is input to the control decision module to form a new feedback loop.

7. The online thickness closed-loop control method for ceramic thin films based on machine learning according to claim 6, characterized in that, In the process of membrane thickness control, a dynamic coupling control strategy is adopted, taking into account the relationship between membrane weight, membrane thickness, and membrane density. When the membrane weight or membrane density changes, the system automatically adjusts the membrane thickness control strategy as follows: The following fundamental physical relationships exist between membrane weight (m), membrane thickness (h), and membrane density (ρ): m=h·A·ρ Where n is the membrane weight; h is the membrane thickness; A is the membrane surface area; and ρ is the membrane density. During the membrane thickness control process, when the membrane density ρ or membrane weight m changes, the membrane thickness h is adjusted to maintain product quality. The adjustment amount is dynamically calculated using the following formula. Define the change in membrane weight or membrane density as Δm(t) or Δρ(t). Based on the relationship between membrane weight and membrane thickness, calculate the adjustment amount for membrane thickness. The formula for the change in membrane weight is: Δm(t)=m(t)-m target Where, m target Target membrane weight; Calculate the adjustment amount Δh(t) for membrane thickness based on the change in membrane weight: Change in membrane density: Δρ(t)=ρ(t)-ρ target Where, ρ target Target film density; Calculate the adjustment amount Δh(t) of the membrane thickness based on the change in membrane density: When both membrane weight and membrane density change simultaneously, the combined adjustment amount for membrane thickness, Δh, is calculated by combining these two factors. total The formula for calculating (t) is: Based on the changing trend of membrane thickness, as well as the changes in membrane weight and membrane density, the control strategy is dynamically adjusted, and the control system generates a new control signal S(t) based on these changes. The control signal S(t) is based on the adjustment amount Δh of the film thickness. total (t) and other relevant factors are used to generate a control signal, and a proportional relationship is established between the generated control signal and the film thickness adjustment amount. The control signal is expressed as: S(t)=κ·Δh total (t)+λ·E(t) Where κ is a proportionality constant; E(t) is the real-time state of the production equipment; and θ is the influence coefficient of the equipment state on the control signal. The system transmits the control signal S(t) to the execution module. The execution module adjusts the production parameters according to the control signal. The adjusted production parameters affect the membrane thickness, membrane weight, and membrane density. The online detection module continuously monitors the membrane thickness parameter and feeds back the new detection data to the system.

8. An electronic device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements a machine learning-based online thickness closed-loop control method for ceramic thin films as described in any one of claims 2-7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements a machine learning-based online thickness closed-loop control method for ceramic thin films as described in any one of claims 2-7.

Images