WO2022225117A1 - 실시간 이상 탐지를 위한 gnn 기반의 마스터 상태 생성 방법 - Google Patents
실시간 이상 탐지를 위한 gnn 기반의 마스터 상태 생성 방법 Download PDFInfo
- Publication number
- WO2022225117A1 WO2022225117A1 PCT/KR2021/016481 KR2021016481W WO2022225117A1 WO 2022225117 A1 WO2022225117 A1 WO 2022225117A1 KR 2021016481 W KR2021016481 W KR 2021016481W WO 2022225117 A1 WO2022225117 A1 WO 2022225117A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- state
- value
- node
- graph
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000001514 detection method Methods 0.000 title claims description 7
- 239000011159 matrix material Substances 0.000 claims description 16
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000004590 computer program Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000012545 processing Methods 0.000 abstract description 4
- 238000007781 pre-processing Methods 0.000 abstract description 2
- 238000012549 training Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 24
- 230000006870 function Effects 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- YYJNOYZRYGDPNH-MFKUBSTISA-N fenpyroximate Chemical compound C=1C=C(C(=O)OC(C)(C)C)C=CC=1CO/N=C/C=1C(C)=NN(C)C=1OC1=CC=CC=C1 YYJNOYZRYGDPNH-MFKUBSTISA-N 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000001151 other effect Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/05—Programmable logic controllers, e.g. simulating logic interconnections of signals according to ladder diagrams or function charts
- G05B19/056—Programming the PLC
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0275—Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
- G05B23/0281—Quantitative, e.g. mathematical distance; Clustering; Neural networks; Statistical analysis
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1602—Programme controls characterised by the control system, structure, architecture
- B25J9/161—Hardware, e.g. neural networks, fuzzy logic, interfaces, processor
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/05—Programmable logic controllers, e.g. simulating logic interconnections of signals according to ladder diagrams or function charts
- G05B19/058—Safety, monitoring
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/418—Total factory control, i.e. centrally controlling a plurality of machines, e.g. direct or distributed numerical control [DNC], flexible manufacturing systems [FMS], integrated manufacturing systems [IMS] or computer integrated manufacturing [CIM]
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
- G05B23/0254—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model based on a quantitative model, e.g. mathematical relationships between inputs and outputs; functions: observer, Kalman filter, residual calculation, Neural Networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
- G06N3/0455—Auto-encoder networks; Encoder-decoder networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B19/00—Programme-control systems
- G05B19/02—Programme-control systems electric
- G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
- G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
- G05B19/0426—Programming the control sequence
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B2219/00—Program-control systems
- G05B2219/20—Pc systems
- G05B2219/24—Pc safety
- G05B2219/24065—Real time diagnostics
Definitions
- the present invention relates to a method of generating a master state of a facility, and more particularly, to a method of analyzing log data of a facility to find a major state in a repeated cycle.
- PLC Programmable Logic Controller
- PLC control logic code PLC control logic code
- Control logic is defined using the memory address of PLC hardware, and the memory address of PLC hardware is called a contact.
- the automation line is operated by defining the input/output relationship to these contacts and controlling the value of the contact point for each situation.
- PLC control logic has numerous contacts depending on the scale of the automation line. Accordingly, by analyzing the contents of the PLC control logic code between the contacts, the relationship and sequence between the contacts are analyzed to generate a reference master pattern to determine whether the automation line is in the normal operating state, and the abnormal situation of the automation line is detected. Attempts are being made to
- Patent Document 1 Republic of Korea Patent Publication No. 10-1527419
- An object of the present specification is to provide a method for generating a master state based on GNN by looking at the operational state of equipment and processes as a finite state.
- a method for generating graph data according to the present specification for solving the above-described problem includes the steps of: (a) classifying into one state for each section in which a contact value changes in log data expressed in a Gantt chart; (b) identifying a main state in the divided state, adding at least one sensor value for each section corresponding to the main state, and converting the log data into a node matrix according to the order of occurrence of the main state to which the sensor value is added converting to data; and (c) converting the log data into edge index data by defining a connection relationship between the divided states, expressing the divided state as a node, and expressing the connection relationship of the divided state as an edge; may include.
- the step (a) may further include providing an identification characteristic for classifying a state according to the changed contact value.
- step (b) may be a step of counting the number of states having the same identification characteristic and identifying a state having a number greater than or equal to a preset value as a main state.
- the step (b) may further include assigning an identification code to the identified main state in a One Hot Encoding format.
- step (b) when there are two or more sensor values output from one sensor in the section, one representative value may be selected and added.
- connection relationship may be a necessary condition or an exclusive condition.
- a node that does not correspond to the main state is deleted from the node matrix data, and before and after nodes connected to the deleted node are connected to convert the data into edge index data.
- a method for generating a master state according to the present specification for solving the above-described problem is a method for generating a master state by using a plurality of graph data generated according to the method for generating graph data according to the present specification, (a) the plurality of graphs inputting one graph data that has not yet been input among the data into the GNN AutoEncoder as input data; (b) calculating a difference value (hereinafter “analog level loss”) for a sensor value between the reconstruction data output by the GNN AutoEncoder and the input data; (c) calculating the average value of the analog level loss (hereinafter, “node level loss”) for each node, and calculating the average value of the node level loss (hereinafter “graph level loss”) for each graph; (d) re-learning the GNN AutoEncoder using the graph level loss; and (e) repeating steps (a) to (d) when graph data that has not yet been input among the plurality of graph data remains.
- analog level loss a difference value
- the graph data generating method according to the present specification may be implemented as a computer program written to perform each step of the graph data generating method in a computer and recorded in a computer-readable recording medium.
- the master state generating method according to the present specification may be implemented as a computer program recorded in a computer-readable recording medium written to perform each step of the master state generating method in a computer.
- a master state may be generated based on the GNN from log data and control logic.
- the device to be analyzed while the device to be analyzed is controlled, it is analyzed and graphed as a correlation of static and dynamic data flow, and control logic kendo, control logic generation, It has the effect of providing diversified services such as real-time anomaly detection, reproduction, and productivity and quality analysis.
- 1 is a reference diagram of the overall flow of the invention disclosed herein.
- FIG. 2 is a schematic flowchart of a method for generating graph data according to the present specification.
- 3 is a reference diagram for the collection of log data.
- 4 is a reference diagram for state classification and state identification features.
- 5 is a reference diagram for identifying major states.
- 6 and 7 are reference diagrams for adding a sensor value for each section.
- FIG. 9 is a reference diagram illustrating that a connection relationship is defined between adjacent states.
- 10 is a reference diagram for removing a node corresponding to a minor state.
- 11 is a reference diagram for the relationship between log data, node matrix data, edge index data, and graph data.
- FIG. 12 is a reference diagram of a relationship between graph data and a cycle.
- FIG. 14 is a schematic flowchart of a method for generating a master state according to the present specification.
- 15 is a reference diagram of a method for generating a master state according to the present specification.
- PLC Programmable Logic Controller
- basic sequence control replacement of functions such as relays, timers, and counters with semiconductor devices such as ICs and transistors.
- digital operation that is used in programmable memory to perform special functions such as logic, sequence, timer, counter, and operation through digital or analog input/output modules and controls various types of machines or processors. is defined as "the electronic device of
- Log data is a result obtained by collecting PLC contact data at regular intervals. According to the operation of the equipment on the line, the value of the contacts on the PLC related to the operation is changed. Whenever the contact value on PLC changes, a log is collected.
- the log data is data expressed by [contact name, value, time] and is value data of a specific contact point at a corresponding time.
- a cycle means a section in which the contact data is constantly repeated.
- the unit of a cycle may be various, such as a plant, a line, a process, etc.
- 1 is a reference diagram of the overall flow of the invention disclosed herein.
- FIG. 2 is a schematic flowchart of a method for generating graph data according to the present specification.
- the log data collected in step S10 may be expressed as a Gantt chart.
- 3 is a reference diagram for the collection of log data.
- a log is generated in the PLC.
- data expressed in [contact name, value, time] that is, log data can be collected.
- “1” means that the contact is changed from the "off” state to the "on” state
- "0” means that the contact is changed from the "on” state to the "off” state.
- the collected log data may be expressed as a Gantt chart as shown in FIG. 3 .
- each section in which the contact value changes may be divided into one state.
- 4 is a reference diagram for state classification and state identification features.
- the number of states having the same identification characteristic may be counted, and a state having a number greater than or equal to a preset value may be identified as a main state.
- 5 is a reference diagram for identifying major states.
- the identified key state may give an identification code in the form of One Hot Encoding. Through One Hot Encoding, it is possible to classify the main state with a single attribute value, and the machine learning process in the future can be easier.
- step S13 at least one sensor value may be added to each section corresponding to the main state.
- 6 and 7 are reference diagrams for adding a sensor value for each section.
- various sensors such as a voltage sensor and a temperature sensor may be attached to the facility, and a sensing value may be output from each sensor (1).
- the output sensing value can be collected as a log (2). If the sensed value is expressed corresponding to each section according to the output time, it can be expressed as shown in FIG. 6 (3).
- values output from the two sensors “D1000” and “D2000” are illustrated in FIG. 6 , the types and number of sensors may vary.
- two or more sensor values output from one sensor within one section may exist.
- the "D1000” sensor outputs "299, 300, 301" values.
- a representative value eg, average value "300" can be selected (1) and added to the main "State 1" (2).
- the log data may be converted into node matrix data according to the order of occurrence of the main state to which the sensor value is added in step S14.
- each major state corresponds to one node, and each node corresponds to an identification code.
- the node When converting to the node matrix data, the node must maintain the order of the log data.
- connection relationship between the divided states may be defined in step S15.
- the connection relationship may be a necessary condition or an exclusive condition.
- FIG. 9 is a reference diagram illustrating that a connection relationship is defined between adjacent states.
- step S16 the divided state is expressed as a node and the connection relationship of the divided state is expressed as an edge, so that the log data can be converted into data using an edge index.
- states are expressed as graphs connected by edges and edge indexes.
- the edge index also leaves only the node for the main state, and also it is necessary to remove the node corresponding to the state (Minor) that occurs once in a while.
- a node that does not correspond to the main state may be deleted from the node matrix data, and before and after nodes connected to the deleted node may be connected to convert the data into edge index data.
- 10 is a reference diagram for removing a node corresponding to a minor state.
- node No. 5 which does not correspond to the main state, is the target of deletion.
- nodes 3 and 4 enter node 5, and node 5 exits node 6. Therefore, it is possible to delete node 5 and change nodes 3 and 4 to go directly to node 6.
- node 6 is changed to node 5 according to the node order, and node 7 is changed to node 6.
- node 4 goes into node 5, and node 5 has a relationship that goes to node 6 and node 7. Therefore, it is possible to delete node 5 and change node 4 to go out to node 6 and node 7.
- node 6 is changed to node 5 according to the node order, and node 7 is changed to node 6.
- the node of the node matrix data and the node of the edge index data may have a mutually corresponding relationship.
- log data (Raw Data) is converted into node matrix data and edge index data.
- node matrix data and edge index data are combined, they may be expressed as graph data.
- the collected log data (Raw Data) will also repeat a cycle including similar data, and in this case, one graph data may correspond to one cycle.
- FIG. 12 is a reference diagram of a relationship between graph data and a cycle.
- the AutoEncoder may be composed of an encoder and a decoder.
- the encoder compresses the input data (Input 'X') into low-dimensional embeddings (Z), and the decoder converts the compressed low-dimensional embeddings (Z) into high-dimensional data (Z). ) is reconstructed.
- the encoder and decoder are trained in a direction to minimize the difference value (loss) between the two data.
- the difference value (Loss) of the data reconstructed for the major state among the input data will be small, and the difference between the data reconstructed for the minor state (Minor) will be small.
- the value (Loss) will be large. It is possible to learn a model that detects abnormal data included in the data set by using the difference in the size of the difference (loss) of the reconstructed data.
- AutoEncoder can be freely applied by changing the network used for encoder and decoder according to the type of target data.
- CNN Convolution Neural Network
- MLP Multi Layered Perceptron
- the master state generation method according to the present specification is expanded to graph data by using a graph neural network (GNN) for the encoder and the decoder.
- GNN graph neural network
- 15 is a reference diagram of a method for generating a master state according to the present specification.
- one graph data among the plurality of graph data may be input to the GNN AutoEncoder (1 in FIG. 15) as input data.
- step S21 it is possible to calculate a difference value (hereinafter, “analog level loss”) for the sensor value between the reconstruction data (2 in FIG. 15) output by the GNN AutoEncoder and the input data (3 in FIG. 15) .
- analog level loss a difference value for the sensor value between the reconstruction data (2 in FIG. 15) output by the GNN AutoEncoder and the input data (3 in FIG. 15) .
- node level loss the average value of the analog level loss (hereinafter referred to as "node level loss”) is calculated for each node (4 in FIG. 15), and the average value of the node level loss (hereinafter “graph level loss”) is calculated for each graph. (5 in FIG. 15) can be done.
- next step S24 it may be determined whether graph data that has not yet been inputted among the plurality of graph data remains.
- the process can proceed to step S20.
- steps S20 to S24 may be repeatedly executed while inputting one graph data that has not yet been input among the plurality of graph data to the GNN AutoEncoder as input data.
- the GNN AutoEncoder when all of the plurality of graph data is input to the GNN AutoEncoder, the GNN AutoEncoder is in a learning state. Thereafter, it is possible to determine whether the process is abnormal by using the GNN AutoEncoder after the learning has been completed. However, it is necessary to set a reference range, ie, an upper limit, for determining whether there is an abnormality.
- the average ( ⁇ ) and standard deviation ( ⁇ ) of the node level loss are calculated, and the average value
- the sum of the standard deviation values in which the preset parameters are reflected can be set as the master state upper limit standard. For example, when the preset parameter is 1.5, the upper limit value for determining whether the node level loss is abnormal may be “ ⁇ +1.5 ⁇ ”.
- the average and standard deviation of the graph level loss are calculated, and a preset parameter is reflected in the average value
- the sum of the standard deviation values can be set as the master state upper limit standard. For example, when the preset parameter is 1.5, the upper limit value for determining whether the graph level loss is abnormal may be “ ⁇ +1.5 ⁇ ”.
- the artificial neural network trained according to the above description can track not only whether there is an error in the cycle when data of a new cycle is input, but also which contact point or/and which link an error occurs.
- the master pattern generation method and cycle analysis model training method according to the present specification is a technology that converts a machine control language (Low-Level Language) that is difficult for a human to analyze into a language that can be analyzed (High-Level Language), that is, a machine language that is executed It is different from the prior art in that it is an MLP (machine language processing)-based technology that can analyze (a language that controls a machine) with a computer and can be understood by humans.
- MLP machine language processing
- cycle analysis model As the cycle analysis model according to the present specification is used, while the analysis target device is controlled, it is analyzed and graphed with the correlation of static and dynamic data flow, and control logic kendo, control through AI models such as GNN (Graph Neural Network) It has the effect of providing various services such as logic generation, real-time anomaly detection, reproduction, and productivity and quality analysis.
- GNN Graph Neural Network
- the graph data generation method and the master state generation method include a processor, an application-specific integrated circuit (ASIC), another chipset, and logic known in the art to which the present invention pertains to execute the described calculation and various control logic. circuits, registers, communication modems, data processing devices, and the like.
- the processor may be implemented as a set of program modules.
- the program module may be stored in the memory device and executed by the processor.
- the program is, in order for the computer to read the program and execute the methods implemented as a program, C/C++, C#, JAVA, Python, which the processor (CPU) of the computer can read through the device interface of the computer, It may include code coded in a computer language such as machine language. Such code may include functional code related to a function defining functions necessary for executing the methods, etc. can do. In addition, the code may further include additional information necessary for the processor of the computer to execute the functions or code related to memory reference for which location (address address) in the internal or external memory of the computer should be referenced. have.
- the storage medium is not a medium that stores data for a short moment, such as a register, a cache, a memory, etc., but a medium that stores data semi-permanently and can be read by a device.
- examples of the storage medium include, but are not limited to, ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage device. That is, the program may be stored in various recording media on various servers accessible by the computer or in various recording media on the computer of the user.
- the medium may be distributed in a computer system connected by a network, and computer-readable codes may be stored in a distributed manner.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Theoretical Computer Science (AREA)
- Mathematical Physics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Manufacturing & Machinery (AREA)
- Quality & Reliability (AREA)
- Algebra (AREA)
- Pure & Applied Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Fuzzy Systems (AREA)
- Robotics (AREA)
- Mechanical Engineering (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Programmable Controllers (AREA)
Abstract
Description
Claims (12)
- (a) 간트 차트로 표현된 로그 데이터에서 접점값이 변화하는 구간마다 하나의 상태로 구분하는 단계;(b) 상기 구분된 상태에서 주요 상태를 식별하고, 상기 주요 상태에 해당하는 구간마다 적어도 하나 이상의 센서값을 추가하고, 상기 센서값이 추가된 주요 상태의 발생 순서에 따라 상기 로그 데이터를 노드 매트릭스 데이터로 변환하는 단계; 및(c) 상기 구분된 상태 사이의 연결관계를 정의하고, 상기 구분된 상태를 노드로 표현하고 상기 구분된 상태의 연결관계를 에지로 표현하여 상기 로그 데이터를 에지 인덱스로 데이터로 변환하는 단계;를 포함하는 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 1에서,상기 (a) 단계는,변화된 접점값에 따라 상태를 구분하기 위한 식별 특징을 부여하는 것을 더 포함하는 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 2에서,상기 (b) 단계는,상기 동일한 식별 특징을 가진 상태의 개수를 카운팅하고 미리 설정된 값 이상의 개수를 가진 상태를 주요 상태로 식별하는 단계인 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 3에서,상기 (b) 단계는,식별된 주요 상태는 One Hot Encoding 형식으로 식별 코드를 부여하는 것을 더 포함하는 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 1에 있어서,상기 (b) 단계는,구간 내 하나의 센서로부터 출력된 2이상의 센서값이 존재할 때, 하나의 대표값을 선정하여 추가하는 것을 특징으로 하는 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 1에 있어서,상기 연결관계는 필요조건 또는 배타조건인 것을 특징으로 하는 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 1에 있어서,상기 (c) 단계는,상기 노드 매트릭스 데이터에서 주요 상태에 대응하지 않는 노드를 삭제하고, 삭제된 노드와 연결된 전후 노드를 연결하여 에지 인덱스로 데이터로 변환하는 것을 특징으로 하는 이상 상태 탐지를 위한 그래프 데이터 생성 방법.
- 청구항 1 내지 7 중 어느 한 청구항에 따라 생성된 복수의 그래프 데이터들을 이용하여 마스터 상태를 생성하는 방법으로서,(a) 상기 복수의 그래프 데이터들 중 아직 입력되지 않은 하나의 그래프 데이터를 입력 데이터로 GNN AutoEncoder에 입력하는 단계;(b) 상기 GNN AutoEncoder에 의해 출력된 재구성 데이터와 상기 입력 데이터 사이의 센서값에 대한 차이값(이하 "아날로그 레벨 로스")을 산출하는 단계;(c) 각 노드마다 상기 아날로그 레벨 로스의 평균값(이하 "노드 레벨 로스")를 산출하고, 각 그래프마다 상기 노드 레벨 로스의 평균값(이하 "그래프 레벨 로스")를 산출하는 단계;(d) 상기 그래프 레벨 로스를 이용하여 상기 GNN AutoEncoder를 재 학습시키는 단계; 및(e) 상기 복수의 그래프 데이터들 중 아직 입력되지 않은 그래프 데이터가 남아 있을 때, 상기 단계 (a) 내지 단계 (d)를 반복 실행하는 단계;를 포함하는 마스터 상태 생성 방법.
- 청구항 8에 있어서,(f) 상기 복수의 그래프 데이터들이 모두 입력 데이터로 GNN AutoEncoder에 입력되었을 때, 상기 노드 레벨 로스의 평균 및 표준편차를 산출하고, 상기 평균값에 미리 설정된 파라미터가 반영된 표준편차값을 합산한 값을 마스터 상태 상한 기준으로 설정하는 단계;를 더 포함하는 마스터 상태 생성 방법.
- 청구항 8에 있어서,(f) 상기 복수의 그래프 데이터들이 모두 입력 데이터로 GNN AutoEncoder에 입력되었을 때, 상기 그래프 레벨 로스의 평균 및 표준편차를 산출하고, 상기 평균값에 미리 설정된 파라미터가 반영된 표준편차값을 합산한 값을 마스터 상태 상한 기준으로 설정하는 단계;를 더 포함하는 마스터 상태 생성 방법.
- 컴퓨터에서 청구항 1 내지 청구항 7 중 어느 한 청구항에 따른 그래프 데이터 생성 방법의 각 단계들을 수행하도록 작성되어 컴퓨터로 독출 가능한 기록 매체에 기록된 컴퓨터프로그램.
- 컴퓨터에서 청구항 8에 따른 마스터 상태 생성 방법의 각 단계들을 수행하도록 작성되어 컴퓨터로 독출 가능한 기록 매체에 기록된 컴퓨터프로그램.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2023564610A JP2024518749A (ja) | 2021-04-20 | 2021-11-12 | リアルタイム異常検出のためのgnnに基づくマスタ状態の生成方法 |
US17/756,460 US20240168450A1 (en) | 2021-04-20 | 2021-11-12 | Method of generating master state based on graph neural network for real-time anomaly detection |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR1020210050982A KR102471829B1 (ko) | 2021-04-20 | 2021-04-20 | 실시간 이상 탐지를 위한 gnn 기반의 마스터 상태 생성 방법 |
KR10-2021-0050982 | 2021-04-20 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022225117A1 true WO2022225117A1 (ko) | 2022-10-27 |
Family
ID=83722412
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2021/016481 WO2022225117A1 (ko) | 2021-04-20 | 2021-11-12 | 실시간 이상 탐지를 위한 gnn 기반의 마스터 상태 생성 방법 |
Country Status (4)
Country | Link |
---|---|
US (1) | US20240168450A1 (ko) |
JP (1) | JP2024518749A (ko) |
KR (1) | KR102471829B1 (ko) |
WO (1) | WO2022225117A1 (ko) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001356801A (ja) * | 2000-06-13 | 2001-12-26 | Toshiba Corp | プラント監視装置および記憶媒体 |
JP5769138B2 (ja) * | 2013-04-22 | 2015-08-26 | 横河電機株式会社 | イベント解析装置およびコンピュータプログラム |
KR20200074677A (ko) * | 2018-12-17 | 2020-06-25 | 주식회사 유디엠텍 | Plc 제어에 따른 오류 검출 시스템 |
KR20200094514A (ko) * | 2019-01-30 | 2020-08-07 | 경기대학교 산학협력단 | 설비 오류 발생 패턴 추출 장치 및 방법 |
CN111860769A (zh) * | 2020-06-16 | 2020-10-30 | 北京百度网讯科技有限公司 | 预训练图神经网络的方法以及装置 |
CN111966076A (zh) * | 2020-08-11 | 2020-11-20 | 广东工业大学 | 基于有限状态机和图神经网络的故障定位方法 |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101527419B1 (ko) | 2013-09-23 | 2015-06-09 | 주식회사 유디엠텍 | 자동화 라인의 정상 상태 판단을 위한 기준 마스터 패턴 생성 방법 |
-
2021
- 2021-04-20 KR KR1020210050982A patent/KR102471829B1/ko active IP Right Grant
- 2021-11-12 US US17/756,460 patent/US20240168450A1/en active Pending
- 2021-11-12 WO PCT/KR2021/016481 patent/WO2022225117A1/ko active Application Filing
- 2021-11-12 JP JP2023564610A patent/JP2024518749A/ja active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2001356801A (ja) * | 2000-06-13 | 2001-12-26 | Toshiba Corp | プラント監視装置および記憶媒体 |
JP5769138B2 (ja) * | 2013-04-22 | 2015-08-26 | 横河電機株式会社 | イベント解析装置およびコンピュータプログラム |
KR20200074677A (ko) * | 2018-12-17 | 2020-06-25 | 주식회사 유디엠텍 | Plc 제어에 따른 오류 검출 시스템 |
KR20200094514A (ko) * | 2019-01-30 | 2020-08-07 | 경기대학교 산학협력단 | 설비 오류 발생 패턴 추출 장치 및 방법 |
CN111860769A (zh) * | 2020-06-16 | 2020-10-30 | 北京百度网讯科技有限公司 | 预训练图神经网络的方法以及装置 |
CN111966076A (zh) * | 2020-08-11 | 2020-11-20 | 广东工业大学 | 基于有限状态机和图神经网络的故障定位方法 |
Also Published As
Publication number | Publication date |
---|---|
KR102471829B9 (ko) | 2023-02-23 |
KR20220144539A (ko) | 2022-10-27 |
KR102471829B1 (ko) | 2022-11-30 |
JP2024518749A (ja) | 2024-05-02 |
US20240168450A1 (en) | 2024-05-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2019107614A1 (ko) | 제조 공정에서 딥러닝을 활용한 머신 비전 기반 품질검사 방법 및 시스템 | |
WO2018221948A1 (ko) | 빅데이터 분석을 통한 공정 관리 방법 | |
WO2023043215A1 (ko) | 표준 운용 수준 평가 기반 산업 설비 운용 제어 장치 및 그 동작 방법 | |
WO2022114654A1 (ko) | 공정 모니터링 시스템 및 방법 | |
WO2020246662A1 (ko) | 산업 플랜트 설비의 신호 그룹 기반 학습 모델을 이용한 운전 예측 시스템, 방법 및 컴퓨터 프로그램 | |
WO2022114653A1 (ko) | 데이터 경계 도출 시스템 및 방법 | |
WO2023085717A1 (ko) | 클러스터링을 기반으로 하는 레이블링을 수행하기 위한 장치, 이상 탐지를 위한 장치 및 이를 위한 방법 | |
WO2023101165A1 (ko) | 생산 계획 수립 장치 및 방법 | |
WO2022225117A1 (ko) | 실시간 이상 탐지를 위한 gnn 기반의 마스터 상태 생성 방법 | |
US20230027840A1 (en) | Anomaly detecting method in sequence of control segment of automation equipment using graph autoencoder | |
WO2023277274A1 (ko) | Gnn 기반의 plc 제어 로직 자동 검수 방법 | |
WO2020050456A1 (ko) | 설비 데이터의 이상 정도 평가 방법 | |
WO2021006501A1 (ko) | 스마트 재봉작업 측정 방법 및 이를 수행하기 위한 시스템 | |
WO2024034873A1 (ko) | 화학공정의 기저원인 분석 및 이를 이용한 수율 예측 모델링 방법 | |
WO2023128320A1 (ko) | 인공지능 검증 시스템 및 방법 | |
WO2019221461A1 (ko) | 네트워크 장애 원인 분석 장치 및 방법 | |
WO2022139193A1 (ko) | 회전기 이상 진단 방법 및 장치 | |
WO2022030831A1 (ko) | 제어구간 온도신호의 분석을 통한 공정이상 탐지장치 및 방법 | |
WO2021261901A1 (ko) | 함수 호출 패턴 분석을 통한 이상 검출 장치 및 방법 | |
WO2022211301A1 (ko) | 오토인코더 앙상블 기반 이상행위 탐지 방법 및 시스템 | |
WO2022146020A1 (ko) | 회전기계의 결함을 종합적으로 진단하는 방법 및 시스템 | |
WO2021251745A1 (ko) | 인공 신경망 모델 내 은닉단위의 상관성을 이용한 은닉단위 개수 선정방법 및 이를 이용한 수문기후변수 예측방법 | |
KR20230081759A (ko) | 인공지능 기반의 클린룸 관리를 위한 통합 관리 시스템 및 그 방법 | |
WO2021137395A1 (ko) | 심층신경망 알고리즘 기반의 문제 행동 분류 시스템 및 방법 | |
WO2020111317A1 (ko) | 기계학습 기법에 기반한 기계의 오류 데이터를 검출하기 위한 알고리즘 및 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 17756460 Country of ref document: US |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21938033 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2023564610 Country of ref document: JP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 21938033 Country of ref document: EP Kind code of ref document: A1 |