KR20180042897A - System and method for monitoring state of structure based compressive sensing - Google Patents

Abstract

The present invention relates to a compressive sensing based structure state diagnosis system capable of effectively diagnosing a state of a rotary device, and a method thereof. According to the present invention, the method comprises: a vibration measurement step (S100) of measuring vibration generated in an object to be measured from a plurality of vibration sensors installed in the object to be measured; a compressive sensing step (S200) of using a preset compressive sensing method to a measurement value measured in the vibration measurement step (S100) to generate a compressive measurement signal; and a state diagnosis step (S300) of analyzing the compressive measurement signal generated in the compressive sensing step (S200) to perform a state diagnosis for the object to be measured.

Description

[0001] The present invention relates to a system and method for monitoring a state of a compression sensing infrastructure,

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system and method for diagnosing a state of a structure based on a compression sensing and a method for detecting the state of a structure accurately And more particularly, to a system and method for diagnosing a state of a compression sensing infrastructure.

More particularly, the present invention relates to a method and system for determining damage occurrence of a structure using ROC (Receiver Operating Characteristic) analysis method without any signal restoration process, The present invention relates to a system for diagnosing a state of a structure based on a compression sensing and a method thereof.

Rotating machinery is a typical machine used throughout the industry, and the rotating system including it is the most common mechanical system.

Such a rotating system has a high frequency of occurrence of faults due to a rapid rotation speed and a complicated structure, and causes a great loss when a fault occurs. In particular, in the case of cooling pumps for nuclear power plants, there is a need for a condition diagnosis system that accurately discriminates the abnormality of the rotor system as it is an important facility in the operation of the power plant.

The state diagnosis technology of the conventional rotating body measures the vibration signal generated in the system in real time and determines the state of the rotating body structure through the statistical analysis of the physical characteristics of the defect and the vibration signal, Diagnosis is being made. However, vibration signal measurement is performed in a high sampling frequency region in the signal measurement process, and there is a problem of high energy consumption and massive data processing in the data measurement and data analysis process due to the sensor network that operates a plurality of sensors.

More specifically, the vibration data measurement method of the conventional rotating state diagnostic system is based on the Shannon-Nyquist theory and the Shannon-Nyquist theory shows that when the measurement is performed at a frequency twice as high as that of the analog signal, When a vibration signal having a high frequency component is measured, the number of measured data increases drastically and a large amount of data is measured for a short time. Therefore, there is a disadvantage that the energy consumption of the data measuring equipment is large.

In addition, there is a problem that errors are inevitably involved in the process of compressing and restoring data.

As a result, there is a growing interest in data compression techniques and data measurement techniques for improving signal processing efficiency and energy efficiency of sensor nodes attached to a rotating system.

In Korean Registered Patent No. 10-1374840 (Registered on Apr. 31, 2014, entitled "System and Method for Diagnosing Rotor Condition of Offshore Structures"), temperature and vibration signals are used to determine the state of rotation shaft and bearing Discloses a system and method for diagnosing a rotating body condition of an offshore structure capable of diagnosing whether a rotating shaft and a bearing are faulty or not.

Such a system and method adopt a conventional method as it is for measuring a vibration signal, and the above-described problems are not directly included, but the configuration for solving the problem is not mentioned at all.

Korean Registered Patent No. 10-1374840 (Registered on Apr. 20, 201, Name: System and Method for Diagnosing Rotor Condition of Offshore Structures)

SUMMARY OF THE INVENTION It is an object of the present invention to provide a method and apparatus for real-time measurement of a compression sensing signal for a vibration generated in a structure by using a compressive sensing technique, (ROC) analysis method without using a separate signal restoration process, and it is possible to identify the damage of the structure, and to detect the sensitivity of the damage classification and the damage intensification degree. State diagnostic system and method therefor.

The system for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention includes a vibration sensor unit 100 including a plurality of vibration sensors provided in a measurement object for measuring vibration generated in the measurement object, A signal processing unit 200 for receiving a measurement value measured from the signal processing unit 200 and generating a compressed sensing signal using a compressed sensing technique, And an analysis unit 300 for analyzing the compression sensing signal and diagnosing the state of the measurement object.

A method for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention includes a vibration measurement step (S100) for measuring vibration generated in the measurement object from a plurality of vibration sensors provided in a measurement object, A compression sensing step S200 for generating a compression sensing signal by using a compression sensing technique which is previously set to a measurement value measured in the compression sensing step S200, And a state diagnosis step (S300) of analyzing the state of the measurement object by analyzing the state of the measurement object.

The present invention provides a new data measurement method and a signal processing method for improving the efficiency of a conventional state diagnosis system and a new data measurement method called a compressive sensing method Can be applied to the vibration data measurement process and the data signal of the time domain can be statistically analyzed without restoring the existing compressed signal. In the data acquisition step, the rotation device (or the rotating body system) It is possible to effectively diagnose the condition of the patient.

More particularly, the present invention provides a system and method for diagnosing a state of a structure based on a compression sensing based on a real-time measurement of a compression sensing signal for vibration generated in a structure using a compression sensing technique, It is also possible to identify the damage of the structure using ROC (Receiver Operating Characteristic) line analysis technique for the measured compression sensing signal without additional signal restoration process, There is an advantage that the degree of damage can be diagnosed to a degree.

1 is a simplified block diagram of a system for diagnosing a state of a compression sensing infrastructure according to an embodiment of the present invention.
2 is a flowchart illustrating a method for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention.
3 is a view illustrating a system for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention and a process of utilizing the compression sensing technique of the method.
4 is a diagram illustrating a system for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention and a process of calculating a compression sensing signal in the method.
FIG. 5 is a diagram illustrating two statistical models generated according to the ROC analysis method of the system for diagnosing state of a structure of a compression sensing infrastructure according to an embodiment of the present invention.
FIG. 6 is a diagram illustrating a ROC diagram of a system for diagnosing state of a structure of a compression sensing infrastructure according to an embodiment of the present invention and a ROC diagram analysis method of the method.
7 and 8 are views for comparing measured signals of a measurement object according to the conventional Shannon-Nyquist technique, a system for diagnosing a condition of a compression sensing based structure according to an embodiment of the present invention, and measurement signals of a measurement object according to the method .

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a system and method for diagnosing a state of a compression sensing infrastructure according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings. The following drawings are provided by way of example so that those skilled in the art can fully understand the spirit of the present invention. Therefore, the present invention is not limited to the following drawings, but may be embodied in other forms. In addition, like reference numerals designate like elements throughout the specification.

In this case, unless otherwise defined, technical terms and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. In the following description and the accompanying drawings, A description of known functions and configurations that may unnecessarily obscure the description of the present invention will be omitted.

The present invention provides a system and method for diagnosing a state of a structure based on a compression sensing based on a method of efficiently detecting a state of an equal accuracy in comparison with a conventional Shannon-Nyquist-based signal processing method by applying a compression sensing technique to an active sensing- As a result of experiments, it has been proved that the defect detection and defect classification of the rotating body can be performed using only the compressed data without a separate recovery process for the vibration signal obtained through the compression sensing technique. In addition, through the experimental results, we confirmed the possibility of power consumption reduction when speeding up the signal processing and applying it to actual data measuring instruments. Even if using uncompressed compressed data, it is similar to conventional Shannon-Nyquist based measurement data The results are shown in Fig.

Hereinafter, a system and method for diagnosing a condition of a structure of a compression sensing infrastructure according to the present invention apply a new data measurement method called a compressive sensing method to a data measurement process and, without restoring an existing compressed signal, By statistically analyzing the data signal, it is possible to efficiently grasp the state of the object to be measured (in particular, the rotating device, the rotating system, etc.) using only a small amount of data in the data acquisition process.

As shown in FIG. 3, such a compression sensing technique does not require a decompression process of a compressed signal, so that it is possible to derive a condition diagnosis result of a measurement object more quickly than an existing measurement technique, It is possible to reduce the error of the state diagnosis due to the occurrence of the restoration error.

1 is a block diagram of a system for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention. Referring to FIG. 1, a system for diagnosing state of a structure based on a compression sensing according to an embodiment of the present invention will be described in detail.

1, the system for diagnosing a state of a compression sensing infrastructure according to an exemplary embodiment of the present invention may include a vibration sensor unit 100, a signal processing unit 200, and an analysis unit 300, The vibration sensor unit 100, the signal processing unit 200 and the analysis unit 300 may be connected to each other by wires or wirelessly to transmit and receive signals and data to each other. The signal processing unit 200 and the analysis unit 300 Is preferably performed through a manager terminal of the compression sensing infrastructure condition diagnosis system.

To learn more about each configuration,

The vibration sensor unit 100 includes a plurality of vibration sensors provided on a measurement object, and can measure vibrations generated from the measurement object.

It is most preferable that the object to be measured is a rotating machine that is most representative and commonly used in the industry as a whole. As described above, since the rotating system has a high rotation frequency and a complicated structure, the frequency of occurrence of faults is high, The state of the system for diagnosing a state of the structure based on the compression sensing according to an embodiment of the present invention is also limited to a rotating body system. However, this is only an embodiment of the present invention, and is applicable to all measurement objects capable of measuring the vibration and performing the condition diagnosis.

The signal processing unit 200 receives the measurement value measured from the vibration sensor unit 100 and generates a compression sensing signal by applying a preset compression sensing technique.

The signal processing unit 200 applies the compression sensing technique to the remaining measurement values except for '0' among the measurement values transmitted from the vibration sensor unit 100, and performs a second matrix operation with a normal distribution, A compression sensing signal can be generated.

A quadratic matrix having a rectification distribution is defined by the following mathematical expression.

(Where y is a compression sensing signal,

x is the original measurement (measurement signal),

Ф is a predetermined compression sensing matrix (Compressive measurement matrix)

More specifically, as shown in FIG. 4, the compression sensing technique of the signal processing unit 200 performs data measurement through a linear measurement model. It is possible to generate a compressed sensing signal through a second matrix operation process in the form of a raw measurement value x, M * N (N = data size of original signal and M = data size of compression sensing signal)

At this time, it is preferable that the compression sensing measurement matrix? Satisfy the RIP (Restricted Isometry Property) condition. The RIP condition is that the compression sensing measurement matrix Φ projects the original measured value x with a uniform energy, and the signal projected with a constant energy can be compressed and restored reliably. Accordingly, the generated compression sensing signal reflects the properties and characteristics of the original measurement value x, which is an analog signal, and performs the condition diagnosis of the measurement object using the compressed sensing signal as it is .

In this case, the types of the measurement matrix that generally satisfy the RIP condition are a matrix in which matrix components such as a Random Gaussian matrix and a Random Binary matrix are distributed randomly (irregularly). This is because the compression- This means that it is a process of measuring data at random rather than a process of measuring.

As the length of the measurement signal increases in the continuous compression sensing process, the size of the measurement matrix drastically increases, which inevitably increases the storage memory and the complexity of the system processing.

On the other hand, in the case of the state-of-the-art diagnosis system for a compression sensing infrastructure according to an embodiment of the present invention, considering the data size and optimal compression ratio suitable for relative diagnosis of the rotating body, By initially setting the matrix, the problem of data processing complexity and storage memory increase can be easily solved.

In addition, although the arrangement form of the measurement matrix components in the universal compression sensing varies from one measurement to another, the compression sensing infrastructure condition diagnosis system according to the embodiment of the present invention uses a constant measurement matrix, There is an advantage that defect can be discriminated.

The analysis unit 300 may analyze the state of the measurement object by analyzing the compression sensing signal received from the signal processing unit 200.

In detail, the analysis unit 300 substitutes a compression sensing signal received from the signal processing unit 200 into signal characteristics previously set so as to quantitatively indicate the state of the measurement object, thereby diagnosing the state of the measurement object It is possible to judge whether or not the damage has occurred, and furthermore, the damaged area can be judged.

That is, the analyzer 300 may set various signal characteristics that can quantitatively indicate the state of the measurement object, as shown in Table 1 below, using the statistical diagnostic method, The statistical distribution of the signal feature values can be analyzed to determine whether the damage has occurred.

In addition, it is possible to judge the damaged part according to the matching signal characteristic.

Features Formula Features Formula P1

P11

P2

P12

P3

P13

P4

P14

P5

P15

P6

P16

P7

P17

P8

P18

P9

P19

P10

P20

At this time, as shown in Table 1, it is common that various signal characteristics set in advance indicate signal characteristics in the time domain.

In addition, the analysis unit 300 generates two statistical models for the compression sensing signal transmitted from the signal processing unit 200 using a preset ROC (Receiver Operating Characteristic) analysis scheme, The degree of sensitivity and degree of damage to the damage classification of the measurement object can be determined according to the degree of overlapping of the models.

In detail, the analysis unit 300 can generate two statistical models for the compression sensing signal, as shown in FIG. 5, using the ROC line analysis technique. The degree of overlap of two statistical models can be determined based on the state diagnostic reference line and the sensitivity and specificity of the degree of overlap of two statistical models can be calculated through Equation (2) below.

Based on the calculated sensitivity and specificity, the ROC curve can be obtained as shown in FIG. The value corresponding to the area of the ROC line can be defined as the Area Under Curve (AUC) value, and the degree of overlap of the two statistical models can be quantitatively expressed through the AUC value. Accordingly, the sensitivity of the damage classification due to damage of the measurement object can be grasped, and the degree of damage can be grasped.

2 is a flowchart illustrating a method for diagnosing a state of a structure based on a compression sensing according to an embodiment of the present invention. With reference to FIG. 2, a method for diagnosing a state of a compression sensing based structure according to an embodiment of the present invention will be described in detail.

The method for diagnosing a state of a compression sensing infrastructure according to an embodiment of the present invention may include a vibration measurement step S100, a compression sensing step S200, and a state diagnosis step S300, as shown in FIG.

To learn more about each step,

The vibration measurement step (S100) can measure the vibration generated in the measurement object through the vibration sensor unit (100).

In the signal processing unit 200, the compression sensing step S200 receives a measurement value measured by the vibration measuring step S100 and applies a predetermined compression sensing technique to the compression sensing signal S200, Compressive measurement.

In the compression sensing step S200, the compression sensing technique is applied to the remaining measurement values except for '0' among the measurement values received from the vibration measurement step S100, and a second matrix calculation process having a normal distribution is performed The compression sensing signal can be generated.

The quadratic matrix having the rectification distribution is defined as Equation (1).

More specifically, as shown in FIG. 4, the compression sensing method of the compression sensing step (S200) proceeds with data measurement through a linear measurement model. The compressed sensing signal can be generated through a second matrix operation with a rectified distribution of x * M * N,

At this time, it is preferable that the compression sensing measurement matrix? Satisfy the RIP (Restricted Isometry Property) condition. The RIP condition is that the compression sensing measurement matrix Φ projects the original measured value x with a uniform energy, and the signal projected with a constant energy can be compressed and restored reliably. Accordingly, the generated compression sensing signal reflects the properties and characteristics of the original measurement value x, which is an analog signal, and performs the condition diagnosis of the measurement object using the compressed sensing signal as it is .

In the analysis unit 300, the state diagnosis step S300 may analyze the state of the measurement object by analyzing the compression sensing signal transmitted in the compression sensing step S200.

As shown in FIG. 2, the state diagnosis step S300 preferably includes a quantitative analysis step S310 and an ROC analysis step S320.

In the quantitative analysis step S310, the state of the measurement object is diagnosed by substituting the compression sensing signal transmitted by the compression sensing step S200 on a signal characteristic preset to quantitatively indicate the state of the measurement object It is possible to determine whether or not the damage has occurred, and furthermore, the damaged area can be judged.

In the quantitative analysis step S310, various signal characteristics capable of quantitatively indicating the state of the measurement object are set in advance, as shown in Table 1, using statistical condition diagnosis techniques, and the signal The statistical distribution of the feature values can be analyzed to determine whether the damage has occurred.

In addition, it is possible to judge the damaged part according to the matching signal characteristic.

The ROC analysis step (S320) generates two statistical models for the compression sensing signal transmitted by the compression sensing step (S200) using a preset ROC line analysis technique, and calculates the overlapping degree of the statistical models The degree of sensitivity and degree of damage to the damage classification of the measurement object can be determined.

In more detail, the ROC analysis step (S320) may generate two statistical models for the compression sensing signal, as shown in FIG. 5, using the ROC line analysis technique. The degree of overlap of two statistical models is determined based on the state diagnostic baseline and the sensitivity and specificity of the degree of overlap of the two statistical models can be calculated through Equation (2).

Based on the calculated sensitivity and specificity, the ROC curve can be obtained as shown in FIG. The value corresponding to the area of the ROC line can be defined as the Area Under Curve (AUC) value, and the degree of overlap of the two statistical models can be quantitatively expressed through the AUC value. Accordingly, the sensitivity of the damage classification due to damage of the measurement object can be grasped, and the degree of damage can be grasped.

Compared to the conventional state diagnosis system, the system for diagnosing state of the structure of the compression sensing infrastructure according to the embodiment of the present invention and the method thereof can confirm (diagnose) the state of the measurement object accurately with a very small data value as compared with the measurement sampling frequency As a result of the experiment, as shown in FIGS. 7 and 8, the comparison result can be confirmed.

In detail, FIG. 7 shows the signal characteristics of the normal and bearing defects extracted from the data measured based on the Shannon-Nyquist sampling frequency, the normal and bearing defects extracted by the compression sensing signal corresponding to 1/4 of the Shannon-Nyquist sampling frequency Fig.

As shown in FIG. 7, it can be seen that the difference in the signal characteristic values between the normal and the defect is similar. More specifically, in the case of the signal characteristics of the normal and the bearing defects extracted from the data measured based on the Shannon-Nyquist sampling frequency , The signal feature value difference between normal and defect is about 1.91 times, and in case of signal characteristics of normal and bearing defect extracted by compression sensing signal corresponding to 1/4 of Shannon-Nyquist sampling frequency, Can be confirmed to be about 1.97 times.

8 is a graph showing the relationship between the AUC value of the ROC curve of the normal and bearing defect states of the data measured based on the Shannon-Nyquist sampling frequency and the AUC value of the data measured with the compression sensing signal corresponding to 1/4 of the Shannon- Lt; RTI ID = 0.0 > ROC < / RTI > curve of normal and bearing fault conditions.

As shown in FIG. 8, it can be seen that the distribution model of the statistical model and the AUC value of the ROC curve coincide with each other according to the damaged area and the type of damage.

Accordingly, it is possible to construct a low-cost, low-frequency-area DAQ device while deriving the same diagnosis result as that of the conventional state diagnostic system, and a compression-sensing-based condition diagnosis system and method thereof according to an embodiment of the present invention, There is an effect that the signal processing and the energy efficiency are remarkably increased as compared with the diagnostic system.

In addition, as described above, there is no need to restore a compressed compression sensing signal, so that an error generated in a restoration process can be minimized, and thus, it can be efficiently applied.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, And various modifications and changes may be made thereto by those skilled in the art to which the present invention pertains.

Therefore, it is to be understood that the subject matter of the present invention is not limited to the described embodiments, and all of the equivalents or equivalents of the claims are included in the scope of the present invention will be.

100: vibration sensor unit
200: Signal processor
300: Analysis section

Claims (7)

A vibration sensor unit (100) comprising: a vibration sensor unit (100) including a plurality of vibration sensors provided on a measurement object to measure vibrations generated in the measurement object;
A signal processor 200 receiving the measurement values measured by the vibration sensor 100 and generating a compressed sensing signal using a compressive sensing technique; And
An analysis unit (300) for analyzing the compression sensing signal transmitted from the signal processing unit (200) and diagnosing the state of the measurement object;
And a system for diagnosing condition of a structure of a compression sensing infrastructure.
The method according to claim 1,
The signal processing unit 200
(Y = [phi] x) calculation process using a compression sensing technique for the remaining measured values except for '0' among the measured values transmitted from the vibration sensor unit 100, Wherein the system is configured to generate the state of the compression sensing based structure.
(Where y is a compression sensing signal,
x is the original measurement (measurement signal),
Ф is a predetermined compression sensing matrix (Compressive measurement matrix)
The method according to claim 1,
The analyzing unit 300
The state of the measurement object is diagnosed by substituting the compression sensing signal transmitted from the signal processing unit 200 into the predetermined signal characteristic so as to quantitatively indicate the state of the measurement object, Based on the result of the comparison.
The method of claim 3,
The analyzing unit 300
And generates two statistical models for the compression sensing signal transmitted from the signal processing unit 200 by using a predetermined ROC (Receiver Operating Characteristic) leading analysis technique, Wherein the sensitivity and the degree of intensification of the damage classification of the damage detection system are determined.
A vibration measuring step (S100) of measuring vibrations generated in the measurement object from a plurality of vibration sensors provided in the measurement object;
A compression sensing step (S200) of generating a compression sensing signal by using a predetermined compression sensing technique based on the measured value measured in the vibration measuring step (S100); And
A state diagnosis step (S300) of analyzing the compression sensing signal generated in the compression sensing step (S200) and performing a state diagnosis of the measurement object;
Wherein the method comprises the steps of:
6. The method of claim 5,
The compression sensing step (S200)
(Y = [phi] x) calculation process using a compression sensing technique for the remaining measurement values except for '0' among the measurement values received from the vibration measurement step S100, Based on the first signal and the second signal.
(Where y is a compression sensing signal,
x is the original measurement (measurement signal),
Ф is a predetermined compression sensing matrix (Compressive measurement matrix)
6. The method of claim 5,
The state diagnosis step (S300)
The compression sensing signal transmitted from the compression sensing step (S200) is substituted into a predetermined signal characteristic to quantitatively indicate the state of the measurement object to determine whether or not a damage has occurred to the measurement object, Analysis step S310; And
The method comprising: generating two statistical models for the compression sensing signal using a pre-established ROC (Receiver Operating Characteristic) leading analysis technique; and then, based on the degree of overlap of the statistical models, An ROC analysis step (S320) of judging the degree of damage intensification;
Wherein the first and second sensors are configured to detect the state of the structure based on the compression sensing.

Images