KR102578489B1 - Digital transformation prediction system based on big data and machine learning - Google Patents

Abstract

The present invention relates to a digital transformation prediction system based on big data and machine learning, which detects an operation signal value from a client that is a measurement target using one or more signal receivers and analyzes the operation signal value to determine the operation of the measurement target. A prediction system comprising: a signal receiver that communicates with at least one external client and receives data from the external client; A data collection unit that collects data obtained through the signal receiver that receives a signal from the client, and collects data collected at a certain period in a structured or unstructured form; A pattern detection unit that analyzes the pattern of the data collected by the data collection unit and compares the analyzed pattern with a standard pattern to detect abnormal patterns, and detects the usage amount of the data collected by the data collection unit and compares it with the standard usage amount. It is comprised of an abuse detection unit that detects abuse data corresponding to abnormal usage, and a time series detection unit that analyzes the data collected by the data collection unit in a time-series manner and detects abnormal data from the time-series data, and each detection detected here. an operation server that operates the client through signals; Each data analyzed through pattern detection, abuse detection, and time series detection in the data detected by the operation server is stored in the big data DB, and each data stored in the big data DB is learned to generate normal data and abnormal data. machine learning department; an administrator terminal that processes control commands to the client or sends control commands to the responsible worker according to the results detected through the operation server; and an external terminal that provides reference data and weight values for each data analyzed by the operation server, wherein the abuse detection unit determines the criteria based on the amount of usage of the client in the data collected through the data collection unit. The usage amount and excess usage amount are detected, and the excessive usage amount is compared with the standard usage amount and the excess usage amount based on time, daily, weekly, and monthly periods to determine whether the client is abusing the client.

Description

Digital transformation prediction system based on big data and machine learning}

The present invention relates to a digital transformation prediction system based on big data and machine learning, and more specifically, to early predict and manage failures, errors, security, abnormal signals, monitoring, and system control that occur in various industrial environments. and a digital transformation prediction system based on big data and machine learning that can be operated.

In modern society, convenience resulting from the development of information systems cannot be ruled out. In particular, artificial intelligence, which is being used in various fields these days, is an example. Artificial intelligence, which is the basis for automation functions, is distributed in various markets.

Technology that predicts failures using artificial intelligence can extract only necessary core data through direct definition of failure requirements for each IT resource component, and provides the function of frequently adding necessary data during the learning process. Additionally, you can learn about failure prediction information based on real-time operation status information for each IT resource component rather than based on system logs.

Deep Learning, an example of artificial intelligence, unlike machine learning, minimizes human intervention and learns the data as it is, so the machine learns the characteristics of the data on its own. Deep Learning is one of the Artificial Neural Network techniques and is a model that has multiple hidden layers between the input layer and the output layer. Deep Learning generates results with high accuracy from a large number of data through a neural network structure and has the advantage of requiring no prior understanding of a specific environment.

In addition, various methodologies in the field of artificial intelligence were applied to compare and analyze existing methodologies and predictive power. The field of artificial intelligence, represented by terms such as machine learning and deep learning, uses computer engineering to develop a system in which computers, like the human brain, go through a learning process and make decisions such as prediction processes. it means.

In the past, it did not receive attention due to physical limitations in processing the number of diverse and simultaneous cases, but recently, the deep learning system represented by Google's 'AlphaGo' has reached the level of actual human judgment. It is receiving great attention around the world by proving that the speed can be equal to or even surpass it.

Big data and artificial intelligence are receiving a lot of attention in various fields as core technologies of the 4th Industrial Revolution, but there are still not many cases of their application in research in the management and financial fields. Therefore, this study attempts to apply this new methodology to the management system. Artificial intelligence has the characteristic of improving prediction ability as the amount of learning data increases, so better results can be expected in the prediction process that uses big data such as text data as the source.

Meanwhile, the Fourth Industrial Revolution is emerging as the Internet-based connection technology between all objects is developed. One of the many technological fields leading this fourth industrial revolution is smart factory technology. Smart factory refers to a comprehensive set of technologies that connects and easily manages factory machines and facilities. One of the many technologies that make up this smart factory is the technology to manage machines. In other words, it monitors whether the machine is operating normally and predicts and manages errors in advance.

In the conventional case, in order to predict machine malfunctions, a threshold defined in the form of an absolute value independent of time is set, and the machine's operation data is collected frequently, and when operation data that exceeds the threshold is generated, a specific error occurs. It is judged that it has been done. The conventional threshold is related to the allowable setting of the machine for production. In other words, it is a combination of the upper and lower limits of operation to prevent defects in the produced product. And, the conventional threshold is a fixed value (an absolute value that is fixed regardless of time within each section set by the worker) directly set by the factory worker. Therefore, if an operator wants to lower the defect rate by arbitrarily lowering the quality of a product, he or she only needs to set the upper and lower limits of the threshold to be wide. In the opposite case, the upper and lower limits of the threshold can be set narrow. In addition, a threshold value is set for a time section during which a product is manufactured during a cycle corresponding to one cycle in which the machine operates, and no separate threshold value is set for a time section unrelated to product manufacturing. Therefore, although it is possible to determine whether a defect has occurred in the product, it is difficult to accurately determine whether there is an error in the machine.

With the recent development of IoT (Internet of Things) technology, various devices (mobile phones, mechanical equipment, sensors, home appliances, etc.) are connected to the Internet to provide users with more valuable information and services.

Therefore, there is a need to develop a data extraction and transmission method that can reduce power consumed by the sensor and improve data transmission speed by extracting and transmitting only meaningful information data from the information data measured by the IoT-based sensor.

KR 10-2281431 KR 10-2208696 KR 10-1538843 KR 10-1856543 KR 10-2373787 KR 10-2180149 KR 10-1802866 KR 10-2384742

The present invention to solve the above problems is a failure type prediction system that detects abnormal situations in mechanical equipment and predicts the time of failure in advance by inferring the situation or pattern in which the failure occurred through big data analysis. The purpose is to provide a solution that predicts potential failures or various operating functions of mechanical equipment using data.

In particular, the main purpose of the present invention is to detect various data characteristic information such as abuse detection and symptom detection, and to provide the purpose of predicting and detecting specific characteristic information through various types of client data.

In addition, another purpose of the present invention is to identify the status of the mechanical equipment based on data collected in real time, so that the status of the mechanical equipment can be determined more objectively and accurately, regardless of the experience and skill level of the management personnel. The purpose is to provide a failure prediction analysis system for mechanical equipment using big data analysis.

Another purpose of the present invention is to reduce the manpower, time and cost required for maintenance by identifying not only the current operating status of mechanical equipment but also signs of failure in advance, and increasing the lifespan and operation rate of mechanical equipment through appropriate preventive inspection. The purpose is to provide a failure prediction analysis system for mechanical equipment using big data analysis that improves customer satisfaction by improving customer satisfaction.

The present invention for achieving the above object is a prediction system that detects an operation signal value from a client that is a measurement target using one or more signal receivers and predicts the operation of the measurement target by analyzing the operation signal value, A signal receiver that communicates with at least one external client and receives data from the external client; A data collection unit that collects data obtained through the signal receiver that receives a signal from the client, and collects data collected at a certain period in a structured or unstructured form; A pattern detection unit that analyzes the pattern of the data collected by the data collection unit and compares the analyzed pattern with a standard pattern to detect abnormal patterns, and detects the usage amount of the data collected by the data collection unit and compares it with the standard usage amount. It is comprised of an abuse detection unit that detects abuse data corresponding to abnormal usage, and a time series detection unit that analyzes the data collected by the data collection unit in a time-series manner and detects abnormal data from the time-series data, and each detection detected here. an operation server that operates the client through signals; Each data analyzed through pattern detection, abuse detection, and time series detection in the data detected by the operation server is stored in the big data DB, and each data stored in the big data DB is learned to generate normal data and abnormal data. machine learning department; an administrator terminal that processes control commands to the client or sends control commands to the responsible worker according to the results detected through the operation server; and an external terminal that provides reference data and weight values for each data analyzed by the operation server, wherein the abuse detection unit determines the criteria based on the amount of usage of the client in the data collected through the data collection unit. The usage amount and excess usage amount are detected, and the excessive usage amount is compared with the standard usage amount and the excess usage amount based on time, daily, weekly, and monthly periods to determine whether the client is abusing the client.

In addition, the pattern detection unit detects a pattern of data collected through the data collection unit, and defines a pattern that exceeds a certain range among the patterns compared with the reference pattern as the first pattern, and further exceeds the first pattern. A symptom detection unit that defines a range as a second pattern and then compares the first pattern and the second pattern with the reference pattern, respectively, to detect the first abnormal symptom and the second abnormal symptom; and the first abnormal symptom detected by the symptom detection unit. It is configured to include a cycle model generator that generates a cycle model to detect the final abnormal signal by comparing it with the normal cycle of the pattern based on the abnormal symptom and the second abnormal signal.

In addition, the abuse detection unit detects the standard usage amount and excess usage amount based on the usage amount of the client from the data collected through the data collection unit, and the excess usage amount is detected periodically based on time, daily, weekly, and monthly periods. It further includes a cycle-by-cycle analysis module that analyzes usage, and a usage comparison module that detects anomalies by comparing each cycle according to the cycle-by-cycle usage detected by the cycle-by-cycle analysis module, wherein the cycle-by-cycle usage is determined by the client. It is characterized by analyzing the usage by cycle after giving weight to the usage time and number of usage.

The present invention, constructed and operated as described above, has the effect of preventing failures through rapid and accurate prediction of an automated monitoring system. In other words, failures can be accurately predicted by combining HW log analysis and big data to establish a technologically accessible prediction model. In particular, there is an effect of predicting failure of a specific part by detecting the occurrence of system data types that are correlated with the failure of a specific part.

In addition, the present invention has the effect of predicting the possibility of occurrence of a failure in real time using an IT service management system, or predicting specific failure details by detecting the occurrence of system data types that are correlated with a specific failure.

In addition, the present invention can increase the accuracy of inspection of failures through a systematic and standardized IT Infra management system (integrating SMS (System Management System), NMS (Network Management System), and FMS (Facility Management system)) and There is an advantage in being able to accurately predict failures by establishing failure prediction models for various failures using intelligent technologies (data mining, regression analysis models, etc.).

1 is a schematic diagram of a digital transformation prediction system based on big data and machine learning according to the present invention;
Figure 2 is a detailed configuration diagram of a digital transformation prediction system based on big data and machine learning according to the present invention;
Figure 3 is a detailed view of the pattern detection unit in the big data and machine learning-based digital transformation prediction system according to the present invention;
Figure 4 is a diagram showing an example of analysis of the abuse detection unit in the big data and machine learning-based digital transformation prediction system according to the present invention;
Figure 5 is a diagram showing an example of analysis of the time series detection unit in the big data and machine learning-based digital transformation prediction system according to the present invention.

Hereinafter, the digital transformation prediction system based on big data and machine learning according to the present invention will be described in detail with reference to the attached drawings.

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.

The “terminal” mentioned below may be implemented as a computer or portable terminal that can connect to a server or other terminal through a network. Here, the computer is, for example, a laptop equipped with a web browser, a desktop, a laptop, a VR HMD (e.g., HTC VIVE, Oculus Rift, GearVR, DayDream, PSVR, etc.), etc. may include.

Here, VR HMD is for PC (e.g. HTC VIVE, Oculus Rift, FOVE, Deepon, etc.), mobile (e.g. GearVR, DayDream, Storm Magic, Google Cardboard, etc.), and console (PSVR). Includes independently implemented Stand Alone models (e.g. Deepon, PICO, etc.). Portable terminals are, for example, wireless communication devices that ensure portability and mobility, including smart phones, tablet PCs, and wearable devices, as well as Bluetooth (BLE, Bluetooth Low Energy), NFC, RFID, and ultrasonic devices. , may include various devices equipped with communication modules such as infrared, WiFi, and LiFi. In addition, “network” refers to a connection structure that allows information exchange between nodes such as terminals and servers, including a local area network (LAN), a wide area network (WAN), and the 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, 3GPP (3rd Generation Partnership Project), LTE (Long Term Evolution), WIMAX (World Interoperability for Microwave Access), Wi-Fi, Bluetooth communication, infrared communication, and ultrasound. This includes, but is not limited to, communication, Visible Light Communication (VLC), LiFi, etc.

Hereinafter, 'operation data' refers to data that can directly or indirectly indicate the operation of a machine, and may mean, for example, information such as temperature, humidity, pressure, power, etc. of the machine. Additionally, 'object' refers to a part produced in a factory or a sub-component of that part, and refers to an object manufactured/produced by a single machine.

Additionally, 'error data' refers to data regarding at least one of machine errors and defects in objects produced by machines.

The big data and machine learning-based digital transformation prediction system according to the present invention detects the operation signal value from the client, which is the measurement target, using one or more signal receivers, and analyzes the operation signal value to predict the operation of the measurement target. A prediction system comprising: a signal receiver that communicates with at least one external client and receives data from the external client; A data collection unit that collects data obtained through the signal receiver that receives a signal from the client, and collects data collected at a certain period in a structured or unstructured form; A pattern detection unit that analyzes the pattern of the data collected by the data collection unit and compares the analyzed pattern with a standard pattern to detect abnormal patterns, and detects the usage amount of the data collected by the data collection unit and compares it with the standard usage amount. It is comprised of an abuse detection unit that detects abuse data corresponding to abnormal usage, and a time series detection unit that analyzes the data collected by the data collection unit in a time-series manner and detects abnormal data from the time-series data, and each detection detected here. an operation server that operates the client through signals; Each data analyzed through pattern detection, abuse detection, and time series detection in the data detected by the operation server is stored in the big data DB, and each data stored in the big data DB is learned to generate normal data and abnormal data. machine learning department; an administrator terminal that processes control commands to the client or sends control commands to the responsible worker according to the results detected through the operation server; and an external terminal that provides reference data and weight values for each data analyzed by the operation server, wherein the abuse detection unit determines the criteria based on the amount of usage of the client in the data collected through the data collection unit. The usage amount and excess usage amount are detected, and the excessive usage amount is compared with the standard usage amount and the excess usage amount based on time, daily, weekly, and monthly periods to determine whether the client is abusing the client.

The digital transformation prediction system based on big data and machine learning according to the present invention detects signals obtained from various external clients and collects them as data to detect the flow of data patterns, motion misuse detection, prediction detection through time series analysis, and The main technical feature is to obtain data from various external clients, such as predicting symptoms through these data, and analyze them to provide a prediction system that can monitor in real time, such as device operation status, operation flow, failure prediction, and cycle analysis. .

The present invention broadly consists of at least one signal receiver 100 that detects signals from various clients 10, collects and stores data obtained from the signal receiver 100, and analyzes various detection results from the acquired data. an operation server 200, a machine learning unit 300 that performs data learning through machine learning based on data obtained from the operation server, and an administrator terminal 400 that manages and operates the data of the operation server, and It is configured to include an external terminal 500 that assigns variable values and weights to the data analyzed by the operation server 200.

Figure 1 is a schematic configuration diagram of a digital transformation prediction system based on big data and machine learning according to the present invention.

The client of the present invention may correspond to a hardware or software system implemented in various industrial sites, such as client operation signals, operating signals, system flows, networks, communications, finance, and industrial equipment, and the client's Predictive signals are analyzed by analyzing the operating state or system flow.

The signal receiver 100 receives digital signals about the operating state or system flow from the client through network communication (wired or wireless), and the signal values detected here are various, such as physical, chemical, and digital signals.

The signal receiver 100 detects online sensing data collected through a sensor network, Ethernet, and Wi-Fi, and detects offline sensor data measured through analysis equipment (dedicated analyzer) from an IoT device or offline sensor. . Alternatively, a signal is received from a sensor attached to an IoT device, or in the case of various mechanical equipment, the signal receiver 100 receives detection values detected from various sensors including a temperature sensor, a humidity sensor, or a pressure sensor.

Accordingly, the signal receiving unit communicates with at least one external client and receives data, and can receive data from a plurality of clients in bulk through one network or through separate equipment such as a communication intermediary or hub.

Various data received from the signal receiver 100 are collected in the operation server 200.

The operation server 200 is a main component of the prediction system according to the present invention, and stores and analyzes various data obtained from external clients to analyze the client's operation, performance analysis, failure prediction, error analysis, event occurrence, and device Analyzes various information values, such as prediction of lifespan or replacement period, productivity improvement, and environmental prediction.

To this end, the pattern of data collected by the data collection unit of the operation server 200 is analyzed, abnormal pattern data is detected by comparing the analyzed pattern with the standard pattern, and usage is detected from the collected data and compared with the standard usage. It is possible to detect a predicted value by detecting time-series operation data that detects abuse data corresponding to abnormal usage.

Various signals detected in the operation server 200 are stored in the big data DB 310 according to another main technical purpose of the present invention, and each data stored in the big data DB is learned to obtain normal data. and a machine learning unit 300 that generates abnormal data.

That is, when the machine learning unit 300 receives a signal from a client and analyzes various analysis values through the operation server 200, the reliability of the data analyzed through machine learning is used for more accurate signal analysis. It is operated to secure.

Therefore, the present invention is implemented to precisely predict various variables of the detection signal through more accurate generation and learning of the prediction signal through the machine learning unit 300 in the process of detecting the prediction signal.

The management terminal 300 processes a control command to the client or sends a control command to the responsible worker (worker terminal) according to the analysis result of the operation server.

In addition, the manager terminal 300 is used to enable various experts to participate to support data analysis and to generate, store, and manage reference data by applying a group analysis system through an internal or external network.

Meanwhile, according to another technical feature of the present invention, it is configured to include an external terminal 500 that provides reference data or weight values for each data analyzed by the operation server 200.

For example, the external terminal is a terminal used by a client manufacturer, a client agent, a hardware manufacturer applied to the client, a separate client external expert, etc., and is an organization that best understands the basic information of the client, and the operation server 200 It is possible to provide reference data or weight data for signals analyzed in the operation server. For example, when determining whether the signal value detected in the operation server is normal or abnormal data, the client's operating characteristics, period of use, and current environment are used. More accurate analysis is possible by providing weight values that reflect variable values that may occur depending on characteristics, etc.

At this time, the external terminal 500 is configured to access the data DB 310, monitors the analyzed data, and verifies the reliability of the analysis value. For example, in the case of mechanical equipment, weight values are provided by reflecting mechanical equipment characteristic information such as durability value, operation time, number of operations, and number of days elapsed from the last inspection date for each mechanical equipment. It is possible to assign variable values to the relevant data.

Therefore, the external terminal 500 allows multiple groups to access the data DB from the outside through a network, and can perform data reliability verification and data change. Additionally, when a data verification request occurs through the manager terminal 400, the external terminal 500 approaches and performs data verification.

Figure 2 is a detailed configuration diagram of a digital transformation prediction system based on big data and machine learning according to the present invention.

A data collection unit 210 that collects various data of the client 10 received through the signal receiver 100 from the operation server 200, a pattern detection unit 220 that detects a pattern of data, and a data It is composed of an abuse detection unit 230 that detects abuse according to the amount of usage, and a time series detection unit 240 that analyzes abnormal signals through time-series analysis of data.

The data collection unit 210 collects the signal generated from the client after receiving it from the signal receiver 100. The data collection unit 210 collects the data at regular intervals and collects it in a structured or unstructured form. do. Data is collected hourly, daily, and over a period of time for a single piece of data, and abnormalities in the data are detected through structured and unstructured collection processes. Data processing speed can be improved depending on structured and unstructured collection methods. For example, in the case of data from industrial sites where data processing is quite large, the data processing speed can be improved by collecting data in a structured way for a certain period of time and then collecting it in an unstructured way when necessary.

The data collected by the data collection unit 210 is first detected by the pattern detection unit 220, which analyzes patterns of the data. For example, when detecting the operating state of mechanical equipment or the operating state of parts of the mechanical equipment, the pattern of the collected data is detected and compared with a previously stored reference pattern to detect whether the pattern is abnormal.

Through this pattern detection, it is possible to analyze the failure type of each mechanical equipment due to external factors, predict the failure of the mechanical equipment, specify the external factors affecting the mechanical equipment, or trace the failure history.

The abuse detection unit 230 detects the client's usage amount, and based on this usage amount, detects an abnormal signal of the client by comparing it with the standard usage amount when the usage amount is greater than the average usage amount or when the usage amount suddenly increases. For example, if the usage of a specific part of mechanical equipment suddenly increases more than the average, predictive information about driver error or intentional manipulation can be detected, and if the amount of access on an online network suddenly increases, information about fraud can be detected. can be detected.

Therefore, the abuse detection unit preferably predicts failure or misconduct by detecting abuse, since errors or failures may occur in the system due to increased usage, such as intentional access or interference from the outside of specific equipment, web traffic, communication data, etc. and can be monitored.

Figure 3 is a detailed diagram of the pattern detection unit in the big data and machine learning-based digital transformation prediction system according to the present invention. FIG. 3 is a detailed configuration of the pattern detection unit described in FIG. 2. In the process of detecting patterns of data collected through the data collection unit, the pattern detection unit 220 detects patterns exceeding a certain range among patterns compared to a reference pattern. Define the first pattern, define the range that exceeds the first pattern as the second pattern, and then compare the first and second patterns with the reference pattern to detect the first and second abnormalities. It further includes a sign detection unit 250.

If a specific signal value deviates from a certain value within a certain range, it corresponds to the first pattern, and if it deviates from a value greater than the first pattern, it is classified as a second pattern, and this is classified into a first abnormality symptom and a second abnormality symptom. Rapid response is possible by classifying the type of disability rather than simply determining the disability.

In addition, the pattern detection unit 220 detects the final abnormal signal through comparison with the normal cycle of the pattern based on the patterns for the first abnormal symptom and the second abnormal symptom corresponding to the categories classified by the symptom detection unit 250. To do so, it includes a cycle model generation unit 260 that generates a cycle model.

The cycle model generator 260 generates a cycle model based on accumulated abnormal symptom data and then uses the cycle model as an analysis standard to be used as analysis data for other clients.

Therefore, the cycle model generated by the cycle model generator 260 can be constructed as one standard data and apply the cycle model when analyzing other clients to quickly derive analysis values.

Figure 4 is a diagram showing an example of analysis of the abuse detection unit in the big data and machine learning-based digital transformation prediction system according to the present invention.

The abuse detection unit according to the present invention can solve the problem of failures caused by client misconduct, external intrusion, or user operation errors, and provides information about when usage data in the client signal deviates from the average value or when usage deviates from the time zone. predict.

Specifically, the abuse detection unit 230 detects the standard usage amount and excess usage amount based on the usage amount of the client from the data collected through the data collection unit, and determines the standard usage amount and excessive usage amount by time, daily, weekly, Includes a cycle-by-cycle analysis module 231 that analyzes usage by cycle based on a monthly period. By performing analysis in seconds, minutes, and hours, it is possible to analyze abuse within a specific range. That is, the cycle-by-cycle analysis module 231 can analyze usage information in detail by implementing cycle-by-cycle analysis when detecting client misuse information.

At this time, a usage comparison module 232 is configured to detect abnormalities by comparing each cycle according to the usage detected by the cycle analysis module 231, and the usage comparison module compares the past history and the current history together. By doing so, the possibility of misuse can be analyzed more accurately.

At this time, the usage by cycle is analyzed after weighting the usage time and number of uses of the client. Since the client's usage time and usage frequency are important variables in the case of mechanical equipment, erroneous judgments about usage information can be resolved when the weight information is assigned.

Figure 5 is a diagram showing an example of analysis of the time series detection unit in the big data and machine learning-based digital transformation prediction system according to the present invention.

The time series detection unit 240 analyzes the collected data in a time series structure, and can easily monitor the flow of operation signals through time series analysis of pattern signals or abuse signals.

In particular, in the present invention, in the process of predicting a client's abnormal signal, failure, or error signal for a pattern signal, a pattern exceeding a certain range is divided into a first pattern, and a range further exceeding the first pattern is divided into a time-series pattern of the second pattern. Reliability can be improved by constructing big data of abnormal signals (failures, etc.) through analysis and time-series analysis of first and second abnormal symptom signals.

The present invention configured in this way has the effect of preventing failures through rapid and accurate prediction of an automated monitoring system. In other words, failures can be accurately predicted by combining HW log analysis and big data to establish a technologically accessible prediction model. In particular, there is an effect of predicting failure of a specific part by detecting the occurrence of system data types that are correlated with the failure of a specific part.

In addition, the present invention has the effect of predicting the possibility of occurrence of a failure in real time using an IT service management system, or predicting specific failure details by detecting the occurrence of system data types that are correlated with a specific failure.

In addition, the present invention can increase the accuracy of inspection of failures through a systematic and standardized IT Infra management system (integrating SMS (System Management System), NMS (Network Management System), and FMS (Facility Management system)) and There is an advantage in being able to accurately predict failures by establishing failure prediction models for various failures using intelligent technologies (data mining, regression analysis models, etc.).

Although the present invention has been described and illustrated in connection with preferred embodiments for illustrating the principles of the present invention, the present invention is not limited to the construction and operation as so shown and described. Rather, those skilled in the art will appreciate that many changes and modifications can be made to the present invention without departing from the spirit and scope of the appended claims. Accordingly, all such appropriate changes, modifications and equivalents shall be considered to fall within the scope of the present invention.

100: signal receiver
200: Operation server
210: data collection unit
220: Pattern detection unit
230: Abuse detection unit
231: Analysis module by cycle
232: Usage comparison module
240: Time series detection unit
250: Sign detection unit
260: Cycle model generation unit
300: Machine Learning Department
310: Big data DB
400: Administrator terminal
500: external terminal

Claims (3)

In a prediction system that detects an operation signal value from a client that is a measurement target using one or more signal receivers and predicts the operation of the measurement target by analyzing the operation signal value,
A signal receiver that communicates with at least one external client and receives data from the external client;
A data collection unit that collects data obtained through the signal receiver that receives a signal from the client, and collects data collected at a certain period in a structured or unstructured form;
a pattern detection unit that analyzes the pattern of the data collected by the data collection unit and compares the analyzed pattern with a standard pattern to detect an abnormal pattern; detects the data usage of the client collected by the data collection unit; It is comprised of an abuse detection unit that detects abuse data corresponding to abnormal usage by comparison, and a time series detection unit that analyzes the data collected by the data collection unit in time series to detect abnormal data from the time series data, and each detected here an operation server that operates the client through a detection signal;
Each data analyzed through pattern detection, abuse detection, and time series detection in the data detected by the operation server is stored in the big data DB, and each data stored in the big data DB is learned to generate normal data and abnormal data. machine learning department;
an administrator terminal that processes control commands to the client or sends control commands to the responsible worker according to the results detected through the operation server; and
It is configured to include; an external terminal that provides reference data and weight values for each data analyzed by the operation server,
The abuse detection unit detects the standard usage amount and excess usage based on the data usage of the client from the data collected through the data collection unit, and the excess usage is determined based on time, daily, weekly, and monthly periods. A digital transformation prediction system based on big data and machine learning, characterized in that it determines whether the client is being misused by comparing usage. The method of claim 1, wherein the pattern detection unit,
Detect patterns in data collected through the data collection unit,
Among the patterns compared to the reference pattern, a pattern exceeding a certain range is defined as the first pattern, and a range further exceeding the first pattern is defined as the second pattern, and then the first pattern and the second pattern are defined as the reference pattern. A symptom detection unit that compares and detects the first abnormal symptom and the second abnormal symptom, respectively;
Big data comprised of a cycle model generator that generates a cycle model to detect the final abnormal signal by comparing it with the normal cycle of the pattern based on the first and second abnormal signs detected by the symptom detection section. , Machine learning-based digital transformation prediction system. The method of claim 1, wherein the abuse detection unit,
From the data collected through the data collection unit, standard usage and excess usage are detected based on the client's usage, and the excess usage is analyzed by cycle based on time, daily, weekly, and monthly periods. module,
It further includes a usage comparison module that detects anomalies by comparing each cycle according to the usage detected by the cycle analysis module,
A digital transformation prediction system based on big data and machine learning, characterized in that the usage by cycle is analyzed by weighting the usage time and number of uses of the client and then analyzing the usage by cycle.