KR20240057866A - Integrated facility monitoring system using wireless sensor and metaverse - Google Patents

Abstract

The present invention collects and analyzes product information of various measuring devices for facility monitoring, represented by wireless vibration sensors, and makes it available through an integrated monitoring system. By combining machine learning and metaverse technology, proactive or rapid response to various failures is possible. It is about an integrated facility monitoring system using wireless sensors and metaverse that makes it possible to recognize the structure and operating status of monitored facilities and identify the cause and situation of defects without a manager's on-site visit.

Description

{Integrated facility monitoring system using wireless sensor and metaverse}

The present invention relates to an integrated facility monitoring system. Specifically, it collects and analyzes product information of various facility monitoring measuring devices, such as wireless vibration sensors, to enable use through an integrated monitoring system, and uses machine learning and metaverse technology. An integrated facility monitoring system using wireless sensors and metabus that enables proactive or rapid response to various failures by incorporating them, and supports the recognition of the structure and operating status of monitored facilities and the cause and situation of defects without a manager's on-site visit. It's about.

Recently, with the rapid development of information and communication technology and the increase in demands for efficient management of various industrial plants, large plants at home and abroad use a variety of wired and wireless measuring devices, such as vibration measuring devices, to monitor and/or monitor the status of various equipment installed in the facility. Through this, managers can check the status through remote monitoring and respond quickly in case of an emergency without having to patrol and check the status of the equipment directly.

However, many measuring devices, such as wireless vibration measuring devices, still analyze measurement results individually through dedicated software produced by manufacturers. Individual monitoring/analysis of various devices allows users to operate multiple systems equipped with dedicated software. Due to the unavoidable nature of the system being supplied as expensive equipment and packages (SW), small and medium-sized plant businesses face difficulties in introducing wireless vibration measurement equipment.

Furthermore, these dedicated software systems require the development of separate software to incorporate new technologies based on artificial intelligence (AI) that have recently been introduced for efficient plant management, so a system that can integrate and utilize them is required.

A representative example of the mentioned new technologies that have recently attracted attention is Metaverse, which includes concepts such as Digital Twin, which replicates the same environment as the actual facility with software to implement digital virtual objects in the digital environment and collects data generated from the facility. One example is a solution that collects data in real time and provides it to the operator.

Through this, managers can check the operating status of facilities in real time through the metaverse space, which is a virtual representation of the actual plant. The advantage is that facility operation data, such as failures or accidents, can be checked and predicted through the metaverse and then responded immediately. There is.

However, due to reasons such as compatibility with the use of dedicated software for individual measuring devices mentioned above, it is not easy to integrate and manage individual wireless measuring devices, and it is still limited to analyzing equipment data and simply providing it, predicting status and processing according to the diagnosis. There are many difficulties in implementing features such as optimization.

Republic of Korea Patent No. 10-2206832 (2021.01.19)

The present invention was created in response to the above-mentioned needs. The purpose of the present invention is to collect and analyze product information of various equipment monitoring measuring devices and utilize it through an integrated monitoring system, and to combine machine learning and metaverse technology. The goal is to provide an integrated facility monitoring system using wireless sensors and metabus that can respond in advance or quickly to various failures.

For the above purpose, the present invention includes a sensing module that is mounted on a rotating device and generates measurement information including vibration; A first analysis module that collects and analyzes the data handling method of the sensing module to generate first analysis information; A second analysis module that collects and analyzes the configuration and process of the monitoring system in operation at the plant and generates second analysis information; a database constructed by storing the first analysis information and the second analysis information, raw data collected from the sensing module, and equipment information to be monitored; A linkage module that applies and processes data based on data retrieved through the database for the monitored equipment and applied sensing module in a linkage method; An integrated monitoring module that includes a conversion unit that converts the signal value collected from the applied sensing module for frequency analysis and a diagnostic unit that analyzes the converted signal value to diagnose the coupling, and monitors the equipment status; It is characterized by having a.

At this time, a learning unit that learns the signal value converted through the conversion unit and generates learning data by extracting the signal value for which a defect has been diagnosed, and classifies the signal value by defect through the learning data and applies it to the diagnosis unit. A learning module having a reflection unit; It may further include.

In addition, a virtual engine unit that creates and manages a digital twin 3D model through digital objects created in response to the monitored facility, and a synchronization unit that reflects the analyzed signal values and diagnosis results in the 3D model in connection with the integrated monitoring module. Metaverse module having wealth; It may further include.

In addition, the linkage module has a direct linkage unit that directly connects the data generated by the sensing module with the integrated monitoring module at a set cycle, extracts data using the API KEY from the operating program of the sensing module, and then performs integrated monitoring at a set cycle using sFTP. It may include an API linkage unit that transmits to the module, and an OPC-UA linkage part that extracts data from the sensing module to the OPC Server and transmits it to the integrated monitoring module through OPC UA data conversion according to the protocol.

Through the present invention, various devices can be integrated into one program to enable integrated monitoring through connection with various devices installed in the plant, and it is possible to build a flexible web-based system that meets user needs.

In addition, through machine learning that sets the initial normal data boundary and defines data outside the boundary as abnormal, and the function of notifying the manager of the initial result of a defect, it can contribute to the stable operation of the plant by responding in advance or quickly to various failures.

In particular, by visualizing equipment through Metaverse, remote managers can intuitively identify the cause and situation of defects without having to visit the site. They can easily recognize the structure and operating status information of monitored equipment, and when an abnormal condition occurs, the phenomenon identification and analysis time can be reduced. can be minimized.

1 is a conceptual diagram of the present invention;
Figure 2 is a block diagram showing the configuration and connection relationship of the present invention;
Figure 3 is a wireless vibration measurement device SW analysis screen according to an embodiment of the present invention;
Figure 4 is a plant monitoring system analysis screen according to an embodiment of the present invention;
5 is a diagram illustrating a direct connection process according to an embodiment of the present invention;
Figure 6 is an explanatory diagram of the API linkage process according to an embodiment of the present invention;
Figure 7 is an explanatory diagram of the OPC UA linkage process according to an embodiment of the present invention.

Hereinafter, the structure of the integrated facility monitoring system using the present invention's wireless sensor and metaverse will be described in detail with reference to the attached drawings.

Figure 1 is a conceptual diagram of the present invention. The present invention analyzes various sensing module 110 products using separate programs to acquire data storage and processing structures and formats, and connects various sensing modules 110 and an integrated database. By implementing PROCESS/algorithm, compatibility is improved so that it can be immediately integrated into the monitoring system of various facilities with functions such as data monitoring/analysis/alarm. In addition, it supports the diagnosis of facility defects by analyzing sensing data.

Figure 2 is a block diagram showing the configuration and connection relationship of the present invention. The main components of the present invention are a sensing module 110, a first analysis module 120 and a second analysis module 130, and a database 140. In addition to the linkage module 150 and the integrated monitoring module 160, it is equipped with a learning module 170 for machine learning and a metaverse module 180 for incorporating metaverse technology as an expansion function.

The sensing module 110 is a sensor for measuring the state of equipment. In the present invention, a representative sensor is a sensor that is mounted on rotating equipment and generates measurement information including vibration. In the embodiment of the present invention, the sensing module 110 is a wireless vibration measurement device in use in Korea, including the W-CMS of the Korean company Assetcare, the 'W' product of the Belgian company 'I", and the 'R' product of the American company 'B'. Three types were selected.

The first analysis module 120 is a component that collects and analyzes the data handling method of the sensing module 110 to generate first analysis information. Figure 3 is a wireless vibration measurement device SW analysis screen according to an embodiment of the present invention. Basically, the data storage and analysis SW of the applicable sensing module 110 is analyzed, and the sensor's data storage format, database, and data connection method are analyzed. It is included in the first analysis information.

The second analysis module 130 is a component that collects and analyzes the configuration and process of the monitoring system in operation in the plant to generate second analysis information. Figure 4 is a plant monitoring system analysis screen according to an embodiment of the present invention. Second analysis information is generated by analyzing information for interoperability according to data integration through process analysis for plant configuration and data integration. do.

Related standards can be reflected for interoperability through data integration, and as ISO18431/ISO 13374 proposes status monitoring procedures and signal processing methods using equipment such as wireless vibration sensors, this is reviewed in the embodiment of the present invention. It was reflected.

The suggestions for ISO18431/ISO 13374 are as follows [Table 1].

suggestions Detail Condition monitoring program definition - Recommended to select equipment subject to monitoring, determine the type of appropriate measurement system, and apply it - Recommend that operating conditions such as speed, load, temperature, etc. be clearly expressed and broadcast on the received vibration data.
- Data representation includes shaft speed (rpm), machine load (force, flow rate, pressure, etc.) and other operating parameters that may affect the measured vibration.
- For general data acquisition, it is recommended that operating conditions be as close to the machine's rating or normal operation as possible.
- Data must be obtained for known, reproducible process parameters when the machine is operated under the same or similar conditions. Condition Assessment Process - Problem detection (deviation from steady state)
- Diagnosis of defects and causes of defects
- Foreknowledge of future defect progression
- Prescription for solving problems or countermeasures
- Postmortem analysis signal processing - Filtering: Time domain signals are filtered through low-pass, high-pass, band-pass, or band-reject filters. - Sampling: Sampling at a rate called sampling frequency fs
Maximum (Nyquist) frequency fN = fs/2
Sampling interval Dt = 1/fs - Signal pre-processing: Before processing the signal to generate time-domain results, digital signals are often pre-processed using linear filters to remove unwanted narrow-band or out-of-band noise. - Fourier Transform: Convert continuous data with physical units in the time domain to the frequency domain by the discrete Fourier transform equation. - Time domain analysis: size analysis/distribution function analysis/moment analysis/correlation analysis/impact response spectrum - Frequency domain analysis: Determination of vibration characteristics as a function of frequency

The database 140 is constructed by storing the first analysis information and the second analysis information, raw data collected from the sensing module 110, and information on equipment to be monitored. Specifically, the database includes signal information, which is information for classification of data acquired from the monitoring target equipment and the sensing module 110, raw data, which is data collected from the sensing module 110 for vibration analysis, and data comparison and big data. Equipment information, which is information on equipment subject to installation of the sensing module 110 for data conversion, defect information on equipment for use in learning, spectrum data converted using the FFT algorithm for frequency analysis as a result of data conversion for analysis, The learning results of classifying by defect by extracting and learning the characteristics of the signal using machine learning data, and the diagnosis results of diagnosing defects in the input data through comparison with the learning results are stored.

In an embodiment of the present invention, data is standardized to build a database 140 in connection with three types of sensing modules 110. At this time, standardization was performed including facility information in order to upgrade to an automatic predictive diagnosis system in the future. Standardization was performed by dividing into properties including facility information/defect information and data including vibration signal information, and considering database and data management. [Table 2] shows the standardized item ‘Properties’.

Standardized items explanation Application field example ID data id data classification "00580010101814070013180503093540" Failure Type Data defect analysis results AI machine learning “misalignment” Measure Point Measuring point symbol of the sensor data classification "POV" Motor Model Motor model name of the equipment being measured Classification of same type/similar facilities "112S" Motor Type Motor type of the equipment being measured Classification of same type/similar facilities "AC induction" Motor Rated Frequency Operating frequency of the equipment motor to be measured Fault frequency selection/classification of similar equipment 3430 Motor Rated Power Output of the equipment motor to be measured Classification of similar equipment 2.2 Motor Bearing Type Bearing type of the equipment motor being measured Selection of defect type and defect frequency "Antifriction" Pump Model Model name of target equipment pump Classification of same type/similar facilities "HES 32-125" Pump Support Bearing support type of target equipment pump Selection of defect type "over-hung" Pump Number of Outlet Vanes Number of vanes of target equipment pump Fault frequency selection 5 Pump Inlet Type Type of inlet of target equipment pump Fault frequency selection "volute" Pump Bearing Type Bearing type of target equipment pump Fault frequency selection "Antifriction" Pump Fluid Type Working fluid type of target equipment pump Selection of defect type and defect frequency "water" Pump Number of Stages Number of stops at the target equipment pump Selection of defect type "one"

[Table 3] shows the standardization item review ‘data’ according to the embodiment of the present invention.

Standardized items explanation Application field example Time sampling time Signal processing of data/data frequency range “8.44595e-05” Value(mm/s) Measured values from the sensor Data analysis and defect assessment “-0.045609199999999996”

In the embodiment of the present invention, the database is a document-oriented database suitable for Json (dynamic schema-type documents, vibration data, etc.), and was designed by implementing it with NoSQL-based mongoDB, which is being used as a database in many wireless vibration measurement equipment.

The linkage module 150 is configured to apply and process data based on data retrieved through the database for the monitoring target equipment and the applied sensing module 110, and in the embodiment of the present invention, direct linkage is a detailed configuration. Three connection methods through the unit 151, the API linkage unit 152, and the OPC-UA linkage unit 153 are explained.

Figure 5 is a diagram illustrating the direct link process according to an embodiment of the present invention. The direct link unit 151 directly links data generated in the sensing module 110 with the integrated monitoring module at a set cycle.

In other words, by applying the same DB (mongoDB) as the DB used in commercial wireless vibration measurement equipment, data is extracted directly from the sensing module 110, synchronized periodically, and the extracted data is standardized and stored.

Figure 6 is a diagram illustrating the API linkage process according to an embodiment of the present invention. The API linkage unit 152 extracts data using the API KEY from the operating program of the sensing module 110 and then performs integrated monitoring at a set cycle using sFTP. Transmit to module.

That is, data is extracted from the operating program of the sensing module 110 using the API KEY, and the extracted data format is Json format. The extracted data is periodically transmitted using sFTP, and the data transmitted to the designated folder is input into the integrated database 140, and the extracted data is stored after standardization.

Figure 7 is an explanatory diagram of the OPC UA linkage process according to an embodiment of the present invention. The OPC-UA linkage unit 153 extracts data from the sensing module 110 to the OPC Server and converts it through OPC UA data conversion according to the protocol. Transmitted to the integrated monitoring module.

OPC UA (Open Platform Communications Unified Architecture) is a communication protocol standard used as a standard in the smart factory field. When data is extracted with OPC Server, the extracted data format is Json format, and the OPC UA data conversion program according to the GE protocol is applied.

The integrated monitoring module 160 outputs the information of the sensing module 110, sensing values, analysis results, facility monitoring, and diagnosis results, and monitors the facility status in an integrated manner. The detailed configuration is converted into a unit 161. and a diagnostic unit 162.

The conversion unit 161 is a component that converts the signal value collected from the sensing module 110 applied for facility monitoring for frequency analysis. Considering the characteristics of the applied sensing module 110, the aforementioned direct connection/ After extracting data considering API linkage/OPC-UA linkage, the data is converted to reflect data standardization.

In the present invention, the conversion unit 161 applies the FFT SPECTRUM conversion algorithm that converts the vibration raw data of the time waveform into the frequency domain by FFT processing it, and in the data obtained at a given sampling rate (rate) to remove aliasing error and noise. Filtering below the maximum usable frequency, sampling to obtain function values for discrete values for fft processing, and reducing errors while processing weighted data points. The truncated function used to apply a HANNING window for condition monitoring, averaging to improve signal accuracy and reduce data measurement time in condition monitoring, and a continuous sample of data x with physical unit U in the time domain. (n) performs a frequency domain transformation process through Fourier transform.

The diagnosis unit 162 is a component that diagnoses defects by analyzing the signal value converted through the conversion unit, and diagnoses the occurrence and type of defect by setting a reference value for coupling judgment and comparing and analyzing the signal value. Machine learning can be used to complement the early defect pre-diagnosis algorithm obtained through prior research to automatically diagnose the type (cause) of the defect.

The learning module 170 provided for this purpose includes a learning unit 171 that learns the signal value converted through the conversion unit 161 and generates learning data by extracting the signal value for which a defect has been diagnosed, and the learning data It is provided with a reflection unit 172 that classifies signal values according to defects and applies them to the diagnosis unit 162.

Raw data (Time Waveform) containing vibration signal information (amplitude, defect, phase) is used to extract features for automatic diagnosis, and noise caused by external factors is removed through filtering to improve the accuracy and reliability of automatic diagnosis. For feature extraction, preprocessing is performed to re-sample the raw data (Time Waveform) and normalize the sampled data.

In an embodiment of the present invention, the time domain (19 types) of [Table 4], the frequency domain (7 types) of [Table 5], and the entropy (4 types) of [Table 6], which represent the statistical and morphological characteristics of the signal, are used. Features were calculated and extracted and used for learning and diagnosis.

Feature Description Frequency Center Indicator of center of density in FFT Spectrum Mean square frequency A value that represents the effective value of the signal and is the average size of the signal. RMS of frequency An indicator showing the density center of the RMS value of each component in the FFT Spectrum. Variance frequency Moment value from center of density in FFT Spectrum Root variance frequency In FFT Spectrum, an indicator expressing the RMS value of each component as a moment from the density center Spectrum overall The sum of the amplitudes of each component in the FFT Spectrum Spectrum rms overall The sum of the effective values of each component in the FFT Spectrum

Feature Description Peak value A value indicating the size of a signal as an indicator of the amplitude value of the signal Root Mean Square(RMS) A value that represents the effective value of the signal and is the average size of the signal. Kurtosis In statistics, it is an indicator of the sharpness of the probability distribution shape and increases as the distribution of signal values is concentrated near a specific value. Crest factor Difference between min/max values compared to the average size of the signal Clearance factor Like the crest factor, it represents the difference between the minimum and maximum values compared to the average size of the signal, and the average size is expressed as the Square Mean Root value instead of RMS. Impulse factor An indicator of the size of the largest impulse (the sharp spike in the signal) in the signal. Shape factor In electronics, it is an indicator of the ratio of DC and AC components, and is the ratio of the size of a continuous waveform between negative and positive compared to the average of the signal. Skewness In statistics, skewness is an indicator of the asymmetry of the probability distribution. As the signal bias (the degree to which the distribution of signal values is concentrated to one side based on the average of the signal) increases, skewness also increases. Square Mean Root(SMR) Same as RMS, it refers to the degree or magnitude of the transition between negative and positive in a continuous waveform, and is more sensitive to the size of the signal than RMS. 5th normalized moment Fifth moment value used in statistics 6th normalized moment 6th moment value used in statistics Mean average over signal magnitude Shape factor2 Like shape factor, an indicator representing the ratio of DC component and AC component, using SMR value that is sensitive to the size of the signal instead of RMS value. Peak to peak A value representing the difference between the maximum and minimum values of the signal Kurtosis factor A modified value of Kurtosis that compensates for the disadvantage of being sensitive to the overall size of the signal. Standard deviation standard deviation value of the signal Smoothness A value indicating the smoothness of the signal Uniformity An indicator of signal uniformity Normal negative log-likelihood An indicator representing the value used for maximum likelihood estimation used in statistics

Feature Description histogram_upper_band Top value of feature histogram histogram_lower_band Low value of feature histogram entropy_estimation_value Approximation of entropy entropy_estimation_error_value Approximation of entropy error

In addition, in an embodiment of the present invention, a Principal Component Analysis method that reduces the dimension for feature analysis and status classification of target equipment, evaluates the distinctiveness of feature combinations, and analyzes feature combinations with high distinctiveness to create similar feature combinations. Genetic Algorithm (GA) optimal feature selection technique that extracts the feature combination with the highest discrimination by generating and repeating it, finding a hyperplane in a high-dimensional or infinite-dimensional space and using it to perform classification, making a discrimination with one data point Support Vector Machine (SVM) facility status classification technique, which is based on linear discrimination with the maximum margin representing the distance to the boundary and does not consider dependencies between attributes, uses new data based on existing data A defect diagnosis technique was applied to classify using k adjacent data and learn the distance scale from the data to find the k closest data and classify them into the class with the highest frequency.

In order to improve the diagnostic performance of the diagnostic algorithm of the diagnostic unit 162, vibration data for each defect for learning is required, and if it is difficult to secure defect data for equipment, data can be secured using specialized equipment.

[Table 7] summarizes the four main defects of rotating equipment by types of learning data according to the embodiment of the present invention.

Defect type defect cause Parameter characteristics unbalance Eccentricity occurs when the axis center of the rotating body is misaligned with the center of gravity, resulting in centrifugal force. -The 1x (operation) component in the radial direction is dominant, and there is no change in phase angle.
- Vibration increases in proportion to rotation speed (centrifugal force action) Bearing Fault Occurs due to improper lubrication management, poor installation, excessive load, normal wear, and scratches/damage of elements. - Characteristic frequency components occur due to bearing damage. Misalignment Occurs when the centers of the two axes are not maintained in a straight line. - Occurrence of 2x (3x, 4x) components other than rotation frequency components
- In excessive cases, the 2x component dominates, and a 180° phase difference appears in both bearings. Looseness Occurs due to excessive gaps due to improper assembly, damage to worn parts, etc., and occurs due to loose bearings, excessive tolerances, etc. - One impact occurs per rotation in the radial direction (1x and multiple components generated)
- Fractional harmonic components occur in excessive cases (1/2, 1/3, 1/4x)

At this time, it is possible to secure defect data using a professional simulator. In the embodiment of the present invention, defects (4 types) of rotating equipment were simulated using a simulator, and defect data was secured using this.

The metaverse module 180 is a configuration that processes the monitoring results described above in connection with the digital twin 3D model and provides them through visualization. It creates and manages the digital twin 3D model through digital objects created in response to the monitored facility. It is provided with a virtual engine unit 181 and a synchronization unit 182 that reflects the signal values and diagnosis results analyzed in connection with the integrated monitoring module to the 3D model.

The virtual engine unit 181 forms a metaverse space by creating a digital twin 3D model of the monitoring target facility with boundary conditions set as a physical model, for example, a rotating facility such as a pump bearing or a monitoring target composing them. At least one digital twin 3D model corresponding to the facility is created and managed, and at least one visualization content corresponding to each monitoring/operation data regarding the monitored facilities is stored and managed.

The synchronization unit 182 performs mapping to link and process the diagnostic result information data received from the digital twin 3D model and at least one visualization content and the monitoring/operation data regarding the facilities to be monitored, for example, pump bearings 3D model-based operation and fault early diagnosis information is visualized and provided, 3D model-based simulation information of the pump bearing system, and pump bearing structure through 3D model-based operation and control are visualized and provided to determine the fault status of the equipment. Make it possible to check.

The rights of the present invention are not limited to the embodiments described above but are defined by the claims, and those skilled in the art can make various changes and modifications within the scope of the claims. This is self-evident.

110: Sensing module 120: First analysis module
130: Second analysis module 140: Database
150: Link module 151: Direct link unit
152: API connection unit 153: OPC-UA connection unit
160: Integrated monitoring module 161: Conversion unit
162: Diagnosis unit 170: Learning module
171: Learning Department 172: Reflection Department
180: Metaverse module 181: Virtual engine unit
182: synchronization unit

Claims (4)

A sensing module 110 mounted on a rotating facility to generate measurement information including vibration;
A first analysis module 120 that collects and analyzes the data handling method of the sensing module 110 to generate first analysis information;
A second analysis module 130 that collects and analyzes the configuration and process of the monitoring system operating in the plant and generates second analysis information;
a database 140 constructed by storing the first analysis information and the second analysis information, raw data collected from the sensing module 110, and information on equipment to be monitored;
A linkage module (150) that applies and processes data based on data retrieved through the database for the monitored equipment and the applied sensing module (110) using a linkage method;
It is provided with a conversion unit 161 that converts the signal value collected from the applied sensing module 110 for frequency analysis, and a diagnostic unit 162 that analyzes the converted signal value to diagnose the coupling, and monitors the equipment status. Integrated monitoring module (160); An integrated facility monitoring system comprising:
According to paragraph 1,
A learning unit 171 learns the signal value converted through the conversion unit and generates learning data by extracting the signal value for which a defect has been diagnosed, and classifies the signal value by defect through the learning data and applies it to the diagnosis unit. A learning module (170) having a reflection unit (172); An integrated facility monitoring system further comprising:
According to paragraph 1,
A virtual engine unit 181 that creates and manages a digital twin 3D model through digital objects created in response to the monitored facility, and a signal value and diagnosis result analyzed in connection with the integrated monitoring module are reflected in the 3D model. A metaverse module including a synchronization unit 182; An integrated facility monitoring system further comprising:
According to paragraph 1,
The linkage module 150 is a direct linkage unit 151 that directly links the data generated in the sensing module 110 with the integrated monitoring module at a set period, and data using the API KEY in the operating program of the sensing module 110. After extraction, the data from the API linkage unit 152 and the sensing module 110, which are extracted and transmitted to the integrated monitoring module at a set cycle using sFTP, are extracted to the OPC Server and transmitted to the integrated monitoring module through OPC UA data conversion according to the protocol. An integrated facility monitoring system comprising an OPC-UA linkage unit (153).