System for Anomaly Diagnosis and Control Parameter Adjustment in Automation Equipment via AI-Based Analysis of PLC / HMI Commissioning Data

Abstract

The present invention relates to an automation equipment abnormality diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis. The system, which analyzes control data collected during the commissioning process of automation equipment to diagnose the cause of abnormalities in the automation equipment and supports the correction of control conditions, comprises: an equipment data collection unit that collects commissioning data including at least one of input signals, output signals, and internal device status of a PLC installed in the automation equipment, alarms and operation history of an HMI, sensor status, actuator operation status, servo motion data, robot operation status, communication status, and interlock conditions; and a normal sequence reference information storage unit that stores normal sequence reference information including at least one of PLC input conditions, PLC output conditions, interlock conditions, sensor input sequence, actuator output sequence, servo position conditions, robot operation conditions, allowable time required for each stage of normal operation of the automation equipment, and HMI alarm occurrence criteria. An AI sequence comparison analysis unit that sorts the above commissioning data in chronological order and compares the actual operation sequence resulting from the sorted commissioning data with the normal operation sequence according to the normal sequence reference information to detect an abnormal section corresponding to at least one of operation delay, signal omission, operation sequence error, repetitive stop, interlock failure, abnormal alarm, or deviation from the set range occurring in the actual operation sequence; and an abnormal cause classification unit that analyzes the correlation between PLC input signals, PLC output signals, internal device status, HMI alarm, interlock condition, sensor status, actuator status, servo motion data, robot operation status, and communication status in the abnormal section, and classifies the cause of the abnormal section into one or more of PLC logic abnormality, sensor signal abnormality, output or wiring abnormality, communication abnormality, servo position abnormality, robot operation standby, interlock condition mismatch, and set parameter abnormality.The system is characterized by comprising: a control condition correction candidate generation unit that generates a control condition correction candidate for at least one of a PLC timer setting value, a sensor input judgment condition, an interlock release condition, a servo position tolerance, a communication response waiting time, an alarm occurrence condition, and an HMI display condition, based on the classified abnormal cause and the normal sequence reference information; and a worker guidance unit configured to output the abnormal section, the classification result of the abnormal cause, the control data that served as the basis for the classification result, the field inspection target, and the control condition correction candidate to an HMI or a worker terminal, and to determine whether to apply the control condition correction candidate according to the worker's approval input.

Description

Technology Field

[0001] The present invention relates to an automation equipment anomaly diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis. Background Technology

[0003] Recently, various automation facilities such as automobile production lines, robotic assembly equipment, logistics transfer equipment, vision inspection equipment, processing equipment, and packaging equipment are being used in the manufacturing industry to improve productivity, stabilize quality, and reduce labor costs.

[0004] These automation facilities may generally include PLC (Programmable Logic Controller), HMI (Human Machine Interface), sensors, actuators, servo drives, industrial robots, vision inspection equipment, safety control devices, and industrial communication devices.

[0005] Based on signals input from sensors and preset control programs, the PLC operates internal relays, timers, counters, comparison commands, interlock conditions, and output coils, thereby controlling the operation of cylinders, motors, servo motors, robots, conveyors, and other actuators.

[0006] The HMI is connected to the PLC to display the operating status, sensor status, actuator status, interlock status, alarm status, production quantity, cycle time, and control parameters of the automation equipment, and can receive operation commands, manual operations, alarm confirmations, parameter changes, and reset inputs from the operator.

[0007] In cases where automation equipment is newly manufactured or installed, or where there is a change in production items, modification of PLC programs, expansion of equipment, replacement of sensors or actuators, replacement of servos or robots, or changes in communication configurations, a commissioning process may be performed to verify whether the equipment operates normally according to the designed sequence and conditions.

[0008] During the commissioning process, verification of PLC input and output, operation of sensors and actuators, manual and automatic operation, interlock conditions, alarm trigger conditions, servo position attainment, robot operation and handshake signals, communication status, and the entire process cycle may be performed.

[0009] However, since the automation equipment operates sequentially with multiple sensors, internal devices, timers, interlock conditions, servo position conditions, robot completion signals, and communication completion signals linked together, if any one condition is not met within a set time, subsequent operations may not be executed or the equipment may stop at an intermediate stage.

[0010] For example, if the servo position arrival signal is not input within the set time, the robot start permission signal may not be generated, and if the robot start permission signal is not generated, an alarm regarding robot operation failure or incomplete process may be displayed on the HMI.

[0011] In this case, the alarm ultimately displayed on the HMI merely indicates that the equipment has stopped, and often fails to directly show whether the actual cause of the malfunction corresponds to servo position deviation, sensor input delay, timer setting value, interlock condition, communication response delay, or PLC internal logic.

[0012] Conventionally, when an equipment malfunction occurred during commissioning, an operator or control engineer would check the alarm displayed on the HMI and manually trace the relevant input contacts, internal relays, timers, output coils, and interlock conditions on the online monitoring screen of the PLC program to identify the cause of the malfunction.

[0013] In addition, a method has been used to restart the equipment by checking the PLC input and output status, HMI alarms, servo status, robot status, communication status, and the actual operation of field devices, respectively, and repeatedly changing timer values, sensor judgment times, servo position tolerances, or communication waiting times.

[0014] However, since this method relies heavily on the experience and proficiency of the worker or engineer, the results of the cause analysis and the corrective measures for the same anomaly may vary depending on the worker.

[0015] Particularly in large-scale automation facilities where multiple PLCs, HMIs, servo drives, robot controllers, and vision inspection equipment are used together, it is difficult to accurately determine which signal was changed first and which signal caused subsequent anomalies, as each device has different data update cycles and time standards.

[0016] While PLC data is updated according to the PLC's scan cycle, HMI alarms can be recorded at the time of alarm occurrence or release, and servo and robot data can be updated separately according to their respective control cycles. Consequently, simply collecting device-specific logs presents a problem in that it is difficult to reconstruct the accurate sequence of events that occurred within a single equipment cycle.

[0017] Furthermore, during the commissioning process, the same equipment may repeatedly operate normally or stop abnormally; however, conventionally, it was often difficult to systematically generate and manage the event sequences and time intervals between events that commonly appear in multiple successfully completed equipment cycles as standard normal sequence information.

[0018] Consequently, it was difficult to automatically determine at which stage the actual equipment operation first deviated from normal operation, what events were missed, whether events were delayed beyond the normal allowable time, or whether the operation sequence was changed.

[0019] Furthermore, if the PLC program, HMI program, electrical drawings, or PLC I / O list are modified, the criteria for normal operation may change. Nevertheless, if the normal sequence corresponding to the program before the change is compared with the actual equipment operation after the change, a difference caused by a normal program modification may be incorrectly judged as an anomaly.

[0020] Conventional equipment monitoring or predictive maintenance technologies have primarily been used to predict the likelihood of failure based on changes in physical quantities such as temperature, vibration, current, pressure, and equipment operating rates.

[0021] However, this technology had limitations in analyzing the logical causal relationships between PLC input signals, output signals, internal device status, interlock conditions, HMI alarms, servo position attainment conditions, robot handshake signals, and communication completion signals that occur during the commissioning process.

[0022] Furthermore, relying solely on simple time-series anomaly detection or alarm analysis, it was difficult to specifically identify the input contact, internal relay, timer completion condition, servo position attainment condition, robot completion signal, or communication completion signal that blocked the occurrence of a specific output event when the event that should have been executed in the normal sequence did not occur.

[0023] In particular, since multiple preconditions and interlock conditions can be connected to a single output event, it is necessary to analyze the logical structure of the PLC program and the relationship between the field device and the PLC I / O address together to identify the cause of the anomaly.

[0024] However, conventional technology has not been sufficiently provided to extract the logical dependency relationships of contacts, internal relays, timers, comparison commands, output coils, and alarm occurrence conditions included in a PLC program, and to analyze the cause of abnormalities by combining them with the event occurrence sequence of actual commissioning data.

[0025] Meanwhile, even when the cause of the anomaly was identified, the process of determining the control conditions to resolve the anomaly often relied on the operator's judgment.

[0026] For example, the extent to which sensor input on-delay values, PLC timer preset values, servo position reach tolerances, robot completion signal waiting times, or communication response waiting times should be adjusted was often determined based on past experience or repeated testing.

[0027] If these control conditions are changed without sufficient verification, even if the current anomaly is resolved, the order of subsequent sequences may be altered or new alarms and repeated stops may occur.

[0028] In addition, if safety interlocks such as emergency stop, safety door, light curtain, motor overload, and collision prevention conditions are included in the correction target or bypassed, there is a risk that the safety of the equipment may be reduced.

[0029] Therefore, when generating control condition correction candidates, it is necessary to exclude safety conditions from the correction target and verify whether the correction candidates actually contribute to the restoration of the normal operation sequence before reflecting them in the actual PLC output.

[0030] However, when testing by directly applying control condition correction candidates to actual automated equipment, incorrect correction values ​​may cause the equipment to operate abnormally or increase the downtime of the production equipment.

[0031] In addition, if multiple control conditions are interrelated, the entire sequence may be incompletely modified if only some control conditions are reflected, and if the PLC program or related interlock logic is changed after the correction value is approved, the correction value at the time of approval may not be suitable for the current program.

[0032] Conventionally, there was insufficient technology to verify correction candidates stepwise by virtually reproducing them using past normal and abnormal cycles without reflecting the correction candidates in the actual output, or by comparing virtual output results based on actual PLC input states with normal output results.

[0033] Furthermore, the technology to check for local changes to the logic subject to correction, to first record the post-correction setting value in a temporary area to verify the read return value, and then to collectively reflect multiple correction values ​​in the actual control area was also insufficient.

[0034] In addition, there was a problem in that a rollback procedure was not systematically provided to monitor normal operation during the re-operation process after calibration and to automatically revert to the control conditions before calibration in the event of a new anomaly.

[0035] Therefore, technology is required to align PLC, HMI, sensor, actuator, servo, robot, and communication data collected during the commissioning process of automation equipment along a common time axis and equipment cycle units, and to detect initial deviation events and abnormal sections by comparing the actual operation sequence with the normal operation sequence.

[0036] In addition, technology is required to analyze the causal relationship between control events based on the PLC program and the PLC I / O list, extract the minimum causal conditions that prevent the occurrence of target output events, and provide the operator with the cause of the abnormality and its basis.

[0037] In addition, there is a need for a technology that can generate control condition correction candidates without changing or bypassing safety conditions, and verify the validity of the correction candidates before actual application through counterfactual commissioning playback and shadow mode verification.

[0038] Furthermore, technology is required to verify whether the program has been modified after correction approval, prevent partial application or data errors that may occur during the recording and application of correction values, and stably restore the state before correction if the re-testing result is abnormal. The problem to be solved

[0040] The present invention aims to solve the problems of the aforementioned prior art by providing an automation equipment anomaly diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis.

[0041] However, the problems that the present invention aims to solve are not limited to those explicitly mentioned above, and other technical problems that a person skilled in the art can recognize from the following description may also be included in the problems that the present invention aims to solve. means of solving the problem

[0043] An automation equipment abnormality diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis according to an embodiment of the present invention is a system that diagnoses the cause of abnormalities in the automation equipment and supports the correction of control conditions by analyzing control data collected during the commissioning process of the automation equipment, comprising: an equipment data collection unit that collects commissioning data including at least one of input signals, output signals and internal device status of a PLC installed in the automation equipment, alarms and operation history of an HMI, sensor status, actuator operation status, servo motion data, robot operation status, communication status, and interlock conditions; and a normal sequence reference information storage unit that stores normal sequence reference information including at least one of PLC input conditions, PLC output conditions, interlock conditions, sensor input sequence, actuator output sequence, servo position conditions, robot operation conditions, allowable time required for each stage of normal operation of the automation equipment, and HMI alarm occurrence criteria. An AI sequence comparison analysis unit that sorts the above commissioning data in chronological order and compares the actual operation sequence resulting from the sorted commissioning data with the normal operation sequence according to the normal sequence reference information to detect an abnormal section corresponding to at least one of operation delay, signal omission, operation sequence error, repetitive stop, interlock failure, abnormal alarm, or deviation from the set range occurring in the actual operation sequence; and an abnormal cause classification unit that analyzes the correlation between PLC input signals, PLC output signals, internal device status, HMI alarm, interlock condition, sensor status, actuator status, servo motion data, robot operation status, and communication status in the abnormal section, and classifies the cause of the abnormal section into one or more of PLC logic abnormality, sensor signal abnormality, output or wiring abnormality, communication abnormality, servo position abnormality, robot operation standby, interlock condition mismatch, and set parameter abnormality.A control condition correction candidate generation unit that generates a control condition correction candidate for at least one of a PLC timer setting value, a sensor input judgment condition, an interlock release condition, a servo position tolerance, a communication response waiting time, an alarm occurrence condition, and an HMI display condition, based on the classified abnormal cause and the normal sequence reference information; and a worker guidance unit configured to output the abnormal section, the classification result of the abnormal cause, the control data that served as the basis for the classification result, the field inspection target, and the control condition correction candidate to an HMI or a worker terminal, and to determine whether to apply the control condition correction candidate according to the worker's approval input.

[0044] In one embodiment, the equipment data collection unit may further include a time synchronization and event conversion unit that converts each data collected from a PLC, HMI, servo drive, robot controller, vision inspection equipment, and upper-level production management system having different communication cycles and time standards into a common time axis, divides each data into the same production cycle unit based on at least one of a PLC scan cycle, an equipment cycle start signal, and an equipment cycle end signal, and converts control data included in each production cycle into event data including signal identification information, signal value, time of occurrence, duration, previous state, state after change, and identification information of a data generating device, and the AI ​​sequence comparison analysis unit may be characterized by generating the actual operation sequence using the event data.

[0045] In one embodiment, the normal sequence reference information storage unit may be characterized by generating a normal sequence candidate using the occurrence order of control events that commonly appear in a plurality of successfully completed commissioning cycles and the time elapsed between events, registering the normal sequence candidate that has received approval input from an operator or engineer as normal sequence reference information, storing the normal sequence reference information in correspondence with at least one of the equipment project name, customer, production line name, equipment name, PLC type, HMI type, PLC program version, HMI program version, electrical drawing version, I / O list version, and commissioning date, and generating a new normal sequence reference information that is distinct from the normal sequence reference information prior to the change when at least one of the PLC program version, HMI program version, electrical drawing version, or I / O list version is changed.

[0046] In one embodiment, the abnormal cause classification unit extracts logical dependency relationships between input contacts, internal relays, timers, counters, comparison commands, output coils, interlock conditions, and alarm occurrence conditions included in a PLC program, and combines the logical dependency relationships with the correspondence relationships between PLC addresses included in a PLC I / O list and field devices to generate an interlock causation graph representing cause-and-effect relationships between multiple control events, searches the interlock causation graph in reverse from the first deviation event where the normal operation sequence and the actual operation sequence first mismatched to extract multiple cause candidates that influenced the occurrence of the first deviation event, calculates cause reliability for each cause candidate using at least two of the following: temporal precedence, frequency of repeated occurrence, signal value deviation relative to the normal cycle, correlation with related HMI alarms, and similarity with past field action history, and outputs the priority of the cause candidates according to the cause reliability, as well as the PLC address, interlock condition, and control event serving as the basis for extracting each cause candidate, through the operator guidance unit.

[0047] In one embodiment, the control condition correction candidate generation unit selects a control condition directly connected to a cause candidate whose cause reliability is greater than or equal to a set reference value as a control condition to be corrected, generates a plurality of control condition correction candidates based on at least one of a set value applied in a past normal commissioning cycle, a set value applied to equipment of the same type, a design tolerance range of the control condition, and a set value stabilized after past field measures for the control condition to be corrected, generates a virtual commissioning sequence to which each control condition correction candidate is applied, evaluates whether the equipment cycle is completed, the time required for each step, whether repeated stops occur, whether an alarm occurs, and whether safety interlocks are satisfied for the virtual commissioning sequence, selects a control condition correction candidate among the plurality of control condition correction candidates that does not change or bypass at least one safety condition among an emergency stop condition, a safety door condition, a light curtain condition, an overload condition, and a collision prevention condition while reducing the difference from the normal operation sequence as an applicable candidate, and generates approval information so that the applicable candidate can be reflected in a PLC or HMI only when an approval input from an operator regarding the applicable candidate is received.

[0048] In one embodiment, the abnormal cause classification unit and the control condition correction candidate generation unit, when an output event that should be executed in the normal operation sequence does not occur in the actual operation sequence, set the unoccurred output event as a target event, perform a reverse search from the output coil or operation command corresponding to the target event in the interlock causation graph to extract at least one of the input contact, internal relay, timer completion condition, counter condition, servo position attainment condition, robot completion signal, communication completion signal, and safety interlock condition that blocked the occurrence of the target event, determine the minimum set of conditions for which the signal transmission path to the target event is restored when the state of one or more of the extracted conditions changes as the minimum blocking cause set, exclude the safety input signal, emergency stop input signal, safety door input signal, light curtain input signal, motor overload input signal, and collision prevention input signal from the conditions included in the minimum blocking cause set from the correction target, and then the PLC timer preset value, sensor input on-delay or off-delay value, servo position attainment tolerance, robot completion signal waiting time, communication response waiting time, and non-safety interlock Generating the above control condition correction candidates limited to release conditions, and for each of the above control condition correction candidates, performing a counter-real-world commissioning replay in chronological order by replaying control events in a state where the above control condition correction candidate is applied using at least three normal commissioning cycles collected prior to the point of anomaly occurrence and at least one commissioning cycle in which anomaly occurred, wherein the result of the above counter-real-world commissioning replay is: a. to generate the above target event; b. to complete subsequent events specified in the above normal operation sequence in a set order; c. to reduce the time required to the above target event or the total equipment cycle time compared to before correction; and d.The control condition correction candidate is selected as a verification candidate only when all of the following conditions are met: not changing the state and signal transmission path of the PLC logic corresponding to the above safety input signal, emergency stop input signal, safety door input signal, light curtain input signal, motor overload input signal, and collision prevention input signal; and the selected verification candidate is executed for at least three consecutive equipment cycles in a shadow mode that reflects it only in a virtual output area without reflecting it in the actual PLC output, thereby verifying whether the output result of the virtual output area based on the input state of the actual PLC and the output result of the normal operation sequence are greater than or equal to a set match rate; and if the match rate is greater than or equal to a set standard, a correction approval package is generated that includes the control condition before correction, the control condition after correction, the address of the PLC to be corrected, the interlock condition affected, the version identification value of the PLC program to be applied, the hash value of the program before correction, and rollback information to return to the state before correction, and is provided to the operator as the final approval target.

[0049] The means for solving the problem described above are merely exemplary and should not be interpreted as intended to limit the present invention. In addition to the exemplary embodiments described above, additional embodiments may exist in the drawings and the detailed description of the invention. Effects of the invention

[0051] According to the present invention, there is an effect of being able to comprehensively analyze commissioning data generated from PLCs, HMIs, sensors, servos, robots, and communication devices.

[0052] In addition, data from different devices can be aligned along a common time axis and in units of equipment cycles to accurately determine the sequence of control events.

[0053] In addition, abnormal sections and the first deviation event can be quickly detected by comparing the actual operation sequence with the normal operation sequence.

[0054] In addition, the cause of the anomaly and the related PLC address can be specifically identified by utilizing the logical dependencies and I / O information of the PLC program.

[0055] In addition, unnecessary changes to control conditions can be reduced by extracting the minimum blocking cause set directly related to the occurrence of anomalies.

[0056] In addition, it is possible to generate control condition correction candidates corresponding to abnormal causes while excluding safety interlocks from the correction target.

[0057] In addition, safety and effectiveness can be verified through counter-testing playback and shadow mode before applying correction candidates to actual equipment.

[0058] In addition, by checking the program hash value and the hash value of the correction influence logic, it is possible to prevent the correction value from being incorrectly applied when the program is modified after approval.

[0059] In addition, the correction values ​​can be verified in a temporary correction area and then collectively reflected in the actual control area, and the system can automatically revert to the state before correction in the event of an anomaly.

[0060] Consequently, the present invention has the effect of improving the accuracy and safety of abnormal diagnosis and control condition correction of automated equipment, and shortening the commissioning period and equipment downtime.

[0061] However, the effects of the present invention are not limited to those explicitly mentioned above, and other effects that a person skilled in the art can recognize from the composition of the present invention may also be included in the effects of the present invention. Brief explanation of the drawing

[0063] FIG. 1 is a block diagram showing the overall configuration of an automation equipment abnormality diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis according to one embodiment of the present invention. FIG. 2 is a conceptual diagram illustrating the process of collecting equipment data, time synchronization, and event conversion according to one embodiment of the present invention. FIG. 3 is a conceptual diagram illustrating the process of comparing a normal operation sequence and an actual operation sequence according to an embodiment of the present invention. FIG. 4 is a conceptual diagram illustrating the process of generating an interlock and graph using a PLC program and PLC I / O information according to one embodiment of the present invention. FIG. 5 is a conceptual diagram illustrating the process of reverse search and determination of the minimum blocking cause set for a target event according to one embodiment of the present invention. FIG. 6 is a conceptual diagram illustrating the process of generating control condition correction candidates and replaying counter-test runs according to one embodiment of the present invention. FIG. 7 is a conceptual diagram illustrating the shadow mode processing process of a correction candidate according to one embodiment of the present invention. FIG. 8 is a flowchart illustrating the worker guidance, correction approval, and rollback processing process according to one embodiment of the present invention. FIG. 9 is a conceptual diagram illustrating the integrity verification and safe application processing of a correction value according to one embodiment of the present invention. Specific details for implementing the invention

[0064] The following detailed description of the invention refers to the accompanying drawings, which illustrate specific embodiments in which the invention may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the invention. It should be understood that various embodiments of the invention are different but need not be mutually exclusive. For example, specific shapes, structures, and characteristics described herein may be implemented in other embodiments without departing from the spirit and scope of the invention in relation to one embodiment.

[0065] Furthermore, it should be understood that the location or arrangement of individual components within each disclosed embodiment may be changed without departing from the spirit and scope of the invention. Accordingly, the following detailed description is not intended to be taken in a limiting sense, and the scope of the invention is limited only by the appended claims, including all equivalents thereof, provided appropriately described. Similar reference numerals in the drawings refer to the same or similar functions across various aspects.

[0066] Meanwhile, throughout this specification, when a part is described as “comprising” a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components. Furthermore, the “part” for a component as used in this specification performs at least one function or operation. And the “part” may perform the function or operation by hardware, software, or a combination of hardware and software.

[0067] Terms including ordinal numbers, such as first, second, etc., may be used to describe various components, but these components are not limited by the aforementioned terms. The aforementioned terms are used solely for the purpose of distinguishing one component from another.

[0068] In this specification, terms such as “comprising” are intended to indicate the existence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof. When a component is referred to as being “connected” to another component, it should be understood that it may be directly connected to or coupled with the other component, or that there may be other components in between.

[0069] Furthermore, in describing the present invention, if it is determined that a detailed description of related known functions or configurations may unnecessarily obscure the essence of the invention, such detailed description is abbreviated or omitted.

[0070] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. However, the embodiments described below are merely examples to aid in understanding the present invention, and the scope of the present invention is not limited thereto. Furthermore, the configurations illustrated in each drawing are schematically depicted for convenience of explanation, and some configurations may be combined, separated, omitted, or added depending on the actual implementation.

[0071] FIG. 1 is a block diagram showing the overall configuration of an automation equipment abnormality diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis according to one embodiment of the present invention.

[0072] FIG. 2 is a conceptual diagram illustrating the process of collecting equipment data, time synchronization, and event conversion according to one embodiment of the present invention.

[0073] FIG. 3 is a conceptual diagram illustrating the process of comparing a normal operation sequence and an actual operation sequence according to an embodiment of the present invention.

[0074] FIG. 4 is a conceptual diagram illustrating the process of generating an interlock and graph using a PLC program and PLC I / O information according to one embodiment of the present invention.

[0075] FIG. 5 is a conceptual diagram illustrating the process of reverse search and determination of the minimum blocking cause set for a target event according to one embodiment of the present invention.

[0076] FIG. 6 is a conceptual diagram illustrating the process of generating control condition correction candidates and replaying counter-test runs according to one embodiment of the present invention.

[0077] FIG. 7 is a conceptual diagram illustrating the shadow mode processing process of a correction candidate according to one embodiment of the present invention.

[0078] FIG. 8 is a flowchart illustrating the worker guidance, correction approval, and rollback processing process according to one embodiment of the present invention.

[0079] FIG. 9 is a conceptual diagram illustrating the integrity verification and safe application processing of a correction value according to one embodiment of the present invention.

[0080] Preferred embodiments of the present invention are described in detail below. However, the embodiments described below are examples intended to specifically explain the technical concept of the present invention, and the scope of the present invention is not limited thereto. In describing the present invention, detailed descriptions of known configurations or functions may be omitted if such detailed descriptions could unnecessarily obscure the essence of the present invention.

[0081] Where in this specification it is described that any one component performs a specific function, said component may be implemented by an industrial computer, edge computing device, server, field terminal, or a combination thereof, comprising one or more processors, memory, storage devices, and communication interfaces. Additionally, the function of each component may be implemented by being integrated into a single device or distributed across multiple devices, and may be performed by program instructions, software modules, firmware, dedicated hardware, or a combination thereof.

[0082] An automation equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis according to one embodiment of the present invention collects multiple types of control data generated during the field commissioning process of automation equipment, detects abnormal sections by comparing a normal sequence when operating normally with an actual operation sequence that appears during the actual commissioning process, classifies the cause of the detected abnormal sections, and generates control condition correction candidates to resolve the classified cause of abnormality and provides them to the operator.

[0083] The automated equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis may include an equipment data collection unit (20), a time synchronization and event conversion unit (21), a normal sequence reference information storage unit (30), an AI sequence comparison analysis unit (40), an abnormality cause classification unit (50), a control condition correction candidate generation unit (60), and a worker guidance unit (70).

[0084] The automation equipment may be an automobile production line, automobile body assembly line, assembly line, welding line, logistics transfer line, robot assembly equipment, vision inspection equipment, servo motion equipment, parts processing equipment, packaging equipment, smart manufacturing equipment, or similar industrial automation equipment.

[0085] The automation equipment may include at least one of a PLC, HMI, sensor, actuator, relay, solenoid valve, motor, inverter, servo drive, servo motor, robot controller, industrial robot, vision inspection equipment, safety control device, communication module, and upper-level production management system.

[0086] The PLC receives input signals from sensors deployed in the field and can operate internal relays, timers, counters, comparison commands, output coils, and sequence steps according to logic conditions set in the PLC program. Output signals generated by the PLC can control the operation of cylinders, solenoid valves, motors, servo motors, robots, conveyors, clamps, lifters, or other actuators.

[0087] The PLC may be an Allen-Bradley family PLC, a Mitsubishi family PLC, a Siemens family PLC, an LS family PLC, or an industrial PLC from another manufacturer. The PLC program may be written in a ladder diagram, structured text, function block diagram, sequential function chart, command list, or a similar industrial control language.

[0088] The HMI can display the automatic or manual operation status of the automation equipment, sensor status, actuator status, interlock status, alarm status, production quantity, equipment cycle time, servo status, robot status, communication status, and control parameters. The HMI can receive automatic operation commands, manual operation commands, reset commands, alarm confirmation inputs, operation mode selections, parameter change inputs, and approval or rejection inputs for control condition correction candidates from the operator.

[0089] In this specification, commissioning may include at least one of the following processes performed after installing a new automation facility or modifying an existing automation facility: power application, PLC input / output verification, sensor and actuator verification, manual operation, automatic operation, interlock verification, alarm verification, servo home position return, robot operation verification, communication verification, process cycle verification, and mass production condition stabilization.

[0090] In addition, commissioning may include a re-verification process performed after changes in production vehicle models or items, modifications to PLC programs, modifications to HMI screens, modifications to electrical drawings, replacement of sensors or actuators, replacement of servos or robots, changes to communication devices, changes in production conditions, and maintenance.

[0091] In this specification, control conditions may include at least one of a timer preset value, a counter setting value, a sensor input judgment condition, a sensor input holding time, a sensor on delay value, a sensor off delay value, an interlock establishment condition, an interlock release condition, a servo position tolerance, a servo position arrival judgment time, a robot completion signal waiting time, a communication response waiting time, an alarm occurrence condition, an alarm occurrence delay time, and an HMI display condition, all of which are set in a PLC or HMI.

[0092] The equipment data collection unit (20) can collect control data generated during the commissioning process of the automation equipment. The equipment data collection unit (20) can collect at least one of a PLC input signal, a PLC output signal, a PLC internal device status, an internal relay status, a timer current value, a timer completion status, a counter current value, a calculation result of a comparison command, a sequence step number, an automatic operation status, a manual operation status, whether an interlock is established, an alarm occurrence condition, and a safety circuit status from the PLC.

[0093] PLC input signals may be signals input from proximity sensors, optical sensors, pressure sensors, position sensors, limit switches, product detection sensors, cylinder forward confirmation sensors, cylinder backward confirmation sensors, and safety sensors. PLC output signals may be solenoid valve outputs, motor start outputs, servo movement commands, robot start permission signals, clamp operation commands, conveyor drive commands, and alarm output signals.

[0094] The PLC internal device status may include at least one of an internal relay, a sequence progress flag, an operation ready flag, an operation completion flag, an interlock established flag, an error flag, and a reset status.

[0095] The equipment data collection unit (20) can collect at least one of the following from the HMI: HMI alarm code, alarm name, alarm occurrence time, alarm release time, operator operation history, manual operation history, screen switching history, parameter change history, alarm confirmation history, and user identification information.

[0096] For example, the time when the operator operated the cylinder forward button on the HMI, the time when the operator operated the reset button for a specific alarm, the time when the timer value was changed, and the time when the operation mode was switched from manual operation to automatic operation can be collected as operator operation history.

[0097] The equipment data collection unit (20) can collect digital on / off status from the sensor as well as pressure value, temperature value, torque value, vibration value, current value, position value, speed value, flow rate value and similar analog measurement values.

[0098] The equipment data collection unit (20) can collect the actuator's operation command, operation start status, operation completion status, operation time required, and operation error status.

[0099] The equipment data collection unit (20) can collect at least one of a command position, actual position, position deviation, command speed, actual speed, torque value, current value, home position return state, position reach state, servo on state, and servo alarm from the servo drive.

[0100] The equipment data collection unit (20) can collect at least one of a robot program number, current execution step, robot operation mode, robot ready state, robot start state, robot operation complete state, robot position reach state, robot standby state, robot alarm, and handshake signal between the PLC and the robot from the robot controller.

[0101] The equipment data collection unit (20) can collect at least one of the following from a communication device or communication module: communication connection status, data transmission time, data reception time, communication response time, number of retransmissions, communication timeout, disconnection history, node error, data update cycle, and protocol error.

[0102] Communication between the PLC, HMI, robot controller, servo drive, and upper production management system can be performed via Ethernet / IP, Profinet, CC-Link, EtherCAT, Modbus TCP, OPC UA, or other industrial communication methods.

[0103] The equipment data collection unit (20) can collect control data by a polling method that periodically reads the value of a PLC tag or PLC address, an event subscription method that receives data when a signal value changes, a method that collects log files of a PLC or HMI, a method that mirrors industrial communication data, or a combination thereof.

[0104] The equipment data collection unit (20) can collect control data through a read-only communication channel so as not to affect the actual control cycle and output operation of the PLC. Even when the equipment data collection unit (20) accesses the memory area or program of the PLC, it is configured to read the value of the monitoring target address by default, and write commands that change the actual output or internal control state can be blocked.

[0105] The equipment data collection unit (20) can correspond at least one of project identification information, production line identification information, equipment identification information, PLC identification information, program version, signal address, tag name, field device name, data type, data value, time of data occurrence and data quality information to each collected control data.

[0106] For example, when the clamp forward confirmation sensor connected to PLC input address X100 changes from an off state to an on state, the equipment data collection unit (20) can generate a data record including project identification information, equipment identification information, PLC identification information, input address X100, sensor name, value before change, value after change, and time of state change.

[0107] The time synchronization and event conversion unit (21) can convert control data collected from different devices into a common time axis.

[0108] PLC data can be updated in increments of a few milliseconds or tens of milliseconds depending on the PLC's scan cycle, HMI alarm data can be generated only when an alarm occurs or is cleared, and data from the robot controller and servo drive can be updated according to their respective control cycles. Accordingly, control data collected from each device may have different communication cycles and time standards.

[0109] The time synchronization and event conversion unit (21) can use at least one of the time of a standard time server, the system time of a PLC, the system time of an industrial computer, the time of an edge computing device, or the cycle reference signal of an automation facility as a common reference time.

[0110] When each device provides a timestamp generated by itself, the time synchronization and event conversion unit (21) can calculate the deviation between each device's timestamp and a common reference time and correct the occurrence time of the control data based on the calculated deviation.

[0111] If the device does not provide its own timestamp, the time at which the equipment data collection unit (20) receives the corresponding control data can be used as the time of data generation. If the communication delay is not constant, the time of reception can be corrected based on the average communication delay per device, the recent communication delay, or the communication round-trip time.

[0112] When analyzing data based on the scan cycle of the PLC, the equipment data collection unit (20) can collect the PLC's scan counter, a reference tag that increases periodically, or a time value inside the PLC together. The time synchronization and event conversion unit (21) can process multiple input signals, output signals, and internal device states collected within the same PLC scan cycle into the same state at the same point in time.

[0113] The time synchronization and event conversion unit (21) can convert continuously collected signal values ​​into state change-centered event data. The event data may include at least one of event identification information, a data generating device, a PLC address or tag name, a value before the state change, a value after the state change, a time of the state change, and a maintenance time of the changed state.

[0114] For example, if the sensor signal changes from 0 to 1, maintains the state of 1 for 1.2 seconds, and then changes back to 0, the time synchronization and event conversion unit (21) can generate a sensor on event, a sensor on state maintenance time, and a sensor off event.

[0115] In the case of analog data, the time synchronization and event conversion unit (21) can generate an event when the measured value changes by more than a set amount of change or deviates from the reference range. For example, if the position deviation between the actual position and the command position of the servo exceeds the set position tolerance, a position deviation exceeding event may be generated.

[0116] For continuously measured values ​​such as communication response time, a moving average, median, standard deviation, or deviation relative to the normal range may be calculated, and if the calculated value deviates from the set criteria, a communication response delay event may be generated.

[0117] The time synchronization and event conversion unit (21) can classify the collected control data into production cycles or equipment operation cycles. The start and end of a cycle can be determined using a PLC cycle start signal, cycle completion signal, product input signal, product output signal, robot start signal, or a reference event pre-specified for each piece of equipment.

[0118] In cases where there is no clear signal indicating the start and end of a cycle, the time synchronization and event conversion unit (21) can estimate the boundaries of the cycle by analyzing the sequence and time intervals of multiple events that appear repeatedly.

[0119] In cases where a single automated facility includes multiple parallel processes or multiple stations, the entire production line cycle and the cycle for each station can be managed separately.

[0120] The normal sequence reference information storage unit (30) can store normal sequence reference information when the automation equipment is operating normally.

[0121] Normal sequence reference information may include at least one of the step-by-step operation sequence of the automation equipment, a prerequisite for initiating each operation, a completion condition for determining that each operation has been completed, an allowable time for each step, an interlock condition, a sensor input sequence, an actuator output sequence, a servo position condition, a robot operation condition, a communication completion condition, and an HMI alarm standard.

[0122] Normal sequence reference information can be generated based on at least one of electrical drawings, PLC programs, HMI programs, PLC I / O lists, equipment specifications, commissioning checklists, input from the equipment manufacturer or operator, and actual commissioning cycles that have been successfully completed.

[0123] For example, normal sequence reference information may include operational relationships in which, when a cycle start signal is input while automatic operation conditions and safety conditions are established, a clamp forward output is generated, a servo movement command is output when the clamp forward confirmation sensor inputs within a set time, and a robot operation permission signal is output when a servo position arrival signal is input.

[0124] The normal sequence reference information storage unit (30) can generate a normal sequence candidate by using the order of occurrence of control events that commonly appear in a plurality of successfully completed commissioning cycles and the time elapsed between events.

[0125] The normal sequence reference information storage unit (30) can classify events that occur commonly in multiple normal cycles as essential events, and can classify events that occur only in some normal cycles according to equipment status, product items, or process branching as selected events or branching events.

[0126] In the process of generating normal sequence candidates, the normal time range between events can be calculated together. For example, in multiple normal cycles, when the time from the clamp forward output until the clamp forward confirmation sensor is input is 1.6 seconds, 1.7 seconds, 1.8 seconds, 1.7 seconds, and 1.9 seconds, respectively, the normal allowable time can be set based on the average value, median value, maximum value, standard deviation, or percentile.

[0127] The normal sequence reference information storage unit (30) can register a normal sequence candidate that has received approval input from a worker or engineer as normal sequence reference information. The worker or engineer can modify the required events, optional events, branching conditions, normal allowable time, and exclusion conditions included in the normal sequence candidate.

[0128] The normal sequence reference information storage unit (30) can store normal sequence reference information by corresponding it to at least one of the project name, customer, production line name, equipment name, PLC type, HMI type, PLC program version, HMI program version, electrical drawing version, I / O list version, commissioning date and person in charge.

[0129] When at least one of the PLC program version, HMI program version, electrical drawing version, or I / O list version is changed, the normal sequence reference information storage unit (30) can generate new normal sequence reference information without deleting or overwriting the existing normal sequence reference information.

[0130] The program version can be identified by a filename, a file creation time, a version identification value inside the program, a checksum, or a hash value. The automation equipment fault diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis can select normal sequence reference information that is identical to or corresponds to the program version currently applied to the automation equipment and compare it with the actual operation sequence.

[0131] The AI ​​sequence comparison analysis unit (40) can generate an actual operation sequence using control data collected, synchronized, and converted by the equipment data collection unit (20) and the time synchronization and event conversion unit (21).

[0132] The actual operation sequence may be a set of events arranged according to the time of occurrence of control events that occurred during a single equipment cycle.

[0133] The AI ​​sequence comparison analysis unit (40) can compare the actual operation sequence with the normal operation sequence according to the normal sequence reference information stored in the normal sequence reference information storage unit (30).

[0134] The AI ​​sequence comparison analysis unit (40) can compare at least one of the event occurrence order, whether an event occurs, the time of event occurrence, the event duration, and the time interval between events between a normal operation sequence and an actual operation sequence.

[0135] If there is an event that exists in the normal operation sequence but does not exist in the actual operation sequence, the AI ​​sequence comparison analysis unit (40) can detect the event as a missing event.

[0136] If an event occurs in a different order than the one specified in the normal operation sequence, the corresponding section may be detected as an operation sequence error.

[0137] If an event occurs later than the allowed time set in the normal sequence reference information, the event may be detected as an operation delay.

[0138] If the same operation, the same sequence step, or the same internal state is repeated more than a set number of times, it may be detected as a repetition stop or repetition retry.

[0139] If the preceding interlock condition is not met and the transition to the next step fails, it may be detected as an unreleased interlock.

[0140] If an HMI alarm occurs that does not occur during the normal operation sequence, it may be detected as an abnormal alarm.

[0141] If servo position deviation, pressure value, temperature value, torque value, communication response time, or other numerical data deviates from the normal range, it may be detected as deviation from the set range.

[0142] The AI ​​sequence comparison analysis unit (40) can determine the event where the normal operation sequence and the actual operation sequence first do not match as the first deviation event.

[0143] The initial deviation event can be used as a criterion to identify the point of origin of the fundamental anomaly that precedes the alarm finally displayed on the HMI in time.

[0144] For example, even if a robot operation failure alarm is displayed on the HMI, the actual initial departure event may be that the servo position arrival signal, which is a prerequisite for the robot operation permission signal, did not occur within the set time.

[0145] The AI ​​sequence comparison analysis unit (40) can compare not only the absolute time of occurrence but also the relative order of occurrence of events and the time required for each interval between events, taking into account that the operating speed of the actual equipment may vary from cycle to cycle.

[0146] If there are multiple acceptable branch paths in a normal operation sequence, the AI ​​sequence comparison analysis unit (40) can detect an abnormal section after selecting the normal branch path most similar to the actual operation sequence.

[0147] In the process of aligning the normal operation sequence and the actual operation sequence, dynamic time matching, edit distance, longest common subsequence, probabilistic sequence matching, time series classification models, or analysis methods combining these may be used.

[0148] The AI ​​sequence comparison analysis unit (40) may use a recurrent neural network, a long short-term memory neural network, a transformer, an autoencoder, a probability model, a statistical anomaly detection model, a clustering model, a classification model, or a learning model that combines these. However, the analysis method of the AI ​​sequence comparison analysis unit (40) is not limited to a specific artificial intelligence model and can be implemented in a manner that combines data-based analysis and PLC logic-based analysis.

[0149] The abnormal cause classification unit (50) can analyze the correlation between the PLC input signal, PLC output signal, internal device status, HMI alarm, interlock condition, sensor status, actuator status, servo motion data, robot operation status and communication status in the abnormal section detected by the AI ​​sequence comparison analysis unit (40).

[0150] The abnormal cause classification unit (50) can classify the abnormal cause into at least one of PLC logic abnormality, sensor signal abnormality, output or wiring abnormality, communication abnormality, servo position abnormality, robot operation standby, interlock condition mismatch and setting parameter abnormality.

[0151] PLC logic abnormalities may include at least one of the following: omission of required output conditions, use of incorrect contacts, negative logic errors, use of incorrect addresses, timer completion condition errors, sequence transition condition errors, reset condition errors, and conflicting output conditions.

[0152] Sensor signal abnormalities may include at least one of sensor non-input, sensor input delay, sensor chatter, insufficient sensor input hold time, sensor always on, sensor always off, and sensor input sequence error.

[0153] Output or wiring abnormalities may include at least one of the following: when the PLC output occurs normally but the field actuator does not operate, when the PLC output address does not match the actual wiring, open circuit, terminal miswiring, relay failure, and driving power abnormality.

[0154] Communication abnormalities may include at least one of device response delay, communication timeout, disconnection, node error, data update delay, and handshake signal mismatch.

[0155] Servo position abnormalities may include at least one of incomplete return to home position, failure to reach position, exceeding position deviation, torque limit, speed condition mismatch, and servo alarm.

[0156] Robot operation standby may include at least one of the following: failure to output robot start permission from the PLC, failure to input a robot ready signal, mismatch in robot program number, failure to input a robot operation completion signal, and mismatch in safety area interlock.

[0157] An interlock condition mismatch may include a state in which an output or sequence transition is blocked because one or more of a plurality of preconditions are not satisfied.

[0158] The abnormal cause classification unit (50) can extract logical dependency relationships between input contacts, internal relays, timers, counters, comparison commands, output coils, interlock conditions, and alarm occurrence conditions included in the PLC program.

[0159] The abnormal cause classification unit (50) can generate an interlock graph representing the cause and effect relationships between multiple control events by combining the extracted logical dependency relationship and the correspondence relationship between the PLC address included in the PLC I / O list and the field device.

[0160] The interlock causation graph can represent input contacts, internal relays, timers, counters, comparison commands, servo status, robot status, communication status, output coils, operation commands, and alarm conditions as nodes.

[0161] Interlocks and graph edges can represent logical relationships where the state of one node affects the establishment or output of another node.

[0162] When a PLC program is written as a ladder diagram, contacts connected in series can be converted into AND conditions, and contacts connected in parallel can be converted into OR conditions. Negative contacts can be converted so that the condition is satisfied when the corresponding signal is in the OFF state.

[0163] When a PLC program is written as structured text, interlocks and graphs can be generated by analyzing conditional statements, branch statements, logical operators, comparison operators, and variable assignment relationships.

[0164] PLC programs from different manufacturers can be converted into a common intermediate representation. The common intermediate representation may include at least one of signal identification information, logical operations, preconditions, result variables, and execution conditions.

[0165] For example, if the operating condition of output Y210 consists of the logical AND of safety door normal signal X001, emergency stop normal signal X002, sensor detection signal X100, timer completion signal T010, and servo position attainment signal M220, a causal path connecting each signal node to the node of output Y210 can be formed in the interlock causal graph.

[0166] When the internal relay M300 combines the above multiple conditions and the output Y210 operates according to the state of the internal relay M300, a path connecting from the input signals to the internal relay M300 and a path connecting from the internal relay M300 to the output Y210 can be formed.

[0167] The abnormal cause classification unit (50) can search the interlock in reverse direction from the first deviation event where the normal operation sequence and the actual operation sequence first do not match.

[0168] In addition, the abnormal cause classification unit (50) can set the output event that did not occur as the target event when the output event that should occur in the normal operation sequence does not occur in the actual operation sequence.

[0169] The abnormal cause classification unit (50) can search in reverse direction for the interlock in and graph from the output coil or operation command corresponding to the target event to extract at least one of the input contact, internal relay, timer completion condition, counter condition, servo position attainment condition, robot completion signal, communication completion signal and safety interlock condition as a cause candidate.

[0170] During the reverse search process, the actual state of each antecedent condition and whether it is satisfied can be verified. Antecedent conditions that are successfully satisfied are classified as normal conditions, while those that are not satisfied can be extracted as cause candidates.

[0171] The abnormal cause classification unit (50) can calculate the cause reliability for each cause candidate by using at least two of the following: temporal sequence, frequency of repeated occurrence, signal value deviation relative to normal cycle, correlation with related HMI alarm, and similarity with past field action history.

[0172] The temporal sequence can indicate whether the abnormal state corresponding to the cause candidate occurred before the initial deviation event or the final alarm.

[0173] The frequency of recurrence can indicate the extent to which the same candidate cause appears repeatedly in multiple abnormal cycles.

[0174] The signal value deviation relative to the normal cycle may indicate the difference in signal occurrence time, signal duration, measurement value, or response time compared to the normal cycle.

[0175] The correlation with related HMI alarms can be calculated based on the ratio of specific cause candidates and specific HMI alarms occurring together in the past.

[0176] Similarity with past field action history can indicate the similarity between the measures actually taken in the past at the same or similar anomalies and the current candidate cause.

[0177] Causal reliability can be calculated by a weighted sum of multiple evaluation items, a probability model, a classification model, or a learned inference model.

[0178] The abnormal cause classification unit (50) can set the priority of multiple cause candidates according to the order of high cause reliability. Each cause candidate may be associated with a PLC address, tag name, interlock condition, related HMI alarm, normal state, actual state, and signal change history that serve as the basis for extracting the corresponding cause candidate.

[0179] The abnormal cause classification unit (50) can determine the minimum number of condition sets in which the signal transmission path to the target event is restored when the state of one or more of the extracted multiple cause candidates changes as the minimum blocking cause set.

[0180] The minimum blocking cause set does not simply include all signals that are off or abnormal, but may include only the minimum cause conditions necessary to actually restore the occurrence of the target event.

[0181] For example, if the output condition of a target event consists of “AND B AND C” and the actual state is A=1, B=0, and C=0, the set including B and C can be determined as the minimum blocking cause set because both B and C must change to a steady state for the target event to occur.

[0182] When the output condition of the target event is composed of “AND (B OR C)” and the actual state is A=1, B=0, and C=0, the target event can occur even if only one of B or C changes to a normal state; therefore, the set containing only B and the set containing only C can each be determined as candidates for the minimum blocking cause set.

[0183] The set of minimum blocking causes can be determined by interlock causation, reverse search of the graph, Boolean simplification, minimum cut set analysis, satisfaction determination, or a combination thereof.

[0184] If multiple sets of minimum blocking causes exist, priority can be determined based on the cause reliability, change risk, past action success rate, and the number of included conditions of the cause candidates included in each set.

[0185] Safety input signals, emergency stop input signals, safety door input signals, light curtain input signals, motor overload input signals, and collision prevention input signals may be analyzed as causes that prevent the occurrence of target events, but may not be selected as targets for control condition correction.

[0186] If safety-related conditions are not met, the automated equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis may not generate control condition correction candidates that forcibly establish or bypass the conditions, and may guide the operator to directly inspect the safety device, site condition, wiring condition, or safety circuit.

[0187] The control condition correction candidate generation unit (60) can generate control condition correction candidates based on abnormal causes classified by the abnormal cause classification unit (50) and normal sequence reference information.

[0188] The control condition correction candidate generation unit (60) can select a control condition directly connected to a cause candidate or a cause candidate included in the minimum blocking cause set, whose cause reliability is greater than or equal to a set reference value, as a control condition to be corrected.

[0189] The control conditions subject to calibration may include at least one of a PLC timer preset value, a sensor input judgment condition, a sensor input hold time, a sensor on delay value, a sensor off delay value, a servo position tolerance, a servo position reach judgment time, a robot completion signal waiting time, a communication response waiting time, an alarm occurrence delay time, a non-safety interlock release condition, and an HMI display condition.

[0190] The control condition correction candidate generation unit (60) can exclude safety input signals, emergency stop input signals, safety door input signals, light curtain input signals, motor overload input signals and collision prevention input signals from the correction targets.

[0191] The control condition correction candidate generation unit (60) can restrictively change only the control conditions defined as items that can be corrected in advance, instead of arbitrarily changing the entire PLC program.

[0192] The control condition correction candidate generation unit (60) can generate a plurality of control condition correction candidates based on at least one of the set value applied in a past normal commissioning cycle, the set value applied to the same type of automation equipment, the design allowable range of the control condition, and the set value stabilized after past field measures.

[0193] For example, if a sensor input occurs normally but occurs later than the normal standard and a timer timeout occurs, the control condition correction candidate generation unit (60) can generate multiple candidates to change the timer preset value currently set to 2.0 seconds to 2.3 seconds, 2.5 seconds, or 2.7 seconds.

[0194] In the case where the actual position deviation of the servo is stably maintained at the level of ±0.6 mm but the current position arrival tolerance is set to ±0.5 mm and no position arrival signal is generated, the control condition correction candidate generation unit (60) can generate a position tolerance candidate of ±0.7 mm or ±0.8 mm within the design tolerance range.

[0195] If the normal response time of a communication device increases due to field network conditions, a correction candidate that increases the communication response waiting time above the current value may be generated.

[0196] The control condition correction candidate generation unit (60) can generate multiple candidates at regular intervals based on the current value or generate candidate values ​​based on the statistical distribution of normal test run data. Additionally, candidate values ​​that minimize the difference between the normal operation sequence and the actual operation sequence may be calculated by a search or optimization algorithm.

[0197] A candidate value may be excluded if it exceeds the design tolerance, has the potential to degrade production quality, increases the risk of collision with equipment, or alters safety conditions.

[0198] The control condition correction candidate generation unit (60) can generate a single correction candidate that changes only one control condition and a composite correction candidate that changes multiple associated control conditions together. Even when a composite correction candidate is generated, the number of control conditions to be changed can be minimized as much as possible.

[0199] The control condition correction candidate generation unit (60) can generate a virtual commissioning sequence to which each control condition correction candidate is applied.

[0200] A virtual commissioning sequence may be a sequence that calculates the operation result when a correction candidate is applied by utilizing commissioning data collected in the past and the PLC logic relationship without changing the control conditions of the actual PLC or HMI.

[0201] The control condition correction candidate generation unit (60) can evaluate whether the equipment cycle is completed, the time required for each step, whether repeated stops occur, whether an alarm occurs, and whether safety interlocks are satisfied for a virtual commissioning sequence.

[0202] In generating a virtual commissioning sequence, a plurality of normal commissioning cycles and at least one abnormal commissioning cycle collected prior to the occurrence of an abnormality may be used. In one embodiment, at least three normal commissioning cycles and at least one abnormal commissioning cycle may be used.

[0203] When generating a virtual commissioning sequence, the actually collected sensor inputs, safety inputs, robot status, servo status, and communication status are maintained, and only the timers, tolerances, wait times, or non-safety interlock conditions selected as calibration targets can be changed according to their respective control condition calibration candidates.

[0204] The control condition correction candidate generation unit (60) can re-evaluate the logic conditions of the PLC in chronological order according to the changed control conditions. Accordingly, the internal relay state, timer completion state, output coil state, alarm state, and sequence step expected to occur at each point in time can be calculated.

[0205] For example, it can be assumed that a candidate to change the timer preset value from 2.0 seconds to 2.5 seconds is applied, and that the corresponding sensor input occurs at 2.3 seconds during an abnormal commissioning cycle. In the existing settings, a timeout occurs when 2.0 seconds have elapsed, but in the virtual commissioning sequence with the correction candidate applied, the sensor input is treated as having occurred before the timer is completed, so a subsequent output may occur.

[0206] The control condition correction candidate generation unit (60) can select a control condition correction candidate as an applicable candidate that reduces the difference from the normal operation sequence among a plurality of control condition correction candidates while not changing or bypassing at least one safety condition among the emergency stop condition, safety door condition, light curtain condition, overload condition and collision prevention condition.

[0207] The control condition correction candidate generation unit (60) can perform counter-test operation playback for each control condition correction candidate.

[0208] Counterfactual commissioning replay may be a process that calculates how the output and subsequent sequence of an automated facility would have changed if only specific control conditions were changed according to correction candidates under a situation where the same input and equipment states collected in the past are given.

[0209] In the replay of the counter-realistic commissioning, the actually observed input signals and equipment status are maintained, and only the control conditions selected for correction can be changed.

[0210] The control condition correction candidate generation unit (60) can determine whether a target event occurs as a result of the counter-test run playback.

[0211] Additionally, the control condition correction candidate generation unit (60) can determine whether subsequent events specified in the normal operation sequence after the target event are completed in the set order.

[0212] The control condition correction candidate generation unit (60) can determine whether the time required to the target event or the total equipment cycle time is shortened compared to before the correction. However, even if the cycle time is not shortened, if the abnormal condition is resolved and the total equipment cycle is completed within the normal allowable time, the candidate may be maintained.

[0213] The control condition correction candidate generation unit (60) can determine whether the state of the PLC logic and the signal transmission path corresponding to the safety input signal, emergency stop input signal, safety door input signal, light curtain input signal, motor overload input signal and collision prevention input signal are changed.

[0214] If the logical expression, input address, output address, precondition, or signal transmission path of a safety-related condition changes before or after correction, the corresponding control condition correction candidate may be excluded.

[0215] The control condition correction candidate generation unit (60) can virtually determine whether equipment driving output is generated by the control condition correction candidate when the safety door is open, the emergency stop is activated, or the light curtain is detected. If it is determined that driving output is generated when the safety condition is not satisfied, the candidate may be excluded.

[0216] The control condition correction candidate generation unit (60) can determine whether a new HMI alarm, repeated stop, conflicting output, or abnormal sequence transition occurs due to the application of the control condition correction candidate.

[0217] The control condition correction candidate generation unit (60) can select a control condition correction candidate as a verification candidate that satisfies all conditions such that a target event occurs as a result of the counter-test run playback, subsequent events are completed in a normal order, safety-related PLC logic and signal transmission paths are not changed, and a new HMI alarm or repeated stop does not occur.

[0218] If applying a control condition correction candidate to a normal commissioning cycle results in the existing normal operation changing to abnormal, that candidate may be excluded. Candidates that restore target events during abnormal commissioning cycles while maintaining the existing normal operation during normal commissioning cycles may be preferentially selected as verification candidates.

[0219] The control condition correction candidate generation unit (60) can perform a shadow mode in which a verification candidate that has passed the counter-actual trial run playback is reflected only in the virtual output area without being reflected in the actual PLC output.

[0220] In shadow mode, real-time input data input to the actual PLC can be input to the existing PLC logic and the virtual control logic to which the verification candidate is applied, respectively.

[0221] Existing PLC logic drives automation equipment by generating outputs in the actual output area, while virtual control logic with applied verification candidates can generate output results in a virtual output area that is not connected to actual output modules or field actuators.

[0222] The virtual output area can be configured in a non-output memory area inside the PLC, a separate industrial computer, an edge computing device, or a simulation server.

[0223] The control condition correction candidate generation unit (60) can compare the output result of the virtual output area with the output result of the normal operation sequence.

[0224] The degree to which the output result of the virtual output area matches the output result of the normal operation sequence can be calculated as the output matching rate.

[0225] The output matching rate can be calculated using at least one of the ratio of events whose occurrence status matches among the output events to be compared, the ratio of events whose order of occurrence matches, and the ratio of events whose error in the time of occurrence is within an allowable range.

[0226] The control condition correction candidate generation unit (60) can perform a shadow mode for at least three consecutive equipment cycles.

[0227] If the agreement rate between the virtual output result and the output result of the normal operation sequence is greater than the set standard during at least three consecutive equipment cycles, and no new alarms, repeated stops, or violations of safety conditions occur, the verification candidate may be selected for final approval.

[0228] The set compliance rate can be set to 90%, 95%, 98%, or other values ​​depending on the type of equipment, process characteristics, and safety importance.

[0229] If the quality of real-time input data is low or the operating status of the automation equipment is unstable, the shadow mode may be suspended and treated as requiring re-verification.

[0230] The control condition correction candidate generation unit (60) can generate a correction approval package for a verification candidate that has passed the counter-real test run playback and shadow mode.

[0231] The calibration approval package may include control conditions before calibration, control conditions after calibration, the address of the PLC to be calibrated, a tag name, an affected interlock condition, a version identification value of the PLC program to be applied, a hash value of the program before calibration, and rollback information for returning to the state before calibration.

[0232] The calibration approval package may further include at least one of project identification information, equipment identification information, PLC identification information, related normal sequence reference information, time of occurrence of anomaly, first deviation event, minimum set of blocking causes, cause reliability, counterfactual commissioning playback result, shadow mode result, and output matching rate.

[0233] The hash value of the program before calibration can be used to verify whether the actual PLC program to be calibrated is identical to the PLC program used for analysis and virtual verification.

[0234] If the hash value of the actual PLC program and the hash value stored in the correction approval package are different, the application of control condition correction candidates may be restricted.

[0235] Rollback information may include at least one of the setting value before correction, the address to be changed, the difference information before and after the change, the program file before correction, the application order, and the restoration procedure.

[0236] The worker guidance unit (70) can output abnormal sections, classification results of abnormal causes, control data that served as the basis for the classification results, field inspection targets, and control condition correction candidates to the HMI or worker terminal.

[0237] The operator terminal may be an industrial computer, a field laptop, a tablet, a smartphone, a remote maintenance terminal, or a web-based management screen.

[0238] The operator guide (70) can display the sequence stage where the abnormality occurred, the first deviation event, the relevant PLC address, the relevant interlock condition, the normal state, the actual state, and the cause reliability.

[0239] For example, the worker guide (70) can indicate that the servo position arrival signal after clamp advancement was delayed by 0.6 seconds compared to the normal standard, and that the robot operation permission interlock was not established.

[0240] The worker guide unit (70) can provide field inspection items such as checking the sensor input status, checking the sensor wiring, checking the servo home position status, checking the servo position deviation, checking the robot ready signal, and checking the communication response time.

[0241] The worker guide section (70) can display at least one of the electrical drawing page related to the cause of the abnormality, the network or program section of the PLC program, the HMI alarm screen, and the location of the PLC I / O list.

[0242] When a candidate for control condition correction is generated, the operator guide unit (70) can display the setting value before correction, the candidate value after correction, the PLC address to be changed, the basis for correction, the expected effect, the result of comparison with the normal operation sequence, the result of the counter-operation trial run, the shadow mode result, and the rollback information.

[0243] When there are multiple control condition correction candidates, the worker guide unit (70) can display the priority of the candidates based on the possibility of resolving abnormalities, safety, number of change targets, equipment cycle time, past action success rate, and ease of rollback.

[0244] The worker guidance unit (70) can receive an input from the worker regarding approval, rejection, or modification of the control condition correction candidate.

[0245] The automation equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis can be configured so as not to automatically reflect control condition correction candidates to the actual PLC or HMI without the operator's approval.

[0246] If the operator modifies the value of the control condition correction candidate, the control condition correction candidate generation unit (60) can generate a virtual commissioning sequence, play back a real commissioning sequence, and re-perform a shadow mode for the modified candidate.

[0247] The worker guidance unit (70) can generate approval information that can be reflected in the actual PLC or HMI for the control condition correction candidate that has received the final approval input from the worker.

[0248] After final approval by the operator, the control condition correction candidates can be reflected in the actual PLC or HMI.

[0249] Candidates for control condition correction can be manually reflected by an operator using a program tool of the PLC or HMI. Alternatively, an automated equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis can reflect approved correction values ​​through an authorized communication interface.

[0250] When the correction value is automatically reflected, the automation equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis can check whether the automation equipment is in a stopped state, whether automatic operation is disabled, whether the safety conditions are normal, whether the program backup is completed, and whether the version or hash value of the actual program matches the information stored in the correction approval package.

[0251] After the correction value is reflected, a re-operation of the automated equipment can be performed.

[0252] During the re-operation process, the equipment data collection unit (20) collects the actual operation sequence after correction again, and the AI ​​sequence comparison analysis unit (40) can compare the actual operation sequence after correction with the normal operation sequence.

[0253] If the normal operation sequence is maintained for more than a set number of equipment cycles, the correction value can be stored as a stabilized control condition.

[0254] If a new alarm, repeated stop, production quality degradation, or safety condition abnormality occurs after the application of the correction value, the system can return to the state before correction according to the rollback information included in the correction approval package.

[0255] Rollback can be performed by operator approval or configured to be performed immediately in the event of a safety-related anomaly.

[0256] As a specific example of application, the automation equipment may be equipment that fixes the product with a clamp, moves the servo to a target position, and starts the robot when the servo's arrival at the position is confirmed.

[0257] The normal operation sequence may consist of cycle start, safety condition check, product detection, clamp advance, clamp advance confirmation, servo movement, servo position attainment, robot operation authorization, robot operation completion, clamp retraction, and cycle completion.

[0258] During the commissioning process, clamp forward movement and servo movement were performed normally, but because the servo position arrival signal did not occur within the set time, the robot operation permission signal was not output, and a robot operation failure alarm may be generated on the HMI.

[0259] The equipment data collection unit (20) can collect the input and output of the PLC, the target position of the servo, the actual position, the position deviation, the current value of the timer, the robot ready signal, and the HMI alarm.

[0260] The time synchronization and event conversion unit (21) can align the collected PLC data, servo data, robot data, and HMI alarm data to a common time axis and convert each signal change into event data.

[0261] The AI ​​sequence comparison analysis unit (40) compares the normal operation sequence with the actual operation sequence and can detect the first deviation event when the servo position arrival signal does not occur within the set time.

[0262] The abnormal cause classification unit (50) can search the interlock ine and graph regarding the robot operation permission signal in reverse.

[0263] If the safety condition, clamp advance confirmation signal, and robot ready signal are normal but only the servo position reach signal is not established, the servo position reach condition can be identified as a candidate cause that blocked the occurrence of the target event.

[0264] It can be assumed that the deviation between the target position and the actual position of the servo is 0.65 mm, the current position reach tolerance is ±0.5 mm, and the maximum design tolerance is ±1.0 mm.

[0265] The abnormal cause classification unit (50) can determine the servo position arrival condition as the minimum blocking cause set.

[0266] The control condition correction candidate generation unit (60) can generate multiple control condition correction candidates that change the position reach tolerance to ±0.7 mm, ±0.8 mm, and ±0.9 mm.

[0267] The control condition correction candidate generation unit (60) can generate a virtual commissioning sequence to which each candidate is applied.

[0268] For candidates within ±0.7 mm, a servo position arrival signal may be generated, but the position arrival judgment may be deemed unstable during some normal commissioning cycles.

[0269] For candidates within ±0.8 mm, the position arrival signal and subsequent robot operation can occur normally in both normal and abnormal commissioning cycles.

[0270] For candidates with ±0.9 mm, equipment operation is possible, but the tolerance may be judged to be excessively wide compared to the product's positional accuracy or process quality standards.

[0271] Accordingly, a candidate with ±0.8 mm can be selected as a priority verification candidate.

[0272] The control condition correction candidate generation unit (60) can perform counter-actual test run regeneration using past normal test run cycles and abnormal test run cycles for candidates of ±0.8 mm.

[0273] In the counterfactual commissioning playback, if the target event, the servo position attainment signal, occurs, the subsequent robot operation permission signal and robot operation completion signal occur in the normal sequence, and no new alarms or safety condition violations occur, the corresponding candidate can be maintained.

[0274] Subsequently, a shadow mode using input data from the actual automation equipment can be performed for at least three consecutive equipment cycles.

[0275] If the virtual output result matches the output result of the normal operation sequence by a set matching rate or higher, and no new alarms, repeated stops, or safety condition violations occur, a candidate of ±0.8 mm can be selected as the final approved candidate.

[0276] The worker guide unit (70) can display to the worker position tolerance before correction, position tolerance after correction, related PLC address, related interlock condition, counter-test run playback result, shadow mode result and rollback information.

[0277] After final approval by the operator, the corresponding correction value can be reflected in the actual control conditions, and normal operation can be verified through re-commissioning.

[0278] As another application example, a timeout alarm may occur if the sensor input of the automation equipment occurs normally but later than the preset value of the currently set PLC timer.

[0279] As a result of analyzing multiple normal commissioning cycles, the time to sensor input is an average of 2.1 seconds and a maximum of 2.4 seconds, but it can be assumed that the PLC timer preset value is 2.0 seconds.

[0280] The AI ​​sequence comparison analysis unit (40) can detect that the timer completion event occurred before the sensor input event as the first exit event.

[0281] The abnormal cause classification unit (50) can extract the timer completion condition and sensor input delay as cause candidates and determine the preset value of the PLC timer as a control condition to be corrected in relation to the minimum blocking cause set.

[0282] The control condition correction candidate generation unit (60) can generate multiple candidates that change the timer preset value to 2.5 seconds, 2.7 seconds, and 3.0 seconds.

[0283] The control condition correction candidate generation unit (60) can perform a counter-actual test run using a normal test run cycle and an abnormal test run cycle for each candidate.

[0284] If a candidate with 2.5 seconds accepts sensor input normally, minimizes delays in the entire equipment cycle, and does not cause new alarms or safety condition violations, that candidate may be selected as a priority verification candidate.

[0285] When a verification candidate generates a normal output result during a set number of equipment cycles in shadow mode, the worker guidance unit (70) can provide the candidate and the basis for analysis to the worker.

[0286] In this way, the automation equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis is not limited to simply displaying the HMI alarm that occurred, but can also analyze PLC input signals, PLC output signals, internal device status, HMI alarms, interlock conditions, sensor status, servo status, robot status and communication status on a common time axis.

[0287] In addition, the automation equipment abnormality diagnosis and control condition correction system (100) through AI-based PLC / HMI commissioning data analysis can detect the point where the normal operation sequence and the actual operation sequence first differ, and specifically identify the cause of the abnormality by searching the interlock in reverse direction and the graph based on the logical dependency relationship of the PLC program.

[0288] Furthermore, by determining the minimum set of blocking causes that prevent the occurrence of the target event, correction candidates can be generated only for limited control conditions while excluding safety-related conditions from the correction targets.

[0289] The generated control condition correction candidates can be reviewed through counter-real-world commissioning playback and shadow mode before being reflected in the actual PLC, and can be reflected in the actual PLC or HMI only after final approval by the operator.

[0290] Accordingly, the time required to trace the cause of anomalies, which previously relied on operator experience during the commissioning of automated equipment, can be shortened, and the accuracy of control condition correction and the efficiency of on-site commissioning can be improved while maintaining safety.

[0291] In one embodiment, the control condition correction candidate generation unit (60) can search the interlock in the reverse and forward directions based on the correction target PLC address to generate a correction influence logic slice including a preceding condition and a subsequent control condition logically connected to the correction target PLC address.

[0292] The above correction influence logic slice may include at least one of an input contact, internal relay, timer, counter, comparison command, interlock condition, output coil, and HMI alarm condition connected to the correction target PLC address.

[0293] For example, if the PLC address to be calibrated is a preset value of a timer that determines whether a robot start permission signal is generated, the calibration influence logic slice may include a sensor input and an internal relay corresponding to the start condition of the timer, and a robot start permission output and related alarm conditions that operate based on the timer completion condition.

[0294] The control condition correction candidate generation unit (60) can arrange the PLC address, instruction type, operand, setting value, and logical connection order included in the correction influence logic slice according to a preset normalization order.

[0295] The above normalization order may be the address order within the PLC program, the execution order of logical operations, the interlock and graph search order, or a combination thereof.

[0296] The control condition correction candidate generation unit (60) can calculate a local logic hash value for the normalized correction influence logic slice.

[0297] The above local logic hash value is a value distinct from the overall program hash value for the entire PLC program, and can be used to verify whether the logic section affected by the actual correction has been changed after approval.

[0298] The control condition correction candidate generation unit (60) may include a list of the entire program hash value, the local logic hash value, the correction target PLC address, the setting value before correction, the setting value after correction, and the PLC address included in the correction influence logic slice in the correction approval package.

[0299] When the worker guidance unit (70) receives the worker's final approval input, the system (100) can recalculate the total program hash value and the local logic hash value from the PLC program currently applied to the actual PLC.

[0300] The system (100) may allow the application of the post-correction setting value only when the recalculated whole program hash value matches the whole program hash value included in the correction approval package and the recalculated local logic hash value matches the local logic hash value included in the correction approval package.

[0301] Even if the entire program hash value does not match but the local logic hash value matches, the system (100) determines that a change has occurred in a program area not directly related to the correction influence logic slice and may request the worker to reconfirm the change program.

[0302] If the local logic hash value does not match, the system (100) may stop applying the correction by determining that the control condition to be corrected or the related interlock condition has been changed after the operator's approval.

[0303] If the correction application is interrupted, the system (100) can re-verify the validity of the normal sequence reference information based on the changed PLC program and guide the re-performance of the counter-test and shadow mode verification.

[0304] In one embodiment, the system (100) can reflect the corrected set value in the actual PLC or HMI only when the automation equipment enters a preset safety application state.

[0305] The above safety application state may be a state in which at least two of the following conditions are satisfied: a state in which automatic operation is stopped, a state in which an equipment cycle completion signal is input, a state in which there are no operating actuators, a state in which the safety input signal is normal, and a state in which the output signal included in the correction influence logic slice maintains the same state during a set number of PLC scans.

[0306] For example, the system (100) can determine the safe application state when automatic operation is deactivated, an equipment cycle completion signal is input, and the output signal included in the correction influence logic slice is not changed during at least 5 consecutive PLC scans.

[0307] The system (100) can first record the set value after correction in a temporary correction area provided in the PLC or HMI before directly recording it in the actual control area.

[0308] The above temporary correction area may be an internal memory area of ​​the PLC, an internal memory area of ​​the HMI, a memory area of ​​an industrial computer, or a separate correction value storage area that does not directly affect the actual output or sequence progress status.

[0309] The system (100) can read the value recorded in the temporary correction area again and compare it with the post-correction setting value included in the correction approval package.

[0310] If the value read from the above temporary correction area matches the above correction post-set value, the system (100) can generate a valid application approval flag during one set PLC scan.

[0311] When the above application approval flag occurs, the PLC or HMI can switch the set value after correction of the above temporary correction area to the actual control area.

[0312] When multiple control conditions are included in a single calibration approval package, multiple post-calibration setting values ​​can be switched together to the actual control area by the same application approval flag.

[0313] Accordingly, partial application, in which only some of the multiple corrected set values ​​are reflected in the actual control area, can be prevented.

[0314] The system (100) can read the value recorded in the actual control area again after the occurrence of the above-mentioned application approval flag and check whether the value read from the actual control area matches the corrected setting value included in the correction approval package.

[0315] If the value read from the actual control area above does not match the set value after correction, or if a correction application completion signal does not occur within the set number of PLC scans, the system (100) may determine that the correction application has not been successfully completed.

[0316] If it is determined that the correction application has not been successfully completed, the system (100) can record the pre-correction setting value included in the correction approval package in the temporary correction area and generate a rollback approval flag to restore the actual control area to the pre-correction state.

[0317] When the set value after correction is properly reflected in the actual control area, the system (100) can perform a re-operation of the automation equipment and prioritize monitoring the control events included in the correction influence logic slice.

[0318] If, during the re-operation process, the target event to be corrected occurs normally, subsequent events specified in the normal operation sequence are completed in the set order, and no new HMI alarm, repeated stop, or safety condition violation occurs, the system (100) can maintain the set value after correction as a confirmed control condition.

[0319] If, during the re-operation process, an output event included in the correction influence logic slice occurs differently from the normal operation sequence, or if a new abnormal event is detected that did not exist prior to the application of the correction post-setting value, the system (100) may determine that there is a correlation between the correction post-setting value and the new abnormal event.

[0320] If it is determined that the above correlation exists, the system (100) can use the rollback information included in the correction approval package to return to the pre-correction setting value and change the corresponding control condition correction candidate into an application exclusion candidate or a re-verification candidate.

[0321] By verifying the local logic hash value of such correction influence logic slices and applying correction values ​​in bulk using a temporary correction area, it is possible to prevent incorrect application of correction values ​​when the relevant PLC logic has been changed after operator approval.

[0322] In addition, by recording the correction-after-setting value in a temporary correction area instead of directly in the actual control area, and then switching using a read-back verification and application approval flag, it is possible to prevent only some of the multiple correction values ​​from being applied due to communication errors or data recording errors.

[0323] In addition, by intensively monitoring control events included in the correction influence logic slice during the re-commissioning process, direct effects caused by the post-correction set value and normal equipment fluctuations can be distinguished, and the system can quickly return to the state before correction in the event of an anomaly. Explanation of the symbols

[0325] 100. System for Automated Equipment Anomaly Diagnosis and Control Condition Correction through AI-Based PLC / HMI Commissioning Data Analysis

Claims

Claim 1 A system for analyzing control data collected during the commissioning process of an automation facility to diagnose the cause of an abnormality in the automation facility and to support the correction of control conditions, comprising: an equipment data collection unit for collecting commissioning data including at least one of an input signal, output signal and internal device status of a PLC installed in the automation facility, an alarm and operation history of an HMI, a sensor status, an actuator operation status, servo motion data, a robot operation status, a communication status, and an interlock condition; a normal sequence reference information storage unit for storing normal sequence reference information including at least one of a PLC input condition, a PLC output condition, an interlock condition, a sensor input sequence, an actuator output sequence, a servo position condition, a robot operation condition, an allowable time required for each normal operation step of the automation facility; and an AI sequence comparison unit for sorting the commissioning data in chronological order and comparing the actual operation sequence resulting from the sorted commissioning data with the normal operation sequence according to the normal sequence reference information to detect an abnormal section corresponding to at least one of an operation delay, a signal omission, an operation sequence error, a repeated stop, an interlock failure, an abnormal alarm, or a deviation from a set range that occurred in the actual operation sequence. Analysis unit; an abnormal cause classification unit that analyzes the correlation between PLC input signals, PLC output signals, internal device status, HMI alarm, interlock condition, sensor status, actuator status, servo motion data, robot operation status, and communication status in the above abnormal section, and classifies the cause of the above abnormal section into one or more of PLC logic abnormality, sensor signal abnormality, output or wiring abnormality, communication abnormality, servo position abnormality, robot operation standby, interlock condition mismatch, and setting parameter abnormality;A control condition correction candidate generation unit that generates a control condition correction candidate for at least one of a PLC timer setting value, a sensor input judgment condition, an interlock release condition, a servo position tolerance, a communication response waiting time, an alarm occurrence condition, and an HMI display condition, based on the classified abnormal cause and the normal sequence reference information; and a worker guidance unit configured to output the abnormal section, the classification result of the abnormal cause, the control data that served as the basis for the classification result, the field inspection target, and the control condition correction candidate to an HMI or a worker terminal, and to determine whether to apply the control condition correction candidate according to the worker's approval input;The equipment data collection unit further includes a time synchronization and event conversion unit that converts each data collected from a PLC, HMI, servo drive, robot controller, vision inspection equipment, and upper-level production management system having different communication cycles and time standards into a common time axis, classifies each data into the same production cycle unit based on at least one of a PLC scan cycle, an equipment cycle start signal, and an equipment cycle end signal, and converts control data included in each production cycle into event data including signal identification information, signal value, occurrence time, maintenance time, previous state, changed state, and identification information of a data generating device; the AI ​​sequence comparison analysis unit generates the actual operation sequence using the event data; the normal sequence reference information storage unit generates a normal sequence candidate using the occurrence order of control events and the time elapsed between events that commonly appear in a plurality of commissioning cycles completed normally, registers the normal sequence candidate that has received approval input from an operator or engineer as normal sequence reference information, and includes the equipment project name, customer, production line name, equipment name, PLC type, HMI type, PLC program version, HMI program version, and electrical drawing. An automation equipment anomaly diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis, characterized by storing at least one of a version, an I / O list version, and a commissioning date in correspondence with the normal sequence reference information, and, when at least one of the PLC program version, the HMI program version, the electrical drawing version, or the I / O list version is changed, generating new normal sequence reference information that is distinct from the normal sequence reference information prior to the change. Claim 2 delete Claim 3 delete Claim 4 The system for diagnosing abnormalities in automated equipment and correcting control conditions through AI-based PLC / HMI commissioning data analysis, wherein the abnormal cause classification unit extracts logical dependency relationships between input contacts, internal relays, timers, counters, comparison commands, output coils, interlock conditions, and alarm occurrence conditions included in a PLC program, combines the logical dependency relationships with the correspondence relationships between PLC addresses included in a PLC I / O list and field devices to generate an interlock causation graph representing cause-and-effect relationships between multiple control events, searches the interlock causation graph in reverse from the first deviation event where the normal operation sequence and the actual operation sequence first mismatched to extract multiple cause candidates that influenced the occurrence of the first deviation event, calculates cause reliability for each cause candidate using at least two of the following: temporal precedence, frequency of repeated occurrence, signal value deviation relative to a normal cycle, correlation with a related HMI alarm, and similarity with a past field action history, and outputs the priority of cause candidates based on the cause reliability, as well as the PLC address, interlock condition, and control event serving as the basis for extracting each cause candidate, through the operator guidance unit. Claim 5 In claim 4, the control condition correction candidate generation unit selects a control condition directly connected to a cause candidate whose cause reliability is greater than or equal to a set reference value as a control condition to be corrected; generates a plurality of control condition correction candidates based on at least one of a set value applied in a past normal commissioning cycle, a set value applied to equipment of the same type, the design tolerance of the control condition, and a set value stabilized after past field measures for the control condition to be corrected; generates a virtual commissioning sequence to which each control condition correction candidate is applied; evaluates whether the equipment cycle is completed, the time required for each stage, whether repeated stops occur, whether an alarm occurs, and whether safety interlocks are satisfied for the virtual commissioning sequence; selects a control condition correction candidate among the plurality of control condition correction candidates as an applicable candidate that does not change or bypass at least one safety condition among an emergency stop condition, safety door condition, light curtain condition, overload condition, and collision prevention condition while reducing the difference from the normal operation sequence; and generates approval information so that the applicable candidate can be reflected in the PLC or HMI only when an operator's approval input regarding the applicable candidate is received. Automated equipment anomaly diagnosis and control condition correction system through commissioning data analysis. Claim 6 In claim 5, the above abnormal cause classification unit and the above control condition correction candidate generation unit, when an output event that should be executed in the above normal operation sequence does not occur in the above actual operation sequence, set the output event that did not occur as a target event, perform a reverse search from the output coil or operation command corresponding to the target event in the interlock causation graph to extract at least one of the input contact, internal relay, timer completion condition, counter condition, servo position attainment condition, robot completion signal, communication completion signal, and safety interlock condition that blocked the occurrence of the target event, determine the minimum number of condition sets in which the signal transmission path to the target event is restored when the state of one or more of the extracted conditions changes as the minimum blocking cause set, and after excluding the safety input signal, emergency stop input signal, safety door input signal, light curtain input signal, motor overload input signal, and collision prevention input signal from the conditions included in the minimum blocking cause set from the correction target, the preset value of the PLC timer, the on-delay or off-delay value of the sensor input, the servo position attainment tolerance, the waiting time of the robot completion signal, the communication response waiting time, and the non-safety Generating the above control condition correction candidates limited to interlock release conditions, and for each of the above control condition correction candidates, performing a counter-real-world commissioning replay in chronological order by replaying control events in a state where the above control condition correction candidate is applied using at least three normal commissioning cycles collected prior to the point of an anomaly and at least one commissioning cycle in which an anomaly occurred, wherein the result of the above counter-real-world commissioning replay is: a. to generate the above target event; b. to complete subsequent events specified in the above normal operation sequence in a set order; c. to reduce the time required to reach the above target event or the total equipment cycle time compared to before correction; and d.An automated equipment fault diagnosis and control condition correction system through AI-based PLC / HMI commissioning data analysis, characterized by selecting a corresponding control condition correction candidate as a verification candidate only when all of the following conditions are satisfied: not changing the state and signal transmission path of the PLC logic corresponding to the above safety input signal, emergency stop input signal, safety door input signal, light curtain input signal, motor overload input signal, and collision prevention input signal; executing the selected verification candidate in a shadow mode for at least three consecutive equipment cycles, in which it is reflected only in a virtual output area without being reflected in the actual PLC output; verifying whether the output result of the virtual output area based on the input state of the actual PLC and the output result of the normal operation sequence are greater than or equal to a set match rate; and if the match rate is greater than or equal to a set standard, generating a correction approval package including the control condition before correction, the control condition after correction, the address of the PLC to be corrected, the affected interlock condition, the version identification value of the PLC program to be applied, the hash value of the program before correction, and rollback information for returning to the state before correction, and providing it to the operator as the final approval target.

Images