PCB production quality tracing early warning method and system based on cloud computing platform

Abstract

The invention discloses a PCB production quality tracing early warning method and system based on a cloud computing platform, and relates to the technical field of quality tracing. The method comprises the following steps: carrying out PCB production chain reconstruction dimensionally by dimensionality according to the quality specification of a PCB, the process quality, the electrical quality and the mechanical resistance, and determining a ternary virtual production line; a traceability interface is additionally arranged at a production line testing end, PCB testing is carried out along with the production process, and a traceability instruction is generated; according to the ternary virtual production line, a quality traceability device is developed on a cloud computing platform, and a quality traceability result is determined by receiving a traceability instruction, matching the virtual production line, writing production flow data, executing traceability branch matching and two-order traceability positioning; and returning a quality traceability result to a production line test end, and performing interface visualization. The technical problems of low PCB production quality tracing efficiency and insufficient tracing accuracy in the prior art are solved, and the technical effects of efficient and accurate quality tracing and early warning based on the cloud computing platform are achieved.

Description

Technical Field

[0001] This invention relates to the field of quality traceability technology, specifically to a method and system for quality traceability and early warning of PCB board production based on a cloud computing platform. Background Technology

[0002] Printed Circuit Boards (PCBs) are crucial components of electronic products, and their production process involves multiple dimensions, including process quality, electrical quality, and mechanical resistance. Current PCB production quality traceability largely relies on localized inspection records and offline data analysis. This not only results in fragmented data and isolated information but also necessitates manual, step-by-step investigation of production stages after defects occur, leading to low traceability efficiency and long location cycles. Furthermore, due to the lack of integrated modeling and real-time interaction mechanisms for multi-dimensional production information, existing traceability methods suffer from significant deficiencies in accuracy and real-time performance, failing to meet the demands of modern PCB production for rapid quality early warning and precise defect location. Summary of the Invention

[0003] This application provides a method and system for PCB production quality traceability and early warning based on a cloud computing platform, which solves the technical problems of low efficiency and insufficient accuracy of traceability in PCB production quality traceability in the prior art.

[0004] The first aspect of this application provides a PCB board production quality traceability and early warning method based on a cloud computing platform, the method comprising: Based on the PCB board quality specifications, and according to process quality, electrical quality, and mechanical resistance, the PCB board production chain is reconstructed dimension by dimension to determine a three-dimensional virtual production line. Each dimension's production thread is constructed using dimensional elements as building entities. A traceability interface is added to the production line testing end to perform PCB board testing as production progresses, generating traceability instructions, with test arrays and product numbers as accompanying information. Based on the three-dimensional virtual production line, a quality traceability device is developed on a cloud computing platform. By receiving the traceability instructions, it matches the virtual production line and writes production flow data, performs traceability branch matching and two-stage traceability positioning, and determines the quality traceability result. The quality traceability result is then transmitted back to the production line testing end through the traceability interface for visualization.

[0005] A second aspect of this application provides a PCB board production quality traceability and early warning system based on a cloud computing platform, the system comprising: Production Chain Restructuring Module: Based on the quality specifications of PCB boards, and according to process quality, electrical quality, and mechanical resistance, the PCB board production chain is restructured dimension by dimension to determine a three-dimensional virtual production line. Each dimension's production thread is constructed using dimensional elements as building entities. Traceability Module: A traceability interface is added to the production line testing end. PCB board testing is performed along with the production process, generating traceability instructions. Test arrays and product numbers are used as accompanying information. Traceability Positioning Module: Based on the three-dimensional virtual production line, a quality traceability device is developed on a cloud computing platform. By receiving the traceability instructions, it matches the virtual production line and writes production flow data, performs traceability branch matching and two-stage traceability positioning, and determines the quality traceability result. Visualization Module: The quality traceability result is transmitted back to the production line testing end through the traceability interface for interface visualization.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, based on the PCB board quality specifications, the PCB board production chain is reconstructed dimension by dimension according to process quality, electrical quality, and mechanical resistance, establishing a three-dimensional virtual production line. Each dimension's production thread is constructed using dimensional elements as building entities. Next, a traceability interface is added to the production line testing end. PCB board testing is performed along with the production process, generating traceability instructions, accompanied by test arrays and product numbers. Then, based on the three-dimensional virtual production line, a quality traceability device is developed on a cloud computing platform. By receiving traceability instructions, it matches the virtual production line and writes production flow data, performing traceability branch matching and two-stage traceability positioning to determine the quality traceability results. Finally, the quality traceability results are transmitted back to the production line testing end through the traceability interface for visualization. This solves the technical problems of low efficiency and insufficient accuracy in PCB board production quality traceability in existing technologies, achieving efficient and accurate quality traceability and early warning based on a cloud computing platform. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 A schematic diagram of the PCB board production quality traceability and early warning method based on a cloud computing platform provided in this application embodiment; Figure 2 A schematic diagram of the structure of a PCB board production quality traceability and early warning system based on a cloud computing platform provided in this application embodiment.

[0009] Figure labeling: Production chain restructuring module 11, traceability module 12, source tracing and positioning module 13, visualization module 14. Detailed Implementation

[0010] This application provides a PCB production quality traceability and early warning method and system based on a cloud computing platform, which solves the technical problems of low efficiency and insufficient traceability accuracy in the existing PCB production quality traceability.

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0013] Example 1, as Figure 1 As shown, this application provides a PCB board production quality traceability and early warning method based on a cloud computing platform, wherein the method includes: Based on the quality specifications of PCB boards, the PCB board production chain is reconstructed dimension by dimension according to process quality, electrical quality and mechanical resistance, and a three-dimensional virtual production line is determined. In this line, each dimension's production thread is constructed with the dimensional element as the entity.

[0014] In this embodiment, the quality specifications and production-related data of the PCB board (including but not limited to equipment logs, process parameters, online testing data, finished product electrical test results, material batches, timestamps, etc.) are received and standardized. Based on the quality specifications, three major quality dimensions are defined: a first process quality dimension (example elements: etching depth / linewidth deviation, drilling diameter / position deviation, solder mask / silk screen coverage, surface roughness, etc.), a second electrical quality dimension (example elements: insulation resistance, conduction resistance, HI-POT withstand voltage, functional test pass / fail indicators, etc.), and a third mechanical resistance dimension (example elements: thermal cycle failure count, damp heat deposition, peel / shear strength, mechanical stress response, etc.). For each dimension, a dimension production thread is constructed to form a corresponding virtual production line, thereby determining the three-dimensional virtual production line.

[0015] Furthermore, based on process quality, electrical quality, and mechanical resistance, the PCB production chain is restructured dimension by dimension, including: For PCB board quality specifications, a first process quality dimension is defined, which includes at least etching, drilling, solder mask printing, and surface condition; a second electrical quality dimension is defined, which includes at least insulation, conductivity, high voltage withstand, and circuit function; and a third mechanical resistance dimension is defined, which includes at least thermal cycling performance, damp heat stability, and mechanical stress. Based on the first process quality dimension, the second electrical quality dimension, and the third mechanical resistance dimension, the PCB board production chain is restructured.

[0016] For the quality specifications of PCB boards, quality-related features are extracted from design parameters, process documents, and historical production and testing data, and then divided into three major quality dimensions according to the quality evaluation index system.

[0017] The first process quality dimension characterizes the process precision and consistency of PCB boards during manufacturing, including at least: line width and spacing control, pattern integrity, and residual copper in the etching process; hole diameter, positional deviation, and hole wall quality in the drilling process; uniform coverage and pattern clarity in the solder mask and screen printing processes; and surface flatness, cleanliness, and plating quality. For each process node, a quality data acquisition and evaluation model corresponding to the first process quality dimension is established and used as a process quality element of the virtual production line.

[0018] The second electrical quality dimension characterizes the electrical performance and stability of the PCB board, including at least: insulation (e.g., insulation resistance and insulation withstand voltage), conductivity (e.g., continuity resistance and connection integrity), high voltage withstand capability (e.g., high voltage breakdown capability), and circuit functionality (e.g., online functional testing, ICT testing, FCT functional pass rate). The production thread for the second electrical quality dimension consists of various electrical testing stations and verification processes, with test parameters and judgment results as dimensional elements.

[0019] The third mechanical resistance dimension characterizes the reliability of PCB boards under environmental stress and mechanical loads, and includes at least: thermal cycling performance (e.g., the number of failures in thermal shock tests), damp heat stability (e.g., dimensional and performance changes under constant temperature and humidity conditions), and mechanical stress (e.g., bending, impact, peel, and shear strength). The production process for the third mechanical resistance dimension consists of environmental testing procedures and mechanical testing stages, using test parameters and performance indicators as dimensional elements.

[0020] Based on quality elements of the first process quality dimension, the second electrical quality dimension, and the third mechanical resistance dimension, the relevant physical process nodes, inspection nodes, and test nodes in the production chain are retrieved. According to the process flow sequence and data association between nodes, the corresponding dimensional production threads are reconstructed to form three virtual production lines. Each virtual production line is constructed with dimensional elements as the building entity, supporting subsequent traceability matching, causal analysis, and multi-dimensional cross-location, thereby realizing full-dimensional quality modeling and traceability of the PCB board production process.

[0021] Furthermore, based on the first process quality dimension, the PCB board production chain is restructured, including: The PCB production chain is retrieved, and physical nodes are located based on the cross-correlation of the first process quality dimension. The physical nodes are production nodes related to the first process quality dimension. For each physical node, a first-order high-dimensional tensor and a second-order decomposition tensor are defined based on the equipment-material status-quality risk control root cause. The physical nodes are sequentially integrated, and the first-order high-dimensional tensor and the second-order decomposition tensor are written into each physical node to form the first virtual production line.

[0022] The entire production chain data is accessed from the Manufacturing Execution System (MES), Manufacturing Execution Data Warehouse, and Online Inspection Database to filter out process records and equipment information corresponding to the first process quality dimension. Based on the historical cross-correlation coefficients between each process node in indicators such as etching, drilling, solder mask printing, and surface condition, an inter-process correlation matrix is ​​constructed. Process nodes with significant correlations are then selected according to a preset correlation threshold (e.g., 0.6~0.8, which can be adjusted according to the actual process) and identified as entity nodes of the first process quality dimension. These entity nodes are production nodes directly related to the first process quality dimension, such as etching machines, drilling machines, solder mask printing machines, and surface treatment equipment, and include information such as their corresponding production batches, equipment parameters, operation times, and inspection results.

[0023] For each entity node, data modeling is performed using a triplet of "equipment identifier - material status - quality risk control root cause." Equipment operating parameters, material batch status information, and potential root cause factors associated with quality defects are integrated into a first-order high-dimensional tensor (axes may include equipment ID, batch number, time window, detection index vector, etc.). Simultaneously, second-order statistics are calculated and decomposed on historical samples (e.g., using Singular Value Decomposition (SVD) or tensor decomposition methods) to obtain a second-order decomposed tensor, representing the node's potential quality influencing factors and their coupling relationships. Following the actual PCB board production process, the aforementioned entity nodes are sequentially integrated to establish an ordered node linked list or directed graph structure. The corresponding first-order high-dimensional tensor and second-order decomposed tensor are written to each node, forming a complete data-driven production thread. This thread serves as the first virtual production line, enabling direct matching of test arrays during subsequent quality traceability, achieving rapid location and causal analysis in the process quality dimension.

[0024] Furthermore, based on the second electrical quality dimension, the PCB production chain is restructured, including: accessing batch data from the Production Equipment (ATE) system, electrical testing database, and production process parameter library to identify the processes and testing nodes corresponding to the second electrical quality dimension. The second electrical quality dimension includes at least four categories of indicators: insulation, continuity, high-voltage withstand capability, and circuit function. Therefore, a multi-indicator correlation matrix needs to be constructed based on historical testing records, and the correlation coefficients between each pair of process nodes need to be statistically analyzed. When the correlation coefficient is higher than a preset threshold (e.g., 0.65), the node is determined to be highly correlated with electrical quality and marked as a physical node of the second electrical quality dimension. Typical physical nodes may include electrical insulation testing stations, online continuity testing equipment, high-voltage withstand test benches, and automatic circuit function testing systems. For each physical node, its equipment operating parameters (such as voltage range, test current, and fixture status), material status information (board type, copper foil thickness, surface treatment method, etc.), and root cause factors corresponding to quality defects in past batches are collected to construct a triplet of "equipment identifier - material status - quality risk control root cause," and a first-order high-dimensional tensor is generated based on this to characterize the node's quality performance under multi-parameter conditions. Simultaneously, using historical production samples, a second-order decomposition (using methods such as Principal Component Analysis (PCA), Singular Value Decomposition (SVD), or tensor decomposition) is performed on the multi-dimensional feature matrix associated with the nodes to form a second-order decomposition tensor to extract hidden electrical performance influencing factors and their interaction patterns. Finally, according to the actual production and inspection sequence of the PCB board, these physical nodes are sequentially ordered to establish an ordered virtual production line data structure for the electrical quality dimension, and the first-order high-dimensional tensor and second-order decomposition tensor of each node are written sequentially to generate a second virtual production line.

[0025] Furthermore, based on the third mechanical resistance dimension, the PCB production chain is reconstructed, including extracting testing and processing records related to the third mechanical resistance dimension from reliability laboratory databases, Environmental Stress Screening (ESS) systems, and production quality inspection archives. This dimension includes at least three types of indicators: thermal cycling performance, damp heat stability, and mechanical stress. Therefore, the correlation coefficients between these indicators and each production node are calculated from historical testing data, and nodes with coefficients higher than a threshold (e.g., 0.7) are selected as physical nodes for the mechanical resistance dimension. Typical nodes include thermal cycling test stations, constant temperature and humidity aging chambers, mechanical stress loading stations, and warpage testing stations. For each physical node, the operating mode and parameters of the equipment (such as temperature curves, humidity curves, stress loading rates, etc.), material batch and lamination structure information, and root causes related to mechanical failure are collected. A triplet of "equipment identifier - material status - quality risk control root cause" is constructed, and a first-order high-dimensional tensor is generated for mechanical performance characterization under multiple conditions. Simultaneously, a second-order decomposition operation (using methods such as tensor decomposition and factor analysis) is performed on historical production and testing data to form a second-order decomposition tensor, revealing the potential driving factors and correlation patterns of mechanical resistance changes. All entity nodes are arranged according to the time sequence of mechanical resistance tests and related production processes to establish a virtual production line structure based on mechanical resistance dimensions. The first-order high-dimensional tensor and the second-order decomposition tensor of each node are written sequentially to form a third virtual production line.

[0026] A traceability interface is added to the production line testing end to perform PCB board testing as the production process progresses and generate traceability instructions, which include test arrays and product numbers as accompanying information.

[0027] An embedded traceability interface module can be installed at the testing station or on-line inspection equipment of existing PCB production lines. This module can be a software API interface, a hardware communication module, or a combination of both, enabling bidirectional data interaction with the quality traceability device on a cloud computing platform. The traceability interface connects to the host computer via industrial Ethernet, OPC-UA protocol, Modbus, or a customized communication protocol to achieve real-time acquisition and uploading of test data.

[0028] As each PCB board passes through the production line testing terminal, it undergoes tests in a preset order, covering aspects such as process quality, electrical quality, and mechanical resistance. The raw test signals collected by the testing equipment are preprocessed, normalized, and encoded by a local data processing unit (such as a PLC or embedded controller) to generate a multi-dimensional numerical matrix (test matrix). The row dimensions of this test matrix can represent test items or indicators, the column dimensions can represent time series, channel numbers, or sensor types, and the matrix elements are the corresponding test values.

[0029] After acquiring the test array, the traceability interface binds it to the unique product number of the corresponding PCB board (which can be a barcode, QR code, RFID tag or batch number generated by MES) to form a standardized traceability instruction.

[0030] Furthermore, PCB board testing is performed as the production process progresses, generating traceability instructions, including: As the PCB boards are processed on the production line in batches, quality specification tests are performed at the production line testing end. If the quality meets the standards, a qualified identifier is generated; if the quality does not meet the standards, a test array is generated based on the quality defects, the test array is encapsulated with the product number, and the traceability instruction is generated. The traceability instruction is an execution instruction that triggers defect tracing.

[0031] During the processing of PCB batches along the production line, when the processing flow reaches the production line testing end, the batch or individual PCBs are tested for quality specifications using configured online testing equipment (such as automated optical inspection (AOI) equipment, flying probe testers, functional testers, or mechanical strength testing devices). Test content may include process quality indicators (etch width, drilling accuracy, solder mask integrity, etc.), electrical quality indicators (insulation resistance, continuity resistance, high voltage withstand, etc.), and mechanical resistance indicators (warpage, thermal cycling performance, etc.).

[0032] If the test results are determined by the local judgment logic (which can be threshold judgment, statistical tolerance range, or machine learning model output) to be in full compliance with the preset quality standards, the test equipment generates a qualified identifier. This identifier can be an electronic tag, a database record status bit, or a "qualified" status directly written into the MES system, used to mark that the product does not need to enter the defect tracing process.

[0033] If any item fails to meet the test results, a corresponding test matrix is ​​generated based on the detected quality defect. This test matrix is ​​a multi-dimensional numerical matrix; the rows correspond to the test items, and the columns correspond to time series, channel numbers, or sensor IDs. The matrix elements are the corresponding test values ​​or feature parameters. After generating the test matrix, it is bound to the unique product number of the PCB board (which can be a barcode, QR code, RFID, or MES production batch number) and encapsulated into a standardized data packet to form a traceability instruction. The traceability instruction must include at least the following fields: product number field (uniquely identifying the target PCB board), test matrix field (recording defect-related test data), defect type and corresponding test indicator code, test timestamp, test station identifier, and equipment number.

[0034] The traceability command is the execution command that triggers defect tracing. After it is uploaded to the quality traceability device on the cloud computing platform, it will immediately start the matching process with the virtual production line and enter the defect location and production process of reverse investigation.

[0035] Based on the aforementioned three-element virtual production line, a quality traceability device is developed on the cloud computing platform. By receiving the traceability instructions, the device performs virtual production line matching and production flow data writing, executes traceability branch matching and two-stage traceability positioning, and determines the quality traceability result.

[0036] Based on the ternary virtual production line, a quality traceability device is developed on a cloud computing platform. As a core software module, the quality traceability device is deployed on a cloud server or edge computing node, possessing high concurrency processing capabilities and real-time data analysis capabilities. The quality traceability device receives traceability instruction data packets uploaded from the production line testing end through a pre-defined API interface. Upon receiving the traceability instruction, the quality traceability device parses the product number and test array in the traceability instruction, and calls the built-in ternary virtual production line database based on the product number to match the corresponding process quality virtual production line, electrical quality virtual production line, and mechanical resistance virtual production line. Finally, the quality traceability device outputs complete quality traceability results, including the traceability path, key node identifiers, and defect types.

[0037] The matching process includes: based on the key indicators in the test matrix, performing preliminary feature matching with the first-order high-dimensional tensor of each entity node in the three-element virtual production line to screen out possible traceability branches (i.e., the set of associated production line nodes); retrieving the historical production flow data within the screened traceability branches, and writing the production flow data corresponding to the real-time test matrix into the corresponding node of the virtual production line based on information such as timestamp and batch number, so as to realize the dynamic updating and association of data.

[0038] The quality traceability system performs two-stage traceability and localization, specifically including: First traceability layer: Through matching of first-order high-dimensional tensors, it quickly identifies production nodes highly correlated with the current test defect, determining the first traceability node; Second verification layer: It performs counterfactual dynamic reasoning verification on the first traceability node, verifying whether the node is indeed a key node causing the defect based on historical data models and causal inference methods, ensuring traceability accuracy; Third traceability layer: After successful verification, it uses second-order decomposition tensors for fine-grained matching to further locate the specific link and root cause of the defect in the production process, achieving high-precision quality traceability results.

[0039] Furthermore, the development of quality traceability systems includes: Based on the first virtual production line, a first tracing layer is deployed using matching based on a first-order high-dimensional tensor; a second verification layer is deployed using counterfactual derivation verification based on the output of the first tracing node; a third tracing layer is deployed using matching based on a second-order decomposition tensor; the first tracing layer, the second tracing layer, and the third tracing layer are cascaded to construct a first tracing branch.

[0040] The first tracing layer: Based on the first virtual production line, coarse-grained defect localization is achieved using first-order high-dimensional tensor matching technology. Specifically, for the input test data matrix, key feature vectors are extracted and their similarity is calculated with the pre-constructed first-order high-dimensional tensors of each entity node in the first virtual production line. Candidate tracing nodes that highly match the features of the test data are selected using methods such as Euclidean distance, cosine similarity, or Mahalanobis distance, thus constructing a preliminary defect tracing path.

[0041] The second verification layer dynamically verifies the candidate traceability nodes output by the first traceability layer using counterfactual reasoning. Utilizing historical production data and causal models, the second verification layer simulates the production line performance if the node were absent or its parameters were normal. By comparing and verifying whether the current node is a key factor in the defect's occurrence, it eliminates false positives and enhances the accuracy and robustness of traceability.

[0042] The third traceability layer: Based on fine-grained matching of second-order decomposition tensors, further in-depth analysis of verified traceability nodes is conducted. By utilizing latent factors obtained through tensor decomposition methods (such as CP decomposition, Tucker decomposition, etc.), multi-dimensional feature matching is performed to achieve precise localization of defect sources, including specific processes, abnormal points in equipment parameters, and related material batches.

[0043] The first traceability layer, the second verification layer, and the third traceability layer are cascaded in sequence to form a complete traceability process, referred to as the first traceability branch. As the core inference link of the quality traceability system, the first traceability branch supports multi-level quality traceability analysis from coarse-grained localization to fine verification and localization, enabling accurate tracing of the root causes of defects in the PCB board production process.

[0044] Furthermore, for the three-element virtual production line, traceability branches are constructed and integrated in parallel to serve as the quality traceability device; the quality traceability device is embedded in the cloud data platform, and communication interaction between the traceability interface and the cloud platform is established.

[0045] For each of the process quality, electrical quality, and mechanical resistance dimensions, corresponding traceability branches are constructed. Each traceability branch independently performs defect localization and root cause analysis based on virtual production line data and tensor matching models for its corresponding dimension. Subsequently, these three traceability branches are merged into a unified quality tracer through a parallel integration mechanism to achieve comprehensive analysis of multi-dimensional quality data and cross-dimensional defect traceability. Parallel integration employs a distributed computing framework (such as containerized deployment based on a microservice architecture, or stream processing technologies such as Apache Kafka and Flink) to achieve asynchronous scheduling and collaborative work among the traceability branches, ensuring the real-time nature and accuracy of traceability results. The quality tracer has a unified interface specification to accept traceability commands from the production line testing end and synchronously transmit the analysis results back.

[0046] The quality traceability device is deployed as an embedded software module on a cloud data platform, leveraging cloud computing resources to achieve elastic scaling and high availability. By establishing a data communication channel between the traceability interface and the cloud platform, it supports multiple communication protocols (such as HTTP / HTTPS, MQTT, WebSocket, etc.) to achieve bidirectional data interaction between the testing end and the cloud, including uploading traceability instructions, real-time feedback of quality traceability results, and monitoring and management of system status.

[0047] Furthermore, the matching of virtual production lines and the writing of production flow data include: The traceability instruction is uploaded to the cloud data platform, and the traceability branch is matched according to the test array to determine the target traceability branch. According to the product number, the production flow data is called in the monitoring database, and the virtual production line in the target traceability branch is written with data. The production flow data is filtered with virtual production line elements. The monitoring database is built into the cloud data platform.

[0048] Traceability instructions generated from the production line testing end are uploaded to the cloud data platform via the network. The test data matrix contained in the traceability instructions serves as multi-dimensional quality feature input. After receiving this data, the quality traceability device performs feature matching between the test data matrix and virtual production line nodes in each dimension, based on a pre-constructed three-dimensional virtual production line model. By calculating the matching degree or similarity index, the traceability branch that best matches the test data is dynamically selected, and the target traceability branch is determined, thereby achieving preliminary coarse-grained defect path localization.

[0049] Based on the unique product number carried in the traceability instruction, the complete production flow data of the product is retrieved from the monitoring database built into the cloud data platform. The production flow data includes multi-dimensional historical information of the PCB board, from raw materials, processing parameters of each process, test results, equipment status to environmental conditions. Using virtual production line elements as filtering conditions, the production flow data is accurately filtered and mapped to ensure that only production node information related to the target traceability branch is written, achieving accurate data aggregation and dynamic updates.

[0050] Furthermore, by performing source tracing branch matching and two-order source tracing localization, the quality tracing results are determined, including: According to the first tracing layer within the target tracing branch, the first tracing node is determined by matching the test array with the one-dimensional high-order tensor of the entity node; according to the second verification layer, counterfactual dynamic reasoning verification is performed on the first tracing node to determine the verification result; if the verification result is yes, according to the third tracing layer, fine-grained matching based on second-order decomposition tensor is performed on the first tracing node to locate the quality tracing result.

[0051] Based on the aforementioned identified target tracing branch, the quality tracing tool utilizes a pre-constructed one-dimensional high-order tensor from the entity nodes of the first tracing layer within that branch, taking the input test matrix as a feature vector, and performs matching calculations. Matching methods can employ Euclidean distance, cosine similarity, or kernel-based similarity measures. By comparing the similarity between the test matrix and the tensor features of each entity node, the tool identifies the first tracing node most relevant to the test defect behavior, serving as a preliminary result for coarse-grained defect localization.

[0052] For the first tracing node, counterfactual reasoning techniques are applied to construct a causal model based on historical production data, dynamically simulating the production performance if this node were absent or its operating parameters returned to normal. By comparing this simulation with actual test results, it is determined whether this node truly caused the defect. This process ensures the elimination of false positives or noise interference; if the verification result is "yes," the node is confirmed as a valid defect source.

[0053] After successful verification, for the confirmed first traceability node, the quality traceability system further invokes a multidimensional feature model based on second-order decomposition tensors to perform fine-grained matching analysis on the influence of multiple factors within the node. This step utilizes the latent factors and interaction patterns extracted by tensor decomposition to accurately locate the specific link of the defect, abnormal equipment status, or material batch problem, ultimately generating a high-precision quality traceability result.

[0054] Furthermore, after determining the quality traceability results, this includes: Write the quality traceability results into a temporary database; set a preset defect frequency, accumulate the same type of frequency of the quality traceability results in the temporary database, and when it exceeds the preset defect frequency, execute batch processing early warning and feedback control.

[0055] First, the obtained quality traceability results are written to a temporary database in the cloud data platform in real time. This temporary database supports high-concurrency writing and fast querying, storing information such as traceability paths, key node identifiers, defect types, relevant production parameters, and timestamps to ensure data integrity and traceability. Second, the system accumulates and statistically analyzes the traceability results of similar defects in the temporary database based on a pre-set defect frequency threshold. The statistical process categorizes and summarizes defects according to category, production batch number, and time window, dynamically monitoring the frequency of specific defects within a certain time or production volume range. When the accumulated frequency of similar defects exceeds the preset threshold, the system automatically triggers a batch processing early warning mechanism. This early warning mechanism includes, but is not limited to, sending alarms to production management personnel, initiating abnormal batch suspension processes, and invoking feedback control algorithms to adjust relevant equipment and process parameters, intervening in the production process in a timely manner to prevent defect spread.

[0056] The quality traceability results are transmitted back to the production line testing end through the traceability interface for visualization.

[0057] The quality traceability results generated on the cloud computing platform are transmitted back to the production line testing end in real time through the traceability interface. This transmission process uses stable communication protocols (such as HTTP / HTTPS, MQTT, or WebSocket) to ensure the real-time performance and security of the data.

[0058] At the production line testing end, the system module receiving the quality traceability results transmits the data to the visualization interface display unit. This interface, in the form of a graphical user interface (GUI), uses various formats such as visual charts, flowcharts, heatmaps, or hierarchical node diagrams to display the traceability path and key defect nodes, facilitating operators' intuitive understanding of the defect sources and their impact scope.

[0059] In summary, the embodiments of this application have at least the following technical effects: First, based on the PCB board quality specifications, the PCB board production chain is reconstructed dimension by dimension according to process quality, electrical quality, and mechanical resistance, establishing a three-dimensional virtual production line. Each dimension's production thread is constructed using dimensional elements as building entities. Next, a traceability interface is added to the production line testing end. PCB board testing is performed along with the production process, generating traceability instructions, accompanied by test arrays and product numbers. Then, based on the three-dimensional virtual production line, a quality traceability device is developed on a cloud computing platform. By receiving traceability instructions, it matches the virtual production line and writes production flow data, performing traceability branch matching and two-stage traceability positioning to determine the quality traceability results. Finally, the quality traceability results are transmitted back to the production line testing end through the traceability interface for visualization. This solves the technical problems of low efficiency and insufficient accuracy in PCB board production quality traceability in existing technologies, achieving efficient and accurate quality traceability and early warning based on a cloud computing platform.

[0060] Example 2 is based on the same inventive concept as the PCB board production quality traceability and early warning method based on a cloud computing platform in the previous examples, such as... Figure 2 As shown, this application provides a PCB board production quality traceability and early warning system based on a cloud computing platform, wherein the system includes: Production Chain Reconstruction Module 11: Based on the quality specifications of the PCB board, and according to process quality, electrical quality, and mechanical resistance, the PCB board production chain is reconstructed dimension by dimension to determine a three-dimensional virtual production line. Each dimension's production thread is constructed using dimensional elements. Traceability Module 12: A traceability interface is added to the production line testing end. PCB board testing is performed along with the production process, generating traceability instructions. Test arrays and product numbers are used as accompanying information. Traceability Positioning Module 13: Based on the three-dimensional virtual production line, a quality traceability device is developed on a cloud computing platform. By receiving the traceability instructions, the virtual production line is matched and production flow data is written. Traceability branch matching and two-stage traceability positioning are performed to determine the quality traceability result. Visualization Module 14: The quality traceability result is transmitted back to the production line testing end through the traceability interface for interface visualization.

[0061] Furthermore, the production chain restructuring module 11 is used to perform the following methods: For PCB board quality specifications, a first process quality dimension is defined, which includes at least etching, drilling, solder mask printing, and surface condition; a second electrical quality dimension is defined, which includes at least insulation, conductivity, high voltage withstand, and circuit function; and a third mechanical resistance dimension is defined, which includes at least thermal cycling performance, damp heat stability, and mechanical stress. Based on the first process quality dimension, the second electrical quality dimension, and the third mechanical resistance dimension, the PCB board production chain is restructured.

[0062] Furthermore, the production chain restructuring module 11 is used to perform the following methods: The PCB production chain is retrieved, and physical nodes are located based on the cross-correlation of the first process quality dimension. The physical nodes are production nodes related to the first process quality dimension. For each physical node, a first-order high-dimensional tensor and a second-order decomposition tensor are defined based on the equipment-material status-quality risk control root cause. The physical nodes are sequentially integrated, and the first-order high-dimensional tensor and the second-order decomposition tensor are written into each physical node to form the first virtual production line.

[0063] Furthermore, the source tracing and location module 13 is used to perform the following method: Based on the first virtual production line, a first tracing layer is deployed using matching based on a first-order high-dimensional tensor; a second verification layer is deployed using counterfactual derivation verification based on the output of the first tracing node; a third tracing layer is deployed using matching based on a second-order decomposition tensor; the first tracing layer, the second tracing layer, and the third tracing layer are cascaded to construct a first tracing branch.

[0064] Furthermore, the source tracing and location module 13 is used to perform the following method: For the three-element virtual production line, traceability branches are constructed and integrated in parallel to serve as the quality traceability device; the quality traceability device is embedded in the cloud data platform, and communication interaction between the traceability interface and the cloud platform is established.

[0065] Furthermore, the traceability module 12 is used to perform the following methods: As the PCB boards are processed on the production line in batches, quality specification tests are performed at the production line testing end. If the quality meets the standards, a qualified identifier is generated; if the quality does not meet the standards, a test array is generated based on the quality defects, the test array is encapsulated with the product number, and the traceability instruction is generated. The traceability instruction is an execution instruction that triggers defect tracing.

[0066] Furthermore, the source tracing and location module 13 is used to perform the following method: The traceability instruction is uploaded to the cloud data platform, and the traceability branch is matched according to the test array to determine the target traceability branch. According to the product number, the production flow data is called in the monitoring database, and the virtual production line in the target traceability branch is written with data. The production flow data is filtered with virtual production line elements. The monitoring database is built into the cloud data platform.

[0067] Furthermore, the source tracing and location module 13 is used to perform the following method: According to the first tracing layer within the target tracing branch, the first tracing node is determined by matching the test array with the one-dimensional high-order tensor of the entity node; according to the second verification layer, counterfactual dynamic reasoning verification is performed on the first tracing node to determine the verification result; if the verification result is yes, according to the third tracing layer, fine-grained matching based on second-order decomposition tensor is performed on the first tracing node to locate the quality tracing result.

[0068] Furthermore, the source tracing and location module 13 is used to perform the following method: Write the quality traceability results into a temporary database; set a preset defect frequency, accumulate the same type of frequency of the quality traceability results in the temporary database, and when it exceeds the preset defect frequency, execute batch processing early warning and feedback control.

[0069] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0070] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0071] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A PCB board production quality traceability and early warning method based on a cloud computing platform, characterized in that, The method includes: Based on the quality specifications of PCB boards, the PCB board production chain is reconstructed dimension by dimension according to process quality, electrical quality and mechanical resistance, and a three-dimensional virtual production line is determined. In this line, each dimension production thread is constructed with the dimension element as the entity. A traceability interface is added to the production line testing end to perform PCB board testing as the production process progresses and generate traceability instructions, in which the test array and product number are the accompanying information. Based on the aforementioned three-element virtual production line, a quality traceability device is developed on the cloud computing platform. By receiving the traceability instructions, the device performs virtual production line matching and production flow data writing, executes traceability branch matching and two-stage traceability positioning, and determines the quality traceability result. The quality traceability results are transmitted back to the production line testing end through the traceability interface for visualization.

2. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 1, characterized in that, Based on process quality, electrical quality, and mechanical resistance, the PCB board production chain is restructured dimension by dimension, including: For the quality specifications of PCB boards, a first process quality dimension is defined, wherein the first process quality dimension includes at least etching, drilling, solder mask printing and surface condition. Define a second electrical quality dimension, wherein the second electrical quality dimension includes at least insulation, conductivity, high voltage withstand, and circuit function; A third mechanical resistance dimension is defined, wherein the third mechanical resistance dimension includes at least thermal cycling performance, damp heat stability and mechanical stress; The PCB production chain is restructured based on the first process quality dimension, the second electrical quality dimension, and the third mechanical resistance dimension.

3. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 2, characterized in that, Based on the first process quality dimension, the PCB board production chain is restructured, including: The PCB production chain is retrieved, and physical nodes are located based on the cross-correlation of the first process quality dimension, wherein the physical nodes are production nodes related to the first process quality dimension. For each entity node, a first-order high-dimensional tensor and a second-order decomposition tensor are defined based on the equipment-material status-quality risk control root cause. The entity nodes are sequentially integrated, and the first-order high-dimensional tensor and the second-order decomposition tensor are written into each entity node to form the first virtual production line.

4. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 3, characterized in that, Develop a quality traceability system, including: Based on the first virtual production line, a first traceability layer is deployed using matching based on a first-order high-dimensional tensor; Deploy a second verification layer based on counterfactual deductions derived from the output of the first tracing node; A third source tracing layer is deployed based on matching of second-order decomposition tensors; The first tracing layer, the second tracing layer, and the third tracing layer are cascaded to construct the first tracing branch.

5. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 4, characterized in that, For the three-element virtual production line, traceability branches are constructed and integrated in parallel to serve as the quality traceability device; The quality traceability device is embedded in a cloud data platform, and the traceability interface is established to communicate with the cloud platform.

6. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 1, characterized in that, PCB board testing is performed as the production process progresses, generating traceability instructions, including: As the PCB boards are processed on the production line in batches, quality specification tests are conducted at the production line testing end. If the quality meets the standards, a qualified identifier is generated. If the quality does not meet the standard, a test array is generated based on the quality defect, the test array is encapsulated with the product number, and the traceability instruction is generated, wherein the traceability instruction is an execution instruction that triggers defect tracing.

7. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 4, characterized in that, Perform virtual production line matching and production flow data writing, including: The tracing instruction is uploaded to the cloud data platform, and the tracing branch is matched according to the test array to determine the target tracing branch. Based on the product number, production flow data is retrieved from the monitoring database, and data is written to the virtual production line within the target traceability branch. The production flow data is filtered using virtual production line elements, and the monitoring database is built into a cloud data platform.

8. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 7, characterized in that, Perform source tracing branch matching and two-order source tracing localization to determine the quality tracing results, including: Based on the first tracing layer within the target tracing branch, the first tracing node is determined by matching the test array with the one-dimensional high-order tensor of the entity node; Based on the second verification layer, counterfactual dynamic reasoning verification is performed on the first tracing node to determine the verification result; If the verification result is yes, then according to the third tracing layer, fine-grained matching based on second-order decomposition tensors is performed on the first tracing node to locate the quality tracing result.

9. The PCB board production quality traceability and early warning method based on a cloud computing platform as described in claim 1, characterized in that, After confirming the quality traceability results, the following is included: Write the quality traceability results into a temporary database; A preset defect frequency is set, and the same type of frequency is accumulated for the quality traceability results in the temporary database. When the frequency exceeds the preset defect frequency, batch processing early warning and feedback control are executed.

10. A PCB board production quality traceability and early warning system based on a cloud computing platform, characterized in that, The system is used to implement the PCB board production quality traceability and early warning method based on a cloud computing platform as described in any one of claims 1-9, the system comprising: Production chain reconstruction module: Based on the quality specifications of PCB boards, the PCB board production chain is reconstructed dimension by dimension according to process quality, electrical quality and mechanical resistance, and a three-dimensional virtual production line is determined. Among them, the production threads of each dimension are constructed entities based on the dimensional elements. Traceability module: A traceability interface is added to the production line testing end. PCB board testing is performed as the production process progresses, and traceability instructions are generated, with the test array and product number as accompanying information. Traceability and Positioning Module: Based on the three-element virtual production line, a quality traceability device is developed on the cloud computing platform. By receiving the traceability instructions, the virtual production line is matched and production flow data is written. Traceability branch matching and two-stage traceability positioning are performed to determine the quality traceability result. Visualization module: The quality traceability results are transmitted back to the production line testing end through the traceability interface for interface visualization.

Images