Cooperative control system and method for multi-process processing of PCB (Printed Circuit Board)

Abstract

The invention discloses a cooperative control system and method for multi-process processing of a PCB (Printed Circuit Board). Multi-source process data of each process is collected, a process state baseline is formed through mutual verification of redundant sensors, process drift characteristics are examined and identified based on process capability, a process reference library is constructed in a layered manner by adopting a dynamic threshold value, and a generated process deviation parameter is detected; performing correlation analysis on adjacent processes to determine a transfer coupling degree, positioning a defect transfer link through bidirectional verification combining forward tracking and reverse backtracking, and cascading and accumulating process deviation parameters along the link to form a process incidence matrix; carrying out backtracking on defect diffusion to determine a sweep range, generating process state parameters according to the sweep range, and extracting trend turning characteristics to construct a process characteristic spectrum; defect amplification factors are extracted through sensitivity decoupling, process critical nodes are identified, a regulation and control rule base is constructed, process characteristic chromatograms are integrated, and process control instructions are output, so that defect source positioning and propagation path tracking are realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of PCB intelligent manufacturing, in particular to a PCB multi-process processing collaborative control system and method. BACKGROUND

[0002] As the core carrier of electronic products, PCB manufacturing process involves multiple processes such as inner layer pattern transfer, lamination, drilling, copper plating, solder mask printing, etc. There is a complex parameter transmission and quality correlation relationship between each process. When the manufacturing parameters of a certain process deviate, the deviation is often not limited to the current process, but is transmitted to the downstream process and accumulates step by step, and finally exposed as a serious quality defect in the finished product stage. For example, improper temperature control of the lamination process will cause abnormal stress distribution in the board, which will cause hole position deviation when transmitted to the drilling process, and will evolve into uneven plating when transmitted to the copper plating process. This cross-process defect propagation characteristic brings great difficulty to PCB quality control.

[0003] The existing PCB process monitoring technology mainly adopts a single-process independent monitoring method, each process sets a parameter threshold and alarms when it is exceeded. This method has obvious limitations: it lacks analysis of the influence transmission mechanism between the previous and subsequent processes, it is difficult to accurately trace back to the source process when the downstream process has quality problems; the monitoring of process parameters is mainly static threshold judgment, it is difficult to give early warning when the parameters are still in the normal range but have shown a deterioration trend; the process control strategy cannot identify sensitive parameters and key node positions, resulting in scattered control resources and limited intervention effect. SUMMARY

[0004] The present application discloses a PCB multi-process processing collaborative control system and method, which collects multi-source process data of each process and performs mutual verification filtering to establish a process state baseline, identifies process drift characteristics and builds a hierarchical process reference library, analyzes the transmission coupling relationship between processes to locate the defect transmission link, traces back to the defect diffusion to determine the affected range and extract the trend turning feature to build a process feature map, and finally identifies the process critical node through sensitivity decoupling and outputs accurate process control instructions to realize full-process collaborative control of the PCB production line.

[0005] The present application discloses a PCB multi-process processing collaborative control system and method, which collects multi-source process data of each process and performs mutual verification filtering to establish a process state baseline, identifies process drift characteristics and builds a hierarchical process reference library, analyzes the transmission coupling relationship between processes to locate the defect transmission link, traces back to the defect diffusion to determine the affected range and extract the trend turning feature to build a process feature map, and finally identifies the process critical node through sensitivity decoupling and outputs accurate process control instructions to realize full-process collaborative control of the PCB production line.

[0006] Collecting multi-source process data of each process of the PCB production line, and performing multi-source mutual verification filtering on the multi-source process data to form a process state baseline;

[0007] The process capability review is performed according to the process state baseline to identify a process drift feature, a dynamic threshold is divided in the process drift feature to build a process benchmark library, offset detection is performed through the process benchmark library to generate a process deviation parameter;

[0008] The process state baseline is subjected to inter-process correlation enhancement processing to determine a transmission coupling degree of adjacent processes, the transmission coupling degree is used to perform cross-process penetration tracking to locate a defect transmission link, the process deviation parameter is cascaded and accumulated along the defect transmission link to form a process correlation matrix;

[0009] The defect diffusion backtracking and influence domain aggregation are performed on the process correlation matrix to determine a defect spread range, a process state parameter is generated according to the defect spread range, and a process feature map is built by extracting a trend turning feature in the process state parameter;

[0010] The defect amplification factor is extracted by decoupling the sensitivity of the process deviation parameter and the process correlation matrix, the process critical node is identified according to the defect amplification factor, the regulation and control rule library is built around the process critical node, and the process control instruction is output by integrating the process feature map through the regulation and control rule library.

[0011] The second aspect of the present application proposes a collaborative control system for PCB multi-process processing, comprising:

[0012] A data acquisition module is configured to acquire multi-source process data of each process of a PCB production line, and to perform multi-source mutual verification filtering on the multi-source process data to form a process state baseline.

[0013] A benchmark detection module is configured to perform process capability review according to the process state baseline to identify a process drift feature, to perform dynamic threshold division in the process drift feature to build a process benchmark library, and to generate a process deviation parameter through offset detection of the process benchmark library.

[0014] An association analysis module is configured to perform inter-process correlation enhancement processing on the process state baseline to determine a transmission coupling degree of adjacent processes, to use the transmission coupling degree to perform cross-process penetration tracking to locate a defect transmission link, to cascade and accumulate the process deviation parameter along the defect transmission link to form a process correlation matrix.

[0015] A map building module is configured to perform defect diffusion backtracking and influence domain aggregation on the process correlation matrix to determine a defect spread range, to generate a process state parameter according to the defect spread range, and to build a process feature map by extracting a trend turning feature in the process state parameter.

[0016] The instruction output module is configured to extract a defect amplification factor by performing sensitivity decoupling on the process deviation parameter and the process sequence correlation matrix, identify a process critical node according to the defect amplification factor, construct a regulation rule library around the process critical node, and output a process sequence control instruction via the regulation rule library based on the process feature map. BRIEF DESCRIPTION OF DRAWINGS

[0017] The accompanying drawings, which are part of the specification, illustrate embodiments of the application and, together with the detailed description, serve to explain the principles of the application, its concepts and effects.

[0018] Figure 1 is a flowchart of a PCB multi-process machining collaborative control method of the application.

[0019] Figure 2 is a structural block diagram of a PCB multi-process machining collaborative control system of the application. DETAILED DESCRIPTION

[0020] In the following description, specific details are set forth, such as particular system configurations, techniques, etc., in order to provide a thorough understanding of the embodiments of the application. However, persons having ordinary skill in the art will readily understand that the application can be practiced without these specific details, which are for illustration and understanding only. In other instances, well-known systems, structures, circuits, and techniques have not been shown in detail in order not to obscure the understanding of this application.

[0021] It should be understood that the term "comprising" as used in the specification and in the claims indicates the presence of the recited features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0022] In the present specification, the reference "one embodiment" or "some embodiments" means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. The appearances of the phrases "in one embodiment", "in some embodiments", "in other embodiments", "in additional embodiments", and so on, in various places in the specification are not necessarily all referring to the same embodiment, unless otherwise specifically stated. The terms "comprising", "including", "having" and their variants, mean "including but not limited to", unless otherwise specifically indicated.

[0023] The technical solutions of the embodiments of the application are described below.

[0024] AsFigure 1 As shown, the embodiment of the present application provides a collaborative control method for PCB multi-process processing, including the following steps S110-S150:

[0025] Step S110, collect multi-source process data of each process of the PCB production line, and perform multi-source mutual verification filtering on the multi-source process data to form a process state baseline.

[0026] Specifically, the multi-source process data of each process of the PCB production line is collected. Data acquisition devices are deployed at each key process of the PCB production line, and the processes include inner layer pattern transfer, lamination, drilling, copper plating, and solder mask printing. The data acquisition device acquires process parameters from multiple information sources, including temperature sensors, pressure sensors, speed encoders, vision detection cameras, and current monitoring modules. For the same process parameter, at least two different types of sensors are used for redundant collection to ensure the reliability of the data. In the lamination process, the lamination temperature is collected by both a thermocouple and an infrared thermometer, and the measurement values of the two sensors are mutually verified. In the drilling process, the hole coordinates are acquired by both a vision detection camera and a mechanical probe, which facilitates subsequent mutual verification. In the inner layer pattern transfer process, parameters such as exposure energy, development temperature, development time, and etching rate are collected. In the lamination process, parameters such as preheating temperature, lamination pressure, temperature rise rate, and pressure holding time are collected. In the drilling process, parameters such as spindle speed, feed rate, drill wear, and hole deviation are collected. The collected raw data is time-stamped to establish a unified time reference, aligning the data of different processes and different sensors in the time dimension. The parameter data collected from each process is classified and organized according to data type to form a multi-source process data set.

[0027] In some embodiments, the multi-source process data is subjected to multi-source mutual verification filtering to form a process state baseline, including: expanding the multi-source process data according to the process flow to form a process data sequence; locating a core process anchor point from the process data sequence by multi-source mutual verification filtering; establishing a process feature imprint with the core process anchor point as a trajectory reference; and generating a process state baseline according to the process feature imprint.

[0028] According to multi-source process data, process data sequences are formed according to process flow. Parameters in multi-source process data are arranged according to the order of process flow of PCB production. The process flow starts from inner layer pattern transfer, and then goes through lamination, drilling, copper plating, outer layer pattern transfer, etching and solder mask printing. Each process corresponds to a set of process parameters. Parameters in multi-source process data belonging to the same process are collected together to form a parameter subset of the process. According to the order of the process, the parameter subsets are arranged in series, and the time sequence correlation between adjacent processes is established. The output time of the previous process and the input time of the next process are continuous. In the lamination process and the drilling process, the interval between the time stamp of lamination completion and the time stamp of drilling start reflects the process connection efficiency. In normal production, the interval is maintained within a fixed range, and an abnormally prolonged interval may indicate a stoppage of material turnover. Through process flow development, scattered multi-source process data is integrated into structured process data sequences, and the process data sequences clearly show the evolution trajectory of process parameters with the process advancing.

[0029] From the process data sequence, multi-source mutual verification is performed to filter and locate the core process anchor point. In the process data sequence, the key parameter position that has the greatest impact on product quality is identified. The sensitivity of the parameters of each process is analyzed to evaluate the impact of parameter changes on the quality of subsequent processes and the final product. The data points corresponding to the parameters with high sensitivity are marked as candidate anchor points. The data at the candidate anchor points are subjected to multi-source mutual verification, and the measurement values of redundant sensors are compared to verify the consistency and reliability of the data. In the drilling process, the hole accuracy parameter directly affects the subsequent copper plating effect and circuit conductivity, and the hole accuracy measurement point is marked as a candidate anchor point. The hole data obtained by visual detection and mechanical probe are mutually verified. When the hole deviation values measured by the two methods differ by more than the allowed range, it is determined that the data point is abnormal, and the abnormal data is removed. The position is determined as the core process anchor point. In the lamination process, the lamination temperature is collected by two sensors, thermocouple and infrared thermometer. When the measurement deviation of the two sensors is within the normal range, the data is confirmed to be valid, and the time when the temperature reaches the set value is determined as the core process anchor point of the process. The candidate anchor points that pass the multi-source mutual verification are screened, and one to two most critical data points are retained as the core process anchor point for each process. The core process anchor point represents the most critical parameter position for quality control in each process, and subsequent process monitoring will be carried out around these anchor points.

[0030] The parameter values of the core process anchor point are extracted as a feature vector representing the process state of the process. For a process containing multiple parameters, the anchor point parameters are weighted and combined according to their importance to form a comprehensive feature value of the process. With the core process anchor point as the center, the parameter value variation characteristics in a certain range in the forward and backward time directions are extracted, including the change rate, fluctuation amplitude and trend direction. In the lamination process, the time when the temperature reaches the set value is taken as the core process anchor point, and the temperature value, pressure value, and temperature rise rate and pressure change rate in the next 30 seconds are extracted, which together form the feature vector of the process. In the drilling process, the time when the first hole is completed is taken as the core process anchor point, and the spindle speed, feed rate, and speed fluctuation and feed stability in the next 10 seconds are extracted to form the feature vector of the drilling process. The feature vectors of each process are arranged in sequence according to the process order to form a process feature mark representing the process state of the entire production process. The process feature mark is like a process fingerprint of the production line, which can uniquely identify the production process state in a specific time period, and the process feature marks of different batches of products can be used for traceability and comparison analysis.

[0031] According to the process feature mark, a process state baseline is generated. The process feature mark data of multiple normal production batches is collected to establish a statistical sample library of process features. Statistical analysis is performed on the process feature mark data in the sample library to calculate the mean value and distribution range of each process feature value. The statistical mean value of each process feature value is connected in sequence according to the process order to form the center curve of the process state baseline. The normal fluctuation range of each process feature value is calculated, and the upper and lower boundaries are established on both sides of the center curve to form the normal interval of the process state baseline. In the copper plating process, the statistical mean value of the current density feature value is the baseline center, and the current density fluctuates within a certain range above and below the baseline center during normal production. Fluctuations beyond this range are considered abnormal. In the inner layer pattern transfer process, the normal fluctuation range of the exposure energy feature value is narrow, and slight deviation may lead to a decrease in line width accuracy, so the baseline interval corresponding to this process is set more strictly. After the process state baseline is established, it can be used for real-time process monitoring. The process feature mark of the current production is compared with the baseline to identify abnormal processes that deviate from the baseline. With the accumulation of production data, the process state baseline is updated regularly to reflect the parameter drift caused by equipment state changes and process optimization, maintaining the timeliness and accuracy of the baseline.

[0032] In step S120, process drift features are identified according to the process state baseline, and a process baseline library is constructed by performing dynamic threshold layering in the process drift features. Process deviation parameters are generated by performing offset detection via the process baseline library.

[0033] Specifically, the process capability review according to the process state baseline identifies the process drift characteristics. The process characteristics of the current production batch are compared with the process state baseline, and the deviation of each process characteristic value from the baseline center curve is calculated. The deviation is continuously tracked, and the law of the accumulation of the deviation over time is identified. If the deviation continuously increases in one direction, it indicates that the process parameters are drifting. The process capability index of each process characteristic value is calculated. The process capability index Cpk is determined by the relationship between the characteristic value distribution and the baseline normal interval. Cpk = min[(USL-μ) / 3σ,(μ-LSL) / 3σ], where USL and LSL are the upper and lower boundaries of the baseline, μ is the mean value of the characteristic value, and σ is the standard deviation of the characteristic value. When the process capability index is lower than the set threshold value, it is determined that there is a risk of process drift in the process. In the drilling process, when the hole position precision characteristic value continuously deviates to the upper boundary of the baseline for multiple batches, and the Cpk value gradually decreases from 1.33 to below 1.0, it is determined that the hole position precision parameter is drifting upward. In the lamination process, although the lamination temperature characteristic value is within the baseline range, it has shown a slow upward trend for five consecutive batches, with an increase of about 0.3°C per batch, indicating that the parameter has a drift tendency. Extract the drift direction, drift amplitude and change rate of each process, and integrate these drift information to form the process drift characteristics. The process drift characteristics include six dimensions of process identification, drift parameter type, drift direction, drift amplitude, change rate and process capability index, which completely describe the deviation of each process from the process state baseline.

[0034] In some embodiments, the dynamic threshold hierarchical construction of the process drift characteristics includes: obtaining the drift amplitude of the process drift characteristics and comparing it with the standard process specification to determine the process standard value; setting an adaptive threshold boundary according to the process standard value and the change rate of the process drift characteristics; dividing the process drift characteristics into different levels according to the adaptive threshold boundary to form a drift level classification; and performing a hazard degree evaluation process on the drift level classification to form a process benchmark library.

[0035] The drift amplitude of the process drift feature is compared with the standard process specification to determine the process standard value. The drift amplitude data of each process is extracted from the process drift feature, which represents the absolute distance of the current process parameter from the baseline center. The drift amplitude in the process drift feature is compared with the parameter tolerance range specified in the industry standard process specification, which is derived from the IPC standard or the enterprise internal control standard document. The proportion of the drift amplitude to the tolerance range is calculated, which reflects the severity of the current drift relative to the allowed deviation. In the copper plating process, the process drift feature shows that the current density drift amplitude is 0.5 A / dm², and the standard process specification specifies that the current density tolerance range is ±1.0 A / dm², and the drift amplitude accounts for 50% of the tolerance. According to the comparison result of the drift amplitude and the tolerance, combined with the process capability index in the process drift feature, the process standard value of each process parameter is determined. The process standard value is defined as the process parameter target value that can be stably achieved under the current equipment state and material conditions, which is within the tolerance range of the standard process specification and matches the current actual production capacity. In the lamination process, the standard process specification specifies that the lamination temperature is 180±5℃, and according to the process drift feature, the current equipment is slightly higher when it is stable, so the process standard value of this process is set to 181℃. The process standard values of each process are summarized to form a complete process standard value data set.

[0036] The adaptive threshold boundary is set according to the process standard value combined with the change rate of the process drift feature. The process standard value is used as the center reference of the threshold boundary, and the allowed deviation range is set on both sides of the standard value. The change rate data of each process parameter is extracted from the process drift feature, which reflects the speed of drift development. The boundary width on both sides of the standard value is dynamically adjusted according to the size of the change rate, and a narrower boundary is set for parameters with a faster change rate to provide early warning, and a wider boundary is set for parameters with a slower change rate to avoid frequent false alarms. The width calculation formula of the adaptive threshold boundary is W=W0×(1-k×v), where W is the adaptive boundary width, W0 is the basic boundary width, k is the rate adjustment coefficient, and v is the normalized change rate, all in dimensionless or consistent with the parameter. In the solder mask printing process, the process standard value is 25μm, and the basic boundary width is ±3μm. When the change rate is fast, the boundary is narrowed to ±2μm, and when the change rate is slow, the boundary remains ±3μm. In the drilling process, the change rate of hole position accuracy is fast, and the basic boundary width of ±20μm is narrowed to ±15μm to ensure that the hole position deviation can be found in time. The process standard value is used as the center to set two adaptive threshold boundaries, the warning boundary and the control boundary, respectively. The warning boundary is used to prompt the process parameter to start to deviate, and the control boundary is used to trigger forced intervention. The adaptive threshold boundaries of each process parameter are summarized, and each process parameter corresponds to a set of boundary data sets containing center value, warning upper and lower limits, and control upper and lower limits.

[0037] The process drift features are classified into drift level categories according to the adaptive threshold boundaries. The current drift amplitude of each process in the process drift features is compared with the adaptive threshold boundaries to determine the boundary interval where the drift amplitude falls. According to the interval position where the drift amplitude is located, the drift state is divided into different levels. When the drift amplitude fluctuates around the center value of the threshold boundary, it is classified as a normal level, indicating that the process parameters are in a stable and controlled state. When the drift amplitude exceeds the warning boundary but does not reach the control boundary, it is classified as a warning level, indicating that the process parameters begin to deviate and need attention. When the drift amplitude exceeds the control boundary, it is classified as an out-of-control level, indicating that the process parameters deviate seriously and need immediate intervention. In the drilling process, the adaptive threshold boundary warning limit of hole position accuracy is ±15μm, the control limit is ±25μm, and the current drift amplitude is 18μm, which exceeds the warning boundary but does not reach the control boundary. The hole position accuracy of this process is classified as a warning level. In the copper plating process, the current drift amplitude of current density is 1.2A / dm², which has exceeded the control boundary ±1.0A / dm², and the parameter is classified as an out-of-control level. The boundary comparison and level classification are performed for each parameter of each process one by one to form the drift level classification results covering all processes and parameters. The key degree weight of each process parameter in the drift level classification is evaluated, and the key degree weight is determined according to the influence of the parameter on the electrical performance, mechanical strength and reliability of the product. The weight of the circuit conduction related parameters is the highest, and the weight of the appearance related parameters is lower. The drift level classification is presented in table form, and each row corresponds to a process parameter. The column fields include process name, parameter type, current drift amplitude, belonging level and key degree weight.

[0038] The process benchmark library is formed by performing the hazard degree evaluation process for the drift level classification. The hazard degree of each process parameter in the drift level classification is evaluated to evaluate the influence degree of the parameter drift on the final product quality. The hazard degree evaluation considers two factors: the level severity of the parameter drift and the criticality of the parameter to the product function. The level information of each parameter is extracted from the drift level classification, and the uncontrolled level has the highest hazard degree, the warning level is followed, and the normal level has the lowest hazard degree. Combined with the function criticality weight of each parameter, the comprehensive hazard degree score is calculated, comprehensive hazard degree H = level coefficient x criticality weight, the level coefficient takes the value of normal 1, warning 3, and uncontrolled 5. The criticality weight is determined according to the influence degree of the parameter on the product electrical performance, mechanical strength, and reliability. The weight of the circuit conduction related parameter is the highest, and the weight of the appearance related parameter is relatively low. In the inner layer pattern transfer process, the line width precision belongs to the circuit function critical parameter, directly affects the circuit impedance and signal transmission quality, and the criticality weight is 0.9. The current level is the warning level, and the comprehensive hazard degree score is 3x0.9=2.7. In the solder mask printing process, the ink thickness mainly affects the appearance and protection performance, and the criticality weight is 0.5. Even if the level is the warning level, the comprehensive hazard degree score is only 3x0.5=1.5. According to the comprehensive hazard degree score, each parameter in the drift level classification is sorted to identify the key drift parameter with the highest hazard degree. The hazard degree evaluation result is integrated with the drift level classification data to establish the process benchmark library. The process benchmark library contains three types of information: the process standard value and the adaptive threshold boundary of each process parameter, the current drift level classification result, and the hazard degree evaluation sorting list. The process benchmark library is stored in the form of a database, and supports multiple retrieval methods such as process query, level filtering, and hazard degree sorting.

[0039] The process deviation parameter is generated by the deviation detection via the process benchmark library. The process standard value and adaptive threshold boundary data of each process parameter are called from the process benchmark library as the basis for determining the deviation detection. The process characteristic data of the current production is collected in real time, and the real-time data is compared with the process standard value of the corresponding process in the process benchmark library to calculate the deviation. The deviation calculation formula is D = X - X0, where D is the deviation, X is the current measured value, and X0 is the process standard value in the process benchmark library. The unit of the deviation is consistent with the unit of the process parameter. The deviation is compared with the adaptive threshold boundary in the process benchmark library to determine whether the deviation exceeds the allowed range. In the etching forming process, the etching rate process standard value is called from the process benchmark library as 3.5 μm / s, the adaptive threshold boundary is ±0.3 μm / s, the current measured value is 3.9 μm / s, and the deviation is +0.4 μm / s, which exceeds the threshold boundary and is determined as abnormal deviation. In the drilling process, the hole position precision process standard value is called from the process benchmark library as 0 μm, the adaptive threshold boundary warning limit is ±15 μm, the current measured hole position deviation is +12 μm, and the deviation does not exceed the warning boundary and is determined as normal. For the parameters detected as abnormal deviation, the hazard degree evaluation information of the parameter is extracted from the process benchmark library, combined with the deviation direction and the deviation amplitude, and the process deviation parameter is generated. The process deviation parameter includes four fields of deviated process, deviated parameter name, deviation direction and deviation. The abnormal deviation parameters detected in each process are summarized, sorted in descending order of hazard degree, and a complete list of process deviation parameters is formed.

[0040] In step S130, the inter-process correlation of the process state baseline is strengthened to determine the transfer coupling degree of adjacent processes. The transfer coupling degree is used to penetrate and track the defect transmission link across processes, and the process deviation parameter is cascaded and accumulated along the defect transmission link to form a process correlation matrix.

[0041] Specifically, the process state baseline is subjected to inter-process correlation enhancement processing to determine the transmission coupling degree of adjacent processes. The characteristic value time series data between adjacent processes is extracted from the process state baseline, and the correlation between the output characteristics of the previous process and the input characteristics of the next process is analyzed. The correlation coefficient of the characteristic values of adjacent processes is calculated, which reflects the degree of influence of the parameter change of the previous process on the parameter of the next process. The historical data in the process state baseline is subjected to sliding window analysis, and the probability and amplitude ratio of the characteristic values of the next process deviating after the characteristic values of the previous process deviate are calculated. Between the lamination process and the drilling process, the historical data in the process state baseline is extracted and analyzed, and it is found that whenever the lamination temperature is too high, the hole position deviation of the subsequent drilling process also increases accordingly. The reason is that uneven heating of the plate causes the internal stress distribution to change, and the plate deforms slightly during drilling, affecting the hole position accuracy. The correlation coefficient and the transmission probability are comprehensively calculated to obtain the transmission coupling degree between adjacent processes, and the transmission coupling degree C = | p | × P, where p is the correlation coefficient, | p | takes its absolute value to represent the correlation strength, and P is the transmission probability. The coupling degree ranges from 0 to 1. The correlation between all adjacent process pairs in the process state baseline is analyzed, and the transmission coupling degree between each pair of adjacent processes is calculated. The transmission coupling degree is stored in the form of a matrix, and the matrix rows and columns correspond to the previous and next processes, respectively. The element value is the coupling degree value of the corresponding process pair, and the higher the coupling degree, the more significant the influence of the previous process on the next process.

[0042] In some embodiments, the use of the transmission coupling degree to penetrate the defect transmission link across the process includes: generating a process topology graph according to the mapping of the transmission coupling degree to the process correlation space; obtaining an influence depth level from the transmission level analysis of the process topology graph; extracting a cross-process transmission channel by performing bidirectional correlation verification through the influence depth level; and forming a defect transmission link by superimposing the transmission strength of the cross-process transmission channel.

[0043] The process topology graph is generated according to the mapping of the transmission coupling degree to the process correlation space. Each process in the transmission coupling degree matrix is taken as a node, and the coupling relationship between the processes is taken as a connection edge to construct a directed graph structure. In the process correlation space, the positions of each process node are arranged in the order of the production process sequence, and the connection edge direction between the nodes indicates the transmission direction of the process parameters. The coupling degree values are extracted from the transmission coupling degree matrix, and the coupling degree values are mapped to the weights of the connection edges. The higher the coupling degree, the greater the weight of the edge. A coupling degree threshold is set, and no connection edge is established between the process pairs with a coupling degree lower than the threshold, and weakly associated relationships are filtered out to simplify the topology structure. In the PCB production line, the exposure parameters of the inner layer pattern transfer process have a significant impact on the interlayer alignment of the laminated layers, and a strong connection edge is established between the two processes. The solder mask printing process is far away from the inner layer pattern transfer process at the front end and has a weak parameter correlation, and no connection edge is established between the two. For non-adjacent process pairs that have a cross-process impact, if the transmission coupling degree data shows that the coupling degree exceeds the threshold, a cross-connection edge is also established. The nodes, connection edges and edge weights are integrated to form a complete process topology graph. The process topology graph intuitively presents the association strength and transmission direction between the processes, and provides a structured basis for subsequent transmission path analysis.

[0044] The influence depth level is obtained by analyzing the transmission level from the process topology graph. In the process topology graph, a target process node is selected, and the level depth of the process affected by the upstream processes is analyzed. Starting from the target process node, the upstream processes that can reach the target node are counted and the path length is traced in reverse along the connection edge. The upstream processes directly connected to the target process are divided into the first level, the more upstream processes connected to the first level processes are divided into the second level, and so on. When analyzing the influence sources of the copper plating process in the process topology graph, the drilling process is directly connected to the first level because the hole wall quality directly affects the copper plating effect; the laminated layer process is indirectly connected to the second level through the drilling process because the laminated layer quality affects the hole wall smoothness during drilling; and the inner layer pattern transfer process belongs to the third level because the integrity of the circuit pattern ultimately affects the plating uniformity through multi-stage transmission. The number of processes contained in each level and the corresponding connection edge weight are counted, and the deeper the level, the longer the transmission path and the more indirect the influence, but there may be cumulative effects. The process list and connection edge weight data of each level are integrated to form the influence depth level of the target process. The influence depth level describes the hierarchical structure of the target process affected by the upstream processes, and the level information is used to determine the source process of the forward tracking. The level of each process node in the process topology graph is analyzed to form a complete influence depth level data set.

[0045] Exemplarily, the cross-process transmission channel is extracted by performing the bidirectional correlation verification according to the influence depth level, including: performing forward transmission tracking according to the influence depth level to form a forward correlation path; performing reverse backtracking verification on the forward correlation path to form a reverse verification path; performing consistency comparison between the forward correlation path and the reverse verification path to screen an effective transmission segment; and performing channel integration according to the effective transmission segment to form the cross-process transmission channel.

[0046] The forward correlation path is formed by performing forward transmission tracking according to the influence depth level. A source process is selected as a tracking starting point from the influence depth level, and a parameter transmission trajectory is tracked downstream along the process flow direction. In the influence depth level structure, the source process at the deepest level is used as a starting point, and each intermediate level process is sequentially passed through until the target process is reached. Each process node passed through in the tracking process and the connection path between the nodes are recorded to form an ordered process sequence from the source to the target. When the inner layer pattern transfer process has a rough line edge caused by insufficient exposure energy, the downstream tracking is performed according to the influence depth level: the rough line edge causes uneven insulation distance between layers during lamination, which in turn causes uneven force on the drill bit during drilling due to differences in material density, resulting in hole position deviation, and finally, uneven thickness of the in-hole plating layer during copper plating. The transmission trajectory from the exposure problem to the plating defect is recorded as the forward correlation path. The transmission coupling degree values between each adjacent process in the forward correlation path are extracted, and the comprehensive transmission strength of the path is calculated. For the case where there are multiple selectable paths, each path is tracked and recorded to form multiple candidate forward correlation paths. The process sequence, transmission strength, and path length of each forward correlation path are summarized, and the forward correlation path describes the possible trajectory of process parameter transmission from the source process downstream.

[0047] The reverse backtracking verification of the forward correlation path forms a reverse check path. Starting from the terminal process of the forward correlation path, the upstream is traced in the reverse direction of the process flow to verify the bidirectional consistency of the transmission relationship. In the backtracking process, it is verified whether the parameter anomaly of the downstream process can indeed be traced back to the parameter change of the upstream process. Starting from the parameter deviation data at the end of the forward correlation path, the causal relationship between the deviation and the parameter of the previous process is analyzed, and if the causal relationship is established, the upstream is continued to be traced. When it is finally detected that the PCB product has a local open circuit fault, the reverse backtracking is performed along the forward correlation path: first, the copper plating process is checked, and it is found that the through-hole plating layer at the fault position is indeed thin; continue to trace the drilling process upstream, and it is found that the burr on the hole wall of the through-hole affects the copper plating adhesion; continue to trace the lamination process upstream, and it is found that the board in the area has a layered bubble, which causes burr during drilling, and the causal chain is verified step by step. Record the verified process nodes and connection relationships in the reverse backtracking process to form a reverse check path. Compare the process sequences of the forward correlation path and the reverse check path, and the part of the same process nodes of the two paths is marked as bidirectional verification passed. The reverse check path is similar in structure to the forward correlation path but opposite in direction, and is used to confirm that the transmission relationship obtained by the forward tracking is also established in the reverse causal analysis.

[0048] The forward correlation path and the reverse check path are compared for consistency to screen effective transmission sections. The process connection relationships in the forward correlation path and the reverse check path are compared section by section to identify the overlapping process pairs in the two paths. For each section of process connection in the forward correlation path, it is checked whether the connection exists in the reverse check path, and the connection that exists in both directions is marked as consistent. When analyzing the transmission relationship from drilling to copper plating, the forward correlation path shows that the hole wall roughness affects the plating quality, and the reverse check path also confirms that the plating defect position and the hole wall roughness position are completely matched, so the section connection passes the consistency comparison; but if the inner layer pattern in the forward correlation path is transferred to a section of the transmission relationship in the lamination process, and it is found in the reverse check path that the lamination problem actually originates from the difference in the pre-impregnated material batch rather than the inner layer pattern, then the section connection does not pass the consistency comparison. The consistency comparison results of each section of connection are counted, and the process connection sections that pass the bidirectional verification are extracted as effective transmission sections. For the connection sections that exist only in one-way path and cannot be verified in the other direction, they are determined as weak correlation or false correlation and are not included in the effective transmission sections. All process connection sections that pass the consistency comparison are summarized, and the effective transmission sections are stored in the form of a process pair list, each process pair containing the previous process, the next process, and the bidirectional verified transmission strength. The effective transmission section represents the real process parameter transmission relationship that has passed the bidirectional check and verification, and excludes the false correlation that is only statistically related in one direction but has no actual causal relationship.

[0049] The channel integration is performed according to the effective transmission segments to form a cross-process transmission channel. The process pairs with continuity in the effective transmission segments are integrated in series to form a complete transmission channel spanning multiple processes. The head-tail connection relationship of each process pair in the effective transmission segment list is checked. If the post-process of a process pair is the same as the pre-process of another process pair, the two process pairs are merged into a longer transmission chain. In the effective transmission segment, the transmission relationship of the lamination to the drilling process has been verified to be established, and the transmission relationship of the drilling process to the copper plating process has also been verified to be established. Since the drilling process is connected in the two process pairs, it is integrated into a three-process transmission channel from lamination to drilling to copper plating. The channel describes the complete path of how the plate quality problem affects the final coating quality step by step. The integrated transmission channel is optimized for path. Channels with the same start and end points but different intermediate paths are merged, and the main channel with the highest transmission strength is retained. The comprehensive transmission efficiency of each cross-process transmission channel is calculated. The comprehensive transmission efficiency is the product of the transmission strengths of each effective transmission segment in the channel. The process sequence, comprehensive transmission efficiency, and channel length of each transmission channel are integrated to form a cross-process transmission channel dataset. The cross-process transmission channel describes the complete path of the process parameters from the source process to the target process through multiple intermediate processes. The longer the channel, the wider the range of parameter transmission.

[0050] The transmission strength of the cross-process transmission channel is superimposed to form a defect transmission link. The transmission strength of each channel in the cross-process transmission channel is analyzed to evaluate the cumulative effect of the defect transmission along the channel. The comprehensive transmission efficiency of each channel is retrieved, which reflects the strength retention ratio of the defect signal after multiple process transmission. At each process node in the cross-process transmission channel, the defect signal may be amplified, maintained, or attenuated. The node transmission coefficient is determined according to the process characteristics of each node. When a local under-pressing occurs in the lamination process, causing a small bubble inside the plate, the defect propagates along the cross-process transmission channel to the drilling process. The heat and vibration generated by the high-speed rotation of the drill bit cause the material in the bubble area to crack, and the defect is amplified to a hole wall damage. When it continues to propagate to the copper plating process, the damaged hole wall cannot form a complete conductive seed layer, and the defect is further amplified to a plating cavity, which may eventually lead to a through-hole open circuit. The transmission coefficients of each node in the cross-process transmission channel are superimposed to obtain the comprehensive defect transmission strength of the channel. The cross-process transmission channel with a transmission strength exceeding a certain threshold is marked as a significant defect transmission path to form a defect transmission link. The defect transmission link is stored in a linked list structure, with process identifiers as nodes and transmission strength information between nodes. The defect transmission link clearly identifies the path and strength of the defect from the source process to the downstream processes, providing a basis for defect root cause analysis and propagation blocking.

[0051] A process deviation correlation matrix is formed by cascading and accumulating process deviation parameters along the defect transfer link. Four field information is extracted from the process deviation parameters of the source process node at the head of the defect transfer link: deviation process determines the location where the deviation occurs, deviation parameter name determines the parameter type, deviation direction indicates whether the parameter is high or low, and deviation amount is the initial deviation value. The deviation is transferred downstream along the defect transfer link, and at each intermediate process node, the upstream accumulated deviation value is superimposed with the deviation amount of the process deviation parameters of the current process, taking into account the transfer intensity coefficient. The formula for calculating the accumulated deviation value is A_n=A_{n-1}×T_n+D_n, where A_n is the accumulated deviation value of the nth process, A_{n-1} is the upstream accumulated value, T_n is the transfer intensity coefficient, D_n is the deviation amount of the current process, and each term is a signed normalized value, with the sign indicating the deviation direction. In actual production, when the inner pattern transfer process has a process deviation parameter of low exposure energy, the deviation is transferred downstream along the defect transfer link, and at the lamination process, the deviation is superimposed with the high temperature deviation of the process itself due to the influence of interlayer alignment. The two deviations in opposite directions partially offset each other. In the drilling process, the deviation is continuously superimposed due to the influence of the positioning reference, and the accumulated deviation value reflects the linkage effect and direction superposition result of multi-process deviation. The accumulated deviation values of each process on the defect transfer link are recorded, and the corresponding deviation parameter name is also recorded, forming a sequence of accumulated deviation values transferred along the process. A matrix structure is constructed with processes as rows and columns, and the matrix element value is the accumulated deviation transfer amount between the corresponding process pairs. The cascading and accumulation results of each defect transfer link are filled into the corresponding positions of the matrix to form a complete process correlation matrix. The row of the process correlation matrix represents the source process, the column represents the target process, the element value represents the influence of the deviation of the source process on the target process after accumulation through the defect transfer link, and the sign of the element reflects the direction of the accumulated deviation.

[0052] In step S140, defect diffusion backtracking and influence domain aggregation are performed on the process correlation matrix to determine the defect coverage range, process state parameters are generated based on the defect coverage range, and a process feature map is constructed by extracting trend turning features from the process state parameters.

[0053] In some embodiments, the defect diffusion backtracking and influence domain aggregation on the process correlation matrix to determine the defect coverage range includes: generating node contribution degrees based on the process correlation matrix; tracking defect development based on the node contribution degrees to obtain a diffusion trajectory; generating a tracking chain by backtracking from the diffusion trajectory and combining the node contribution degrees; and classifying and aggregating influence domains along the tracking chain to determine the defect coverage range.

[0054] The node contribution degree is generated according to the process correlation matrix. The row vector and column vector data of each process node are extracted from the process correlation matrix, the output contribution degree of each process node is calculated, the output contribution degree is equal to the sum of the element values in the row of the process correlation matrix, and the output contribution degree reflects the ability of the process as a defect source to pass down the influence. The input contribution degree of each process node is calculated, the input contribution degree is equal to the sum of the element values in the column of the process correlation matrix, and the input contribution degree reflects the degree of receiving upstream defect influence. The output contribution degree and the input contribution degree are comprehensively weighted to obtain the comprehensive node contribution degree of each process, and the comprehensive contribution degree N = a x O + b x I, wherein O is the output contribution degree, I is the input contribution degree, a and b are weight coefficients and a + b = 1, and each term is a normalized dimensionless value. In the PCB production line, the node contribution degree of the drilling process is very special. This process not only undertakes the problems of board internal stress and interlayer alignment degree passed by the lamination process, but also passes the problems of hole wall roughness and hole position accuracy to the copper plating and solder mask printing processes. The input contribution degree and the output contribution degree are at a high level, and the comprehensive node contribution degree is in the forefront among all processes, like a crossroads in the production process. Any defect passing through this process may be amplified and spread in multiple directions. The node contribution degrees of each process are sorted according to the numerical value, and the key process node with the highest contribution degree is identified.

[0055] The diffusion trajectory is obtained by tracking the development of defects based on the node contribution degree. The process with the highest output contribution degree is selected as the starting point of tracking according to the node contribution degree, and the development path of the defect is tracked along the process flow direction. In the tracking process, the downstream process with a higher node contribution degree is preferentially selected as the next hop node to construct the main path of defect transmission. When a bubble appears in the dry film attachment of the inner layer pattern transfer process, causing a pinhole defect in the circuit pattern, the node contribution degree is used to track downstream: the pinhole defect is first transmitted to the lamination process, and due to the poor combination of copper foil and resin at the pinhole position, a small interlayer gap is formed after lamination; the gap defect continues to be transmitted to the drilling process, and the drill bit is deflected when passing through the gap area due to the sudden change in material density; the deflected via hole is transmitted to the copper plating process, and the inclined hole wall causes uneven current distribution during plating, resulting in a difference in plating thickness, and finally the via resistance exceeds the standard in the finished product test. The diffusion trajectory is formed by recording each process node passed in the tracking process, and the node contribution degree value of each node is included in the diffusion trajectory. For the case of branches, each branch path is tracked, and the complete diffusion trajectory describing the overall development of the defect is integrated.

[0056] For example, the step of generating a traceability chain by tracing back the source from the diffusion trajectory and combining the node contribution includes: establishing a time-reverse correlation based on the diffusion trajectory; setting multiple backtracking verification points along the time-reverse correlation; constructing a defect transmission chain in reverse through the backtracking verification points; and forming a traceability chain by weighting and connecting the defect transmission chain with the node contribution.

[0057] A time-reverse correlation is established based on the diffusion trajectory. Following the chronological order of the process flow, a reverse correlation is established starting from the final node of the diffusion trajectory. The timestamps of each process node in the diffusion trajectory are arranged in ascending order, and a time-reverse correlation is established between adjacent process nodes, with the connection direction pointing from the later process to the earlier process. When the final inspection process discovers large-area blistering and peeling of the solder mask layer in a batch of PCBs, a time-reverse correlation is established based on the diffusion trajectory: tracing back from the final inspection time to the curing time of the solder mask printing process, which is responsible for coating and curing the protective ink on the board surface; then tracing back to the exit time of the copper plating process, which is responsible for forming a conductive copper layer on the inner wall of the through-holes and the board surface; continuing to tracing back to the completion time of the drilling process, which is responsible for drilling various through-holes and blind holes on the board; finally, tracing back to the exit time of the lamination process, which is responsible for laminating the multilayer circuit board into a whole. The time-reverse correlation links each process together, forming a complete time-reverse correlation structure arranged in reverse chronological order, recording the timestamp information of each process node in the correlation.

[0058] Multiple backtracking verification points are set up along a time-reverse correlation. Key locations are selected within the time-reverse correlation process sequence to set backtracking verification points, which are used to verify whether defects actually exist at those locations. Processes with high node contribution in the time-reverse correlation are selected as backtracking verification points, as these processes play a pivotal role in defect propagation. Appropriate intervals are maintained between adjacent backtracking verification points to avoid overly dense verification points causing verification redundancy and increasing traceability costs, and also to avoid overly sparse verification points missing key defect propagation links. For the backtracking analysis of solder resist blistering defects, the first backtracking verification point is set in the copper plating process. This verification point needs to check whether there is oxide film or oil residue on the plating surface causing poor adhesion to the solder resist ink; the second backtracking verification point is set in the drilling process. This verification point needs to check whether there are burrs, edge curls, or copper shavings around the hole opening affecting the uniformity of subsequent coating; the third backtracking verification point is set in the lamination process. This verification point needs to check whether there is resin precipitation or copper foil wrinkling on the board surface affecting coating adhesion. Configure verification rules for each backtracking verification point. The verification rules clearly specify the types of parameters to be checked, the threshold conditions for determining the existence of defects, and the source of verification data.

[0059] The defect conduction chain is reversely constructed by backtracking the check points. The defect existence verification is performed at each backtracking check point, and it is determined whether the position is an effective node of the defect conduction chain according to the check result. Starting from the backtracking check point closest to the defect discovery point, the historical process data and quality detection records of the process are called and compared with the check rules, and the data sources include process parameter acquisition devices, online detection equipment and manual sampling records. If the check passes, the process is confirmed as a node of the defect conduction chain and the verification is continued to the next backtracking check point upstream, and if the check fails, it indicates that the defect source is located between the check point and the previous check point, and temporary check points need to be added in the interval to further narrow down the range. In the backtracking of the solder resist blistering defect, the check point of the copper plating process calls the plating layer surface roughness and cleanliness detection records, and it is found that there is indeed a slight oxide film on the plating layer surface of this batch and the cleanliness index is at the lower limit edge, the check passes, and the copper plating is confirmed as a node of the defect conduction chain; the check point of the drilling process calls the hole mouth burr detection image, and it is found that there are burrs on the hole mouth of some through holes and the burr height exceeds the standard limit, the check passes, and the drilling process is confirmed as a node of the defect conduction chain; the check point of the lamination process calls the plate appearance detection records and surface flatness data, and all the indexes are within the normal range, the check fails, and it is determined that the defect source is located in the drilling process rather than the lamination process upstream. The process nodes that pass the check are connected in reverse order of time to form a complete defect conduction chain from the defect discovery point to the defect source.

[0060] The defect conduction chain is reversely constructed by backtracking the check points. The defect existence verification is performed at each backtracking check point, and it is determined whether the position is an effective node of the defect conduction chain according to the check result. Starting from the backtracking check point closest to the defect discovery point, the historical process data and quality detection records of the process are called and compared with the check rules, and the data sources include process parameter acquisition devices, online detection equipment and manual sampling records. If the check passes, the process is confirmed as a node of the defect conduction chain and the verification is continued to the next backtracking check point upstream, and if the check fails, it indicates that the defect source is located between the check point and the previous check point, and temporary check points need to be added in the interval to further narrow down the range. In the backtracking of the solder resist blistering defect, the check point of the copper plating process calls the plating layer surface roughness and cleanliness detection records, and it is found that there is indeed a slight oxide film on the plating layer surface of this batch and the cleanliness index is at the lower limit edge, the check passes, and the copper plating is confirmed as a node of the defect conduction chain; the check point of the drilling process calls the hole mouth burr detection image, and it is found that there are burrs on the hole mouth of some through holes and the burr height exceeds the standard limit, the check passes, and the drilling process is confirmed as a node of the defect conduction chain; the check point of the lamination process calls the plate appearance detection records and surface flatness data, and all the indexes are within the normal range, the check fails, and it is determined that the defect source is located in the drilling process rather than the lamination process upstream. The process nodes that pass the check are connected in reverse order of time to form a complete defect conduction chain from the defect discovery point to the defect source.

[0061] The defect propagation range is determined by aggregating the influence domains along the traceability chain. The direct influence domain of each node on the traceability chain is determined according to the process correlation matrix, which contains all downstream processes having a direct transmission relationship with the node. The direct influence domains of the nodes on the traceability chain are summarized to form the influence domain set of the entire chain. The processes in the influence domain set are classified and analyzed. The processes affected by multiple chain nodes are subject to the cumulative effect of defects from multiple sources, and the risk level is the highest, marked as high-risk processes, which are included in the core area of the defect propagation range. The processes affected by only a single node have a single defect source, and the risk is relatively controllable, marked as general affected processes, which are included in the peripheral area of the defect propagation range. In the traceability chain with drilling as the source, the influence domain of the drilling process covers three downstream processes: copper plating, outer pattern transfer, and solder mask printing. The influence domain of the copper plating process covers two downstream processes: outer pattern transfer and solder mask printing. Outer pattern transfer and solder mask printing appear in both influence domains, indicating that these two processes are simultaneously affected by hole wall defects and plating layer defects, and are marked as high-risk processes that need to be monitored. The boundaries of each influence domain are aggregated to merge overlapping areas and integrate dispersed areas, eliminating redundant boundaries to form a unified defect propagation range. The defect propagation range is presented in two layers: a high-risk process list and a general affected process list.

[0062] The process state parameters are generated according to the defect propagation range. For each affected process in the defect propagation range, the real-time parameter value and historical parameter statistical value of the process are retrieved from the production monitoring database. The current parameter value of each process in the defect propagation range is compared with its normal baseline value to calculate the deviation percentage, which reflects the deviation of the current state of the process from the normal state. Parameters with a deviation percentage exceeding a preset threshold are marked as abnormal parameters and included in the process state parameters. Parameters that do not exceed the threshold are marked as normal parameters. When the defect propagation range includes two high-risk processes: drilling and copper plating, the drilling process needs to focus on whether the spindle speed is forced to increase due to tool wear, whether the feed rate is adjusted due to changes in plate hardness, and whether the hole position deviation shows a one-way drift trend. The copper plating process needs to focus on whether the current density needs to be compensated due to plating solution aging, whether the plating solution temperature fluctuates due to environmental changes, and whether there are batch-to-batch differences in plating layer thickness. The deviation percentages of the parameters of each process are adjusted by weighting in combination with the influence degree information recorded in the defect propagation range. The parameters of processes with high influence degrees are given higher weights, and the process state parameters are integrated to record the parameter deviation status of each process in the propagation range.

[0063] In some embodiments, the extracting trend turning features in the process state parameters to construct a process feature map comprises: generating a process variation curve graph according to the process state parameters; detecting gradient mutation from the process variation curve graph to track a process parameter variation trajectory; locating an inflection point value where the process parameter variation trajectory turns from normal to abnormal as a trend turning feature; and performing boundary clustering processing on the trend turning feature to generate a process feature map.

[0064] A process variation curve graph is generated according to process state parameters. Time series data of each parameter is extracted from the process state parameters, and a curve of variation of each parameter with time is plotted with time as the horizontal axis and parameter value as the vertical axis, each data point on the curve corresponding to a measured parameter value at a sampling time. Multiple associated parameter curves of the same process are superimposed in the same process variation curve graph, facilitating observation of the linkage change relationship between parameters. For process state parameters of a drilling process, a spindle speed curve, a feed rate curve and a hole position deviation curve are plotted in the same process variation curve graph. When the spindle speed is forced to increase due to tool wear to maintain cutting efficiency, it is observed whether the feed rate is simultaneously reduced to reduce tool load, and whether the hole position deviation systematically drifts due to speed variation. The normal upper and lower boundary lines of each parameter are marked in the process variation curve graph, and the boundary lines divide the curve space into normal and abnormal intervals, facilitating intuitive identification of the period when the parameter exceeds the normal range. Moving average filter smoothing processing is performed on the curve data to eliminate the interference of sampling noise on trend analysis, and the variation curves of each parameter of each process are integrated to form a complete process variation curve graph.

[0065] The gradient mutation is detected from the process variation graph to track the parameter change trajectory. The gradient value of each curve in the process variation graph is calculated, and the gradient value G=(V_{t+1}-V_t) / Δt reflects the rate of change of the parameter value with time, where V_t and V_{t+1} are the parameter values at adjacent times, Δt is the time interval, and the dimension of the gradient is the parameter unit divided by the time unit. The gradient sequence of each curve in the process variation graph is scanned, and the position where the gradient value suddenly changes is identified. The gradient mutation indicates that the change rate of the parameter has changed significantly, which may correspond to the switching of the equipment state, manual intervention operation or abnormal event. When the gradient difference between adjacent times exceeds the threshold, it is marked as a mutation point of the process parameter change trajectory. In the current density curve of the copper electroplating process, the current density slowly decreases with the consumption of copper ions in the plating solution during normal production, and the gradient is negative and small in absolute value. When the operator adds a replenishing solution to the plating tank to adjust the concentration, the current density quickly rises, the gradient suddenly changes from negative to positive, and the absolute value increases. The gradient mutation point is detected and included in the process parameter change trajectory. If the amount of replenishing solution added is too large, the concentration will exceed the standard, the current density will continue to rise and exceed the upper limit of the process, forming another mutation point to mark the parameter from normal to abnormal. The gradient mutation points detected on each curve are connected in time sequence to form a complete process parameter change trajectory describing the evolution of the parameter from one state to another.

[0066] The inflection point value where the positioning process parameter change trajectory changes from normal to abnormal is used as the trend turning feature. Among the mutation points of the process parameter change trajectory, the key inflection points with state conversion significance are screened, the parameter states before and after each mutation point are analyzed, and it is judged whether the mutation point corresponds to the crossing of the parameter from the normal range to the abnormal range. For the case where the parameter value in the process parameter change trajectory is in the normal range before the mutation point and enters the abnormal range after the mutation point, the mutation point is the state conversion inflection point, which represents the critical time when the process state changes from controllable to uncontrollable. The parameter value at the state conversion inflection point is extracted as the feature value of the trend turning feature, which represents the critical level where the parameter changes from normal to abnormal. In the temperature curve of the lamination process, the temperature rises smoothly according to the preset temperature rising curve during the heating stage. When the local aging power attenuation of the heating element occurs, the temperature rising rate slows down and deviates from the preset curve. At a certain time, the temperature stops rising and stabilizes at a level lower than the minimum temperature required by the process. This time is the state conversion inflection point, and the temperature value at the inflection point is the feature value of the trend turning feature. The temperature below this value will cause incomplete curing of the resin and affect the performance of the board. The state conversion inflection points of each parameter are collected to form a trend turning feature data set, which completely records the critical turning level of each parameter.

[0067] The boundary clustering processing is performed on the trend turning feature to generate a process feature map. The feature values in the trend turning feature dataset are clustered according to similarity, and a parameter group with similar trend turning features or affected by the same factor is identified. The parameters with similar turning features are divided into the same cluster, and the parameters in the cluster often have an internal physical correlation or are driven by a common external factor. In multiple processes of a PCB production line, the exposure energy of inner layer pattern transfer is sensitive to environmental humidity, the curing temperature of lamination stacking is sensitive to workshop temperature, and the ink viscosity of solder mask printing is sensitive to environmental temperature and humidity. The trend turning feature values of the three parameters have similar deviation modes and are related to environmental conditions. The three parameters are clustered into an environmental sensitive parameter cluster and included in the process feature map. When the air conditioning in the workshop fails and the temperature and humidity are out of control, these parameters may simultaneously show a turning trend. The spindle speed of drilling, the feed rate, and the cathode moving speed of copper plating are all related to the mechanical state of the equipment. The three parameters are clustered into an equipment state parameter cluster and included in the process feature map. When the equipment has not been maintained for several months, these parameters may show a turning trend in succession. A boundary range is set for each parameter cluster. The lower limit of the boundary is the minimum value of the trend turning feature values of the parameters in the cluster, and the upper limit is the maximum value. The parameter clusters are mapped into a two-dimensional map space with the parameter type as one dimension and the process position as the other dimension, to form a process feature map. The process feature map directly presents the turning feature distribution and boundary range of each parameter in each process.

[0068] In step S150, a defect amplification factor is extracted by decoupling the sensitivity of the process deviation parameter and the process association matrix. A process critical node is identified according to the defect amplification factor. A regulation rule library is constructed around the process critical node. Process feature map is integrated through the regulation rule library to output process control instructions.

[0069] In some embodiments, the sensitivity decoupling of the process deviation parameter and the process association matrix to extract the defect amplification factor includes: performing orthogonal decomposition on the process deviation parameter and the process association matrix to form a sensitivity coefficient; determining a mutation division point from the sensitivity coefficient; performing nonlinear gain processing through the mutation division point to form a gain ratio; and forming a defect amplification factor based on the gain ratio.

[0070] The process deviation parameters are orthogonally decomposed according to the process procedure correlation matrix to form sensitivity coefficients. Four field information of each process procedure is extracted from the process deviation parameters, the deviation procedure and the deviation parameter name determine the analysis object, the deviation direction determines the vector sign, and the deviation amount determines the vector modulus. The process deviation parameters are constructed into a deviation vector according to the above information. The transfer relationship data in the process procedure correlation matrix is constructed into a correlation vector. The deviation vector and the correlation vector are orthogonally decomposed to separate the coupling relationship between the process deviation parameters and the transfer relationship of the process procedure correlation matrix. The projection component of the deviation vector in the direction of the correlation vector is calculated. The projection component reflects the part of the process deviation parameters related to the correlation transfer. The orthogonal component of the deviation vector perpendicular to the correlation vector is calculated. The orthogonal component reflects the part of the deviation parameters independent of the correlation transfer. In the lamination procedure, there is a compensation relationship among the temperature deviation, the pressure deviation, and the pressure holding time deviation. When the temperature is low, the pressure holding time can be extended to compensate for the deviation. The deviation parameters are constructed into a multi-dimensional vector, and the orthogonal decomposition is performed on the multi-dimensional vector and the influence vector in the correlation matrix to separate the sensitive components of each parameter. The sensitivity coefficients of each parameter are calculated according to the orthogonal decomposition results. The sensitivity coefficient S = |P_parallel| / |D|, wherein P_parallel is the projection length of the deviation vector in the correlation direction, and D is the modulus of the deviation vector. Each item is a dimensionless value. The higher the sensitivity coefficient is, the easier the parameter deviation is amplified and transferred.

[0071] The critical point is located according to the sensitivity coefficients to determine the mutation division point. The coefficient values in the sensitivity coefficient data set are analyzed and distributed. Each parameter is sorted according to the sensitivity coefficient value from small to large, and the change trend of the sorted coefficient value is analyzed. In the sensitivity coefficient sequence, the coefficient values of most parameters are at a low level and change gently, indicating that the deviation of these parameters is basically maintained after the process transfer and will not be amplified. The coefficient values of a few parameters suddenly jump to a high level, indicating that the deviation of these parameters will be significantly amplified. The position where the jump occurs is the mutation division point. In the sensitivity analysis of all process parameters, the sensitivity coefficients of most parameters such as the ink thickness of solder mask printing and the plating solution temperature of copper plating are concentrated in the low interval. The impact of these parameters can be basically controlled even if there is a deviation, like the ripples caused by small stones in a river that subside quickly. The sensitivity coefficients of a few parameters such as the feed rate of drilling and the temperature curve slope of lamination suddenly increase to several times of the low interval. The slight deviation of these parameters will grow like a snowball, and the position of these parameters is located as the mutation division point. The mutation division point clearly divides the parameters into a low sensitivity group and a high sensitivity group. Parameters exceeding the mutation division point need to be focused on.

[0072] The gain ratio is formed by performing nonlinear gain processing on the mutation boundary. According to the mutation boundary, the parameters are divided into different sensitive intervals. The parameters below the mutation boundary are processed in a linear relationship, and the parameters above the mutation boundary are analyzed for the nonlinear function relationship between the deviation amplitude and the influence amount. The nonlinear gain curve usually takes the form of an exponential or power function, and when the deviation amplitude exceeds the mutation boundary, the influence amount grows at an accelerating rate. In the feed rate parameter of the drilling machining process, when the feed rate fluctuates slightly within the normal range, the hole wall quality changes linearly with the feed rate, like driving at a slightly changed speed on a flat road has little effect on the journey; when the feed rate exceeds the critical value and continues to increase, the drill cutting force rises sharply, causing the hole wall to tear instead of cutting, like driving too fast causing the brakes to fail, the consequences worsen sharply, and the hole wall quality decays exponentially. According to the nonlinear gain curve, the gain ratio of each high-sensitive parameter at the current deviation level is calculated. The gain ratio is equal to the ratio of the nonlinear gain value to the linear gain value. A gain ratio greater than 1 indicates an amplification effect, and a higher gain ratio indicates a more significant amplification effect.

[0073] The defect amplification factor is formed based on the gain ratio. Parameters with a gain ratio greater than a set threshold are selected, and these parameters have significant defect amplification characteristics and are candidates for defect amplification factors. The selected high-gain ratio parameters are sorted by ratio size, and each high-gain parameter is assigned a defect amplification factor level. In the sensitivity analysis of the PCB production line, the feed rate parameter of the drilling machining process has the highest gain ratio. A slight deviation in this parameter can cause serious hole wall tearing problems and amplify along the transmission link to form copper plating voids and solder mask ink leakage in the plating layer and printing, like the butterfly effect, where a small disturbance causes a major disaster. Marked as a third-level defect amplification factor; the heating curve slope parameter of the lamination process has the second highest gain ratio. Deviation in this parameter can cause uneven resin flow, leading to a chain of problems in subsequent processes, marked as a second-level defect amplification factor; the development time parameter of the inner layer pattern transfer has a relatively low gain ratio but still exceeds the threshold, marked as a first-level defect amplification factor. The process identification, parameter name, gain ratio, and amplification factor level of each parameter are integrated to form a complete defect amplification factor dataset.

[0074] The process critical node is identified according to the defect amplification factor. The number and grade distribution of defect amplification factors contained in each process are counted, and the process containing high-grade defect amplification factors or containing multiple defect amplification factors is marked as a candidate process critical node. The position of the candidate process critical node in the production process is analyzed, and the candidate process located in the middle of the process and affecting multiple downstream processes has higher criticality. The drilling process contains two high-grade defect amplification factors of feed rate and spindle speed, and the process connects the plate processing section and the surface treatment section, like a hub site in the traffic network that will cause multiple lines to be interrupted once it is paralyzed. The quality of the process directly determines whether the subsequent copper plating can form reliable interlayer conduction, whether the outer layer pattern transfer can realize accurate pattern alignment, and whether the solder mask printing can obtain good ink adhesion. The drilling process is determined as the process critical node. The criticality score of each candidate process is calculated by comprehensively considering the number, grade and position of defect amplification factors, and the process whose score exceeds the threshold is determined as the process critical node, which is the object of subsequent key regulation.

[0075] A regulation rule base is constructed around the process critical node. For each process in the process critical node, the normal range and critical boundary of the defect amplification factor parameter are analyzed, and the trigger condition of the regulation rule base is set according to the distance between the current state and the boundary of the parameter. When the parameter value of the process critical node approaches the critical boundary, a warning regulation is triggered, and when the parameter value exceeds the critical boundary, a forced regulation is triggered. These regulation logics are recorded in the regulation rule base. Specific rules in the regulation rule base are formulated for the drilling process, which is a process critical node: when the feed rate parameter is adjusted by the operator to approach the upper limit boundary due to rush demand, like a driver driving at high speed to approach a dangerous area to save time, a speed reduction regulation instruction is triggered to reduce the feed rate to a safe range and send a production warning to the scheduling platform; when the spindle speed continuously rises due to tool wear compensation, like an engine continuously increasing the speed to compensate for power deficiency, indicating that a failure is approaching, a tool replacement reminder instruction is triggered to suggest replacing the drill bit to avoid continuous speed rise; when the hole position deviation shows a one-way drift trend in continuous batches, like a vehicle steering wheel showing a drift indicating that the chassis needs to be calibrated, a device calibration instruction is triggered to arrange the machine tool for positioning accuracy detection. The mapping relationship between the regulation rules in the regulation rule base and the defect types is established, the temperature type amplification factor is configured with a temperature control adjustment strategy, and the speed type amplification factor is configured with a motion parameter adjustment strategy. All regulation rules are indexed and organized according to the process critical node process and parameter type to form a complete regulation rule base.

[0076] The process characteristic map output process control instructions are integrated via the regulation rule base. Trend turning feature values of each process parameter in the process characteristic map are matched with trigger conditions in the regulation rule base, process parameters and process control instructions are generated which need to be started. When the process characteristic map shows that the slope of the temperature rising curve of the pressing lamination process is approaching the trend turning boundary, like the oil temperature is approaching the red line area on the instrument panel, indicating that the heating element may have power attenuation, the temperature rising abnormal regulation rule is matched in the regulation rule base, the heating power compensation action specified in the rule is extracted to generate the corresponding process control instruction. When the process characteristic map shows that the hole deviation of the drilling process has crossed the turning boundary into the abnormal interval, like the alarm light has turned on indicating that the problem has occurred, the precision out-of-control regulation rule is matched in the regulation rule base, the emergency shutdown maintenance action specified in the rule is extracted to generate the corresponding process control instruction. The regulation action is converted into specific process control instructions, and the process control instruction content includes target process, target parameter, adjustment direction and adjustment amplitude. The generated process control instructions are sorted according to priority, and the instructions for process critical nodes are executed first. The sorted process control instructions are sent to the execution device of each process, and the execution device adjusts the process parameters to realize closed-loop control of process deviation according to the instructions.

[0077] In order to perform the above-mentioned method embodiment corresponding to a kind of PCB multi-process processing collaborative control method, to realize corresponding function and technical effect. Referring to Figure 2 , Figure 2 The structure block diagram of a kind of PCB multi-process processing collaborative control system 200 provided by the embodiment of the application is shown. For ease of illustration, only the part related to the embodiment is shown, and the PCB multi-process processing collaborative control system 200 provided by the embodiment of the application includes:

[0078] The data acquisition module 201 is used to collect multi-source process data of each process of PCB production line, and multi-source mutual verification filtering is formed into process state baseline for the multi-source process data;

[0079] The reference detection module 202 is used to identify process drift characteristics according to the process state baseline, and process reference library is constructed by dynamic threshold layering in the process drift characteristics, and process deviation parameters are generated by offset detection via the process reference library;

[0080] The correlation analysis module 203 is used to determine the transfer coupling degree of adjacent processes by strengthening the process state baseline inter-process correlation, and the transfer coupling degree is used to penetrate and track the defect transmission link across processes, and the process deviation parameters are cascaded and accumulated along the defect transmission link to form a process correlation matrix;

[0081] The atlas constructing module 204 is configured to perform defect diffusion backtracking and impact domain aggregation on the process correlation matrix to determine a defect spread range, generate a process state parameter according to the defect spread range, extract a trend turning feature from the process state parameter, and construct a process feature atlas;

[0082] The instruction output module 205 is configured to perform sensitivity decoupling on the process deviation parameter and the process correlation matrix to extract a defect amplification factor, identify a process critical node according to the defect amplification factor, construct a regulation rule library around the process critical node, and output a process control instruction by integrating the process feature atlas via the regulation rule library.

[0083] The PCB multi-process machining collaborative control system 200 described above can implement a PCB multi-process machining collaborative control method according to the method embodiment described above. The optional items in the method embodiment described above are also applicable to the present embodiment, and will not be described in detail here. The remaining content of the present embodiment can be referred to the content of the method embodiment described above, and will not be described in detail in the present embodiment.

[0084] The purpose of the above embodiments is to exemplarily reproduce and deduce the technical solutions of the present application, and to completely describe the technical solutions, purposes and effects of the present application. The purpose is to make the public understand the disclosure of the present application more thoroughly and comprehensively, and does not limit the protection scope of the present application.

[0085] The above embodiments are also not based on an exhaustive enumeration of the present application. In addition to this, there can be many other unlisted embodiments. Any substitution and improvement made without violating the concept of the present application is within the protection scope of the present application.

Claims

1. A method of collaborative control of multi-process PCB processing, characterized by, The method comprises the following steps: Collecting multi-source process data of each process of a PCB production line, performing multi-source mutual verification filtering on the multi-source process data to form a process state baseline; Performing process capability review according to the process state baseline to identify process drift characteristics, performing dynamic threshold layering on the process drift characteristics to construct a process reference library, and performing offset detection via the process reference library to generate process deviation parameters; Performing inter-process correlation enhancement processing on the process state baseline to determine the transfer coupling degree of adjacent processes, using the transfer coupling degree to perform cross-process penetration tracking to locate a defect transmission link, cascading and accumulating the process deviation parameters along the defect transmission link to form a process correlation matrix; Performing defect diffusion backtracking and impact domain aggregation on the process correlation matrix to determine a defect spread range, generating a process state parameter according to the defect spread range, and extracting a trend turning feature in the process state parameter to construct a process feature map; Performing sensitivity decoupling on the process deviation parameters and the process correlation matrix to extract a defect amplification factor, identifying a process critical node according to the defect amplification factor, constructing a regulation rule library around the process critical node, and integrating the process feature map via the regulation rule library to output a process control instruction.

2. The method of claim 1, wherein, The process state baseline formed by performing multi-source mutual verification filtering on the multi-source process data comprises the following steps: According to the multi-source process data, a process data sequence is formed according to the process flow; From the process data sequence, a multi-source mutual verification filtering is performed to locate a core process anchor point; The core process anchor point is taken as a track reference to establish a process feature imprint; According to the process feature imprint, a process state baseline is generated.

3. The method of claim 1, wherein, The process reference library constructed by performing dynamic threshold layering on the process drift characteristics comprises the following steps: The drift amplitude of the process drift characteristics is compared with the standard process specification to determine a process standard value; According to the process standard value, an adaptive threshold boundary is set in combination with the change rate of the process drift characteristics; According to the adaptive threshold boundary, the process drift characteristics are divided into layers to form a drift level classification; A hazard degree evaluation processing is performed on the drift level classification to form a process reference library.

4. The method of claim 1, wherein, The cross-process penetration tracking to locate a defect transmission link by using the transfer coupling degree comprises the following steps: According to the transfer coupling degree, a process topology graph is generated by mapping to a process correlation space; From the process topology graph, a transmission level is obtained by performing transmission level analysis; Through the influence depth level, a cross-process transmission channel is extracted by performing bidirectional correlation verification; The defect transmission link is formed by performing transmission strength superposition on the cross-process transmission channel.

5. The method of claim 1, wherein, The defect spread range is determined by performing defect diffusion backtracking and impact domain aggregation on the process correlation matrix, which comprises the following steps: According to the process correlation matrix, a node contribution degree is generated; Based on the node contribution degree, a diffusion track is obtained by performing defect development tracking; From the diffusion track, a backtracking chain is generated by performing source backtracking in combination with the node contribution degree; The defect spread range is determined by performing impact domain classification and aggregation along the backtracking chain.

6. The method of claim 1, wherein, The process feature map is constructed by extracting a trend turning feature in the process state parameter, which comprises the following steps: Generating a process variation curve according to the process state parameter; Detecting gradient mutation from the process variation curve to track the process parameter change trajectory; Positioning the inflection point value where the process parameter change trajectory changes from normal to abnormal as a trend turning feature; Generating a process feature map by boundary clustering processing according to the trend turning feature.

7. The method of claim 1, wherein, The sensitivity decoupling extraction of the process deviation parameter and the process correlation matrix includes: Orthogonal decomposition is performed according to the process deviation parameter and the process correlation matrix to form a sensitivity coefficient; Critical point positioning is performed from the sensitivity coefficient to determine a mutation boundary point; Nonlinear gain processing is performed through the mutation boundary point to form a gain ratio; A defect amplification factor is formed based on the gain ratio.

8. The method of claim 4, wherein, The bidirectional correlation verification extraction of the influence depth level includes: Forward transmission tracking is performed according to the influence depth level to form a forward correlation path; Reverse backtracking verification is performed on the forward correlation path to form a reverse verification path; The forward correlation path and the reverse verification path are compared for consistency to screen an effective transmission segment; The cross-process transmission channel is formed according to the effective transmission segment.

9. The method of claim 5, wherein, The source backtracking from the diffusion trajectory combined with the node contribution degree to generate a backtracking chain includes: Time reverse correlation is established according to the diffusion trajectory; A plurality of backtracking check points are set along the time reverse correlation; Defect conduction chain is reversely constructed through the backtracking check points; The backtracking chain is formed by weighted series according to the defect conduction chain combined with the node contribution degree.

10. A collaborative control system for multi-process PCB manufacturing, characterized in that, It includes: A data acquisition module is used to acquire multi-source process data of each process of a PCB production line, and to perform multi-source mutual verification filtering on the multi-source process data to form a process state baseline; A reference detection module is used to perform process capability review according to the process state baseline to identify process drift features, to perform dynamic threshold layering on the process drift features to construct a process reference library, and to perform offset detection through the process reference library to generate a process deviation parameter; An association analysis module is used to determine the transmission coupling degree of adjacent processes by performing process inter-correlation strengthening processing on the process state baseline, to locate a defect transmission link by using the transmission coupling degree for cross-process penetration tracking, and to form a process correlation matrix by cascading and accumulating the process deviation parameter along the defect transmission link; A map construction module is used to determine a defect propagation range by defect diffusion backtracking and influence domain aggregation on the process correlation matrix, to generate a process state parameter according to the defect propagation range, and to construct a process feature map by extracting a trend turning feature in the process state parameter; An instruction output module is used to perform sensitivity decoupling extraction on the process deviation parameter and the process correlation matrix to extract a defect amplification factor, to identify a process critical node according to the defect amplification factor, to construct a regulation rule library around the process critical node, and to output a process control instruction by integrating the process feature map through the regulation rule library.

Images