FPC flexible circuit board intelligent production system based on image recognition

Abstract

The invention relates to the technical field of FPC (Flexible Printed Circuit) production control, in particular to an FPC intelligent production system based on image recognition, which comprises a demand analysis module used for carrying out information analysis on a production main body on an FPC production line, constructing a production information model and carrying out production task allocation according to production information; the process monitoring module carries out image recognition on key processes on the FPC production line and carries out processing according to production characteristics of the FPC flexible circuit board in the key processes to generate an FPC label, and the FPC label at least comprises time, key process characteristics, production characteristics and processing data; and the intelligent analysis module performs FPC quality tracing and management, including obtaining an electrical performance test result of the FPC flexible circuit board, generating FPC labels, searching and comparing similar FPC label sets, checking whether the FPC flexible circuit boards in the same batch have problems or not, and feeding back key FPC process steps.

Description

Technical Field

[0001] The present invention relates to the technical field of FPC production control, and particularly to an intelligent production system for FPC flexible circuit boards based on image recognition. Background Art

[0002] In the field of flexible circuit board manufacturing, due to its thin, light, and flexible substrate and complex production processes, higher requirements are put forward for aspects such as automated handling, high-precision alignment, on-line quality inspection, and process parameter control. In existing production workshops, there are generally problems such as low informatization and networking levels, most equipment running independently, and many manual operation links leading to easy wear and inaccurate positioning. Therefore, some experts have started to explore the construction of an intelligent manufacturing unit control system. By establishing a database and knowledge base, and using data analysis and machine learning technologies, self-perception and self-configuration of process parameters, prediction of product quality, and a supervision and control system based on data analysis are realized to evaluate production difficulty, monitor the environment and personnel flow, and conduct fault prediction and simulation. On the other hand, modular production units and intelligent material management systems are also gradually being applied to improve production flexibility and efficiency. However, existing technologies still have deficiencies in achieving refined, real-time, and intelligent comprehensive centralized control and management of the entire FPC production process, especially in aspects such as how to effectively integrate multi-source information, accurately monitor key process parameters and product status, and make quick responses and intelligent decisions according to real-time situations, which are still the key issues that current technologies need to solve.

[0003] Therefore, an intelligent production system for FPC flexible circuit boards based on image recognition is proposed. Summary of the Invention

[0004] The purpose of the present invention is to provide an intelligent production system for FPC flexible circuit boards based on image recognition to improve the supervision and control of the entire production process of FPC flexible circuit boards. First, the requirement analysis module analyzes the information of production entities on the FPC production line, constructs a production information model, and allocates production tasks according to the production information; the process monitoring module performs image recognition on key processes on the FPC production line, and processes according to the production characteristics of the FPC flexible circuit board in key processes to generate an FPC label, where the FPC label includes at least time, key process characteristics, production characteristics, and processing data; the intelligent analysis module conducts FPC quality traceability and management, including obtaining the electrical performance test results of the FPC flexible circuit board, generating an FPC label, searching for and comparing similar FPC label sets, checking whether there are problems with FPC flexible circuit boards of the same batch, and feeding back key FPC process steps.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An intelligent production system for FPC flexible circuit boards based on image recognition, comprising:

[0007] A requirements analysis module is used to analyze the information of the production entities on the FPC production line, construct a production information model, and allocate production tasks according to the production information;

[0008] Furthermore, the production entities at least include the FPC flexible circuit board materials to be produced, production equipment, tooling fixtures, production personnel, and production orders to be executed;

[0009] The production information model constructed based on the information of the production entities at least includes a material sub-model, an equipment sub-model, an order sub-model, a personnel sub-model, and association relationships, where the association relationships record the matching and compatibility rules between materials and orders, materials and equipment, and equipment and orders.

[0010] Furthermore, the process of production task allocation includes:

[0011] Construct a scheduling algorithm based on order priority, delivery date, available material quantity and its quality assessment, real-time equipment load and readiness status, and changeover cost; the scheduling algorithm continuously monitors the data changes in the production information model. When it detects that the production information model receives the information of a new production entity, the scheduling algorithm is started, generates an optimal production task sequence and equipment allocation plan, and executes the allocation plan.

[0012] The process monitoring module performs image recognition on the key processes on the FPC production line, and processes according to the production characteristics of the FPC flexible circuit board in the key processes to generate an FPC label, where the FPC label at least includes time, key process characteristics, production characteristics, and processing data;

[0013] Furthermore, the key processes at least include blanking, drilling, graphic transfer, etching, lamination, cutting, and surface treatment;

[0014] The image recognition is to collect and analyze images of the production characteristics of the FPC flexible circuit board in the key processes, and the production characteristics at least include registration targets, circuit patterns, hole positions, pads, character markings, surface defects, and alignment relationships between layers or materials;

[0015] The recognition algorithms used for the image recognition include but are not limited to edge detection, image segmentation, feature extraction, template matching, defect classification based on convolutional neural networks, and object detection algorithms.

[0016] Furthermore, the process of generating the FPC label includes:

[0017] Directly obtain the visual fingerprint of the production characteristics of the FPC flexible circuit board in each key process through image recognition. The visual fingerprint at least includes the time of the FPC flexible circuit board in the corresponding key process, the key process characteristics, and the processing data characteristics;

[0018] After image recognition and data collection are completed in each key process, link the visual fingerprints of the FPC flexible circuit board in all key processes according to the time sequence and mark a unique identifier, which is generated and attached in the first key process.

[0019] The intelligent analysis module conducts FPC quality traceability and management, including obtaining the electrical performance test results of the FPC flexible circuit board, generating FPC labels, searching for and comparing similar FPC label sets, checking whether there are problems with FPC flexible circuit boards of the same batch, and feeding back the key FPC process steps.

[0020] Furthermore, the process of obtaining the electrical performance test results of the FPC flexible circuit board and generating FPC labels includes:

[0021] After the FPC flexible circuit board completes all manufacturing processes, it enters the electrical performance test link; the electrical performance test at least includes a flying probe tester, an ICT tester, and a high-frequency tester;

[0022] Detect the electrical performance of the FPC flexible circuit board through the electrical performance test. The electrical performance at least includes conduction, insulation, impedance, and capacitance; the test results are automatically uploaded to the intelligent analysis module through the device interface and associated with the unique identifier of the FPC flexible circuit board;

[0023] If the test result of the FPC flexible circuit board is unqualified, immediately mark the status of the FPC flexible circuit board and ensure that detailed failure test data is added to the FPC label. The failure test data at least includes a list of failed test items, failure types, failure location information, comparison of actual measurement values with standard values, test conditions, test equipment and fixture information, and test timestamps.

[0024] Furthermore, the process of searching for and comparing similar FPC label sets and checking whether there are problems with FPC products of the same batch includes:

[0025] Retrieve FPC labels from the database. Similarity searches can be performed through several search methods, where the search methods at least include searching based on product attributes and batches, searching based on production characteristics, and searching based on test results; construct a set of similar FPC labels according to the search results;

[0026] After finding a set of similar FPC labels, analyze the electrical performance test results of the FPC flexible printed circuit boards within the set. If the test results of the FPC flexible printed circuit boards in the set are unqualified in several electrical performance test items and the unqualified ratio exceeds the set threshold ratio, it is determined that there is a quality problem with the set of similar FPC labels;

[0027] After determining that there is a quality problem with the set of similar FPC labels, the intelligent analysis module retrospectively analyzes the FPC label data corresponding to each key process of the unqualified FPC flexible printed circuit boards in the set of similar FPC labels, and compares it with the FPC label data of the qualified FPC flexible printed circuit boards. If the difference exceeds the preset difference threshold, the key process is marked as abnormal;

[0028] The intelligent analysis module counts the number of abnormalities in the key processes and generates a detailed analysis report. The analysis report at least includes the key processes with problems, the reasons for abnormalities, relevant production characteristics, and FPC label data.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] 1. The production information model provides a comprehensive, accurate, and dynamically updated digital mapping of the production site by constructing a digital mapping that reflects the status of personnel, equipment, materials, process methods, environment, etc. in the current production site in real time, providing high-quality input for subsequent scheduling algorithms; the scheduling algorithm continuously monitors the dynamically changing production task allocation process of the production information model and performs production scheduling, which can greatly improve the response ability and execution efficiency of the production plan.

[0031] 2. By deploying intelligent monitoring based on image recognition at key processes and generating digital labels containing full-process visual fingerprints and processing data for each FPC flexible printed circuit board, high-precision, real-time, and non-contact monitoring of the production process of FPC flexible printed circuit boards is achieved, making the originally difficult-to-quantify and traceable production process of flexible printed circuit boards completely transparent and traceable.

[0032] 3. Through similarity search and set analysis of FPC labels, it is possible to automatically discover product groups with common characteristics and potential quality problems from a large amount of production data, changing passive search to active early warning, accurately locating the production batches or process windows where problems occur, and efficiently and scientifically diagnosing the root cause processes leading to problems by retrospectively comparing the detailed label data of problem products and qualified products at key processes. Brief Description of the Drawings

[0033] Figure 1 It is a schematic structural diagram of an intelligent production system for FPC flexible printed circuit boards based on image recognition provided by an embodiment of the present invention;

[0034] Figure 2 It is the flowchart of the requirement analysis module provided by the embodiment of the present invention;

[0035] Figure 3 It is the flowchart of the intelligent analysis module provided by the embodiment of the present invention. Specific embodiments

[0036] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0037] Embodiment 1:

[0038] In order to improve the supervision and control of the entire production process and improve the production efficiency of FPC flexible circuit boards, a certain FPC flexible circuit board manufacturer introduced an FPC flexible circuit board intelligent production system based on image recognition provided by the present invention. The system structure is as Figure 1 shown, and the specific implementation is as follows:

[0039] The requirement analysis module is used to analyze the information of the production entities on the FPC production line, construct a production information model, and allocate production tasks according to the production information;

[0040] Further, the production entities at least include the FPC flexible circuit board materials to be produced, production equipment, tooling fixtures, production personnel, and production orders to be executed;

[0041] The production information model constructed according to the information of the production entities at least includes a material sub-model, an equipment sub-model, an order sub-model, a personnel sub-model, and association relationships, where the association relationships record the matching and compatibility rules between materials and orders, materials and equipment, and equipment and orders.

[0042] Further, after receiving and integrating the original information collected from the above-mentioned various production entities, the requirement analysis module performs information analysis and constructs the production information model. Specifically, the process of constructing the production information model includes: data collection and preprocessing, information extraction and recognition, data association and integration, etc.

[0043] Further, the integrated structured information is used to dynamically update the production information model in the memory. When new materials enter the site, equipment status changes, tooling fixtures are in place, personnel log in, or new orders are imported, the corresponding material sub-model, equipment sub-model, tooling sub-model, personnel sub-model, or order sub-model in the model is created, updated, or marked with a status change. For example, a batch of newly arrived flexible copper clad laminates will create a record in the material sub-model, including its model number, quantity, preliminary quality score based on image analysis, and marked as the "to be assigned" status. A piece of equipment that has completed maintenance will update its status to "idle" in the equipment sub-model and record the type of tooling fixture currently installed on it.

[0044] Further, the core of the production information model lies in the association relationships and rules it contains. These rules are pre-configured according to the product's process flow and equipment capabilities and stored in the rule library of the model. These association relationships and rules at least include that a specific product model needs to use a specific material type and batch range, a specific process step must be executed on a specific type of equipment or equipment with a specific configuration, a specific equipment requires the use of a specific type or version of tooling fixture, the association between materials and orders, etc. These association relationships and rules are referenced during the model construction process to verify the accuracy and matching of information and are called by the scheduling algorithm as constraint conditions during scheduling.

[0045] The production information model is a digital mapping that reflects the status of personnel, equipment, materials, process methods, environment, etc. at the current production site in real time. It is a dynamically changing graph or network structure, where nodes represent production entities and edges represent the association relationships and current status between these entities, providing a comprehensive, accurate, and dynamically updated digital mapping of the production site, providing high-quality input for the subsequent scheduling algorithm, enabling it to more effectively plan the production process, reduce waiting and changeover times, improve equipment utilization and production throughput, and quickly respond to sudden production changes or new high-priority orders, enhancing the flexibility and responsiveness of the entire production system.

[0046] Further, the process of production task allocation includes:

[0047] Construct a scheduling algorithm based on order priority, delivery date, material availability and its quality assessment, real-time equipment load and readiness status, and changeover cost; the scheduling algorithm continuously monitors the data changes in the production information model. When it detects that the production information model receives information about a new production entity, the scheduling algorithm is started, generating an optimal production task sequence and equipment allocation plan, and executing the allocation plan.

[0048] Furthermore, first of all, the scheduling algorithm can be a rule-based expert system, linear programming, mixed integer programming, constraint programming, or more complex heuristic algorithms, simulation-based optimization, or even use reinforcement learning to cope with dynamic changes. The goal of the algorithm is to generate a production task sequence and equipment allocation plan that can optimize one or more objectives; the scheduling algorithm runs continuously as a background service and monitors the data changes in the production information model in real time. Once the model receives new production entity information, such as the import of a new production order, these changes will trigger the restart or adjustment of the scheduling algorithm; the system can also set trigger thresholds or periods, such as starting the scheduling every fixed time or when the status of key resources changes significantly.

[0049] Furthermore, the scheduling algorithm reads the current latest panoramic data in the production information model, including all tasks to be executed, the status of all available and in-use production resources, and their associated relationships. The algorithm calculates the benefits of different task sequences and resource allocation plans according to the preset optimization objectives and constraint conditions, and selects the optimal plan from them. This plan details which products, which batches, which equipment will process them in which time period, which batch of materials will be used, and the corresponding process parameter program numbers, etc. in the next time period.

[0050] Furthermore, after generating the optimal production task sequence and equipment allocation plan, the demand analysis module converts this plan into specific execution instructions. These instructions are sent to the control system layer at the production site through the network.

[0051] The scheduling algorithm continuously monitors the dynamic production task allocation process in the production information model and performs production scheduling, which can greatly improve the response ability and execution efficiency of the production plan; it gets rid of the limitations of static plans, can quickly respond to emergencies at the production site, make dynamic adjustments, and reduce production interruptions and resource waste caused by information lag or inaccuracy.

[0052] The process monitoring module performs image recognition on the key processes on the FPC production line and processes them according to the production characteristics of the FPC flexible circuit board in the key processes to generate FPC labels, and the FPC labels at least include time, key process characteristics, production characteristics, and processing data;

[0053] Furthermore, the key processes at least include blanking, drilling, graphic transfer, etching, lamination, cutting, and surface treatment;

[0054] The image recognition is for image acquisition and analysis of the production characteristics of the FPC flexible circuit board in the key processes, and the production characteristics at least include registration targets, circuit patterns, hole positions, pads, character markings, surface defects, and the alignment relationship between layers or materials;

[0055] The recognition algorithms used for the image recognition include, but are not limited to, edge detection, image segmentation, feature extraction, template matching, and defect classification and object detection algorithms based on convolutional neural networks.

[0056] Further, the process monitoring module deploys automated vision inspection stations at multiple key process points in the production process of the FPC flexible circuit board to perform real-time image acquisition and analysis on the products. These processes are selected as key points because they are crucial to the final performance and reliability of the FPC flexible circuit board, are vulnerable to fluctuations in process parameters, and are prone to defects.

[0057] By using image recognition algorithms to perform image acquisition and analysis on production features at these key processes, significant quality control benefits have been achieved, enabling the system to perform real-time, high-precision, and objective monitoring and measurement on key production features such as registration targets, circuit patterns, hole positions, pads, character markings, surface defects, and interlayer alignment of the FPC flexible circuit board during the manufacturing process.

[0058] Further, the process of generating the FPC label includes:

[0059] Directly obtaining the visual fingerprint of the production features of the FPC flexible circuit board at each key process through image recognition, where the visual fingerprint at least includes the time of the corresponding key process of the FPC flexible circuit board, key process features, and processing data features;

[0060] After image recognition and data acquisition are completed at each key process, the visual fingerprints of the FPC flexible circuit board in all key processes are linked in chronological order and marked with a unique identifier, which is generated and attached at the first key process.

[0061] Further, the visual fingerprint at least includes: the precise timestamp of the key process for this image acquisition; key process features obtained through image analysis, such as key geometric measurement values such as the average line width of the circuit, minimum line pitch, hole position deviation, cover film alignment offset, etc.; a list of detected production feature defects, such as defect type, coordinates of the defect on the board, defect severity assessment, etc.; and any processing data feature identifiers read from the image.

[0062] Further, in addition to the visual fingerprint obtained through image recognition, the equipment at this key process will also provide processing data through sensors or communication interfaces, such as actual temperature, pressure, speed, current, chemical solution concentration, etc., and these processing data are associated with the visual fingerprint of this visual inspection in terms of time and space.

[0063] Further, when the FPC flexible circuit board enters the first key process of the production process, the system generates a unique identifier for it, such as a two-dimensional code etched by laser or a specific coded pattern formed by identifying drilling, and attaches the identifier to the FPC or physically associates it with the FPC. After image recognition and data collection are completed in each subsequent key process, the generated visual fingerprint of the process and the associated processing data are packaged and uploaded to the central database through the network, and are associated and stored with the unique identifier of the FPC. According to the recorded timestamp, the system automatically links and organizes all visual fingerprints and processing data belonging to the same unique identifier in chronological order to form a complete FPC label file, which is the full-process digital record of the FPC from raw materials to the current process. Table 1 shows the detailed visual fingerprint data obtained through image recognition in the etching process of some FPC flexible circuit boards.

[0064] Table 1. Partial Visual Fingerprint Data

[0065]

[0066] By deploying intelligent monitoring based on image recognition in key processes and generating digital labels containing full-process visual fingerprints and processing data for each FPC flexible circuit board, high-precision, real-time, and non-contact monitoring of the FPC flexible circuit board production process is achieved, which makes the flexible circuit board production process that was originally difficult to quantify and trace completely transparent and traceable.

[0067] The intelligent analysis module conducts FPC quality traceability and management, including obtaining the electrical performance test results of the FPC flexible circuit board, generating FPC labels, searching for and comparing similar FPC label sets, checking whether there are problems with FPC flexible circuit boards in the same batch, and feeding back key FPC process steps.

[0068] Further, the process of obtaining the electrical performance test results of the FPC flexible circuit board and generating FPC labels includes:

[0069] After the FPC flexible circuit board completes all manufacturing processes, it enters the electrical performance test link; the electrical performance test includes at least a flying probe tester, an ICT tester, and a high-frequency tester;

[0070] The electrical performance of the FPC flexible circuit board is detected through the electrical performance test, and the electrical performance includes at least conduction, insulation, impedance, and capacitance; the test results are automatically uploaded to the intelligent analysis module through the device interface and associated with the unique identifier of the FPC flexible circuit board;

[0071] If the test result of the FPC flexible circuit board does not meet the standard, immediately mark the status of the FPC flexible circuit board and ensure that detailed failure test data is added to the FPC label. The failure test data shall at least include a list of failed test items, failure types, failure location information, comparison of actual measurement values with standard values, test conditions, test equipment and fixture information, and test timestamps.

[0072] Automatically obtain the electrical performance test results of the FPC flexible circuit board and associate them with the FPC label. Especially for non-compliant products, add detailed failure test data to their labels, providing indispensable data for the intelligent analysis module to conduct accurate root cause analysis, enabling quality engineers and intelligent systems to quickly and accurately understand "where" and "how" electrical failures occur in the product, and being able to trace back its manufacturing process based on this, efficiently identifying potential process problems or visual abnormalities that lead to failures, greatly improving the problem diagnosis efficiency.

[0073] Furthermore, the process of finding similar FPC label sets and comparing them to check if there are problems with FPC products in the same batch includes:

[0074] Retrieve FPC labels from the database. Similarity searches can be performed through several search methods, where the search methods shall at least include searching based on product attributes and batches, searching based on production characteristics, and searching based on test results; construct a set of similar FPC labels according to the search results;

[0075] After finding the set of similar FPC labels, analyze the electrical performance test results of the FPC flexible circuit boards within the set. If the test results of the FPC flexible circuit boards in the set are unqualified in several electrical performance test items and the unqualified ratio exceeds the set threshold ratio, it is determined that there are quality problems with the set of similar FPC labels;

[0076] When it is determined that there are quality problems with the set of similar FPC labels, the intelligent analysis module retrospectively analyzes the FPC label data corresponding to each key process of the unqualified FPC flexible circuit boards in the set of similar FPC labels, compares the FPC label data of the qualified FPC flexible circuit boards. If the difference exceeds the preset difference threshold, mark the key process as abnormal;

[0077] The intelligent analysis module counts the number of abnormalities in the key processes and generates a detailed analysis report. The analysis report shall at least include the key processes with problems, reasons for abnormalities, relevant production characteristics, and FPC label data.

[0078] Furthermore, this process aims to identify groups of products with similar characteristics by intelligently mining and analyzing the labeled data accumulated from a large number of FPC flexible circuit board products, find common problems among these groups, and ultimately trace back to the possible root processes.

[0079] Furthermore, searching based on product attributes and batches is the most basic search method. The system directly groups the FPC product label sets that belong to the same category or batch according to information such as product model, production order number, production batch number, or raw material batch number in the FPC label;

[0080] Furthermore, searching based on production characteristics is that the intelligent analysis module reads the "production characteristics" and "key process characteristics" of each key process recorded in the FPC label, and uses feature vector matching, clustering algorithms, or dimensionality reduction techniques to find groups of FPC labels with dense data distribution or close distances in the multi-dimensional production feature space. For example, all FPC labels with line width measurement values fluctuating within a certain specific range in the image detection of the etching process can be grouped into one category; or all FPC labels with similar bubble defect patterns in the lamination process can be grouped into one category.

[0081] Furthermore, searching based on test results is that the intelligent analysis module reads the electrical performance test results recorded in the FPC label and searches for FPC labels that show similar results in specific test items. For example, search for all product labels that fail at a specific short-circuit test point, or search for all product labels with similar but unqualified measurement values in a certain impedance test item.

[0082] Furthermore, after finding the set of similar FPC labels, the intelligent analysis module first conducts statistical analysis on the electrical performance test results of all FPC flexible circuit boards in the set, and will check the test results of the FPC flexible circuit boards in the set on several electrical performance test items. If the unqualified proportion of the FPC flexible circuit boards in the set exceeds the set threshold proportion in a certain test item, for example, more than 5% of the products fail in a specific short-circuit test, the system determines that there is a quality problem with this set of similar FPC labels.

[0083] Furthermore, when it is determined that there is a quality problem with a set of similar FPC labels, the intelligent analysis module retrospectively analyzes the FPC label data corresponding to each key process of the unqualified FPC flexible circuit boards in this set. At the same time, the system retrieves and compares the label data of qualified FPC flexible circuit boards produced during the same period from the database. The focus of the comparison is to find the differences in the production characteristics of the key processes between the unqualified product set and the qualified product set.

[0084] By performing similarity searches and set analyses on FPC labels, it is possible to automatically discover product groups with common characteristics and potential quality problems from a vast amount of production data, turning passive searches into proactive warnings. It can accurately locate the production batches or process windows where problems occur, and by retrospectively comparing the detailed label data of defective products and qualified products in key processes, efficiently and scientifically diagnose the root cause processes that lead to problems.

[0085] By deeply integrating requirements analysis, process monitoring, and intelligent analysis functions based on image recognition, this system provides a comprehensive intelligent solution for FPC flexible circuit board production, significantly enhancing the automation and intelligence levels of the production process and overcoming the limitations brought by traditional manual dependence and information silos. The precise image monitoring throughout key processes enables real-time, objective, and detailed control of product quality and process status, capable of detecting and preventing defects earlier and more accurately, thereby greatly improving production yields and reducing rework and waste. The integrated data collection and intelligent analysis, especially the quality traceability and root cause diagnosis combined with visual information throughout the process, make problem-solving more efficient and accurate, providing strong data support for continuous process optimization and management decision-making.

[0086] Example Two:

[0087] During the production process of FPC flexible circuit boards, a certain company introduced an intelligent production system for FPC flexible circuit boards based on image recognition provided by the present invention to improve the supervision and control efficiency of the entire production process. The specific implementation method is as follows:

[0088] The requirements analysis module is used to analyze the information of production entities on the FPC production line, construct a production information model, and allocate production tasks according to the production information.

[0089] Furthermore, the production entities at least include FPC flexible circuit board materials to be produced, production equipment, tooling fixtures, production personnel, and production orders to be executed.

[0090] The production information model constructed based on the information of production entities at least includes a material sub-model, an equipment sub-model, an order sub-model, a personnel sub-model, and association relationships, where the association relationships record the matching and compatibility rules between materials and orders, materials and equipment, and equipment and orders.

[0091] Furthermore, the process of production task allocation includes:

[0092] Construct a scheduling algorithm based on order priority, delivery date, material availability and its quality assessment, real-time equipment load and readiness status, and changeover cost; the scheduling algorithm continuously monitors data changes in the production information model. When it detects that the production information model receives information of a new production entity, the scheduling algorithm is activated to generate an optimal production task sequence and equipment allocation plan, and execute the allocation plan.

[0093] The process monitoring module performs image recognition on key processes on the FPC production line and processes according to the production characteristics of the FPC flexible circuit board in key processes to generate an FPC label, where the FPC label includes at least time, key process characteristics, production characteristics, and processing data.

[0094] Further, the key processes at least include blanking, drilling, graphic transfer, etching, lamination, cutting, and surface treatment.

[0095] The image recognition is to collect and analyze images of the production characteristics of the FPC flexible circuit board in the key processes, and the production characteristics at least include registration targets, circuit patterns, hole positions, pads, character markings, surface defects, and the alignment relationship between layers or materials.

[0096] The recognition algorithms used for the image recognition include but are not limited to edge detection, image segmentation, feature extraction, template matching, defect classification based on convolutional neural networks, and object detection algorithms.

[0097] Further, the process of generating the FPC label includes:

[0098] Directly obtain the visual fingerprints of the production characteristics of the FPC flexible circuit board in each key process through image recognition, and the visual fingerprints at least include the time, key process characteristics, and processing data characteristics of the FPC flexible circuit board in the corresponding key process.

[0099] After image recognition and data collection are completed in each key process, link the visual fingerprints of the FPC flexible circuit board in all key processes in chronological order and mark a unique identifier, which is generated and attached in the first key process.

[0100] The intelligent analysis module performs FPC quality traceability and management, including obtaining the electrical performance test results of the FPC flexible circuit board, generating an FPC label, finding and comparing similar FPC label sets, checking whether there are problems with FPC flexible circuit boards of the same batch, and feeding back key FPC process steps.

[0101] Further, the process of obtaining the electrical performance test results of the FPC flexible circuit board and generating an FPC label includes:

[0102] After the FPC flexible circuit board completes all manufacturing processes, it enters the electrical performance testing phase; the electrical performance testing at least includes a flying probe tester, an ICT tester, and a high-frequency tester;

[0103] The electrical performance of the FPC flexible circuit board is detected through electrical performance testing, and the electrical performance at least includes conduction, insulation, impedance, and capacitance; the test results are automatically uploaded to the intelligent analysis module through the device interface and associated with the unique identifier of the FPC flexible circuit board;

[0104] If the test result of the FPC flexible circuit board is unqualified, immediately mark the status of the FPC flexible circuit board and ensure that detailed failure test data is added to the FPC label. The failure test data at least includes a list of failed test items, failure types, failure location information, comparison of actual measurement values with standard values, test conditions, test equipment and fixture information, and test timestamps. Table 2 shows the test data of some FPC flexible circuit boards.

[0105] Table 2, Test Data of Some FPC Flexible Circuit Boards

[0106]

[0107] Furthermore, the process of finding similar FPC label sets and comparing them to check if there are problems with FPC products of the same batch includes:

[0108] Retrieve FPC labels from the database. Similarity searches can be performed through several search methods, and the search methods at least include searching based on product attributes and batches, searching based on production characteristics, and searching based on test results; construct a set of similar FPC labels according to the search results;

[0109] After finding the set of similar FPC labels, analyze the electrical performance test results of the FPC flexible circuit boards within the set. If the test results of the FPC flexible circuit boards in the set are unqualified in several electrical performance test items and the unqualified proportion exceeds the set threshold proportion, it is determined that there are quality problems with the set of similar FPC labels;

[0110] When it is determined that there are quality problems with the set of similar FPC labels, the intelligent analysis module retrospectively analyzes the FPC label data corresponding to each key process of the unqualified FPC flexible circuit boards in the set of similar FPC labels, compares the FPC label data of the qualified FPC flexible circuit boards. If the difference exceeds the preset difference threshold, mark the key process as abnormal;

[0111] The intelligent analysis module counts the number of anomalies in key processes and generates a detailed analysis report, which at least includes the key processes with problems, the reasons for anomalies, relevant production characteristics, and FPC label data. Part of the content of the analysis report is shown in Table 3.

[0112] Table 3. Analysis Report

[0113]

[0114] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent production system for FPC flexible circuit boards based on image recognition, characterized in that, Including: The requirement analysis module is used to analyze the information of the production entities on the FPC production line, construct a production information model, and allocate production tasks according to the production information. The process monitoring module performs image recognition on the key processes of the FPC production line, and processes according to the production characteristics of the FPC flexible circuit board in the key processes to generate an FPC label, where the FPC label at least includes time, key process characteristics, production characteristics, and processing data. The intelligent analysis module conducts FPC quality traceability and management, including obtaining the electrical performance test results of the FPC flexible circuit board, generating an FPC label, searching for and comparing similar FPC label sets, checking whether there are problems with the FPC flexible circuit boards of the same batch, and feeding back the key FPC process steps.

2. The intelligent production system of an FPC flexible circuit board based on image recognition according to claim 1, wherein, The production entities at least include the FPC flexible circuit board materials to be produced, production equipment, tooling fixtures, production personnel, and production orders to be executed. The production information model constructed according to the information of the production entities at least includes a material sub-model, an equipment sub-model, an order sub-model, a personnel sub-model, and association relationships, where the association relationships record the matching and compatibility rules between materials and orders, materials and equipment, and equipment and orders.

3. An intelligent production system for FPC flexible circuit boards based on image recognition according to claim 1, characterized in that, The process of production task allocation includes: Construct a scheduling algorithm based on order priority, delivery date, available material quantity and its quality assessment, real-time equipment load, readiness status, and changeover cost; the scheduling algorithm continuously monitors the production information model. When it detects that the production information model receives the information of a new production entity, the scheduling algorithm starts, generates an optimal production task sequence and equipment allocation plan, and executes the allocation plan.

4. An intelligent production system for FPC flexible circuit boards based on image recognition according to claim 1, characterized in that, The key processes at least include blanking, drilling, graphic transfer, etching, lamination, cutting, and surface treatment. The image recognition is to collect and analyze images of the production characteristics of the FPC flexible circuit board in the key processes, and the production characteristics at least include registration targets, circuit patterns, hole positions, pads, character markings, surface defects, and the alignment relationship between materials. The recognition algorithms used for the image recognition include but are not limited to edge detection, image segmentation, feature extraction, template matching, defect classification based on convolutional neural networks, and object detection algorithms.

5. An intelligent production system for FPC flexible circuit boards based on image recognition according to claim 1, characterized in that, The process of generating the FPC label includes: Directly obtain the visual fingerprints of the production characteristics of the FPC flexible circuit board in each key process through image recognition, where the visual fingerprints at least include the time, key process characteristics, and processing data characteristics of the FPC flexible circuit board in the corresponding key process. After image recognition and data collection are completed in each key process, link the visual fingerprints of the FPC flexible circuit board in all key processes in chronological order and mark a unique identifier, which is generated and attached in the first key process.

6. The intelligent production system of an FPC flexible circuit board based on image recognition according to claim 1, characterized in that, The process of obtaining the electrical performance test results of the FPC flexible circuit board and generating an FPC label includes: After the FPC flexible circuit board completes all manufacturing processes, it enters the electrical performance test link; the electrical performance test at least includes a flying probe tester, an ICT tester, and a high-frequency tester. The electrical performance of the FPC flexible circuit board is detected through electrical performance tests, and the electrical performance at least includes conduction, insulation, impedance, and capacitance reactance; the test results are automatically uploaded to the intelligent analysis module through the device interface and associated with the unique identifier of the FPC flexible circuit board; If the test result of the FPC flexible circuit board is unqualified, the status of the FPC flexible circuit board is immediately marked, and it is ensured that detailed failure test data is added to the FPC label. The failure test data at least includes a list of failed test items, failure types, failure location information, comparison of actual measurement values with standard values, test conditions, test equipment and fixture information, and test timestamps.

7. An intelligent production system for FPC flexible circuit boards based on image recognition according to claim 1, characterized in that, The process of finding a similar set of FPC labels and comparing them to check if there are problems with FPC products of the same batch includes: Retrieve FPC labels from the database and perform similarity searches through several search methods. The search methods at least include searching based on product attributes and batches, searching based on production characteristics, and searching based on test results; construct a set of similar FPC labels according to the search results; After finding the set of similar FPC labels, analyze the electrical performance test results of the FPC flexible circuit boards in the set. If the test results of the FPC flexible circuit boards in the set are unqualified in several electrical performance test items and the unqualified ratio exceeds the set threshold ratio, it is determined that there is a quality problem with the set of similar FPC labels; When it is determined that there is a quality problem with the set of similar FPC labels, the intelligent analysis module backtracks and analyzes the FPC label data corresponding to each key process of the unqualified FPC flexible circuit boards in the set of similar FPC labels, and compares it with the FPC label data of the qualified FPC flexible circuit boards. If the difference exceeds the preset difference threshold, the key process is marked as abnormal; The intelligent analysis module counts the number of abnormalities in the key processes and generates a detailed analysis report. The analysis report at least includes the key processes with problems, the reasons for the abnormalities, relevant production characteristics, and FPC label data.

Images