JP2003156415A - Method of inspecting soundness of large-sized structure by adaptive parameter estimation method using physical model, and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method by which the natural frequency of a large-sized structure can be estimated by an impact vibration, and the soundness thereof is easily and precisely diagnosed. SOLUTION: In this soundness inspection method for the large-sized structure by an adaptive parameter estimation method using a physical model, a vibration model constituted to be set preliminarily is combined to a vibration data from a vibration sensor to estimate the natural frequency of the structure by a maximum likelihood method, and a result therein is displayed and stored.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a soundness inspection method for a large structure by a nondestructive inspection for estimating the natural vibration frequency of the large structure by impact vibration and diagnosing the soundness thereof, and an apparatus therefor. .

[0002]

2. Description of the Related Art In order to ensure the safe running of trains and cars,
It is essential to check the safety of large structures such as bridges spanning rivers on railways and roads. As a method of inspecting the soundness of such large structures by nondestructive inspection, conventionally, a shock was applied to the bridge pier head with a weight, the vibration waveform at that time was measured, and frequency analysis such as fast Fourier transform was performed. A method has been adopted in which the natural vibration frequency is obtained and the soundness of the pier is determined from the value.

[0003]

However, according to the above-mentioned method, about half of bridge bridge piers have not only poor frequency spectrum resolution but also a large number of peaks, and even if phase characteristics are used, the 180 deg phase suddenly changes. Since the natural vibration frequency cannot be obtained, for example, the frequency at which the peak takes place cannot be found. Therefore, it is practically difficult to determine the soundness of the structure. Moreover, even if a weight is used, a turret for suspending the weight must be temporarily installed for each pier, which is extremely troublesome.

Therefore, according to the present invention, by combining a simple and effortless striking method and an adaptive parameter estimation method using a physical model, the reliability of the diagnosis regarding the safety of the structure can be greatly improved. An object of the present invention is to solve the above problems by providing a method and an apparatus for checking the soundness of an object.

The present invention comprises a vibration sensor for observing the vibration of a large structure and a means for digitalizing the signal.
Comprised of arithmetic means for estimating the natural vibration frequency of a large structure by the maximum likelihood method by combining the digitalized vibration data with a vibration model that can be set in advance, and means for displaying or saving the result. By doing so, the natural vibration frequency of the large structure is easily and accurately estimated, and the soundness of the large structure is inspected.

[0006]

For this reason, the technical solution adopted by the present invention is based on the vibration data from the vibration sensor.
Large structure by the adaptive parameter estimation method using a physical model that estimates the natural vibration frequency of a large structure by the maximum likelihood method by combining vibration models that can be set in advance and displays the result. It is a soundness inspection method of. Further, it comprises a vibration sensor, a signal input unit for inputting a signal from the vibration sensor, a data processing unit, and an output unit for outputting an output signal from the data processing unit, wherein the data processing unit is a Kalman filter and a maximum likelihood method. Is a soundness inspection apparatus for a large structure by an adaptive parameter estimation method using a physical model, which is characterized by calculating a fundamental mode frequency of a structure which is an object to be inspected.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION A soundness inspection method for a large structure and an apparatus therefor according to an embodiment of the present invention will be described below with reference to the drawings. 2 is a block diagram of an apparatus for realizing a soundness inspection method for a large structure, FIG. 3 is a diagram of an eigenfunction of a pier vibration model, and FIG. 4 is a soundness of the present invention. FIG. 5 is a conceptual diagram of a cantilever for explaining the degree inspection method, and FIG. 5 is a data processing flowchart in the data processing unit and the output unit.

In FIG. 1, 1 is a concrete pier, 2 is a pier head, 3 is a hammer (weight) for hitting the pier, 4 is a vibration sensor (for example, an acceleration sensor unit), and 5 is sound according to the present invention. Degree checker and trigger switch, 6 is a bridge girder.

The concrete pier 1 as shown
It vibrates at its own vibration frequency due to impact. This natural vibration frequency can be regarded as the cantilever structure in which the pier 1 has the free end of the pier head.

For this reason, when the bridge pier 1 shown in FIG. 1 is regarded as a beam, assuming that the equivalent bridge pier length is h ₁ when many parts of the pier 1 are buried in soil, h ₁ is the top surface of the sediment. The value is longer than the length from to the pier head. Next, in the unhealthy state where the sand flow is carried out by the vortex flow around the bridge pier and the scour occurs due to the heavy water flow, the equivalent pier length h ₂ reduces the debris pressure, and for example, as shown in FIG. It becomes longer than h ₁ as shown by h ₂ . As a result, the natural angular frequency ω ₁ of the fundamental mode of vibration of the pier 1 becomes smaller. The present invention is characterized in that the decrease in the soundness of the pier is estimated by the decrease in the value of the natural angular frequency ω ₁ of the fundamental mode described above.

The estimation method in the case of a bridge pier will be described below. As shown in FIG. 1, the hammer 3 impacts the bridge pier head 2 of the pier 1 horizontally in the direction y. The direction y is perpendicular to the railroad track direction and is the same as the direction of the water flow of the river flowing around the pier. The basis is to strike the y-axis with the longest horizontal direction of the pier as the y-axis.

The acceleration sensor unit 4 is installed on the pier head 2 in such a direction that the vibration component in the y direction can be measured. The trigger switch provided in the inspection device 5 is pressed to allow the signal input, and the acceleration signal of the vibration due to the impact generated at the subsequent time is taken into the inspection device 5. The striking start time and the trigger switch pressing time may be mixed. When hitting with the hammer 3 held in the hand, a plurality of hits should be performed on the assumption of a missed hit. The acceleration signal transmitted from the acceleration sensor unit 4 is an analog signal, and after this signal is digitized, the estimated natural frequency of the pier is calculated and output in this inspection device (details will be described later).

The structure of the above-mentioned inspection device will be described below with reference to FIG. 2. The inspection device 5 has a computing means composed of a computer or a dedicated circuit, and as shown in the figure, receives a signal from the acceleration sensor unit. A signal input section 7 for
The data processing unit 8 includes an output unit 9 that outputs an output signal from the data processing unit, and an appropriate storage device 10 that stores a signal from the output unit. Further, the data processing unit 8 includes a noise reduction processing unit 81, an arithmetic processing unit 82,
It comprises a parameter setting means 83, a natural vibration frequency theoretical value calculating means 84, a basic mode frequency determining means 85, a judgment parameter setting means 86 and a soundness judging means 87. Note that the noise reduction processing unit can be omitted if necessary. In the arithmetic processing section 82, the natural vibration model of the pier is determined by the information from the parameter setting means 83 and the natural vibration frequency theoretical value calculation means 84, and the fundamental mode frequency is derived by the Kalman filter and the maximum likelihood method. See below).

Here, as the vibration model of the pier, the vibration model of the cantilever beam shown in FIG. 4 is used. Given the beam length h, cross-sectional area S, density ρ, beam moment of inertia I, and longitudinal elastic modulus E of the beam, the following formula (1) can be used within a range in which the effects of accessories such as bridge girders can be ignored. Is established.

[0015]

[Equation 1]

The solution y (x, t) of the equation (4) is separated into variables.

[Equation 2] Is expressed as

[0017]

[Equation 3] Then, from equations (1) and (2),

[0018]

[Equation 4]

[0019]

[Equation 5] Holds. Here, the equation (5) represents an equation satisfied by the eigenfunction of the vibration mode, and the equation (4) represents an equation satisfied by the time variation of each mode. For details, see the literature [Kamiishi / Tanaka:
Abnormality diagnosis of bridge piers using accelerometer, The 44th Joint Lecture on Automatic Control], but by solving the eigen equations satisfied by this vibration model, and from this and Equation (3),
It can be seen that ω in the equation (4) satisfies the following equation.

[0020]

[Equation 6] Here, ω _j (j = 1,2,3, ...) is expressed by equation (2) and equation (4).
As is clear from the equation, the natural angular frequency of the lateral vibration (y direction) at an arbitrary position x in the vertical direction of the pier is represented, and if j is changed, the natural angular frequency for the natural vibration of the higher order mode can be obtained. Since the angular frequency of the natural vibration when the bridge pier is regarded as an elastic body is given by equation (6), it is difficult to directly observe the natural vibration of the first mode by using this physical model, In the case where there are many cases, the reliability can be greatly improved by comprehensively evaluating the vibrations of a plurality of modes including the first order, the second order, the third order, .... Further, in order to increase the SN ratio of the sensor output, it can be understood from the shape of the eigenfunction shown in FIG. 3 that the sensor should be installed on the pier head. Actually, the observed vibration data is modeled by the sum of the natural vibrations of the three modes as follows.

[0021]

[Equation 7] Where ω_jIs the natural angular frequency of the j-th mode,
Given by equation (6), these values are₁ : Ω₂ : Ω₃ =
Satisfies 1: 6.21: 17.40. Also unique to the 4th mode and above
Vibrations and acceleration noises that are naturally mixed are collectively referred to as observation noises.
is doing. At this time, the first mode, the second mode, the third mode
X is the vibration component of the₁(t), x₂(t), x ₃(t) and
And state vector

[0022]

[Equation 8] As a sample value type dynamic model

[0023]

[Equation 9] As an observation equation

[0024]

[Equation 10] You get what you get. here

[0025]

[Equation 11] In addition,

[0026]

[Equation 12] Where I ₆ is the 6th order identity matrix, and L ⁻¹ [·] is the inverse Laplace transform. At this time, the estimation of the state vector is performed by the following Kalman filter.

[0027]

[Equation 13] And R is the observation noise v_kRepresents the variance of. As a premise
In equation (6), the physical parameters of the pier such as ρ, S, E, and I
The beam length h and the above dispersion R are not
Knowledge. If this is the case, equations (15) to (21)
Kalman filter cannot be applied. So in the present invention
Represents two unknown parameters h and R θ = (h, R) ^TRepresents
We give candidates to θ and apply the Kalman filter.
It was

In the fundamental mode frequency deciding means, the peak value having the smallest frequency is selected from the frequency giving the maximum likelihood or a plurality of peaks giving a large likelihood and decided as the eigenfrequency of the fundamental mode. The function will be described below. Likelihood function is a function that evaluates the credibility of a given parameter.
It is the one that gives the likelihood of realization of the observed value for an arbitrary parameter θ. Therefore, if θ that maximizes this probability is obtained, h that is the component of θ at this time is obtained, and the purpose of diagnosis is achieved. This likelihood function to be maximized is then given by the following equation, which can be used by the Kalman filter described above.

[0029]

[Equation 14] In reality, the physical parameters ρ, S, E, I are unknown or inaccurate, so it is better to actually obtain the natural angular frequencies such as ω ₁ , ω ₂ , ω ₃ . In addition,
Since there is a constraint condition of 1: 6.21: 17.40 between ω ₁ , ω ₂ , and ω ₃ , in applying this realistic method, the parameter that maximizes the likelihood function is θ = Consider {ω ₁ , R}.

Here, the parameter setting means 83 in FIG.
In the natural vibration frequency theoretical value calculation means 84, the physical parameters of the target large structure are input and a parameterized natural vibration model is created. Then, the arithmetic processing unit 82 calculates the state vector x _K by (1
Estimated as shown in equations (5) to (21), the realization probability likelihood (likelihood function) of this parameter is represented by equations (22) and (23).
Calculate as shown in the formula.

FIG. 5 shows a data processing flowchart in the data processing section. In the figure, when the program starts, the physical parameters of the diagnostic object are set in step S1. In step S2, the input signal is a file,
Or read in memory. In step S3, the state Becktle is estimated and calculated using the Kalman filter. In step S4, the calculation result of step S3 and the observed value are shown in FIG.
The likelihood function as shown in is calculated. In step S5, the vibration frequency of the fundamental mode that gives the maximum likelihood is determined. In step S6, the soundness determining means 87 determines whether the vibration frequency of the basic mode obtained in step S5 is within the normal control range.
Determined by The judgment is made by comparing the vibration frequency of the normal mode in the calculation of the diagnostic object in a normal time with that obtained by the diagnosis, or with the frequency of the fundamental mode vibration measured in the past that can be judged to be sound from experience. It is performed by comparing with that obtained by the diagnosis. The threshold value of the management width varies depending on the type of the diagnosis target and the installation status. In step S7, it is selected whether to output the soundness determination result. In step S8, the soundness determination result is output to the storage means.

Below, the experimental results for the bridge pier in the form shown in FIG. 1 are shown. For comparison, weight (30 kg)
Also, the measurement results when the analysis method by the proposed method and the conventional method are used for the data when using a magic hammer (4.5 kg) are also shown. First, FIG. 6 and FIG. 7 show the output signals of the acceleration pickup when the weight and the magic hammer are applied with an impact, respectively.
The accelerometer adopts a hot melt adhesive system directly above the pier so that the pier vibration in the impact direction can be sampled.

An unknown parameter θ =
FIG. 8 and FIG. 9 respectively depict likelihood functions when the proposed method with (h, R) ^T is applied. The horizontal axis shows each frequency ω ₁ of the fundamental mode of vibration in terms of frequency, and the vertical axis shows the likelihood for each frequency. Regarding the calculation of the likelihood, the observation noise variance R used for the Kalman filter should also be a variable, but since it has been confirmed that R has almost no effect on the likelihood, it can be a constant, Where the standard deviation of the observed noise (ie √R)
Was set to 10% of the effective value of the observed value. As can be seen from these figures, the frequency of the fundamental mode that gives the maximum likelihood is 17.0 Hz when the weight is used and 17.7 Hz when the magic hammer is used, which is almost the same. 30kg from this
It was found that the basic mode of vibration of the pier can be accurately measured with a magic hammer of only 4.5 kg that is convenient to carry without using the weight of. 6 and 7, the estimated waveform of the sensor output by the Kalman filter when the fundamental mode frequency is used according to the actual sensor output is shown by a broken line, but both are well estimated.
It can be seen that the physical model given here is valid.

From the above, it can be said that the proposed method using the magic hammer is effective even in the method in which the physical parameters of the bridge pier are not assumed. Therefore, if the physical parameters of the pier such as ρ, S, E, I are given correctly,
The equivalent length of the pier up to the fixed end can be accurately measured by the proposed method with unknown parameters R, h (or h one unknown parameter with R as a constant), but these parameters ρ, S, Even if E and I are not known accurately, the proposed method with unknown parameters R and ω ₁ (or one ω _{1 with} R as a constant is an unknown parameter) makes the stable state of the pier, that is, the progress of scour, It can be seen that it can be monitored though it is relatively. In addition, the measurement results when the conventional method is used for the data when the weight and the magic hammer are used are shown below.

FIG. 10 shows the frequency spectrum of the acceleration signal by the FFT (Fast Fourier Transform) of the acceleration pickup when using the weight and the magic hammer, respectively. It can be seen that the vibration frequency of the fundamental mode cannot be obtained only by analysis. Moreover, in order to perform a comparative study with the conventional method for obtaining the natural frequency of the fundamental mode in relation to the phase characteristic, FIGS. 11 and 12 show the spectrum and the phase characteristic together, particularly in the low frequency band up to 100 Hz. , The weight and the magic hammer did not find the frequency that peaks at the point where the phase suddenly changes by 180 deg, and it can be seen that the frequency of the fundamental mode cannot be obtained by the conventional method.

In the above embodiment, the bridge pier has been described as the object to be measured, but it goes without saying that the present inspection method can be applied not only to the pier but also to the estimation of the natural vibration frequency of a large structure having the same structure. Also, when considering a physical model in which a concentrated load is placed on the upper part of the beam, the natural vibration frequency can be obtained by the same means. Note that it is difficult to obtain the natural frequency by FFT, but if the parameter M (the number of repeated calculations) is appropriately selected by using MEM (maximum entropy method), the peak frequencies of the first, second, and third modes are The natural angular frequency of the fundamental mode can also be measured by comprehensively evaluating. Furthermore, the present invention can be embodied in various forms without departing from the spirit and the main features thereof. Therefore, the above-mentioned embodiment is merely an example and should not be limitedly interpreted. Further, all modifications and changes belonging to the equivalent scope of the claims are within the scope of the present invention.

[0037]

As described in detail above, according to the present invention,
By estimating the natural vibration frequency of a large structure by the maximum likelihood method using a vibration model that can be set in advance, the reliability of diagnosis is greatly improved, and inspection can be performed with a simple and effortless striking method. Becomes

[Brief description of drawings]

FIG. 1 is a diagram illustrating an inspection method of the present invention by taking a case of a bridge pier as an example.

FIG. 2 is a block diagram of an inspection device that realizes a soundness inspection method for a large structure.

FIG. 3 is a diagram of eigenfunctions of a pier vibration model.

FIG. 4 is a conceptual diagram of a cantilever for explaining the soundness check method of the present invention.

FIG. 5 is a data processing flowchart in a data processing unit and an output unit.

FIG. 6 is a sensor output signal when a shock vibration is given by a 30 kg weight.

FIG. 7 is a sensor output signal when a shock vibration is given by a 4.5 kg magic hammer.

FIG. 8 is a relationship diagram between a likelihood function and a natural vibration frequency of a fundamental mode when an impact vibration is applied with a 30 kg weight.

FIG. 9 is a relationship diagram between a likelihood function and a fundamental mode natural vibration frequency when an impact vibration is applied by a 4.5 kg magic hammer.

FIG. 10 is a frequency spectrum diagram by FFT (Fast Fourier Transform).

FIG. 11 is a frequency spectrum and phase characteristic diagram when a weight is used.

FIG. 12 shows frequency spectrum and phase characteristics when a magic hammer is used.

[Explanation of symbols]

1 Bridge pier 2 Bridge pier head 3 Hammer hitting the pier 4 Vibration sensor 5 Soundness checker and trigger switch 6 Bridge girder 7 Signal input section 8 Data processing section 9 Output section for outputting output signal from data processing section 10 From output section An appropriate storage device for storing the signal of

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Yoichi Kamiishi             5-15-5 Shimbashi, Minato-ku, Tokyo Kotsu Building             6F Synchro Co., Ltd. (72) Inventor Tetsuyuki Wada             5-15-5 Shimbashi, Minato-ku, Tokyo Kotsu Building             6F Synchro Co., Ltd. F-term (reference) 2G024 AD34 BA12 CA13 DA12                 2G064 AA05 AB01 CC41

Claims (2)

[Claims] 1. A physics capable of estimating a natural vibration frequency of a large structure by a maximum likelihood method by combining vibration data from a vibration sensor with a vibration model which can be set in advance, and displaying and storing the result. A method for checking the soundness of large structures by an adaptive parameter estimation method using a model. 2. A vibration sensor, a signal input section for inputting a signal from the vibration sensor, a data processing section, and an output section for outputting an output signal from the data processing section, the data processing section comprising a Kalman filter and An apparatus for inspecting soundness of a large structure by an adaptive parameter estimation method using a physical model characterized by calculating a fundamental mode frequency of a structure to be inspected using a maximum likelihood method.