Flexible circuit board production quality monitoring method and system based on data feedback

Abstract

The invention discloses a flexible circuit board production quality monitoring method and system based on data feedback, and the method comprises the steps: collecting the physical parameters of development etching, drilling and copper plating processes, such as line width, line distance, aperture, roundness and copper thickness, and workshop temperature and humidity and dust concentration data; generating a single-variable control chart for a single-process physical parameter, generating a multivariable T2 control chart for a multi-coupling parameter process, and calculating a process stability index and a comprehensive fluctuation index; inputting the abnormal data into a BP neural network, identifying an abnormal mode type and outputting characteristic parameters; generating a parameter correction instruction according to the abnormal mode and the characteristic parameters, and adjusting equipment parameters in real time; recalculating the process capability index based on the adjusted data, and triggering secondary feedback if the process capability index does not reach the standard; and dynamically adjusting the threshold value of the control chart according to the standard deviation and the mean value of the process capability indexes of the continuous batches. Production key parameters are comprehensively covered, abnormity is found in time, accurate and comprehensive detection is achieved, real-time adjustment is achieved, stability and controllability are ensured, and the product quality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of circuit board quality detection, and in particular to a method and system for monitoring the production quality of flexible circuit boards based on data feedback. Background Art

[0002] As a key electronic component, circuit boards (PCBs) integrate various electronic components, miniaturizing and simplifying circuits. They play a crucial role in the mass production of fixed circuits and optimizing the layout of electrical appliances. With the rapid advancement of technology, PCBs are increasingly used in a wide range of applications, from simple household appliances to complex industrial control systems. This widespread application has led to increasingly stringent market requirements for PCB quality.

[0003] Flexible printed circuits (FPCs) are circuit boards made of a flexible insulating substrate and conductive lines. They are widely used in electronic devices and are favored for their thinness, lightness, and flexibility. As electronic devices move toward miniaturization, lightweighting, and high integration, the quality requirements for FPCs are becoming increasingly stringent. Traditional quality inspection methods rely primarily on manual visual inspection and simple mechanical measurements. These methods are not only inefficient but also susceptible to human factors, leading to inconsistent inspection results and failing to meet the high precision and efficiency demands of modern production.

[0004] During the production of circuit boards, quality inspection is a crucial step in ensuring stable and reliable product performance. Traditional circuit board inspection methods mostly rely on manual visual inspection, which is not only inefficient but also easily affected by human factors, resulting in inaccurate inspection results. Patent publication number CN118762001B discloses a flexible circuit board quality inspection system and method based on data analysis. The system first captures a reference image of the flexible circuit board, extracts the positional relationship between the bending axis of the flexible circuit board and the electronic components, and constructs a model of image changes and flexible circuit board quality. Using image recognition technology, an assessment value for the failure rate of the flexible circuit board is generated, thereby reducing the workload of manual inspection. Clearly, circuit board inspection has received significant attention. With the continuous development of the circuit board industry, the requirements for circuit board inspection technology are becoming increasingly stringent. Modern circuit board inspection technology needs to possess efficient and accurate inspection capabilities. Summary of the Invention

[0005] The purpose of the present invention is to overcome the shortcomings of the existing technology and provide a flexible circuit board production quality monitoring method and system based on data feedback, which adopts multi-source data collection, abnormal fluctuation joint detection and feedback process adjustment methods to ensure data reliability; timely discover single or multiple process abnormalities in the production process, avoid potential quality problems, and thus improve product quality.

[0006] The object of the present invention is achieved through the following technical solutions:

[0007] A method for monitoring the production quality of a flexible circuit board based on data feedback, comprising:

[0008] S1. Multi-source data acquisition: Collect physical parameter data from the FPC development, etching, drilling, and copper plating processes, including line width, line spacing, hole diameter, roundness, and copper thickness, as well as workshop temperature, humidity, and dust concentration data;

[0009] S2. Joint detection of abnormal fluctuations: Generate a single variable control chart for the physical parameters of a single process to calculate the process stability index, and generate a multivariate T for the multi-coupling parameter process 2 The control chart calculates comprehensive volatility indicators and triggers an abnormal signal when any indicator exceeds the preset threshold;

[0010] S3. Neural Network Pattern Analysis: Input abnormal data into a pre-trained BP neural network to identify step, trend, or periodic abnormal pattern types and output characteristic parameters;

[0011] S4. Feedback process adjustment: Generate parameter correction instructions based on the abnormal pattern type and characteristic parameters, and adjust the equipment temperature, pressure or motion accuracy parameters of the corresponding process in real time;

[0012] S5. Closed-loop quality assessment: Recalculate the process capability indices Cp and Cpk based on the adjusted process data. If the indices do not meet the standards, secondary feedback is triggered until the process stabilizes.

[0013] S6. Dynamic threshold update: based on the process capability index standard deviation σ of N consecutive batches a and mean Dynamically adjust the thresholds of univariate and multivariate control charts to adaptively optimize monitoring sensitivity.

[0014] Through the multi-source data acquisition module, the key physical parameters of the flexible circuit board production process are comprehensively collected, such as the line width, line spacing, hole diameter, roundness and copper thickness of the development and etching, drilling and copper plating processes, as well as the temperature, humidity and dust concentration data of the workshop environment. These data provide the basis for subsequent anomaly detection and quality assessment. Then, the abnormal fluctuation joint detection module uses the single variable control chart and multivariate T 2Control charts calculate stability indicators for physical parameters of single processes and comprehensive fluctuation indicators for processes with multiple coupled parameters. When any indicator exceeds a preset threshold, the system triggers an anomaly signal, indicating a potential quality issue in the production process. The neural network pattern analysis module then inputs the anomaly data into a pretrained BP neural network, which can identify different anomaly pattern types, such as steps, trends, or periodicity, and outputs corresponding characteristic parameters. These characteristic parameters help further understand the specific circumstances and causes of the anomaly. Based on the identified anomaly pattern type and characteristic parameters, the feedback process adjustment module generates parameter correction instructions, adjusting parameters such as equipment temperature, pressure, or motion accuracy in real time for the corresponding process to eliminate the anomaly and restore production process stability. The closed-loop quality assessment module recalculates the process capability indices Cp and Cpk based on the adjusted process data to evaluate the effectiveness of the adjustments. If the indices do not meet the standards, secondary feedback is triggered, and adjustments are continued until the process stabilizes. This process ensures continuous improvement and optimization of production quality. Finally, the dynamic threshold update module calculates the process capability indices based on the standard deviation σ of the process capability indices for N consecutive batches. a and mean Dynamically adjust thresholds for single- and multivariate control charts. This adaptive optimization mechanism adjusts monitoring sensitivity as the production process changes, ensuring the accuracy and effectiveness of the monitoring system.

[0015] As a preferred embodiment, the univariate control chart in S2 includes a mean-range control chart, and the control limit calculation formula is:

[0016]

[0017] in, is the total sample mean, is the mean of the range, A2 is the control chart coefficient, UCL X and LCL X are the upper and lower control limits respectively.

[0018] As a preferred embodiment, the multivariate T in S2 2 The upper control limit of the control chart is calculated as:

[0019]

[0020] Among them, p is the variable dimension, n is the sample size, F α is the critical value of the F distribution under the confidence level α.

[0021] As a preferred method, the training of the BP neural network in S3 includes: generating a control chart sample set containing normal and abnormal patterns, extracting the statistical feature vector of the sample as input, the pattern type and feature parameters as output, and optimizing the network weights through the back propagation algorithm, wherein the statistical features include mean, range and standard deviation.

[0022] As a preferred embodiment, the parameter correction instruction in S4 includes:

[0023] Adjust the equipment temperature or pressure parameters for step anomalies, and the correction amplitude is proportional to the step amplitude;

[0024] The mechanical motion accuracy is calibrated for trend anomalies, and the correction slope is inversely proportional to the trend anomaly slope;

[0025] For periodic abnormal replacement of worn parts, the replacement period should match the fluctuation wavelength.

[0026] As a preferred method, the secondary feedback described in S5 includes: marking the substandard process as a high-risk batch, starting the manual re-inspection process, and comparing the re-inspection results with the neural network diagnosis results to optimize the model weights.

[0027] As a preferred method, it also includes: environmental coupling compensation, calculating the environmental interference coefficient HJ based on temperature, humidity and dust data, and dynamically compensating the physical parameters when HJ exceeds the threshold. The formula is:

[0028] HJ=α1·|T-T0|+α2·(H max -H)+α3·D

[0029] Among them, T is the real-time temperature, T0 is the standard temperature, H is the real-time humidity, H max is the upper limit of humidity, D is the dust concentration, and α1, α2, and α3 are weight coefficients.

[0030] As a preferred embodiment, the specific formula for updating the dynamic threshold in S6 is:

[0031]

[0032] Among them, σ Cp is the standard deviation of the process capability index, is the mean, and β is the adjustment factor.

[0033] A flexible circuit board production quality monitoring system based on data feedback, comprising:

[0034] The data acquisition module integrates a high-precision microscope, a copper thickness gauge, and environmental sensors to obtain real-time data on line width, line spacing, aperture, roundness, copper thickness, and workshop environment.

[0035] Statistical analysis module, performs univariate control chart analysis, multivariate T 2 Control chart analysis, output process stability and comprehensive fluctuation indicators;

[0036] The neural network module has a built-in BP neural network that identifies abnormal patterns in the control chart and outputs characteristic parameters; the feedback execution module generates parameter adjustment instructions based on the abnormal patterns and dynamically controls the production equipment;

[0037] Closed-loop evaluation module recalculates the process capability index, determines whether secondary feedback is required, and stores the results in the database;

[0038] The environmental compensation unit calculates the environmental interference coefficient HJ based on temperature, humidity and dust data, and dynamically corrects the physical parameters; the module realizes real-time data interaction and closed-loop control through industrial bus and Ethernet communication.

[0039] Preferably, the data acquisition module includes a vibration compensation unit and a data verification unit. The vibration compensation unit eliminates the influence of mechanical vibration on detection accuracy through an acceleration sensor. The data verification unit marks a timestamp on abnormal data and associates it with historical batches.

[0040] The present invention has at least the following beneficial effects: A multi-source data acquisition strategy comprehensively covers key physical parameters and environmental variables in the production process, providing a detailed data foundation for subsequent anomaly detection and pattern analysis. A combined abnormal fluctuation detection mechanism can promptly detect anomalies in a single or multiple processes during the production process, effectively avoiding potential quality issues. The combined use of single- and multivariate control charts improves the accuracy and comprehensiveness of anomaly detection. A feedback-based process adjustment strategy can adjust equipment parameters in real time based on abnormal patterns and characteristic parameters, ensuring the stability and controllability of the production process and thus improving product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] To reveal the technical details of the embodiments of the present invention, the following is a brief introduction to the drawings involved in the embodiments. It should be emphasized that these drawings only illustrate several embodiments of the present invention and should not be considered as defining the scope of the invention. Those skilled in the art can deduce other relevant drawings based on these drawings without engaging in creative work.

[0042] Figure 1 Schematic diagram of a process of an embodiment of the method of the present invention;

[0043] Figure 2 Schematic diagram of the structure of a flexible circuit board production quality monitoring system based on data feedback in an embodiment;

[0044] Figure 3 for. DETAILED DESCRIPTION

[0045] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the following.

[0046] In the following, embodiments of the present disclosure are described in detail with the aid of accompanying drawings. However, please be aware that the present disclosure is not limited to the specific forms shown herein. Rather, it should be understood to encompass various variations, equivalents, and / or alternatives to the embodiments of the present disclosure. In describing the drawings, the same reference numerals will be used to indicate similar components.

[0047] It should be understood that while the following description provides extensive specific details intended to facilitate a comprehensive understanding of the example embodiments, those skilled in the art will appreciate that the example embodiments can be implemented without these specific details. For example, systems may be presented in block diagram form to avoid excessive detail that would obscure the clarity of the examples. In other cases, unnecessary details regarding well-known processes, structures, and techniques may be omitted to maintain clarity of the examples.

[0048] like Figure 1 As shown, a flexible circuit board production quality monitoring method based on data feedback includes:

[0049] S1. Real-time multi-source data acquisition: A high-precision microscope inspection platform and copper thickness gauge are used to simultaneously collect physical parameter data from the FPC development, etching, drilling, and copper plating processes. These physical parameters include line width, line spacing, hole diameter, roundness, and copper thickness. Workshop temperature, humidity, and dust concentration data are also collected.

[0050] S2. Joint detection of abnormal fluctuations: Generate a single variable control chart for the physical parameters of a single process to calculate the process stability index, and generate a multivariate T for the multi-coupling parameter process 2 The control chart calculates comprehensive volatility indicators and triggers an abnormal signal when any indicator exceeds the preset threshold;

[0051] S3. Neural Network Pattern Analysis: Input abnormal data into a pre-trained BP neural network to identify step, trend, or periodic abnormal pattern types and output characteristic parameters;

[0052] S4. Feedback process adjustment: Generate parameter correction instructions based on the abnormal pattern type and characteristic parameters, and adjust the equipment temperature, pressure or motion accuracy parameters of the corresponding process in real time;

[0053] S5. Closed-loop quality assessment: Recalculate the process capability indices Cp and Cpk based on the adjusted process data. Cp measures the potential capability of the process under ideal conditions, that is, whether the process variation range can meet the specification requirements when the process mean is completely consistent with the specification center. Cpk introduces the impact of mean shift based on Cp to reflect the actual process capability. These two indices are relatively common and will not be discussed here. If the index does not meet the standard, secondary feedback is triggered until the process stabilizes.

[0054] S6. Dynamic threshold update: based on the process capability index standard deviation σ of N consecutive batches a and mean Dynamically adjust the thresholds of univariate and multivariate control charts to adaptively optimize monitoring sensitivity.

[0055] This embodiment monitors the production quality of flexible circuit boards based on data feedback. Through a high-precision microscope detection platform and a copper thickness measuring instrument, physical parameter data of each process of the flexible circuit board, such as line width, line spacing, hole diameter, roundness, and copper thickness, are collected in real time. At the same time, workshop environmental data, such as temperature, humidity, and dust concentration, are monitored. Using single-variable control charts and multivariable T 2 The control chart calculates the process stability of single processes and multi-coupled parameter processes, and immediately triggers an abnormal signal once an abnormal fluctuation is detected. The abnormal data is then input into the pre-trained BP neural network to identify the type of abnormal pattern, such as step, trend or periodic anomalies, and output characteristic parameters. Based on the identified abnormal pattern and characteristic parameters, parameter correction instructions are generated to adjust the equipment parameters of the corresponding process, such as temperature, pressure or motion accuracy, in real time to correct the anomaly. After adjusting the parameters, the process capability index Cp and Cpk are recalculated. If the index does not meet the standard, secondary feedback is triggered and adjustments are made continuously until the process is stable. Finally, based on the standard deviation and mean of the process capability index of consecutive batches, the threshold of the control chart is dynamically adjusted, and the sensitivity of the monitoring system is adaptively optimized to ensure the accuracy and effectiveness of monitoring.

[0056] In a preferred embodiment, the univariate control chart in S2 includes a mean-range control chart, and the control limit calculation formula is:

[0057]

[0058] in, is the total sample mean, is the mean of the range, A2 is the control chart coefficient, UCL X and LCL X are the upper and lower control limits respectively.

[0059] The mean-range control chart assesses production process stability by monitoring sample means and ranges. The sample mean reflects the concentration trend of the data, while the range reflects the degree of data dispersion. Control limits are calculated based on the overall mean and average range of historical data. The upper and lower limits are centered on the overall mean and are calculated by adding or subtracting the coefficient A2 multiplied by the average range to form a reasonable fluctuation range. Data points outside this range indicate possible abnormal factors and require investigation and adjustment.

[0060] The mean-range control chart monitors the stability of the production process by analyzing the central tendency and dispersion of sample data. This control chart consists of two parts: a mean chart, which monitors fluctuations in the data mean, and a range chart, which monitors changes in the data range. In flexible circuit board production, for example, when monitoring line width or aperture, the system first collects multiple sample groups, each containing several consecutive measurements. The mean and range (the difference between the maximum and minimum values) of each group are calculated. The mean values ​​of all groups are then summed to produce the overall mean and the average of the range values ​​of each group.

[0061] Control limits are set based on the grand mean and range mean of historical data. The upper and lower control limits of the mean chart are centered around the grand mean and extend outward by a certain multiple (determined by the control chart coefficient) of the range mean to form a reasonable fluctuation range. The control limits of the range chart are similarly calculated based on the range mean and coefficient. If equipment anomalies or material fluctuations occur during production, the sample mean or range may exceed the control limits, triggering a system alert indicating the need to investigate process parameters or equipment status. This control chart can effectively identify sudden anomalies or gradual trends in the process, ensuring the quality stability of high-precision flexible circuit boards.

[0062] In a preferred embodiment, the multivariate T in S2 2 The upper control limit of the control chart is calculated as:

[0063]

[0064] Among them, p is the variable dimension, n is the sample size, F α is the critical value of the F distribution under the confidence level α.

[0065] Multivariate T 2 Control charts can simultaneously monitor the combined fluctuations of multiple related quality parameters. In flexible circuit board production, parameters such as line width and line spacing in the development and etching process, and hole diameter and roundness in the drilling process are often interrelated, and a control chart for a single parameter cannot fully reflect the overall state of the process. By calculating the "multidimensional distance" between each sample and the historical data center, the control chart integrates the fluctuations of multiple variables into a single statistic, forming a unified quality fluctuation indicator. 2 Essentially, this statistic is a "multivariate distance," measuring the "combined deviation" between sample data and the process mean. This distance is always non-negative, and the only concern is whether it exceeds a safety threshold (upper control limit). When the statistic exceeds the preset threshold, it indicates that the process may be affected by abnormal factors.

[0066] T 2The threshold setting of the control chart comprehensively considers the variable dimension, sample capacity and confidence level. The variable dimension refers to the number of parameters monitored simultaneously (such as 2 dimensions when monitoring line width and line spacing at the same time), the sample capacity refers to the amount of data collected in each batch, and the confidence level determines the sensitivity to abnormal fluctuations. When the process requirements are strict, a higher confidence level (such as 99%) can be selected. At this time, the threshold will be increased accordingly to reduce the risk of false alarms. This embodiment can not only capture the collaborative anomalies between variables, but also avoid the error superposition problem when monitoring multiple parameters separately. It is particularly suitable for high-precision control scenarios of flexible circuit board parameters.

[0067] In a preferred embodiment, the training of the BP neural network in S3 includes: generating a control chart sample set containing normal and abnormal patterns, extracting the statistical feature vector of the sample as input, the pattern type and feature parameters as output, and optimizing the network weights through the back propagation algorithm, wherein the statistical features include mean, range and standard deviation.

[0068] Control chart anomaly pattern recognition based on a BP neural network achieves intelligent diagnosis by training the network with simulated data. The system first generates a control chart sample set containing both normal and abnormal patterns, simulating abnormalities such as step, trend, and cycle patterns. For example, step anomalies manifest as sudden jumps in the data, trend anomalies as sustained increases or decreases, and cycle anomalies as regular fluctuations. Three statistical features, namely mean, range, and standard deviation, are extracted for each sample. The mean reflects the trend in the data, the range captures the range of fluctuations, and the standard deviation quantifies the degree of dispersion. Together, these three constitute the input feature vector. The network employs a three-layer architecture, with the input layer corresponding to three feature nodes. The hidden layer learns feature associations through nonlinear transformations. The output layer is divided into two parts: pattern classification and parameter estimation. The classification module outputs probabilities of normal, step, trend, and cycle patterns, while the parameter module predicts numerical values ​​such as step amplitude, trend slope, or cycle wavelength. Training utilizes a backpropagation algorithm, comparing the predicted results with the true labels and adjusting weights layer by layer. After tens of thousands of iterations, the network establishes a mapping between features and patterns. In practice, the system collects process data in real time and calculates statistical features. These are then fed into the trained network to determine anomaly types and quantitative parameters. In a preferred embodiment, the parameter modification instruction in S4 includes:

[0069] Adjust the equipment temperature or pressure parameters for step anomalies, and the correction amplitude is proportional to the step amplitude Δ;

[0070] The mechanical motion accuracy is calibrated for trend anomaly, and the correction slope is inversely proportional to the trend anomaly slope k;

[0071] For periodic abnormal replacement of worn parts, the replacement period should match the fluctuation wavelength λ.

[0072] The system employs differentiated process adjustment strategies for different types of anomalies in flexible circuit board production. For step anomalies, manifested as a sudden jump in physical parameters (such as a 1.5μm increase in line width), the system corrects by adjusting equipment temperature or pressure parameters. The magnitude of the correction is proportional to the step amplitude Δ. For example, a larger Δ increases the temperature compensation, thereby quickly offsetting the sudden change. These anomalies are often caused by incorrect equipment parameter settings or material batch fluctuations. Linear compensation can restore process stability.

[0073] In the production process of flexible circuit boards, physical parameter jumps (such as a sudden increase of 1.5μm in line width) are usually caused by sudden changes in process conditions or misalignment of equipment parameters, and adjustments to temperature or pressure parameters can correct such anomalies. Taking temperature as an example, the etching rate is positively correlated with temperature. In the development and etching process, the temperature of the etching solution directly affects the chemical reaction rate. An increase in temperature will accelerate the etching rate of the copper layer by the etching solution, resulting in a narrower line width; conversely, a decrease in temperature will slow down the corrosion rate, resulting in a wider line width. Flexible substrate materials (such as polyimide) are sensitive to temperature. Temperature fluctuations may cause micron-level deformation of the substrate, indirectly changing the alignment accuracy of the etching mask and the copper layer, thereby affecting the line width. When a sudden increase in line width (Δ>0) is detected, the system corrects it through the following steps. See Figure 3 , ① Abnormal attribution (identifying abnormalities): The step signal indicates a sudden drop in the etching rate, which may be caused by a temperature sensor failure, an abnormal cooling system, or a sudden drop in ambient temperature. ② Temperature compensation formula (applying temperature compensation): The adjustment amount is calculated using a proportional model: ΔT = k·Δ, where Δ is the step amplitude, k is the temperature correction coefficient, and ΔT is the temperature adjustment amount. ③ Real-time feedback (providing real-time feedback): Increase the etching tank temperature (such as from 25°C to 26.5°C) to accelerate the etching solution reaction so that the line width gradually returns to the target value in subsequent processes. Temperature adjustment directly affects the line width accuracy through the dual effects of chemical reaction rate and material deformation. The real-time correction mechanism based on data feedback, combined with the quantitative model of process parameters and physical parameters, enables the system to quickly respond to step anomalies and ensure the stability of micron-level manufacturing of flexible circuit boards.

[0074] In the development and etching process of flexible circuit boards, the adjustment of pressure parameters can correct sudden increases in line width. Etching uniformity control: Pressure directly affects the contact tightness between the etching roller and the substrate. When the pressure is insufficient, the etching liquid is unevenly distributed, and the copper layer in some local areas is not fully removed, resulting in a wider line width; if the pressure is too high, the mask may be squeezed, causing abnormal etching liquid penetration, which in turn causes line width fluctuations. Mechanical stability: Pressure fluctuations can cause equipment vibrations, resulting in displacement of the mask and substrate. If the roller pressure is unbalanced, the etching area will be misaligned, directly amplifying the line width deviation. Pressure correction, when a sudden increase Δ in line width is detected, perform the following steps: ① Abnormal attribution: Determine whether it is caused by pressure abnormality by comparing historical data with the process model. If accompanied by a decrease in etching rate and a mask displacement alarm, the pressure factor can be locked. ② Calculation of correction amount: Use the linear proportional model: ΔP = k·Δ, where ΔP is the pressure adjustment amount (e.g., unit: MPa), k is the pressure correction coefficient (calibrated through experiments, for example, k = 0.3 MPa / μm), and Δ is the line width sudden increase value (e.g., 1.5 μm). ③ Real-time feedback: Dynamically adjust the hydraulic device through the servo system to increase the roller pressure, improve the contact efficiency between the etching solution and the copper layer, and inhibit the line width from continuing to widen. The pressure parameters directly act on the copper layer removal rate by regulating the etching solution distribution and the mechanical stability of the equipment, thereby quickly correcting the line width anomaly. Combining the Δ proportional model with closed-loop feedback, fully automatic control from "detecting anomalies" to "dynamic pressure regulation" is achieved, providing key guarantees for the micron-level manufacturing of high-density flexible circuit boards.

[0075] When an abnormal trend is detected (such as a continuous decrease in aperture diameter), the system identifies it as a gradual deviation caused by a shift in mechanical motion accuracy. The equipment's motion mechanism needs to be calibrated. The correction slope is inversely proportional to the slope of the abnormal trend, k. That is, the steeper the trend (larger k values), the smaller the calibration margin to avoid overshoot; the flatter the trend (smaller k values), the larger the calibration margin. For example, if wear on the drilling equipment's guide rails causes a linear change in aperture diameter, cumulative errors can be eliminated by inversely compensating for motion accuracy.

[0076] When an abnormal trend is detected (such as a continuous reduction in the aperture), the system corrects the deviation by calibrating the equipment's motion mechanism. The trend slope k reflects the cumulative rate of deviation of the process parameters. A large k value (steep trend) indicates that the deviation accumulates rapidly in a short period of time, which may be caused by sudden wear of mechanical parts or transient interference (such as vibration). At this time, a large adjustment may exceed the dynamic response capability of the equipment, causing overshoot or oscillation. A small k value (flat trend) indicates that the deviation accumulates slowly, which is usually caused by long-term wear or gradual environmental changes (such as temperature drift), and a larger correction is required to offset the long-term cumulative effect.

[0077] This embodiment uses an inverse relationship (correction amplitude ∝1 / |k|) to achieve stable compensation: Steep trend (large k): small adjustments to avoid violent disturbances. For example, if the sudden wear of the drilling machine guide rail causes the aperture to shrink by 2μm per hour (k=2μm / h), the system will fine-tune the guide rail pressure (such as reducing it by 0.5%) and gradually approach the target value through multiple iterations. Gentle trend (small k): Increase the correction amount to cover long-term effects. If the aperture shrinks by 0.5μm per month (k=0.5μm / month), the system may calibrate the mechanical transmission ratio (such as increasing it by 3%) to completely compensate for the systematic deviation caused by wear.

[0078] Dynamic hysteresis compensation: The mechanical system has response hysteresis, and directly correcting it according to the k value may cause overshoot. The inverse strategy balances response speed and stability through adaptive adjustment. Formula expression: Correction value Correction value Where λ is the baseline correction coefficient, and ∈ is the zero-prevention constant. The inverse correction strategy quantifies the rate of deviation accumulation and dynamically matches the mechanical system's response characteristics. This effectively eliminates deviations while suppressing overshoot, ensuring stability in micron-level manufacturing.

[0079] Periodic anomalies (such as regular fluctuations in copper thickness) are often caused by periodic component wear or vibration. The system dynamically adjusts component replacement cycles based on the fluctuation wavelength λ, increasing replacement frequency when λ is short and extending maintenance intervals when λ is long. For example, if a copper plating roller exhibits fluctuations with an amplitude of A = 0.8μm every λ = 8 hours, an 8-hour replacement cycle is applied, effectively eliminating the abnormal fluctuations at the source. These three adjustment strategies form a closed-loop control system, ensuring a rapid return to steady-state production.

[0080] In a preferred embodiment, the data in S1 is transmitted through a C / S architecture, the client is deployed at the process detection terminal, and the server integrates historical batch data and automatically compares it with the Gerber production standard file. The historical data includes process parameter setting values ​​and defect records. Distributed management of flexible circuit board production data is achieved through the client-server (C / S) architecture. The client is deployed at each process detection terminal (such as development and etching, drilling, and copper plating equipment), responsible for real-time collection of physical parameters such as line width and aperture obtained by high-precision microscopes and copper thickness measuring instruments, and recording workshop temperature, humidity and dust concentration data. The collected data is transmitted to the server via an industrial bus or Ethernet to ensure the timeliness and integrity of the test results.

[0081] The server integrates all historical batch data uploaded by the client and builds a comprehensive database containing process parameter settings, defect records, and environmental parameters. The system automatically compares real-time detection data with the Gerber production standard file. The Gerber file serves as a digital blueprint for circuit board design and stores the nominal values ​​and tolerance ranges of parameters such as line width and aperture. If the detection data exceeds the threshold defined by the Gerber file, the server immediately marks the anomaly and triggers an early warning, helping operators quickly locate process deviations. The linked analysis of historical data and real-time data is one of the core functions of the system. The server optimizes the current production parameter settings by analyzing the fluctuation patterns of process parameters in historical batches. For example, when the aperture roundness of a batch of drilling processes is abnormal and occurs frequently, the system can trace back historical data, associate it with equipment wear cycles or changes in environmental parameters, and generate targeted maintenance recommendations. At the same time, the accumulation of defect records provides training data for the neural network model, continuously improving the accuracy of abnormal pattern recognition.

[0082] The permissions management module ensures data security. Operators in different processes can only access the inspection data and statistical analysis results for their respective processes. Administrators can view global data and adjust system parameters through the server. This layered design ensures data isolation while supporting cross-process collaborative optimization. Through a closed-loop mechanism of real-time data collection, intelligent analysis, and dynamic adjustment, the system significantly improves the stability and yield rate of flexible circuit board manufacturing.

[0083] In a preferred embodiment, the secondary feedback described in S5 includes: marking the substandard process as a high-risk batch, starting a manual re-inspection process, and comparing the re-inspection results with the neural network diagnosis results to optimize the model weight. The secondary feedback mechanism plays a key corrective role in the closed-loop quality assessment. When the process capability index (Cp / Cpk) recalculated after the process adjustment still does not meet the standard, the system automatically marks the batch as high-risk and triggers the manual re-inspection process. The operator conducts a full inspection of the high-risk batch and records the actual defect type and location, such as the specific value of the line width deviation or the hole position offset. These manual re-inspection results are compared with the abnormal patterns and parameter estimates previously diagnosed by the neural network. If the neural network is found to have misjudged (such as identifying the trend anomaly caused by mechanical wear as environmental interference), the system will use the difference data as a new training sample and dynamically adjust the network weight. For example, if a batch re-inspection shows that the periodic fluctuation amplitude of the aperture is 1.2μm, while the original estimate of the neural network is 0.9μm, the system uses the back-propagation algorithm to correct the neuron connection strength, so that the parameter estimation error in subsequent similar situations is reduced. This closed-loop mechanism of automatic adjustment, manual verification, and model iteration not only avoids the limitations of relying solely on an algorithm, but also continuously optimizes the model through real data, ultimately achieving a dual improvement in the accuracy and adaptability of the monitoring system.

[0084] In a preferred embodiment, it also includes: environmental coupling compensation, calculating the environmental interference coefficient HJ based on temperature, humidity and dust data, and dynamically compensating the physical parameters when HJ exceeds a threshold. The formula is:

[0085] HJ=α1·|T-T0|+α2·(H max -H)+α3·D

[0086] Among them, T is the real-time temperature, T0 is the standard temperature, H is the real-time humidity, H max is the upper humidity limit, D is the dust concentration, and α1, α2, and α3 are weight coefficients. HJ (environmental interference coefficient) is the trigger condition used to determine whether compensation needs to be activated.

[0087] During flexible circuit board production, fluctuations in workshop environmental parameters can directly impact the measurement accuracy of micron-level process parameters. For example, rising temperatures can cause thermal expansion of the substrate material, causing line width measurements to deviate from the true value. Low humidity can exacerbate dust absorption, interfering with high-precision microscope imaging clarity. High dust concentrations can contaminate the etching solution, causing errors in copper thickness measurements. The environmental coupling compensation mechanism quantifies these interference factors, constructs a comprehensive evaluation model, and calculates the environmental interference coefficient (HJ) in real time.

[0088] The calculation of HJ incorporates three key environmental parameters: temperature, humidity, and dust. Temperature deviation reflects the difference between actual temperature and standard process temperature. The larger the deviation, the more significant the impact on detection accuracy. The humidity deficiency index measures the dryness of the air; the lower the humidity, the stronger the electrostatic adsorption effect. Dust concentration is directly related to the cleanliness of the workshop; the higher the concentration, the more severe the interference with optical detection. These three parameters are weighted and summed using preset weight coefficients. The weight values ​​are calibrated based on historical data to reflect the differences in sensitivity of different processes to environmental factors. For example, the developing and etching process is more sensitive to temperature changes, so the α1 value is larger; the drilling process is more significantly affected by dust, so the α3 weight is increased accordingly.

[0089] When the HJ value exceeds the set threshold (the HJ threshold is set according to historical data or process requirements), the system automatically triggers dynamic compensation. If HJ = 1.2 (threshold HJ = 1.0) is detected, it indicates that environmental interference has affected process stability. At this time, the system locates the main source of interference based on the contribution value of each sub-item: if the temperature deviation accounts for 60%, the etching tank constant temperature system is adjusted first; if the dust concentration is abnormally prominent, the workshop fresh air system is linked to reduce dust. The compensation amplitude is proportional to the HJ limit value, ensuring that the micron-level parameter measurement value restores the true process state. This mechanism effectively solves the problem of misjudgment caused by environmental noise in traditional detection, so that high-density flexible circuit boards can still maintain a detection accuracy of ±0.5μm under complex workshop conditions.

[0090] The compensation amount of each physical parameter is related to its sensitive environmental parameters. The compensation amount of each parameter is the weighted sum of each environmental deviation component, and the weight is determined by the sensitivity coefficient: Δ 参数 =∑(k i ·Δ 环境i ).

[0091] For line width: Δ line width = k T1 |T-T0|+k H1 ·(H max -H)+k D1 D, k T1 ,k H1 ,k D1 These are the sensitivity coefficients of line width to temperature, humidity, and dust, respectively (experimentally calibrated). Compensation Direction: When temperature increases, causing the line width to increase, subtract Δ line width for correction; when temperature decreases, causing the line width to decrease, add Δ line width for correction. When insufficient humidity decreases the line width, add Δ line width for correction; when excessive humidity increases the line width, subtract Δ line width for correction. When dust concentration increases, causing the line width to decrease, add Δ line width for correction; when dust concentration decreases, causing the line width to increase, subtract Δ line width for correction.

[0092] For line spacing: Δ line spacing = k T2 |T-T0|+k H2 ·(H max -H)+k D2 ·D;k T2 ,k H2 ,k D2 These are the sensitivity coefficients of wire spacing to temperature, humidity, and dust, respectively. Compensation Direction: When temperature increases, causing wire spacing to decrease, add Δ wire spacing to the correction; when temperature decreases, causing wire spacing to increase, subtract Δ wire spacing from the correction. When insufficient humidity decreases wire spacing, increase Δ wire spacing; when excessive humidity increases wire spacing, decrease Δ wire spacing from the correction. When dust concentration increases, causing wire spacing to decrease, add Δ wire spacing to the correction; when dust concentration decreases, causing wire spacing to increase, subtract Δ wire spacing from the correction.

[0093] For pore size: Δpore size = k T3 |T-T0|+k D3 ·D;k T3 ,k D3 These are the sensitivity coefficients of aperture to temperature and dust, respectively (dust has a higher weight). Compensation Direction: When temperature increases, causing the aperture to widen, ΔAperture is subtracted during correction; when temperature decreases, causing the aperture to narrow, ΔAperture is added during correction. When dust concentration increases, causing the aperture to narrow, ΔAperture is increased during correction; when dust concentration decreases, causing the aperture to widen, ΔAperture is decreased during correction.

[0094] For roundness: Δ roundness = k H4 ·(H max -H)+k D4 D, k H4 ,k D4 These are the sensitivity coefficients of roundness to humidity and dust, respectively. Compensation Direction: When insufficient humidity causes a decrease in roundness, corrections should be made by increasing Δroundness. When excessive humidity causes an abnormal increase in roundness, corrections should be made by decreasing Δroundness. When increased dust concentration causes a larger roundness deviation, corrections should be made by increasing Δroundness. When decreased dust concentration causes a smaller roundness deviation, corrections should be made by maintaining Δroundness.

[0095] For copper thickness: Δ 铜厚 =k T5 |T-T0|+k H5 ·(H max -H), k T5 ,k H5 These are the sensitivity coefficients of copper thickness to temperature and humidity, respectively. Compensation: When rising temperature accelerates the etching rate and reduces copper thickness, corrections require increasing ΔCopperThickness. When falling temperature slows the etching rate and increases copper thickness, corrections require decreasing ΔCopperThickness. When insufficient humidity increases copper thickness, corrections require decreasing ΔCopperThickness. When excessive humidity reduces copper thickness, corrections require increasing ΔCopperThickness.

[0096] In a preferred embodiment, the specific formula for updating the dynamic threshold in S6 is:

[0097]

[0098] Among them, σ Cp is the standard deviation of the process capability index, is the mean, and β is the adjustment factor.

[0099] The dynamic threshold update mechanism automatically adjusts the control chart's alarm threshold by quantifying the degree of fluctuation in the production process. Traditional fixed thresholds are prone to misjudgment when process stability fluctuates. For example, when a production line experiences frequent small fluctuations (such as differences in material batches or slight equipment aging), the fixed threshold may frequently trigger false alarms. Conversely, if process capability continues to improve (such as an increase in the Cp value after process improvements), the original threshold may be too loose, resulting in missed alarms. This mechanism dynamically expands or contracts the threshold range by calculating the standard deviation to mean ratio of the recent process capability index (Cp) in real time. A larger standard deviation indicates more severe process fluctuations, and the threshold tolerance is proportionally increased to avoid false alarm interference. A smaller standard deviation tightens the threshold to improve anomaly detection sensitivity.

[0100] The adjustment factor β plays a balancing role in this process. When the value of β is large (such as β = 1.5), the system is more sensitive to fluctuations, and the threshold adjustment range is large, which is suitable for scenarios with extremely high stability requirements and low tolerance; when the value of β is small (such as β = 0.5), the adjustment range is reduced to maintain a relatively stable threshold, which is suitable for conventional civilian product production. This embodiment can not only adapt to the natural fluctuations of the process, but also match the control requirements of different production stages through parameter configuration, realizing an intelligent upgrade from rigid thresholds to flexible adaptation.

[0101] In a preferred embodiment, the characteristic parameters described in S3 include the amplitude Δ and occurrence time t of step anomalies, the slope k and duration τ of trend anomalies, and the wavelength λ and amplitude A of periodic anomalies. The error rate for anomaly pattern recognition and parameter estimation is less than 5%. Using a trained neural network model, anomaly pattern recognition accurately identifies the type of control chart anomaly and quantifies its key parameter characteristics. The system first models the characteristics of three anomaly patterns: step, trend, and periodic. Step anomalies manifest as sudden changes in data at a specific time point. The neural network analyzes the jump amplitude Δ and the location t of the jump to identify equipment parameter missetting or batch material anomalies. Trend anomalies reflect gradual deviations in data, such as continuous increases or decreases. The system extracts the slope k and duration τ to identify long-term influencing factors such as mechanical wear or environmental drift. Periodic anomalies exhibit regular fluctuations. The algorithm captures the wavelength λ and amplitude A to identify periodic fault sources such as equipment vibration or component fatigue. To ensure an error rate below 5%, the system employs a two-stage learning strategy. In the first phase, the network was trained using tens of thousands of simulated data sets, covering scenarios with varying amplitudes, slopes, and periods, strengthening the model's ability to discern abnormal features. In the second phase, actual production data was injected to fine-tune the weights and eliminate discrepancies between simulation and reality. For example, for step anomalies, the network controlled the amplitude error to within ±0.1μm by comparing it to historical mutation cases. For trend slopes, the model was calibrated using equipment maintenance records to ensure slope estimation error of less than 2%. This dual mechanism of simulation training and real-world validation enables the system to maintain high-precision recognition even under complex operating conditions. In actual application, the system receives control chart data streams in real time and compares them point by point against pre-set pattern features. When an abnormal signal is detected, it simultaneously outputs a type determination and parameter estimation results. For example, when a periodic wavelength λ = 8 hours and an amplitude A = 1.2μm is identified, it automatically associates it with the equipment maintenance cycle, triggering a bearing replacement alert. Through parameterized feedback, the system converts abstract fluctuations into executable process instructions, forming a closed-loop control chain from anomaly detection to precise intervention.

[0102] A flexible circuit board production quality monitoring system based on data feedback, see Figure 2 ,include:

[0103] The data acquisition module integrates a high-precision microscope, a copper thickness gauge, and environmental sensors to obtain real-time data on line width, line spacing, aperture, roundness, copper thickness, and workshop environment.

[0104] Statistical analysis module, performs univariate control chart analysis, multivariate T 2 Control chart analysis, output process stability and comprehensive fluctuation indicators;

[0105] Neural network module, with built-in BP neural network, identifies abnormal patterns in control charts and outputs characteristic parameters;

[0106] Feedback execution module generates parameter adjustment instructions based on abnormal patterns and dynamically controls production equipment;

[0107] Closed-loop evaluation module recalculates the process capability index, determines whether secondary feedback is required, and stores the results in the database;

[0108] The environmental compensation unit calculates the environmental interference coefficient HJ based on temperature, humidity and dust data, and dynamically corrects the physical parameters; the module realizes real-time data interaction and closed-loop control through industrial bus and Ethernet communication.

[0109] This flexible circuit board production quality monitoring system achieves intelligent monitoring of the entire process through multi-module collaboration. The data acquisition module is equipped with a high-precision microscope and copper thickness measuring instrument to capture micron-level parameters such as line width and aperture in real time during processes such as etching, drilling, and copper plating. It also integrates temperature, humidity, and dust sensors to monitor the workshop environment. The collected raw data is processed by the statistical analysis module, and the single-variable control chart tracks the independent fluctuations of each parameter. The multi-variable T 2 Control charts evaluate combined anomalies of related parameters, such as simultaneously monitoring the co-existence of line width and line spacing.

[0110] When an abnormal signal is detected, the neural network module initiates analysis. The pre-trained BP neural network identifies abnormal patterns such as steps, trends, and cycles, and outputs specific parameters. If a step abnormality is identified, the jump amplitude (1.5μm) and the time of occurrence are simultaneously displayed. The feedback execution module then coordinates production equipment, such as adjusting etching temperature to compensate for sudden line width changes or calibrating drilling pressure to correct for aperture deviations, providing an immediate response for detection, diagnosis, and adjustment.

[0111] The environmental compensation unit dynamically corrects measurement errors. For example, if high temperatures cause substrate expansion, the actual line width is recalculated based on the temperature and humidity weighting coefficients to ensure data authenticity. After process adjustments, the closed-loop assessment module verifies that the process capability index meets the target through secondary testing. If it fails, manual re-inspection is triggered and the neural network model is optimized. Each module exchanges data in real time via the industrial network, forming a complete closed loop from parameter acquisition to process optimization, significantly improving the stability and yield rate of high-precision flexible circuit board production.

[0112] In a preferred embodiment, the data acquisition module includes a vibration compensation unit and a data verification unit. The vibration compensation unit eliminates the influence of mechanical vibration on detection accuracy through an acceleration sensor. The data verification unit timestamps abnormal data and associates it with historical batches.

[0113] The data acquisition module of the flexible circuit board manufacturing system ensures detection accuracy and data reliability through a vibration compensation unit and a data verification unit. The vibration compensation unit uses an accelerometer to monitor mechanical vibrations during operation in real time. When a high-precision microscope or copper thickness gauge experiences micron-level deviations due to equipment vibration, the sensor captures the vibration amplitude and frequency signals and uses an algorithm to generate reverse compensation instructions to dynamically adjust the position of the detection platform or the optical focal length. For example, when a drilling machine is operating at high speed, the system detects a 10Hz horizontal vibration signal and immediately drives the servo motor to fine-tune the stage position in the reverse direction, eliminating ±0.8μm displacement errors and ensuring the accuracy of aperture measurements.

[0114] The data verification unit performs multiple verifications when collecting data. When a sudden increase or decrease in parameters such as line width and aperture is detected, the unit will compare the fluctuation range of historical data of the same batch. If a certain line width measurement value deviates from the historical mean by more than 3σ, the system automatically marks the abnormal timestamp and associates the production equipment number, environmental parameters and operator information of the batch. If a batch detects that the line width jumps by 1.2μm three times in a row, the verification unit will associate it with the vibration compensation log of the same model substrate last week, and quickly locate it as a systematic deviation caused by loose guide screws, providing structured data support for subsequent quality traceability. Through the dual mechanisms of vibration suppression and data cross-validation, the system can still maintain micron-level detection accuracy in complex industrial environments.

[0115] In a preferred embodiment, the feedback execution module includes an instruction priority queue and a parameter correction verification unit. The priority queue is sorted by the severity of the abnormality and allocates adjustment resources. The verification unit collects three sets of verification data after the adjustment, and triggers a manual intervention signal if it does not meet the standard. The feedback execution module ensures the accuracy of process adjustment through intelligent instruction scheduling and effect verification mechanism. After the system receives the abnormality type and parameters determined by the neural network, the instruction priority queue hierarchically sorts the adjustment instructions according to preset rules. If the step abnormality causes the line width to increase by more than ±2μm, it is classified as high priority, triggering immediate adjustment of the equipment temperature or pressure; if the trend abnormality slope is less than 0.1μm / batch, it is classified as medium priority, and the mechanical accuracy is calibrated after the equipment is idle. This grading strategy can avoid resource conflicts caused by the simultaneous adjustment of multiple processes, and give priority to high-risk abnormalities to reduce scrap rates.

[0116] The parameter correction verification unit starts the closed-loop verification process after the instruction is executed. The system automatically collects three consecutive sets of process data after adjustment, recalculates statistics such as mean and range, and compares them with the target value. If all three sets of data are within the control limit, the adjustment is deemed effective and the process parameters are updated; if any set exceeds the limit, the manual intervention signal is immediately triggered and the abnormal details are pushed to the engineer terminal. For example, after a certain drilling aperture correction, the verification data shows that the fluctuation amplitude has not decreased. The system will link the equipment maintenance log, prompting that the compensation failure may be caused by guide rail wear, and guide manual intervention for in-depth maintenance. The dynamic adjustment and multiple verification mechanisms effectively balance automation efficiency and process reliability.

[0117] In a preferred embodiment, the database uses a time-series structure to store data, supporting multi-dimensional searches by batch, time, and process, and automatically comparing discrepancies with production standard Gerber files. The system's database utilizes a time-series structured storage method to comprehensively record all types of data from the flexible circuit board production process. Each batch's measured values ​​for physical parameters such as line width and aperture are accurately timestamped and archived by process, such as developing, etching, drilling, and copper plating. This time-series storage mechanism enables historical data traceability. For example, if aperture deviation occurs in a batch, the test records for that batch across different processes can be quickly traced back to analyze the root cause. The database supports flexible, multi-dimensional search capabilities. Users can query production data for a specific product's entire lifecycle by batch number, filter quality trends by week or month by time range, or extract all relevant parameters for a single process (such as drilling). This multi-angle query capability helps engineers quickly locate abnormal periods, such as filtering all records of line width exceeding the standard within the past 24 hours and analyzing the impact of equipment temperature fluctuations in conjunction with equipment logs. The system automatically and intelligently compares test data with Gerber design files. Gerber files serve as digital blueprints for circuit board designs, storing the nominal values ​​and tolerance ranges of parameters such as line width and aperture. Each time new inspection data is entered, the system matches the corresponding design standards in real time and marks any abnormal points that exceed the tolerance. For example, if the designed diameter of a hole is 20μm and the measured value is 19.2μm, the system automatically triggers an alert and associates the processing equipment number of the hole, providing precise positioning for process adjustments. This dynamic comparison mechanism implements closed-loop verification from design standards to production practice, ensuring that micron-level process parameters strictly meet quality requirements. In a preferred embodiment, it also includes a mobile monitoring terminal that receives high-risk alerts from the closed-loop assessment module and displays real-time video streams, parameter curves, and neural network diagnostic reports, supporting remote confirmation of adjustment instructions. The mobile monitoring terminal provides remote real-time monitoring and decision support capabilities for the flexible circuit board production quality monitoring system. The terminal is connected to the main control system via a wireless network. When the closed-loop assessment module detects a high-risk batch or process anomaly, the system automatically pushes an alert to the terminal interface. Alarm information includes the anomaly type (e.g., sudden line width changes, periodic aperture fluctuations), the location of the anomaly (development and etching process or drilling equipment number), real-time parameter curves (e.g., temperature fluctuation trends), and a neural network diagnostic report (anomaly pattern classification and feature parameter estimation). Operators can view real-time footage from workshop cameras through the terminal to observe equipment operating status and the effectiveness of process adjustments, such as confirming whether the drilling machine's aperture deviation is caused by guide rail wear.

[0118] The terminal supports interactive operation. After receiving an alarm, engineers can directly call up historical data for comparative analysis on the interface. If the system prompts that the abnormal slope of the line width trend of a batch exceeds the standard, the engineer can remotely view the line width control chart of the same process in the past 24 hours, and determine whether the machine needs to be shut down for maintenance based on the equipment maintenance record. After confirming the abnormality, the terminal provides a one-click command issuance function, such as sending an adjustment command of "copper thickness compensation +0.5μm" to the copper plating process. After receiving the command, the equipment automatically executes and feedbacks the execution status. To adapt to complex industrial environments, the terminal has built-in multi-level authority management. The workshop director can view global data and authorize emergency operations, and the process manager can only access the information of this process. All operation records and data changes are encrypted and stored to ensure traceability and security. The closed-loop remote management mechanism of early warning, analysis, decision-making, and execution has significantly improved the response speed and cross-regional collaboration efficiency of high-density flexible circuit board production.

[0119] The flexible circuit board production quality monitoring system of the present invention builds a closed-loop control system through the collaboration of multiple modules, realizing full-process monitoring from data acquisition to intelligent optimization. The data acquisition module acquires physical parameters such as line width and aperture and environmental data in real time through a high-precision microscope, copper thickness measuring instrument and environmental sensor, and transmits the data to the statistical analysis module via the industrial network. The data is analyzed by the single variable mean-range control chart and the multivariate T 2 Control charts simultaneously detect single-process stability and multi-parameter coupling anomalies. The neural network module performs pattern recognition on abnormal data, distinguishing between step, trend, or cycle types and quantifying characteristic parameters. The feedback execution module dynamically adjusts the temperature, pressure, or motion accuracy of the equipment according to the abnormal pattern. The closed-loop evaluation module recalculates the process capability index to verify the adjustment effect. If the standard is not met, manual re-inspection and model optimization will be triggered. The environmental compensation unit dynamically corrects measurement deviations based on temperature, humidity, and dust data. Mobile terminals enable remote real-time monitoring and command issuance. The database supports time series storage and compares differences with the Gerber standard. Each module is organically connected through data flow and feedback chain, forming a closed-loop self-optimization mechanism of perception, analysis, decision-making, and verification, significantly improving the accuracy and yield of micron-level manufacturing of flexible circuit boards.

[0120] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as covering the preferred embodiments and all changes and modifications that fall within the scope of the invention. The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. A method for monitoring the production quality of flexible circuit boards based on data feedback, characterized in that: include: S1. Multi-source data acquisition: Collect physical parameter data for the FPC development, etching, drilling, and copper plating processes, including line width, line spacing, hole diameter, roundness, and copper thickness, as well as workshop temperature, humidity, and dust concentration data; S2. Joint detection of abnormal fluctuations: Generate a single variable control chart for the physical parameters of a single process to calculate the process stability index, and generate a multivariate T for the multi-coupling parameter process 2 The control chart calculates comprehensive volatility indicators and triggers an abnormal signal when any indicator exceeds the preset threshold; S3. Neural Network Pattern Analysis: Input abnormal data into a pre-trained BP neural network to identify step, trend, or periodic abnormal pattern types and output characteristic parameters; S4. Feedback process adjustment: Generate parameter correction instructions based on the abnormal pattern type and characteristic parameters, and adjust the equipment temperature, pressure or motion accuracy parameters of the corresponding process in real time; S5. Closed-loop quality assessment: Recalculate the process capability indices Cp and Cpk based on the adjusted process data. If the indices do not meet the standards, secondary feedback is triggered until the process stabilizes. S6. Dynamic threshold update: based on the process capability index standard deviation σ of N consecutive batches a and mean Dynamically adjust the thresholds of single-variable and multivariable control charts to adaptively optimize monitoring sensitivity.

2. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: The univariate control chart described in S2 includes a mean-range control chart, and the control limit calculation formula is: in, is the total sample mean, is the mean of the range, A2 is the control chart coefficient, UCL X and LCL X are the upper and lower control limits respectively.

3. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: S2 Multivariate T 2 The upper control limit of the control chart is calculated as: Among them, p is the variable dimension, n is the sample size, F α is the critical value of the F distribution under the confidence level α.

4. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: The training of the BP neural network described in S3 includes: generating a control chart sample set containing normal and abnormal patterns, extracting the statistical feature vector of the sample as input, the pattern type and feature parameters as output, and optimizing the network weights through the back propagation algorithm, wherein the statistical features include mean, range and standard deviation.

5. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: The parameter correction instruction in S4 includes: Adjust the equipment temperature or pressure parameters for step anomalies, and the correction amplitude is proportional to the step amplitude; The mechanical motion accuracy is calibrated for trend anomalies, and the correction slope is inversely proportional to the trend anomaly slope; For periodic abnormal replacement of worn parts, the replacement cycle should match the fluctuation wavelength.

6. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: The secondary feedback described in S5 includes: marking the substandard process as a high-risk batch, initiating a manual re-inspection process, and comparing the re-inspection results with the neural network diagnosis results to optimize the model weights.

7. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: It also includes: environmental coupling compensation, calculating the environmental interference coefficient HJ based on temperature, humidity and dust data, and dynamically compensating the physical parameters when HJ exceeds the threshold. The formula is: HJ=α1·|T-T0|+α2·(H max -H)+α3·D Among them, T is the real-time temperature, T0 is the standard temperature, H is the real-time humidity, H max is the upper limit of humidity, D is the dust concentration, and α1, α2, and α3 are weight coefficients.

8. The method for monitoring the production quality of a flexible circuit board based on data feedback according to claim 1, characterized in that: The specific formula for dynamic threshold update in S6 is: Among them, σ Cp is the standard deviation of the process capability index, is the mean, and β is the adjustment factor.

9. A flexible circuit board production quality monitoring system based on data feedback, characterized in that: include: Data acquisition module, used to obtain line width, line spacing, aperture, roundness, copper thickness and workshop environment data; Statistical analysis module, performs univariate control chart analysis, multivariate T 2 Control chart analysis, output process stability and comprehensive fluctuation indicators; Neural network module, with built-in BP neural network, identifies abnormal patterns in control charts and outputs characteristic parameters; Feedback execution module generates parameter adjustment instructions based on abnormal patterns and dynamically controls production equipment; Closed-loop evaluation module recalculates the process capability index, determines whether secondary feedback is required, and stores the results in the database; The environmental compensation unit calculates the environmental interference coefficient HJ based on temperature, humidity and dust data, and dynamically corrects the physical parameters; the module realizes real-time data interaction and closed-loop control through industrial bus and Ethernet communication.

10. The flexible circuit board production quality monitoring system based on data feedback according to claim 9, characterized in that: The data acquisition module includes a vibration compensation unit and a data verification unit. The vibration compensation unit eliminates the influence of mechanical vibration on detection accuracy through an acceleration sensor. The data verification unit marks a timestamp on abnormal data and associates it with historical batches.

Images