TWI760975B - SMT machine throwing analysis and abnormal notification system and method - Google Patents

Abstract

A placement machine throwing analysis and abnormal reporting system and method, which connect a communication interface to a placement machine to collect throwing data, and convert the data into serialized data by a data processing module and store it in a database, and can pass The data learning and analysis module independently checks whether the database has updated data, and then categorizes the abnormality of the thrown material according to the results of the learning and analysis of the serialized data, including the abnormal suction nozzle of the placement machine, the abnormality of the feeder, and the vacuum pressure value. In case of abnormality or abnormal incoming materials, the learning module can be notified by the abnormal feedback to self-learn and automatically adjust the threshold of the throwing rate according to the throwing data stored in the database. When checking that the throwing rate of the placement machine is not less than the feedback threshold , the abnormal classification results of the data learning analysis module can be immediately reported to the engineer for processing, so as to improve the throwing rate.

Description

SMT machine throwing analysis and abnormal notification system and method

The present invention provides a throwing analysis and anomaly reporting system and method, in particular to a throwing analysis and anomaly reporting system and method for a placement machine.

Press, in the process of manufacturing electronic products today, various electronic components must be assembled on the circuit board. These electronic components can usually be manufactured by surface mount technology (SMT) or dual in-line packaging (DIP), and in In the SMT production line, the metropolis uses a placement machine to place the leadless or short lead surface mount components (SMC/SMD for short) on the pads on the surface of the circuit board or other substrates at high speed and high precision, and then pass reflow soldering. Or dip soldering and other methods to be welded and assembled, which can improve the production efficiency of surface mount.

In the surface mount production line, the most concerned issue is how to effectively reduce the production cost and improve the production efficiency, which involves the problem of the throwing rate during the production of the placement machine. The level of the throwing rate will not only affect the production cost, but also It will seriously affect the production efficiency and quality of the surface mount, and the current technology to reduce the throwing rate of the placement machine is generally based on the experience of engineers, to define a threshold of the throwing rate. When the throwing rate of the placement machine is greater than At this threshold, the equipment representing the placement machine is abnormal, and the engineer needs to conduct an on-site inspection, and then after the actual test and replacement of the hardware verification, can it be confirmed whether the throwing rate of the placement machine can be reduced, but such a complicated test and The verification process, not only can not accurately adjust the threshold of its throw rate, coupled with different surfaces Assembled components or parts have different production conditions and characteristics. Relying on human experience and thinking analysis alone, it is not only impossible to quickly solve the problem of high throwing rate, but also requires a lot of man-hours to observe and analyze. , in addition to the difficulty of defining the threshold or standard deviation range of the throwing rate, it is also difficult to bring good improvement measures.

In addition, in the actual production process, there are many key factors that will affect the throwing rate of the placement machine, which can be mainly divided into problems such as nozzle deformation, poor feeder, abnormal vacuum pressure value, and poor incoming materials. These keys will each produce different abnormal details. For example, the feeder may accumulate dust, the gears may be old, damaged, etc., resulting in poor feeding or poor reclaiming and throwing materials, etc., thus affecting the throwing rate. Therefore, in In the traditional experience, in addition to labor hours and labor costs, engineers are more difficult to accurately classify abnormal problems through human experience, so as to achieve effective abnormal analysis. That is, the direction of research and improvement that those engaged in this industry are eager to study and improve.

Therefore, in order to improve and solve the above-mentioned problem of material throwing of the placement machine, the inventor collected relevant data through various evaluations and considerations, and continued to test and revise with years of experience accumulated in this industry. The material throwing data of the placement machine is independently studied and analyzed, the threshold value of the throwing rate is automatically adjusted, and the throwing analysis and abnormal notification system and method of the placement machine for abnormal classification and real-time notification of the throwing material are invented. Patent birth.

The main purpose of the present invention is that the communication interface of the placement machine throwing analysis and abnormal reporting system is connected to the placement machine to collect the data of the throwing data, and is converted into serialized data by the data processing module and stored in the database for data processing. Sharing, you can independently check whether the database has updated data through the data learning and analysis module, and then throw it according to the results of learning and analyzing the serialized data. The abnormal classification of materials, including the abnormal suction nozzle of the placement machine, the abnormality of the feeder, the abnormality of the vacuum pressure value or the abnormality of the incoming material, can be reported to the learning module through the abnormal feedback to learn and analyze independently according to the material throwing data in the database. Automatic feedback to adjust the threshold of the placement machine throwing rate with different materials. When checking that the throwing rate of the placement machine is not less than (ie greater than or equal to) the feedback threshold, the data learning and analysis module can be classified according to the abnormality The results are immediately reported to engineers to help engineers reduce the process of analyzing different material throwing anomalies and reduce the man-hours spent in manual analysis, more accurately define the threshold of the throwing rate, and classify different throwing exceptions, and then Improving the throwing rate of the placement machine can effectively improve the quality and yield of production, and can reduce costs.

100: SMT machine throwing analysis and abnormal notification system

110: Communication interface

120: Data processing module

130:Database

140: Data Learning Analysis Module

150: Abnormal feedback notification learning module

160:Monitoring Platform

200: Mounter

[FIG. 1] is a block diagram of a preferred embodiment of the present invention.

[Fig. 2] is a flow chart of the steps of a preferred embodiment of the present invention.

In order to achieve the above-mentioned purpose and effect, the technical means and structure adopted by the present invention, the preferred embodiments of the present invention are described in detail in the drawings and the structure and function are as follows, so as to be fully understood.

Please refer to Figures 1 to 2, which are respectively a block diagram and a step flow chart of a preferred embodiment of the present invention. It can be clearly seen from the figures that the chip placement machine throwing analysis and abnormal reporting system 100 of the present invention It includes a communication interface 110 which is connected to a placement machine 200 and captures its related raw material throwing data; a data processing module 120 is connected with the communication interface 110 for converting the material throwing data into a serialized data; a database 130 is connected with the communication interface 110 and the data processing module 120 to collect and store the throwing data and the serialized data; a data The data learning and analysis module 140 is connected with the database 130 to independently check whether the database 130 has updated data, obtain a learning and analysis result according to the serialized data, and then perform the The abnormal classification of the material thrown by the placement machine 200 includes, but is not limited to, four groups of abnormal suction nozzles, abnormal feeders, abnormal vacuum pressure values or abnormal incoming materials on the placement head of the placement machine 200; And an abnormal feedback notification learning module 150 is connected with the data learning and analysis module 140 and the database 130, and according to the throwing data stored in the database 130, self-learning and analysis and automatic feedback adjust the throwing rate of the placement machine 200 When checking that the throwing rate of the placement machine 200 is not less than (ie greater than or equal to) the feedback threshold, it will output a prompt message according to the result of the abnormal classification corresponding to the data learning and analysis module 140 and actively communicate with The engineer notified that the material throwing state during the production process of the placement machine 200 was immediately improved.

In this embodiment, when the data learning and analysis module 140 analyzes the abnormality of the suction nozzle in the material throwing abnormality classification of the placement machine 200, the material throwing data captured by the data learning and analysis module 140 includes the number of the material picked up by the suction nozzle (referred to as the material number for short). No.), nozzle identification code, material number of the thrown material (referred to as the throwing material number), the number of thrown materials and the total number of sucked materials, and after analyzing a specific nozzle identification code, the behavior of throwing materials is particularly prone to occur. , it can be determined that the nozzle may be damaged or blocked and other abnormal events may occur, causing it to fail to take materials normally.

When analyzing the abnormality of the feeder, the collected throwing data includes the feeder identification code, the position of the feeder, the number of the material thrown from the feeder, the number of throws thrown from the feeder, and The total number of feeds, and when it is analyzed that the feeder at a specific position is particularly prone to throwing behavior, it can be determined that the feeder is abnormal after accumulating certain data according to the combination of different thrown material numbers. The possible reasons for the abnormality include dust accumulation in the feeder, old gears, damage, etc., which cause the feeding to be unsmooth, thus affecting the throwing rate.

If the analysis of the vacuum pressure value is abnormal, the collected throwing data includes the material number sucked by the suction nozzle, the suction nozzle identification code and the vacuum pressure value, etc. When it is analyzed that the suction pressure is greater than the threshold value, it can be determined that the suction nozzle may be damaged, air leakage, or the vacuum press has problems and other abnormal events. Similarly, when the suction pressure is less than the threshold value, it can be determined that the quality of the incoming material is abnormal (such as uneven surface of the material, different quality and other quality differences will affect the suction pressure).

Furthermore, when analyzing the abnormality of incoming materials, the captured raw material throwing data includes the number of materials picked up by the nozzle, the number of materials thrown, the total number of materials thrown, the total number of materials picked up, and the number of materials extracted from all the above groups. Data, when a thrown material number often occurs in different nozzle or feeder positions, but no centralized material throwing phenomenon occurs, it can be judged as abnormal incoming material.

In addition, the placement machine throwing analysis and abnormal reporting system 100 further includes a monitoring platform 160. The monitoring platform 160 is preferably implemented as a display device, but is not limited to this, and may further include an alarm device. For example, a display device has an operating system or a user interface, and is connected to the database 130 and the data learning and analysis module 140 to display the data stored in the database 130 and the learning and analysis module 140 performs learning With the analysis result, it can also be further connected with the abnormal feedback notification learning module 150 to receive the prompt message output by the abnormal feedback notification learning module 150, and actively notify the engineer for processing by displaying or issuing an alarm, and then To achieve the function of improving the throwing rate in real time.

As shown in FIG. 2, the present invention further provides a method for analyzing and reporting abnormality of material thrown by a chip mounter, which is suitable for the system 100 for analyzing and reporting abnormality of material thrown by a chip mounter, so as to analyze the abnormal return of the material thrown by the chip mounter 200. For similar reasons, the placement machine throwing analysis and abnormal reporting system 100 includes the above-mentioned communication interface 110, data processing module 120, database 130, data learning and analysis The module 140 and the abnormal feedback notification learning module 150, wherein the placement machine throwing analysis and abnormal notification method includes the following implementation steps:

(S101) starts.

( S102 ) Connect the communication interface 110 to the placement machine 200 .

(S103) Collect the data of the throwing data.

(S104) The data processing module 120 performs data serialization conversion.

(S105) Store in the database 130.

(S106) The data learning analysis module 140 checks whether there is updated data? If yes, execute step ( S107 ), if no, execute step ( S103 ) repeatedly.

(S107) Data learning and analysis, and then synchronously execute steps (S108) and (S109).

(S108) The abnormality is classified, and then the step (S110) is executed.

(S109) Abnormal feedback notification and the learning module 150 feeds back and adjusts the threshold value of the material throwing rate of the placement machine 200, and then executes the step (S110).

(S110) Check whether the throwing rate is greater than the threshold value? If yes, execute step ( S111 ), if no, execute step ( S107 ) repeatedly.

(S111) A result of classifying the abnormality is actively notified.

(S112) ends.

It can be clearly seen from the figure and the above-mentioned implementation steps that the placement machine throwing analysis and abnormal reporting system 100 of the present invention is an automatic control system connected to the placement machine 200 through the communication interface 110 , and captures the detection including The values related to the operation of the suction nozzle, feeder, vacuum press, etc. of the patch head are collected to collect the data as the original throwing data, and the data processing module 120 can After the original throwing data is serialized and converted, it is then stored in the database 130. Then, the data learning and analysis module 140 automatically checks that the database 130 has updated data, and will perform the process of data learning and analysis. According to the analysis results, the abnormality of the material thrown by the placement machine 200 is classified, and at the same time, the abnormal feedback notification learning module 150 will learn and analyze independently according to the throwing data stored in the database 130 and automatically adjust the placement machine 200 according to the automatic feedback of different materials. Threshold value of the throwing rate. When checking that the throwing rate of the placement machine 200 is not less than the feedback threshold, the result of the abnormal classification of the analysis corresponding to the data learning analysis module 140 will be actively notified to the engineer for subsequent related processing. The process can instantly improve the throwing rate of the placement machine 200.

In detail, the above-mentioned data learning and analysis module 140 performs the process of data learning and analysis of the serialization of throwing data, mainly using the regression model algorithm of machine learning to serialize and convert the data processing module 120 The latter data are processed, including linear regression model and logistic regression model, wherein linear regression model can be further divided into simple linear regression (Simple Linear Regression), polynomial regression (Polynomial Regression) and multiple regression (Multivariable Regression), preferably, the system Use multiple regression to curve fit the feature data of multiple continuous variables, and group/cluster similar feature data with unsupervised learning, including central clustering (or K-means clustering), hierarchical clustering , proximity propagation clustering or density clustering to construct a regression model for classification, but not limited to this, supervised learning can also be used to classify feature data, including support vector machines, Bayesian classifiers Or a random forest model (a model containing multiple decision trees), etc., and then can autonomously perform classification inference or analyze and predict the results of abnormal classification.

In addition, the abnormal feedback notification learning module 150 defines and feeds back the process of adjusting the throwing rate threshold of the placement machine 200, which also uses the throwing data shared by the database 130. And use the regression model algorithm of machine learning to learn and analyze independently to make predictions, so as to define the threshold of the 200 throw rate of the placement machine in the process of actual production, and it can also be automatically adjusted according to the production conditions and characteristics of different materials. Threshold of the 200 throw rate of the placement machine, or a standard deviation can be obtained for the 200 throw rate of the mounter, and the upper and lower limits of the standard deviation are taken as whether there is a sudden increase in the throw rate, and whether it is necessary to notify the engineer to throw When the abnormality feedback notification learning module 150 checks that the material throwing rate of the placement machine 200 is greater than the feedback threshold or standard deviation range, it will actively notify the engineer of the abnormal classification result of the data learning analysis module 140 The subsequent processing flow is performed to instantly improve the throwing rate of the placement machine 200 .

Therefore, the placement machine throwing analysis and abnormal reporting system 100 of the present invention can collect the throwing data of the placement machine 200 through the communication interface 110, and the data processing module 120 serializes and converts the data and stores it in Data sharing is carried out in the database 130. After the data learning and analysis module 140 and the abnormal feedback notification learning module 150 use the regression model algorithm of machine learning to analyze and learn independently, in addition to the automatic feedback adjustment of the placement machine according to different materials In addition to the threshold of 200 material throwing rate, it can also classify abnormal material throwing anomalies (but not limited to the above four abnormal classification groups) and notify them in real time, so as to help engineers reduce the process of analyzing different material throwing abnormalities At the same time, it reduces the man-hours spent on manual analysis, and can more accurately define the threshold of the 200 material throwing rate of the placement machine. Quality and yield, and can reduce costs.

The above detailed description is for a preferred feasible embodiment of the present invention, but the embodiment is not intended to limit the scope of the present invention. All other equivalent changes and modifications are completed without departing from the technical spirit disclosed in the present invention. Changes should be included in the patent scope covered by the present invention.

To sum up, the system and method for material throwing analysis and abnormal notification of the chip mounter of the present invention can indeed achieve its effect and purpose. Therefore, the present invention is an invention with excellent practicability. In order to meet the application requirements for an invention patent, The application is filed in accordance with the law, and I hope that the review committee will approve the case as soon as possible to protect the inventor's hard work.

100: SMT machine throwing analysis and abnormal notification system

110: Communication interface

120: Data processing module

130:Database

140: Data Learning Analysis Module

150: Abnormal feedback notification learning module

160:Monitoring Platform

200: Mounter

Claims (12)

A material throwing analysis and abnormal notification system for a placement machine, comprising: A communication interface is connected to a placement machine and captures its raw material throwing data; A data processing module is connected with the communication interface for converting the throwing data into serialized data; A database is connected with the communication interface and the data processing module for storing the throwing data and the serialized data; A data learning and analysis module is connected to the database to independently check whether the database has updated data, obtain a learning and analysis result according to the serialized data, and then throw materials according to the learning and analysis results. Abnormal classification, including an abnormal suction nozzle, abnormal feeder, abnormal vacuum pressure value or abnormal incoming material of the placement machine; and An abnormal feedback notification learning module is connected with the data learning and analysis module and the database, and according to the throwing data stored in the database, it can learn and analyze automatically and adjust the throwing rate threshold of the placement machine automatically. When the abnormal feedback notification learning module checks that the throwing rate of the placement machine is not less than the feedback threshold, it will actively send a notification of the abnormal classification result of the data learning analysis module. The material throwing analysis and anomaly reporting system for a placement machine according to claim 1, wherein the data learning and analysis module uses a regression model algorithm of machine learning to process the serialized data, and uses the similar characteristic data to supervise It can be used to classify the anomaly through unsupervised or unsupervised learning, and then it can autonomously predict the result of the abnormal classification. The placement machine throwing analysis and anomaly reporting system according to claim 2, wherein the machine learning uses a multiple regression algorithm of a linear regression model to have the serialized data Curve fitting is performed on the feature data of multiple continuous variables, and similar feature data is grouped/clustered by unsupervised learning, and the clustering includes central clustering, hierarchical clustering, proximity propagation clustering or density clustering , to construct a regression model for classification. The placement machine throwing analysis and abnormal reporting system according to claim 2, wherein the machine learning uses a multiple regression algorithm of a linear regression model to perform curve fitting on the serialized data with a plurality of continuous variables, the characteristic data, And use the Bayesian classifier of supervised learning to classify the feature data. The placement machine throwing analysis and abnormal reporting system as described in claim 1 further includes a monitoring platform, which is connected to the database and the data learning and analysis module for displaying the learning and analysis of the data stored in the database The result of the module's learning and analysis. The placement machine throwing analysis and abnormal reporting system according to claim 5, wherein the monitoring platform is connected with the abnormal feedback notification learning module to receive the abnormal classification result output by the abnormal feedback notification learning module a prompt message, and proactively notify by displaying or sounding an alert. A method for material throwing analysis and abnormal notification of a chip mounter, which is applicable to a chip mounter throwing analysis and abnormal notification system, includes the following steps: Provide a communication interface to connect to a placement machine and collect the data of the original throwing data of the placement machine; A data processing module is provided to convert the throwing data of the placement machine into serialized data and store it in a database; Provide a data learning analysis module to check whether the database has updated data, and obtain a learning and analysis result according to the serialized data, and then throw the data according to the learning and analysis results. Abnormal classification of materials, including an abnormal suction nozzle of the placement machine, an abnormal feeder, an abnormal vacuum pressure value or an abnormal incoming material; Provide an abnormal feedback notification. The learning module learns and analyzes the material throwing data stored in the database independently and adjusts the threshold of the material throwing rate of the placement machine according to the automatic feedback. When checking that the material throwing rate of the placement machine is not less than the feedback rate When the threshold is exceeded, a notification will be proactively sent to the result of the abnormal classification corresponding to the data learning and analysis module. The material throwing analysis and abnormal reporting method for a placement machine as claimed in claim 7, wherein the data learning and analysis module uses a regression model algorithm of machine learning to process the serialized data, and uses the similar characteristic data to supervise It can be used to classify the anomaly through unsupervised or unsupervised learning, and then it can autonomously predict the result of the abnormal classification. The method for material throwing analysis and anomaly notification of a placement machine according to claim 8, wherein the machine learning uses a multiple regression algorithm of a linear regression model to perform curve fitting on the characteristic data of the serialized data having a plurality of continuous variables, Similar feature data are grouped/clustered by unsupervised learning, and the clustering includes central clustering, hierarchical clustering, proximity propagation clustering or density clustering to construct a regression model for classification. The method for material throwing analysis and anomaly notification of a placement machine according to claim 8, wherein the machine learning uses a multiple regression algorithm of a linear regression model to perform curve fitting on the characteristic data of the serialized data having a plurality of continuous variables, And use the Bayesian classifier of supervised learning to classify the feature data. The method for analyzing and reporting an abnormality of a placement machine according to claim 7 further provides a monitoring platform, and the monitoring platform displays the results of learning and analysis performed by the data learning and analysis module stored in the database. According to the method for material throwing analysis and abnormal notification of a placement machine according to claim 11, wherein the monitoring platform actively notifies the abnormal classification result of the abnormal feedback to the learning module by displaying or issuing an alarm.

Images