KR102334694B1 - Cloud server for managing defect determination algorithm applied to manufactured products and smart factory management system including the same - Google Patents

Abstract

The present invention relates to a cloud server that manages a number of manufacturing sites. The cloud server includes a storage unit, a communication unit, and a processor operatively connected to the communication unit. The processor is configured to receive quality data obtained in a process from an on-site server deployed at a manufacturing site, calculate pass/fail determination accuracy of products generated in the process by substituting the quality data into a plurality of pre-stored defect determination algorithms, select a defect determination algorithm applicable to the process based on the pass/fail determination accuracy, and provide the selected defect determination algorithm to the on-site server, or the processor is configured to receive manufacturing data of equipment installed at the manufacturing site from the on-site server, obtain an inspection cycle prediction result of the equipment by substituting the manufacturing data into a plurality of pre-stored predictive maintenance algorithms, select a predictive maintenance algorithm applicable to the equipment based on the inspection cycle prediction result, and provide the selected predictive maintenance algorithm to the on-site server. According to the present invention, the cloud server stores and analyzes industrial data and operation/control data generated at the manufacturing site and forms big data for factory automation operations to manage production, quality, equipment, and energy data of factories.

Description

CLOUD SERVER FOR MANAGING DEFECT DETERMINATION ALGORITHM APPLIED TO MANUFACTURED PRODUCTS AND SMART FACTORY MANAGEMENT SYSTEM INCLUDING THE SAME

The present invention relates to an algorithm management method for efficiently operating process facilities arranged at each site in a large-scale industrial complex performing manufacturing and maintaining product quality, and a cloud server and a smart factory management system for performing the same .

Recently, attempts to converge ICT (Information and Communication Technology) technologies such as Internet of Things (IoT), wireless communication, big data, cloud computing, and augmented reality with various industries are increasing.

In the case of the manufacturing industry, development of a smart factory capable of autonomous management by monitoring, analyzing, and judging manufactured products and facilities by itself using artificial intelligence (AI) software is being developed.

As a representative example, many studies are being conducted to monitor various error situations (facility error, production load, network error, etc.) occurring in a factory in real time, and to cope with and judge by themselves.

Accordingly, various algorithms and learning models have been developed to keep the production efficiency of products constant, but there is a limit to improving problems in a dynamic manufacturing environment in which individual servers controlling a factory change frequently.

In particular, since the burden of computation is high to store and manage a large amount of data generated in a factory that is running 24 hours a day, the server has no choice but to identify errors by extracting a small number of samples, which in turn lowers the accuracy of error determination.

The description underlying the invention has been prepared to facilitate understanding of the invention. It should not be construed as an admission that the matters described in the background technology of the invention exist as prior art.

Accordingly, the problem to be solved by the present invention is to overcome the computational limitations of individual servers, and to store all industrial data (eg, long-term manufacturing data, large-capacity quality data) collected/generated in processes and facilities for efficient operation of factories, and , to provide a method for recommending and managing algorithms suitable for various sites by analyzing them, and a cloud server and management system that performs them.

Another problem to be solved by the present invention is to collect unstructured data (error data) in addition to the standardized data from the cloud server linked with the on-site server while the process/facility manager is in progress by the on-site server deployed on the site in real time However, it is to provide a method that can quickly suggest whether there is a problem and an algorithm for solving it, as well as a cloud server and management system that performs it.

The problems of the present invention are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the following description.

In order to solve the above problems, there is provided a cloud server according to an embodiment of the present invention. The server includes a storage unit, a communication unit, the storage unit, and a processor operatively connected to the communication unit, wherein the processor receives quality data obtained in a process process from a field server disposed at a manufacturing site, and the quality By substituting data into a plurality of pre-stored defect determination algorithms, the pass/fail determination accuracy of the product generated through the process process is calculated, and a defect determination algorithm applicable to the process process is selected based on the pass/fail determination accuracy and provide the determined failure determination algorithm to the on-site server.

According to a feature of the present invention, the processor receives, from the on-site server, process data of at least one of a type of process process, an operating condition, and a standard image and attribute of a product manufactured through the process process, and based on the process data can set the type of product defect that may occur in the process process.

According to another feature of the present invention, when the process process is a new process process, the defect determination algorithm having the highest pass/fail determination accuracy for each defective type of the product may be selected as an algorithm applicable to the process process. have.

According to another feature of the present invention, the processor, when the process process is a previously registered process process, receives the pass/fail determination accuracy for the product produced in the process process from the on-site server for each defective type of the product and, if the difference between the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through the plurality of failure determination algorithms is within a preset range, maintain the pass/fail determination criterion for the type of failure, or maintain the received pass/fail determination accuracy and the When the difference from the pass/fail determination accuracy calculated through the plurality of failure determination algorithms is out of a preset range, a failure determination algorithm for the corresponding failure type may be newly selected.

According to another feature of the present invention, the processor may receive long-term quality data collected by the on-site server for a predetermined period or large-capacity quality data exceeding a specified size.

According to another feature of the present invention, the quality data may include an image of a product produced through any one process process and a pass/fail determination result of the corresponding product pre-stored in the on-site server.

According to another feature of the present invention, the processor receives the defective rate of the product calculated according to the defect determination algorithm applied to the process process from the on-site server through the communication unit, and monitors whether the received defective rate is less than or equal to a preset value and when the received defective rate is less than or equal to a preset value according to the monitoring result, manufacturing data for checking the equipment related to the process process is received from the on-site server through the communication unit, and a plurality of pre-stored predictive maintenance algorithms are used. Therefore, it is possible to determine whether to change the inspection period of the relevant equipment.

According to another feature of the present invention, the communication unit may periodically receive from the field server based on the time when any one product is produced according to the process process.

In order to solve the above problems, there is provided a method for managing a failure determination algorithm according to another embodiment of the present invention. The method is a method in which a cloud server manages a defect determination algorithm for each process of a manufactured product, comprising: receiving quality data acquired in a process process from an edge server disposed at a manufacturing site; Substituting into the determination algorithm, calculating the pass/fail determination accuracy of the product generated through the process process, selecting a failure determination algorithm applicable to the process process based on the pass/fail determination accuracy, and the selected failure determination algorithm and providing to the edge server.

In order to solve the above problems, there is provided a smart factory management system according to another embodiment of the present invention. The system receives quality data for any one manufactured product from the field server and the field server that determine the defect of the manufactured product for each process process of the manufactured product, and applies the quality data to a plurality of pre-stored defect determination algorithms Thus, it is configured to include a cloud server that selects a failure determination algorithm applicable to the corresponding process process.

Specific details of other embodiments are included in the detailed description and drawings.

The present invention stores and analyzes industrial data and operation/control data generated at the manufacturing site by a cloud server linked with a plurality of field servers that control the manufacturing site, and forms big data for factory automation operation, A smart factory can be built that can manage production, quality, equipment, and energy data.

The present invention selects an algorithm for the cloud server to maintain the quality of products manufactured for each manufacturing site based on production, quality, equipment, and energy data of factories, and a predictive maintenance algorithm to prevent deterioration and damage of factory equipment and it can be provided to a field server deployed at each site. In particular, by newly implementing factory automation, factory automation can be implemented using the cloud server even at new sites that do not have countermeasures for various problem situations, and even at sites that generate large amounts of industrial data due to many types of production facilities. As the cloud server processes data quickly, efficient operation of the factory may be possible.

According to the present invention, by additionally performing performance evaluation before providing various algorithms stored in the cloud server for each factory facility and process, reckless application of algorithms can be prevented and factory production efficiency can be increased according to appropriate algorithm application.

The present invention provides an interface for the cloud server to collect various industrial data (manufacturing data, quality data, such as production, quality, equipment, energy data, etc.) and visually analyze it, and a manufacturing factory established in a smart factory We can provide you with customized analysis consulting.

The effect according to the present invention is not limited by the contents exemplified above, and more various effects are included in the present invention.

1 is a schematic diagram for explaining a smart factory management system according to an embodiment of the present invention.
2 is a block diagram illustrating a configuration of a cloud server according to an embodiment of the present invention.
3A to 3D are schematic diagrams for explaining an industrial data analysis interface implemented in a cloud server according to an embodiment of the present invention.
4 is a flowchart illustrating a method for managing a failure determination algorithm according to an embodiment of the present invention.
5 is a flowchart illustrating step S130 illustrated in FIG. 4 .
6 is a flowchart illustrating a method for managing a predictive maintenance algorithm according to an embodiment of the present invention.
7 is a flowchart illustrating step S230 illustrated in FIG. 6 .

Advantages and features of the present invention, and a method of achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but will be implemented in a variety of different forms, only these embodiments allow the disclosure of the present invention to be complete, and common knowledge in the technical field to which the present invention pertains It is provided to fully inform the possessor of the scope of the invention, and the present invention is only defined by the scope of the claims. In connection with the description of the drawings, like reference numerals may be used for like components.

In this document, expressions such as "have," "may have," "includes," or "may include" refer to the presence of a corresponding characteristic (eg, a numerical value, function, operation, or component such as a part). and does not exclude the presence of additional features.

In this document, expressions such as “A or B,” “at least one of A or/and B,” or “one or more of A or/and B” may include all possible combinations of the items listed together. . For example, "A or B," "at least one of A and B," or "at least one of A or B" means (1) includes at least one A, (2) includes at least one B; Or (3) it may refer to all cases including both at least one A and at least one B.

As used herein, expressions such as "first," "second," "first," or "second," may modify various elements, regardless of order and/or importance, and refer to one element. It is used only to distinguish it from other components, and does not limit the components. For example, the first user equipment and the second user equipment may represent different user equipment regardless of order or importance. For example, without departing from the scope of rights described in this document, the first component may be named as the second component, and similarly, the second component may also be renamed as the first component.

A component (eg, a first component) is "coupled with/to (operatively or communicatively)" to another component (eg, a second component); When referring to "connected to", it will be understood that the certain element may be directly connected to the other element or may be connected through another element (eg, a third element). On the other hand, when it is said that a component (eg, a first component) is "directly connected" or "directly connected" to another component (eg, a second component), the component and the It may be understood that other components (eg, a third component) do not exist between other components.

The expression "configured to (or configured to)" as used in this document, depending on the context, for example, "suitable for," "having the capacity to ," "designed to," "adapted to," "made to," or "capable of." The term “configured (or configured to)” may not necessarily mean only “specifically designed to” in hardware. Instead, in some circumstances, the expression “a device configured to” may mean that the device is “capable of” with other devices or parts. For example, the phrase "a processor configured (or configured to perform) A, B, and C" refers to a dedicated processor (eg, an embedded processor) for performing the corresponding operations, or by executing one or more software programs stored in a memory device. , may mean a generic-purpose processor (eg, a CPU or an application processor) capable of performing corresponding operations.

Terms used in this document are only used to describe specific embodiments, and may not be intended to limit the scope of other embodiments. The singular expression may include the plural expression unless the context clearly dictates otherwise. Terms used herein, including technical or scientific terms, may have the same meanings as commonly understood by one of ordinary skill in the art described in this document. Among terms used in this document, terms defined in a general dictionary may be interpreted with the same or similar meaning as the meaning in the context of the related art, and unless explicitly defined in this document, ideal or excessively formal meanings is not interpreted as In some cases, even terms defined in this document cannot be construed to exclude embodiments of this document.

Each feature of the various embodiments of the present invention may be partially or wholly combined or combined with each other, and as those skilled in the art will fully understand, technically various interlocking and driving are possible, and each embodiment may be implemented independently of each other. It may be possible to implement together in a related relationship.

For clarity of interpretation of the present specification, terms used herein will be defined below.

1 is a schematic diagram for explaining a smart factory management system according to an embodiment of the present invention.

Referring to Figure 1, the smart factory management system 1000 is a cloud server (Cloud Server) 100 that manages all the sites, a plurality of field servers (edge server, Edge Server) 200 arranged for each site, the site The server 200 may include a factory 300 controlled by the server 200 and a plurality of process facilities 350 installed in the factory 300 .

The cloud server 100 collects industrial data from a plurality of field servers 200 for efficient operation of the factory 300, and an algorithm or factory ( 300), an algorithm for maintaining the production efficiency of the process equipment 350 may be managed. Here, the industrial data may include quality data of a manufactured product acquired in the process and manufacturing data of the process facility 350 . More specifically, the quality data includes an image of a product produced through any one process and a result of determining whether the product is pass/fail, and the manufacturing data includes programmable logic controller (PLC) control data that can be measured in the process facility 350, noise , vibration, temperature, humidity, dust, operating time, and number of products produced.

In the present invention, "algorithm" means a sequence that analyzes industrial data, makes a judgment on an item based on the analysis result, or makes a prediction, and is defective depending on the type of item that can be determined/predicted in the factory 300 It may be defined as a decision algorithm, a predictive maintenance algorithm, and the like.

In addition, in the present invention, "managing the algorithm" means that the cloud server 100 monitors the field server 200, while the field server 200 determines or predicts any item based on the algorithm, It can be understood that the accuracy is lowered, and accordingly, the cloud server 100 adjusts the data value specified in the algorithm possessed by the on-site server 200 or changes or adds a new algorithm.

That is, the cloud server 100 does not newly generate and provide an algorithm for detecting product defects or diagnosing the replacement cycle of equipment for each factory 300 producing products, but rather an algorithm stored in the cloud server 100 and An algorithm suitable for the factory 300 or process equipment 350 may be recommended based on industrial data (quality data, manufacturing data) It can help in a field with limitations.

Specifically, the cloud server 100 performs rapid analysis and algorithm diagnosis of industrial data, which is impossible in the field server 200 that controls the factory 300 or process equipment 350 based on a large amount of algorithms and industrial data. can For example, the cloud server 100 detects defects for each product and each process, or stores specific and various algorithms (eg, defect detection algorithm, equipment inspection (predictive maintenance) algorithm) for inspecting equipment to detect a large amount of equipment. It is possible to determine whether an appropriate algorithm is applied to the factory 300 and the process equipment 350 based on the industrial data received from / after the construction of the / learning data from the field server 200 .

In addition, the cloud server 100 may collect industrial data (quality data, manufacturing data) collected in the factory 300 or process facility 350 that produces the product in real time, and analyze the dynamic failure, and the analysis result Based on this, a pre-stored algorithm may be updated.

Accordingly, it is possible to prevent failures that may occur due to various variables while the factory 300 or the process equipment 350 is operated according to the algorithm recommended by the cloud server 100 thereafter.

The on-site server 200 is an on-site control server and an edge server that controls the process equipment 350 in the factory 300 and manages the quality of products, and includes manufacturing data and quality data from the process equipment 350 in real time. , and may manage the process equipment 350 .

The field server 200 may transmit industrial data (manufacturing data, quality data) received in real time to the cloud server 100 . According to an embodiment, the on-site server 200 may periodically transmit industrial data of all types and sizes to the cloud server 100, and the transmission period is variable according to the setting of the administrator, but the time at which the transmission starts is It may be a time when the process equipment 350 is installed or a time when any one product is produced.

The on-site server 200 may determine whether a product is defective by substituting the quality data collected in real time to a pre-specified defect detection algorithm, and substituting the manufacturing data collected in real time to the pre-specified predictive maintenance algorithm to process equipment 350 ) can be predicted by itself. Here, the algorithm specified in the on-site server 200 may be an algorithm established based on control data for the process equipment 350 accumulated by the manager or an algorithm provided from the cloud server 100 .

In addition, the field server 200 may perform analysis only on industrial data that does not require advanced data analysis. Specifically, the field server 200 can learn in advance and analyze in real time in the case of standardized industrial data or low-capacity industrial data, in addition to unstructured industrial data, large-capacity industrial data, or industrial data that must be collected for a long period of time. An analysis may be requested from the cloud server 100 . Here, the unstructured industrial data refers to new industrial data that has not been previously collected, and for example, the industrial data may include an image file of a low pixel so that the quality of the product cannot be confirmed.

In addition, in the case of industrial data related to the process equipment 350 that requires an immediate response according to the type of industrial data, the on-site server 200 may analyze even if unstructured industrial data is received from the process equipment 350, An analysis may be immediately requested to the cloud server 100 . Thereafter, the field server 200 may maintain or change the currently set failure determination/predictive maintenance algorithm according to the analysis result transmitted by the cloud server 100 .

The factory 300 included in the smart factory management system 1000 and the process equipment 350 installed in the factory 300 may vary depending on the type of manufacturing, and the process processes performed within the process equipment 350 are also diverse. can do. For example, in the case of the factory 300 corresponding to the machinery field, semiconductor equipment and parts, precision equipment, heavy equipment parts, mechanical parts, copper rolling, extrusion and stretching products, machine tools, semiconductor manufacturing machines, welding machines, and imaging devices It may include a process facility 350 for manufacturing, and in the case of the factory 300 corresponding to the electric and electronic field, electric power, automation business, transformers, heavy electric parts, precision optical parts and materials, electronic parts, batteries, power equipment It may include a process facility 350 for manufacturing a product, and in the case of the factory 300 corresponding to the steel/metal field, iron making, steel making, rolling and steel materials, metal surface treatment steel, structural metal plate products/workpieces, It may include a process facility 350 for manufacturing the metal assembly structure, and in the case of the factory 300 corresponding to the automobile field, automobile parts, automobile engines, gears, power transmission devices, steering devices, engine pistons, tire valves , nozzles, bottom manufacturing, bands, industrial plastics, production line parts, injection molding molds, vacuum pump machines, process equipment 350 for manufacturing industrial non-curing rubber, and factories belonging to various other fields (300 ), medical devices, microscopes, food hygiene paper boxes, containers, general paints, textile and fabric dyeing processing, cosmetics, shochu, food manufacturing, basic inorganic chemicals, golf equipment, yogurt and dairy products, etc. It may include a process facility 350 for

The factory 300 and the process equipment 350 described above are not clustered and may be installed in a distributed manner, and the cloud server 100 divides the factory 300 and the process equipment 350 by field, and then the field server 200 ), by substituting the industrial data received from the algorithm for each field, the algorithm for the factory 300 and the process equipment 350 can be managed.

So far, the configuration of the smart factory management system 1000 according to an embodiment of the present invention has been schematically described, and the cloud server 100 capable of managing algorithms applicable to the factory 300 and the process equipment 350 below. ) to be explained.

2 is a block diagram illustrating a configuration of a cloud server according to an embodiment of the present invention.

Referring to FIG. 2 , the cloud server 100 may include a storage unit 110 , a communication unit 120 , and a processor 130 .

The storage unit 110 may store industrial data received from the field server 200 through the communication unit 120 . For example, the industry data may include quality data of a manufactured product obtained through a process process and manufacturing data of the process facility 350 . More specifically, the quality data may include an image of a product produced through any one process process and a result of determining whether the product is pass/fail, and the manufacturing data is a programmable logic controller (PLC) control that can be measured in the process facility 350 . It can include data, noise, vibration, temperature, humidity, dust, uptime and number of products produced.

For another example, the industrial data may include unstructured industrial data, large-capacity industrial data, or industrial data to be collected for a long period of time, and the storage unit 110 may classify the data by type and store it.

In addition, the storage unit 110 may store algorithms and control data specified in advance in the field server 200 , and various types of defects/defects that may occur in the process process according to the factory 300 and the process equipment 350 . can be defined For example, the storage unit 110 stores information about the position and background of the surface teeth, screws, sockets, and holes in relation to the field of machinery for manufacturing the rack gear, and occurs due to physical impact during the manufacturing process of the rack gear. Defect/defect information such as scratches, air bubbles, dents/pressed, foreign materials/chips, etc. can be stored.

As such, the information stored by the storage unit 110 learns this when receiving industrial data (quality data, manufacturing data) from the field server 200 that controls the other factories 300 and process equipment 350, and learns each It may be utilized in the process of analyzing an algorithm suitable for the factory 300 or the process facility 350 .

The storage unit 110 may store any one or more of an application program, data, and commands required for an operation of providing a failure determination/predictive maintenance algorithm according to an embodiment of the present invention. According to an embodiment, the storage unit 110 is implemented as various storage devices such as ROM, RAM, EPROM, flash drive, hard drive, etc., or web storage ( web storage), network storage storage, cloud, and block chain database may be implemented as at least one type.

The communication unit 120 may send and receive data to and from the field server 200 using a wired/wireless communication network. Specifically, the communication unit 120 may receive industrial data (quality data, manufacturing data) of the process equipment 350 installed at the manufacturing site from the field server 200 .

According to an embodiment, the communication unit 120 may periodically receive quality data from the on-site server 200 based on the time when any one product is produced as the process progresses. Specifically, the quality data may include an image of a product produced through a process process and a pass/fail determination result of the corresponding product determined by the on-site server 200 . In addition, the quality data may include long-term quality data collected for a predetermined period that the on-site server 200 cannot immediately respond to by itself or large-capacity quality data exceeding a specified size.

According to another embodiment, the communication unit 120 may periodically receive manufacturing data from the field server 200 based on the time when the process facility 350 is installed. That is, the communication unit 120 may periodically receive manufacturing data from the time of operation of the process facility 350 in order to prevent the deterioration of the quality of the manufactured product due to the aging of the process facility 350 . In addition, the communication unit 120 may receive the type and operating conditions of the factory equipment 350 from the on-site server 200 to obtain specific data for predicting the inspection period of the process equipment 350 . In addition, as mentioned above, the manufacturing data may also include long-term manufacturing data collected for a predetermined period that the on-site server 200 cannot immediately respond to, or large-capacity manufacturing data exceeding a specified size.

The processor 130 is operatively connected to the storage 110 and the communication unit 120 to control the overall operation of the cloud server 100, and drives an application or program stored in the storage 110 to operate the factory ( 300), the process equipment 350, and various commands for managing the failure determination/predictive maintenance algorithm applied to each process process may be performed.

The processor 130 is ASICs (Application Specific Integrated Circuit), DSPs (Digital Signal processors), DSPDs (Digital Signal Processing Devices), PLD (Programmable Logic Devices), FPGAs (Field Programmable Gate Arrays), controllers (Controllers), microcontroller (Micro-controllers) and microprocessors (Microprocessors) may be implemented as any one.

The processor 130 may manage an algorithm for each process facility 350 and each process process so that a plurality of factories 300 and process facilities 350 belonging to the manufacturing industry can be operated without a manager, and based on the following algorithm, the processor 130 ) to describe the functions it performs.

[Bad Decision Algorithm]

According to an embodiment, the processor 130 substitutes the quality data received from the on-site server 200 into a plurality of pre-stored defect determination algorithms, so that the on-site server 200 maintains the quality of the product for each process process. You can monitor whether you are using .

However, in order for the processor 130 to monitor the algorithm, a process of defining the defective type of the product may be preceded. That is, since various types of defects exist even for one product, and various quality control conditions may be applied accordingly, the processor 130 may manage a defect determination algorithm for each type of defect occurring in the product.

The processor 130 transmits at least one of the type of process equipment 350 according to the process process, operating conditions, and standard images and attributes of products manufactured through the communication unit 120 together with quality data from the field server 200 The included process data may be received, and a defective type of a product that may be generated in the process process/process facility 350 may be set based on the process data.

If the defect type is set, the processor 130 may calculate pass/fail determination accuracy of the product generated through the process process by substituting the quality data into a plurality of pre-stored defect determination algorithms, and apply to the process process based on the pass/fail determination accuracy Possible failure determination algorithms can be selected. For example, the processor 130 may calculate pass/fail determination accuracy through a plurality of pre-stored defect determination algorithms in relation to each case of a dent defect and a polishing defect that may occur in a mechanical product.

Here, if the factory 300 is newly established and a new process process is operated, an algorithm for determining a product defect may not be specified in the field server 200 . Accordingly, the processor 130 may select a failure determination algorithm having the highest pass/fail determination accuracy for each defective type as an algorithm applicable to the process process.

Contrary to this, if the factory 300 operates a previously registered process process, the processor 130 pays for the product produced through the process process for each defective type of the product from the on-site server 200 through the communication unit 120 . The determination accuracy may be received, and the processor 130 may compare it with the previously calculated pass/fail determination accuracy.

That is, if the difference between the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through the plurality of failure determination algorithms is within a preset range, the algorithm currently applied to the process process by the on-site server 200 is an appropriate algorithm, and accordingly, the product It can be understood that the pass/fail determination is being performed accurately, and the processor 130 may maintain the pass/fail determination criteria for the corresponding defect type.

Conversely, if the difference between the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through a plurality of failure determination algorithms is out of a preset range, the algorithm currently applied by the on-site server 200 to the process process is an inappropriate algorithm, and Accordingly, it may be understood that the product pass/fail determination is not properly performed, and the processor 130 may newly select a failure determination algorithm for the corresponding failure type. For example, the processor 130 may select an algorithm having the highest pass/fail determination accuracy as a new failure determination algorithm for the corresponding defect type in the same manner as in the method of applying the failure determination algorithm to the new process process.

Meanwhile, a preset range for comparison determination may be variably set by an operator who manages the cloud server 100 .

In addition, the processor 130 may evaluate the performance of a plurality of failure determination algorithms for each type of failure through Receiver Operating Characteristics (ROC), and may selectively use the algorithm according to the evaluation result. For example, in relation to the type A of the defect, assuming that the pre-stored failure determination algorithms a, b, c, and d exist in the storage unit 110 , the processor 130 performs each algorithm a, b, c, and d. The algorithm with the highest AUC (Area under the curve) value in the ROC Curve can be selected as a failure determination algorithm to be applied to the new factory 300 or as a failure determination algorithm to be compared with the failure determination algorithm applied to the existing factory 300 . . As such, the processor 130 selects an algorithm having the highest AUC value of the ROC curve among a plurality of pre-stored failure determination algorithms, so that the factory 300 and the process equipment 350 are operated through an algorithm with optimized performance. can [Predictive maintenance algorithm]

As described above, while the processor 130 manages the failure determination algorithm, at the same time, the processor 130 may monitor whether a failure has occurred in the process equipment 350 . Specifically, the processor 130 receives the defective rate of the product calculated according to the defect determination algorithm applied to the process process from the field server 200 through the communication unit 120, and monitors whether the received defective rate is greater than or equal to a preset value, It may be determined whether a failure of the process equipment 350 has occurred.

That is, if the defective rate of the product is greater than or equal to a certain value even though the appropriate algorithm is provided to the on-site server 200 , the processor 130 may determine that a problem has occurred in the process equipment 350 . Accordingly, the processor 130 receives the manufacturing data for checking at least one process facility 350 included in the corresponding process process from the field server 200 through the communication unit 120 , and a plurality of pre-stored predictive maintenance Whether to change the inspection period of the process equipment 350 may be determined using the algorithm.

According to an embodiment, the processor 130 substitutes the manufacturing data of the process equipment 350 received from the on-site server 200 into a plurality of pre-stored predictive maintenance algorithms, so that the on-site server 200 controls the process equipment 350 . You can monitor that you are using the right algorithms to meet your maintenance intervals.

In this case, the processor 130 may additionally check the operating conditions of the process facility 350 in order to determine whether an appropriate algorithm is applied to the process facility 350 . That is, the processor 130 may check the type and operating conditions of the process equipment 350 received by the communication unit 120 , and obtain the inspection period prediction result of the process equipment 350 for each of a plurality of predictive maintenance algorithms. Here, the inspection cycle prediction result may be derived based on any one of the lifetime of the process equipment 350, the probability of occurrence of a failure (%), and the production efficiency of the manufactured product.

Here, if the process equipment 350 is the new process equipment 350, data for predicting the inspection period may not be secured. Accordingly, the processor 130 calculates the average value of the inspection cycle prediction results derived for each of the plurality of predictive maintenance algorithms, and uses the predictive maintenance algorithm with the smallest difference from the average value as the predictive maintenance algorithm applicable to the process facility 350 . can be selected

On the other hand, if the process equipment 350 is the previously installed process equipment 350 , the processor 130 receives control data for the process equipment 350 or the on-site server from the on-site server 200 through the communication unit 120 . 200 may receive an inspection period prediction date predicted based on the stored predictive maintenance algorithm, and compare it with the previously calculated inspection period prediction result.

That is, if the difference between the received inspection period prediction date and the average value of the inspection period prediction results obtained through a plurality of predictive maintenance algorithms is within a preset range, the algorithm currently applied by the on-site server 200 to the process equipment 350 As this is an appropriate predictive maintenance algorithm, it can be understood that the process equipment 350 is managed according to an inspection cycle capable of maintaining product quality and production efficiency. ) can be maintained. Here, the control information may include a predictive maintenance algorithm currently specified in the on-site server 200 or an inspection period set by an administrator.

Conversely, if the difference between the received inspection period prediction date and the average value of the inspection period prediction results obtained through a plurality of predictive maintenance algorithms is out of a preset range, the current on-site server 200 is applied to the process equipment 350 It can be understood that the selected algorithm is an inappropriate algorithm, and accordingly, a decrease in product quality and production efficiency through the process equipment 350 may occur, and the processor 130 performs predictive maintenance applicable to the process equipment 350 Algorithms can be newly selected. For example, in the same manner as in the method of applying the predictive maintenance algorithm to the new process equipment 350 , the processor 130 uses an algorithm with the smallest difference from the average value of the inspection cycle prediction result for the new process equipment 350 . It can be selected by predictive maintenance algorithm.

Meanwhile, a preset range for comparison determination may be variably set by an operator who manages the cloud server 100 .

In addition, the processor 130 may evaluate the performance of the predictive maintenance algorithm for each process facility 350 through Receiver Operating Characteristics (ROC), and may selectively use the algorithm according to the evaluation result. For example, in relation to process facility B, assuming that predictive maintenance algorithms z, y, x, w already stored in the storage unit 110 exist, the processor 130 is configured for each z, y, x, w algorithm. The algorithm with the highest AUC (Area under the curve) value in the ROC Curve can be selected as the predictive maintenance algorithm to be applied to the new plant 300 or the predictive maintenance algorithm to be compared with the predictive maintenance algorithm applied to the existing plant equipment 350. have. As such, the processor 130 selects an algorithm having the highest AUC value of the ROC curve among a plurality of pre-stored predictive maintenance algorithms, so that the factory 300 and the process equipment 350 are operated through an algorithm with optimized performance. can

On the other hand, the processor 130 estimates the potential risk that a failure may occur 1) based on the industrial data (quality data, manufacturing data) collected in real time from the factory 300 or process facility 350 that produces the product, 2 ) The algorithm can be updated by analyzing the correlation of factors for each plant 300 or process facility 350 according to the potential risk level, and 3) analyzing the root cause based on the analysis result. Specifically, the processor 130 may obtain a change characteristic graph of industrial data in consideration of time series change characteristics for each factory 300 or process facility 350 , and then the change characteristic graph monitored by the processor 130 is gradually In the case of an increasing form, the potential risk level is estimated as “risk”, and when the change characteristic graph monitored by the processor 130 maintains the rising value, the potential risk level is estimated as “significant”, and the processor ( 130), the potential risk level can be estimated as “severe” if the peak value repeatedly appears in the graph of the change characteristics. Here, the potential risk level may be designated in the order of risk>severe>warning>significant.

That is, the processor 130 acquires a change characteristic graph as a reference for estimating the potential risk based on the existing industrial data, and when the amount of change of the resource (industrial data) collected and monitored in real time is greater than or equal to the reference level (eg. serious), the correlation between the variables for each plant 300 or process equipment 350 may be analyzed. For example, in any one process facility 350, variables are CPU, Memory, Net Traffic, Event Message, Sys Message, and Net Message, and industrial data obtained from the variables are performance log, event log, system log, and network log. can be

The processor 130 may select the variable most highly correlated with the current potential risk based on the case of the correlation pre-stored in the storage unit 110 , and may take action to solve the selected variable. As an example, the processor 130 may control to transmit variable information highly related to the potential risk along with a notification for an emergency action to an administrator who manages the on-site server 200 through the communication unit 120 .

If there is no case of the correlation pre-stored in the storage unit 110 , the processor 130 may perform a root cause analysis. Specifically, the processor 130 collects continuous measurement values and frequency of occurrence of industry data within the variable group for the variable that has increased the potential risk level, and selects a factor with a high continuous measurement value or a frequent occurrence as the root cause, and , it can be used to update the algorithm. Here, the variable group means a combination of explanatory variables that can explain the variable. For example, when the variable is Memory, description variables such as Svr_Mem5, Net_Packet3, Event#2321MsgA, Svr_Cpu3_Rate, NEW IP, etc. can form a description group.

In this way, when a specific situation occurs, the processor 130 not only solves the problem shown, but also all of the various industrial data constituting the specific situation in the factory 300 and the process facility 350 are related. , and by learning the algorithm, problems that have not occurred since then can be solved more quickly.

So far, the cloud server 100 according to an embodiment of the present invention has been described. According to the present invention, the cloud server 100 is an algorithm for maintaining the quality of products manufactured for each manufacturing site based on the production, quality, and equipment data of the factories 300, to prevent deterioration and damage of factory equipment By managing the predictive maintenance algorithm, it is possible to build complete automation of the factories 300 in which various variable data are generated according to dynamic changes.

3A to 3D are schematic diagrams for explaining an industrial data analysis interface implemented in a cloud server according to an embodiment of the present invention.

Referring to FIGS. 3A to 3D , an administrator managing the cloud server 100 analyzes industrial data collected from a plurality of field servers 200 through an analysis method (Scatter plot, Numeric Fields, Categorical Fields) suitable for the data type. It can analyze and update the algorithm accordingly.

For example, an administrator who manages the cloud server 100 through the search window 31 as shown in FIG. 3A may search for an existing algorithm, or newly register and change a variable in the algorithm, and set the searched algorithm to another factory. After arbitrarily applied to 300 or process equipment 350, it is possible to monitor whether the process is operated efficiently.

For another example, as shown in FIG. 3B , the manager who manages the cloud server 100 checks the industrial data of the factory 300 or the process equipment 350 to which the algorithm is applied through a scatterplot matrix, and between the variables By discovering outliers that can distort a relationship, or visually examining negative (-) or positive (+) correlations that are supposed to influence each other, you can discover the rationale for improving the algorithm.

For another example, as shown in FIG. 3C , the manager managing the cloud server 100 visualizes the functional relationship of the variables identified in FIG. 3B and checks the index value according to the regression analysis result, between the two variables. relationship can be identified.

As another example, as shown in FIG. 3D , the manager managing the cloud server 100 visually confirms the efficiency prediction value of the factory 300 or the process facility 350 through the variables related to the numerical variables and , it is possible to check the difference value of how the actual factory 300 or process equipment 350 was operated. In this way, by visually checking the efficiency prediction value of the factory 300 or the process equipment 350, the manager can adjust the parameters in the algorithm of the specific plant 300 or the process equipment 350, or select a new algorithm. Also, the manager may provide an analysis service for improving the process efficiency of the factory 300 or the process equipment 350 included in the smart factory management system 1000 through this.

Hereinafter, a method of managing an algorithm for determining a product defect through the above-described cloud server 100 and a method of managing a predictive maintenance algorithm for maintaining the inspection cycle of the process equipment 350 will be described.

4 is a flowchart illustrating a method for managing a failure determination algorithm according to an embodiment of the present invention, and FIG. 5 is a flowchart illustrating step S130 illustrated in FIG. 4 .

Referring to FIG. 4 , the cloud server 100 receives quality data acquired through a process process from the field server 200 ( S110 ). Here, the quality data may include an image of a product produced in the process and a pass/fail determination result of the corresponding product determined by the on-site server 200 .

According to an embodiment, the cloud server 100 may periodically receive quality data of all types and sizes from the on-site server 200 , among them, atypical quality data that has not been processed in the on-site server 200 or Quality data requiring a high amount of computation can be received in real time. For example, the quality data requiring a high amount of computation may be long-term manufacturing data collected for a predetermined period by the on-site server 200 or large-capacity quality data exceeding a specified size.

On the other hand, the period at which the cloud server 100 receives the quality data from the on-site server 200 may be set based on the time when any one product is produced according to the process process, and the cloud server 100 receives the quality data. In accordance with the receiving cycle, it is possible to monitor whether a defect determination algorithm suitable for the process process is applied.

After step S110, the cloud server 100 substitutes the quality data into a plurality of pre-stored failure determination algorithms to calculate pass/fail determination accuracy of the product generated through the process process (S120). In this case, the cloud server 100 may calculate pass/fail determination accuracy according to various types of failures that may occur in the product, rather than calculating the pass/fail determination accuracy according to any one failure determination type related to the product.

That is, the cloud server 100 is a process including at least one of a standard image and attributes of the type of process equipment 350 according to the process process, operating conditions, and standard images and properties of products manufactured through the cloud server 100 together with quality data from the on-site server 200 The data may be received, and a defective type of a product that may be generated in the process process/process facility 350 may be set based on the process data.

If the defect type is set, the cloud server 100 may calculate the pass/fail determination accuracy of the product generated through the process process by substituting the quality data into a plurality of pre-stored failure determination algorithms.

After step S120, the Claus server 100 selects a failure determination algorithm applicable to the process process based on the calculated pass/fail determination accuracy (S130). Here, the process of selecting an algorithm may vary depending on whether the process process is new.

Referring to FIG. 5 , in the case of a new process process, the cloud server 100 may select a defect determination algorithm having the highest accuracy of pass/fail determination for each defective type as an algorithm applicable to the process process of the factory 300 ( S130-1).

Conversely, in the case of the existing process process, the cloud server 100 receives the pass/fail determination accuracy for the product produced through the process process for each defective type of the product from the field server 200 (S130-10), and the received pass/fail It may be compared whether the determination accuracy difference is within a preset range (S130-20). Here, the preset range for the comparison determination may be variably set by an operator who manages the cloud server 100 .

As a result of the comparison in step S130-20, if the difference between the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through the plurality of failure determination algorithms is within a preset range, the algorithm currently applied to the process by the on-site server 200 is It is a suitable algorithm, and accordingly, it can be understood that the product pass/fail determination is being performed accurately, and the cloud server 100 may maintain the pass/fail determination standard for the corresponding defective type (S130-31, yes).

As a result of the comparison in step S130-20, if the difference between the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through the plurality of failure determination algorithms is out of a preset range, the on-site server 200 currently applies to the process process. It can be understood that the algorithm is an inappropriate algorithm, and accordingly, the product pass/fail determination is not performed properly, and the cloud server 100 may newly select a failure determination algorithm for the corresponding defective type (S130-32, no). For example, the cloud server 100 may select an algorithm having the highest pass/fail determination accuracy as a new failure determination algorithm for the corresponding defect type in the same manner as in the method of applying the failure determination algorithm to the new process process.

After steps S130-1, S130-31, and S130-32, the cloud server 100 provides the selected failure determination algorithm to the field server 200 (S140). That is, finally, the cloud server 100 is based on a vast amount of quality data and algorithms, so that the on-site server 200 can use an appropriate algorithm in the factory 300 and factory equipment 350 where various variables occur. can supervise.

In addition, while the cloud server 100 manages the failure determination algorithm, at the same time, the cloud server 100 may monitor whether a failure has occurred in the process equipment 350 . Specifically, the cloud server 100 receives the defective rate of the product calculated according to the defect determination algorithm applied to the process process from the on-site server 200, and monitors whether the received defective rate is greater than or equal to a preset value, thereby processing equipment 350 It is possible to determine whether a malfunction has occurred. That is, if the defective rate of the product is greater than or equal to a certain value even though an appropriate algorithm is provided to the on-site server 200 , the cloud server 100 may determine that a problem has occurred in the process equipment 350 , and accordingly, the cloud server 100 ) may manage the predictive maintenance algorithm of the process facility 350 based on the manufacturing data.

6 is a flowchart illustrating a method for managing a predictive maintenance algorithm according to an embodiment of the present invention, and FIG. 7 is a flowchart illustrating step S230 illustrated in FIG. 6 .

Referring to FIG. 6 , the cloud server 100 receives manufacturing data of the process equipment 350 from the field server 200 ( S210 ). Here, the manufacturing data may include programmable logic controller (PLC) control data measurable in process equipment, noise, vibration, temperature, humidity, dust, operating time, and the number of products produced.

According to an embodiment, the cloud server 100 may periodically receive manufacturing data of all types and sizes from the field server 200 , among them, atypical manufacturing data that has not been processed in the field server 200 or Manufacturing data requiring a high amount of computation can be received in real time.

On the other hand, the period in which the cloud server 100 receives the manufacturing data from the on-site server 200 may be set based on the time when the process equipment 350 is installed, and the cloud server 100 is in the period of receiving the quality data. Accordingly, it is possible to monitor whether a failure determination algorithm suitable for the process process is applied.

After step S210 , the cloud server 100 substitutes the manufacturing data into a plurality of pre-stored predictive maintenance algorithms to obtain the inspection cycle prediction result of the process equipment 350 ( S220 ). At this time, the cloud server 100 may check the type and operating conditions of the process equipment 350 through the on-site server 200 , and obtain the inspection cycle prediction results of the process equipment 350 for each predictive maintenance algorithm. have. Here, the inspection cycle prediction result may be derived based on any one of the lifetime of the process equipment 350, the probability of occurrence of a failure (%), and the production efficiency of the manufactured product.

After step S220, the Klaus server 100 selects a failure determination algorithm applicable to the process equipment 350 based on the inspection cycle prediction result (S230). Here, the process of selecting an algorithm may vary depending on whether the process facility 350 is new.

Referring to FIG. 7 , in the case of the new process facility 350 , the Klaus server 100 calculates an average maintenance cycle value based on the maintenance cycle prediction result, and provides a predictive maintenance algorithm with the smallest difference from the calculated average value. It can be selected as a predictive maintenance algorithm applicable to the facility (S230-1).

Conversely, in the case of the existing process equipment 350 , the cloud server 100 predicts based on the predictive maintenance algorithm in which the on-site server 200 or control data for the process equipment 350 from the on-site server 200 is stored. One inspection period prediction date may be received (S230-10), and it may be compared with the previously calculated inspection period prediction result (S230-20). Here, the preset range for the comparison determination may be variably set by an operator who manages the cloud server 100 .

As a result of the comparison in step S230-20, if the difference between the received inspection period prediction date and the average value of the inspection period prediction results obtained through a plurality of predictive maintenance algorithms is within a preset range, the current on-site server 200 operates the process facility It can be understood that the algorithm applied to 350 is an appropriate predictive maintenance algorithm, and accordingly, the process equipment 350 is managed according to an inspection cycle that can maintain product quality and production efficiency, and the cloud server ( 100) may maintain control information on the process equipment 350 (S230-31, yes).

As a result of the comparison in step S230-20, if the difference between the received inspection period prediction date and the average value of the inspection period prediction results obtained through a plurality of predictive maintenance algorithms is out of a preset range, the current on-site server 200 is It can be understood that the algorithm applied to the process equipment 350 is an inappropriate algorithm, and accordingly, the quality and production efficiency of the product through the process equipment 350 may decrease, and the cloud server 100 is the corresponding process equipment. A predictive maintenance algorithm applicable to 350 may be newly selected (S230-32, No).

After steps S230-1, S230-31, and S230-32, the cloud server 100 provides the selected predictive maintenance algorithm to the field server 200 (S240). That is, finally, the cloud server 100 is based on a vast amount of manufacturing data, control data, and algorithms, and in the factory 300 and factory equipment 350 where various variables occur, the on-site server 200 executes an appropriate algorithm. It can be managed and supervised so that it can be used.

Meanwhile, the cloud server 100 may periodically update the algorithm (after steps S140 and S240) in the process of managing and supervising the failure determination algorithm and the predictive maintenance algorithm. According to an embodiment, the cloud server 100 evaluates the performance of a plurality of defect determination algorithms for each defect type through Receiver Operating Characteristics (ROC), or evaluates the performance of the predictive maintenance algorithm for each process facility 350, and the evaluation result Algorithms can be selectively used according to

In addition, according to another embodiment, in the process in which the cloud server 100 manages and supervises the defect determination algorithm and the predictive maintenance algorithm, the factory 300 or process facility ( 350) can be collected in real time and industrial data (quality data, manufacturing data) can be collected in real time, based on this 1) estimating the potential risk of failure, and 2) depending on the potential risk of the plant 300 or process facility ( 350) It is possible to update the algorithm by analyzing the correlation of star elements and 3) analyzing the root cause based on the analysis result.

So far, an algorithm management method according to an embodiment of the present invention has been described. According to the present invention, the cloud server stores and analyzes industrial data and operation/control data generated at the manufacturing site, forms big data for factory automation operation, and manages production, quality, equipment, and energy data of factories. You can build a smart factory that can

Accordingly, by newly implementing factory automation, factory automation can be implemented using the cloud server 100 even at new sites where there are no countermeasures for various problem situations, and a large amount of industrial data is generated due to many types of production facilities. Even in the field where the cloud server 100 processes data quickly, efficient operation of the factory may be possible.

Although one embodiment of the present invention has been described in more detail with reference to the accompanying drawings, the present invention is not necessarily limited to these embodiments, and various modifications may be made within the scope without departing from the technical spirit of the present invention. . Therefore, the embodiments disclosed in the present invention are not intended to limit the technical spirit of the present invention, but to explain, and the scope of the technical spirit of the present invention is not limited by these embodiments. Therefore, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. The protection scope of the present invention should be construed by the following claims, and all technical ideas within the scope equivalent thereto should be construed as being included in the scope of the present invention.

1000: smart factory management system
100: cloud server
110: storage unit
120: communication department
130: processor
200: field server
300: factory
350: process equipment

Claims (17)

storage;
communication department;
a processor operatively connected to the storage unit and the communication unit; including,
The processor is
Receive quality data acquired in the process process from the field server disposed at the manufacturing site through the communication unit,
Receives process data including the type of process process, operating conditions, standard images and attributes of products manufactured through the process process from the on-site server through the communication unit,
Based on the process data, the type of product defect that can occur in the process process is set,
A plurality of failure determination algorithms pre-stored for each defective type of the received quality data - The plurality of failure determination algorithms are learned based on industrial data received in real time from other field servers that control a plurality of factories or process equipment- By substituting in to calculate the pass/fail judgment accuracy for each defective type of the product generated through the process process,
Selecting a failure determination algorithm applicable to the process process based on the pass/fail determination accuracy,
When the process process is a new process process, a defect determination algorithm having the highest pass/fail determination accuracy for each defective type of the product is selected as an algorithm applicable to the process process, and when the process process is a previously registered process process , by receiving the pass/fail determination accuracy for the product produced in the process process from the on-site server for each defective type of the product through the communication unit, the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through the plurality of failure determination algorithms If the difference is within a preset range, the pass/fail decision standard for the corresponding defect type is maintained, but the difference between the received pass/fail decision accuracy and the pass/fail decision accuracy calculated through the plurality of reject decision algorithms is within the preset range. If it does, a new defect determination algorithm for the type of defect is newly selected,
The processor is
Selects a failure determination algorithm applicable to the process process based on the pass/fail determination accuracy, and provides the selected failure determination algorithm to the on-site server,
AUC (Receiver Operating Characteristics) in a plurality of defect determination algorithms that are different for each defective type by checking the defective types of products that can occur in the process process according to the type of process process received from the on-site server The algorithm with the highest Area Under the Curve) value is selected as the defect determination algorithm applicable to the process process,
Receives the defective rate of the product calculated according to the defect determination algorithm applied to the process process from the on-site server through the communication unit, and monitors whether the received defective rate is greater than or equal to a preset value,
When the received defective rate is greater than or equal to a preset value according to the monitoring result, manufacturing data for checking the process equipment related to the process process is received from the on-site server through the communication unit, and the manufacturing data is stored in a plurality of predictive maintenance By substituting into the algorithm, the inspection cycle prediction result of the process equipment is obtained,
When the process facility is a new process facility, the maintenance cycle average value is calculated based on the maintenance cycle prediction result, and a predictive maintenance algorithm with the smallest difference from the calculated maintenance cycle average value among the plurality of pre-stored predictive maintenance algorithms is used. Selected as a predictive maintenance algorithm applicable to the process equipment, or
When the process equipment is an existing process equipment, the calculated average value of the inspection period and the inspection period predicted date received from the on-site server-the inspection period predicted date are predicted based on the manufacturing data for the process equipment or the on-site Determining whether to select a new predictive maintenance algorithm applicable to the process equipment according to whether the difference from - predicted based on the predictive maintenance algorithm stored in the server is within a preset range;
A cloud server that provides the selected predictive maintenance algorithm to the on-site server, and determines whether to change the inspection period of the process equipment based on the selected predictive maintenance algorithm. delete delete delete According to claim 1,
The processor is
A cloud server that receives, through the communication unit, long-term quality data collected by the on-site server for a predetermined period or large-capacity quality data exceeding a specified size. 6. The method of claim 5,
The quality data is
A cloud server that includes an image of a product produced through any one process process and a result of determining whether or not the product is pre-stored in the on-site server. delete According to claim 1,
The communication unit,
A cloud server that periodically receives from the on-site server based on the time when any one product is produced according to the process process. As a method for a cloud server to manage a defect determination algorithm for each process of a manufacturing product,
receiving quality data acquired in the process process from a field server disposed at a manufacturing site;
receiving process data including a type of the process process, operating conditions, and standard images and attributes of products manufactured through the process process from the on-site server;
setting a defective type of a product that may be generated in the process process based on the process data;
A plurality of failure determination algorithms pre-stored for each defective type of the received quality data-The plurality of failure determination algorithms are learned based on industrial data collected in real time from other field servers that control a plurality of factories or process equipment- Calculating pass/fail determination accuracy for each defective type of the product generated through the process process by substituting it;
Selecting a failure determination algorithm applicable to the process process based on the pass/fail determination accuracy,
When the process process is a new process process, selecting a defect determination algorithm with the highest accuracy of pass/fail determination for each defective type of the product as an algorithm applicable to the process process;
If the process process is a previously registered process process, by receiving the pass/fail determination accuracy for the product produced in the process process from the on-site server for each defective type of the product,
If the difference between the received pass/fail decision accuracy and the pass/fail decision accuracy calculated through the plurality of reject decision algorithms is within a preset range, maintain the pass/fail decision standard for the corresponding defect type;
selecting a new failure determination algorithm for the corresponding failure type when a difference between the received pass/fail determination accuracy and the pass/fail determination accuracy calculated through the plurality of failure determination algorithms is out of a preset range; and
providing a selected failure determination algorithm to the on-site server; including,
After the step of providing,
Checking the types of product defects that may occur in the process process according to the process process type received from the on-site server,
Selecting an algorithm with the highest AUC (Area Under the Curve) value of Receiver Operating Characteristics (AUC) as a failure determination algorithm applicable to the process process in a plurality of failure determination algorithms different for each defective type of the product;
receiving the defective rate of the product calculated according to the defect determination algorithm applied to the process process from the on-site server, and monitoring whether the received defective rate is greater than or equal to a preset value; and
When the received defective rate is greater than or equal to a preset value according to the monitoring result, manufacturing data for checking the process equipment related to the process process is received from the on-site server, and the manufacturing data is substituted into a plurality of pre-stored predictive maintenance algorithms. , obtaining a prediction result of the inspection cycle of the process equipment;
When the process facility is a new process facility, the maintenance cycle average value is calculated based on the maintenance cycle prediction result, and a predictive maintenance algorithm with the smallest difference from the calculated maintenance cycle average value among the plurality of pre-stored predictive maintenance algorithms is used. Selected as a predictive maintenance algorithm applicable to the process equipment, or
When the process equipment is an existing process equipment, the calculated average value of the inspection period and the inspection period predicted date received from the on-site server-the inspection period forecast date are predicted based on the manufacturing data for the process equipment, or Determining whether to select a new predictive maintenance algorithm applicable to the process equipment according to whether the difference from - predicted based on the predictive maintenance algorithm stored in the server is within a preset range;
providing the selected predictive maintenance algorithm to the on-site server, and determining whether to change the inspection cycle of the process equipment based on the selected predictive maintenance algorithm; delete delete delete 10. The method of claim 9,
Receiving the quality data comprises:
Receiving the long-term quality data collected for a predetermined period by the on-site server or large-capacity quality data exceeding a specified size, the method for managing a manufacturing product defect determination algorithm. 14. The method of claim 13,
The quality data is
A manufacturing product defect determination algorithm management method, including an image of a product produced in the process process and a pass/fail determination result of the product stored in advance in the on-site server. delete 10. The method of claim 9,
Receiving the quality data comprises:
A method of managing a manufacturing product defect determination algorithm, which is a step of periodically receiving from the on-site server based on the time when any one product is produced according to the process process. an on-site server configured to determine a defect of a manufactured product for each process process of the manufactured product; and
Receive quality data for any one manufactured product from the field server, and receive process data including the type of process process, operating conditions, and standard images and attributes of products manufactured through the process process from the field server and, based on the process data, sets the defective types of products that can occur in the process process, and sets the received quality data for each defective type of product. It is learned based on industrial data received in real time from other field servers that control factories or process equipment, and calculates the pass/fail judgment accuracy for each defective type of product created through the process process, and calculates the pass/fail judgment accuracy. A defect determination algorithm applicable to the process process is selected based on the process process, but when the process process is a new process process, the defect determination algorithm with the highest pass/fail determination accuracy for each defective type of the product is selected as an algorithm applicable to the process process selected, and if the process process is a previously registered process process, by receiving the pass/fail decision accuracy for the product produced in the process process from the on-site server for each defective type of the product, the received pass/fail decision accuracy and the plurality of If the difference from the pass/fail determination accuracy calculated through the failure determination algorithm of a cloud server configured to newly select a failure determination algorithm for a corresponding failure type when the difference from the pass/fail determination accuracy is out of a preset range; includes,
The cloud server,
Checking the types of product defects that may occur in the process process according to the process process type received from the on-site server,
It is configured to select an algorithm having the highest AUC (Area Under the Curve) value of Receiver Operating Characteristics as a failure determination algorithm applicable to the process process in a plurality of failure determination algorithms that are different for each defective type of the product,
Receive the defective rate of the product calculated according to the defect determination algorithm applied to the process process from the on-site server, and monitor whether the received defective rate is greater than or equal to a preset value,
When the received defective rate is greater than or equal to a preset value according to the monitoring result, manufacturing data for checking the process equipment related to the process process is received from the on-site server, and the manufacturing data is substituted into a plurality of pre-stored predictive maintenance algorithms. , to obtain the inspection cycle prediction result of the process equipment,
When the process facility is a new process facility, the maintenance cycle average value is calculated based on the maintenance cycle prediction result, and a predictive maintenance algorithm with the smallest difference from the calculated maintenance cycle average value among the plurality of pre-stored predictive maintenance algorithms is used. Selected as a predictive maintenance algorithm applicable to the process equipment, or
When the process equipment is an existing process equipment, the calculated average value of the inspection period and the inspection period predicted date received from the on-site server-the inspection period forecast date are predicted based on the manufacturing data for the process equipment, or Determining whether to select a new predictive maintenance algorithm applicable to the process equipment according to whether the difference from - predicted based on the predictive maintenance algorithm stored in the server is within a preset range;
The smart factory management system, configured to provide the selected predictive maintenance algorithm to the on-site server, and to determine whether to change the inspection cycle of the process equipment based on the selected predictive maintenance algorithm.

Images