# US 11,029,260 B2 · Solder paste printing quality inspection system and method > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,029,260 B2 - Document ID: US11029260B2 - Status: Granted - Kind: B2 - Application number: 16/398,764 - Filed: 2019-04-30 - Published: 2020-09-17 - Granted: 2021-06-08 - Earliest claimed: 2019-03-15 - Assignees: HONGFUJIN PRECISION ELECTRONICS (CHENGDU) Co., Ltd.; Hon Hai Precision Industry Co., Ltd. - Applicants: HONGFUJIN PRECISION ELECTRONICS (CHENGDU) Co., Ltd.; Hon Hai Precision Industry Co., Ltd. - Inventors: Zi-Qing Xia; Xiao-Lei Liu; Min Chen; Yi-Kun Wang; Fu-Ju Zeng - Web page: https://patentgrep.com/patent/US11029260B2 ## CPC - G01N 21/95684: {Patterns showing highly reflecting parts, e.g. metallic elements} - G06T 7/0004: {Industrial image inspection} - G01N 21/95607: {using a comparative method} - G06T 7/001: {using an image reference approach} - G01N 2021/8883: {involving the calculation of gauges, generating models} - G01N 2021/95615: {with stored comparision signal} - G01N 2021/95646: {Soldering} - G06T 2207/20081: Training; Learning - G06T 2207/20084: Artificial neural networks [ANN] - G06T 2207/30152: Solder - G01N: INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES (measuring or testing apparatus or processes other than immunoassay, involving enzymes or microorganisms C12M, C12Q) - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL ## Abstract A solder paste printing quality inspection (SPI) method includes establishing at least one data inspection model, acquiring real-time test data, preprocessing the real-time test data, and determining a type of the real-time test data and inputting the real-time test data into the data inspection model corresponding to the type of the real-time test data to obtain a judgment result of the solder paste printing quality inspection. The real-time test data includes one or both of a test image and a test value of solder paste printing quality inspection. The preprocessing includes obtaining key parameters of the test value or obtaining image data of the test image and standardizing the real-time test data. ## Claims ### Claim 1 A solder paste printing quality inspection (SPI) method comprising: establishing at least one data inspection model comprising a data inspection model based on the random forest algorithm and a data inspection model based on the convolutional neural network algorithm; acquiring real-time test data comprising one or both of a test image and a test value of solder paste printing quality inspections; preprocessing the real-time test data, the preprocessing comprising obtaining parameters of the test value or obtaining image data of the test image, and standardizing the real-time test data; and determining a type of the real-time test data, and inputting the real-time test data into the data inspection model corresponding to the type of the real-time test data to obtain a judgment result of the solder paste printing quality inspection, wherein when the type of the real-time test data is the test value, the data inspection model based on the random forest algorithm is implemented; when the type of the real-time test data is the test image, the data inspection model based on the convolutional neural network algorithm is implemented; when the type of the real-time test data is the test value and the test image, the data inspection model based on the random forest algorithm and the data inspection model based on the convolutional neural network algorithm are both implemented; the judgment result of the test value is output by the data inspection model based on the random forest algorithm; and the judgment result of the test image is output by the data inspection model based on the convolutional neural network algorithm. ### Claim 2 (depends on 1) The SPI method of claim 1, wherein establishing the data inspection model comprises: acquiring historical test data; determining an algorithm corresponding to the data inspection model according to the historical test data; subjecting the historical test data to feature engineering processing to extract the parameters; and establishing and training the data inspection model according to the parameters. ### Claim 3 (depends on 2) The SPI method of claim 2, wherein: the historical test data comprises the test value, the test image, test principles, and domain knowledge. ### Claim 4 (depends on 2) The SPI method of claim 2, wherein: the algorithm comprises a random forest algorithm and a convolutional neural network algorithm. ### Claim 5 (depends on 1) The SPI method of claim 1 further comprising: analyzing a rate of acceptability and a rate of unacceptability, statistically analyzing reasons for unacceptability, and outputting a statistical report. ### Claim 6 (depends on 5) The SPI method of claim 5 further comprising: initiating an alarm process if the rate of unacceptability increases abnormally to alert a worker that a solder paste printing machine may be malfunctioning. ### Claim 7 A data processing device comprising: a processor; and a memory storing a plurality of instructions, which when executed by the processor, cause the processor to: establish at least one data inspection model comprising a data inspection model based on the random forest algorithm and a data inspection model based on the convolutional neural network algorithm; acquire real-time test data comprising one or both of a test image and a test value of solder paste printing quality inspections; preprocess the real-time test data, the preprocessing comprising obtaining parameters of the test value or obtaining image data of the test image, and standardizing the real-time test data; and determine a type of the real-time test data, and input the real-time test data into the data inspection model corresponding to the type of the real-time test data to obtain a judgment result of the solder paste printing quality inspection, wherein when the type of the real-time test data is the test value, the data inspection model based on the random forest algorithm is implemented; when the type of the real-time test data is the test image, the data inspection model based on the convolutional neural network algorithm is implemented; when the type of the real-time test data is the test value and the test image, the data inspection model based on the random forest algorithm and the data inspection model based on the convolutional neural network algorithm are both implemented; the judgment result of the test value is output by the data inspection model based on the random forest algorithm; and the judgment result of the test image is output by the data inspection model based on the convolutional neural network algorithm. ### Claim 8 (depends on 7) The data processing device of claim 7, wherein the data inspection model is established by: acquiring historical test data; determining an algorithm corresponding to the data inspection model according to the historical test data; subjecting the historical test data to feature engineering processing to extract the parameters; and establishing and training the data inspection model according to the parameters. ### Claim 9 (depends on 8) The data processing device of claim 8, wherein: the historical test data comprises the test value, the test image, test principles, and domain knowledge. ### Claim 10 (depends on 8) The data processing device of claim 8, wherein: the algorithm comprises a random forest algorithm and a convolutional neural network algorithm. ### Claim 11 (depends on 7) The data processing device of claim 7, wherein the processor is further configured to: analyze a rate of acceptability and a rate of unacceptability, statistically analyze reasons for unacceptability, and output a statistical report. ### Claim 12 (depends on 11) The data processing device of claim 11, wherein the processor is further configured to: initiate an alarm process if the rate of unacceptability increases abnormally to alert a worker that a solder paste printing machine may be malfunctioning.