Intelligent dynamic optimization method and system for printed circuit board production parameters

Abstract

The present invention discloses a method and system for intelligent dynamic optimization of printed circuit board production parameters, relating to the field of circuit board processing optimization technology. The method comprises: real-time monitoring of real-time temperature data of the lamination site; processing the real-time temperature data to identify temperature-instable areas during the production process; selecting a hot press roller process section with a stable temperature distribution to perform the lamination operation; assessing the risk level of the current lamination process; selecting a low-risk temperature distribution pattern based on the risk level for lamination production, obtaining a quality deviation coefficient, and determining whether the lamination process meets the preset process standards based on the quality deviation coefficient; and for products that do not meet the preset process standards, tracing the temperature control parameters during the production process and dynamically adjusting the temperature set point of the hot press roller. The advantages of the present invention are: it enables real-time quantitative assessment of the wear status of the hot press roller, significantly improving the line width precision control capability and interlayer alignment consistency.

Description

Technical Field

[0001] The present invention relates to the technical field of circuit board processing optimization technology, and in particular to a method and system for intelligent dynamic optimization of printed circuit board production parameters. Background Art

[0002] The lamination process for printed circuit boards (PCBs) is a critical step in the manufacture of multi-layer circuit boards. Its core lies in applying temperature and pressure to the substrate through hot-press rollers to achieve precise bonding between the layers. In traditional processes, temperature control relies heavily on fixed parameter settings and manual adjustments based on experience, making it difficult to respond in real time to dynamic temperature changes in the contact area between the hot-press roller and the substrate. Especially during long-term operation, problems such as surface wear and uneven heat transfer from the hot-press roller can easily lead to local temperature drift, which in turn can cause defects such as line width deviation and interlayer misalignment. While existing technologies use temperature sensors for monitoring, they lack a comprehensive analysis of the frequency and amplitude of temperature fluctuations and the risk of wear. This results in delayed adjustment of process parameters and makes it difficult to effectively suppress quality fluctuations.

[0003] The growing demand for high-density interconnected circuit boards (HDIs) places higher demands on the precision and stability of the lamination process. Current methods have significant shortcomings in identifying abnormal temperatures, predicting risks, and dynamically compensating for them. Traditional frequency-domain analysis struggles to capture transient temperature anomalies and accurately distinguish normal process fluctuations from equipment degradation signals. Furthermore, quality inspection and parameter optimization are disconnected, and adjustment strategies rely on trial and error, making closed-loop control difficult to achieve. Therefore, an intelligent control system that integrates multi-source sensor data and offers real-time anomaly diagnosis and adaptive optimization capabilities is urgently needed to address the impact of complex coupled factors such as uneven temperature distribution and hot press roller wear on product quality. Summary of the Invention

[0004] In order to solve the above technical problems, a method and system for intelligent dynamic optimization of printed circuit board production parameters are provided. This technical solution solves the problem that although the above-mentioned existing technology uses temperature sensors for monitoring, it lacks a comprehensive analysis of the frequency, amplitude and wear risk of temperature fluctuations, resulting in delayed adjustment of process parameters and difficulty in effectively suppressing quality fluctuations.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is:

[0006] A method for intelligent dynamic optimization of printed circuit board production parameters, comprising:

[0007] During the lamination production process of printed circuit boards, the temperature data of the lamination part is monitored in real time by the temperature sensor installed in the contact area between the hot pressing roller and the substrate;

[0008] Process real-time temperature data, analyze the frequency and amplitude changes of temperature fluctuations, calculate the temperature anomaly coefficient based on the comprehensive fluctuation trend of frequency and amplitude, and identify temperature unstable areas in the production process;

[0009] According to the evaluation results of the temperature anomaly coefficient, the hot pressing roller process section with stable temperature distribution is selected to perform the lamination operation;

[0010] Optical sensors collect surface wear data from the hot pressing rollers used in lamination operations, analyze the risk of uneven temperature distribution caused by wear, and assess the risk level of the current lamination process.

[0011] Based on the risk level, low-risk temperature distribution patterns are screened for lamination production. After completion, the line width accuracy and inter-layer alignment of the circuit board are tested to obtain the quality deviation coefficient. Based on the quality deviation coefficient, it is determined whether it meets the preset process standards;

[0012] For products that do not meet the preset process standards, the temperature control parameters in the production process are traced and the temperature setting value of the hot pressing roller is dynamically adjusted until the quality deviation coefficient is lower than the preset threshold.

[0013] Preferably, the processing of real-time temperature data, analyzing the frequency and amplitude changes of temperature fluctuations, calculating the temperature anomaly coefficient based on the comprehensive fluctuation trend of frequency and amplitude, and identifying the temperature unstable area in the production process specifically includes:

[0014] Perform multi-scale decomposition on the frequency of temperature fluctuations and extract the energy proportion of the abnormal frequency band as the frequency anomaly index;

[0015] Perform frequency domain energy analysis on the amplitude of temperature fluctuations and calculate the energy ratio of abnormal amplitude as the amplitude anomaly index;

[0016] The frequency anomaly index and the amplitude anomaly index are normalized and weighted to generate a temperature anomaly coefficient that comprehensively reflects the temperature fluctuation characteristics;

[0017] If the temperature anomaly coefficient exceeds a preset threshold, the process section of the hot pressing roller is determined to be a temperature unstable area.

[0018] Preferably, performing multi-scale decomposition on the frequency of temperature fluctuations and extracting the energy proportion of the abnormal frequency band as the frequency anomaly index specifically includes:

[0019] Perform multi-level wavelet decomposition on the temperature signal to separate the high-frequency sudden change component and the low-frequency slowly changing component;

[0020] Extract the energy value of the high-frequency component and calculate the proportion of the energy value of the high-frequency component in the total signal energy. If the proportion exceeds the preset range, it will be marked as abnormal frequency fluctuation;

[0021] Specifically, the calculation formula for the proportion of the energy value of the high-frequency component in the total signal energy is:

[0022] ;

[0023] Where, is the proportion of the energy value of the high-frequency component in the total signal energy, is the approximate coefficient in wavelet decomposition, reflecting the low-frequency slowly varying component of the temperature signal. is the detail coefficient in wavelet decomposition, reflecting the high-frequency mutation component of the temperature signal. is the abnormal frequency band index set, is the number of decomposition levels of the approximate coefficients, is the number of decomposition levels of detail coefficients.

[0024] Preferably, performing frequency domain energy analysis on the amplitude of temperature fluctuation and calculating the abnormal amplitude energy ratio as the amplitude abnormality index specifically includes:

[0025] Convert the temperature signal to the frequency domain and divide the normal process bandwidth in the frequency domain ;

[0026] Calculate the signal energy in the two frequency bands respectively, and use the ratio of the abnormal frequency band energy to the total energy as the amplitude anomaly index;

[0027] Specifically, the calculation formula for the amplitude anomaly index is:

[0028] ;

[0029] is the amplitude anomaly index, which indicates the energy ratio within the abnormal frequency range. is the power spectrum density function, which reflects the energy distribution of the temperature signal in the frequency domain. For the normal process frequency range, is half the signal sampling frequency.

[0030] Preferably, collecting surface wear data of the hot pressing roller during the lamination operation by an optical sensor, analyzing the risk of uneven temperature distribution caused by wear, and evaluating the risk level of the current lamination process specifically include:

[0031] A convolutional neural network is used to automatically extract the deep features of the hot pressing roller surface morphology from historical production data, build a sample library containing normal wear and abnormal wear, and annotate the corresponding risk level labels;

[0032] Training risk prediction models based on sample libraries;

[0033] Extract features from the surface topography images collected by the optical sensor, including the area of ​​the worn area and edge sharpness;

[0034] The extracted features are input into the pre-trained risk prediction model, and an index value representing the degree of wear risk is output. When the index exceeds the threshold, it is judged as high risk.

[0035] Preferably, low-risk temperature distribution patterns are screened based on risk levels for lamination production. After completion, the line width accuracy and interlayer alignment of the circuit board are tested to obtain the quality deviation coefficient. Based on the quality deviation coefficient, it is determined whether the preset process standards are met. Specifically, the following steps are included:

[0036] Measure the line width deviation and inter-layer alignment offset at key locations on the PCB, and calculate their relative errors to the target values.

[0037] The errors of multiple detection points are weighted and summed to generate the quality deviation coefficient;

[0038] Determine whether the quality deviation coefficient exceeds the deviation threshold, if so, it is determined that it does not meet the preset process standards, if not, it is determined that it meets the preset process standards;

[0039] Specifically, the calculation formula of the mass deviation coefficient is:

[0040] ;

[0041] Where, is the quality deviation coefficient, which comprehensively reflects the process deviation of line width and alignment. is the measured line width of the i-th detection point, is the target line width, is the measured coordinate of the i-th detection point, Align the coordinates for the target, Process tolerance, is the total number of detection points, for and The distance between them.

[0042] Preferably, for products that do not meet the preset process standards, tracing back the temperature control parameters in the production process and dynamically adjusting the temperature setting value of the hot pressing roller until the quality deviation coefficient is lower than the preset threshold specifically includes:

[0043] When the mass deviation coefficient is continuously high, the hot pressing temperature setting value is adaptively adjusted according to the deviation direction, and the temperature is increased to compensate for the decrease in heat transfer efficiency caused by wear;

[0044] The adjustment range is associated with the current deviation coefficient, historical adjustment records and process constraints, and stable optimization is achieved through fuzzy logic rules.

[0045] Furthermore, this solution proposes a system for intelligent dynamic optimization of printed circuit board production parameters, which is used to implement the above-mentioned intelligent dynamic optimization method for printed circuit board production parameters, specifically including:

[0046] The temperature monitoring module includes a distributed temperature sensor array installed in the contact area between the hot pressing roller and the substrate, which is used to collect temperature distribution data of the lamination part in real time;

[0047] A data processing module, which is in communication with the temperature monitoring module and includes a multi-channel signal acquisition unit and a real-time storage unit, and is used to perform noise reduction processing and time series alignment on the temperature data;

[0048] A temperature analysis module is communicatively connected to the data processing module, and is configured to: perform wavelet decomposition on the temperature fluctuation signal to extract high-frequency mutation components, calculate the energy proportion of the abnormal frequency band as a frequency anomaly index; generate a power spectrum density distribution through FFT conversion, calculate the energy proportion of the abnormal amplitude as an amplitude anomaly index; and normalize the frequency anomaly index and the amplitude anomaly index to generate a temperature anomaly coefficient;

[0049] The process section screening module is connected to the output end of the temperature analysis module and is configured to dynamically generate a stable process section mapping map of the hot pressing roller based on the temperature anomaly coefficient threshold comparison result;

[0050] The surface detection module, which includes a high-resolution optical sensor and a coaxial light source assembly, is located above the lamination station to capture the surface topography of the hot pressing roller and extract wear characteristic parameters;

[0051] A risk assessment engine, the risk assessment engine being in communication with the surface detection module, the risk assessment engine being configured to: analyze surface topography images using a pre-trained deep feature extraction network to output a wear risk index; and generate a risk heat map of the hot pressing area based on risk level classification rules;

[0052] The quality inspection unit includes a laser microscopy measurement device and an image registration module, which is used to quantify the line width deviation and inter-layer alignment offset of the circuit board, and generate the quality deviation coefficient through weighted error calculation;

[0053] The dynamic control module interacts with the hot pressing roller temperature control system in real time. The dynamic control module is configured to: start the temperature setting value backtracking analysis function according to the degree of exceeding the quality deviation coefficient; generate temperature compensation parameters through the genetic algorithm optimizer, and send PID control instructions to the actuator.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] The present invention collects temperature distribution data in the contact area between the hot pressing roller and the substrate in real time, and combines multi-scale wavelet decomposition and frequency domain energy analysis technology to accurately identify abnormal frequency and amplitude characteristics of temperature fluctuations, significantly improving the positioning accuracy of temperature unstable areas and effectively avoiding the misjudgment problem caused by the traditional threshold alarm method that ignores dynamic fluctuation characteristics. By integrating optical surface morphology detection and convolutional neural network risk prediction models, real-time quantitative assessment of the wear state of the hot pressing roller is achieved, solving the problem of the coupling effect between uneven temperature distribution and equipment degradation. At the same time, the fuzzy logic dynamic compensation mechanism based on quality deviation coefficient feedback greatly shortens the process parameter adaptive adjustment cycle, significantly improves the line width precision control capability and inter-layer alignment consistency, and systematically optimizes the production yield and process stability of high-density circuit boards. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Flowchart of the intelligent dynamic optimization method for printed circuit board production parameters proposed in this solution;

[0057] Figure 2 Flowchart of the method proposed for identifying temperature unstable areas in the production process;

[0058] Figure 3 Flowchart of the method proposed for this solution to assess the risk level of the current lamination process;

[0059] Figure 4 This is a flow chart of the method proposed in this plan to determine whether the preset process standards are met based on the quality deviation coefficient;

[0060] Figure 5 This is a flow chart of the method for dynamically adjusting the temperature setting value of the hot pressing roller proposed in this solution. DETAILED DESCRIPTION

[0061] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0062] Reference Figure 1 As shown, a method for intelligent dynamic optimization of printed circuit board production parameters includes:

[0063] During the lamination production process of printed circuit boards, the temperature data of the lamination part is monitored in real time by the temperature sensor installed in the contact area between the hot pressing roller and the substrate;

[0064] By deploying temperature sensors in the contact area between the hot pressing roller and the substrate and monitoring the temperature distribution in real time, we achieve comprehensive awareness of the dynamic changes in the thermal field during the lamination process, overcoming the limitations of traditional manual sampling or single-point monitoring. Combined with the real-time storage and analysis capabilities of the data processing unit, we can quickly capture abnormal temperature signals, providing highly timely data support for subsequent precise control, significantly improving the controllability of the process.

[0065] Process real-time temperature data, analyze the frequency and amplitude changes of temperature fluctuations, calculate the temperature anomaly coefficient based on the comprehensive fluctuation trend of frequency and amplitude, and identify temperature unstable areas in the production process;

[0066] The temperature anomaly coefficient is calculated based on the frequency and amplitude variation characteristics of temperature fluctuations. This breaks through the traditional static judgment mode of single threshold alarm and can identify the root cause of temperature instability from the perspective of joint analysis in the time and frequency domains. By quantifying the comprehensive fluctuation trend, this method effectively distinguishes between normal process fluctuations and abnormal equipment conditions, significantly reducing the misjudgment rate and providing a scientific basis for process stage screening.

[0067] According to the evaluation results of the temperature anomaly coefficient, the hot pressing roller process section with stable temperature distribution is selected to perform the lamination operation;

[0068] By screening stable process sections based on the temperature anomaly coefficient, we can dynamically avoid areas prone to defects caused by localized hot pressing roller aging or uneven heat transfer, ensuring that the lamination operation is always carried out under optimal thermal conditions. This active avoidance mechanism significantly improves product yield while reducing material waste and equipment loss caused by temperature runaway.

[0069] Optical sensors collect surface wear data from the hot pressing rollers used in lamination operations, analyze the risk of uneven temperature distribution caused by wear, and assess the risk level of the current lamination process.

[0070] By quantifying the surface wear of hot-pressing rollers using optical sensors and combining them with a risk assessment model, we can proactively predict the risk of abnormal temperature distribution caused by equipment degradation. This method models the relationship between surface topography degradation and thermal conductivity, enabling a shift from "reactive maintenance" to "predictive maintenance," effectively preventing process parameter drift caused by wear.

[0071] Based on the risk level, low-risk temperature distribution patterns are screened for lamination production. After completion, the line width accuracy and inter-layer alignment of the circuit board are tested to obtain the quality deviation coefficient. Based on the quality deviation coefficient, it is determined whether it meets the preset process standards;

[0072] Low-risk temperature distribution patterns are screened based on risk levels, and process standard compliance is verified using quality deviation coefficients, forming a closed-loop control chain of "risk prediction-process execution-quality verification." This approach uses a multi-dimensional feedback mechanism to ensure that the lamination process always approaches the optimal state during dynamic changes, systematically improving product line width accuracy and inter-layer alignment consistency.

[0073] For products that do not meet the preset process standards, the temperature control parameters in the production process are traced and the temperature setting value of the hot pressing roller is dynamically adjusted until the quality deviation coefficient is lower than the preset threshold.

[0074] For products that don't meet quality standards, we backtrack temperature parameters and dynamically adjust hot pressing settings, establishing a reverse optimization path of "detection-traceability-compensation." This mechanism uses fuzzy logic and adaptive algorithms to intelligently correct parameters, avoiding the efficiency bottlenecks of traditional trial-and-error adjustments. Ultimately, it fosters a self-healing process parameter capability, ensuring continuous stability in the production process and iterative optimization of product quality.

[0075] Reference Figure 2 As shown, the real-time temperature data is processed to analyze the frequency and amplitude changes of temperature fluctuations. The temperature anomaly coefficient is calculated based on the comprehensive fluctuation trend of frequency and amplitude. The temperature unstable areas in the production process are identified. Specifically, the following are included:

[0076] Perform multi-scale decomposition on the frequency of temperature fluctuations and extract the energy proportion of the abnormal frequency band as the frequency anomaly index;

[0077] Specifically, the frequency of temperature fluctuations is decomposed at multiple scales, and the energy proportion of the abnormal frequency band is extracted as the frequency anomaly index. Specifically, the following steps are performed:

[0078] Perform multi-level wavelet decomposition on the temperature signal to separate the high-frequency sudden change component and the low-frequency slowly changing component;

[0079] Extract the energy value of the high-frequency component and calculate the proportion of the energy value of the high-frequency component in the total signal energy. If the proportion exceeds the preset range, it will be marked as abnormal frequency fluctuation;

[0080] Specifically, the calculation formula for the proportion of the energy value of the high-frequency component in the total signal energy is:

[0081] ;

[0082] Where, is the proportion of the energy value of the high-frequency component in the total signal energy, is the approximate coefficient in wavelet decomposition, reflecting the low-frequency slowly varying component of the temperature signal. is the detail coefficient in wavelet decomposition, reflecting the high-frequency mutation component of the temperature signal. is the abnormal frequency band index set, is the number of decomposition levels of the approximate coefficients, is the number of decomposition layers of detail coefficients;

[0083] Perform frequency domain energy analysis on the amplitude of temperature fluctuations and calculate the energy ratio of abnormal amplitude as the amplitude anomaly index;

[0084] Specifically, frequency domain energy analysis is performed on the amplitude of temperature fluctuations, and the abnormal amplitude energy ratio is calculated as the amplitude abnormality index. Specifically, the following are included:

[0085] Convert the temperature signal to the frequency domain and divide the normal process bandwidth in the frequency domain ;

[0086] Calculate the signal energy in the two frequency bands respectively, and use the ratio of the abnormal frequency band energy to the total energy as the amplitude anomaly index;

[0087] Specifically, the calculation formula for the amplitude anomaly index is:

[0088] ;

[0089] is the amplitude anomaly index, which indicates the energy ratio within the abnormal frequency range. is the power spectrum density function, which reflects the energy distribution of the temperature signal in the frequency domain. For the normal process frequency range, is half of the signal sampling frequency;

[0090] The frequency anomaly index and the amplitude anomaly index are normalized and weighted to generate a temperature anomaly coefficient that comprehensively reflects the temperature fluctuation characteristics;

[0091] The specific calculation formula of the temperature anomaly coefficient is:

[0092] ;

[0093] Where, is the temperature anomaly coefficient, α and β are weight factors, which are used to adjust the contribution ratio of the frequency anomaly index and the amplitude anomaly index in the anomaly coefficient, and α+β=1;

[0094] If the temperature anomaly coefficient exceeds a preset threshold, the process section of the hot pressing roller is determined to be a temperature unstable area.

[0095] Through the synergistic effect of multi-scale wavelet decomposition and frequency-domain energy analysis, the precise separation and quantitative characterization of high-frequency transient anomalies and low-frequency slow-changing anomalies in temperature fluctuations have been achieved for the first time. By calculating the frequency anomaly index, it is possible to effectively capture sudden temperature changes caused by mechanical vibration of the hot press roller and uneven material contact, thus avoiding the sensitivity of traditional time-domain analysis to high-frequency noise. At the same time, based on the calculation of the frequency-domain amplitude anomaly index, it can reveal the persistent temperature drift caused by the degradation of the heat transfer performance of the hot press roller. The two are then normalized and weighted to generate a temperature anomaly coefficient, which breaks through the limitations of single-dimensional anomaly detection and significantly improves the accuracy of identifying temperature-unstable areas. This technology not only provides early warning of hidden faults such as local wear of the hot press roller and failure of the heat transfer medium, but also locates the source of process interference through energy tracing of abnormal frequency bands, providing a multi-dimensional decision-making basis for the dynamic optimization of process parameters, thereby systematically reducing the risk of lamination defects caused by uncontrolled temperature distribution.

[0096] Reference Figure 3 As shown in the figure, optical sensors are used to collect surface wear data of the hot pressing rollers performing the lamination operation, analyze the risk of uneven temperature distribution caused by wear, and assess the risk level of the current lamination process. Specifically, the following are included:

[0097] A convolutional neural network is used to automatically extract the deep features of the hot pressing roller surface morphology from historical production data, build a sample library containing normal wear and abnormal wear, and annotate the corresponding risk level labels;

[0098] The risk prediction model is trained based on the sample library, and the risk prediction model corresponds to the temperature distribution pattern one by one;

[0099] Extract features from the surface topography images collected by the optical sensor, including the area of ​​the worn area and edge sharpness;

[0100] The extracted features are input into the pre-trained risk prediction model, and an index value representing the degree of wear risk is output. When the index exceeds the threshold, it is judged as high risk.

[0101] Specifically, a feasible risk prediction model is:

[0102] ;

[0103] Where, is the wear risk index, is the proportion of the wear area, is the edge sharpness, is a bias term used to adjust the output baseline of the model, is the Sigmoid function, which maps the linear combination result to the probability interval of 0~1. Train the weights for the model, is a natural constant.

[0104] By integrating optical sensing technology with deep learning models, an innovative quantitative assessment system for hot pressing roller wear risk has been constructed. Using a convolutional neural network, this system automatically extracts deep features of the surface topography, transcending the traditional reliance on manually set thresholds. This system enables correlation analysis of multi-dimensional degradation features, such as wear area and edge sharpness, accurately capturing early signs of degradation, such as oxide layer flaking and microcrack propagation on the hot pressing roller surface. A wear risk index, output by a sigmoid function mapping, establishes a nonlinear mapping relationship between complex topographical features and temperature distribution risk. This not only identifies the immediate impact of the current wear state on thermal field uniformity but also predicts the remaining life within the future process window. This technology represents a leap from "image grayscale analysis" to "risk probability prediction," upgrading equipment maintenance strategies from scheduled inspections to predictive intervention. This effectively prevents wear-induced lamination temperature drift. Furthermore, a risk-based screening mechanism reduces ineffective downtime, ensuring stable and controllable equipment status for high-precision lamination processes.

[0105] Reference Figure 4 As shown, based on the risk level, low-risk temperature distribution patterns are screened for lamination production. After completion, the line width accuracy and inter-layer alignment of the circuit board are tested to obtain the quality deviation coefficient. Based on the quality deviation coefficient, it is determined whether the preset process standards are met. Specifically, the following are included:

[0106] Measure the line width deviation and inter-layer alignment offset at key locations on the PCB, and calculate their relative errors to the target values.

[0107] The errors of multiple detection points are weighted and summed to generate the quality deviation coefficient;

[0108] Determine whether the quality deviation coefficient exceeds the deviation threshold, if so, it is determined that it does not meet the preset process standards, if not, it is determined that it meets the preset process standards;

[0109] Specifically, the calculation formula of the mass deviation coefficient is:

[0110] ;

[0111] Where, is the quality deviation coefficient, which comprehensively reflects the process deviation of line width and alignment. is the measured line width of the i-th detection point, is the target line width, is the measured coordinate of the i-th detection point, Align the coordinates for the target, Process tolerance, is the total number of detection points, for and The distance between them.

[0112] By constructing a quality deviation coefficient that comprehensively considers line width accuracy and inter-layer alignment, the limitations of traditional single-index judgment have been overcome, and a global quantitative assessment of the quality of the lamination process has been achieved. By weightedly fusing the line width relative error and coordinate offset distance of multiple detection points, it not only reflects the process deviation details of local key areas, but also takes into account the cumulative deformation effect of the entire circuit board, significantly improving the comprehensiveness and objectivity of quality inspection. The introduction of a process tolerance adaptive adjustment mechanism in the formula enables the quality judgment standard to dynamically match different design specification requirements, avoiding the problem of over-inspection or missed inspection caused by traditional fixed thresholds. This technology converts discrete process defect characteristics into continuously traceable quality deviation coefficients, which not only provides accurate feedback for process parameter optimization, but also locates sensitive areas of hot pressing temperature field distortion through deviation tracing, forming a closed-loop response link of "quality anomaly-temperature control", and ultimately achieving a coordinated improvement in the linear accuracy and stacking structure stability of high-density circuit boards.

[0113] Reference Figure 5 As shown in the figure, for products that do not meet the preset process standards, the temperature control parameters in the production process are traced back and the temperature setting value of the hot pressing roller is dynamically adjusted until the quality deviation coefficient is lower than the preset threshold. Specifically, it includes:

[0114] When the mass deviation coefficient is continuously high, the hot pressing temperature setting value is adaptively adjusted according to the deviation direction, and the temperature is increased to compensate for the decrease in heat transfer efficiency caused by wear;

[0115] The adjustment range is associated with the current deviation coefficient, historical adjustment records and process constraints, and stable optimization is achieved through fuzzy logic rules;

[0116] Specifically, the feedback optimization model is:

[0117] ;

[0118] Where, is the temperature adjustment value of the kth control cycle, is the proportionality coefficient, is the integration coefficient, is the differential coefficient, is the quality deviation coefficient of the kth control cycle, is the time interval of the control cycle, where , , According to the real-time quality deviation coefficient adjustment, the adjustment strategy is stored in the knowledge base, and the adjustment strategy is optimized through self-learning based on the knowledge base.

[0119] By constructing a feedback optimization model that integrates fuzzy logic and self-learning mechanisms, intelligent closed-loop control of process parameter adjustment is achieved. The PID parameter weights are dynamically adjusted based on the quality deviation coefficient, so that the temperature compensation amount accurately matches the current degree of heat transfer efficiency attenuation. This can not only quickly respond to sudden process deviations, but also suppress the overshoot oscillation phenomenon in traditional PID control. By storing historical adjustment strategies in the knowledge base and continuously optimizing them, the system can adapt to changes in the characteristics of hot pressing rollers at different stages of wear, breaking through the limitations of fixed parameter control in adaptability to equipment aging conditions. This technology embeds process tolerance constraints into the adjustment logic to ensure that parameter corrections are always within the safe operating boundaries, forming a full-process self-optimization capability of "deviation traceability-parameter compensation-strategy iteration", significantly improving the stability of temperature control under complex working conditions, and ultimately achieving dual optimization of reduced scrap rate and good product consistency.

[0120] Furthermore, based on the same inventive concept as the above-mentioned method for intelligent dynamic optimization of printed circuit board production parameters, this solution also proposes an intelligent dynamic optimization system for printed circuit board production parameters, comprising:

[0121] The data processing module is connected to the temperature monitoring module and includes a multi-channel signal acquisition unit and a real-time storage unit, and is used to perform noise reduction and time series alignment on the temperature data;

[0122] The temperature analysis module is in communication with the data processing module and is configured to: perform wavelet decomposition on the temperature fluctuation signal to extract the high-frequency mutation component, calculate the energy ratio of the abnormal frequency band as the frequency anomaly index; generate the power spectrum density distribution through FFT conversion, calculate the energy ratio of the abnormal amplitude as the amplitude anomaly index; and normalize the frequency anomaly index and the amplitude anomaly index to generate the temperature anomaly coefficient;

[0123] The process section screening module is connected to the output end of the temperature analysis module and is configured to dynamically generate a stable process section mapping map of the hot pressing roller based on the temperature anomaly coefficient threshold comparison result;

[0124] The surface detection module, which includes a high-resolution optical sensor and a coaxial light source assembly, is located above the lamination station to capture the surface topography of the hot pressing roller and extract wear characteristic parameters;

[0125] The risk assessment engine is in communication with the surface detection module. The risk assessment engine is configured to: analyze the surface topography image through a pre-trained deep feature extraction network and output a wear risk index; and generate a risk heat map of the hot pressing area based on risk level classification rules;

[0126] The quality inspection unit includes a laser microscopy measurement device and an image registration module, which is used to quantify the line width deviation and inter-layer alignment offset of the circuit board, and generate the quality deviation coefficient through weighted error calculation;

[0127] The dynamic control module interacts with the hot pressing roller temperature control system in real time. The dynamic control module is configured to: start the temperature setting value backtracking analysis function according to the degree of exceeding the quality deviation coefficient; generate temperature compensation parameters through the genetic algorithm optimizer, and send PID control instructions to the actuator.

[0128] In summary, the advantages of the present invention are: by collecting temperature distribution data of the contact area between the hot pressing roller and the substrate in real time, combined with multi-scale wavelet decomposition and frequency domain energy analysis technology, the frequency and amplitude abnormal characteristics of temperature fluctuations can be accurately identified, which significantly improves the positioning accuracy of the temperature unstable area and effectively avoids the misjudgment problem caused by ignoring the dynamic fluctuation characteristics of the traditional threshold alarm method. By integrating optical surface morphology detection and convolutional neural network risk prediction model, real-time quantitative evaluation of the wear state of the hot pressing roller is realized, solving the problem of the coupling effect between uneven temperature distribution and equipment degradation; at the same time, the fuzzy logic dynamic compensation mechanism based on quality deviation coefficient feedback greatly shortens the process parameter adaptive adjustment cycle, significantly improves the line width precision control capability and inter-layer alignment consistency, and systematically optimizes the production yield and process stability of high-density circuit boards.

[0129] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent dynamic optimization of printed circuit board production parameters, characterized in that: include: During the lamination production process of printed circuit boards, the temperature data of the lamination part is monitored in real time by the temperature sensor installed in the contact area between the hot pressing roller and the substrate; Process real-time temperature data, analyze the frequency and amplitude changes of temperature fluctuations, calculate the temperature anomaly coefficient based on the comprehensive fluctuation trend of frequency and amplitude, and identify temperature unstable areas in the production process; According to the evaluation results of the temperature anomaly coefficient, the hot pressing roller process section with stable temperature distribution is selected to perform the lamination operation; Optical sensors collect surface wear data from the hot pressing rollers used in lamination operations, analyze the risk of uneven temperature distribution caused by wear, and assess the risk level of the current lamination process. Based on the risk level, low-risk temperature distribution patterns are screened for lamination production. After completion, the line width accuracy and inter-layer alignment of the circuit board are tested to obtain the quality deviation coefficient. Based on the quality deviation coefficient, it is determined whether it meets the preset process standards; For products that do not meet the preset process standards, the temperature control parameters in the production process are traced and the temperature setting value of the hot pressing roller is dynamically adjusted until the quality deviation coefficient is lower than the preset threshold; The processing of real-time temperature data, analyzing the frequency and amplitude changes of temperature fluctuations, calculating the temperature anomaly coefficient based on the comprehensive fluctuation trend of frequency and amplitude, and identifying temperature unstable areas in the production process specifically include: Perform multi-scale decomposition on the frequency of temperature fluctuations and extract the energy proportion of the abnormal frequency band as the frequency anomaly index; Perform frequency domain energy analysis on the amplitude of temperature fluctuations and calculate the energy ratio of abnormal amplitude as the amplitude anomaly index; The frequency anomaly index and the amplitude anomaly index are normalized and weighted to generate a temperature anomaly coefficient that comprehensively reflects the temperature fluctuation characteristics; If the temperature anomaly coefficient exceeds a preset threshold, the process section of the hot pressing roller is determined to be a temperature unstable area.

2. The method for intelligent dynamic optimization of printed circuit board production parameters according to claim 1, characterized in that: The multi-scale decomposition of the temperature fluctuation frequency and the extraction of the energy proportion of the abnormal frequency band as the frequency anomaly index specifically include: Perform multi-level wavelet decomposition on the temperature signal to separate the high-frequency sudden change component and the low-frequency slowly changing component; Extract the energy value of the high-frequency component and calculate the proportion of the energy value of the high-frequency component in the total signal energy. If the proportion exceeds the preset range, it will be marked as abnormal frequency fluctuation; Specifically, the calculation formula for the proportion of the energy value of the high-frequency component in the total signal energy is: ； Where, is the proportion of the energy value of the high-frequency component in the total signal energy, is the approximate coefficient in wavelet decomposition, reflecting the low-frequency slowly varying component of the temperature signal. is the detail coefficient in wavelet decomposition, reflecting the high-frequency mutation component of the temperature signal. is the abnormal frequency band index set, is the number of decomposition levels of the approximate coefficients, is the number of decomposition levels of detail coefficients.

3. The method for intelligent dynamic optimization of printed circuit board production parameters according to claim 2, characterized in that: The frequency domain energy analysis of the temperature fluctuation amplitude and the calculation of the abnormal amplitude energy ratio as the amplitude abnormality index specifically include: Convert the temperature signal to the frequency domain and divide the normal process bandwidth in the frequency domain ; Calculate the signal energy in the two frequency bands respectively, and use the ratio of the abnormal frequency band energy to the total energy as the amplitude anomaly index; Specifically, the calculation formula for the amplitude anomaly index is: ； is the amplitude anomaly index, which indicates the energy ratio within the abnormal frequency range. is the power spectrum density function, which reflects the energy distribution of the temperature signal in the frequency domain. For the normal process frequency range, is half the signal sampling frequency.

4. The method for intelligent dynamic optimization of printed circuit board production parameters according to claim 3, characterized in that: The optical sensor is used to collect surface wear data of the hot pressing rollers performing the lamination operation, analyze the risk of uneven temperature distribution caused by wear, and evaluate the risk level of the current lamination process, specifically including: A convolutional neural network is used to automatically extract the deep features of the hot pressing roller surface morphology from historical production data, build a sample library containing normal wear and abnormal wear, and annotate the corresponding risk level labels; Training risk prediction models based on sample libraries; Extract features from the surface topography images collected by the optical sensor, including the area of ​​the worn area and edge sharpness; The extracted features are input into the pre-trained risk prediction model, and an index value representing the degree of wear risk is output. When the index exceeds the threshold, it is judged as high risk.

5. The method for intelligent dynamic optimization of printed circuit board production parameters according to claim 4, characterized in that: Based on the risk level, low-risk temperature distribution patterns are screened for lamination production. After completion, the line width accuracy and inter-layer alignment of the circuit board are tested to obtain the quality deviation coefficient. Based on the quality deviation coefficient, it is determined whether it meets the preset process standards. Specifically, it includes: Measure the line width deviation and inter-layer alignment offset at key locations on the PCB, and calculate their relative errors to the target values. The errors of multiple detection points are weighted and summed to generate the quality deviation coefficient; Determine whether the quality deviation coefficient exceeds the deviation threshold, if so, it is determined that it does not meet the preset process standards, if not, it is determined that it meets the preset process standards; Specifically, the calculation formula of the mass deviation coefficient is: ； Where, is the quality deviation coefficient, which comprehensively reflects the process deviation of line width and alignment. is the measured line width of the i-th detection point, is the target line width, is the measured coordinate of the i-th detection point, Align the coordinates for the target, Process tolerance, is the total number of detection points, for and The distance between them.

6. The method for intelligent dynamic optimization of printed circuit board production parameters according to claim 5, characterized in that: For products that do not meet the preset process standards, tracing the temperature control parameters in the production process and dynamically adjusting the temperature setting value of the hot pressing roller until the quality deviation coefficient is lower than the preset threshold specifically include: When the mass deviation coefficient is continuously high, the hot pressing temperature setting value is adaptively adjusted according to the deviation direction, and the temperature is increased to compensate for the decrease in heat transfer efficiency caused by wear; The adjustment range is associated with the current deviation coefficient, historical adjustment records and process constraints, and stable optimization is achieved through fuzzy logic rules.

7. An intelligent dynamic optimization system for printed circuit board production parameters, characterized in that: The method for implementing the intelligent dynamic optimization method of printed circuit board production parameters according to any one of claims 1 to 6 specifically comprises: The temperature monitoring module includes a distributed temperature sensor array installed in the contact area between the hot pressing roller and the substrate, which is used to collect temperature distribution data of the lamination part in real time; A data processing module, which is in communication with the temperature monitoring module and includes a multi-channel signal acquisition unit and a real-time storage unit, and is used to perform noise reduction processing and time series alignment on the temperature data; A temperature analysis module is communicatively connected to the data processing module, and is configured to: perform wavelet decomposition on the temperature fluctuation signal to extract high-frequency mutation components, calculate the energy proportion of the abnormal frequency band as a frequency anomaly index; generate a power spectrum density distribution through FFT conversion, calculate the energy proportion of the abnormal amplitude as an amplitude anomaly index; and normalize the frequency anomaly index and the amplitude anomaly index to generate a temperature anomaly coefficient; The process section screening module is connected to the output end of the temperature analysis module and is configured to dynamically generate a stable process section mapping map of the hot pressing roller based on the temperature anomaly coefficient threshold comparison result; The surface detection module, which includes a high-resolution optical sensor and a coaxial light source assembly, is located above the lamination station to capture the surface topography of the hot pressing roller and extract wear characteristic parameters; A risk assessment engine, the risk assessment engine being in communication with the surface detection module, the risk assessment engine being configured to: analyze surface topography images using a pre-trained deep feature extraction network to output a wear risk index; and generate a risk heat map of the hot pressing area based on risk level classification rules; The quality inspection unit includes a laser microscopy measurement device and an image registration module, which is used to quantify the line width deviation and inter-layer alignment offset of the circuit board, and generate the quality deviation coefficient through weighted error calculation; The dynamic control module interacts with the hot pressing roller temperature control system in real time. The dynamic control module is configured to: start the temperature setting value backtracking analysis function according to the degree of exceeding the quality deviation coefficient; generate temperature compensation parameters through the genetic algorithm optimizer, and send PID control instructions to the actuator.

Images