CN110276410B - Method and device for determining bad reason, electronic equipment and storage medium - Google Patents
Method and device for determining bad reason, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN110276410B CN110276410B CN201910570414.0A CN201910570414A CN110276410B CN 110276410 B CN110276410 B CN 110276410B CN 201910570414 A CN201910570414 A CN 201910570414A CN 110276410 B CN110276410 B CN 110276410B
- Authority
- CN
- China
- Prior art keywords
- data
- production
- production equipment
- model
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000003860 storage Methods 0.000 title claims abstract description 12
- 238000004519 manufacturing process Methods 0.000 claims abstract description 247
- 230000007547 defect Effects 0.000 claims abstract description 69
- 238000012549 training Methods 0.000 claims abstract description 62
- 238000001514 detection method Methods 0.000 claims abstract description 44
- 238000009826 distribution Methods 0.000 claims description 12
- 238000002372 labelling Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 description 23
- 238000012545 processing Methods 0.000 description 19
- 238000004458 analytical method Methods 0.000 description 13
- 230000000875 corresponding effect Effects 0.000 description 11
- 238000012216 screening Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 7
- 239000011521 glass Substances 0.000 description 7
- 238000007689 inspection Methods 0.000 description 7
- 238000005259 measurement Methods 0.000 description 7
- 238000007405 data analysis Methods 0.000 description 6
- 238000013075 data extraction Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000004140 cleaning Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 230000001960 triggered effect Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000002950 deficient Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 230000008676 import Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Strategic Management (AREA)
- Primary Health Care (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Manufacturing & Machinery (AREA)
- General Factory Administration (AREA)
Abstract
The invention discloses a method, a device, equipment and a storage medium for determining a bad reason, wherein the method for determining the bad reason comprises the following steps: when the detected object has known defects, extracting production data of the detected object, wherein the production data at least comprises one of production record data, an identifier of production equipment and production equipment parameters; and inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known defects, wherein the model is obtained by training according to production sample data of the detection object, and the correlation between each piece of data in the production sample data and the known defects is marked.
Description
Technical Field
The present invention relates to the field of computer data mining technologies, and in particular, to a method and an apparatus for determining a cause of a failure, an electronic device, and a storage medium.
Background
Currently, the production processes of some products, such as Organic Light-Emitting diodes (OLED) panels, are highly integrated, the processes, processes and equipment involved are many, and the production line has many burst known defects and unknown defects. If the cause of the failure needs to be determined, analysis needs to be performed based on data in each link involved in the production process of the product, but the data are distributed in databases of a plurality of different systems, data which may have problems need to be manually pulled from the databases for failure analysis, and the analysis process is complicated and the efficiency is low.
Disclosure of Invention
In view of the above, the present invention provides a method, an apparatus, a device and a storage medium for determining a cause of a failure, which can improve efficiency of determining a cause of a failure of a detection object in a production process.
According to a first aspect of the present invention, there is provided a method of determining a cause of a failure, comprising: when the detected object has known defects, extracting production data of the detected object, wherein the production data at least comprises one of production record data, an identifier of production equipment and production equipment parameters; and inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known defects, wherein the model is obtained by training according to production sample data of the detection object, and the correlation between each piece of data in the production sample data and the known defects is marked.
Optionally, the method further includes: acquiring unknown bad data to form a first data set; calculating a potential factor outlier value based on the first data set to obtain an outlier factor; sending out alarm prompt information based on the outlier factor; acquiring first mark information aiming at the alarm prompt information; marking the unknown bad data by using the first marking information to obtain marked unknown bad data; and when the marked unknown bad data reach a threshold value, importing the marked unknown bad data into a training data set of the model, wherein data in the training data set is used for training the model.
Optionally, inputting the production data into a model obtained by pre-training to obtain at least one of a production process, a production equipment and a production equipment parameter related to the known failure, including: and respectively inputting the production history data, the identification of the production equipment and the production equipment parameters into the model to respectively obtain the production process, the production equipment and the production equipment parameters related to the known defects.
Optionally, the method further includes: before extracting the production data of the detection object, obtaining production sample data of the detection object; marking each sample data according to the relevance of each sample data and known defects in the production sample data; and training the model by using the marked sample data.
Optionally, the method further includes: inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known failure, and then acquiring second labeling information aiming at the production equipment parameters; when the second label information indicates that the detection object has a defect in the production process, determining at least one of a production batch related to the defect, a production process passed by the detection object after the defect is generated and stock distribution information of the detection object according to a time period for generating the production equipment parameter.
According to a second aspect of the present invention, there is provided an apparatus for determining a cause of a failure, comprising: the extraction module is used for extracting production data of the detection object when the detection object has known defects, wherein the production data at least comprises one of production record data, an identifier of production equipment and production equipment parameters; and the input module is used for inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known defects, wherein the model is obtained by training according to production sample data of the detection object, and the correlation between each piece of data in the production sample data and the known defects is marked.
Optionally, the apparatus further comprises: the first acquisition module is used for acquiring unknown bad data to form a first data set; the calculation module is used for calculating the outlier of the potential factor based on the first data set to obtain an outlier factor; the prompting module is used for sending out alarm prompting information based on the outlier factor; the second acquisition module is used for acquiring first mark information aiming at the alarm prompt information; the marking module is used for marking the unknown bad data by using the first marking information to obtain marked unknown bad data; and the importing module is used for importing the marked unknown bad data into a training data set of the model when the marked unknown bad data reaches a threshold value, wherein data in the training data set is used for training the model.
Optionally, the input module is configured to: and respectively inputting the production history data, the identification of the production equipment and the production equipment parameters into the model to respectively obtain the production process, the production equipment and the production equipment parameters related to the known defects.
According to a third aspect of the present invention, there is provided an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing any of the methods for determining a cause of failure according to the first aspect of the present invention when executing the program.
According to a fourth aspect of the present invention, there is provided a non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the method of determining a cause of failure according to any one of the first aspect of the present invention.
As can be seen from the above description, in the method for determining the cause of the failure according to the embodiment of the present invention, for the detected object with the known failure, the production data of the detected object is input into the pre-trained model, so that the production data related to the known failure can be determined, and the cause of the failure in the production process of the detected object can be conveniently and quickly known.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the embodiments or the prior art descriptions will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart illustrating a method of determining a cause of an adverse event in accordance with an exemplary embodiment;
FIG. 2 is a flow diagram illustrating the processing of unknown bad data in a method of determining a cause of a bad according to an example embodiment;
FIG. 3 is a schematic diagram illustrating an apparatus for determining a cause of an undesirable condition in accordance with an exemplary embodiment;
fig. 4 is a block diagram illustrating an apparatus for determining a cause of an undesirable condition according to an example embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to specific embodiments and the accompanying drawings.
It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are used for distinguishing two entities with the same name but different names or different parameters, and it should be noted that "first" and "second" are merely for convenience of description and should not be construed as limitations of the embodiments of the present invention, and they are not described in any more detail in the following embodiments.
Fig. 1 is a flow chart illustrating a method of determining a cause of an undesirable condition, as shown in fig. 1, according to an exemplary embodiment, the method comprising:
step 101: when the detected object has known defects, extracting production data of the detected object, wherein the production data at least comprises one of production record data, an identifier of production equipment and production equipment parameters;
for example, the detection object may be an OLED panel.
The production history data may include, for example, data of a production process that the detection target passes through in the production process.
The identification of the production apparatus may include information such as a number, a name, a model number, and the like of the production apparatus for producing the detection object.
The production equipment parameters may include operation parameters of the production equipment when the production equipment is producing the inspection object, abnormal parameters of the production equipment when the production equipment is producing the inspection object, and the like.
In step 101, production data of the inspection object may be extracted from a data source, where the data source may include structured data such as process history information, bad information, inspection information, etc. generated in real time during the production process, and these data are respectively stored in different systems such as the production process execution, fault detection and classification, bad files, yield management, etc., and the data source may further include semi-structured and unstructured data imported by a third party, such as bad pictures, equipment logs, etc.
Step 102: and inputting the production data into a model obtained by training in advance to obtain at least one of production processes, production equipment and production equipment parameters related to the known defects, wherein the model is obtained by training according to production sample data of the detection object, and each piece of data in the production sample data is marked with the correlation with the known defects.
Wherein the production data relating to the known defect is, for example, production data that may lead to the known defect; for example, if the output of the model is production equipment 005 or 90% after the production data is input to a model trained in advance, this means that the production data related to the known defect is production equipment 005 and the probability that the production equipment 005 causes the known defect is 90%.
In one implementation, after the production data is input into the model obtained by pre-training, the model may output at least one result of the sorted production processes, production equipment and production equipment parameters according to the correlation between the production data and the known defects, for example, the model may output a process a, a process B and a process C, wherein the influence of the processes a to C on the known defects is sequentially increased, or the model may output a correlation corresponding to the process a of 80%, a correlation corresponding to the process B of 60% and a correlation corresponding to the process C of 10%.
The method for determining the cause of the badness inputs the production data of the detected object with the known badness into the pre-trained model, and can determine the production data related to the known badness, thereby being capable of conveniently and quickly knowing the cause of the badness generated in the production process of the detected object.
In the method for determining the cause of the failure, the extracted data may be processed after the data is extracted from the data source, for example, after the data is incrementally extracted from the data source, at least one of field screening, field type conversion, time slicing partitioning, data deduplication, rule matching screening, data field verification, field expansion, field calculation and the like may be performed on the extracted data. These processing methods will be explained in turn below.
And field screening treatment: and screening required data fields according to the service requirements, and removing invalid data fields.
And field type conversion processing: and converting the character data type in the source data into a time and digital data type which is easy to calculate and saves the storage space.
Time slicing partition processing: and data with large data volume is sliced according to the time dimension, so that the data processing efficiency and the processing limit are improved.
Data deduplication: and carrying out deduplication operation on the data according to the screened fields, so as to reduce the total amount of the data.
And (3) rule matching and screening treatment: and screening the data according to the data range screening rule of each field in the service requirement.
And (3) data field checking processing: and checking whether the value distribution and the value of the data field accord with the rules.
And field extension processing: one field is expanded into a plurality of fields, so that the subsequent service is convenient to use, for example, for a time field, the field can be expanded into the current day, week, month, season and the like.
And field calculation processing: in the data extraction process, the date is digitalized, and if the current month is expressed as (year x 12+ current month), the corresponding number can be directly added or subtracted to the digital date, so that the purpose of screening data according to the date is achieved.
After the data is processed by at least one of the above methods, the processed data can be combined into various business data subjects according to the business classification of the data and stored in a data warehouse. The data warehouse can be divided into model training data and full-volume big data, wherein the model training data can be extracted from the full-volume big data and used for training the model.
In order to reduce the learning cost of data cognition and the uniformity of service implementation, the data in the data warehouse can be divided into different data subjects, the data stored in the data warehouse is taken as the OLED panel production data, and the data in the data warehouse is divided into six data subjects, namely a production history data subject, a bad detection data subject, a bad point measurement data subject, a site discarded data subject, an equipment data subject and a dimension data subject. The six data topics are explained in turn below.
Production resume data topic: production history data of the whole life cycle of the GLASS grade (one piece of GLASS), HALF GLASS (half piece of GLASS) grade and PANEL grade granularity records data of processes, equipment, CST (PANEL shelf) and the like which pass through the production process of the product. The data volume of the theme is maximum, the column type storage can be adopted, the compression format storage is set, and the storage space can be saved.
Bad detection data topic: recording detection records of a detection station and a maintenance station; and recording the detection grade of the PANEL, the defect type, the defect coordinate, the maintenance record, the grade of the PANEL after maintenance, defect information and the like, wherein the data of the theme can be stored in a column mode.
Bad point location measurement data subject: the method comprises the description information of the product measurement point location, such as the machine coordinate and the measurement type of the measurement point location, and the measurement value of the point location, the measurement type and the time of pressing GLASS as the main key. The data in the data topic may be stored in a columnar manner.
Site discarding data topic: and recording the broken, irreparable and discarded GLASS, HALF GLASS and PANEL data of each site, wherein the data in the data theme can be stored in a column mode.
Device data topic: the method comprises the steps of recording relevant information such as a home site, a position, a function and classification of equipment, wherein the equipment dimension information of the whole factory is contained; recording the operating conditions of the equipment, such as temperature, humidity, voltage, current and other working environment data in production operation; and recording the information of events in the running process of the equipment, such as the time and influence of events such as shutdown, part replacement, cleaning, early warning and the like. The data in the data topic can be stored in a line mode, and the query efficiency and the concurrency of the data can be improved.
Dimension data topic: for data unification and availability among upper-level applications, a common dimension theme based on a data warehouse is established, and dimension data including production process flow dimension data of products, description dimensions of product defects, associated data of product models of all products in various factories and the like are established. The data in the data subject can be stored in a line mode, and the query efficiency and concurrency of the data can be improved.
It should be noted that the data topics are not limited to the above six topics, and new data topics can be added or reconstructed according to business needs.
Fig. 2 is a flowchart illustrating a processing of unknown bad data in a method for determining a cause of a bad according to an exemplary embodiment, where as shown in fig. 2, the method may further include, on the basis of the method shown in fig. 1:
acquiring unknown bad data to form a first data set;
where unknown bad data, such as bursts of unknown bad data generated during the production of a product, are generated, the data may be flagged as unknown bad data. The data volume of the data is small, and the burst of unknown bad data can be extracted from the data source and processed (using the data processing method described above) to form the first data set.
Calculating a potential factor outlier value based on the first data set to obtain an outlier factor;
sending out alarm prompt information based on the outlier factor;
for example, after the warning prompt information is sent, the service personnel judges the warning prompt information, and if the warning prompt information is determined to be accurate and the cause of the failure is determined to be located, the service personnel can allocate a number or an identifier to the data corresponding to the warning prompt information and can mark the data, so that the known failure is obtained.
Acquiring first mark information aiming at the alarm prompt information;
for example, first tag information input by a business person is received.
Marking the unknown bad data by using the first marking information to obtain marked unknown bad data;
and when the data volume of the marked unknown bad data reaches a threshold value, importing the marked unknown bad data into a training data set of the model, wherein data in the training data set is used for training the model.
For example, after the marked unknown bad data reaches the threshold value, the marked unknown bad data is imported into the training data set of the model, and then retraining of the model can be triggered, so that the model can identify more factors influencing bad.
In one implementation, inputting the production data into a pre-trained model to obtain at least one of a production process, a production equipment, and a production equipment parameter related to the known defect may include:
and respectively inputting the production history data, the production equipment identification and the production equipment parameter into the model to respectively obtain the production process, the type of the production equipment and the production equipment parameter related to the known failure. For example, the model may sequentially input the production history data, the identification of the production equipment, and the production equipment parameter, and the model may sequentially output information on the production processes related to the known defects (for example, the identification of the production processes ranked from high to low in correlation with the known defects), information on the production equipment related to the known defects (for example, the identification of the production equipment ranked from high to low in correlation with the known defects), and the production equipment parameter related to the known defects (for example, the production equipment parameter ranked from high to low in correlation with the known defects).
In one implementation, the method may further include: before extracting the production data of the detection object, obtaining production sample data of the detection object; marking each sample data according to the relevance of each sample data and known defects in the production sample data; and training the model by using the marked sample data. For example, model training data is extracted from the above-mentioned total large data, the data is labeled according to bad CODEs (for example, a known bad corresponds to a bad CODE), and a trained model is obtained by training based on the labeled data by a random forest model training method.
In one implementation, the method may further include: after inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known failure, receiving second labeling information aiming at the production equipment parameters; for example, after the model outputs the parameter of the production equipment, the field personnel judges that the parameter of the production equipment exceeds the preset numerical range corresponding to the parameter, and can allocate a bad CODE (which is an example of the second labeled information) to the parameter, so that the parameter indicates that the detected object has a bad property in the production process; or, if the field personnel judges that the parameter of the production equipment is within the preset numerical range corresponding to the parameter, the field personnel can mark the parameter as a normal parameter. When the second labeling information indicates that the detection object has a defect in the production process, at least one of a production batch related to the defect, a subsequent production process of the detection object after the defect is generated and the stock distribution of the detection object is determined according to the time period for generating the production equipment parameter, so that a basis can be provided for intercepting defective products.
Fig. 3 is a schematic diagram illustrating an apparatus for determining a cause of a failure according to an exemplary embodiment, as shown in fig. 3, the apparatus including:
data source 100, data extraction and processing module 110, data warehouse 120, data distribution module 130, data analysis module 140, and result presentation module 150. The modules implement a method for determining the cause of the failure through data interaction and processing, for example, the data extraction and processing module 110 incrementally extracts production data from the data source 100, and stores the data in the data warehouse 120 after cleaning and processing the data; data distribution module 130 pulls data from data warehouses 120, triggers data analysis module 140; the data analysis module 140 trains, evaluates and deploys the model, and presents the calculation results in the result presentation module 150.
The above modules will be explained below.
The data source 100: structured data 101 such as process history information, defect information, detection information and the like generated in real time in the production process are respectively stored in systems such as production process execution, fault detection and classification, defect documents, yield management and the like, and the structured data 101 and semi-structured and unstructured data 102 imported by a third party are used as data sources for defect analysis of big data.
The data extraction and processing module 110 is configured to extract data from the data source 100 incrementally and perform data processing, where the data processing may include field screening, field type conversion, time-slicing partitioning, data deduplication, rule matching screening, data field verification, field expansion, field calculation, and the like.
After the data is processed through a series of data processing, the data may be organized into various business data topics according to the business classification of the data, and stored in the data warehouse 120.
Data warehouse 120, may include model training data 121 and full volume big data 122. Model training data 121 may be extracted from the full amount of big data 122 for model training 141 in data analysis module 140. To reduce the learning cost of data awareness and the uniformity of business implementation, the data in data warehouse 120 may be divided into the six data topics 123 described above.
The data distribution 130 module can be divided into a model training trigger 131, a model training data extraction module 132, and a machine learning analysis trigger module 133. Depending on the job created by result presentation module 150 and the set scheduling configuration, machine learning analysis module 133 is triggered for known failures, while for unknown failures, data distribution module 130 extracts model training data 121 from data warehouse 120 and triggers model training module 131. Wherein, the model training data extraction module 132 is configured to extract a portion of data from the full amount of big data 122 for model training. And a machine learning analysis triggering module 133, configured to extract the full amount of data 122 periodically according to a timing mechanism, import the extracted data into a deployed model 143 (the data analysis module 140 may be divided into a model training module 141, a model evaluation module 142, and a deployed model 143), and perform result analysis.
The model training 141 module may be used for feature engineering and model construction, where feature engineering may search for features (e.g., at least one of process, equipment, and equipment parameters) that have significant effects on the undesirable effects, such as searching for outliers via box maps or searching for features that have significant corresponding effects on the undesirable effects via one-hot coded transformation of data. And (4) building a module, combining the business and the data, selecting a proper machine learning model, and discovering the internal association between the characteristics and the defects.
The model evaluation 142 module may utilize the test data set and expert experience to perform a comprehensive evaluation of the model operation results.
Deployed model 143: and deploying the trained model to a production environment for bad analysis.
The data analysis module 140 can be used for known failure analysis, unknown failure analysis, and failure prediction. For example, the unknown failure analysis may be implemented by the process shown in fig. 2, and the failure prediction function, for example, establishes a reference model for each "main process-sub process-equipment" combination (the combination refers to the main process, sub process and equipment that the product passes through in the processing process), calculates the difference between the currently obtained production equipment parameter and the reference value in real time, and stores the difference as a difference value; and when the difference value exceeds a preset threshold value, sending an abnormity early warning. The field personnel can judge whether the early warning is correct according to experience, data corresponding to the early warning can be marked according to the judgment result, and the marked data can be used for retraining the model, so that the improvement of the model is realized. The system can automatically trace the risk LOT (LOT number) ID (number), the providing list (list of bad LOT ID) and the subsequent process/inventory distribution condition thereof generated in the period to field personnel and a production process execution system according to the time range of bad data, and the risk LOT is intercepted.
The result presenting module 150 is used for providing error localization impact factors (such as poor processes, devices or equipment) and importance ranking thereof, providing risk LOT lists, presenting the process/inventory distribution thereof, and providing an interactive interface for personnel to create workflow and scheduling configuration through model analysis.
Fig. 4 is a block diagram illustrating an apparatus for determining a cause of an undesirable condition according to an exemplary embodiment, and as shown in fig. 4, the apparatus 40 includes:
an extracting module 41 (which may be a component of the data extracting and processing module 110) configured to extract production data of a detected object when the detected object has a known defect, where the production data at least includes one of production history data, an identifier of a production equipment, and a parameter of the production equipment;
and an input module 42, configured to input the production data into a model obtained through pre-training, so as to obtain at least one of a production process, a production equipment, and a production equipment parameter related to the known defect, where the model is obtained through training according to production sample data of the detection object, and each piece of data in the production sample data is marked with a correlation with the known defect.
In one implementation, the apparatus may further include: the first acquisition module is used for acquiring unknown bad data to form a first data set; the calculation module is used for calculating the outlier of the potential factor based on the first data set to obtain an outlier factor; the prompting module is used for sending out alarm prompting information based on the outlier factor; the second acquisition module is used for acquiring first mark information aiming at the alarm prompt information; the marking module is used for marking the unknown bad data by using the first marking information to obtain marked unknown bad data; and the importing module is used for importing the marked unknown bad data into a training data set of the model when the data volume of the marked unknown bad data reaches a threshold value, wherein data in the training data set is used for training the model.
In one implementation, the input module is operable to: and respectively inputting the production history data, the identification of the production equipment and the production equipment parameters into the model to respectively obtain the production process, the production equipment and the production equipment parameters related to the known defects.
In one implementation, the apparatus may further include: the third acquisition module is used for acquiring production sample data of the detection object before extracting the production data of the detection object; the marking module is used for marking each sample data according to the correlation between each sample data in the production sample data and the known badness; and the training module is used for training the model by using the marked sample data.
In one implementation, the apparatus further comprises: a third obtaining module, configured to obtain second labeling information for the production equipment parameter after inputting the production data into a model obtained through pre-training to obtain at least one of production processes, production equipment, and production equipment parameters related to the known defect; and a determining module, configured to determine, according to a time period during which the parameter of the production equipment is generated, at least one of a production lot (for example, a list of production lots related to the defect may be provided), a subsequent production process of the inspection object after the defect is generated, and inventory distribution information of the inspection object, when the second label information indicates that the inspection object has a defect in the production process.
It should be noted that the apparatus in the foregoing embodiment is used for implementing the corresponding method in the foregoing embodiment, and has the beneficial effects of the corresponding method embodiment, which are not described herein again.
Based on the same inventive concept, an embodiment of the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the method for determining the cause of the failure according to any of the above embodiments is implemented.
Based on the same inventive concept, embodiments of the present invention further provide a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform the method for determining a cause of failure as described in any of the above embodiments.
Those of ordinary skill in the art will understand that: the discussion of any embodiment above is meant to be exemplary only, and is not intended to intimate that the scope of the disclosure, including the claims, is limited to these examples; within the idea of the invention, also features in the above embodiments or in different embodiments may be combined, steps may be implemented in any order, and there are many other variations of the different aspects of the invention as described above, which are not provided in detail for the sake of brevity.
In addition, well known power/ground connections to Integrated Circuit (IC) chips and other components may or may not be shown within the provided figures for simplicity of illustration and discussion, and so as not to obscure the invention. Furthermore, devices may be shown in block diagram form in order to avoid obscuring the invention, and also in view of the fact that specifics with respect to implementation of such block diagram devices are highly dependent upon the platform within which the present invention is to be implemented (i.e., specifics should be well within purview of one skilled in the art). Where specific details (e.g., circuits) are set forth in order to describe example embodiments of the invention, it should be apparent to one skilled in the art that the invention can be practiced without, or with variation of, these specific details. Accordingly, the description is to be regarded as illustrative instead of restrictive.
While the present invention has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those of ordinary skill in the art in light of the foregoing description. For example, other memory architectures (e.g., dynamic ram (dram)) may use the discussed embodiments.
The embodiments of the invention are intended to embrace all such alternatives, modifications and variances that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, substitutions, improvements and the like that may be made without departing from the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (7)
1. A method of determining a cause of an undesirable condition, comprising:
when the detected object has known defects, extracting production data of the detected object, wherein the production data at least comprises one of production record data, an identifier of production equipment and production equipment parameters, and the detected object is an OLED panel;
inputting the production data into a model obtained by pre-training to obtain at least one of production processes, production equipment and production equipment parameters related to the known defects, wherein the model is obtained by training according to production sample data of the detection object, and the correlation between each piece of data in the production sample data and the known defects is marked;
the method further comprises the following steps:
acquiring unknown bad data to form a first data set;
calculating a potential factor outlier value based on the first data set to obtain an outlier factor;
sending out alarm prompt information based on the outlier factor;
acquiring first mark information aiming at the alarm prompt information;
marking the unknown bad data by using the first marking information to obtain marked unknown bad data;
when the data volume of the marked unknown bad data reaches a threshold value, importing the marked unknown bad data into a training data set of the model, wherein data in the training data set is used for training the model;
inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known failure, and then acquiring second labeling information aiming at the production equipment parameters; the second labeling information comprises numerical range information corresponding to the production equipment parameters;
when the second labeling information indicates that the detection object has a defect in the production process, determining at least one of a production batch related to the defect, a subsequent production process of the detection object after the defect is generated and inventory distribution information of the detection object according to a time period for generating the production equipment parameter.
2. The method of claim 1, wherein inputting the production data into a pre-trained model to obtain at least one of production process, production equipment, and production equipment parameters associated with the known defects comprises:
and respectively inputting the production history data, the identification of the production equipment and the production equipment parameters into the model to respectively obtain the production process, the production equipment and the production equipment parameters related to the known defects.
3. The method of claim 1, further comprising:
before extracting the production data of the detection object, obtaining production sample data of the detection object;
marking each sample data according to the relevance of each sample data and known defects in the production sample data;
and training the model by using the marked sample data.
4. An apparatus for determining a cause of an undesirable condition, comprising:
the extraction module is used for extracting the production data of the detection object when the detection object has known defects, wherein the production data at least comprises one of production record data, an identifier of production equipment and production equipment parameters, and the detection object is an OLED panel;
an input module, configured to input the production data into a model obtained through pre-training, so as to obtain at least one of a production process, a production equipment, and a production equipment parameter related to the known defect, where the model is obtained through training according to production sample data of the detection object, and each piece of data in the production sample data marks a correlation between the piece of data and the known defect;
the device further comprises:
the first acquisition module is used for acquiring unknown bad data to form a first data set;
the calculation module is used for calculating the outlier of the potential factor based on the first data set to obtain an outlier factor;
the prompting module is used for sending out alarm prompting information based on the outlier factor;
the second acquisition module is used for acquiring first mark information aiming at the alarm prompt information;
the marking module is used for marking the unknown bad data by using the first marking information to obtain marked unknown bad data;
the importing module is used for importing the marked unknown bad data into a training data set of the model when the data volume of the marked unknown bad data reaches a threshold value, wherein data in the training data set is used for training the model;
inputting the production data into a model obtained by pre-training to obtain at least one of production process, production equipment and production equipment parameters related to the known failure, and then acquiring second labeling information aiming at the production equipment parameters; the second labeling information comprises numerical range information corresponding to the production equipment parameters;
when the second labeling information indicates that the detection object has a defect in the production process, determining at least one of a production batch related to the defect, a subsequent production process of the detection object after the defect is generated and inventory distribution information of the detection object according to a time period for generating the production equipment parameter.
5. The apparatus of claim 4, wherein the input module is configured to:
and respectively inputting the production history data, the identification of the production equipment and the production equipment parameters into the model to respectively obtain the production process, the production equipment and the production equipment parameters related to the known defects.
6. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of determining the cause of the failure according to any one of claims 1 to 3 when executing the program.
7. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method of determining a cause of failure of any one of claims 1 to 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910570414.0A CN110276410B (en) | 2019-06-27 | 2019-06-27 | Method and device for determining bad reason, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910570414.0A CN110276410B (en) | 2019-06-27 | 2019-06-27 | Method and device for determining bad reason, electronic equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110276410A CN110276410A (en) | 2019-09-24 |
| CN110276410B true CN110276410B (en) | 2022-06-03 |
Family
ID=67962527
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910570414.0A Active CN110276410B (en) | 2019-06-27 | 2019-06-27 | Method and device for determining bad reason, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110276410B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112825161A (en) * | 2019-11-21 | 2021-05-21 | 鸿富锦精密电子（成都）有限公司 | Product yield prediction method, device, computer device and storage medium |
| CN111159645B (en) * | 2019-12-19 | 2023-06-06 | 成都数之联科技股份有限公司 | Bad root cause positioning method based on product production histories and parameters |
| WO2021142622A1 (en) * | 2020-01-14 | 2021-07-22 | 京东方科技集团股份有限公司 | Method for determining cause of defect, and electronic device, storage medium, and system |
| CN114492898A (en) * | 2020-10-28 | 2022-05-13 | 北京中祥英科技有限公司 | Product failure prediction method and device, electronic device and storage medium |
| CN114429256A (en) * | 2020-10-29 | 2022-05-03 | 北京中祥英科技有限公司 | Data monitoring method and device, electronic equipment and storage medium |
| WO2022088084A1 (en) * | 2020-10-30 | 2022-05-05 | 京东方科技集团股份有限公司 | Data processing method, apparatus and system, and electronic device |
| WO2022116107A1 (en) * | 2020-12-03 | 2022-06-09 | Boe Technology Group Co., Ltd. | Data management platform, intelligent defect analysis system, intelligent defect analysis method, computer-program product, and method for defect analysis |
| DE112021001736T5 (en) * | 2021-03-26 | 2023-01-05 | Beijing Zhongxiangying Technology Co., Ltd. | DATA PROCESSING METHODS AND DEVICES, ELECTRONIC DEVICE AND STORAGE MEDIA |
| CN113139734A (en) * | 2021-04-30 | 2021-07-20 | 重庆城市管理职业学院 | Intelligent manufacturing management system based on data mining |
| CN115623872A (en) * | 2021-04-30 | 2023-01-17 | 京东方科技集团股份有限公司 | Data processing method, device, equipment and storage medium |
| CN113570184B (en) * | 2021-06-07 | 2024-07-12 | 东莞盛翔精密金属有限公司 | Tracing method, tracing system, electronic device and computer readable storage medium |
| CN114022059B (en) * | 2022-01-06 | 2022-05-27 | 锱云(上海)物联网科技有限公司 | Method, system, terminal device and storage medium for analyzing bad reason of product |
| CN114418011B (en) * | 2022-01-21 | 2023-05-12 | 京东方科技集团股份有限公司 | Method, equipment and system for analyzing bad causes of products and storage medium |
| CN114912898A (en) * | 2022-05-27 | 2022-08-16 | 上海哥瑞利软件股份有限公司 | System for rapidly analyzing equipment difference root cause in semiconductor manufacturing |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108008332A (en) * | 2017-11-29 | 2018-05-08 | 国网山东省电力公司电力科学研究院 | A kind of new energy Remote testing device method for diagnosing faults based on data mining |
| CN109711659A (en) * | 2018-11-09 | 2019-05-03 | 成都数之联科技有限公司 | A kind of industrial Yield lmproved management system and method |
| CN109739902A (en) * | 2018-12-29 | 2019-05-10 | 联想(北京)有限公司 | A kind of data analysing method, equipment and computer readable storage medium |
Family Cites Families (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP5948262B2 (en) * | 2013-01-30 | 2016-07-06 | 株式会社日立ハイテクノロジーズ | Defect observation method and defect observation apparatus |
| CN104123298B (en) * | 2013-04-26 | 2017-09-29 | 华为技术有限公司 | The analysis method and equipment of product defects |
| CN108241925A (en) * | 2016-12-23 | 2018-07-03 | 重庆邮电大学 | A kind of discrete manufacture mechanical product quality source tracing method based on outlier detection |
| JP7077553B2 (en) * | 2017-09-21 | 2022-05-31 | 富士フイルムビジネスイノベーション株式会社 | Quality control equipment |
| CN108257121B (en) * | 2018-01-09 | 2019-01-25 | 北京百度网讯科技有限公司 | Method, apparatus, storage medium and the terminal device that product defects detection model updates |
| CN108596229B (en) * | 2018-04-13 | 2021-09-10 | 北京华电智慧科技产业有限公司 | Method and system for monitoring and diagnosing online abnormity |
| CN109446189A (en) * | 2018-10-31 | 2019-03-08 | 成都天衡智造科技有限公司 | A kind of technological parameter outlier detection system and method |
-
2019
- 2019-06-27 CN CN201910570414.0A patent/CN110276410B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108008332A (en) * | 2017-11-29 | 2018-05-08 | 国网山东省电力公司电力科学研究院 | A kind of new energy Remote testing device method for diagnosing faults based on data mining |
| CN109711659A (en) * | 2018-11-09 | 2019-05-03 | 成都数之联科技有限公司 | A kind of industrial Yield lmproved management system and method |
| CN109739902A (en) * | 2018-12-29 | 2019-05-10 | 联想(北京)有限公司 | A kind of data analysing method, equipment and computer readable storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 基于改进决策树算法识别LCD产品质量影响因素;郑晓笑等;《现代制造工程》;20151231(第4期);第21-26页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110276410A (en) | 2019-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110276410B (en) | Method and device for determining bad reason, electronic equipment and storage medium | |
| CN102611565B (en) | Regular expression-based alarm correlation analysis method for monitoring system | |
| CN106933691B (en) | Identification method and system for abnormal points in seasonal time series data sequence | |
| CN108197177B (en) | Business object monitoring method and device, storage medium and computer equipment | |
| US20180312180A1 (en) | Determinging root cause of locomotive failure | |
| CN104424539A (en) | Maintenance information management system and method, and maintenance information display method | |
| US20150149541A1 (en) | Leveraging Social Media to Assist in Troubleshooting | |
| CN110109908A (en) | Analysis system and method based on the potential relationship of social base information excavating personage | |
| CN110716539A (en) | Fault diagnosis and analysis method and device | |
| US10839303B2 (en) | Automatic detection and correction of license plate misidentification | |
| CN113220875A (en) | Internet information classification method and system based on industry label and electronic equipment | |
| US20220303188A1 (en) | Managing telecommunication network event data | |
| CN111143404A (en) | Service processing method and device | |
| CN111330871B (en) | Quality classification method and device | |
| CN111413952A (en) | Robot fault detection method and device, electronic equipment and readable storage medium | |
| CN103577964A (en) | Method and system for acquiring product circulation regional information and verifying authenticity of product | |
| CN114384885B (en) | Process parameter adjusting method, device, equipment and medium based on abnormal working conditions | |
| JP2017167578A (en) | Incident management system | |
| CN113781068A (en) | Online problem solving method and device, electronic equipment and storage medium | |
| CN114996080A (en) | Data processing method, device, equipment and storage medium | |
| CN113656287B (en) | Method and device for predicting software instance faults, electronic equipment and storage medium | |
| CN105227333B (en) | The method and apparatus of network element device data maintenance | |
| CN115185792A (en) | Fault hardware processing method, device and system | |
| CN115470953A (en) | Exhibition hall management method, electronic device and computer-readable storage medium | |
| CN115903720A (en) | Fault diagnosis system and method for rail transit, and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |