JP2834393B2 - Automatic detection of abnormal objects ahead in tunnel construction - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for automatically detecting an abnormal front object in a tunnel construction method.

[0002]

2. Description of the Related Art In a tunnel construction method for performing a subway work or a burial work such as a power cable, an underground tunnel is excavated by a tunnel excavator. At this time, there are rocks, underground structures, foundation piles of above-ground buildings, steel materials left behind in the ground after construction, concrete blocks, etc. in front of the excavator, and digging continues without noticing Not only does the excavator break, but there is a significant delay in construction due to their removal and repair of the excavator.

For this reason, conventionally, for example, an ultrasonic sensor has been used to detect the position of an abnormal object buried in front of the excavator, and an ultrasonic wave is emitted from a sonic oscillator provided on the front of the excavator. An abnormal object is detected by detecting a reflected wave reflected by the abnormal object.

[0004]

By the way, in general, large and small crushable stones are scattered in the ground in addition to abnormal objects which are difficult to crush and need to be removed. However, in such a system that detects even large and small stones and detects only the reception time of the reflected wave by the envelope shape as in the related art, an error increases due to noise, or a plurality of abnormal objects are detected. In some cases, there is a problem that only one abnormal object can be recognized, and the position of the abnormal object to be actually removed cannot be accurately detected.

Accordingly, an object of the present invention is to provide a method for automatically detecting an abnormal front object in a tunnel construction method capable of solving the above problem.

[0006]

In order to solve the above problems, the present invention provides a method for automatically detecting an abnormal front object in a tunnel construction method. The object is sent from the transmitter
And the reflected wave is received by the receiver.
Is a method of automatic detection by
The received signal waveform of the abnormal object is sent to the transmitter and receiver.
With parameters of attenuation coefficient and natural angular frequency
Model with a secondary damping vibration system model
Between the predicted received signal waveform and the actual received signal waveform
And perform normal matching to distinguish normal buried objects from abnormal objects.
At the time of separation, the received signal waveform obtained at each observation point
Cut out a predetermined time window every time, and cut out
Time vector of received signal waveform and predicted received signal waveform
Optimal to minimize the angle between time waveform vectors
Find the matching angle and determine the optimal matching angle
The received signal waveform when it becomes smaller than the threshold of
Judging from the body, and the above optimal matching angle
So that the optimal matching angle is further minimized when obtaining
Another method is to adjust the values of the attenuation coefficient and the natural angular frequency .

[0007]

According to the above configuration, the actual received signal waveform is
By performing pattern matching with the modeled predicted received signal waveform of the abnormal object, the abnormal object is determined, so that the abnormal object can be reliably detected.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below with reference to FIGS. The gist of the present invention is that when excavating a tunnel with a tunnel excavator, an abnormal object to be removed buried underground in the front (for example, a rock that is not excavable or difficult to excavate, an underground structure, an underground structure, Foundation pile of objects,
Steel and concrete blocks left underground after construction)
Is a method that uses ultrasonic waves to automatically distinguish from buried stones and other commonly buried objects scattered underground.

Normally, in order to know the distance s _j between the observation point (x _j , y _j , z _j ) and the abnormal object, the wave reflected from the abnormal object is received by the receiver at the observation point. Time,
That is, it is only necessary to know the reception time. At this time, the distance to the abnormal object is calculated by multiplying half (1/2) of the time difference between the transmission time and the reception time by the sound speed in the soil. In calculating the distance to the abnormal object at this time, the least squares method is used.

Here, a method of calculating the distance to the abnormal object will be described. First, consider the case of finding the position of an abnormal object when there are a plurality of data items for the distance to the abnormal object in front.

That is, one transmitter and several receivers are mounted on the cutter surface in front of the tunnel excavator, and observation is performed several times each time the vehicle excavates an appropriate distance (about 1 m). Think about doing it. At this time, distance data is obtained for an abnormal object ahead by the number of observations. The specific distance can be obtained by multiplying a half (1/2) of the time from the transmission of the wave to the reception of the reflected wave from the abnormal object by the speed of sound in the soil.

Now, the traveling direction of the excavator is x-axis, the directions of the cutter surface perpendicular to the x-axis are y-axis and z-axis, and the origin is x-axis.
Place at an appropriate point on the axis. At this time, points reflected by the abnormal object at M points (hereinafter referred to as observation points) (x _j , y _j ,
z _j ) (1 ≦ j ≦ M), and the distance s _j (1 ≦ j ≦ M) to the abnormal object is determined by the above-described method.

At this time, the position of the abnormal object is set to (x, y,
z), the following equation holds.

[0014]

(Equation 1)

By the way, since s _j (1 ≦ j ≦ M) includes a measurement error, a solution (x, y, z) that satisfies the above equation cannot be generally obtained. Therefore, as described above, it is reasonable to calculate by the least squares method.

For this purpose, by using an appropriate optimization method,
The following equation (2) may be minimized for (x, y, z).

[0017]

(Equation 2)

Using the three formulas of the above formula (1),
When an approximate solution of (x, y, z) is obtained, and the approximate solution is used as an initial value, the optimum solution can be easily obtained. In this way, the distance to the abnormal object is obtained, and at this time, in the received signal, which part is the reflected wave from the abnormal object,
It is important to detect accurately. However, in order to detect a distant abnormal object, the frequency of a sound wave to be used needs to be extremely low (for example, about several kHz), and therefore, it overlaps with a reflected wave from another abnormal object and is scattered. Due to the influence of scattered waves due to large and small stones and the like, it is difficult to accurately detect the reception time from an abnormal object.

In view of this, the present embodiment proposes a method for detecting the reception time of a reflected wave which is hardly affected by disturbance by considering modeling of a reflected wave from an abnormal object.

In creating a prediction model of the reflected wave, it is necessary to first consider the characteristics of the transmitter and the receiver. In other words, the transmitter is a transducer that converts an electric signal into a sound pressure signal, and the receiver is a transducer that converts a sound pressure signal into an electric signal, and the mechanism having both ultrasonic transducers is almost the same, These input / output relationships can be expressed by secondary damped vibration system models.

Also, since it takes time for the sound pressure signal sent from the transmitter to propagate through the ground and return to the receiver,
It is necessary to include a dead time element in consideration of the propagation time in series with the two secondary damping vibration systems.

That is, as shown in the block diagram of FIG. 1, if the transmitter electrical signal and the receiver electrical signal that can be directly observed are regarded as the input and output of the sound wave propagation system, there is a
It can be considered that the following transfer function exists. In FIG. 1, [G ₀ (s)] represents the interaction between a sound wave and a disturbance.

[0023]

(Equation 3)

[0024] Here, it is _{^{ζ i j, ω i j (}} (1 ≦ i ≦ 2) is the attenuation coefficient of each secondary damping vibration system, natural angular frequency. These damping factor and natural angular frequency is acoustically In addition to the shape and material of the vibrator and the characteristics of the electric circuit connected to it, it is affected by the condition and quality of the soil surface with which the vibrator is in contact. Must be considered as a parameter.

Further, the dead time T _j of the dead time element is:
This is the reception time of the reflected wave from the abnormal object, and these unknown parameters are collectively obtained for each observation point by checking the matching of two waveform vectors described below. The gain K _{j in} equation (5) has no meaning in waveform matching, and is not considered. Also, in this modeling, in order to consider the on-line of the detection system, the underground waveform distortion is ignored and handled.

Therefore, the electric signal input to the transmitter is u
Assuming (t), the electric signal output to the receiver is obtained by the following equation.

[0027]

(Equation 4)

Where L ^-1 is the inverse Laplace transform and U
(s) represents the Laplace transform of u (t). In this embodiment, as the transmitted electric signal,

[0029]

(Equation 5)

Consider: Here, p is a wave number, and in this embodiment, it is in a range of 2 ≦ p ≦ 5. (6), the reflected wave from the abnormal object obtained by the receiver was modeled.
Next, it is necessary to examine the degree of matching between the model waveform and the received signal (electric signal) waveform actually received by the receiver at each observation point.

In order to do this, the received wave obtained at each observation point is
Signal (electric signal), start time T _j = KΔT
Cut at Indian [kΔT, (k + N-1) ΔT]
This and the prediction (in the same window) given by Eq. (6)
The degree of pattern matching with the measured signal waveform
T_j = Evaluate while changing kΔT as a parameter
No.

Here, ΔT is a sampling time when data is recorded. Incidentally, the attenuation coefficient, natural angular frequency _{^{ζ i j, ω i j (}} 1 ≦ i ≦ 2) Since it is unknown, each T _j
It is necessary to optimally adjust these parameters so that pattern matching is optimally performed.

As for the degree of pattern matching,
An angle between two waveform vectors that are easy to understand geometrically can be adopted. To minimize this angle, the Powell method is used to save computation time.

The angle between these two waveform vectors is firstly the actual angle obtained from the receiver.

[0035]

[Outside 1]

[K + N-1] ΔT] is displayed by the following number vector.

[0037]

(Equation 6)

This is an N-dimensional vector, and the time kΔT
To (k + N-1) ΔT. Of course, the width of the window is considered to be such that the entire waveform of the reflected wave from the abnormal object can be grasped.

Next, the waveform vector having the same window width is calculated for the waveform calculated based on the model of the equation (6).

[0040]

(Equation 7)

Consider: At this time, the angles of the two time waveform vectors defined by the equations (8) and (9) are given by the following equations.

[0042]

(Equation 8)

Thus, for each T _j = kΔT, the parameters ω _i ^j , ζ _i ^{j are} determined by the Powell method so as to maximize equation (9).
By adjusting (1 ≦ i ≦ 2), the optimum matching angle θ _j ^# _{, k at} each T _j can be obtained.

In order to determine whether or not the reflected wave is from an abnormal object, the optimum matching angle is compared with an appropriately determined threshold value ρ, and θ _j ^# _{, k} ≦ ρ is satisfied. It is only necessary to consider whether or not.

Assuming that k satisfying this condition is k ^# , k ^#
It can be determined that ΔT is the reception time of the reflected wave from the abnormal object. Therefore, if this operation is performed on the received signal waveform obtained at each observation point, the round trip time of the sound wave from each observation point to the abnormal object can be obtained. Then, the distance from each observation point to the abnormal object can be obtained by multiplying a half (1/2) of the round trip time by the speed of sound.

As described above, the received signal waveform of the abnormal object which can be predicted is modeled, and the modeled predicted received signal waveform and the received signal waveform from the actually embedded abnormal object are compared with each other. Pattern matching, and because this pattern matching is performed based on the optimal matching angle of the Powell method, it is possible to easily and reliably distinguish normal buried objects such as stones from abnormal objects. be able to.

In the above description, modeling of reflected waves from one abnormal object has been performed, but modeling of reflected waves from a plurality of abnormal objects is also important. It is important to detect an abnormal object near the excavator first, and it is sufficient to detect an abnormal object at a long distance after an abnormal object at a short distance is detected.

Next, the case where there are two abnormal objects will be described. If modeling in such a case is performed, it becomes possible to detect an abnormal object very close to the excavator (of course, it may be assumed to be one).

[0049] This is because the transmission signal is directly transmitted to the receiver by bypassing the resonance of the body portion of the excavator or the soil immediately before the cutter surface from the transmitter mounted on the cutter surface. That's why.

The equations (6) reflected from the two abnormal objects are expressed by two windows shown in the following equations (11) and (12).

[0051]

(Equation 9)

Then, the pattern matching between the predicted waveform on the synthetic window W _j = W _j ¹ ∪W _j ² obtained by adding these and the actual waveform on the same window W _j is evaluated. Needless to say, parameters for creating a predicted waveform on each window are unknown, and therefore, as described above, each T _j ¹ , T _j , T _{j is set} so that the angle between the synthesized predicted waveform vector and the actual waveform vector is minimized. unknown parameter is determined for _j ^2.

[0053] Thus when Motoma' optimum matching angle has exceeded the appropriate threshold, and determines in the T _j ^1, T _j ² and give the two received signals. Here, an experiment was conducted in an anechoic tank in which the medium was more uniform than underground to see how well the reflected wave from the actual abnormal object could be subjected to pattern matching using the model waveform described in this embodiment. The results will be described. In addition. The aquarium used in the experiment is an anechoic aquarium having a length, width and depth of 4 m, 5 m and 4 m as shown in FIG.

As shown in FIG. 2, one end of the water tank 1 is provided with a transmitter (acoustic sensor for transmitting wave) 2 and a receiver (acoustic sensor for receiving wave) 3 one by one, and the other end. At a distance of 3 m from the transducers 2 and 3, the thickness is 9 mm as an abnormal object.
A steel plate 4 having a length and a width of 2 m was arranged.

The frequency of the electric signal input to the transmitter 2 is 6 so that the transmission signal and the reception signal do not overlap.
kHz. Of course, the transmitted signal is a sine wave, and the wave number p
Was changed from 1 to 5. When recording the received data, the sampling frequency was 100 kHz, that is, the sampling time was ΔT = 10 μs. FIG. 3 shows a received signal (electric signal) waveform when the wave number p is 3 for reference.

In order to confirm the effectiveness of the above-described propagation model and the method of detecting the position of an abnormal object, the window is shifted for each wave number, and the pattern of actual time-series data and the time-series data predicted by the model are compared. Consider matching. However, the window width N is 120, 15 according to the wave number of the transmitted signal from 1 to 5, respectively.
0, 180, 180, 180 were employed.

For each wave number, the minimum optimum matching angle θ ₁ ^# when the window is shifted.
_{, k #} and the reception time T ₁ ^# = k ^# ΔT at that time, and the distance to the abnormal object calculated using the standard sound velocity in water at 20 ° C. of 1480 m / s is shown in [Table 1].
However, the transmission time was set to 0.

[0058]

[Table 1]

From Table 1, it can be seen that good matching is obtained particularly when the wave number is 3 or more. In addition, in each case of the wave numbers, the distance to the steel plate 4, which is an abnormal object, was approximately 3 m, and it can be seen that the distance to the steel plate 4 was found very accurately.

For reference, FIGS. 4 and 5 show the transition of the optimum matching angle θ ₁ ^# _{, k} when the wave number is 3 and the estimation parameter (ω ₁₎ at the time k ^# ΔT that gives the minimum optimum matching angle. ^{_{1 = 32.9, ζ 1 1 =}} 7.26 × 10 -2, ω 2 1
= 43.8, shows the predicted waveform of zeta ₂ ¹ = 1.22 × reception electric signals based on the 10 ^-1). The angular frequency omega ₁ ^1, a frequency corresponding to omega ₂ ¹ Each 5.24kHz, 6.97kH
z, which is very close to the resonance frequency of the transmitter and receiver (when frequency characteristics are obtained in water) of 5.5 to 5.7 kHz, supporting the validity of the above physical considerations and 3 and FIG. 5 show that the predicted received waveform can accurately represent the actual received waveform (the waveform in the range of 4 to 6 ms in FIG. 3).

However, in FIG. 3, it can be seen that the reflected wave from the abnormal object is received in about 0.4 ms. This is due to the direct propagation from the transmitter to the receiver through the medium, as seen in the received signal at the tunnel construction site described later. It can be excluded from reception time candidates. In fact, in this experiment, the distance between the transmitter and the receiver (distance between the centers of the transducers) is 0.6 m, and 0.4 ms is appropriate as the propagation time. Naturally, the model proposed for this directly propagating wave is also effective. In fact, as shown in FIG. 4, the optimum matching angle is particularly small only at one point in the vicinity of 0.40 ms in addition to 4.12 ms. Furthermore, near these extreme values, the matching angle changes sharply, and the time of reception of the reflected wave from the abnormal object becomes highly sensitive.
Moreover, it can be seen that it is detected accurately. As shown in FIG. 4, the received signal and the predicted signal waveform are both spindle-shaped sinusoidal functions. Therefore, if the window of the predicted signal waveform is shifted, the matching becomes good every half cycle, and the matching is performed in the middle. Worsens. Drawing this in detail makes it difficult to see, and it is important to find a small optimum matching angle. Therefore, in FIG. 4, portions where the optimum matching angle is small are smoothly connected. This is the same in the drawing of FIG. 9 and FIG. 11 in the below-mentioned soil experiment, but in FIG. 11 in particular, since only one extremum appears, the shape near the time when the optimal matching angle becomes the minimum is accurately determined. It is drawn in. From these figures, it can be understood once again that the time at which the minimum value is given is determined with high sensitivity. In addition,
In this connection, unknown parameters omega _i ^j when work to minimize the matching angle by Powell method, a behavior at the optimal value near the ^{_{ζ i j (1 <i <}} 2), to minimize According to data analysis It has been found that the power evaluation function is a convex function with respect to the parameters, the minimum value can be easily obtained, and it does not converge to another minimum value to cause an error in the reception time.

Next, the effectiveness of the above-described propagation model was examined using underground (soil) data actually collected at the tunnel construction site. FIG. 6 shows the state of underground data collection at this time, and FIG. 7 shows the arrangement of the transmitter 13 and the receiver 14 on the cutter surface 12 of the excavator 11. The soil at the construction site is sandy soil from the Diluvial World.

At the time of data collection, the frequency of the electric signal input to the transmitter 13 was selected to be 2.4 kHz and the wave number p was set to 5 so that an abnormal object at a relatively long distance could be detected. As an abnormal object, a vertical concrete wall 15 is provided 12.3 m in front of the excavator's initial position,
Receiving data was obtained by three receivers 14 every time one meter was excavated up to 9.3 m. The sampling frequency at this time is 25 kHz, that is, the sampling cycle is ΔT = 40 μs
And Further, the window width N when checking the matching degree is N which corresponds to a time width substantially similar to that of the water tank.
= 300 was adopted. At that time, the NO.1 receiver 14A,
The minimum optimum matching angles θ _j ^# _{, k # by the} NO. 3 receiver 14C (denoted as S ₁ and S ₃ ) and the time k ^# ΔT at which they are given, and the standard sound velocity of 1480 m / s in water were used. The distance from each receiver 14 to the concrete wall 15, which is an abnormal object, at this time is shown in [Table 2]. However, three receivers 14 attached to the cutter surface 12 of the excavator 11
A, 14B and 14C, the No. 2 receiver 14B is the transmitter 1
3, the gain of the amplifier was made smaller than that of the other two receivers 14A and 14C so that the influence would not appear. Therefore, the data obtained by the NO.2 receiver 14B was not used.

[0064]

[Table 2]

As shown in [Table 2], the minimum optimum matching angle is slightly larger than the experiment in the anechoic tank. This is because, in actual earth, unlike underwater, the medium is generally non-uniform. Looking at the distance to the concrete wall 15 where the sound speed was calculated as the standard sound speed in water, it was found that the distance was approximately the same as the distance obtained by laser ranging in the cavity after excavation.

For reference, FIG. 8 shows a received signal waveform (electric signal waveform) obtained by the No. 1 receiver 14A when the distance from the excavator 11 to the concrete wall 15 is 10.3 m.
FIG. 9 shows a window [kΔT, (k + N−1)].
ΔT] is a time change of the optimum matching angle θ _j ^# _{, k} obtained for the received waveform and the model waveform within the estimated received time t = k
4 shows a graph drawn against ΔT.

As in the water tank experiment, a point (4.40 ms) where the optimum matching angle is reduced before the time (13.36 ms) indicating the reception of the reflected wave of the abnormal object.
Is seen. This is considered to be caused by multiple reflections in the muddy water chamber behind the cutter surface, because the error always appears in the same time zone regardless of the receiver and the measurement point.

Therefore, when actually detecting an abnormal object, it is possible to remove the received signal received around 4 ms. The results shown in Table 2 are based on the received signals excluding the spindle-shaped propagation signals. In FIG. 8, small reflected wave-like signals appear at two places immediately after the transmission and at about 26 ms. The former is a signal that has propagated linearly on the cutter surface, and the latter is a signal having a cutter surface and an abnormal object. It is considered that the signal is a reflected wave signal that has been received by making two round trips between.
These are all removable. However, in FIG. 9, the matching angle is not small in the former case,
This is because the small propagating signal is buried in the above-mentioned large propagating wave signal starting at 4.40 ms, and effective matching cannot be obtained only with a single window. However, for the latter, the matching angle is smaller, and the reception time (about 26 ms) is the first reflected wave reception time 1
It is almost twice as long as 3.36 ms. Furthermore, the received signal is small, which supports the above-mentioned idea of repeating reflection and propagation through the ground twice.

Also, according to the present experiment, the
Optimal matching angle threshold for judging a launch wave
Is about 15 to 17 degrees.
Was. In other words, the degree of matching at the angle defined by equation (10)
When considering the degree, the value of this level is significant for detecting abnormal objects.
You can see the taste. Of course, θ_{j, k} Instead of cos
θ _{j, k} It is possible to consider a certain amount (correlation coefficient).
Also, the value tends to increase or decrease from the functional relationship between θ and cos θ.
Are different but correspond to cosine values of 15 to 17 (degrees)
It can be seen that the value that has a significance as a threshold value. Above
Using the cosine value described above, an evaluation that indicates the degree of matching
The function is smoother near the extremes, but elsewhere
Then, on the contrary, the inclination increases. Therefore, near the extremum
Has the advantage that it becomes sharper,
The direction is weakened, making it difficult for humans to search for extreme values
Seems like. However, as we have mentioned
In the meantime, there is no other extremum near the meaningful extremum,
There seems to be no problem with machine optimization.

Next, the excavator is operated to receive the NO. 1 and NO. 3 receivers 14 from 12.3 m to 9.3 m from the abnormal object.
Using the total of eight measured data by A and 14C, the abnormal object is regarded as a point (x, y, z), and the position of the abnormal object and the sound velocity are obtained by the least square method. Taking the origin, the position of the abnormal object is (x, y, z) =
(12.13, -0.218, -0.004), sound velocity is 1510.8m /
s.

From this, it can be seen that the abnormal object is located at a distance of 12.13 m from the initial position in the traveling direction of the excavator, and is very close to the actually measured value of 12.3 m by the laser ranging. It was also found that the sound speed underground at the construction site was slightly higher than the standard underwater sound speed of 1480 m / s.
As described above, the standard underwater sound speed of 1480 m / s may be used, but if the sound speed is also treated as a variable by the least square method, the position of an abnormal object with high reliability can be determined even when little information is obtained on soil properties. Is possible.

For reference, if the standard sound velocity in water is used, the distance to the abnormal object is 11.90 m, and it can be seen that the measurement accuracy is slightly deteriorated. The sound velocity in the ground was calculated using the data at a place slightly closer to the abnormal object, that is, four observation data points at 10.3 m and 9.3 m from the abnormal object. When the position of the abnormal object is determined, including the position 10.3 m ahead of the abnormal object, the sound velocity is 1504.3 m / s, and the position of the abnormal object (x, y, z) = (10.23, -0.084, 0.00) I was asked.

In this case, since the distance to the abnormal object has become somewhat shorter, it is considered that more reliable observation data was obtained, and the measurement error related to the position of the abnormal object was only-
0.07 (= 10.23-10.3) m. In addition, a slight change in the speed of sound due to the change in soil quality is also observed.

Although not described in the above description, in actual application, it is not necessary to shift the window by one sampling time from the zero time to obtain the optimum matching angle, and use the envelope shape of the received signal. The calculation of the optimum matching angle may be started before or after the time when a thing like a reflected wave from the abnormal object appears. Since the envelope of the reflected wave from this abnormal object has a spindle shape, rough registration is easy. Therefore, in addition to the high speed of the Powell method, this method requires little calculation time and can be used online.

In the above description, the case where the reflected wave from the abnormal object is relatively easy to handle, in which the reflected wave is separated from other reflected waves or directly propagated waves, is described next. To examine the effectiveness of the overlap, consider the case where the excavator is close to the abnormal object in the previous underground experiment, and the reflected wave from the abnormal object overlaps with the direct propagation wave from the transmitter.

As described above, the directly propagated wave is represented by a waveform similar to the reflected wave from the abnormal object. Therefore, if this effectiveness can be confirmed, not only the position detection of a short-range abnormal object can be performed, but also the abnormal object detection when there are a plurality of abnormal objects, as described above.

That is, the distance from the abnormal object is 4.3 m
FIG. 10 and FIG. 11 show the transition of the received signal waveform appearing at the No. 1 receiver 14A at that time and the optimum matching angle at that time.
Shown in Of course, the reflected wave reception waveform model at this time is (6)
It is a linear combination of two waveforms (with different windows) in the form of an equation.

As shown in FIG. 10, in the judgment based on the simple envelope shape, only one reflected wave from the abnormal object is known, and it is difficult to determine the time of receiving the reflected wave from the true abnormal object. However, in the proposed method (primary coupling model), as shown in FIG. 11, the minimum optimal matching angle is 7.49 degrees, and the start time of the two reflected waves (to be linearly coupled) that gives this is 3.49 degrees. 40 ms, 5.
92 ms, that is, reception of the directly propagated wave and the reflected wave from the abnormal object is 3.40 ms and 5.92 ms, respectively, and the two received signal waveforms can be clearly separated. The horizontal axis of FIG. 11, there for the start time T ¹ of the one window, start time T ² of the other windows are is optimized in conjunction with other unknown parameters. That is, one window start time that minimizes the optimal matching angle is
From FIG. 11, T ^{1 #} = 3.40 ms, and the start time of the other window at this time is separately T ^{2 #} = 5.
92 ms (details are omitted).

Therefore, if the distance to the abnormal object is calculated using the standard underwater sound speed, 1480 m / s × (5.92 ×
10 ⁻³ /2)s=4.44 m, which is a value very close to the actually measured distance of 4.3 m by laser distance measurement. On the other hand, 3.4
Other received waves appearing from around 0 ms are direct propagation signals that always appear in the initial time zone during transmission as described above, and can be excluded from the beginning. For reference,
When indicating the future parameters to achieve the best match of the received signals of FIG. 10, for the direct propagation wave, ω _{¹ 1} = 15.
^{_{86, ζ 1 1 = 0.418,}} ω 2 1 = 15.51, ζ 2 1 =
0.01, and for the reflected wave from the abnormal object, ω ₁ ^{2
_{= 14.39, ζ 1 2 = 0.049}} , ω 2 2 = 15.91,
ζ _{² 2} = 0.01 was obtained. In both cases, values of about 2.3 to 2.5 kHz are obtained for the angular frequency, and these are the resonance frequencies of the transmitter and the receiver in water 5.5 to 5.7.
kHz. This means that the resonance frequency of the vibrator is lower in the soil than in the water,
Physically reasonable. By the way, 16kHz in the water tank
The optimal angular frequency ω for realizing the optimal pattern matching is also 6 kHz when the transmission signal of the frequency
The value is equivalent to 5 to 6 kHz as in the case of z, and the model parameter hardly changes in water regardless of the magnitude of the electric signal input frequency.

In the actual application, which waveform model is to be applied is determined by the equation (6) if a single reflected wave is considered due to the envelope shape, and a plurality of models (here, two).
If the reflected wave is a synthetic wave, a model obtained by linearly combining the model of Expression (6) with changing the start time may be used. For reference, spherical objects and inclined steel plates were also tested as abnormal objects, and almost the same results as above were obtained. In particular, in the case of the inclined steel plate, data similar to the present experiment (that is, similar to the vertical steel plate) was obtained up to the inclination of about 10 degrees. It can be seen that there is a tendency that a point near the shortest distance point tends to be determined as the position of the abnormal object, and that the present measurement system is also useful for preventing excavator accidents in the case of an inclined wall. This is because the directivity of the transmitter is not so high because the used frequency is low.

As described above, the method for detecting the position of an abnormal object described in this embodiment was confirmed to be effective through experiments in an anechoic tank and underground.

[0082]

As described above, according to the method for automatically detecting an abnormal object of the present invention, a received signal waveform of an abnormal object which can be predicted is modeled, and the modeled predicted received signal waveform and the actual Since the pattern matching with the received signal waveform from the abnormal object buried in the object is performed, it is possible to reliably distinguish the normal buried object such as a crushable stone from the abnormal object. In particular, pattern match
Signal waveforms to be transmitted are reduced at the transmitter and receiver.
Taking into account parameters such as attenuation coefficient and natural angular frequency
Using the one modeled with the secondary damping vibration system model
Abnormalities in the soil where large and small stones are scattered in the detection area
When detecting an object, the scattering of sound waves due to crushable stones, etc.
Eliminates the effects of disturbance and reflects waves from multiple abnormal objects
To determine the position of each abnormal object
It is possible to respond intelligently, such as the ability to do.

[Brief description of the drawings]

FIG. 1 is a block diagram showing input and output of a sound wave propagation system in an automatic detection method according to an embodiment of the present invention.

FIG. 2 is a plan view of a water tank used in a water tank experiment of the automatic detection method of the embodiment.

FIG. 3 is a graph showing a received signal waveform obtained in the water tank experiment.

FIG. 4 is a graph showing a temporal change of an optimum matching angle obtained in the water tank experiment.

FIG. 5 is a graph showing a waveform of a received electric wave predicted in the water tank experiment.

FIG. 6 is a sectional view showing an underground experiment of the automatic detection method according to the embodiment.

FIG. 7 is a front view showing an arrangement state of the transducer on the cutter surface of the excavator used in the underground experiment.

FIG. 8 is a graph showing a received signal waveform obtained in the underground experiment.

FIG. 9 is a graph showing a temporal change of an optimum matching angle obtained in the underground experiment.

FIG. 10 is a graph showing a waveform of a received signal with respect to an abnormally close object in the underground experiment.

FIG. 11 is a graph showing a time change of an optimum matching angle with respect to a proximity abnormal object in the underground experiment.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Water tank 2 Transmitter 3 Receiver 4 Steel plate 11 Excavator 12 Cutter surface 13 Transmitter 14 Receiver 15 Concrete wall

Continuation of the front page (72) Inventor Masao Kinoshita 5-28 Nishikujo, Konohana-ku, Osaka-shi, Osaka Inside Hitachi Zosen Corporation (56) References JP-A-5-52949 (JP, A) JP-A Sho 58-223771 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G01S 15/88 G01S 13/88 E21D 9/06 301 G01V 1/00

Claims (1)

(57) [Claims] When excavating a tunnel, an abnormal object to be removed buried in the ground in front of the tunnel is sonicated from a transmitter.
And the reflected wave is received by the receiver.
Is a method of automatic detection by
The received signal waveform of the abnormal object is sent to the transmitter and receiver.
With parameters of attenuation coefficient and natural angular frequency
Model with a secondary damping vibration system model
Between the predicted received signal waveform and the actual received signal waveform
And perform normal matching to distinguish normal buried objects from abnormal objects.
At the time of separation, the received signal waveform obtained at each observation point
Cut out a predetermined time window every time, and cut out
Time vector of received signal waveform and predicted received signal waveform
Optimal to minimize the angle between time waveform vectors
Find the matching angle and determine the optimal matching angle
The received signal waveform when it becomes smaller than the threshold of
Judging from the body, and the above optimal matching angle
So that the optimal matching angle is further minimized when obtaining
A method for automatically detecting an abnormal front object in a tunneling method, wherein a value of a damping coefficient and a value of a natural angular frequency are adjusted .