KR102684101B1 - System for providing facility predictive maintenance and tool breakage predict service using intelligent robotic process automation - Google Patents

Abstract

A facility anomaly detection service providing system using intelligent RPA is provided, and includes at least one monitoring terminal that collects at least one manufacturing environment data from facilities in the manufacturing environment, and the facility as a result of collecting and analyzing at least one manufacturing environment data. A collection unit that collects at least one manufacturing environment data collected from equipment in the manufacturing environment using a user terminal that outputs abnormal signs and RPA (Robotic Process Automation) from at least one monitoring terminal, and at least one manufacturing environment data It includes a detection service providing server that includes a detection unit that detects anomalies by detecting them using a pre-built anomaly detection model, and an alarm unit that transmits to the user terminal for predictive maintenance when an abnormality is detected.

Description

System for detecting equipment abnormalities and providing tool breakage prediction service using intelligent RP{SYSTEM FOR PROVIDING FACILITY PREDICTIVE MAINTENANCE AND TOOL BREAKAGE PREDICT SERVICE USING INTELLIGENT ROBOTIC PROCESS AUTOMATION}

The present invention relates to a system for providing equipment anomaly detection service using intelligent RPA. It provides a solution that detects abnormalities in equipment using intelligent RPA and records and stores the results.

A smart factory is a digital factory that connects, collects, and analyzes numerous sensor data by connecting sensors installed in various production facilities with IoT, and uses the keywords of the 4th Industrial Revolution: Internet of Things (IoT), artificial intelligence (AI), and machine learning ( It encompasses cutting-edge technologies such as machine learning (ML), big data, and 5G, and is a key field leading the digital transformation of the manufacturing sector. A smart factory is a factory that collects and analyzes process data based on ICT technology and controls itself according to the purpose. For efficient management, preventive maintenance is used to monitor signs of equipment failure in real time based on a large amount of collected data. ) In addition, predictive maintenance is possible through real-time monitoring of facilities, operation status monitoring, advance failure prediction, and maintenance. In addition, by using this predictive maintenance in industrial sites, it is possible to predict when, where, and how an accident will occur and respond before an accident occurs.

At this time, a method of using predictive maintenance in a smart factory was researched and developed. In this regard, the prior art, Korean Patent No. 10-2488820 (announced on January 17, 2023) and Korean Patent No. 10-2265459 (2021) Announcement on June 15, 2006) measures the energy of the facility with a sensor, calculates and determines the utilization and usage energy of the facility, controls predictive maintenance based on surrounding energy and energy utilization rate, and provides sensor data from the facility. And image data is collected, learning data is selected based on sensor data and image data, the learning data is grouped and labeled, a deep learning model is trained using the labeled learning data, and then the sensor is collected in real time. Configurations for predicting equipment defects using data and image data are disclosed.

However, in the former case, it is said that predictive maintenance is provided based on energy utilization rate, but the structure of how to model the artificial intelligence model for predictive maintenance is not disclosed. In the latter case, it is disclosed that a labeler groups and labels each data individually, but it takes a lot of manpower and time for humans to directly label each data. Accordingly, research and development of solutions that can minimize human intervention when building an abnormality detection model to perform predictive maintenance at manufacturing sites are required.

In one embodiment of the present invention, when collecting manufacturing environment data, RPA (Robotic Process Automation) is used to collect manufacturing environment data from facilities in the manufacturing environment, input the manufacturing environment data into the anomaly detection model, and then detect abnormalities. When an abnormality is detected, it is stored as manufacturing environment data and used to build big data, so that it can be applied to the abnormality detection model in a virtuous cycle and detect abnormality before the equipment breaks down. By detecting and predicting damage and wear, it is possible to provide an optimal production environment, and to provide a system that provides equipment anomaly detection services using intelligent RPA, which can increase sales by reducing defect rates and improving productivity. However, the technical challenge that this embodiment aims to achieve is not limited to the technical challenges described above, and other technical challenges may exist.

As a technical means for achieving the above-described technical problem, an embodiment of the present invention includes at least one monitoring terminal that collects at least one manufacturing environment data from equipment in the manufacturing environment, and collecting and analyzing at least one manufacturing environment data. As a result, at least one user terminal that outputs abnormal signs about equipment and a collection unit that collects at least one manufacturing environment data collected from equipment in the manufacturing environment using RPA (Robotic Process Automation) from at least one monitoring terminal. A detection service providing server that includes a detection unit that detects abnormalities by detecting manufacturing environment data using a pre-established abnormality detection model, and an alarm unit that transmits to the user terminal for predictive maintenance when abnormalities are detected. Includes.

According to one of the above-described means of solving the problem of the present invention, when collecting manufacturing environment data, RPA (Robotic Process Automation) is used to collect manufacturing environment data from equipment in the manufacturing environment, and manufacturing environment data is included in an abnormality detection model. After entering, abnormal symptoms are detected, and if an abnormal symptom is detected, it is saved as manufacturing environment data and used to build big data, so that it can be applied to the abnormal symptom detection model in a virtuous cycle, and if the equipment fails, By detecting abnormal signs before they occur and predicting damage and wear, an optimal production environment can be provided, and sales can be increased by reducing defect rates and improving productivity.

Figure 1 is a diagram illustrating a system for providing equipment anomaly detection service using intelligent RPA according to an embodiment of the present invention.
FIG. 2 is a block diagram illustrating a detection service providing server included in the system of FIG. 1.
Figures 3 and 4 are diagrams to explain an embodiment in which a facility anomaly detection service using intelligent RPA is implemented according to an embodiment of the present invention.
Figure 5 is an operation flowchart illustrating a method of providing equipment anomaly detection service using intelligent RPA according to an embodiment of the present invention.

Below, with reference to the attached drawings, embodiments of the present invention will be described in detail so that those skilled in the art can easily implement the present invention. However, the present invention may be implemented in many different forms and is not limited to the embodiments described herein. In order to clearly explain the present invention in the drawings, parts unrelated to the description are omitted, and similar parts are given similar reference numerals throughout the specification.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected," but also the case where it is "electrically connected" with another element in between. . In addition, when a part is said to "include" a certain component, this means that it does not exclude other components, but may further include other components, unless specifically stated to the contrary, and one or more other features. It should be understood that it does not exclude in advance the presence or addition of numbers, steps, operations, components, parts, or combinations thereof.

The terms “about,” “substantially,” and the like used throughout the specification are used to mean at or close to that value when manufacturing and material tolerances inherent in the stated meaning are presented, and are used to enhance the understanding of the present invention. Precise or absolute figures are used to assist in preventing unscrupulous infringers from taking unfair advantage of stated disclosures. The term “step of” or “step of” as used throughout the specification of the present invention does not mean “step for.”

In this specification, 'part' includes a unit realized by hardware, a unit realized by software, and a unit realized using both. Additionally, one unit may be realized using two or more pieces of hardware, and two or more units may be realized using one piece of hardware. Meanwhile, '~ part' is not limited to software or hardware, and '~ part' may be configured to reside in an addressable storage medium or may be configured to reproduce one or more processors. Therefore, as an example, '~ part' refers to components such as software components, object-oriented software components, class components, and task components, processes, functions, properties, and procedures. , subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided within the components and 'parts' may be combined into a smaller number of components and 'parts' or may be further separated into additional components and 'parts'. Additionally, components and 'parts' may be implemented to regenerate one or more CPUs within a device or a secure multimedia card.

In this specification, some of the operations or functions described as being performed by a terminal, apparatus, or device may instead be performed on a server connected to the terminal, apparatus, or device. Likewise, some of the operations or functions described as being performed by the server may also be performed in a terminal, apparatus, or device connected to the server.

In this specification, some of the operations or functions described as mapping or matching with the terminal mean mapping or matching the terminal's unique number or personal identification information, which is identifying data of the terminal. It can be interpreted as

Hereinafter, the present invention will be described in detail with reference to the attached drawings.

Figure 1 is a diagram illustrating a system for providing equipment anomaly detection service using intelligent RPA according to an embodiment of the present invention. Referring to FIG. 1, the facility abnormality symptom detection service providing system 1 using intelligent RPA may include at least one user terminal 100, a detection service providing server 300, and at least one monitoring terminal 400. You can. However, since the equipment anomaly detection service providing system 1 using the intelligent RPA shown in FIG. 1 is only an embodiment of the present invention, the present invention is not limitedly interpreted through FIG. 1.

At this time, each component of FIG. 1 is generally connected through a network (Network, 200). For example, as shown in FIG. 1, at least one user terminal 100 may be connected to the detection service providing server 300 through the network 200. In addition, the detection service providing server 300 may be connected to at least one user terminal 100 and at least one monitoring terminal 400 through the network 200. Additionally, at least one monitoring terminal 400 may be connected to the detection service providing server 300 through the network 200.

Here, the network refers to a connection structure that allows information exchange between each node, such as a plurality of terminals and servers. Examples of such networks include a local area network (LAN) and a wide area network (WAN). Wide Area Network, Internet (WWW: World Wide Web), wired and wireless data communication network, telephone network, wired and wireless television communication network, etc. Examples of wireless data communication networks include 3G, 4G, 5G, 3rd Generation Partnership Project (3GPP), 5th Generation Partnership Project (5GPP), 5G New Radio (NR), 6th Generation of Cellular Networks (6G), and Long Term Evolution (LTE). , WIMAX (World Interoperability for Microwave Access), Wi-Fi, Internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network) ), RF (Radio Frequency), Bluetooth (Bluetooth) network, NFC (Near-Field Communication) network, satellite broadcasting network, analog broadcasting network, DMB (Digital Multimedia Broadcasting) network, etc., but are not limited thereto.

In the following, the term at least one is defined as a term including singular and plural, and even if the term at least one does not exist, each component may exist in singular or plural, and may mean singular or plural. This should be self-explanatory. In addition, whether each component is provided in singular or plural form may be changed depending on the embodiment.

At least one user terminal 100 may be a terminal that outputs abnormal signs detected using a web page, app page, program, or application related to a facility abnormal sign detection service using intelligent RPA.

Here, at least one user terminal 100 may be implemented as a computer capable of accessing a remote server or terminal through a network. Here, the computer may include, for example, a laptop equipped with a navigation system and a web browser, a desktop, a laptop, etc. At this time, at least one user terminal 100 may be implemented as a terminal capable of accessing a remote server or terminal through a network. At least one user terminal 100 is, for example, a wireless communication device that guarantees portability and mobility, and includes navigation, personal communication system (PCS), global system for mobile communications (GSM), personal digital cellular (PDC), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet) ) It may include all types of handheld-based wireless communication devices such as terminals, smartphones, smartpads, and tablet PCs.

The detection service providing server 300 may be a server that provides a facility anomaly detection service web page, app page, program, or application using intelligent RPA. In addition, the detection service providing server 300 may be a server that detects anomalies after querying the manufacturing environment data of the equipment to an anomaly detection model, and when an anomaly is detected, it transmits it to the user terminal 100. It may be a server that assigns work to an engineer terminal (not shown). In addition, the detection service providing server 300 may be a server that retrains the anomaly detection model based on the results when result data is uploaded from the technician terminal.

Here, the detection service providing server 300 may be implemented as a computer that can connect to a remote server or terminal through a network. Here, the computer may include, for example, a laptop equipped with a navigation system and a web browser, a desktop, a laptop, etc.

At least one monitoring terminal 400 detects and monitors various manufacturing environment data generated from processes and facilities using a web page, app page, program, or application related to equipment anomaly detection service using intelligent RPA, and meets RPA requirements. It may be a terminal that transmits manufacturing environment data at the time or point in time.

Here, at least one monitoring terminal 400 may be implemented as a computer capable of accessing a remote server or terminal through a network. Here, the computer may include, for example, a laptop equipped with a navigation system and a web browser, a desktop, a laptop, etc. At this time, at least one monitoring terminal 400 may be implemented as a terminal capable of accessing a remote server or terminal through a network. At least one monitoring terminal 400 is, for example, a wireless communication device that ensures portability and mobility, and includes navigation, personal communication system (PCS), global system for mobile communications (GSM), personal digital cellular (PDC), PHS (Personal Handyphone System), PDA (Personal Digital Assistant), IMT (International Mobile Telecommunication)-2000, CDMA (Code Division Multiple Access)-2000, W-CDMA (W-Code Division Multiple Access), Wibro (Wireless Broadband Internet) ) It may include all types of handheld-based wireless communication devices such as terminals, smartphones, smartpads, and tablet PCs.

Figure 2 is a block diagram for explaining the detection service providing server included in the system of Figure 1, and Figures 3 and 4 are implementations of a facility anomaly detection service using intelligent RPA according to an embodiment of the present invention. This drawing is for explaining an embodiment.

Referring to Figure 2, the detection service providing server 300 includes a collection unit 310, a detection unit 320, an alarm unit 330, an access expansion unit 340, a re-learning unit 350, and a no-coding unit. (360) and may include a dashboard unit (370).

The detection service providing server 300 according to an embodiment of the present invention or another server (not shown) operating in conjunction with at least one user terminal 100 and at least one monitoring terminal 400 uses intelligent RPA. When transmitting an anomaly detection service application, program, app page, web page, etc., at least one user terminal 100 and at least one monitoring terminal 400 are equipped with an equipment anomaly detection service application, program using intelligent RPA. , you can install or open app pages, web pages, etc. Additionally, a service program may be run on at least one user terminal 100 and at least one monitoring terminal 400 using a script executed in a web browser. Here, a web browser is a program that allows the use of web (WWW: World Wide Web) services and refers to a program that receives and displays hypertext written in HTML (Hyper Text Mark-up Language), for example, Chrome. , Microsoft Edge, Safari, FireFox, Whale, UC Browser, etc. Additionally, an application refers to an application on a terminal and includes, for example, an app running on a mobile terminal (smartphone).

Referring to FIG. 2 , the collection unit 310 may collect at least one manufacturing environment data collected from equipment in the manufacturing environment from at least one monitoring terminal 400 using RPA (Robotic Process Automation). At least one monitoring terminal 400 may collect at least one manufacturing environment data from facilities within the manufacturing environment. At this time, the manufacturing environment data may be data collected from at least one sensor among a temperature sensor, humidity sensor, audio sensor, vibration sensor, current sensor, and image sensor, but is not limited to those listed and is not excluded for reasons not listed. .

The detection unit 320 may detect an anomaly by detecting at least one manufacturing environment data using a pre-built anomaly detection model. At this time, the abnormality detection model may be a predictive maintenance model. At this time, predictive maintenance is a technology that predicts the possibility of failure and remaining lifespan of factory facilities and equipment based on artificial intelligence algorithms. Vibrations and currents that occur while facilities and equipment are in operation are analyzed based on data, abnormal signs are checked in real time and notified to managers, so measures can be taken before the equipment breaks down, reducing the risk of factory downtime. Damage from low operation rates can be reduced. At this time, the stages of predictive maintenance are standardized and may be as shown in Table 2, for example. To this end, each manufacturing environment data is processed in real time to perform modeling, then predictive analysis is performed using an artificial intelligence algorithm. After real-time monitoring, repair and replacement times are predicted to diagnose and take action.

step detail Step 1 Data acquisition -Detect abnormal signs Step 2 Diagnosis -Diagnose whether there is a defect and the cause of the defect
-Severity classification Step 3 prognosis -Prediction of future defect progression
-Prediction of remaining useful life Step 4 Presentation and Action -Suggesting solutions to problems and taking action Step 5 postmortem -Check problem resolution
-Post-condition diagnosis

For example, process data such as pressure, temperature, and pH levels indicate the security status of the physical manufacturing process. Analyzing control process data means analyzing these process data to detect abnormalities. You can also perform processes such as detecting abnormal signs by analyzing process data and making predictions, or predicting the results of control commands through control command analysis. At this time, the following will explain the control and predictive maintenance of the air conditioning system as an example. Of course, predictive maintenance according to an embodiment of the present invention is not limited to the air conditioning system, and of course, the following concepts can be applied and utilized to each smart factory. <Predictive Maintenance>

<Air conditioning system>

First, the model for fine dust control is defined as follows. To express the concentration of fine dust in an arithmetic expression, the concentration of PM2.5 (ultrafine dust) with a diameter smaller than 2.5㎛ is expressed against time t as it ^{(1) (㎍/㎥), and PM10 with a diameter smaller than 10㎛ is expressed as it (1)} (㎍/㎥). (Fine dust) concentration can also be expressed with respect to time t as it ⁽²⁾ (㎍/㎥).

Then, for the previously defined time t, K blowers and L air conditioning equipment use power equal to vt(1),...,vt(K) and wt(1),...,wt(L), respectively. Thus, the concentration of fine dust is controlled under these conditions. At this time, when the power price given at time t is calculated as pt, the total power cost ct is equal to Equation 1. Here, pt(X) includes other devices except blowers and air conditioning equipment.

As time changes, the fine dust concentration at time change t+1 changes to i(1)t+1 and i(2)t+1, respectively, due to automatic control of air conditioning equipment and blowers.

In order to apply artificial intelligence learning, that is, reinforcement learning with rewards, to the system modeling defined in Equation 1, the process of fine dust adjustment through control of air conditioning equipment and blowers must be expressed as a causal relationship before and after the operation of the air conditioning equipment and blower. And for this, it must be expressed as a probabilistic model based on the Markov Decision Process (MDP). The Markov decision-making process can be composed of five elements: S, A, P, R, and T.

First, S is the current state and includes the current time t, history fine dust concentration i(1)t, i(2)t, and external environmental factors such as temperature, humidity, and fine dust concentration that affect change. , elements that are correlated with changes in fine dust concentration within subway stations are included. A refers to the action currently performed by the blower and air conditioning equipment, and in the above modeling, the action at at time t can be defined as Equation 2.

P is a transition function and can be defined as the probability of moving to st+1 when an action called at is taken from the current state st. At this time, in modeling based on the Markov decision process, the transition probability of changing the state from st to st+1 is the Markov Property, which can be defined as being determined only by the previous state, st, among all the various stages of the past. Follow. These properties are as shown in Equation 3.

R is reward, which means the reward received when the action at is taken in state st. The reward rt obtained at time t can be expressed as a function of st and at as shown in Equation 4.

The relationship between the change in fine dust concentration within the station and the resulting power consumption due to state factors, air conditioning equipment, and blower control that are correlated with the concentration of fine dust within the station is defined as a compensation function.

T is the discount factor, and τ∈(0,1) corresponds to a variable that controls the importance between the current reward and the reward that can be obtained in the future. If the depreciation rate has a small value, it means that the reward obtained now can be considered more valuable than the reward that can be obtained in the future. This is because as time passes, the depreciation rate increases, further reducing the value of future rewards. If the time considered in machine learning reinforcement learning is finite, τ=1 (maximum) can be set. When predictive maintenance is applied to an air conditioning system, numerous functions are implemented within the system, such as detection of equipment abnormalities, diagnosis of vibration and noise, failure prediction, and life prediction. In one embodiment of the present invention, predictive maintenance of an air conditioning system can be performed as an example of predictive maintenance.

<Data collection>

Data can be collected from subway management agencies through information disclosure requests (open.go.kr). There are many types of failures in subway air conditioning equipment, and each organization maintains a list of failure types or chain failure types. According to this failure list, there are 36 major failure types and approximately 330 failure types in detail. Among these, the failure types in which chain failures occur based on problem symptoms can be organized according to frequency as shown in Table 2 below.

Problem Symptoms Chain Failure Surface Fatigue gear Chemical Wear gear Scuffing gear Abrasive Wear gear Activation Failure gear Excessive Vibration gear Excessive Noise gears, bearings Design Drafting Error gear Sudden Knocking Noise mechanical system High Temperature of Sleeve Bearings mechanical system High Temperature of Rolling Bearings mechanical system Tooth Breakage gear Plastic Deformation gear Case Cracking gear Excessive Vibration gear Excessive Noise gears, bearings Surface Fatigue gears, bearings Chemical Wear gears, bearings Scuffing gears, bearings Abrasive Wear gears, bearings Tooth Breakage gears, bearings Excessive Noise mechanical system Activation Failure electrical system Excessive Vibration assembly Sudden Knocking Noise mechanical system High Temperature of Sleeve Bearings mechanical system Repetitive Belt Breakage operating conditions Design/Installation Sudden Belt Wear wear and tear operating conditions Design/Installation Crack in The Belt wear and tear operating conditions Design/Installation

According to the above-described Markov decision-making process model and the chain failure pattern in Table 1, it can be controlled to maintain optimal fine dust conditions based on predictive maintenance. At this time, the action at with random probability ε is selected, and the other at is expressed as Equation 5.

At this time, the st+1 state is calculated using Equation 6 below. At this time, λ(t) is the failure rate function.

Then, after calculating the compensation τt and calculating the approximate slope of the loss function, each parameter is updated and repeated for a preset period.

<Average number of failure days per year after introduction of predictive maintenance>

When predictive maintenance is introduced, it is possible to predict continuous chain failures in the system, thereby reducing the average number of days of failure per year. This is also related to the reliability of the system, and the series structure is often used in reliability analysis. This structure means that when any one of the system components fails, there is a chain of failures. The failure time Tseq of the series structure is calculated as Equation 7.

Here, Ti, i=1,...,n is the failure time of the ith component, and T1,...,Tn are assumed to be independent. The reliability function of Tseq can be calculated as Equation 8 and is calculated as the product of each probability.

At this time, if the failure rate function of the ith component is called λ(t) and the inverse function is obtained, Ri(t) can be calculated as shown in Equation 9.

Therefore, the reliability function of the serial structure system is as shown in Equation 10.

From this relation, the failure rate function λ(t) of the system can be calculated as the sum of the failure rate functions of each component.

According to this formula, it can be seen that the life distribution of a serial system in which the life distribution of each component follows an exponential distribution has a failure rate equal to the sum of the failure rates of each component, which also follows an exponential distribution. Applying Equation 12 to find the average lifespan of a series structure system, and assuming that the lifespan distribution of each component follows an exponential distribution, MTTF (Mean Time to Failure) is calculated through the same process as Equation 13.

For this calculation, assume λi(t)=λi>0 and i=1,...,n for all 0<t<∞. Here, MTTFi, i=1,...,n represents the average failure time of the ith component. Equation 13 shows that when the failure time of each component follows an exponential distribution, the average failure time of a system with a serial structure is the harmonic average of the average failure times of each component.

In particular, if λ1=...=λn=λ0, MTTF1=...=MTTFn=MTTF0 is established, and therefore Equation 13 can be simplified to MTTF=MTTF0/n. In other words, the average lifespan of a system under the average lifespan of all components cited in the calculation may be approximately 1/n of the average lifespan of the parts. Through this, it is possible to calculate the average number of days of failure per year after introducing the predictive maintenance system.

<Rotation equipment>

In addition to the Markov decision-making and reinforcement learning described above, in the case of rotating equipment, it is also possible to analyze vibration signals generated from the rotating equipment and perform predictive maintenance using the Auto Encoder (AE) algorithm. Due to recent developments in deep learning technology, machine learning and deep learning technology have been applied to predict failures in rotating equipment, showing high prediction accuracy. Outliers can also be detected for unlabeled data in rotating equipment using the autoencoder algorithm described above. Accordingly, in addition to Markov decision-making and reinforcement learning, outliers can also be detected using an autoencoder. To this end, it is possible to predict failure using an autoencoder by analyzing vibration signals from data collected directly through the rotor and DAQ (Data Acquisition).

<Data collection>

There are various methods such as machine heat detection and noise analysis to predict failure or detect abnormal conditions in rotating equipment, but the most accurate and fastest way to detect abnormal conditions is by predicting defects using vibration signals. Accordingly, data on the normal state and abnormal state of the rotating body can be extracted using the vibration signal and used for predictive maintenance. Vibration data can be collected by attaching an acceleration sensor to a rotating facility and using DAQ (Data Acquisition) to collect vibration signals. Data samples can be collected by changing the RPM of the rotating equipment. In addition, even if the data is extracted and visualized so that it can be checked in real time, it is difficult to distinguish with the naked eye. Accordingly, predictive maintenance can be performed by conducting data analysis using deep learning. In addition, the normal state is defined as the state in which the rotating equipment operates without weight running. An abnormal state is defined as an unbalanced load state that can occur due to imbalance or defects in components. Accordingly, in order to collect data, the state in which an unbalanced load exists by artificially adding a weight to the rotating body can be defined as an abnormal state and data can be extracted.

<Feature extraction>

Well-known physical and statistical state indicators can be used to use as input for deep learning from the measured time series vibration data, and these can be as shown in Table 3 below, but it is obvious that it is not limited to this indicator and various indicators can be used. something to do.

characteristic official Benefit Absolute Mean

Detect changes in the overall level of vibration signals Variance

Detection of spread of vibration signal Skewness

Detection of asymmetry in vibration signals Kurtosis

Flatness detection of vibration signals Peak-to-Peak

Measures the full range of vibration amplitudes RMS

Measurement of average energy of vibration signal Impulse factor

Measurement of high frequency content of vibration signal

<Data Preprocessing> Physical and statistical state indicators extracted from time series vibration data can be normalized by using the Min-Max Scaling method. Using the min-max scaling method ensures that all input data has values between 0 and 1, helping the model make more accurate predictions.

<Anomaly detection>

You can model a model that reconstructs the input using an autoencoder. An autoencoder is an unsupervised learning neural network with the same input and output structure. It operates by compressing input data to create a latent vector and then restoring it to reconstruct the input data. Anomaly detection can be performed using an autoencoder, a representative unsupervised learning model. In order to perform anomaly detection, the autoencoder model must first be trained with normal data. After learning, abnormal data is input into the model, the reconstruction error is calculated using MSE (Mean Squared Error), and compared to a preset threshold, if the reconstruction error is greater than the threshold, it is determined to be abnormal data.

The alarm unit 330 may transmit to the user terminal 100 for predictive maintenance when abnormal signs are detected. The user terminal 100 may output abnormal signs about equipment as a result of collecting and analyzing at least one manufacturing environment data.

The access expansion unit 340 may collect manufacturing environment data from at least one cloud or database using RPA.

When an on-site inspector is deployed for a detected abnormality, the re-learning unit 350 uploads the action result data from the technician terminal and includes the action result data as a dataset of the anomaly detection model to re-create the abnormality detection model. It can be learned.

When a cell corresponding to at least one step is selected and dropped between different steps so that the order of at least one step is changed, the no-coding unit 360 stores the cell in the changed order. Accordingly, the order of execution code can be changed so that RPA operates. At this time, automation and no-coding of RPA (Robotic Process Automation) are described in detail in Korean Patent No. 10-2543064 (announced on June 13, 2023), which is the applicant's pre-registered patent, but is briefly described for understanding. Do this. For information not described in an embodiment of the present invention, refer to the applicant's previously registered patent.

For example, let's assume that A, an employee of a fish cake manufacturer that mass-produces fish cakes, wants to identify abnormal signs in the fish cake fryer equipment among the manufacturing facilities. Assume that Employee A is an ordinary person, not a coding expert. At this time, Employee A must select an artificial intelligence algorithm to identify abnormalities, but this process is already difficult. I don't know which artificial intelligence algorithm to choose or what data to input. At this time, RPA possesses in advance artificial intelligence algorithms and types of data sets that can identify abnormal signs of each manufacturing facility, and when the user inputs only the data [Fryer] and [Symptoms of abnormalities], the type of data set to be input is determined. Select an artificial intelligence algorithm that can learn and verify it. In this way, if the user inputs only [Fryer] and [Symptoms of Abnormalities] and presses the create button, the user's work is completed and the flow for the artificial intelligence algorithm is output.

At this time, the types of data sets for identifying abnormalities in the fryer are the camera filming the fryer, the current output from the fryer, and the vibration of the fryer. Even if the user does not collect all of this, each manufacturing environment data includes the type of equipment and ID. , Since the type of detected data and the detected value are all tagged and labeled, this data can be collected from the database, and then an artificial intelligence algorithm for image analysis (Algorithm #1) can be used to determine whether there is an abnormality in the vibration data. Loads an artificial intelligence algorithm (Algorithm #2) that can identify abnormalities in current data (Algorithm #3), and data that matches each artificial intelligence algorithm, that is, Algorithm #1, is loaded with data collected from the camera. By inputting image data, vibration data into Algorithm #2, and current data into Algorithm #3, the artificial intelligence algorithm (Algorithm #1, #2, #3) performs learning and verification to enable modeling. And, when learning and verification are completed, the flow created by the user, that is, information about whether the fryer is broken or not, is output to the user terminal 100. The artificial intelligence algorithm created in this way continuously guides users to the results of monitoring the fish cake fryer and predicts whether there is a breakdown. The process continues until the user deletes this artificial intelligence algorithm. Each cell is a step, and if you want to eliminate the artificial intelligence algorithm itself, just drag and delete all cells. If you want to delete just one step, select the cell with the mouse and press the DEL button to delete it. If you want to add a step, you can add it from the activity list.

When a cell corresponding to at least one step is selected and dropped between different steps so that the order of at least one step is changed, the no-coding unit 360 allows the RPA to operate according to the changed order. The order of execution code can be changed. For example, let's assume there is a flow of [A→B→C→D→E]. The flow includes each step. Assuming that the user wants to change level B to B', select the activity in the activity list corresponding to B', drag and drop it into the cell where level B exists, and the user's work is completed. Even a person who does not know the C language can complete the maintenance work of modifying the flow of [A→B→C→D→E] into the flow of [A→B'→C→D→E]. At this time, the activity list is written in text format, but you can overlay it with a guide, explanation, or graphical GUI to intuitively select which task you want to add. Alternatively, you can select an activity in the form of a keyword search. In addition, the steps in the flow can be easily changed by overwriting (changing) the activity in the list in the left area with the cell corresponding to the activity in the right area or inserting (adding) the cell corresponding to the activity. At this time, a cell refers to a box in which each activity is written.

The dashboard unit 370 may provide a dashboard to visualize and output the results of anomaly detection in the user terminal 100.

Hereinafter, the operation process according to the configuration of the detection service providing server of FIG. 2 described above will be described in detail using FIGS. 3 and 4 as an example. However, it will be apparent that the embodiment is only one of various embodiments of the present invention and is not limited thereto.

Referring to Figure 3a, RPA collects manufacturing environment data from the monitoring terminal 400 (NeoCollector), inputs it into the built anomaly detection model (NeoRPA&NeoAI), and visualizes the results (NeoBoard). Looking at an embodiment applied to a CNC, it may be as shown in Figure 3b, and based on the actions taken by the engineer terminal 500, the anomaly detection model can continuously re-learn, and the information on [cause-process-result] Since time series data can be collected, it is possible to lay the foundation for performing predictive maintenance using time series analysis algorithms such as LSTM. For this purpose, the data collection device shown in Figure 4a, automated programming tool, AI engine shown in Figure 4b, and web visualization service provided by the applicant's company can be used. The visualization service may be provided as shown in Figure 4c, but is not limited thereto. The AI engine may be as shown in Figure 4D, and the RPA may be applied as shown in Figure 4E and provide an automated process as shown in Figure 4F.

Matters that are not explained regarding the method of providing facility anomaly detection service using the intelligent RPA of FIGS. 2 to 4 are the same as those previously described regarding the method of providing facility anomaly detection service using intelligent RPA through FIG. 1. Since it can be easily inferred from the explained contents, the following description will be omitted.

FIG. 5 is a diagram illustrating a process in which data is transmitted and received between each component included in the equipment anomaly detection service providing system using the intelligent RPA of FIG. 1 according to an embodiment of the present invention. Hereinafter, an example of the process of transmitting and receiving data between each component will be described with reference to FIG. 5, but the present application is not limited to this embodiment, and the process shown in FIG. 5 according to the various embodiments described above It is obvious to those skilled in the art that the process of transmitting and receiving data can be changed.

Referring to FIG. 5, the detection service providing server collects at least one manufacturing environment data collected from equipment in the manufacturing environment from at least one monitoring terminal using RPA (Robotic Process Automation) (S5100).

In addition, the detection service providing server detects anomalies by detecting at least one manufacturing environment data with a pre-built anomaly detection model (S5200), and if an abnormality is detected, the user terminal performs predictive maintenance. Transmit to (S5300).

The sequence between the above-described steps (S5100 to S5300) is only an example and is not limited thereto. That is, the order between the above-described steps (S5100 to S5300) may change, and some of the steps may be executed simultaneously or deleted.

Matters not explained regarding the method of providing facility anomaly detection service using the intelligent RPA of FIG. 5 are the same as those previously described regarding the method of providing facility anomaly detection service using intelligent RPA through FIGS. 1 to 4. Since it can be easily inferred from the explained contents, the following description will be omitted.

The method of providing equipment abnormality symptom detection service using intelligent RPA according to an embodiment described in Figure 5 is also in the form of a recording medium containing instructions executable by a computer, such as an application or program module executed by a computer. It can be implemented. Computer-readable media can be any available media that can be accessed by a computer and includes both volatile and non-volatile media, removable and non-removable media. Additionally, computer-readable media may include all computer storage media. Computer storage media includes both volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data.

The method of providing equipment anomaly detection service using intelligent RPA according to an embodiment of the present invention described above includes an application installed by default on a terminal (this may include programs included in the platform or operating system, etc., which are installed by default on the terminal). It may be executed by, and may also be executed by an application (i.e. program) installed by the user directly on the master terminal through an application providing server such as an application store server, an application, or a web server related to the service. In this sense, the method of providing equipment anomaly detection service using intelligent RPA according to an embodiment of the present invention described above is implemented as an application (i.e., program) installed by default on the terminal or directly installed by the user, and is implemented as an application (i.e., program) installed by default on the terminal or directly installed by the user. It may be recorded on a computer-readable recording medium.

The description of the present invention described above is for illustrative purposes, and those skilled in the art will understand that the present invention can be easily modified into other specific forms without changing the technical idea or essential features of the present invention. will be. Therefore, the embodiments described above should be understood in all respects as illustrative and not restrictive. For example, each component described as single may be implemented in a distributed manner, and similarly, components described as distributed may also be implemented in a combined form.

The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modified forms derived from the meaning and scope of the claims and their equivalent concepts should be construed as being included in the scope of the present invention. do.

Claims (7)

At least one monitoring terminal that collects at least one manufacturing environment data from equipment in the manufacturing environment;
a user terminal that outputs abnormal signs about the equipment as a result of collecting and analyzing the at least one manufacturing environment data; and
A collection unit that collects at least one manufacturing environment data collected from equipment in the manufacturing environment using RPA (Robotic Process Automation) from the at least one monitoring terminal, and detects pre-established abnormalities in the at least one manufacturing environment data A detection service providing server including a detection unit that detects an abnormality by detecting a model, and an alarm unit that transmits the abnormality to the user terminal for predictive maintenance when the abnormality is detected;
Including,
The detection service providing server,
an access extension unit that collects the manufacturing environment data from at least one cloud or database using the RPA;
When an on-site inspector is deployed for a detected anomaly, a re-learning unit that uploads action result data from the technician terminal and includes the action result data as a dataset of the anomaly detection model to retrain the anomaly detection model. ;
Provides at least one activity corresponding to at least one executable code to create, modify, and delete a flow including at least one step performed in the RPA, and the at least one A no-coding unit that allows the RPA to run according to the changed flow when an activity is inserted into a step in the flow using a drag and drop method;
Including,
The no-coding unit,
When selecting a cell corresponding to the at least one step and dropping it between different steps so that the order of the at least one step is changed, the order of the execution code so that the RPA operates according to the changed order change ,
The no-coding unit,
When applied to the flow of [ABCDE], the flow includes each step, and if the user wants to change step B to B', select the activity in the activity list corresponding to B' and enter the cell where step B exists. By dragging and dropping into the (Cell), the user's work is completed. At this time, the activity list is written in text format, and a guide, explanation, or graphical GUI is overlaid to intuitively let you know what kind of work you want to add. You can select an activity in the form of a keyword search, and you can overwrite (change) the activity in the list in the left area with the cell corresponding to the activity in the right area by inserting (adding) a cell corresponding to the activity. You can change the steps in the flow,
The collection unit 310 may collect at least one manufacturing environment data collected from equipment in the manufacturing environment using RPA (Robotic Process Automation) from at least one monitoring terminal,
The monitoring terminal is capable of collecting at least one manufacturing environment data from equipment in the manufacturing environment,
The manufacturing environment data is,
Data collected from at least one sensor among temperature sensor, humidity sensor, audio sensor, vibration sensor, current sensor, and image sensor,
The detection unit 320 detects anomalies by detecting at least one manufacturing environment data using a pre-built anomaly detection model,
The abnormality detection model is a predictive maintenance model,
The predictive maintenance model measures the possibility of failure and remaining lifespan of factory facilities or equipment based on an artificial intelligence algorithm, and analyzes the vibration and current values that occur while the facility and equipment are in operation based on data to detect abnormal signs. Check in real time and notify the manager,
The predictive maintenance model processes each manufacturing environment data in real time to perform modeling, then performs predictive analysis using an artificial intelligence algorithm, predicts repair and replacement times after real-time monitoring, and makes diagnoses and measures.
The predictive maintenance model analyzes data including measured pressure, temperature, and pH values to detect abnormalities when they deviate from a preset standard range, and detects abnormalities by analyzing process data and making predictions. and predict the result of the control command through control command analysis,
The predictive maintenance model performs the functions of detecting equipment condition abnormalities, diagnosing vibration and noise, predicting failure, and predicting lifespan, and performing predictive maintenance of the air conditioning system,
The predictive maintenance model is,
In order to apply artificial intelligence learning, that is, reinforcement learning with rewards, a probabilistic model based on the Markov Decision Process (MDP) is used to coordinate the fine dust adjustment process through control of air conditioning equipment and blowers. Expressed as a causal relationship before and after the operation of equipment and blowers,
The Markov decision-making process consists of five elements: S, A, P, R, and T,
The predictive maintenance model is,
When applied to rotating equipment, vibration signals generated from rotating equipment are analyzed, predictive maintenance is performed using an auto encoder (AE) algorithm, and the auto encoder algorithm is used to detect abnormalities in unlabeled data from rotating equipment. In addition to Markov decision-making and reinforcement learning, outliers are detected using an autoencoder. To this end, vibration signals are analyzed from data collected directly through the rotating body and DAQ (Data Acquisition) to predict failure using an autoencoder. do,
The detection service providing server,
a dashboard unit providing a dashboard to visualize and output the results of the abnormality detection on the user terminal;
It further includes,
The detection service providing server 300 is a server that detects anomalies after querying the manufacturing environment data of the equipment to an anomaly detection model, and when an anomaly is detected, it transmits it to the user terminal 100 to the technician terminal. It is a server that assigns tasks, and when result data is uploaded from the technician terminal, it is a server that retrains the anomaly detection model based on the results. A facility anomaly detection service providing system using intelligent RPA. delete delete delete delete delete delete

Images