JP7770646B2 - Detection device, detection method, and program - Google Patents

Description

The present invention relates to a detection device, a detection method, and a program.

Conventionally, techniques for detecting scouring of target structures such as bridge piers are known (see, for example, Patent Documents 1 and 2). In Patent Document 1, a search range for vibration frequencies that includes the natural frequency of the target pier is set, and the frequency at which the amplitude of the Fourier spectrum is maximum is considered to be the natural frequency. In Patent Document 2, sensors are installed at both ends of the top of the structure, ground vibrations are estimated based on the vibrations of the structure detected by the sensors, and the natural frequency is estimated based on the phase difference between the vibrations of the pier and the ground.
Patent Document 1: Japanese Patent No. 4698466 Patent Document 2: Japanese Patent Application Laid-Open No. 2017-166922

With the method in Patent Document 1, it can be difficult to identify natural frequencies with high accuracy, limiting its scope of application. Furthermore, with the method in Patent Document 2, there is a risk that the number of sensors required to be installed on the structure may increase.

A first aspect of the present invention provides a detection device that includes a data acquisition unit that acquires measurement data from a target structure, an index calculation unit that analyzes the vibration characteristics of the target structure based on the measurement data and calculates a diagnostic index related to the soundness of the target structure, a diagnosis unit that diagnoses the soundness of the target structure based on the diagnostic index, and an output unit that outputs the soundness diagnosis results.

The output unit may output at least one of measurement data, diagnostic indicators, or diagnostic results.

The detection device may be equipped with a memory unit that stores at least one of the measurement data, diagnostic indicators, or diagnostic results.

Diagnostic indicators may include natural frequencies and statistical distributions of natural frequencies based on measurement data of the target structure.

The diagnostic index may include a natural frequency based on measurement data of the target structure and an alarm issuance probability index value based on a probability density function identified from the statistical distribution of the natural frequency.

The index calculation unit may calculate the statistical distribution of the natural frequencies using fast Bayesian FFT.

The data acquisition unit may acquire free damped vibration data after a predetermined vibration occurs in the target structure.

The data acquisition unit may acquire constant microtremor vibration data when no predetermined vibrations are occurring in the target structure.

The output section may display the diagnostic index and the statistical distribution of the diagnostic index.

The output section may display the diagnostic index, the statistical distribution of the diagnostic index, and the probability density function identified from the statistical distribution.

The detection device may be equipped with an information acquisition unit that acquires information for analyzing the vibration characteristics of the target structure.

The data acquisition unit may acquire time-series measurement data from the target structure. The index calculation unit may analyze the vibration characteristics of the target structure based on the time-series measurement data and calculate diagnostic indices related to the soundness of the target structure in time series. The diagnosis unit may diagnose the soundness of the target structure based on the diagnostic indices calculated in time series. The output unit may output the soundness diagnosis results in time series.

The output unit may further display structural environment information indicating the environmental status of the target structure.

A second aspect of the present invention provides a detection method for detecting the soundness of a target structure using a detection device, the detection method comprising the steps of acquiring measurement data from the target structure, analyzing the vibration characteristics of the target structure based on the measurement data and calculating a diagnostic index related to the soundness of the target structure, diagnosing the soundness of the target structure based on the diagnostic index, and outputting the diagnostic results of the soundness.

A third aspect of the present invention provides a program for causing a computer to execute the detection method according to the second aspect of the present invention.

Please note that the above summary of the invention does not list all of the features of the present invention. Subcombinations of these features may also constitute inventions.

1 shows an example of the configuration of a detection device 100. An example of the statistical distribution of the natural frequency estimation results is shown. An example of the statistical distribution of the natural frequency estimation results is shown. 1 shows an example of an estimated distribution of natural frequencies. The time history of the natural frequency and the histogram corresponding to the natural frequency are shown. 10 is a graph showing the estimation results of the natural frequency, the modal external force, the S/N ratio, and the time history fluctuation of the water level. 10 shows an example of an operational flowchart of a detection method using the detection device 100. An example of the estimation process in step S110 in FIG. 6A is shown. An example of the diagnostic process in step S120 in FIG. 6A is shown. An example of the statistical distribution of the natural frequency estimation results under normal conditions and the identified probability density function is shown below. An example of the statistical distribution of the natural frequency estimation results during flooding and the identified probability density function is shown below. 10 shows an example of a time history graph of the warning issuance probability index value and water level. 22 illustrates an example computer 2200 in which aspects of the present invention may be embodied in whole or in part.

The present invention will be explained below through embodiments of the invention, but the following embodiments do not limit the scope of the invention as claimed. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

Figure 1 shows an example of the configuration of a detection device 100. The detection device 100 includes a data acquisition unit 10, an information acquisition unit 20, an index calculation unit 30, a diagnosis unit 40, an output unit 50, and a memory unit 60.

The data acquisition unit 10 acquires measurement data from the target structure 150. In this example, the data acquisition unit 10 acquires measurement data of the target structure 150 from sensors 110 installed in the target structure 150. By acquiring measurement data of the target structure 150, the vibration characteristics of the target structure 150 can be acquired. For example, the measurement data is acceleration data acquired from an acceleration sensor. Furthermore, the measurement data is not limited to acceleration data, and may be any data such as velocity data or displacement data.

The sensor 110 is installed on the target structure 150. The detection device 100 may install one or more sensors 110 on the target structure 150. The sensor 110 transmits the measured data to the detection device 100 via a network or the like. In this example, the sensor 110 is attached to the top 152 of the target structure 150. The top 152 is the highest part of the target structure 150. The sensor 110 may also be installed on other parts of the target structure 150, such as the side.

The target structure 150 is a pier of a railway or road bridge. The detection device 100 analyzes measurement data from the target structure 150 to diagnose the soundness of the target structure 150. The soundness of the target structure 150 may be the possibility of scour, which indicates whether scouring occurring in the target structure 150 has reached a dangerous level that could lead to collapse or other problems. Scour refers to the erosion of soil around the target structure 150 by rain, wind, or waves. Note that even when the scour possibility of the target structure 150 is described in this specification, it is also applicable to the evaluation of structural soundness other than scouring, such as damage to concrete structures or buckling of steel frames.

The information acquisition unit 20 acquires information for analyzing the vibration characteristics of the target structure 150. The information acquisition unit 20 may also acquire information necessary for diagnosing the soundness of the target structure 150. For example, the information acquisition unit 20 acquires information such as parameters used by the index calculation unit 30 or the diagnosis unit 40. The information acquisition unit 20 may also acquire information entered via input means such as a keyboard or mouse. The information acquisition unit 20 may be able to communicate with the outside world via communication means such as a wired LAN or wireless LAN.

The index calculation unit 30 analyzes the vibration characteristics of the target structure 150 based on the measurement data and calculates a diagnostic index related to the health of the target structure 150. In one example, the index calculation unit 30 estimates the vibration characteristics of the target structure 150 based on the measurement data. The vibration characteristics may be at least one of the natural frequency, damping constant, or vibration mode of the target structure 150. For example, the index calculation unit 30 may calculate the statistical distribution of the natural frequency using the fast Bayes FFT described below. The fast Bayes FFT is an example of a Bayesian operating mode identification method based on Bayesian estimation.

In addition, the index calculation unit 30 may calculate diagnostic indices related to the soundness of the target structure 150 based on the measurement data acquired by the data acquisition unit 10 and various information acquired by the information acquisition unit 20.

The diagnostic index is an index for diagnosing the soundness based on vibration characteristics. For example, the diagnostic index may be a comparison between the natural frequency of the target structure 150 and a threshold value based on past statistical data. The diagnostic index may include a statistical distribution of vibration characteristics. The diagnostic index may also include an alarm issuance probability index value based on a probability density function identified from the statistical distribution of the vibration characteristics of the target structure. The alarm issuance probability index value will be described later. The diagnostic index is not limited to this, as long as it can be used by the diagnosing unit 40 to diagnose the soundness of the target structure 150.

The diagnosis unit 40 diagnoses the soundness of the target structure 150 based on the diagnostic index. For example, the diagnosis unit 40 diagnoses the soundness by comparing the diagnostic index with a predetermined diagnostic threshold. The diagnosis unit 40 transmits the soundness diagnosis results to the output unit 50. Specific methods for diagnosing soundness will be described later.

The output unit 50 outputs the health diagnosis results. The output unit 50 may include a display, a transmission device, a printer, etc. The output unit 50 outputs at least one of the measurement data acquired by the data acquisition unit 10, various information acquired by the information acquisition unit 20, the vibration characteristics of the target structure 150, the diagnostic index calculated by the index calculation unit 30, or the diagnostic results obtained by the diagnosis unit 40. The output method by the output unit 50 may be display, transmission, printing, or other methods.

The output unit 50 displays the diagnostic index and the statistical distribution of the diagnostic index. For example, the output unit 50 simultaneously displays the diagnostic index and the statistical distribution of the diagnostic index to the user. By displaying the diagnostic index and the statistical distribution of the diagnostic index, it is possible to diagnose the health of the system based on information about uncertainty.

The storage unit 60 stores the measurement data acquired by the data acquisition unit 10, various information acquired by the information acquisition unit 20, diagnostic indices calculated by the index calculation unit 30, or diagnostic results by the diagnostic unit 40. In this example, the storage unit 60 stores at least one of the measurement data, diagnostic indices, or health diagnostic results of the target structure 150. For example, the storage unit 60 includes a hard disk, BD, DVD, CD-R, MO, floppy disk, or memory.

The detection device 100 of this example uses the sensor 110 to detect scouring of the target structure 150, thereby preventing the collapse of the target structure 150 due to scouring. This prevents disasters caused by the collapse of the target structure 150 due to scouring of the ground around the bridge pier caused by rising river levels, etc.

The detection device 100 of this example estimates the vibration characteristics of the target structure 150 based on the measurement data acquired by the data acquisition unit 10. To estimate the vibration characteristics of the target structure 150, a system identification method, the stochastic subspace method, or a method based on Bayes' theorem may be applied. In Bayesian estimation, the parameters that make up a statistical model are evaluated as a probability distribution, so information regarding the uncertainty of the estimated value can be directly and quantitatively estimated from the observed value. The stochastic subspace method and the method based on Bayes' theorem will be described later.

The detection device 100 of this example does not require the implementation of risky impact vibration tests, such as actually impacting a bridge pier and measuring the vibration frequency. Furthermore, because the detection device 100 does not require irregular impact vibration tests, it can constantly monitor the target structure 150.

In one example, the data acquisition unit 10 acquires free damping vibration data after a predetermined vibration occurs in the target structure 150. For example, if the target structure 150 is a bridge, the predetermined vibration is vibration generated when a vehicle passes over the bridge of the target structure 150. By using the free damping vibration data after the vehicle passes over, the detection device 100 can perform vibration analysis with reduced noise effects.

The data acquisition unit 10 may also acquire constant microtremor vibration data when no predetermined vibrations are occurring in the target structure 150. For example, the data acquisition unit 10 acquires constant microtremor vibration data when no vehicles are passing over the bridge of the target structure 150. By using the constant microtremor data when no vehicles are passing over, the detection device 100 can understand the vibration characteristics when the structure is healthy.

The detection device 100 of this example can also be used when the river is flooded. When the river is flooded, the risk of scouring increases, but if vehicle traffic is prohibited during floods, it becomes impossible to estimate vibration characteristics using vibrations caused by passing vehicles. Therefore, in situations where vibration characteristics analysis using microtremor data is unavoidable, analysis using microtremor data is significantly affected by noise, and it may not be possible to estimate the natural frequency with high accuracy. Even in such cases, the detection device 100 of this example applies a method that can directly evaluate the uncertainty of the estimated value and also shows the statistical distribution of the estimated results, thereby improving reliability.

Therefore, the detection device 100 can diagnose the soundness of the target structure 150 even in an environment where ground vibrations are significantly mixed with surrounding noise. This allows the range of application of the detection device 100 to be expanded while maintaining the accuracy of estimating the natural frequency. Furthermore, the scour diagnosis method using the detection device 100 allows for soundness diagnosis even with a small number of sensors.

Figure 2 shows an example of the statistical distribution of the natural frequency estimation results. The vertical axis shows the natural frequency [Hz], and the horizontal axis shows the number of hits [items]. In this example, the statistical distribution of the natural frequency of the target structure 150 is displayed.

The index calculation unit 30 calculates the natural frequency and the statistical distribution of the natural frequency based on the measurement data of the target structure 150 as a diagnostic index. The detection device 100 can inform the user of the likelihood of the natural frequency by displaying the statistical distribution of the natural frequency in addition to the natural frequency. The index calculation unit 30 may also calculate an alarm issuance probability index value based on the statistical distribution of the natural frequency estimation results as a diagnostic index. By calculating the alarm issuance probability index value, the user can be informed of the probability that the diagnostic index will fall below a threshold and issue an alarm.

The detection device 100 can also display the reliability of the estimated value by displaying the statistical distribution of the estimation results over a certain period of time, such as every 10 minutes. In one example, the detection device 100 displays the average value and standard deviation of the natural frequency. In this example, the detection device 100 displays the average value and variance of the natural frequency. In this example, the average value of the natural frequency is 15.07, and the variance is 0.42.

Table 1 shows an example of diagnostic indices, categories, and judgment methods for diagnosing the health. The output unit 50 may also display information necessary for diagnosing the health, such as that shown in Table 1, to the user.

The detection device 100 can set an evaluation index for the natural frequency when the structure is healthy and make a judgment based on the distribution of estimated natural frequency values over a certain period. The natural frequency when the structure is healthy may be calculated using microtremor data when no vehicles are passing through. x is expressed by the following equation:
x = Estimated natural frequency / Natural frequency in healthy condition

In this example, the detection device 100 divides the condition into four categories depending on the evaluation index x, and diagnoses the condition's soundness. For example, if x≦0.70, it falls under category A1, and it is determined that an abnormality exists. If an abnormality exists, the target structure 150 is repaired or reinforced. If 0.70<x≦0.85, it falls under category A2, and it is determined that inspection is necessary. If 0.85<x≦1.00, it falls under category B, and it is determined that there is a low possibility of an abnormality occurring. And if 1.00<x, it falls under category S, and it is determined that the target structure 150 is sound.

Figure 3 shows an example of the statistical distribution of the natural frequency estimation results. In this example, the positions corresponding to the predetermined evaluation indexes are indicated by dashed lines.

The dashed line indicates the case where the evaluation index is x = 0.85 when the natural frequency in a healthy state is 15.9 Hz. Referring to Table 1, natural frequencies greater than the dashed line can be determined to be Category B or S, which indicate a low possibility of scour occurring. On the other hand, natural frequencies in the range below the dashed line can be determined to be Category A1 or A2, which indicate a high possibility of scour occurring. The detection device 100 can determine the possibility of scour occurring by displaying the statistical distribution of the natural frequencies along with the estimated natural frequencies. By indicating the range of the evaluation index, as in this example, the soundness of the structure can be immediately diagnosed based on the estimated natural frequencies.

Figure 4A shows an example of the estimated distribution of natural frequencies. The vertical axis shows the natural frequency [Hz], and the horizontal axis shows the date of acquisition of the data. The detection device 100 in this example displays the time-series changes in the natural frequency. In this example, the natural frequency is estimated using a Bayesian operating mode identification method using fast Bayesian FFT based on Bayesian estimation.

The Bayesian operational modal identification method using fast Bayesian FFT is an operational modal analysis method using frequency-domain microtremor data based on Bayesian theory. Bayesian theory evaluates the parameters that make up the statistical model as a probability distribution, so information about the uncertainty of the estimated values can be directly and quantitatively estimated from the observed values. Using the estimated results, it is possible to calculate the signal-to-noise ratio (S/N ratio) to evaluate the impact of observation errors for each modal response.

Consider the problem of estimating the governing parameter θ of the statistical model in which a set of appropriate observed data D occurs. In Bayesian estimation, the parameter θ is estimated using the probability density as a measure. A probability distribution of the parameter θ, called the prior distribution, is assumed based on information prior to the observation data D being obtained, and its probability density function is represented as p(θ). The probability distribution of the parameter θ estimated from the observation data D is called the posterior distribution, and is expressed as the conditional probability density function p(θ|D) based on the following Bayes' theorem.
Here, the conditional probability density function p(D|θ) on the right-hand side is called the likelihood function, and is a function that represents the likelihood of a model when observed data D is obtained in a statistical model configured with parameter θ. In Bayesian estimation, the observed data D is interpreted as a deterministic value that has already been observed, so the term in the denominator on the right-hand side of equation (1) can be ignored when estimating parameter θ.

In the Bayesian operational modal analysis method, the power spectral density (PSD) curve obtained by performing a fast Fourier transform (FFT) on the measured values of the acceleration, velocity, and displacement of a structure is used as the observed data D, and various parameters related to the vibration mode are used as the governing parameters θ of the statistical model. When a large number of observed values are obtained, such as FFT data of acceleration time series in long-term measurements, it is known that the likelihood function p(D|θ) dominates the posterior distribution p(θ|D), and the prior distribution term can be ignored. Assuming that a non-informative prior distribution is applied as the prior distribution, the probability density function of the posterior distribution is proportional to the likelihood function. In other words, the following proportional relationship holds:

Therefore, for data D observed in vibration measurements, the posterior distribution of θ can be estimated by formulating the likelihood function p(D|θ) as an appropriate function of the parameter θ related to the vibration mode. The θ that maximizes the posterior distribution p(θ|D) is called the Most Probable Value (MPV), and in cases such as equation (2) that assume an uninformative prior distribution, it coincides with the θ that maximizes the likelihood function p(D|θ). Here, θ in the posterior distribution p(θ|D) is composed of modal frequencies, modal damping, S/N ratio, etc., and the natural frequency is estimated by calculating the θ that maximizes the likelihood function p(D|θ). This point estimation method for calculating an estimate of θ that maximizes the likelihood function is known as maximum likelihood estimation, and many vibration mode identification methods, including stochastic subspace methods, are based on the least-squares method, which is one type of estimation problem. One of the advantages of introducing a Bayesian estimation approach to vibration mode identification is that by evaluating the posterior distribution of vibration characteristics, it is possible to estimate the degree of uncertainty in the most probable value (MPV) estimation results.

Next, a method for identifying vibration characteristics using the stochastic subspace method will be described. The target structure 150 can be expressed by a discretized state equation as follows:

Here,
indicates the observed value at point m of the structure, and C is a matrix that extracts the observed value y(k) from the state variable x(k),
The coefficient matrix, external force vector, and observation matrix are expressed as follows:

When performing system identification using vibration data from bridges and other structures, it is often difficult to observe the external forces that are the inputs to the system due to environmental effects such as river flow loads. Therefore, we apply the Operational Modal Analysis (OMA) method, which treats the inputs as unknown and uses only the observed values of the outputs. In this case, the external forces are considered to be a time series of disturbance noise that is specific to the system. Due to stationarity and linearity of the process, this disturbance noise can be considered as stationary white noise. Furthermore, the observation noise when observing the output from the system can be expressed in a similar way. By formulating the above, we obtain the following equation:
where w(k) and v(k) are the system noise and the observation noise, respectively, and both are white noise time series. That is, for any covariance matrix R, the following equation holds:

δ _pq represents the Kronecker delta, and E[ ] represents the mathematical mean. A similar process is performed for the observation noise v(k). The state variables in equation (4) can be given as follows using a non-stationary Kalman filter with one-period-ahead prediction based on the observation vector y(t):

However, here
is the estimate of the state variable x(k) by the non-stationary Kalman filter. K _k-1 is the non-stationary Kalman gain, which is updated each time a prediction is made based on information specific to the system. The mathematical mean of the error between this predicted value and the true value is minimized using a least-squares approach, making it possible to algebraically estimate the vector space of the state variable vector.

Consider the Oblique Matrix O _i defined by the following equation:

however,
Let Y p and Y t denote the Moore-Penrose pseudoinverse matrix, and Y _p and Y _t are block Hankel matrices that contain past and future information on the observations, respectively, and are defined as follows:

In addition, the system matrix is assumed to be observable, meaning that the governing parameters can be uniquely determined from the input and output data of the system. "Observability" means that the current state can be described from past responses. The observable matrix of the discretized state equation is expressed as follows:

When rank(T _p ), the system becomes observable. Here, we assume that m×p=2n holds, where m is the degree of freedom of the output vector y(t) and n is the degree of freedom of the dynamic system. In other words, p indicates the relationship between 2n state variables and m observation points, and represents the correspondence between 2n state variables and m observation points with a time delay of p points. Therefore, the state equation for 2n variables can be expressed as a p-th-delayed differential equation for m observation variables.

When j in Y _p and Y _f is sufficiently large, it can be expressed as follows using the observable matrix T _p defined by equation (10):

is the vector of estimated values of the state variables by the Kalman filter
In other words, the row space of the matrix O _i is the matrix
If we perform singular value decomposition on this matrix _Oi as shown in the following equation, we obtain the following equation.

Here, _W1 and _W2 are appropriate weight matrices. Both of these weight matrices can be set as unit matrices of appropriate size. In singular value decomposition, matrices U and V are unitary matrices, and the singular values of matrix S are arranged in the diagonal elements in descending order of absolute value. In other words, the basis vectors corresponding to the row and column spaces that make up the matrix are arranged in descending order of their contribution to each element of the matrix. By utilizing this property, the original matrix can be effectively approximated by a matrix of lower rank.

The time series of the estimated values of the state variables based on the singular value decomposition of Equation (12)
and its one-stage ahead time series
can be calculated, and the coefficient matrix of the state space model expressed by Equation 4 is
is identified by the following equation:
However, the following equation is satisfied for Y _i .

The detection device 100 may diagnose the health of the target structure 150 based on the time-series changes in the natural frequency. The detection device 100 may also simultaneously display the range of the health diagnostic index. The detection device 100 may also display the average value or standard deviation of the natural frequency over a predetermined period.

Figure 4B shows the time history of the natural frequency and a histogram corresponding to the natural frequency. The vertical axis represents the estimated natural frequency value [Hz], and the horizontal axis represents time. In this example, the time history of the estimated natural frequency value represents the average value over a predetermined period. The detection device 100 in this example shows time history fluctuations over a 12-month period, but this is not limited to this.

For example, the natural frequency when the structure is healthy is 9.0 Hz, and the diagnostic threshold is 7.65 Hz, which corresponds to Category B. The detection device 100 in this example simultaneously displays the natural frequency when the structure is healthy and the diagnostic threshold, making it possible to diagnose the structure's health and evaluate the reliability of the estimated natural frequency.

Figure 5 is a graph showing the estimated natural frequency, the PSD of the modal external force, the S/N ratio, and the time history variation of the water level. In this example, the detection device 100 shows the time history variation over 24 hours, but this is not limited to this. The scour detection device 100 estimates the PSD of the modal external force and the time series data observation error. Using these estimation results, the scour detection device 100 can calculate the signal-to-noise ratio (S/N ratio) to evaluate the impact of the observation error for each modal response.

Referring to Figure 5, as the water level rises, the PSD and S/N ratio of the modal external forces increase, and the variance in the estimated values of the natural frequencies decreases. This indicates that signals corresponding to the modal responses of the target structure are prominently present in the measured acceleration time series. In other words, the reliability of the target frequency identification results can be evaluated as high. In this way, the detection device 100 can evaluate the reliability of the estimates using the variance in the estimated values of the natural frequencies, the PSD of the modal external forces, and the magnitude of the S/N ratio.

The detection device 100 may also filter data using the calculated diagnostic index. For example, the detection device 100 filters the natural frequency using at least one of the PSD and S/N ratio of the modal external force. This allows the reliability of the estimated value to be evaluated based on more reliable data.

In this example, the data acquisition unit 10 acquires time-series measurement data from the target structure 150. The index calculation unit 30 analyzes the vibration characteristics of the target structure 150 based on the time-series measurement data and calculates diagnostic indices related to the soundness of the target structure 150 in time series. The diagnosis unit 40 diagnoses the soundness of the target structure 150 based on the diagnostic indices calculated in time series. The output unit 50 then outputs the soundness diagnosis results in time series.

In this way, the detection device 100 of this example can output time-series diagnostic indicators based on time-series measurement data. In other words, the detection device 100 of this example can acquire vibration data without having to apply external shocks each time a diagnosis is performed.

As described above, the detection device 100 can diagnose the soundness of the target structure 150 regardless of whether flooding is occurring or whether vehicles are passing through. Therefore, the range of application of the detection device 100 can be expanded while maintaining the accuracy of estimating the natural frequency. Furthermore, the scour diagnosis method using the detection device 100 allows for the use of a small number of sensors.

Figure 6A shows an example of an operational flowchart of a detection method using the detection device 100. In step S100, it is determined whether a predetermined vibration has occurred. If the predetermined vibration has not occurred, data from the sensor 110 is acquired as continuous micro-tremor vibration data (S102). On the other hand, if the predetermined vibration has occurred, it is determined whether the predetermined vibration has subsided (S104), and if it has subsided, data from the sensor 110 is acquired as free damped vibration data (S106).

In step S110, an estimation process is performed to estimate the vibration characteristics of the target structure 150 based on the data acquired by the sensor 110. Specific examples of the estimation process will be described later. In step S120, a diagnosis process is performed to diagnose the soundness of the target structure 150. Specific examples of the diagnosis process will be described later. In step S130, the diagnosis results are output.

For example, the detection device 100 may output at least one diagnostic index, such as the evaluation index x, the PSD of the modal external force, or the S/N ratio, in addition to the natural frequency. Furthermore, the detection device 100 may output structural environment information indicating the environmental conditions of the target structure 150 as a diagnostic index. For example, the structural environment information may be additional information indicating the conditions of a river, such as water level or rainfall. Displaying the diagnostic index and its statistical distribution makes it possible to directly evaluate the uncertainty of the estimated value, thereby improving reliability.

Figure 6B shows an example of the estimation process in step S110 in Figure 6A. In step S112, the natural frequency is calculated from each piece of data. In step S114, the natural frequency is estimated using the stochastic subspace method or fast Bayesian FFT, and a statistical distribution is calculated.

Figure 6C shows an example of the diagnostic processing of step S120 in Figure 6A. In step S122, an evaluation index is calculated from the estimated natural frequency and the natural frequency in a healthy state. In step S124, the health of the target structure 150 is determined based on the calculated evaluation index.

(Explanation of warning issuance probability index value)
Here, a method for determining the possibility of scouring occurrence using an alarm issuance probability index value will be described. The detection device 100 sets an evaluation index for the natural frequency and calculates an alarm issuance probability index value for the natural frequency estimation results for a certain period. For example, the detection device 100 sets an evaluation index for the natural frequency when the structure is healthy and calculates an alarm issuance probability index value for the natural frequency estimation results for a certain period. Specifically, the detection device 100 derives a probability density function using the probability characteristics of the natural frequency obtained by fast Bayesian FFT when the structure is healthy (normal), and calculates an alarm issuance probability index value. Note that in this example, a case will be described in which the target structure 150 is a bridge pier, but the target structure 150 is not limited to this.

First, a probability density function is derived for the statistical distribution of the natural frequency (distribution of the MPV of the frequency) estimated during healthy (normal) conditions. For example, the parameter values of the probability density curve of the stable distribution are identified using the maximum likelihood estimation method. The characteristic function E(e ^itZ ) of the stable distribution is given by the following equation. The detection device 100 calculates E(e ^itZ ) from a sample of the most probable value of the natural frequency identified by finding θ that maximizes the likelihood function p(D|θ).
The probability density function of the stable distribution is expressed as S(α, β, γ, d; Z).

Table 2 shows the parameters used in the probability density function of a stable distribution. The first shape parameter, α, controls the thickness of the tail of the probability density function and is called the characteristic index. As the shape parameter α decreases, the tail of the probability density function becomes thicker. The second shape parameter, β, represents the skewness of the distribution. When β = 0, the distribution is symmetric. When β > 0, the probability density function has a long right tail. When β < 0, the probability density function has a long left tail. The scale parameter γ controls the spread of the probability density function and is called the magnitude index. The location parameter d represents the mean value of the probability density function. Z represents the random variable, and t represents the argument of the characteristic function.

Figure 6D shows the probability density function of the stable distribution identified from the statistical distribution of natural frequencies estimated under healthy (normal) conditions (distribution of MPV of frequencies).

The solid line shows the probability density function of the target frequency of the pier estimated under normal conditions. This probability density function is considered to be the distribution characteristic of the pier's natural frequency estimated under normal conditions, and after calculating the cumulative probability density function Ψ (α _N , β _N , γ _N , d _N , X), this is used as the "alert issuance probability index value" when the estimated pier's natural frequency falls to a certain value. This makes it possible to determine whether there is a risk of disaster due to scouring based on the calculated alert issuance probability index value.

For example, if the natural frequency in a healthy state is 9.39 Hz, the warning probability index value will be 0.0027 when the estimated natural frequency drops to 7.98 Hz. The warning probability index value when the average value of the estimated natural frequency of the bridge pier during flooding is X can be defined as follows:
Ψ _C = Ψ (α _N , β _N , γ _N , d _N , X)

Table 3 shows the relationship between the health index value and the warning issuance probability index value.

The threshold value for the warning probability index is set to, for example, 0.01, and if the warning probability index value falls to this value during times of flooding, etc., it is determined that there is a risk of disaster due to scouring. This threshold value corresponds to a value greater than the limit value (7.98 Hz, κ = 0.85) for judgment rank B of the soundness judgment category shown in Table 3. Therefore, it indicates that the threshold is set so that there is a high probability of a Type 1 error (judging that scouring has occurred despite the fact that the river is sound) and a low probability of a Type 2 error (judging that the river is sound despite scouring having occurred).

Figure 6E shows the probability density function curve identified from the statistical distribution of the natural frequencies estimated during flooding (the distribution of MPV of frequencies). As the distribution characteristics of the natural frequencies of the bridge piers differ in this way between normal times and flooding, it is also possible to calculate an alert issuance probability index value that takes into account the different distribution characteristics during flooding. If the probability density function during flooding is S _F (α _F , β _F , γ _F , d _F ; Z), the alert issuance probability index value that takes into account the different distribution characteristics during flooding can be calculated as follows:

Figure 6F is an example of a display of the time history of the alarm issuance probability index value and water level [m]. In this example, a threshold value (e.g., 0.01) for the alarm issuance probability index value is shown. In this example, the alarm issuance probability index value obtained from the estimated natural frequency never falls below the threshold value, resulting in the conclusion that the target bridge pier is sound. The detection device 100 in this example can visualize the possibility of an alarm being issued by graphing the alarm issuance probability index value.

In the above-described embodiment, the alarm issuance probability index value is calculated using the probability characteristics of the natural frequency calculated by fast Bayes FFT. However, the embodiment is not limited to this. The alarm issuance probability index value may also be calculated using the probability characteristics of the natural frequency calculated by a method other than fast Bayes FFT. For example, instead of fast Bayes FFT, a probability density function may be derived using the probability characteristics of the natural frequency calculated by a stochastic subspace method (SSI: Stochastic Subspace Identification), a multidimensional AR model (VAR), a frequency domain decomposition (FDD), an eigensystem realization algorithm (ERA), or a regular FFT, and the alarm issuance probability index value may be calculated. For example, when calculating the alarm issuance probability index value using the stochastic subspace method, the detection device 100 calculates parameters related to the mode (e.g., the mode shape and poles (the frequency and damping constant may be calculated from the poles)) using A and C in equation (13).

Figure 7 shows an example of a computer 2200 in which aspects of the present invention may be embodied, in whole or in part. Programs installed on the computer 2200 may cause the computer 2200 to function as or perform operations associated with an apparatus or one or more sections of the apparatus according to embodiments of the present invention, and/or to perform a process or steps of a process according to embodiments of the present invention. Such programs may be executed by the CPU 2212 to cause the computer 2200 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

The computer 2200 according to this embodiment includes a CPU 2212, RAM 2214, a graphics controller 2216, and a display device 2218, which are interconnected by a host controller 2210. The computer 2200 also includes input/output units such as a communications interface 2222, a hard disk drive 2224, a DVD-ROM drive 2226, and an IC card drive, which are connected to the host controller 2210 via an input/output controller 2220. The computer also includes legacy input/output units such as a ROM 2230 and a keyboard 2242, which are connected to the input/output controller 2220 via an input/output chip 2240.

The CPU 2212 operates according to programs stored in the ROM 2230 and RAM 2214, thereby controlling each unit. The graphics controller 2216 retrieves image data generated by the CPU 2212 into a frame buffer or the like provided in the RAM 2214 or into the graphics controller itself, and causes the image data to be displayed on the display device 2218.

The communication interface 2222 communicates with other electronic devices via a network. The hard disk drive 2224 stores programs and data used by the CPU 2212 in the computer 2200. The DVD-ROM drive 2226 reads programs or data from the DVD-ROM 2201 and provides the programs or data to the hard disk drive 2224 via the RAM 2214. The IC card drive reads programs and data from an IC card and/or writes programs and data to an IC card.

ROM 2230 stores boot programs and the like that are executed by computer 2200 upon activation, and/or programs that depend on the hardware of computer 2200. I/O chip 2240 may also connect various I/O units to I/O controller 2220 via parallel ports, serial ports, keyboard ports, mouse ports, etc.

The programs are provided on computer-readable media such as a DVD-ROM 2201 or an IC card. The programs are read from the computer-readable media, installed on the hard disk drive 2224, RAM 2214, or ROM 2230, which are also examples of computer-readable media, and executed by the CPU 2212. The information processing described in these programs is read by the computer 2200, and brings about cooperation between the programs and the various types of hardware resources described above. An apparatus or method may be configured by implementing information manipulation or processing in accordance with the use of the computer 2200.

For example, when communication is performed between computer 2200 and an external device, CPU 2212 may execute a communication program loaded into RAM 2214 and instruct communication interface 2222 to perform communication processing based on the processing described in the communication program. Under the control of CPU 2212, communication interface 2222 reads transmission data stored in a transmission buffer processing area provided in RAM 2214, hard disk drive 2224, DVD-ROM 2201, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes received data received from the network to a reception buffer processing area, etc., provided on the recording medium.

The CPU 2212 may also cause all or necessary portions of files or databases stored on external recording media such as the hard disk drive 2224, DVD-ROM drive 2226 (DVD-ROM 2201), IC card, etc. to be read into the RAM 2214, and perform various types of processing on the data on the RAM 2214. The CPU 2212 then writes the processed data back to the external recording media.

Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and may undergo information processing. The CPU 2212 may perform various types of processing on data read from RAM 2214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search/replacement, etc., as described throughout this disclosure and specified by the program's instruction sequence, and write the results back to RAM 2214. The CPU 2212 may also search for information in files, databases, etc. on the recording medium. For example, if multiple entries, each having an attribute value of a first attribute associated with an attribute value of a second attribute, are stored on the recording medium, the CPU 2212 may search for an entry that matches a condition specified by the attribute value of the first attribute from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

The programs or software modules described above may be stored on computer-readable media on or near the computer 2200. Recording media such as a hard disk or RAM provided within a server system connected to a dedicated communications network or the Internet can also be used as computer-readable media, thereby providing the programs to the computer 2200 via the network.

The present invention has been described above using embodiments, but the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be clear to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

The order of execution of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before" or "prior to," and it should be noted that processes can be performed in any order unless the output of a previous process is used in a later process. Even if the operational flow in the claims, specifications, and drawings is explained using terms such as "first," "next," etc. for convenience, this does not mean that the processes must be performed in that order.

10...Data acquisition unit, 20...Information acquisition unit, 30...Index calculation unit, 40...Diagnosis unit, 50...Output unit, 60...Memory unit, 100...Detection device, 110...Sensor, 150...Target structure, 152...Top, 2200...Computer, 2201...DVD-ROM, 2210...Host controller, 2212...CPU, 2214...RAM, 2216...Graphics controller, 2218...Display device, 2220...I/O controller, 2222...Communication interface, 2224...Hard disk drive, 2226...DVD-ROM drive, 2230...ROM, 2240...I/O chip, 2242...Keyboard

Claims (16)

a data acquisition unit that acquires measurement data from the target structure;
an index calculation unit that analyzes vibration characteristics of the target structure based on the measurement data and calculates a diagnostic index related to the soundness of the target structure;
a diagnostic unit that diagnoses the soundness of the target structure based on the diagnostic index;
an output unit that outputs the diagnosis result of the soundness,
the index calculation unit calculates an evaluation index of the soundness based on the estimated value of the natural frequency estimated based on the measurement data and the natural frequency in a sound state;
the output unit displays the range of the evaluation index for the category related to the soundness of the target structure superimposed on the statistical distribution of the estimation result of the natural frequency ,
the diagnostic index is an index for diagnosing the soundness of the target structure based on the vibration characteristics of the target structure, and includes the natural frequency and the statistical distribution;
A detection device in which the evaluation index is an index for evaluating the soundness of the target structure . The detection device according to claim 1 , wherein the output unit outputs at least one of the measurement data, the diagnostic index, and the diagnostic result. The detection device according to claim 1 or 2, further comprising a storage unit that stores at least one of the measurement data, the diagnostic index, and the diagnostic result. The detection device according to claim 1 , wherein the evaluation index is x that satisfies the following formula: x=estimated value of natural frequency/natural frequency in a healthy state. The detection device according to any one of claims 1 to 4, wherein the diagnostic index includes a natural frequency based on measurement data of the target structure and an alarm issuance probability index value based on a probability density function identified from a statistical distribution of the natural frequency. The detection device according to claim 1 , wherein the index calculation unit calculates the statistical distribution of the natural frequencies by a fast Bayesian FFT. The detection device according to claim 1 , wherein the data acquisition unit acquires free damped vibration data after a predetermined vibration occurs in the target structure. The detection device according to claim 1 , wherein the data acquisition unit acquires constant microtremor vibration data in a state where a predetermined vibration is not occurring in the target structure. The detection device according to claim 1 , wherein the output unit displays the diagnostic index and a statistical distribution of the diagnostic index. The detection device according to claim 1 , wherein the output unit displays the diagnostic index, a statistical distribution of the diagnostic index, and a probability density function identified from the statistical distribution. The detection device according to claim 1 , further comprising an information acquisition unit that acquires information for analyzing vibration characteristics of the target structure. The data acquisition unit acquires time-series measurement data from the target structure,
the index calculation unit analyzes vibration characteristics of the target structure based on the time-series measurement data and calculates diagnostic indices related to the soundness of the target structure in time series;
the diagnosis unit diagnoses the soundness of the target structure based on the diagnostic index calculated in time series;
The detection device according to claim 1 , wherein the output unit outputs the diagnosis results of the soundness in chronological order. The output unit further displays structural environment information indicating the environmental status of the target structure.
Detecting device according to any one of claims 1 to 12. A detection method for detecting the soundness of a target structure using a detection device, comprising:
acquiring measurement data from a target structure;
a step of analyzing vibration characteristics of the target structure based on the measurement data and calculating a diagnostic index related to the soundness of the target structure;
diagnosing the soundness of the target structure based on the diagnostic index;
and outputting the diagnosis result of the healthiness,
the step of calculating the diagnostic index includes a step of calculating an evaluation index of the health level based on the estimated value of the natural frequency estimated based on the measurement data and the natural frequency in a healthy state,
The step of outputting the diagnosis result of the soundness includes a step of displaying the range of the evaluation index for the category related to the soundness of the target structure by superimposing it on the statistical distribution of the estimation result of the natural frequency ,
the diagnostic index is an index for diagnosing the soundness of the target structure based on the vibration characteristics of the target structure, and includes the natural frequency and the statistical distribution;
A detection method in which the evaluation index is an index for evaluating the soundness of the target structure . The detection method according to claim 14 , wherein the evaluation index is x that satisfies the following formula: x=estimated value of natural frequency/natural frequency in healthy state. A program for causing a computer to execute the detection method described in claim 14 or 15.