WO2017084186A1

WO2017084186A1 - System and method for automatic monitoring and intelligent analysis of flexible circuit board manufacturing process

Info

Publication number: WO2017084186A1
Application number: PCT/CN2015/100210
Authority: WO
Inventors: 罗家祥; 李致富; 吕斯俊; 王加朋; 李康婧; 胡跃明
Original assignee: 华南理工大学
Priority date: 2015-11-18
Filing date: 2015-12-31
Publication date: 2017-05-26
Also published as: CN105372581B; CN105372581A

Abstract

A system and method for automatic monitoring and intelligent analysis of a flexible circuit board manufacturing process. The system comprises a basic information module, a parameter acquisition module, a comprehensive database module, an intelligent data analysis module and a monitoring display and data report module. The parameter acquisition module comprises a microscope automatic data acquisition device and a copper thickness testing device; the microscope automatic data acquisition device acquires the line width and line distance of an etching procedure and the aperture of a drilling procedure; and the copper thickness testing device is adopted to measure the copper thickness, and the statistics of procedure defect data is carried out. A multivariate statistical process control method, a neural network and a support vector machine are adopted as basic analysis methods, so as to perform intelligent analysis on acquired data, predict procedure and production line abnormalities, and identify generated abnormality sources. Automatic monitoring on the manufacturing process and the quality of a flexible circuit board can be realized, and the stability of procedures and the yield of products can be improved.

Description

Flexible circuit board manufacturing process automatic monitoring and intelligent analysis system and method

Technical field

The invention relates to the field of automatic monitoring and intelligent analysis technology of a flexible printed circuit board (Flexible Printed Circuit Board) manufacturing process, and particularly relates to an automatic monitoring and intelligent analysis system and method for a flexible circuit board manufacturing process.

Background technique

Flexible circuit boards are widely used in electronic products with requirements for miniaturization, light weight, and mobility, including display driver chips, camera modules, RF function modules, MEMS modules, fingerprint recognition modules, and finance. IC cards, etc., involving computers, mobile communications, displays, instrumentation, medical devices, smart cards, etc., as well as aerospace, defense, military and other fields, in the development trend of electronic products are getting smaller and thinner, flexible circuits The board will further replace the hard circuit board and usher in more room for development.

Flexible circuit board manufacturing process is more complicated, in the general manufacturing process, it needs to be laminated copper foil, paste dry film, laser direct writing exposure, development, etching, stripping, laser drilling, copper plating, Solder mask ink, continuity test and molding process. Among them, the key processes mainly include: etching, laser drilling, copper thinning and so on.

In order to improve the stability and yield of flexible circuit board manufacturing, the main physical parameters and defects of the key processes of the flexible circuit board can be automatically monitored during the manufacturing process of the flexible circuit board, so as to take timely action in the event of critical process abnormalities. Emergency measures to reduce the risk of line failures. This kind of automatic monitoring system will help reduce the cost of the enterprise and is therefore receiving more and more attention from the enterprise. The main physical parameters of the key processes of flexible circuit boards are line width, line spacing and aperture size; and the product defects involved mainly include open circuit, short circuit, line gap, bump, residual copper and so on. Automatic detection and monitoring of key physical parameters of these flexible circuit boards can detect abnormal conditions of key processes in a timely manner, and can perform statistics and analysis on data in a certain period of time to improve and optimize the flexible circuit board manufacturing process.

In the current flexible circuit board manufacturing industry, traditional manual inspection is still the main method of detection. However, as the line width and line spacing become smaller and smaller, the image density becomes higher and higher, and the traditional manual detection has a long detection time and a high false positive rate. The law meets the needs of the industry. For the key processes of flexible circuit boards, image patterning and processing are performed by pattern recognition technology, and the key physical parameters of the flexible circuit board are compared with the standard parameters stored in the computer to judge the abnormal state of the key process, thus the abnormal process Processed to automate the detection and monitoring of critical processes.

Summary of the invention

In view of the technical problems existing in the prior art, the object of the present invention is to provide an intelligent monitoring and intelligent analysis system and device for automatically monitoring, defect identification and defect cause analysis of a production process of a flexible circuit board manufacturing key process. And methods.

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

Automatic monitoring and intelligent analysis system for flexible circuit board manufacturing process, including basic data module, data acquisition module, comprehensive database module, intelligent data analysis module, and monitoring display and data reporting module. The data acquisition module includes a microscope automatic data acquisition device, a copper thickness test device and other data acquisition devices, which are mainly used for collecting key physical parameters of the flexible circuit board manufacturing process and flexible circuit board quality data; the basic data module includes process information. And quality inspection specifications, mainly used as standard information materials for evaluation and inspection of various process quality and defect inspection of flexible circuit boards; intelligent data analysis module includes: T ² control method using multivariate statistical process for single process anomalies It is predicted that the fusion neural network and support vector machine method are used to predict the abnormality of the whole manufacturing line of flexible circuit board. The neural network based on genetic algorithm optimization is used to identify the abnormal source of the flexible circuit board manufacturing process. The main realization is the flexible circuit. Intelligent quality control of the production process, automatic abnormal state identification and abnormal positioning of the production process, provide reference for the maintenance personnel to eliminate abnormal faults; the comprehensive database module is mainly used to store the data collected by the data acquisition module, process information data and quality inspection specification information. And wisdom And data analysis reports generated by the analysis module.

Further, the process information of the basic data module mainly includes: 1) a list of process personnel, a record process worker and a responsible process thereof; 2) device information, including basic information such as a device name, a category, and a model; a basic data module The quality inspection specifications include: IPC-6013B "Quality and Performance Specifications for Flexible Printed Boards", and quality inspection procedures developed within the company.

Further, the microscope automatic data acquisition device comprises: a host computer system and a microscope detection platform. The upper computer system includes a motion control module and a microscope vision control processing module. The microscope inspection platform includes an electric precision stage, a shifting rod, a motor control box, a microscope mounting bracket, a microscope, a light source, and a digital camera. The microscope is mounted on a microscope mount, the digital camera is mounted above the microscope, the light source is mounted directly behind the microscope, and the motorized precision stage is mounted directly below the microscope and connected to the motor control box. The microscope vision control processing module is connected to the digital camera, and the motion control module is connected to the motor control box. A servo driver and a power supply are installed in the motor control box, and the servo driver is connected to the host computer through a control card.

Further, the host computer system collects images through the motion control module and the microscope vision control processing module, and identifies key physical parameters and defect data of the flexible circuit board corresponding to each key process. The key processes of the flexible circuit board corresponding to the microscope automatic data acquisition device are an etching process and a laser drilling process; key physical parameters include line width, line spacing, aperture size, etc.; defect data mainly includes open circuit, short circuit, and line Notches, bumps, residual copper, etc.

Further, the host computer system includes a motion control module that is responsible for moving the electric precision stage on which the flexible circuit board is placed. After the upper computer sends the image acquisition command through the motion control module, the servo driver in the motor control box drives the servo motor to control the X and Y axis motion guides to move the electric precision stage respectively, and collect images with the digital camera and the light source.

Further, the host computer system includes a microscope vision control processing module, which is responsible for controlling a digital camera to obtain a partial image of the flexible circuit board, image stitching and image processing after being enlarged by the microscope, and according to standard data of the production process of the flexible circuit board process, A comparison of the data is achieved, ultimately showing a flexible circuit board image with defective areas and specific defect details.

Further, the copper thickness testing device comprises a copper thickness measuring instrument and a data communication software, the copper thickness measuring instrument is mainly used for measuring the thickness of the copper foil, and the data communication software is mainly used for collecting and transmitting the copper thickness data of the data.

Further, the comprehensive database is mainly used for storing basic information of key processes, initial data collected, and statistical information of the data and final results of intelligent analysis, including statistical information of each physical parameter, The diagnosis result of the abnormality of the sequence or the production line and the abnormal source information when there is an abnormality; all the data information is finally stored in the database in the form of a report for the engineers and managers to query in real time.

The method for monitoring and analyzing the automatic monitoring and intelligent analysis system of the flexible circuit board manufacturing process includes monitoring the key processes such as copper thinning, etching, drilling, etc. in the manufacturing process of the flexible circuit board, and specifically includes the following steps:

1. According to the process selected by the user, the algorithm will query and read the corresponding parameter data set from the database module.

2. After the completion of step 1, judge whether to perform preprocessing according to the data type included in the data set. If it is a measurement value type, all the data in the parameter data set is standardized and preprocessed, and if it is a count type, the parameter is The data is scaled.

3. After the completion of step 2, based on the parameter variable category and the standardized data, a multivariate statistical T ² model is established to visually monitor the abnormal situation of the selected process. If the T ² control map detects that production abnormal fluctuations are out of control, an alarm is issued and uploaded to the monitoring display and data reporting module, otherwise no response is made.

An intelligent analysis method is designed for the abnormal recognition of the whole process of the flexible circuit board manufacturing process, including the following steps:

1. According to the user's choice, read the training data or the data to be monitored for the abnormal process identification of the flexible circuit board manufacturing. Determine whether to perform preprocessing according to the data type included in the data set. If it is a measurement value type (including line width, line spacing, aperture, and hole roundness), all data in the parameter data set is standardized and preprocessed, if it is a count The type (including the number of occurrences of various defects in the statistical time), the parameter data is scaled.

2. After the completion of step 1, the neural network method extracts features.

3. After the completion of step 2, if the user selects the training model, the support vector machine model is trained with the normal and abnormal feature data, and the Gaussian kernel function is used and the grid parameters are used to determine the relevant parameters, thereby completing the establishment of the support vector machine model. . Otherwise, follow step 4 for intelligent analysis of the data.

4. After the completion of step 2, the batch data of the processing process is monitored using the support vector machine model. If check If it is detected that the abnormal fluctuation of production is out of control, an alarm is issued and uploaded to the monitoring display and data report module, otherwise no response is made.

The abnormal source identification (abnormal positioning) algorithm for monitoring a flexible circuit board production process includes the following steps:

1. According to the user's choice, read the training data or the data to be monitored for the abnormal positioning of the whole process of the flexible circuit board manufacturing. The key parameters of the flexible circuit board manufacturing process involved are queried and read from the database module to form batch data. It is judged whether the batch data is preprocessed. If the metering value type requires preprocessing, all the data in the parameter data set is subjected to standardization preprocessing, and if it is a counting type, it is not processed.

2. After the completion of step 1, if the user selects the training model, the fusion genetic algorithm and the abnormal source recognition model of the deep learning neural network are established on the pre-processed feature data set. The genetic algorithm adopts binary coding technology, and uses the total error square function as the fitness function. Through the selection, crossover, mutation and other evolutionary operators, the structure and weight of the optimized deep learning neural network are selected. Otherwise, follow step 3 for intelligent analysis of the data.

3. After the completion of step 2, use the abnormal source identification model established by 2 to monitor the whole process of the flexible circuit board manufacturing process. If abnormal fluctuation occurs, the output of the abnormal source identification model can be located to locate the process of the out-of-control abnormality. Send the results to the monitor display and data report module, otherwise no response will be made.

The flexible circuit board quality evaluation method of the microscope automatic data acquisition device is characterized in that it comprises the following steps:

1. After placing the flexible circuit board on the electric precision stage and fixing it with the fixing device, the operator logs into the host computer system, turns on the light source, and switches to the digital camera drawing mode through the shift lever after manual focusing.

2. After the completion of step 1, the servo motor drives the guide rail to move the electric precision stage to return the system to the detection origin.

3. After the completion of step 2, the operator inputs or downloads the standard file of the flexible circuit board to be tested, such as Gerber file, CAD file, etc., in the database, and then parses the flexible circuit board standard file to obtain the standard. Standard data required for graphs and quality evaluation.

4. After the completion of step 3, the digital camera captures the image, and the servo motor moves the X and Y axis motion guides to make the system recognize and align the reference point.

5. After the completion of step 4, the motion control module controls the servo driver in the motor control box to drive the servo motor, and moves the electric precision stage by moving the X and Y axis motion guides. According to the order from left to right and from top to bottom, the digital camera is enlarged by a microscope, and then the flexible circuit board to be tested is partially taken. At the same time, the microscopic vision control processing module is to be tested for a partial view of the flexible circuit board. Perform pre-processing, then use the splicing method based on feature template matching feature points to splicing the image, and complete the smoothing of the image, so repeat the drawing and splicing until the flexible circuit board is scanned, and finally obtain the flexible circuit board to be tested. Global map.

6. After the completion of step 5, the image is binarized and the connected domain is searched, and the connected domain is compared with the connected domain in the circuit diagram template by using the connected domain statistical centroid and area as the matching criterion to determine the mismatched area (defect area); The thinning method detects the line width and the line spacing; uses the Hough transform to identify the position of the hole, and obtains the aperture size based on the area information. Compared with the line width, line spacing and aperture size in the standard figure and standard data, the quality evaluation information is obtained; the contrast method is used to identify the defects such as open circuit, short circuit and residual copper.

7. After the completion of step 6, the upper computer system displays the full image of the defect area and the specific defect details, and provides an alarm according to the pre-recorded threshold information, so that the operator can timely handle the abnormal process.

8. After the completion of step 7, the host computer system stores the detection result in the local computer, and uploads related images, quality evaluation information, defect information, and defect data to the comprehensive database module for subsequent statistical processing.

The invention realizes the whole intelligence of automatic monitoring, defect identification and defect cause analysis on the production process of the key processes of the flexible circuit board manufacturing process by collecting, monitoring and intelligently analyzing the key physical parameters of the flexible circuit board corresponding to the key processes. Monitoring and intelligent analysis.

The present invention has the following positive advantages and effects over the prior art:

1. Automated monitoring and analysis of the manufacturing process of flexible circuit boards using automated devices and systems. Compared with the traditional manual detection, it not only reduces the false alarm rate of detection, increases the type of detection, but also greatly improves the production efficiency and automation level of flexible circuit boards.

2. The microscope and precision electric platform are introduced into the monitoring of the key processes of the flexible circuit board. On one hand, the accuracy of the system monitoring object is effectively improved, and on the other hand, the detection process is simplified and the efficiency of the system detection is improved.

3. Using statistical process control, neural network and other methods to achieve intelligent quality monitoring and fault diagnosis, thereby more effectively ensuring the quality of flexible circuit board production and improving the ability of production process fault diagnosis.

DRAWINGS

Figure 1 is a block diagram showing the automatic monitoring and intelligent analysis system of the flexible circuit board manufacturing process;

2 is a perspective view of a microscope automatic data acquisition device;

3 is a structural block diagram of a microscope automatic data acquisition device;

Figure 4 is a detailed implementation diagram of the automatic monitoring and intelligent analysis system for the manufacturing process of the flexible circuit board.

detailed description

The present invention will be further described in detail below with reference to the embodiments and the accompanying drawings. However, the embodiments of the present invention are not limited thereto, and it should be noted that the following procedures, which are not specifically described in detail, can be implemented by those skilled in the art according to the conventional practice. of.

Example

As shown in FIG. 1, the automatic monitoring and intelligent analysis system of the flexible circuit board manufacturing process includes a basic data module, a data acquisition module, a comprehensive database module, an intelligent data analysis module, and a monitoring display and data report module. The data acquisition module includes a microscope automatic data acquisition device, a copper thickness test device and other data acquisition devices, which are mainly used for collecting key physical parameters of the flexible circuit board manufacturing process and flexible circuit board quality data; the basic data module includes process information. And quality inspection specifications, mainly used as standard information materials for evaluation and inspection of various process quality and defect inspection of flexible circuit boards; intelligent data analysis module includes statistical process control algorithm to identify abnormal state, fusion neural network and support vector machine algorithm automatically Identifying anomalies, genetic algorithm-based neural network deep learning methods for the identification of abnormal source of flexible circuit board production process (abnormal positioning), mainly to achieve scratch Intelligent quality control of the production process of the circuit board, automatic abnormal state identification and abnormal positioning of the production process, provide reference for the maintenance personnel to eliminate abnormal faults; the comprehensive database module is mainly used to store the data collected by the data acquisition module, process information data and quality inspection. Specification information and analysis results and data reports generated by the intelligent analysis module.

The process information mainly includes information such as personnel and equipment of the process. The quality inspection specifications mainly include: IPC-6013B "Qualification and Performance Specifications of Flexible Printed Board", and quality inspection procedures formulated by the company.

As shown in FIG. 2, the microscope automatic data acquisition device includes a host computer system and a microscope inspection platform. The upper computer system includes a motion control module and a microscope vision control processing module. The microscope inspection platform includes a motorized precision stage 1, a shifting rod 3, a motor control box 11, a microscope fixing bracket 6, a microscope 5, a light source 2, and a digital camera 4. The electric precision stage 1 includes an X-axis servo motor 10, a Y-axis servo motor 13, an X-axis motion guide 7, and a Y-axis motion guide 12. The microscope 5 is mounted on the microscope fixing bracket 6, the digital camera 4 is mounted above the microscope 5, the light source 2 is mounted directly behind the microscope 5, and the electric precision stage 1 is mounted directly below the microscope 5, and is connected to the motor control box 11. .

As shown in Fig. 3, the microscope vision control processing module is connected to the digital camera, and the motion control module is connected to the motor control box. A servo driver and a power supply are installed in the motor control box, and the servo driver is connected to the host computer through a control card.

The microscope inspection platform described in this example uses a digital camera from Basler's piA2400 model, using a model MJ51 microscope and a halogen light source with a blue filter. The electric precision stage uses Panasonic's servo motor and servo drive, and is connected to the host computer using GTS-400-PV motion control card from Gutech (Shenzhen) Co., Ltd.

As shown in FIG. 4, the flexible circuit board manufacturing process automatic monitoring and intelligent analysis system uses a distributed structure to construct the system. The automatic data acquisition and collection of the microscope is mainly responsible for data acquisition of key parameters and defects of the etching process and the drilling process; the copper thickness test device is mainly used for data collection of key parameters of the copper thinning process. The etching process monitoring station, the copper thinning process monitoring station and the drilling process monitoring station respectively implement the etching process, the copper thinning process and the drilling On-site automatic monitoring and intelligent analysis of key parameters and defect data of the process. The database server is mainly used to store system user data and historical data of monitoring and intelligent analysis. The work tasks of the quality department monitoring room mainly realize the data analysis, monitoring and intelligent analysis and fault diagnosis of the process comprehensive data. The general manager's office monitoring room mainly realizes the data review, monitoring and process comprehensive data analysis and fault diagnosis results of each process.

The key processes of the flexible circuit board corresponding to the microscope automatic data acquisition device of the present example are an etching process and a laser drilling process; key physical parameters include line width, line spacing, aperture size, etc.; defect data mainly includes open circuit, short circuit, and line Notches, bumps, residual copper, etc.

The upper computer system of the microscope automatic data acquisition device includes a motion control module, which is responsible for moving the electric precision stage on which the flexible circuit board is placed. After the upper computer sends the image acquisition command through the motion control module, the servo driver in the motor control box drives the servo motor to control the X and Y axis motion guides to move the electric precision stage respectively, and collect images with the digital camera and the light source.

The upper computer system of the microscope automatic data acquisition device further comprises a microscope vision control processing module, which is responsible for controlling the digital camera to obtain a partial image of the flexible circuit board, image stitching and image processing after being enlarged by the microscope, and according to the production process of the flexible circuit board. Standard data, which compares the data and ultimately displays a flexible board image with defective areas and specific defect details.

The method for evaluating the quality of a flexible circuit board of a microscope automatic data acquisition device includes the following steps:

3. After the completion of step 2, the operator inputs or downloads the standard file of the flexible circuit board to be tested, such as Gerber file, CAD file, etc. in the database, and then parses the flexible circuit board standard file to obtain the standard figure and quality. Evaluate the required standard data.

4. After the completion of step 3, the digital camera captures the image, and the servo motor moves the X and Y axis motion guides to make the system Identify and align the reference points.

6. After the completion of step 5, the image is binarized and the connected domain is searched, and the connected domain is compared with the connected domain in the circuit diagram template by using the connected domain statistical centroid and area as the matching criterion to determine the mismatched area (defect area); The thinning method detects the line width and the line spacing; uses the Hough transform to identify the position of the hole, and obtains the aperture size based on the area information. Compared with the line width, line spacing and aperture size in the standard map and standard data, the quality evaluation information is obtained; and the defects such as open circuit, short circuit and residual copper are identified by the comparison method.

The copper thickness test device of the present example includes a copper thickness measuring instrument and a data communication software. The copper thickness measuring instrument is mainly used for measuring the thickness of the copper foil, and the data communication software is mainly used for collecting and transmitting the copper thickness data of the data.

The comprehensive database of this example is mainly used to store the basic information of key processes, the initial data collected, and various intermediate information obtained in the process of automatic monitoring and intelligent analysis and reasoning, and output result information after solving the problem. The results of the intelligent analysis are finally stored in the database in the form of reports for engineers and managers to query in real time.

This example designs a multivariate statistical process monitoring method for key processes such as copper thinning, etching, and drilling in the manufacturing process of flexible circuit boards, including the following steps:

1. According to the operation selected by the user, the algorithm will query and read the corresponding parameter number from the database module. According to the collection.

2. After the completion of step 1, determine whether to perform pre-processing according to the data type included in the data set, if it is a type of measurement value, such as the line width and line spacing of the etching process, the roundness, position and copper plating of the drilling process The copper thickness of the process, etc., is standardized for all data in the parameter data set. Let parameter i collect n data, denoted as {x _i1 ,...,x _in }, calculate the average value

And the standard deviation s _i , then the normalized processed data x' _ij is:

If it is a count value type, such as the number of boards with open circuit, short circuit, line gap, bump, and copper residual defect, the parameter data is scaled, that is, the number of boards with a certain defect is divided by the production. The total number of boards.

3. After the completion of step 2, based on the parameter variable category and the standardized data, a multivariate statistical χ ² model is established to visually monitor the abnormal conditions of the selected process. Taking the drilling process as an example, the key physical parameters are the aperture size and the roundness of the hole. The acquired n sets of aperture size and roundness data (ie, the number of attributes p=2) are expressed as x _i =(x). _I1 , x _i2 ) (i = 1, 2, ..., n), calculate the two data mean

Then calculate the covariance s _jh (j, h = 1, 2) between the attributes j and h and the corresponding covariance matrix S is

And the ith T ² statistic,

will

Compared with the upper and lower limits:

(where B(·) is the parameter

The beta distribution, 1-alpha is the confidence level). Finally, taking n as the horizontal axis and the vertical axis to draw the T ² control map, according to

Whether the control limit is exceeded or not is judged whether the quality of the product is out of control. If the data exceeds the upper and lower control limits, it is detected that the abnormal production fluctuation is out of control, and an alarm is issued and uploaded to the monitoring display and data report module, otherwise no response is made.

This example designs an intelligent analysis method for the abnormal recognition of the whole process of the flexible circuit board manufacturing process. Including the following steps:

1. According to the user's choice, read the training data or the data to be monitored for the abnormal process identification of the flexible circuit board manufacturing. According to the data type included in the data set, whether to perform preprocessing, if it is a measurement value type, all the data in the parameter data set is standardized and preprocessed, and if it is a counting type, the parameter data is proportionalized.

2. After the completion of step 1, the neural network method (such as a 3-layer BP network) is used to extract features, that is, input key parameters of all processes in the manufacturing process of the flexible circuit board, and output main characteristic data that affects the quality of the flexible circuit board.

3. After the completion of step 2, if the user selects the training model, the support vector machine model is trained with the normal and abnormal feature data, and the Gaussian kernel function is used and the penalty parameter C and the Gaussian kernel parameter γ in the model are determined using the grid method. , thus completing the establishment of the support vector machine model. Otherwise, follow step 4 for intelligent analysis of the data.

4. After the completion of step 2, the batch data of the processing process is monitored using the support vector machine model. If it is detected that the abnormal production fluctuation is out of control, an alarm is issued and uploaded to the monitoring display and data reporting module, otherwise no response is made.

This example designs an anomaly source identification (abnormal positioning) algorithm for monitoring the production process of flexible circuit boards, including the following steps:

2. After the completion of step 1, if the user selects the training model, a neural network anomaly source recognition model of the fusion genetic algorithm is established for the pre-processed feature data set. The genetic algorithm adopts binary coding technology, and uses the total error square function as the fitness function. Through the selection, crossover, mutation and other evolutionary operators, the structure and weight of the optimized neural network are selected. Otherwise, follow step 3 for intelligent analysis of the data.

As described above, the present invention can be preferably implemented.

Claims

Automatic monitoring and intelligent analysis system for flexible circuit board manufacturing process, including basic data module, parameter acquisition module, comprehensive database module, intelligent data analysis module, and monitoring display and data report module, characterized in that the parameter acquisition module includes microscope automatic data acquisition The device and copper thickness test device are mainly used to collect the key physical parameters of the flexible circuit board manufacturing process and the flexible circuit board quality data; the basic data module includes process information and quality inspection specifications, mainly used for flexible circuit boards. As the standard information data for evaluation and inspection during process quality and defect inspection; intelligent data analysis module includes: T 2 control method using multivariate statistical process to predict the anomaly of single process, using fusion neural network and support vector machine to predict The abnormality of the whole circuit of flexible circuit board manufacturing, the neural network based on genetic algorithm optimization is used to identify the abnormal source of the flexible circuit board manufacturing process, which mainly realizes the intelligent quality control of the flexible circuit board production process and the automatic abnormal state of the production process. Identification and It was located, to provide a reference for the maintenance personnel to remove the unusual fault; integrated database module is mainly used for analysis of the data storage and data reporting the collected data acquisition module, process information and quality inspection data intelligent analysis and specification information generated by the module.
The automatic monitoring and intelligent analysis system for manufacturing a flexible circuit board according to claim 1, wherein the process information of the basic data module mainly comprises: 1) a list of process personnel, a record process worker and a responsible process thereof; 2) Equipment information, including basic information such as the name, category and model of the equipment; the quality inspection specifications of the basic data module include: quality inspection specification IPC-6013B "Quality and performance specifications of flexible printed boards", quality established within the enterprise Inspection procedures.
The automatic monitoring and intelligent analysis system for manufacturing a flexible circuit board according to claim 1, wherein the microscope automatic data acquisition device comprises: a host computer system and a microscope detection platform; the host computer system comprises a motion control module and a microscope. Vision control processing module; microscope detection platform includes electric precision stage, conversion rod, motor control box, microscope fixed bracket, microscope, light source and digital camera; microscope is mounted on microscope fixed bracket, digital camera is installed above microscope, light source is installed Immediately behind the microscope, the electric precision stage is mounted directly under the microscope and connected to the motor control box; the microscope vision control processing module is connected to the digital camera, the motion control module is connected to the motor control box; the motor control box is installed There are servo drive, power supply, and servo drive connected to the host computer through the control card.
A flexible circuit board manufacturing process automatic monitoring and intelligent analysis system according to claim 3, wherein said upper computer system of the microscope automatic data acquisition device collects images through a motion control module and a microscope vision control processing module, and Identifying key physical parameters and defect data of the flexible circuit board corresponding to each key process; wherein the key processes of the flexible circuit board corresponding to the microscope automatic data acquisition device are an etching process and a laser drilling process; key physical parameters include a line Width, line spacing, aperture size; defect data mainly includes open circuit, short circuit, line gap, bump and residual copper.
3 . The automatic monitoring and intelligent analysis system for manufacturing a flexible circuit board according to claim 3 , wherein the upper computer system comprises a motion control module, and is responsible for moving an electric precision stage on which the flexible circuit board is placed; After the machine sends a command to acquire images through the motion control module, the servo drive in the motor control box drives the servo motor to control the X and Y axis motion guides to move the electric precision stage, and collect images with the digital camera and light source;

The upper computer system further comprises a microscope vision control processing module, which is responsible for controlling the digital camera to obtain partial image, image mosaic and image processing of the flexible circuit board after being enlarged by the microscope, and realizing the standard data according to the production process of the flexible circuit board process. The comparison of the data ultimately shows a flexible circuit board image with defective areas and specific defect details.
The automatic monitoring and intelligent analysis system for manufacturing a flexible circuit board according to claim 1, wherein the copper thickness measuring device comprises a copper thickness measuring instrument and a data communication software, and the copper thickness measuring instrument is mainly used for measuring copper foil. The thickness of the data communication software is mainly used for the collection and transmission of copper thick data of data.
The automatic monitoring and intelligent analysis system for manufacturing a flexible circuit board according to claim 1, wherein the integrated database module is mainly used for storing basic information of key processes, initial data collected, and statistics of the data. The final result of the information and intelligent analysis, including the statistical information of each physical parameter, the diagnosis result of whether the process or the production line is abnormal, and the abnormal source information when there is an abnormality; all the data information is finally stored in the database in the form of a report for the engineer. And managers in real time.
Automatic monitoring and intelligent analysis system monitoring using the flexible circuit board manufacturing process of claim 1. And analysis method, characterized in that: monitoring the key processes of copper thinning, etching, drilling, etc. in the manufacturing process of the flexible circuit board, including the following steps:

8.1 Query and read the corresponding parameter data set from the database module according to the operation selected by the user;

8.2 After the completion of step 8.1, it is judged whether preprocessing is performed according to the data type included in the data set, and if it is a measurement value type, all data in the parameter data set is standardized and preprocessed, and if it is a counting type, the parameter data is used. Proportional processing;

8.3 After the completion of step 8.2, based on the parameter variable category and the standardized data, establish a T 2 control chart model of multivariate statistics to visually monitor the abnormal conditions of the selected process; if the T 2 control chart detects abnormal production fluctuations If it is out of control, an alarm is issued and uploaded to the monitoring display and data reporting module, otherwise no response is made;

Intelligent analysis of the anomaly identification of the entire process of the flexible circuit board manufacturing process, including the following steps:

8.4 According to the user's choice, read the training data or the data to be monitored for the abnormal process of the whole process of manufacturing the flexible circuit board; determine whether to perform preprocessing according to the data type included in the data set, and if it is the type of the measurement value, concentrate the parameter data All data is standardized and preprocessed, and if it is a count type, the parameter data is scaled;

8.5 After the completion of step 8.4, the neural network method extracts data features;

8.6 After step 8.5 is completed, if the user selects the training model, the support vector machine model is trained with the normal and abnormal feature data, and the Gaussian kernel function is used and the relevant parameters are determined by the grid method, thereby completing the establishment of the support vector machine model; otherwise, , according to step 8.7 for intelligent analysis of data;

8.7 After the completion of step 8.5, use the support vector machine model to monitor the batch data of the machining process; if it detects that the abnormal production fluctuation is out of control, an alarm is issued and uploaded to the monitoring display and data report module, otherwise no response is made.
The method according to claim 8, further comprising: identifying an abnormal source that is monitored by the flexible circuit board manufacturing process, that is, an abnormal positioning, and realizing the positioning of the abnormal process and the abnormal variable in the whole process, specifically including Next steps:

9.1 According to the user's choice, read the training data or the data to be monitored for the abnormal positioning of the flexible circuit board manufacturing process; query and read the key parameter data of the flexible circuit board manufacturing process from the database module to form the batch data; Whether the batch data is preprocessed. If the metering value type requires preprocessing, all the data in the parameter data set is standardized and preprocessed, and if it is a counting type, it is not processed;

9.2 After the completion of step 9.1, if the user selects the training model, the fusion genetic algorithm and the abnormal source recognition model of the deep learning neural network are established for the pre-processed feature data set; the genetic algorithm adopts the binary coding technique and takes the total error square function as The fitness function selects the structure and weight of the optimized deep learning neural network by selecting, crossing, and mutating the evolutionary operator; otherwise, intelligent analysis of the data is performed according to step 9.3;

9.3 After the completion of step 9.2, use the abnormal source identification model established in 9.2 to monitor the whole process of the flexible circuit board manufacturing process. If abnormal fluctuation occurs, the output of the abnormal source identification model can be located to locate the operation of the runaway abnormality and The result is sent to the monitor display and data report module, otherwise no response is made.
The method of claim 8 further comprising the step of evaluating the quality of the flexible circuit board using said microscope automatic data acquisition device, specifically comprising the steps of:

10.1 After placing the flexible circuit board on the electric precision stage and fixing it with the fixing device, the operator logs into the host computer system, turns on the light source, and switches to the digital camera drawing mode through the shift lever after manual focusing;

10.2 After the completion of step 10.1, the servo motor drives the guide rail to move the electric precision stage to return the system to the detection origin;

10.3 After the completion of step 10.2, the standard file of the flexible circuit board to be tested is input by the operator or downloaded in the database. The standard file includes the Gerber file, the CAD file, and then the flexible circuit board standard file is parsed to obtain the standard figure and quality. Evaluate the required standard data;

10.4 After the completion of step 10.3, the digital camera captures the image, and the servo motor moves the X and Y axis motion guides. Make the system recognize and align the reference point;

10.5 After the completion of step 10.4, the motion control module controls the servo drive in the motor control box to drive the servo motor, and moves the electric precision stage by moving the X and Y axis motion guides, in order from left to right, from top to bottom. After the digital camera is enlarged by the microscope, the flexible circuit board to be tested is locally taken, and at the same time, the partial view of the flexible circuit board to be tested by the microscope vision control processing module is preprocessed, and then the feature points are matched based on the feature template. The splicing method performs image splicing, and completes the smoothing processing of the image, so that the drawing and splicing are repeated until the flexible circuit board is scanned, and finally the global map of the flexible circuit board to be tested is obtained;

10.6 After the completion of step 10.5, the image is binarized and the connected domain is searched, and the connected domain statistical centroid and area are used as the matching criterion to compare with the connected domain in the circuit diagram template to determine the mismatched area, that is, the defect area; The method detects line width and line spacing; uses Hough transform to identify the position of the hole, and obtains the aperture size according to the area information; compares with the line width, line spacing and aperture size in the standard figure and standard data to obtain quality evaluation information; The method identifies the open circuit, short circuit, and residual copper defects;

10.7 After the completion of step 10.6, the host computer system displays the full image of the defect area and the specific defect details, and provides an alarm according to the pre-recorded threshold information, so that the operator can timely handle the abnormal process;

10.8 After the completion of step 10.7, the host computer system stores the detection result in the local computer, and uploads related images, quality evaluation information, defect information, and defect data to the comprehensive database module for subsequent statistical processing.