Time synchronization system, time synchronization program, and time synchronization method

Abstract

To provide a time synchronization system, a time synchronization program, and a time synchronization method capable of ensuring the accuracy of time synchronization between a plurality of vibration data through highly-free and simple processing or method.SOLUTION: The time synchronization system synchronizes time between first vibration data and second vibration data by using the first vibration data and the second vibration data acquired at separate positions of structures in which vibration occurs. The time synchronization system comprises a calculation unit which is used for estimating time deviation between the first vibration data and the second vibration data, sequentially calculates a plurality of estimation transmission functions each including one of a plurality of temporary estimation values of time deviation, and determines, as a first optimal value for synchronizing time, a temporary estimation value of the time deviation in which a phase value at the estimation transmission function most concentrates in a direction of a binary of 0 and 180 degrees among the plurality of temporary estimation values of the time deviation. The time synchronization program and the time synchronization method are configured to correspond to the above.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a time synchronization system, a time synchronization program, and a time synchronization method for synchronizing the time between a plurality of vibration data acquired in a structure in which vibration occurs. [Background technology]

[0002] When multiple vibration sensors for measuring vibrations are installed in a structure such as an architectural structure where vibrations such as earthquake motions occur, it is necessary to synchronize the times of the multiple vibration sensors in order to accurately analyze the vibration data acquired by the multiple vibration sensors. A sensor unit equipped with a vibration sensor is equipped with a built-in clock for identifying the time when the vibration occurred, but the accuracy of the built-in clock is not high, and the time of the built-in clock becomes misaligned over time. As a method for adjusting the time of the built-in clock, for example, a communication environment is constructed in which multiple sensor units are connected by wire or wirelessly, and the time is adjusted using a Global Positioning System (GPS) server, a Network Time Protocol (NTP) server, or an atomic clock transmitted from a standard radio wave transmitting station.

[0003] However, when adjusting the time via a wired or wireless connection, high-precision time adjustment may not be guaranteed due to problems in the communication environment such as radio interference, communication delays, etc. In addition, there are structures in which it is difficult to create a communication environment, whether via a wired or wireless connection, such as nuclear power plant buildings, studios, MRI (Magnetic Resonance Image) rooms, and laboratories, where the walls, floors, or ceilings are electromagnetically shielded.

[0004] Even if it is possible to create a suitable communication environment in a structure, incorporating a device for receiving GPS or the like into the sensor unit increases the power consumption of the sensor unit. This requires an external power source for the sensor unit or a large capacity internal battery, which leads to an increase in the size of the sensor unit or an increase in operating costs. In addition, costs are incurred when connecting sensors to each other, such as wiring work being required for wired connections and a parent device other than the sensor or a repeater depending on the communication situation being required for wireless connections.

[0005] Therefore, in vibration data analysis, a method is often adopted in which, after acquiring the vibration data, multiple pieces of vibration data are time-synchronized, regardless of whether a wired or wireless connection is possible (e.g., Patent Documents 1 to 3). [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2007-327873 A [Patent Document 2] JP 2018-091824 A [Patent Document 3] JP 2018-100875 A Summary of the Invention [Problem to be solved by the invention]

[0007] Patent Document 1 discloses a method for extracting the waveforms of the rising portions of earthquake record data acquired by multiple seismometers, comparing the rising portion waveforms with each other to calculate the time shift, and performing time synchronization between each earthquake record data.

[0008] However, in Patent Document 1, it is essential to record the rising part of the vibration, and if the rising part of the vibration cannot be recorded for some reason, it may cause problems in time synchronization. Also, since only the rising part of the waveform of the earthquake record data is used, time synchronization calculations must be performed with a short waveform, and there is room for improvement in terms of ensuring accuracy. Furthermore, the result of time synchronization may differ depending on the set length of the range for extracting the rising part waveform from the earthquake record data.

[0009] Also, Patent Document 2 discloses a method for calculating a time shift between first vibration waveform data on the input side and second vibration waveform data on the output side using a transfer function between the first vibration waveform data and the second vibration waveform data. In Patent Document 2, the time shift between the first vibration waveform data and the second vibration waveform data is calculated based on the correlation between the first vibration waveform data and third vibration waveform data obtained by removing the influence of the transfer function from the second vibration waveform data.

[0010] However, in Patent Document 2, the first vibrometer that acquires the first vibration waveform data needs to be installed on the vibration source side, and therefore the positional relationship between the first vibrometer and the second vibrometer that acquires the second vibration waveform data is limited.

[0011] Patent Document 2 also discloses a method for calculating a transfer function that approximates a structure using a theoretical formula for the transfer function of a one-mass system when the structure on which the vibrometer is installed is a simple structure such as a wooden house that can be regarded as a one-mass system. Patent Document 2 also discloses a method for calculating a transfer function when the structure on which the vibrometer is installed is a complex structure that can be regarded as a vibration model other than a one-mass system, for example, a vibration model of a multi-mass system.

[0012] However, if the structure on which the vibrometer is installed is a complex structure that can be regarded as a vibration model other than a single mass point system, it is necessary to prepare a complex structure model using the finite element method or the like, and it takes a lot of time and effort to calculate the transfer function. In addition, since the accuracy of the transfer function depends on the accuracy of the created structure model, the modeling work of the structure may affect the result of time synchronization. Therefore, if the structure in Patent Document 2 is a vibration model other than a single mass point system, it is difficult to calculate the transfer function, and even if the transfer function is calculated, the accuracy of time synchronization may not be ensured.

[0013] Furthermore, Patent Document 3 discloses a method of calculating a correlation coefficient between first acceleration data and second acceleration data, and correcting a time lag between the first acceleration data and the second acceleration data based on the correlation coefficient.

[0014] However, since Patent Document 3 is intended for a low-rise wooden building that has a large first-order natural frequency and can be regarded as a vibration model of a single-mass system, it may not be possible to ensure the accuracy of time synchronization in a structure that has a small first-order natural frequency and can be regarded as a vibration model of a multi-mass system.

[0015] The present invention is intended to solve the above-mentioned problems, and aims to provide a time synchronization system, a time synchronization program, and a time synchronization method that can ensure the accuracy of time synchronization between multiple vibration data through a highly flexible and simple process or method. [Means for solving the problem]

[0016] The time synchronization system of the present invention is a time synchronization system that uses first vibration data and second vibration data acquired at separate positions of a structure in which vibration occurs to synchronize the time between the first vibration data and the second vibration data, and includes a calculation unit that sequentially calculates a plurality of estimated transfer functions that are used to estimate a time shift between the first vibration data and the second vibration data, each of which includes one of a plurality of tentative estimates of the time shift, and determines, among the plurality of tentative estimates of the time shift, the tentative estimate of the time shift in which the phase values ​​in the estimated transfer function are most concentrated in the two-value directions of 0 degrees and 180 degrees, as a first optimal value for synchronizing the time.

[0017] According to the time synchronization system of the present invention, it is possible to synchronize time using vibration data of any two points of a structure without determining reference vibration data. Also, according to the time synchronization system of the present invention, it is possible to synchronize time between multiple vibration data without being restricted by a known vibration model of the structure. Therefore, according to the time synchronization system of the present invention, it is possible to achieve highly accurate time synchronization between vibration data with a high degree of freedom.

[0018] Furthermore, according to the time synchronization system of the present invention, the time between the first vibration data and the second vibration data can be synchronized by using the entirety of the acquired first vibration data and second vibration data. Therefore, according to the time synchronization system of the present invention, the time between the vibration data can be automatically synchronized without using a specific part of the vibration data such as the rising part of the waveform, a frequency band below the low-order natural frequency, or a resonance frequency, and therefore the time between the vibration data can be easily synchronized with high accuracy.

[0019] In one aspect of the present invention, the calculation unit synchronizes the time between the first vibration data and the second vibration data based on the first optimal value, calculates a coherence value between the first vibration data after the time is synchronized and the second vibration data after the time is synchronized, weights phase values ​​in a plurality of the estimated transfer functions using the coherence value, and determines, among the plurality of tentative estimates of the time shift, the tentative estimate of the time shift in which the phase values ​​are most concentrated in two directions of 0 degrees and 180 degrees as a second optimal value for synchronizing the time.

[0020] According to one aspect of the present invention, a true estimate of the time offset can be more accurately determined.

[0021] Further, a time synchronization program of the present invention is a time synchronization program that uses first vibration data and second vibration data acquired at separate positions of a structure in which vibration occurs to synchronize the time between the first vibration data and the second vibration data, and sequentially calculates a plurality of estimated transfer functions that are used to estimate a time shift between the first vibration data and the second vibration data, each of which includes one of a plurality of tentative estimates of the time shift, and determines, among the plurality of tentative estimates of the time shift, the tentative estimate of the time shift in which the phase value in the estimated transfer function is most concentrated in the two-value directions of 0 degrees and 180 degrees, as a first optimal value for synchronizing the time.

[0022] According to the time synchronization program of the present invention, it is possible to synchronize time using vibration data of any two points of a structure without determining reference vibration data. Also, according to the time synchronization system of the present invention, it is possible to synchronize time between vibration data without being restricted by a known vibration model of the structure. Therefore, according to the time synchronization system of the present invention, it is possible to achieve highly accurate time synchronization between vibration data with a high degree of freedom.

[0023] Furthermore, according to the time synchronization program of the present invention, the first vibration data and the second vibration data can be used in their entirety to synchronize the time between them. Therefore, according to the time synchronization system of the present invention, the time between the vibration data can be automatically synchronized without using a specific part of the vibration data such as the rising part of the waveform, a frequency band below the low-order natural frequency, or a resonance frequency, and therefore the time between the vibration data can be easily synchronized with high accuracy.

[0024] Further, a time synchronization method of the present invention is a method for synchronizing time between first vibration data and second vibration data by using first vibration data and second vibration data acquired at separate positions of a structure in which vibration occurs, and includes a step of sequentially calculating a plurality of estimated transfer functions used for estimating a time shift between the first vibration data and the second vibration data, each of the estimated transfer functions including one of a plurality of tentative estimates of the time shift, and a step of determining, as a first optimal value for synchronizing the time, the tentative estimate of the time shift among the plurality of tentative estimates of the time shift, the tentative estimate of the time shift in which the phase value in the estimated transfer function is most concentrated in the two-value directions of 0 degrees and 180 degrees.

[0025] According to the time synchronization method of the present invention, it is possible to synchronize time using vibration data of any two points of a structure without determining reference vibration data. Also, according to the time synchronization system of the present invention, it is possible to synchronize time between vibration data without being restricted by a known vibration model of the structure. Therefore, according to the time synchronization system of the present invention, it is possible to achieve highly accurate time synchronization between vibration data with a high degree of freedom.

[0026] Furthermore, according to the time synchronization method of the present invention, the time synchronization between the first vibration data and the second vibration data can be performed by using the entirety of the acquired first vibration data and second vibration data. Therefore, according to the time synchronization system of the present invention, the time synchronization between the vibration data can be automatically performed without using a specific part of the vibration data such as the rising part of the waveform, a frequency band below the low-order natural frequency, or a resonance frequency, and therefore the time synchronization between the vibration data can be easily performed with high accuracy. Effect of the Invention

[0027] According to the present invention, it is possible to provide a time synchronization system, a time synchronization program, and a time synchronization method that are capable of ensuring the accuracy of time synchronization between multiple vibration data using a highly flexible and simple process or method. [Brief description of the drawings]

[0028] [Figure 1] FIG. 1 is an overall view showing an example of the relationship between a time synchronization system according to an embodiment of the present invention and a sensor unit installed in a structure. [Diagram 2] FIG. 2 is a schematic diagram showing an example of functional blocks of a time synchronization system and a sensor unit according to an embodiment of the present invention. [Diagram 3]FIG. 3 is a graph showing the characteristics of a transfer function representing the relationship between the first vibration data and the second vibration data, obtained by a simulation using actual vibration data. FIG. 3(a) is a graph showing the amplification of the transfer function with respect to the frequency (Hz) when there is no time lag between the first vibration data and the second vibration data, with the horizontal axis showing the frequency (Hz) and the vertical axis showing the amplification of the transfer function. FIG. 3(b) is a graph showing the phase of the transfer function with respect to the frequency (Hz) when there is no time lag between the first vibration data and the second vibration data, with the horizontal axis showing the frequency (Hz) and the vertical axis showing the phase of the transfer function in degrees. FIG. 3(c) is a graph showing the amplification of the transfer function with respect to the frequency (Hz) when the time lag between the first vibration data and the second vibration data is 1 second, with the horizontal axis showing the frequency (Hz) and the vertical axis showing the amplification of the transfer function. FIG. 3(d) is a graph showing the phase of the transfer function versus frequency (Hz) when the time lag between the first vibration data and the second vibration data is 1 second, where the horizontal axis shows frequency (Hz) and the vertical axis shows the phase of the transfer function in degrees. [Figure 4] FIG. 4 is a flowchart showing a calculation process according to an embodiment of the present invention. [Diagram 5] Fig. 5 is a graph showing the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the time lag between the first vibration data and the second vibration data is 1 second, based on a simulation using actual vibration data. In the graph of Fig. 5(a), the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from -5 seconds to 5 seconds, and the vertical axis shows the phase variance. The graph of Fig. 5(b) is an enlarged version of the horizontal axis of the graph of Fig. 5(a), where the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from 0 seconds to 1.1 seconds, and the vertical axis shows the phase variance. [Figure 6]Fig. 6 is a graph showing the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the time lag between the first vibration data and the second vibration data is 0.5 seconds, based on a simulation using actual vibration data. In the graph of Fig. 6(a), the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from -5 seconds to 5 seconds, and the vertical axis shows the phase variance. The graph of Fig. 6(b) is an enlarged version of the horizontal axis of the graph of Fig. 6(a), where the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from 0 seconds to 1.1 seconds, and the vertical axis shows the phase variance. [Figure 7] Fig. 7 is a graph showing the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the time lag between the first vibration data and the second vibration data is 0.1 seconds, based on a simulation using actual vibration data. In the graph of Fig. 7(a), the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from -5 seconds to 5 seconds, and the vertical axis shows the phase variance. The graph of Fig. 7(b) is an enlarged version of the horizontal axis of the graph of Fig. 7(a), where the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from 0 seconds to 1.1 seconds, and the vertical axis shows the phase variance. [Figure 8] FIG. 8 is a schematic diagram showing an example of functional blocks of a time synchronization system and a sensor unit according to a modified example of the embodiment of the present invention. [Figure 9]FIG. 9 is a graph showing the relationship between the transfer function representing the relationship between the first vibration data and the second vibration data and the coherence value between the first vibration data after synchronization and the second vibration data after synchronization, which is obtained by simulation using actual vibration data. FIG. 9(a) is the same graph as FIG. 3(a), and shows the amplification of the transfer function with respect to the frequency (Hz) when there is no time lag between the first vibration data and the second vibration data. In FIG. 9(a), the horizontal axis shows the frequency (Hz), and the vertical axis shows the amplification of the transfer function. FIG. 9(b) is the same graph as FIG. 3(b), and shows the phase of the transfer function with respect to the frequency (Hz) when there is no time lag between the first vibration data and the second vibration data. In FIG. 9(b), the horizontal axis shows the frequency (Hz), and the vertical axis shows the phase of the transfer function in degrees. FIG. 9(c) is a graph showing the coherence value between the first vibration data after synchronization and the second vibration data after synchronization. In Fig. 9(c), the horizontal axis indicates the frequency (Hz), and the vertical axis indicates the coherence value. In Fig. 9(c), a continuous curve in the case where there is no time lag between the first vibration data and the second vibration data is shown by a thin black solid line, and a continuous curve in the case where the time lag between the first vibration data and the second vibration data is 0.1 seconds is shown by a black dashed line. [Figure 10] FIG. 10 is a flowchart showing the calculation process according to this modified example. [Figure 11]FIG. 11 is a graph showing the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the time lag between the first vibration data and the second vibration data is 1 second, based on a simulation using actual vibration data. FIG. 11(a) and FIG. 11(b) show the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the phase value is not weighted by the coherence value in the estimated transfer function. The graph in FIG. 11(a) is the same as FIG. 5(a), with the horizontal axis showing the tentative estimate of the time lag (seconds) in the range from -5 seconds to 5 seconds, and the vertical axis showing the phase variance. The graph in FIG. 11(b) is the same as FIG. 5(b), with the horizontal axis of the graph in FIG. 11(a) enlarged, with the horizontal axis showing the tentative estimate of the time lag (seconds) in the range from 0 seconds to 1.1 seconds, and the vertical axis showing the phase variance. Figures 11(c) and 11(d) show the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the phase value is weighted by the coherence value in the estimated transfer function. In Figure 11(c), the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from -5 to 5 seconds, and the vertical axis shows the phase variance. The graph in Figure 11(d) is an enlarged horizontal axis of the graph in Figure 11(c), where the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from 0 to 1.1 seconds, and the vertical axis shows the phase variance. [Figure 12]FIG. 12 is a graph showing the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the time lag between the first vibration data and the second vibration data is 0.5 seconds, based on a simulation using actual vibration data. FIG. 12(a) and FIG. 12(b) show the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the phase value is not weighted by the coherence value in the estimated transfer function. The graph in FIG. 12(a) is the same as FIG. 6(a), with the horizontal axis showing the tentative estimate of the time lag (seconds) in the range from -5 seconds to 5 seconds, and the vertical axis showing the phase variance. The graph in FIG. 12(b) is the same as FIG. 6(b), with the horizontal axis of the graph in FIG. 12(a) enlarged, with the horizontal axis showing the tentative estimate of the time lag (seconds) in the range from 0 seconds to 1.1 seconds, and the vertical axis showing the phase variance. Figures 12(c) and 12(d) show the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the phase value is weighted by the coherence value in the estimated transfer function. In Figure 12(c), the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from -5 to 5 seconds, and the vertical axis shows the phase variance. The graph in Figure 12(d) is an enlarged horizontal axis of the graph in Figure 12(c), where the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from 0 to 1.1 seconds, and the vertical axis shows the phase variance. [Figure 13]FIG. 13 is a graph showing the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the time lag between the first vibration data and the second vibration data is 0.1 seconds, based on a simulation using actual vibration data. FIG. 13(a) and FIG. 13(b) show the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the phase value is not weighted by the coherence value in the estimated transfer function. The graph in FIG. 13(a) is the same as FIG. 7(a), with the horizontal axis showing the tentative estimate of the time lag (seconds) in the range from -5 seconds to 5 seconds, and the vertical axis showing the phase variance. The graph in FIG. 13(b) is the same as FIG. 7(b), with the horizontal axis of the graph in FIG. 13(a) enlarged, with the horizontal axis showing the tentative estimate of the time lag (seconds) in the range from 0 seconds to 1.1 seconds, and the vertical axis showing the phase variance. Figures 13(c) and 13(d) show the relationship between the tentative estimate of the time lag and the phase variance of the estimated transfer function when the phase value is weighted by the coherence value in the estimated transfer function. In Figure 13(c), the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from -5 to 5 seconds, and the vertical axis shows the phase variance. The graph in Figure 13(d) is an enlarged horizontal axis of the graph in Figure 13(c), where the horizontal axis shows the tentative estimate of the time lag (seconds) in the range from 0 to 1.1 seconds, and the vertical axis shows the phase variance. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0029] A time synchronization system 1 according to an embodiment will be described with reference to Fig. 1. In the following drawings, the same components or parts, or components or parts having the same functions, are denoted by the same reference numerals or the reference numerals are omitted.

[0030] The time synchronization system 1 uses first vibration data and second vibration data acquired at separate positions in a vibration-generating structure 5 to synchronize the time between the first vibration data and the second vibration data. The first vibration data is one of a plurality of vibration data acquired at separate positions in a vibration-generating structure 5. The second vibration data is the other of the above-mentioned plurality of vibration data.

[0031] The multiple vibration data are acquired by multiple sensor units 2 installed at separate positions in the structure 5. The first vibration data is acquired by a first sensor unit 2a, which is one of the multiple sensor units 2. The second vibration data is acquired by a second sensor unit 2b, which is another one of the multiple sensor units 2. In this embodiment, the first sensor unit 2a and the second sensor unit 2b are arbitrarily selected from the multiple sensor units 2.

[0032] The multiple sensor units 2 are installed at any position of the structure 5. In FIG. 1, the multiple sensor units 2 are installed on the floor, but they may be installed on the ceiling or wall. Also, the multiple sensor units 2 may be installed on the roof or basement. Also, as shown in FIG. 1, the multiple sensor units 2 may be installed on the same floor of the structure 5.

[0033] 1, three-dimensional coordinate axes are defined based on the lower left corner of the front of the structure 5. The three-dimensional coordinate axes are defined by an X-axis extending from the left side to the right side of the structure 5, a Y-axis extending from the front side to the rear side of the structure 5, and a Z-axis extending from the bottom end of the structure 5 upward. However, the time synchronization system 1 of this embodiment is applicable to various structures 5, from simple structures 5 that can be expressed by a mass system model to complex structures 5 that cannot be expressed without using a finite element method or the like, and the arrangement of the multiple sensor units 2 is not limited to being aligned along a specific axis (for example, the Z-axis).

[0034] Furthermore, in this embodiment, the multiple sensor units 2 do not need to be fixed to fixed positions on the structure 5. For example, the sensor units 2 may be mounted on an autonomous mobile robot, and vibration data may be acquired while the robot is moved to an arbitrary position.

[0035] The vibration-generating structure 5 to which the time synchronization system 1 of the present embodiment can be applied is, for example, a wooden, steel-framed, or reinforced concrete architectural structure such as a wooden house or a high-rise building, but is not limited thereto. For example, the structure 5 may be a special architectural structure such as a nuclear power plant building, or a civil engineering structure such as a bridge, tower, or highway.

[0036] In addition, the type of vibration to which the time synchronization system 1 of the present embodiment can be applied is not limited to vibration caused by earthquakes. For example, the time synchronization system 1 of the present embodiment can be applied to vibration of architectural structures or civil engineering structures caused by the passage of vehicles, trains, or airplanes, or the movement of people, vibration of bridges or towers caused by natural phenomena such as strong winds, and vibration of architectural structures such as factories caused by the operation of industrial machines.

[0037] Next, the structures of the time synchronization system 1 and the sensor unit 2 will be described with reference to FIG.

[0038] The time synchronization system 1 is configured as a computer device or a microcomputer, dedicated hardware, or a combination thereof. As shown in FIG. 2, the time synchronization system 1 includes a calculation unit 11, a data output unit 13, and a data transfer unit 15.

[0039] When the time synchronization system 1 is a computer device or a microcomputer, the calculation unit 11 is configured as a CPU (Central Processing Unit) or MPU (Micro Processing Unit). When the calculation unit 11 is configured as a CPU or MPU, each function executed by the calculation unit 11 is realized by software, firmware, or a combination of software and firmware. The software or firmware is described as a program in a program language. As the program language, C language, Python, R language, etc. are used, but are not limited to them. The program is stored in an internal memory (not shown) of the time synchronization system 1, and the program stored in the internal memory (not shown) is read and executed by the CPU or MPU by the control process of the control unit (not shown). Each function in the calculation unit 11 is realized by the CPU or MPU reading and executing the program stored in the internal memory. The internal memory (not shown) is, for example, a non-volatile or volatile semiconductor memory such as a RAM, a ROM, a flash memory, an EPROM, or an EEPROM.

[0040] The type of computer device is not limited, but may be, for example, a virtual server that provides cloud services, or a physical server that provides various services to groups in a local network. The computer device may be a quantum computer, a general-purpose desktop or laptop computer, or a portable terminal such as a tablet terminal or a smartphone. The time synchronization system 1 may perform distributed processing using multiple computer devices.

[0041] When the time synchronization system 1 is a dedicated hardware, the calculation unit 11 is realized by, for example, a single circuit, a composite circuit, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a circuit that combines these. Each function realized by the calculation unit 11 may be realized by individual hardware, or all of the functions may be realized by a single hardware.

[0042] The calculation unit 11 has a plurality of functional blocks, for example, a data reading unit 11a1, a transfer function calculation unit 11a2, a phase variance value calculation unit 11a3, a time shift calculation unit 11a4, and a time shift correction unit 11a5. The data reading unit 11a1, the transfer function calculation unit 11a2, the phase variance value calculation unit 11a3, the time shift calculation unit 11a4, and the time shift correction unit 11a5 form a first time synchronization unit 11a, which is one functional block.

[0043] The data reading unit 11a1 reads the first vibration data and the second vibration data acquired at separate positions in the structure 5 (see FIG. 1) where vibration occurs into the inside of the calculation unit 11 via the data transfer unit 15. The reading of the first vibration data and the second vibration data in the data reading unit 11a1 can be performed by, for example, control processing in a control unit (not shown) provided inside the time synchronization system 1.

[0044] The transfer function calculation unit 11a2 calculates a transfer function representing a relationship between the first vibration data and the second vibration data based on the first vibration data and the second vibration data read by the data reading unit 11a1. The transfer function calculation unit 11a2 sequentially calculates a plurality of estimated transfer functions, which are used to estimate a time shift between the first vibration data and the second vibration data and include a plurality of tentative estimates of the time shift. In the transfer function calculation unit 11a2, a plurality of estimated transfer functions are sequentially calculated so that each of the plurality of estimated transfer functions includes one of a plurality of tentative estimates of the time shift between the first vibration data and the second vibration data. The calculated plurality of estimated transfer functions may be at least temporarily stored in a storage unit (not shown) such as an internal memory or an auxiliary storage device such as a hard disk provided in the time synchronization system 1 by a control process in a control unit (not shown).

[0045] The phase variance calculation unit 11a3 calculates a phase variance value for each of the multiple estimated transfer functions calculated by the transfer function calculation unit 11a2. The phase variance value is calculated for each provisional estimate of the time shift between the first vibration data and the second vibration data. The multiple phase variance values ​​calculated for each provisional estimate of the time shift are at least temporarily stored in a storage unit (not shown) such as an internal memory or an auxiliary storage device such as a hard disk provided in the time synchronization system 1 by a control process in a control unit (not shown).

[0046] The time shift calculation unit 11a4 determines the tentative time shift estimate value in which the phase value in the estimated transfer function is most concentrated in the binary directions of 0 degree and 180 degree, among the tentative time shift estimate values ​​in the estimated transfer function calculated in the phase variance value calculation unit 11a3, as a first optimum value which is a true estimate value for time synchronization. In the time shift calculation unit 11a4, the multiple phase variance values ​​calculated for each tentative time shift estimate value in the phase variance value calculation unit 11a3 are read out from a storage unit (not shown) by a control process in a control unit (not shown) for numerical comparison.

[0047] The time lag correction unit 11a5 synchronizes the time between the first vibration data and the second vibration data based on the first optimal value determined by the time lag calculation unit 11a4. In the time lag correction unit 11a5, the time synchronization between the first vibration data and the second vibration data is performed automatically. Note that the time synchronization between the first vibration data and the second vibration data can also be performed manually by operating an input device (not shown) such as a keyboard or a mouse, as long as the operator can know the first optimal value. When the time synchronization is performed manually, the calculation process in the time lag correction unit 11a5 may be omitted.

[0048] The data output unit 13 outputs the first vibration data and the second vibration data, the time of which is synchronized based on the first optimal value in the time lag correction unit 11a5, for data analysis. The data output unit 13 may be an output device such as a monitor or a printer, or an output port for connecting an output device. When the data output unit 13 is an output port, the data output unit 13 is configured with a USB port, an HDMI (registered trademark) port, a parallel port, a serial port, or the like.

[0049] The data transfer unit 15 is an interface for reading the first vibration data and the second vibration data acquired in the structure 5 (see FIG. 1) into the time synchronization system 1. The data transfer unit 15 may be a wireless communication port for Wi-Fi communication, NFC communication, WiMAX (registered trademark) communication, or the like, or may be a wired communication port for Ethernet (wired LAN) communication, or the like. In addition, when the structure 5 (see FIG. 1) is an environment in which it is difficult to establish a communication environment, such as a nuclear power plant building, the data transfer unit 15 may be a connection port for connecting the sensor unit 2 removed from the structure 5 and directly receiving vibration data from the sensor unit 2. The connection port for directly receiving vibration data from the sensor unit 2 may be, but is not limited to, a USB connection port for USB connection with the sensor unit 2, or a wired connection port such as an IEEE1394 port for IEEE1394 connection with the sensor unit 2. In addition, the connection port for directly receiving vibration data from the sensor unit 2 may be a wireless connection port for Bluetooth (registered trademark) communication or infrared communication. Furthermore, the data transfer section 15 may be a removable media reader for receiving the vibration data acquired by the sensor unit 2 offline via a removable media such as a USB memory or an SD card.

[0050] As shown in FIG. 2, the sensor unit 2 has a processing unit 21, a measuring unit 23, a built-in clock 25, a storage unit 27, and a data transfer unit 29 as functional blocks.

[0051] The processing unit 21 controls the measurement unit 23, the built-in clock 25, the storage unit 27, and the data transfer unit 29. The processing unit 21 is configured as a microcomputer, dedicated hardware, or a combination thereof.

[0052] When the processing unit 21 is a microcomputer, the processing unit 21 has an MPU. Each function executed by the MPU is realized by software, firmware, or a combination of software and firmware. The software or firmware is described as a program. The program is stored in an internal memory (not shown) provided in the processing unit 21, and the program stored in the internal memory (not shown) is read out and executed by the MPU. Each function in the sensor unit 2 is realized by the MPU reading out and executing the program stored in the internal memory. The internal memory (not shown) is, for example, a non-volatile or volatile semiconductor memory such as a RAM, a ROM, a flash memory, an EPROM, or an EEPROM.

[0053] When the processing unit 21 is a dedicated hardware, the processing unit 21 is realized by, for example, a single circuit, a composite circuit, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a circuit that combines these. Each function realized by the processing unit 21 may be realized by individual hardware, or all of the functions may be realized by a single hardware.

[0054] The measuring unit 23 is a vibration sensor that converts vibrations detected in the structure 5 (see FIG. 1) into an electronic signal. The vibration sensor is also called a vibrometer, and has an element that detects vibrations and a signal processing circuit that adjusts the electrical signal from the element.

[0055] The measuring unit 23 may be a contact type vibration sensor that is in direct contact with the structure 5 (see FIG. 1), or a non-contact type vibration sensor that measures vibration from the structure 5 (see FIG. 1) without contact. The contact type vibration sensor may be an acceleration sensor that is an acceleration detection type vibration sensor, a velocity sensor that is a velocity detection type vibration sensor, or a displacement sensor that is a displacement detection type vibration sensor. The acceleration sensor may be, for example, a piezoelectric type acceleration sensor, or an electrodynamic type, strain gauge type, or semiconductor type acceleration sensor. For example, an electrodynamic type velocity sensor is used as the velocity sensor. For example, a strain gauge type displacement sensor is used as the contact type displacement sensor. Also, for example, a displacement sensor that is a displacement detection type vibration sensor is used as the non-contact type vibration sensor. For example, the non-contact type displacement sensor may be an eddy current type displacement sensor, or a capacitance type, optical type, or differential transformer type displacement sensor.

[0056] In this embodiment, vibration sensors suitable for detecting vibrations in the structure 5 (see FIG. 1) are, but are not limited to, piezoelectric acceleration sensors, electrodynamic velocity sensors, and eddy current displacement sensors. Among them, piezoelectric acceleration sensors are suitable for detecting vibrations in the structure 5 (see FIG. 1) because they have a wider range of detectable vibration frequencies than electrodynamic velocity sensors and eddy current displacement sensors. In addition, piezoelectric acceleration sensors have the characteristics of being miniaturizable and having good mechanical strength and environmental resistance.

[0057] As the piezoelectric acceleration sensor, for example, a MEMS (Micro Electro Mechanical Systems) type piezoelectric acceleration sensor is used. By employing a MEMS type piezoelectric acceleration sensor, it is possible to reduce the size and increase the intelligence of the sensor unit 2. Note that, although a MEMS type piezoelectric acceleration sensor is generally a three-axis acceleration sensor, it may also be a one-axis acceleration sensor.

[0058] The built-in clock 25 is an integrated circuit (IC) that implements a timekeeping function in the sensor unit 2, and is also called a real-time clock (RTC) or a CMOS clock. The time kept by the built-in clock 25 is correlated with the electrical signal of the vibration acquired by the measurement unit 23 by the control process in the processing unit 21, and is incorporated into the vibration data.

[0059] The storage unit 27 is an auxiliary storage device such as a hard disk, and at least temporarily stores vibration data in which the electrical signal of vibration acquired by the measurement unit 23 is associated with the time kept by the built-in clock 25. The storage unit 27 can be omitted when the vibration data is transmitted from the sensor unit 2 to the time synchronization system online in real time.

[0060] The data transfer unit 29 is an interface for transferring the vibration data generated by the control process in the processing unit 21 to the time synchronization system 1. The data transfer unit 29 may be a wireless communication port for Wi-Fi communication, NFC communication, WiMAX (registered trademark), or the like, or may be a wired communication port for Ethernet (wired LAN) communication or the like. The data transfer unit 29 may also be a connection port that is directly connected to the time synchronization system 1 and transmits the vibration data directly to the time synchronization system 1. The connection port that transmits the vibration data directly to the time synchronization system 1 may be, but is not limited to, a USB connection port for USB connection with the time synchronization system 1, or a wired connection port such as an IEEE1394 port for IEEE1394 connection with the time synchronization system 1. The connection port that transmits the vibration data directly to the time synchronization system 1 may be a wireless connection port for Bluetooth (registered trademark) communication or infrared communication. Furthermore, the data transfer unit 29 may be a removable media writing unit for transferring the vibration data acquired by the sensor unit 2 to the time synchronization system 1 offline via a removable medium such as a USB memory or an SD card.

[0061] In FIG. 2, the dashed arrow indicates that vibration data is transferred from the data transfer unit 29 of the sensor unit 2 to the data transfer unit 15 of the time synchronization system 1, but it is of course also possible to transfer various information in the opposite direction.

[0062] Next, a specific flow of the calculation process in this embodiment will be described with reference to Fig. 3 and Fig. 4. Note that Fig. 3 describes the results of a simulation based on vibration data of an actual earthquake. The graphs relating to time lag in Fig. 3 and subsequent figures are intentionally created with respect to the vibration data of an actual earthquake for the purpose of performing a simulation. Therefore, it should be noted that the graphs relating to time lag in Fig. 3 and subsequent figures are not the time lag in the vibration data of an actual earthquake.

[0063] 4, in the arithmetic processing in this embodiment, in step S1, first vibration data and second vibration data acquired at separate positions in a structure 5 (see FIG. 1) where vibration occurs are acquired for arithmetic processing. Step S1 is executed, for example, by a data reading unit 11a1 of a calculation unit 11 in a time synchronization system 1 (see FIG. 2).

[0064] In addition, although not limited to this embodiment, when performing time synchronization processing of vibration data, the number of data pieces of the first vibration data and the number of data pieces of the second vibration data must be the same. The number of data pieces of the first vibration data and the second vibration data is not limited, but may be made the same by acquiring data for 10 seconds at 0.01 second intervals. In addition, the number of data pieces of the first vibration data and the second vibration data may be made the same by adding 0 to the end of data with a short data length among the first vibration data and the second vibration data.

[0065] In step S1, the first vibration data and the second vibration data can be acquired from any position of the structure 5 (see FIG. 1) where vibration occurs, so that a time synchronization method using a process or method with a high degree of freedom is provided.

[0066] When the above-mentioned three-axis acceleration sensor is used as the measurement section 23 in the sensor unit 2 (see FIG. 2), vibration data is acquired along the X-axis, Y-axis, and Z-axis shown in FIG. 1. In this case, the vibration data used for time synchronization only needs to determine a time shift for the vibration data along any one of the X-axis, Y-axis, and Z-axis, and the vibration data along the other two axes can be time-synchronized based on the determined time shift value.

[0067] Next, in step S2, a plurality of estimated transfer functions used to estimate a relationship between the first vibration data and the second vibration data are calculated based on the first vibration data and the second vibration data acquired in step S1. Step S2 is executed, for example, in the transfer function calculation unit 11a2 of the calculation unit 11 in the time synchronization system 1 (see FIG. 2).

[0068] In step S2, first, a transfer function H(f) representing the relationship between the first vibration data and the second vibration data without taking the time lag into consideration is calculated from the following formula (1).

[0069]

number

[0070] In equation (1), f is a variable that represents the frequency (Hz), and F X (f) is a Fourier transform value of one of the first vibration data and the second vibration data, and F Y (f) is a Fourier transform value of one of the first vibration data and the second vibration data. Note that the frequency is synonymous with the frequency, and may be referred to as the frequency.

[0071] The transfer function H(f) for vibrations occurring in the structure 5 (see Fig. 1) is expressed as a transfer function of a second-order lag system. Therefore, as shown in Fig. 3(b), when there is no time lag between the first vibration data and the second vibration data, it can be seen that the phase is concentrated at two values, 0 degrees and 180 degrees. In Fig. 3(b), it can be seen that the phase is concentrated at two values, 0 degrees and 180 degrees, especially in the low frequency band (for example, a band of 10 Hz or less), except near low-order natural frequencies.

[0072] Furthermore, when a time lag of 1 second occurs between the first vibration data and the second vibration data, no change can be confirmed in the amplification of the transfer function H(f), as shown in Figures 3(a) and 3(c). On the other hand, it is confirmed that the variance of the phase of the transfer function H(f) becomes larger compared to Figure 3(b) (Figure 3(d)). Therefore, when a time lag occurs between the first vibration data and the second vibration data, it has almost no effect on the amplification of the transfer function H(f), but it has a large effect on the phase of the transfer function H(f).

[0073] The estimated transfer function H used to estimate the time delay between the first vibration data and the second vibration data, which reflects the phase characteristics and amplification characteristics due to the time delay as described above, L (f) can be expressed by the following equation (2).

[0074]

number

[0075] In equation (2), L is a variable that represents a tentative estimate of the time lag, and is expressed in units of seconds. Also in equation (2), e is Napier's constant, and i is an imaginary number.

[0076] In the right-hand side of equation (2), the part expressed by the power of Napier's constant e, which is multiplied by the transfer function H(f), has all values ​​on the unit circle in the complex plane, regardless of the value of the variable L, which is the tentative estimate of the time lag. Therefore, the absolute value of the part expressed by the power of Napier's constant e in equation (2) is 1, regardless of the value of the variable L. Therefore, the estimated transfer function H L The amplification of (f) is the same as the amplification of the transfer function H(f).

[0077] On the other hand, the estimated transfer function H L The phase of (f) will deviate from the transfer function H(f) by the tentative estimate of the time lag.

[0078] Therefore, the estimated transfer function H L If a tentative estimate of the time shift at which the phase values ​​in (f) are most concentrated in the two directions of 0 degrees and 180 degrees can be determined, the time shift between the first vibration data and the second vibration data can be identified.

[0079] As will be clear from the description of this embodiment, the "estimated transfer function H L "The phase values ​​in (f) are most concentrated in the two directions of 0 degrees and 180 degrees" means that, in a statistical sense, the estimated transfer function H L This means that the phase values ​​in the frequency domain of (f) are most concentrated in the two directions of 0 degrees and 180 degrees. Therefore, each estimated transfer function H L It should be noted that the number of phase values ​​that match the binary values ​​of 180 degrees or 0 degrees in (f) does not necessarily relate to whether or not the phase values ​​are "most concentrated in the binary directions of 0 degrees and 180 degrees." In addition, each estimated transfer function H L Please note that even if the phase values ​​in (f) are most concentrated at a binary value other than 180 degrees or 0 degrees, if they are most concentrated at the binary values ​​of 180 degrees or 0 degrees, this falls under the category of "most concentrated in the directions of the binary values ​​of 0 degrees and 180 degrees."

[0080] In step S2, a variable L, which is a tentative estimate of the time lag, is changed based on the above description to obtain a plurality of estimated transfer functions HL In step S2, the estimated transfer function H for each tentative estimated value of the time lag is calculated by changing the variable L at a sampling interval of 0.01 seconds in the range from -5 seconds to +5 seconds, for example, but not limited to, L (f) can be calculated sequentially. L The calculation of (f) can be performed quickly and easily by, for example, loop processing.

[0081] In step S3, the multiple estimated transfer functions H L The phase variance values ​​of each of the signals (f) are calculated. Step S3 is executed by the phase variance value calculation unit 11a3 of the calculation unit 11 in the time synchronization system 1 (see FIG. 2), for example.

[0082] In step S3, the estimated transfer function H L A function showing the relationship between frequency and phase in (f) is calculated.

[0083]

number

[0084] The calculation of the phase variance value in step S3 is, for example, calculated by a mean resultant length. Based on the idea of ​​the mean resultant length, the estimated transfer function H L The phase in (f) is considered to be a vector on a unit circle in the complex plane, and the average direction of the phase data is considered to be the direction on the complex plane of the average vector calculated by the vector average of each vector on the unit circle in the complex plane. The closer the magnitude of the calculated average vector is to 1, the more concentrated the phase data is considered to be in the direction of the average vector. In summary statistics, the magnitude of the calculated average vector minus 1 is considered to be the variance of the phase data.

[0085] In this embodiment, the estimated transfer function H LThe tentative time shift estimate in (f) where the phase values ​​are most concentrated in the directions of 0 and 180 degrees is determined to be the true time shift estimate. On the other hand, on the complex plane, 0 and 180 degrees are phase vectors in opposite directions on the real axis, so standardization is required to treat 0 and 180 degrees as equivalent.

[0086] In the above formula (3), when θ(f) is doubled, the binary values ​​of 0 degrees and 180 degrees become binary values ​​of 0 degrees and 360 degrees, and on the complex plane, the binary values ​​of 0 degrees and 180 degrees are standardized to vectors in the same direction. Therefore, the estimated transfer function H having the provisional estimated value of the time lag that maximizes the value of the real part of the average vector calculated by standardizing θ(f) is L (f) is the estimated transfer function H having the true estimate of the time shift that is most concentrated in the two directions of phase values ​​of 0 degrees and 180 degrees. L In other words, the estimated transfer function H with the provisional estimated value of the time lag that has the smallest value when the real part of the mean vector calculated by standardizing θ(f) is subtracted from 1 is L (f) is the estimated transfer function H that has the true estimated value of the time shift with the smallest variance among the two phase values ​​of 0 degrees and 180 degrees. L The result is (f).

[0087] Therefore, the calculation of the phase dispersion value in step S3 is performed, for example, by the following formula (4).

[0088]

number

[0089] In equation (4), V is the phase variance value, and Re(A) is the real part of the average vector A calculated by standardizing θ(f). The average vector A calculated by standardizing θ(f) in equation (4) is calculated by the following equation (5). Note that the reason that the real part Re(A) of the average vector A is calculated in equation (4) is that the phase values ​​of 0 degrees and 180 degrees are values ​​on the real axis, and the magnitude of the average vector A on the real axis indicates the degree of concentration on the two values ​​of 0 degrees and 180 degrees.

[0090]

number

[0091] In equation (5), N is the number of phase data of the first vibration data and the second vibration data.

[0092] In step S4, a first optimum value that is a true estimated value for time synchronization is determined based on the value of the real part Re(A) of the average vector A calculated in step S3. Step S4 is executed, for example, by the time offset calculation unit 11a4 of the calculation unit 11 in the time synchronization system 1 (see FIG. 2).

[0093] In FIG. 5, the tentative estimate of the time lag and the estimated transfer function H L FIG. 6 shows the relationship between the phase variance value of (f) and the estimated transfer function H L FIG. 7 shows the relationship between the phase variance value of (f) and the estimated transfer function H L The relationship between the phase dispersion value V and the phase shift coefficient V is shown in (f).

[0094] In all cases of Figures 5 to 7, the estimated transfer function H L The tentative estimated value of the time shift when the phase variance value of (f) is the minimum coincides with the value of the time shift set for the simulation. Therefore, in step S4, the estimated transfer function H L The tentative estimate of the time offset that minimizes the phase variance value in (f) can be determined as the first optimum value, which is the true estimate for time synchronization.

[0095] In step S4, the estimated transfer function H LThe tentative estimate of the time lag at which the real part Re(A) of the average vector A in (f) is maximized may be determined as the first optimum value, which is the true estimate. L The tentative estimated value of the time shift for which the phase values ​​in (f) are most concentrated in the two directions of 0 degrees and 180 degrees can be determined as the first optimum value.

[0096] In step S5, the first vibration data and the second vibration data are synchronized in time based on the first optimal value determined in step S4. Step S5 is executed, for example, by the time lag correction unit 11a5 of the calculation unit 11 in the time synchronization system 1 (see FIG. 2).

[0097] In addition, in step S5, the operator can manually synchronize the time between the first vibration data and the second vibration data by knowing the first optimum value. When the time synchronization is performed manually, the calculation process in step S5 can be omitted.

[0098] As described above, the time synchronization system 1 according to the present embodiment is a time synchronization system 1 that uses the first vibration data and the second vibration data acquired at separate positions of the structure 5 where vibration occurs to synchronize the time between the first vibration data and the second vibration data, and includes a plurality of estimated transfer functions H L (f) is calculated sequentially, and the estimated transfer function H L The calculation unit 11 determines the tentative estimated value of the time shift in which the phase value in (f) is most concentrated in the directions of two values ​​of 0 degrees and 180 degrees as a first optimum value for time synchronization.

[0099] A time synchronization program according to the present embodiment is a time synchronization program for synchronizing the time between the first vibration data and the second vibration data by using first vibration data and second vibration data acquired at separate positions of a structure where vibration occurs, the time synchronization program including a plurality of estimated transfer functions H L (f) is calculated, and the estimated transfer function H L The tentative estimated time offset value for which the phase values ​​in (f) are most concentrated in the directions of 0 degrees and 180 degrees is determined as the first optimum value for time synchronization.

[0100] Furthermore, a time synchronization method according to this embodiment is a time synchronization method for synchronizing time between first vibration data and second vibration data by using first vibration data and second vibration data acquired at separate positions of a structure in which vibration occurs, and includes the steps of: sequentially calculating a plurality of estimated transfer functions used for estimating a time shift between the first vibration data and the second vibration data, each of the estimated transfer functions including one of a plurality of tentative estimates of the time shift; and determining, as a first optimal value for synchronizing time, a tentative estimate of the time shift among the plurality of tentative estimates of the time shift, the tentative estimate of the time shift in which the phase values ​​in the estimated transfer function are most concentrated in the two directions of 0 degrees and 180 degrees.

[0101] According to the time synchronization system, the time synchronization program, and the time synchronization method of the present embodiment, it is possible to perform time synchronization using vibration data of any two points of the structure 5 without determining reference vibration data. Furthermore, according to the present embodiment, it is possible to perform time synchronization between vibration data without being restricted by a known vibration model of the structure 5. Therefore, it is possible to achieve highly accurate time synchronization between vibration data with a high degree of freedom.

[0102] Furthermore, according to the time synchronization system, the time synchronization program, and the time synchronization method of the present embodiment, the first vibration data and the second vibration data can be synchronized in time between them by using the entire acquired first vibration data and the entire acquired second vibration data. Therefore, according to the present embodiment, the time between the vibration data can be automatically synchronized without using a specific part of the vibration data such as the rising part of the waveform, a frequency band below the low-order natural frequency, or a resonance frequency, and the time between the vibration data can be easily synchronized with high accuracy.

[0103] Next, a modified example of the time synchronization system 1 according to the above embodiment will be described with reference to Figures 8 to 13. In the following description, the description of the same configuration as in the above embodiment will be omitted.

[0104] 8, in this modification, the calculation unit 11 of the time synchronization system 1 includes a second time synchronization unit 11b. The second time synchronization unit 11b includes a coherence calculation unit 11b1, a phase dispersion value calculation unit 11b2, a time shift calculation unit 11b3, and a time shift correction unit 11b4.

[0105] In this modified example, it is assumed that the time synchronization between the first vibration data and the second vibration data is automatically performed based on the first optimal value in the time lag correction unit 11a5 (see FIG. 2) of the first time synchronization unit 11a.

[0106] The coherence calculation unit 11b1 calculates a coherence value between the first vibration data after time synchronization and the second vibration data after time synchronization. The calculated coherence value may be at least temporarily stored in a storage unit (not shown) such as an internal memory or an auxiliary storage device such as a hard disk provided in the time synchronization system 1 by a control process in a control unit (not shown).

[0107] The phase variance calculation unit 11b2 calculates a phase variance value for each of a plurality of estimated transfer functions in which the phase value is weighted by using the coherence value calculated by the coherence calculation unit 11b1. The phase variance value is calculated for each tentative estimate of the time shift between the first vibration data and the second vibration data. The plurality of phase variance values ​​calculated for each tentative estimate of the time shift are at least temporarily stored in a storage unit (not shown) such as an internal memory or an auxiliary storage device such as a hard disk provided in the time synchronization system 1 by a control process in a control unit (not shown).

[0108] The time offset calculation unit 11b3 determines a provisional estimated value of the time offset in which the phase values ​​in the multiple estimated transfer functions calculated by the phase variance calculation unit 11b2 are most concentrated in the binary directions of 0 degree and 180 degree as a second optimum value for time synchronization. The other configurations of the time offset calculation unit 11b3 are the same as those of the time offset calculation unit 11a4 of the first time synchronization unit 11a according to the above-mentioned embodiment.

[0109] The time lag correction unit 11b4 synchronizes the time between the first vibration data and the second vibration data based on the second optimal value determined by the time lag calculation unit 11b3. The other configurations of the time lag correction unit 11b4 are the same as those of the time lag correction unit 11a5 of the first time synchronization unit 11a according to the above-mentioned embodiment.

[0110] Next, a specific flow of the calculation process in this modified example will be described with reference to Figures 9 and 10. Note that since steps S1 to S4 in Figure 10 are the same as those in the above-described embodiment, their description will be omitted.

[0111] In the calculation process of this modification, it is assumed that the time synchronization between the first vibration data and the second vibration data is automatically performed based on the first optimum value in step S5. According to the above embodiment, it is possible to highly accurately suppress the error between the actual time difference between the first vibration data and the second vibration data and the first optimum value.

[0112] However, when the coherence between the first vibration data and the second vibration data is not good, that is, when the phase distribution in the estimated transfer function has a relatively small frequency band in which the phase values ​​of the estimated transfer function are concentrated at two values ​​of 0 degrees and 180 degrees, there is a rare possibility that an error of about ±0.1 seconds will occur in the first optimal value, which is the true estimate of the time shift, even when time synchronization is achieved.

[0113] On the other hand, as shown in Fig. 9(c), the coherence value when there is no time lag between the first vibration data and the second vibration data is almost the same as the coherence value when the time lag between the first vibration data and the second vibration data is about 0.1 seconds. That is, even if there is an error of about ±0.1 seconds in the estimated value of the time lag, it is possible to calculate a coherence that is almost the same as when there is no time lag.

[0114] Moreover, in the two-value frequency bands where the phase of the transfer function is 0 degrees and 180 degrees, the coherence value is relatively large. Therefore, if the coherence value between the first vibration data after time synchronization and the second vibration data after time synchronization is calculated and the phase value of the estimated transfer function is weighted by the calculated coherence value, the influence of the resonance point near 3 (Hz) as shown in FIG. 9(b) on the estimation of the time shift can be reduced. Furthermore, if the phase value of the estimated transfer function is weighted by the calculated coherence value, the influence of the frequency band of 13 (Hz) or more, in which the phase variance is relatively large, on the estimation of the time shift can be reduced among all the frequency bands of the estimated transfer function as shown in FIG. 9(b). Therefore, if the phase value of the estimated transfer function is weighted by the calculated coherence value, it becomes possible to perform time synchronization with even higher accuracy.

[0115] In step S6, the Fourier transform value F' of the first vibration data after the time synchronization is performed X (f) and the Fourier transform value F' of the second vibration data after time synchronization. Y (f) and the coherence value γ 2(f) is calculated. The coherence value γ 2 The calculation of (f) is as shown in the following formula (6): F' X F' is the complex conjugate of (f) X (f) * And, F' Y The square of the absolute value of the average value of the inner product with (f) in the frequency domain is F' x (f) and F' X (f) * The average value of the dot product with F' in the frequency domain Y (f) and F' Y F' is the complex conjugate of (f) Y (f) * The coherence value γ 2 (f) is 0≦γ 2 (f) is a value between 1 and 2. Note that F' X (f) and F' Y The average value of the inner product values ​​in the frequency domain of (f) may be calculated using, but is not limited to, a Parzen window of 0.5 (Hz).

[0116]

number

[0117] Step S6 is executed by the coherence calculation unit 11b1 of the calculation unit 11 in the time synchronization system 1, for example.

[0118] In step S7, the phase variance value is calculated using the standardized average vector A in the same manner as in step S3, but each phase vector used to calculate the average vector has a coherence value γ 2 (f). Therefore, the standardized mean vector A is weighted by the coherence value γ 2 The weighting of the phase value by (f) is reflected and the calculation is performed by the following formula (7). In order to further emphasize the relative difference of the coherence value for each frequency, the coherence value γ 2Weighting is done by the square of (f).

[0119]

number

[0120] Other calculation processes in step S7 are the same as those in step S3. Step S7 is executed in the phase dispersion value calculation unit 11b2 of the calculation unit 11 in the time synchronization system 1, for example.

[0121] In step S8, a second optimum value that is a true estimated value for time synchronization is determined based on the value of the real part Re(A) of the average vector A calculated in step S7. Step S8 is executed, for example, in the time offset calculation unit 11b3 of the calculation unit 11 in the time synchronization system 1. Other calculation processes in step S8 are the same as those in step S4.

[0122] In step S9, the time between the first vibration data and the second vibration data is synchronized based on the second optimal value determined in step S8. Step S9 is executed, for example, in the time lag correction unit 11b4 of the calculation unit 11 in the time synchronization system 1. Other calculation processes in step S8 are the same as those in step S4 of the above-mentioned embodiment.

[0123] As described above, in the time synchronization system 1 of this modified example, the calculation unit 11 synchronizes the time between the first vibration data and the second vibration data based on the first optimal value, and calculates a coherence value γ between the first vibration data after the time synchronization and the second vibration data after the time synchronization. 2 (f) is calculated and the coherence value γ 2 (f) is used to weight the phase values ​​in a plurality of estimated transfer functions, and the tentative time shift estimate value that is most concentrated in the binary directions of phase values ​​of 0 degree and 180 degree is determined as the second optimum value that is the true estimate value for time synchronization. The time synchronization program and the time synchronization method of this modified example also have the configuration corresponding to the above.

[0124] In FIG. 11(a) and FIG. 11(b), the time lag between the first vibration data and the second vibration data is 1 second, and the estimated transfer function H L In (f), the coherence value γ 2 (f) The tentative time lag estimate and the estimated transfer function H L 11(c) and 11(d), the time lag between the first vibration data and the second vibration data is 1 second, and the estimated transfer function H L In (f), the coherence value γ 2 (f) The tentative time lag estimate and the estimated transfer function H L The relationship between the phase dispersion value V and the phase shift coefficient V is shown in (f).

[0125] In FIG. 12(a) and FIG. 12(b), the time difference between the first vibration data and the second vibration data is 0.5 seconds, and the estimated transfer function H L In (f), the coherence value γ 2 (f) The tentative time lag estimate and the estimated transfer function H L 12(c) and 12(d), the time lag between the first vibration data and the second vibration data is 0.5 seconds, and the estimated transfer function H L In (f), the coherence value γ 2 (f) The tentative time lag estimate and the estimated transfer function H L The relationship between the phase dispersion value V and the phase shift coefficient V is shown in (f).

[0126] In FIG. 13(a) and FIG. 13(b), the time difference between the first vibration data and the second vibration data is 0.1 seconds, and the estimated transfer function H L In (f), the coherence value γ 2 (f) The tentative time lag estimate and the estimated transfer function H L13(c) and 13(d), the time lag between the first vibration data and the second vibration data is 0.1 seconds, and the estimated transfer function H L In (f), the coherence value γ 2 (f) The tentative estimate of the time lag and the estimated transfer function H L The relationship between the phase dispersion value V and the phase shift coefficient V is shown in (f).

[0127] In all of Figures 11 to 13, the tentative estimate of the time lag is calculated using the estimated transfer function H L The tentative estimate of the time lag when the phase variance value V in (f) is at a minimum coincides with the time lag value set for the simulation. On the other hand, in all of Figs. 11 to 13, the estimated transfer function H L In (f), the coherence value γ 2 Compared with the case where the phase value is not weighted by (f) (Figs. 11(a), 11(b), 12(a), 12(b), 13(a), and 13(b)), the estimated transfer function H L In (f), the coherence value γ 2 When the phase values ​​are weighted by (f) (Figures 11(c), 11(d), 12(c), 12(d), 13(c), and 13(d)), the phase variance value V is small.

[0128] Therefore, the estimated transfer function H L In (f), the coherence value γ 2 By weighting the phase values ​​according to (f), a true estimate of the time offset can be determined more accurately.

[0129] The weighting based on the coherence value may be weighted according to the amplitude of the transfer function.

[0130] Furthermore, the phase value at a frequency may be weighted according to the magnitude of the average vector within a frequency band of a certain width centered on the frequency to be weighted. [Explanation of symbols]

[0131] 1 time synchronization system, 2 sensor unit, 2a first sensor unit, 2b second sensor unit, 5 structure, 11 calculation unit, 11a first time synchronization unit, 11a1 data reading unit, 11a2 transfer function calculation unit, 11a3 phase variance value calculation unit, 11a4 time shift calculation unit, 11a5 time shift correction unit, 11b second time synchronization unit, 11b1 coherence calculation unit, 11b2 phase variance value calculation unit, 11b3 time shift calculation unit, 11b4 time shift correction unit, 13 data output unit, 15 data transfer unit, 21 processing unit, 23 measurement unit, 25 built-in clock, 27 memory unit, 29 data transfer unit.

Claims

1. A time synchronization system for synchronizing a time between a first vibration data and a second vibration data, the first vibration data and the second vibration data being acquired at separate positions of a structure where vibration occurs, the time synchronization system comprising: Sequentially calculating a plurality of estimated transfer functions used to estimate a time delay between the first vibration data and the second vibration data, each of the estimated transfer functions including one of a plurality of tentative estimates of the time delay; Among the plurality of tentative estimates of the time shift, a tentative estimate of the time shift in which the phase value in the estimated transfer function is most concentrated in two directions of 0 degrees and 180 degrees is determined as a first optimum value for synchronizing the time. Equipped with a calculation unit Time synchronization system.

2. The calculation unit is Synchronizing the time between the first vibration data and the second vibration data based on the first optimum value; Calculating a coherence value between the first vibration data after the time is synchronized and the second vibration data after the time is synchronized; using the coherence value to weight phase values ​​in a plurality of the estimated transfer functions; Among the plurality of tentative estimates of the time shift, the tentative estimates of the time shift in which the phase value is most concentrated in the directions of two values ​​of 0 degrees and 180 degrees are determined as a second optimum value for synchronizing the time. The time synchronization system according to claim 1 .

3. A time synchronization program for synchronizing a time between first vibration data and second vibration data, the first vibration data and the second vibration data being acquired at separate positions of a structure where vibration occurs, the time synchronization program comprising: Sequentially calculating a plurality of estimated transfer functions used to estimate a time delay between the first vibration data and the second vibration data, each of the estimated transfer functions including one of a plurality of tentative estimates of the time delay; Among the plurality of tentative estimates of the time shift, a tentative estimate of the time shift in which the phase value in the estimated transfer function is most concentrated in two directions of 0 degrees and 180 degrees is determined as a first optimum value for synchronizing the time. Time synchronization program.

4. A time synchronization method for synchronizing times between first vibration data and second vibration data, the first vibration data and the second vibration data being acquired at separate positions of a structure where vibration occurs, comprising: sequentially calculating a plurality of estimated transfer functions used to estimate a time delay between the first vibration data and the second vibration data, each of the estimated transfer functions including one of a plurality of tentative estimates of the time delay; determining, among the plurality of tentative estimates of the time shift, a tentative estimate of the time shift in which the phase value in the estimated transfer function is most concentrated in two directions of 0 degree and 180 degree, as a first optimum value for synchronizing the time; Includes Time synchronization method.

Images