A component mounting quality detection method for a chip mounter

Abstract

The application relates to the technical field of image data processing, and discloses a component mounting quality detection method for a chip mounter, which comprises the following steps: synchronously acquiring images before and after component mounting and a vacuum pressure signal of a mounting suction nozzle, compensating vibration by using an inherent substrate background area in the images, obtaining optical position deviation through image difference and centroid calculation, and finally logically associating the deviation with pressure signal change characteristics to classify a mounting event physical process. Through the cooperative diagnosis of optical detection and physical signals, the physical causes of the mounting event can be clearly distinguished, the inherent limitations of traditional detection methods caused by vibration interference and unclear fault attribution during high-speed operation are avoided, and the method has good adaptability to components with different appearances.

Description

TECHNICAL FIELD

[0001] The application relates to a component mounting quality detection method for a chip mounter and belongs to the technical field of image data processing. BACKGROUND

[0002] In surface mounting technology, the position of a component after mounting is detected on line by using machine vision, which is a general technical means to ensure the reliability of a circuit board, and the common working mode is to identify and measure the image of the component or its target pad, and the difference between the obtained geometric position and the theoretical value is calculated to evaluate the quality.

[0003] However, under the large-scale and low-cost production demand of the current electronic manufacturing industry, especially when facing a large number of, multiple sources and appearance batch differences of general passive components, the above detection method relies on the identification of the characteristics of the component body, which must deal with the interference caused by the color, silk screen and reflection of the component, thereby increasing the complexity and cost of the detection algorithm and the supporting illumination system, and when higher detection accuracy is pursued, the increase in cost is particularly prominent, which forms a continuous contradiction with the low-cost application scene of general components.

[0004] Specifically, the existing technical route faces a dilemma in actual application, that is, if the reliability of detection is to be ensured, the hardware and software costs that are not commensurate with the economic value of the detection object need to be invested, otherwise a higher misjudgment rate caused by material differences or mechanical micro-vibration of the equipment under high-speed running state needs to be tolerated, and pure optical image analysis often presents similar detection results for two kinds of faults with completely different physical causes, that is, a large deviation of the component position and complete absence of the component caused by failure of the suction nozzle, and cannot provide attribution judgment for rapid troubleshooting of the production process. Therefore, how to establish a low-cost detection method that can eliminate the influence of the appearance variation of the component and the vibration of the equipment in the detection process, and the detection result can clearly diagnose the physical process of the mounting event, has become a technical problem to be solved by the application. SUMMARY

[0005] The application provides a component mounting quality detection method for a chip mounter, and the main purpose is to solve the problem of how to establish a low-cost detection method that can eliminate the influence of the appearance variation of the component and the vibration of the equipment in the detection process, and the detection result can clearly diagnose the physical process of the mounting event.

[0006] To achieve the above object, the application provides a component mounting quality detection method for a chip mounter, which establishes a diagnosis logic for arbitrating optical detection results by physical process signals to classify component mounting events under high-speed operation conditions, and the method comprises the following steps: before a component is mounted on a target pad of a printed circuit board, a first image containing the target pad and a background area of the printed circuit board substrate adjacent to the target pad is synchronously captured by an imaging device as a dynamic real-time reference image; after the component is mounted, a second image containing the component, the target pad and the background area of the substrate is captured by the imaging device at the same position as that of capturing the first image, and a vacuum pressure signal of a mounting suction nozzle is synchronously monitored during the component mounting process to obtain a pressure time-domain curve; based on the background area of the substrate in the first image and the second image, sub-pixel level displacement introduced by equipment vibration is calculated and compensated to generate an aligned second image; the first image and the aligned second image are differentially processed to generate a residual image; the centroid position of a highlight area in the residual image is calculated to determine the planar mounting position deviation of the component, and the mounting event physical process of the component is classified according to the numerical value of the planar mounting position deviation and the characteristics of the pressure time-domain curve.

[0007] Preferably, the step of capturing the first image is performed at the moment when the mounting head carrying the component is lowered to a predetermined imaging height, and the predetermined imaging height makes the imaging device in the same focal plane as the final mounting position without contacting the printed circuit board.

[0008] Preferably, the method further comprises the following steps before the differential processing: before performing the mounting task, an ideal pad mask limiting the geometric shape of the target pad is generated according to pre-stored computer-aided design data corresponding to the target pad, the first image is logically ANDed with the ideal pad mask to avoid any image information in the first image located outside the geometric shape of the target pad, thereby generating a purified reference image, and the subsequent differential processing is performed using the purified reference image and the aligned second image, and the first image is logically ANDed with the inverse of the ideal pad mask to detect whether there is foreign matter outside the target pad area, and a pre-warning signal is output when foreign matter is detected.

[0009] Preferably, the step of classifying the mounting event physical process of the component specifically comprises the following steps: if the planar mounting position deviation is greater than a predetermined deviation threshold, and the pressure change rate of the pressure time-domain curve exceeds a predetermined transition slope threshold at the mounting moment, the category of the event is determined as high-deviation mounting; if the residual image shows that the pixel value of the entire target pad area is higher than a brightness threshold, and the pressure value of the pressure time-domain curve still maintains within the adsorption pressure range after the mounting moment, the category of the event is determined as component loss or empty mounting.

[0010] Preferably, the method further comprises a step of diagnosing coplanarity defects of the component, which is performed after the residual image is generated and comprises: calculating the intensity standard deviation of all pixels in the residual image whose pixel values are higher than a background noise threshold When the intensity standard deviation is higher than a predetermined coplanarity defect threshold, it is determined that the component has coplanarity defects.

[0011] Preferably, the method is also capable of handling optically transparent components and comprises: calculating the geometric residual energy and the texture residual energy of the residual image; when the geometric residual energy is lower than a predetermined geometric threshold and the texture residual energy is higher than a predetermined texture threshold, it is determined that the current component is an optically transparent component and the method switches to perform the following step to evaluate the mounting quality: comparing the texture residual energy of the residual image with a reference texture energy set for the type of optically transparent component, and evaluating the mounting quality according to the difference between the two.

[0012] Preferably, the method further comprises a step of diagnosing the upstream solder paste printing process: at the initial stage of a production batch, the average pixel intensity of the residual images generated by the components whose planar mounting position deviations are within the allowed tolerance range is recorded, and a residual intensity mean value and a residual intensity standard deviation are statistically calculated, which together constitute the residual intensity historical fingerprint of the batch, for each residual image generated in the subsequent production, its average pixel intensity is compared with the residual intensity historical fingerprint, and when the average pixel intensity deviates from the residual intensity mean value by more than a predetermined multiple of the residual intensity standard deviation, it is determined that there is an abnormality in the solder paste printing process.

[0013] Preferably, the step of calculating and compensating for the sub-pixel level displacement between images introduced by equipment vibration specifically comprises: extracting the background image blocks corresponding to the substrate background area from the first image and the second image respectively, calculating the sub-pixel level displacement vector between the two background image blocks using a phase correlation algorithm, and digitally shifting the second image according to the displacement vector to align the substrate background areas of the two images before performing the difference processing.

[0014] Preferably, the step of determining that there is an abnormality in the solder paste printing process further comprises determining the type of abnormality: if the average pixel intensity is lower than the residual intensity mean value and its deviation is more than a predetermined multiple of the residual intensity standard deviation, it is determined that there is too much solder paste; if the average pixel intensity is higher than the residual intensity mean value and its deviation is more than a predetermined multiple of the residual intensity standard deviation, it is determined that there is too little solder paste or missing printing.

[0015] Preferably, the method further comprises the step of performing systematic process control: continuously recording and storing a series of planar placement position deviation data determined by component placement, analyzing the statistical trend of the deviation data to determine whether there is a one-way deviation trend representing mechanical drift of the placement head or a periodic deviation pattern representing abnormal state of the feeder, and outputting instructions for predictive maintenance or equipment calibration according to the determined deviation trend or deviation pattern.

[0016] Compared with the prior art, the present application has the following beneficial effects: 1. The present application establishes a detection method of logically associating and judging optical imaging detection and physical process signals. By synchronously monitoring the vacuum pressure signal of the placement suction nozzle while obtaining the images before and after placement, the detection system can not only judge the position deviation of the component in the two-dimensional plane, but also classify the physical process of the placement event according to the change characteristics of the pressure signal, so as to clearly distinguish between high deviation placement and component loss or empty placement, which are two similar optical results caused by completely different physical reasons. This cooperative diagnosis mechanism of optical information and physical information avoids the blind area of traditional pure optical detection in fault attribution, makes the positioning of equipment problems more accurate, and provides a direct basis for subsequent process adjustment or equipment maintenance.

[0017] 2. The present application captures the target pad and its adjacent substrate background area as part of the printed circuit board body before and after placement, and uses the background area as a dynamic reference system to calculate and compensate for mechanical vibration caused by high-speed operation of the equipment by using the relative displacement of the background area between the two times of imaging. This mechanism makes the subsequent image difference processing immune to vibration artifacts, so as to obtain residual information that truly reflects the change of component position. This online dynamic calibration method based on image information itself solves the problem of decreased detection reliability in high-speed placement scenarios without adding any additional hardware, and makes high-precision process online monitoring possible without sacrificing equipment running efficiency.

[0018] 3、The detection logic adopted by the present application is centered on extracting residual information by comparing images of the same position before and after mounting, to verify whether the shielding relationship between the component and the target pad is correct, rather than recognizing the optical or geometric features of the component itself. This relationship verification method completely decouples the detection results from the appearance differences such as color, character, reflective characteristics, and dimensional tolerance between different batches of the component body, so the method is suitable for processing low-cost, mass-produced universal materials from different suppliers with low appearance consistency, avoiding the tedious work of frequently adjusting algorithm parameters or rebuilding templates due to differences in incoming materials in the prior art, improving the overall operation efficiency and flexible production capacity of the production line, and the present application has the ability to extend the diagnosis and control to the upstream and downstream of the process chain. On the one hand, by introducing computer-aided design data to generate ideal pad masks, not only can the reference image be purified to deal with pre-existing pad defects, but also can detect whether there are foreign objects on the substrate before mounting, realizing the functional expansion from post-quality detection to pre-substrate inspection. On the other hand, by continuously recording and statistically analyzing a series of mounting deviation data, it can be judged whether the equipment has systematic mechanical drift or periodic supply instability, and then output instructions for predictive maintenance, so that quality control is improved from post-single-point correction to pre-process prevention and system-level health monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The element mounting quality collaborative diagnosis method flowchart of the present application.

[0020] Figure 2 The diagnostic classification logic state machine diagram of a single mounting event of the present application.

[0021] Figure 3 The core function and extended use case model diagram of the detection system of the present application.

[0022] Figure 4 The distributed deployment and communication architecture diagram of the mounting quality detection system of the present application. DETAILED DESCRIPTION

[0023] To make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail below in conjunction with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and do not limit the protection scope of the present application.

[0024] The application discloses a component mounting quality detection method for a chip mounter, which is based on logical correlation of optical imaging detection and physical process signals, and comprises the following steps: synchronously acquiring image information and vacuum pressure signals of a mounting suction nozzle before and after component mounting, calculating and compensating mechanical vibration introduced by equipment operation by using a substrate background area in the image, obtaining optical position deviation through image difference and centroid calculation, and finally logically correlating the magnitude of the deviation with the change characteristics of the pressure signals to classify and diagnose the physical causes of the mounting event; in surface mounting applications, the detection method relying on a static template is difficult to cope with uneven lighting, circuit board bending and appearance differences of incoming components in actual working conditions, to cope with this challenge, the method is configured to capture a first image containing a target pad and a neighboring substrate background area by using an imaging device before the component is mounted on the target pad of the printed circuit board, the execution timing of this step is determined by setting the mounting head to carry the component to a predetermined imaging height, the height makes the focal plane of the imaging device objective lens coincide with the plane on which the component is finally mounted on the pad, and at the same time maintains a safe distance from the circuit board, so that the first image obtained is a reference image containing all instantaneous environment and workpiece disturbance factors, which provides a reference system for subsequent differential processing to effectively suppress common-mode noise; in other words, under the high-speed running state of the chip mounter, the mechanical micro-vibration of the equipment itself will cause relative displacement between images within a small time interval between two imaging, thereby generating a false residual signal, in view of this, the method adopts an online dynamic alignment mechanism based on image background, after the component is mounted, the imaging device captures a second image containing the component, the target pad and the substrate background area at the same position as capturing the first image, and performs a deterministic sub-pixel level displacement compensation step, the procedure is as follows: extracting background image blocks corresponding to the substrate background area from the first image and the second image respectively, calculating a sub-pixel level translation vector between the two background image blocks by using a phase correlation algorithm , and digitally translating the second image according to the displacement vector to generate an aligned second image, this process compensates for the displacement introduced by vibration, so that the subsequent differential processing mainly reflects the image change introduced by component mounting.

[0025] To reduce the dependence on the identification of the element body features, the method adopts image difference and centroid positioning to determine the deviation, the process is: performing pixel-level difference processing on the first image and the aligned second image to generate a residual image, as an embodiment, to avoid the interference of the possible pre-stored defects on the target pad on the reference image, the purification and pre-checking steps of the reference image can also be performed, this step generates an ideal pad mask that limits the geometry of the target pad according to the pre-stored computer-aided design data before performing the mounting task, and performs logical AND operation on the first image and the ideal pad mask to generate a purified reference image for subsequent difference, and also performs logical operation on the first image and the inverse of the mask to detect whether there is foreign matter outside the pad area and output a warning, after generating the residual image, the system calculates the centroid position of the highlight area in the residual image, and determines the planar mounting position deviation of the element based on the deviation of the centroid position from the pre-set ideal center position, this verification method reduces the sensitivity of the detection result to the appearance differences such as color, characters and reflective characteristics of the element body; simple optical analysis may present similar detection results for faults with different physical causes, to realize fault attribution, the method establishes a set of diagnostic logic for arbitrating optical detection results by physical process signals, the logic synchronously monitors the vacuum pressure signal of the mounting suction nozzle to obtain the pressure time domain curve while performing optical imaging , the system classifies the physical process of the mounting event according to the value of the determined planar mounting position deviation and the characteristics of the pressure time domain curve, the determination procedure is as follows: if the planar mounting position deviation is greater than the predetermined offset threshold, and the pressure change rate of the pressure time domain curve at the mounting time exceeds the predetermined transition slope threshold, the event category is determined as high offset mounting, in contrast, if the residual image shows that the pixel value of the entire target pad area is higher than the brightness threshold, and the pressure value of the pressure time domain curve still maintains within the adsorption pressure range after the mounting time, the event category is determined as element loss or empty mounting.

[0026] The procedure of calculating and compensating sub-pixel displacement and acquiring time-domain curve features of pressure is further determined, specifically, before performing the phase correlation algorithm, an image block of no less than four times the projected area of the component to be mounted and not containing the substrate background area of the pad or high-contrast circuit is cut from the first and second images, and a 5x5 Gaussian low-pass filter is applied to the image block to suppress image noise, then a Hanning window function is applied to reduce spectral leakage, then in the calculated phase correlation matrix, the maximum value point and its 3x3 neighborhood are fitted by a two-dimensional Gaussian surface to obtain a sub-pixel precision translation vector; at the same time, the original pressure signal collected by the vacuum pressure sensor at a frequency of no less than 1 kHz is smoothed by a Savgol filter with a window length of 9 and a polynomial order of 3, which is a digital filter that directly performs least squares fitting in the time domain to smooth the data, to obtain a smoothed pressure curve , and the time when the mounting head Z-axis servo motor controller sends a signal that the preset mounting height is reached is defined as the mounting time , and finally the pressure rate of change at the mounting time is determined by the following formula, which is calculated by calculating the average slope across an asymmetric time window to obtain a stable physical quantity that is not sensitive to noise.

[0027] In addition, the method can also be applied to process diagnosis of extensibility, on the one hand, the method can be used to diagnose the coplanarity defects of the component, the steps of which are to calculate the intensity standard deviation of all pixels in the residual error image whose pixel values are higher than the background noise threshold after generating the residual error image When the intensity standard deviation exceeds the predetermined coplanarity defect threshold, it is determined that the component has coplanarity defects, which is based on the fact that the edge of the tilted component will produce a gradual change in grayscale in the residual error image, resulting in an increase in the intensity standard deviation, on the other hand, the method can be used for online monitoring of the upstream solder paste printing process, the procedure of which is to record the average pixel intensity of the residual error image generated by the components with position deviation within the allowable range in the initial stage of a production batch, and statistically calculate the residual error intensity mean and standard deviation , which together constitute the residual error intensity historical fingerprint of the batch, for the residual error image in subsequent production, compare its average pixel intensity with the historical fingerprint, when the deviation from the mean a predetermined multiple of the standard deviation of the historical stable process, it is determined that there is an abnormality in the solder paste printing process. Finally, through continuous recording and statistical trend analysis of a series of mounting deviation data, it can be determined whether the equipment has a one-way mechanical drift or a periodic supply instability. The results of the statistical trend analysis can be used to output predictive maintenance instructions to achieve early intervention on systematic deviation. In addition, the procedures for processing optical penetration elements and performing systematic process control are further specified. The texture residual energy is calculated based on the gray level co-occurrence matrix of the residual image, which is a statistical matrix that describes the frequency of occurrence of gray levels in a specific spatial relationship in the image. The angular second moment is calculated as a quantitative indicator of image texture uniformity, and the formula is wherein is the normalized element value in the gray level co-occurrence matrix. The statistical trend analysis of the deviation data is achieved by maintaining an exponentially weighted moving average control chart for each mounting head X and Y axis deviation sequence. This is a trend monitoring method that gives more weight to recent data. The sequence value is calculated by the recursive formula wherein is the current deviation, is the previous cycle sequence value, and the weight factor is set to 0.2. When the sequence value continuously exceeds the three standard deviation control limit determined by the historical stable process standard deviation for five times, i.e. , it is determined that there is a systematic deviation trend representing an abnormal equipment state.

[0028] Example 1: In a continuously running surface mount production line for consumer electronics, the placement machine is operating at its upper limit of rated speed, with tens of thousands of generic capacitors and resistors being placed by the machine, which are supplied by different vendors, and have batch-to-batch variations in color, surface gloss, and dimensional tolerance, meanwhile, the high speed operation of the machine induces persistent high frequency mechanical micro-vibration on the machine, this working condition poses double challenges to a vision inspection system based on static template feature matching, i.e. the system not only has to deal with the difficulty of recognizing the inconsistent appearance of the components, but also has to handle the image capture position drift caused by vibration, which further leads to the increase of detection false positive rate, forcing the production process to make a trade-off between operation efficiency and quality control; in this scenario, for the placement of each component, at the moment when the placement head carrying the component is about to touch the printed circuit board, the imaging device captures a first image, which records the image of the target pad under the current lighting and circuit board state, and synchronously captures the background area of the substrate adjacent to it as the subsequent dynamic reference frame, after the component is placed, the suction nozzle is lifted, and the imaging device captures a second image at the same position, then, the online dynamic alignment mechanism based on image background is triggered, the system calculates the sub-pixel level translation vector introduced by mechanical vibration by comparing the background area of the substrate in the first image and the second image , and uses the vector to digitally shift the second image to generate an aligned second image The capture of the first image provides a dynamic time and space reference for the vibration compensation step, while the vibration compensation step provides a calibrated and aligned data input for the subsequent differential processing, the combination of these mechanisms enables the image comparison baseline to remain stable under high speed motion working conditions.

[0029] Subsequently, the system performs pixel-level differential processing to generate a residual image In this residual image, the difference information about the appearance of the component body is cancelled out as the common background in the two imaging processes, leaving only the highlight residual representing the unshielded pad area or the component overflow area caused by placement offset, which converts a problem of complex feature recognition on a variable object into a problem of locating the geometric centroid of a signal containing only differential information, on this basis, the system further calculates the planar placement position deviation, and synchronously acquires the pressure time domain curve The features are logically associated. When the end of a batch of feeders is empty, causing the suction nozzle to be empty, the optical part detects a large residual covering the entire pad, and the pressure signal part detects that the pressure before and after mounting does not jump. The system classifies the physical process of the event as component loss or empty mounting, and outputs the corresponding replenishment instruction. Finally, the production line can continue to run at its rated speed, the number of detection-related downtime events caused by component appearance differences and equipment vibration is reduced, and the system can distinguish between defects caused by position offset and defects caused by feed interruption, providing executable maintenance guidance for operators, and the basis for quality control is changed from recognition of the ontology of variable objects to verification of the correctness of the physical relationship between mounting behavior and target position, and synchronous monitoring of the physical state of the actuator.

[0030] Example 2: To quantitatively evaluate the detection stability of the method of the present application in dealing with component appearance differences and equipment mechanical vibration interference, the present comparative test is set up, the purpose of which is to compare the performance of the method of the present application with a conventional detection method based on recognition of the profile features of the component ontology under controlled conditions, to verify the effectiveness of the present application scheme with data; the test is carried out on a standard placement machine, which has a programmable vibration table integrated on the PCB bearing platform to apply controlled mechanical micro-vibration, the test uses two visual detection methods for data acquisition and analysis, method A is the present application, based on dynamic real-time reference image and differential residual centroid calculation, this method includes an online dynamic alignment mechanism based on image background, method B is the control group, which uses a conventional detection method based on static CAD template for component edge and corner feature matching, the test objects are three batches of passive components, batch A, 0402 size dark matte resistors, batch B, 0402 size gray highlight resistors with characters on the surface, and batch C, 0603 size light brown ceramic capacitors, a key parameter in the test, vibration frequency, is set to simulate the typical vibration state of the placement machine during high-speed operation, the main factors affecting the parameter value are the motion harmonics of the device main shaft and cross beam, the technical trade-off point is that the set frequency should be sufficient to interfere with the conventional detection method B, but it cannot exceed the realistic range of industrial scenarios, based on this decision logic, the vibration frequency is set to 100 Hz, and the amplitude is calibrated to produce a peak displacement of 25 microns in the imaging field of view.

[0031] The test procedure is as follows: for each component in batches A, B and C, 100 mounting operations are performed under two conditions of no vibration and 100Hz vibration, respectively, in each mounting, a random plane offset in the range of 0 to 60 microns is applied to the mounting position, and a high-precision laser displacement sensor is used to record the actual offset of this mounting, methods A and B simultaneously analyze the image after each mounting is completed, and calculate the respective plane mounting position deviation measurement values, if the difference between the measurement value and the actual offset exceeds 15 microns, it is recorded as one misjudgment; if the average measurement error of each group is plotted, it can be observed that under all test conditions, the error value corresponding to method A maintains at a low level of less than 5 microns, while the error value of method B increases sharply under certain conditions, specifically, when detecting component B with high surface light under the condition of no vibration, the average measurement error of method B is 18.7 microns, and the misjudgment rate is 21%, while the error of method A is only 3.3 microns, and the misjudgment rate is 0, when 100Hz vibration is introduced, the average measurement error of method B when dealing with component B further increases to 35.2 microns, and the misjudgment rate is as high as 58%, while the average measurement error of method A under this combined condition still maintains at 4.8 microns, and the misjudgment rate is only 2%, for components A and C with regular appearance, under 100Hz vibration, the average measurement error of method B also reaches 24.1 microns and 26.5 microns respectively, while method A maintains below 4.4 microns.

[0032] The mechanism of the above data difference is that the detection logic of method B relies on the stable extraction of component edge features, so when facing the surface high light and character interference of batch B or the dynamic blur introduced by vibration, the feature extraction process will produce a large deviation, while method A performs pixel-level difference processing on the images before and after mounting, and its detection result is largely independent of the optical properties of the component itself, at the same time, its built-in online dynamic alignment mechanism uses the substrate background in the image as a reference, and the vibration displacement is compensated before performing difference processing, thereby suppressing the influence of vibration on the measurement result.

[0033] Embodiment 3: This embodiment combines Figures 1 to 4 to describe a component mounting quality detection method for a chip mounter, like Figure 1As shown, the method starts with the first and second images before and after the online capture element mounting, and synchronously monitors the vacuum pressure signal of the mounting nozzle, the core processing path of the method includes: using the substrate background area in the front and rear images, performing compensation for the displacement introduced by vibration to avoid mechanical vibration interference, then performing image difference processing to generate residual images reflecting the position change, and determining the planar mounting deviation of the element by calculating the mass center position, the optical position deviation information and the synchronous acquired pressure signal change characteristics are input into the cooperative diagnosis and classification link, which finally classifies the physical process of the mounting event as element loss or empty mounting, qualified mounting or high deviation mounting according to the logical association rules. In addition, the method also includes two extended diagnostic and control loops, one is to diagnose the upstream solder paste printing process and output a warning signal when abnormal, and the other is to perform systematic process control by statistical trend analysis of historical deviation data, and output predictive maintenance or equipment calibration instructions accordingly.

[0034] As shown in Figure 2 The flow is triggered by the event of the mounting head descending to the predetermined imaging height, so that the system enters the waiting mounting state, and then the first image capturing event is performed, so that the system is converted to the dynamic reference captured state. When the mounting and synchronous monitoring event occurs, the system enters the core analysis state, and a series of actions are sequentially performed in this state, including capturing the second image, compensating for the vibration displacement, generating the residual image, calculating the position deviation, and acquiring the pressure curve feature. Finally, the system determines the final state according to a set of mutually exclusive logical conditions and exits, if the conditions residual image highlight, pressure value maintained in the adsorption range are met, the element loss or empty mounting final state is entered, if the conditions deviation> offset threshold, pressure change rate> slope threshold are met, the high deviation mounting final state is entered, and if the conditions deviation< threshold, pressure normal are met, the qualified mounting final state is entered.

[0035] As shown in Figure 3 The pick-and-place machine as the main executor, and the operator / maintenance system as the external participant, the core function execution quality detection is shown as a collection of a series of interrelated sub-function use cases, these use cases include synchronous acquisition of images and pressure signals as the basis for input and compensation mechanism and compensation for displacement introduced by vibration, logical association diagnosis and classification as the core diagnostic logic, and diagnosis of upstream solder paste printing process, systematic process control, diagnosis of element coplanarity defects and processing of optically transparent elements as the extension ability of the method.

[0036] As shown in Figure 4As shown, the core of the whole system is the placement machine as the device node, whose data processing is carried out around an industrial control computer as the execution node, which is responsible for receiving and processing data from two types of physical devices: one is the imaging device that transmits image information via the image data bus, and the other is the vacuum pressure sensor that transmits pressure signals via the sensor bus. Inside the computer, three core software components are deployed to perform the detection task: the core analysis engine is responsible for basic data processing, the collaborative diagnosis and classification module performs the core logical judgment, and the extended diagnosis module handles advanced functions such as process diagnosis. In order to achieve systematic data management and remote monitoring, the collaborative diagnosis and classification module communicates with a central server through the standard TCP / IP protocol. The database in the server is specifically used to persistently store historical fingerprint and deviation data for long-term analysis. Finally, the engineer workstation serves as the terminal for human-computer interaction. The monitoring dashboard component running on it also acquires data from the server via the TCP / IP protocol, providing operators with an interface for monitoring and managing the entire detection system.

[0037] Example 4: Before the start of a new printed circuit board assembly task, a set of systematic offline calibration procedures is performed for the new model of circuit board and components used in the task to determine a series of key threshold values in the detection method of the present application. The calibration procedures are performed on a placement machine that has already installed a detection system, and its initial state is to prepare five printed circuit board samples to be tested from the same production batch and a sufficient amount of components to be mounted. The first step of the procedure is to calibrate the planar mounting position deviation threshold value and the pressure transition slope threshold value. The operator controls the placement machine to perform twenty normal mountings and twenty mountings with a preset large deviation on the sample boards in low-speed mode, where the large deviation is set to be greater than fifty percent of the component size. The system synchronously records the planar mounting position deviation value and the pressure time-domain curve generated by each mounting event , and calculates the pressure change rate of the pressure curve at the mounting time. Thus, two data sets of the normal mounting group and the high deviation mounting group are obtained. The distribution of the planar mounting position deviation values in the two data sets is analyzed, and a value between the upper limit of the deviation distribution of the normal mounting group and the lower limit of the deviation distribution of the high deviation mounting group is selected as the planar mounting position deviation threshold value. Similarly, the distribution of the pressure change rate in the two data sets is analyzed, and a slope value that can distinguish whether the component has successfully detached from the suction nozzle is selected as the pressure transition slope threshold value.

[0038] The second step of the procedure is to calibrate the brightness threshold for missing components. The operator controls the placement machine to perform ten empty placements without picking up a component. The system analyzes the residual image generated by each empty placement, which is a high-brightness image of the entire target pad area, records the average pixel intensity, and multiplies the minimum average pixel intensity value obtained from the ten measurements by a safety factor of 0.9. The result is set as the brightness threshold. The third step of the procedure is to calibrate the coplanarity defect threshold. The operator places a sheet of known thickness on one side of a sample board to introduce a certain tilt angle during component placement. The placement is repeated ten times, and the standard deviation of the pixel intensity of the residual image generated by each placement is calculated The minimum standard deviation value obtained from the ten measurements is set as the coplanarity defect threshold. In subsequent production, any placement event with a standard deviation of intensity exceeding the threshold will be identified as having a coplanarity defect. After completing the calibration of all the thresholds mentioned above, a trial production run of one hundred placements is started. During the trial production run, the system uses the calibrated thresholds to monitor the quality of each placement and only records the average pixel intensity of the residual image for those placements that are judged to be of good quality. After the trial production run is completed, the system performs statistical analysis on the average pixel intensity data of all the qualified samples recorded, calculates the arithmetic mean as the residual intensity mean value of the batch and calculates the standard deviation These two values and together constitute the residual intensity historical fingerprint for diagnosing the upstream solder paste printing process.

[0039] In the processing of a production task involving a type of optical penetrating component, specifically glass-encapsulated diodes, it was observed that the low masking property of this type of component resulted in excessively low geometric residual energy, making it impossible to obtain effective positional deviations based on the calculation of the geometric residual centroid. Therefore, before entering formal production, a baseline texture energy calibration procedure was performed on this type of component. In the calibration procedure, ten samples of this type of optical penetrating component were taken, and each sample was placed at the theoretical center position of the target pad using a high-precision coordinate measuring device. For each of the ten placements, the detection system captured the first and second images before and after placement, respectively, and generated a residual image. Subsequently, the system calculated and recorded the texture residual energy of the residual image corresponding to each of the ten placements, and defined the arithmetic mean of the ten measurement results as the baseline texture energy of this type of optical penetrating component. This baseline texture energy value was stored in the material database of the system and associated with the material number of the component.

[0040] In the subsequent online automatic production process, when the system identifies that the currently mounted component is an optical penetrating component of this model, the calibrated reference texture energy value is called. After generating the residual image of this mounting, the system does not calculate the centroid position, but calculates the texture residual energy, and compares the measured energy value with the called reference texture energy value. If the difference between the two exceeds a tolerance range set based on the statistical distribution of the aforementioned calibration experiment data, the system determines that there is a quality defect in this mounting. This calibration and detection process enables the application of the method of the present application to the quality detection of such optical penetrating components.

[0041] In the initial stage of the system entering continuous production operation, in order to establish a baseline reference model capable of representing the normal mechanical state of the equipment for health state monitoring in the subsequent long-time operation process, the system performs the following procedures: for each mounting head, a sliding window containing the deviation data of the last one hundred mountings is dynamically maintained, and the arithmetic mean values of the deviation data in the X and Y axis directions within the window are continuously calculated. The core parameter of this model, i.e. the maintenance warning threshold, is set to fifty percent of the mounting tolerance. When the absolute value of the sliding average of the deviation data of any mounting head exceeds the maintenance warning threshold for ten consecutive calculation results, the baseline reference model determines that the equipment state has deviated from the normal baseline, and triggers a predictive maintenance instruction to record the drift state and data of the mounting head for subsequent calibration.

[0042] In the continuous operation of the system, its health state monitoring logic also includes an immediate intervention procedure for defects introduced by upstream processes. When the system performs the substrate quality pre-checking step before component mounting, a non-zero result is obtained by performing logical AND operation between the dynamic real-time reference image and the inverse of the ideal pad mask. This state indicates that there is a pre-existing foreign matter outside the target pad area. The system immediately suspends the mounting action on the target pad, and stores the unique identification code of the printed circuit board and the image coordinates of the foreign matter into the quality traceability log. Subsequently, the chip mounter skips the mounting task at this point and continues to perform mounting operations at other positions. The combination of this baseline model and the abnormal processing procedure enables the system not only to judge the quality of single-point mounting, but also to evaluate the long-term trend of its own running state based on historical data, and to intervene immediately when defects are introduced by upstream processes.

[0043] It is obvious to those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0044] Finally, it should be noted that the above examples are merely intended to illustrate the technical solutions of the present application and not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced without departing from the spirit and scope of the present application.

Claims

1. A component mounting quality detection method for a chip mounter, characterized by, The method establishes a diagnosis logic arbitrated by physical process signals to optical detection results, to classify component mounting events under high-speed running conditions, the method comprises: Before mounting a component onto a target pad on a printed circuit board, a first image containing the target pad and a substrate background area adjacent to the target pad is synchronously captured by an imaging device as a dynamic real-time reference image; After the component is mounted, a second image containing the component, the target pad and the substrate background area is captured by the imaging device at the same position as the first image, and a vacuum pressure signal of a mounting suction nozzle is synchronously monitored during the component mounting process to obtain a pressure time-domain curve, based on the substrate background area in the first image and the second image, a sub-pixel level displacement between the images introduced by device vibration is calculated and compensated, and an aligned second image is generated, the first image and the aligned second image are differentially processed to generate a residual image; The centroid position of the highlight area in the residual image is calculated to determine the planar mounting position deviation of the component, and the physical process of the component mounting event is classified according to the numerical value of the planar mounting position deviation and the characteristics of the pressure time-domain curve; Meanwhile, the step of classifying the physical process of the component mounting event specifically comprises: if the planar mounting position deviation is greater than a predetermined offset threshold, and the pressure change rate of the pressure time-domain curve exceeds a predetermined transition slope threshold at the mounting time, the category of the event is determined as high offset mounting, if the residual image shows that the pixel value of the entire target pad area is higher than a brightness threshold, and the pressure value of the pressure time-domain curve still maintains within the adsorption pressure range after the mounting time, the category of the event is determined as component loss or empty mounting.

2. The component mounting quality detection method for a chip mounter according to claim 1, wherein The step of capturing the first image is executed at the moment when the mounting head carrying the component is lowered to a predetermined imaging height, which makes the imaging device in the same focal plane as the final mounting position without contacting the printed circuit board.

3. The component mounting quality detection method for a chip mounter according to claim 1, wherein Before differential processing, the method further comprises: before executing the mounting task, an ideal pad mask limiting the geometric shape of the target pad is generated according to the pre-stored computer-aided design data corresponding to the target pad, the first image is logically ANDed with the ideal pad mask to generate a purified reference image, and the subsequent differential processing is performed using the purified reference image and the aligned second image, and the inverse of the first image and the ideal pad mask is logically ANDed to detect whether there is foreign matter outside the target pad area, and a warning signal is output when foreign matter is detected.

4. The component mounting quality detection method for a chip mounter according to Claim 1, wherein The method also comprises a step of diagnosing a coplanarity defect of the element, which is performed after the generation of the residual image and comprises: calculating the intensity standard deviation of the pixels of the residual image whose value is higher than a background noise threshold When the intensity standard deviation exceeds a predetermined coplanarity defect threshold, it is then decided that the element has a coplanarity defect.

5. The component mounting quality detection method for a chip mounter according to Claim 1, wherein The method can also process optical penetrating components and comprises: calculating the geometric residual energy and the texture residual energy of the residual image; when the geometric residual energy is lower than a predetermined geometric threshold, and the texture residual energy is higher than a predetermined texture threshold, it is determined that the current component is an optical penetrating component, and the following steps are executed to evaluate the mounting quality: comparing the texture residual energy of the residual image with a reference texture energy set for the same type of optical penetrating component, and evaluating the mounting quality according to the difference between the two.

6. The component mounting quality detection method for a chip mounter according to Claim 1, wherein The method further comprises a step of diagnosing the upstream solder paste printing process: at an initial stage of a production batch, recording the average pixel intensity of the residual image generated by the components with planar placement position deviations within the allowed tolerance range, and statistically calculating a residual intensity mean value and a residual intensity standard deviation, which together constitute the residual intensity historical fingerprint of the batch, for each residual image generated in the subsequent production, comparing its average pixel intensity with the residual intensity historical fingerprint, and when the average pixel intensity deviates from the residual intensity mean value by more than a predetermined multiple of the residual intensity standard deviation, it is determined that there is an abnormality in the solder paste printing process.

7. The component mounting quality detection method for a chip mounter according to Claim 1, wherein The step of calculating and compensating for the sub-pixel level displacement between images introduced by equipment vibration specifically comprises: extracting the background image blocks corresponding to the substrate background area from the first image and the second image respectively, calculating the sub-pixel level displacement vector between the two background image blocks using a phase correlation algorithm, and digitally translating the second image according to the displacement vector to align the substrate background areas of the two images before performing difference processing.

8. The component mounting quality detection method for a chip mounter according to Claim 6, wherein The step of determining that there is an abnormality in the solder paste printing process further comprises determining the type of abnormality: if the average pixel intensity is lower than the residual intensity mean value and its deviation exceeds a predetermined multiple of the residual intensity standard deviation, it is determined that there is too much solder paste, and if the average pixel intensity is higher than the residual intensity mean value and its deviation exceeds a predetermined multiple of the residual intensity standard deviation, it is determined that there is too little solder paste or missing printing.

9. The component mounting quality detection method for a chip mounter according to Claim 1, wherein The method further comprises a step of performing systematic process control: continuously recording and storing a series of component placement determined planar placement position deviation data, analyzing the statistical trend of the deviation data to determine whether there is a one-way deviation trend representing mechanical drift of the placement head or a periodic deviation pattern representing abnormality of the feeder state, and outputting instructions for predictive maintenance or equipment calibration according to the determined deviation trend or deviation pattern.

Images