A passive excitation type bridge flaw detection device

Abstract

The present invention discloses a passive excitation type bridge flaw detection device, comprising: an excitation part, a damage location and evaluation part, and a counterweight connection part. Among them, the excitation part includes an excitation wheel with uniformly distributed excitation teeth on its surface, which is used for passive excitation of the bridge deck at a fixed excitation frequency. Both ends of the central bearing of the excitation wheel are fixed with fixing parts having rectangular end faces, and the fixing parts are fixedly connected to the end grooves of the approach slab through bolts. An acceleration sensor is arranged on the wheel shaft and is used for collecting the acceleration signals transmitted from the bridge surface to the excitation teeth; the damage location and evaluation part determines the moment of stiffness mutation detected by extracting the maximum points of the equivalent acceleration from the time-frequency analysis results, and determines the bridge damage location and evaluates the damage degree corresponding to the inflection points in the power spectrum diagram.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge detection, and particularly relates to a passive excitation type bridge flaw detection device. The excitation teeth uniformly distributed on a single excitation wheel perform passive excitation with a fixed excitation frequency on the bridge deck, and an acceleration sensor is used to collect the acceleration signal transmitted from the bridge surface to the excitation wheel. By considering the influence of the derivative term of the equivalent stiffness with respect to time on the acceleration amplitude, the stiffness change edge is obtained. By analyzing the moment of stiffness mutation within the time domain controlled by the second derivative term of the equivalent stiffness with respect to time, and then finding the inflection point in the power spectrum diagram near this moment to determine the bridge disease location and evaluate the disease degree. Background Art

[0002] With the rapid economic development in the past more than 30 years, the infrastructure scale of highway bridges has also increased rapidly. At present, there are more than 800,000 highway bridges in service in China, with a total mileage of more than 40 million meters. While promoting economic development, it is inevitable that over time, different degrees of damage will occur due to factors such as construction technology, materials, overload effects or the environment, and timely maintenance and repair are required.

[0003] At present, the discovery of highway bridge diseases mainly relies on manual identification. However, the huge number of bridges and the mileage base pose great challenges to bridge maintenance work. If diseases cannot be discovered in time, their degrees cannot be evaluated, and corresponding strategies cannot be formulated, it is very likely to cause safety accidents and losses of life and property. Therefore, improving the detection efficiency and accuracy through automated detection is the only way for bridge disease detection.

[0004] The invention patent with the application number CN201810339418.3 and the name of "a passive percussion type material damage detection device and method" proposes a self-balancing detection vehicle. The double wheels of the self-balancing vehicle are replaced with excitation wheels with excitation teeth on the surface. When driving, it percusses the material surface, and the sensor collects the signals in the detection area. By calculating the spectral envelope of the collected detection signals, the damage indication value of the detection surface is calculated, and the position where the damage indication value suddenly drops is determined as the damage position, and the degree of damage is measured by the magnitude of the mutation. However, since the damage indication values at non-damage positions are close, the damage indication value determination method leads to large and complex calculations. The design of the double excitation wheels also causes interference to easily form between the feedback signals due to different percussion positions, increasing the difficulty of signal analysis and reducing the accuracy of the detection results at the same time. Summary of the Invention

[0005] In view of the above problems, the present invention provides a passive excitation type bridge flaw detection device, which uses excitation teeth uniformly distributed on a single excitation wheel to perform passive excitation with a fixed excitation frequency on the bridge deck, and collects the acceleration signal transmitted from the bridge surface to the excitation wheel through an acceleration sensor. By considering the influence of the derivative term of the equivalent stiffness with respect to time on the acceleration amplitude, the stiffness change edge is obtained. By analyzing the stiffness mutation moment in the time domain controlled by the second derivative term of the equivalent stiffness with respect to time, and then finding the inflection point in the power spectrum diagram near this moment to determine the bridge disease location and evaluate the disease degree.

[0006] A passive excitation type bridge flaw detection device, comprising: an excitation part, a damage location and evaluation part, and a counterweight connection part, characterized in that

[0007] The excitation part includes an excitation wheel with uniformly distributed excitation teeth on its surface, which is used to perform passive excitation with a fixed excitation frequency on the bridge deck. Both ends of the central bearing of the excitation wheel are fixed with fixing parts having rectangular end faces. The fixing parts are fixedly connected to the end grooves of the coping slab through bolts. An acceleration sensor is arranged on the wheel shaft, which is used to collect the acceleration signal transmitted from the bridge surface to the excitation teeth;

[0008] The damage location and evaluation part determines the detected stiffness mutation moment through the maximum point of the equivalent acceleration extracted from the time-frequency analysis result, and determines the bridge damage location and evaluates the damage degree corresponding to the inflection point in the power spectrum diagram;

[0009] The counterweight connection part includes a square groove member and a coping slab. The square groove member is connected to the front body drive part and the rear body excitation part through a slot and bolts; the coping slab is welded and fixed on the square groove member, and includes a top triangular connection part and a support arm for enhancing the torsional stiffness, and a bottom square steel connection part for enhancing the bending stiffness.

[0010] Furthermore, the steps of filtering the acceleration signal by the damage location and evaluation part are as follows:

[0011] First, perform time-frequency analysis on the acceleration signal to obtain n signals, each signal corresponding to a specific moment, with a corresponding frequency spectrum vector Y i , i = 1, 2,..., n. First, sum up each vector to obtain the total energy E at this moment i , and then find the mean value of the energies at all moments and the standard deviation Calculate the mean value of the frequency spectrum vector at the i-th moment and the standard deviation Furthermore, obtain the coefficient of variation Q for measuring the fluctuation of the frequency spectrum at this moment i = σ i / μ i , and the mean value of the coefficients of variation at all moments and standard deviation The coefficient of variation fluctuation represents the stiffness mutation caused by damage.

[0012] Furthermore, the damage location step of the damage location and evaluation part is as follows:

[0013] First, record the position point vector loc after removing noise according to the filtering algorithm s :

[0014] loc s = {i | E i > μ E + B × σ E ||Q i > μ Q + B × σ Q , i = 1, 2,..., n};

[0015] where B is the noise interference degree;

[0016] Then, calculate the equivalent acceleration to obtain a two-dimensional curve, select the maximum value point, and count its position into the vector loc peak , where PSD max is the maximum value of the power spectral density in the frequency domain at each moment, and the damage location with an obvious stiffness mutation is loc f = loc s ∩ loc peak .

[0017] Furthermore, the evaluation of the damage degree of the damage location and evaluation part is determined by the following formula:

[0018]

[0019] where G and H are dimensionless coefficients calibrated in advance.

[0020] The excitation teeth evenly distributed on the excitation wheel of the present invention perform passive excitation with a fixed excitation frequency on the bridge deck, simplify the detection structure and avoid interference between excitation sources, determine the rigid mutation point through the power spectrum corresponding to the acceleration peak mutation time, the detection method is simple and efficient, and the detection accuracy is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is the front view of the bridge inspection vehicle body;

[0022] Figure 2 is the side view of the bridge inspection vehicle body;

[0023] Figure 3 is the assembly schematic diagram of the excitation part and the counterweight connection part;

[0024] Figure 4 It is a schematic diagram of the counterweight connection part;

[0025] Figure 5 It is a schematic diagram of the bridging slab;

[0026] Figure 6 It is a schematic diagram of the square groove component;

[0027] Figure 7 It is a schematic diagram of the excitation part;

[0028] Figure 8 It is a schematic diagram of the basic model of the knocking scanning method;

[0029] Figure 9 It is a schematic diagram of the damaged beam model;

[0030] Figure 10 It is a diagram of the natural frequency distribution form of the trolley when there is a stiffness mutation in the beam;

[0031] Figure 11 is the diagram of the influence of time-varying ω Vd on acceleration. Among them, the excitation frequency of Figure 11(a) = 141Hz, and the excitation frequency of Figure 11(b) = 131Hz;

[0032] Figure 12 is the diagram of the distribution of the cross-sectional moment of inertia. Among them, Figure 12(a) is the actual stiffness ratio curve diagram, and Figure 12(b) is the weighted stiffness ratio curve diagram;

[0033] Figure 13 is the distribution diagram of θ w under different window lengths. Among them, the window length of Figure 13(a) is 2m, the window length of Figure 13(b) is 1m, and the window length of Figure 13(c) is 0.5m;

[0034] Figure 14(a) is the acceleration curve diagram of the trolley when θ = 1 / 1.07. Among them, Figure 14(b) is the power spectral density diagram of the trolley when θ = 1 / 1.07;

[0035] Figure 15 is when considering the extreme points at the stiffness mutation. Among them, Figure 15(a) is the distribution data point diagram of and θ, Figure 15(b) is the quadratic curve diagram fitted according to the data points in Figure 15(a), and Figure 15(c) is the distribution data point diagram of and θ s and Figure 15(d) is the quadratic curve diagram fitted according to the data points in Figure 15(c);

[0036] Figure 16 shows the conversion of characteristic points within the influence region of the second derivative at different values of θ. Specifically, in Figure 16(a), θ = 1 / 1.2; in Figure 16(b), θ = 1 / 1.07; in Figure 16(c), θ = 1; in Figure 16(d), θ = 1.03; in Figure 16(e), θ = 1.07; and in Figure 16(f), θ = 1.2.

[0037] Figure 17(a) is a schematic diagram of the structure of the experimental beam; Figure 17(b) is a side view of the experimental beam.

[0038] Figure 18 is the natural frequency spectrum of the inspection vehicle model.

[0039] Figure 19(a) is the acceleration curve of the inspection vehicle, and Figure 19(b) is the corresponding power spectral density diagram.

[0040] Figure 20(a) shows the distribution of the equivalent acceleration Y of the inspection vehicle max and the detection points, and Figure 20(b) is the corresponding power spectral density diagram.

[0041] Figure 21 is the distribution diagram of the equivalent acceleration of the inspection vehicle.

[0042] In Figure 22(a), the vertical vibration frequency is approximately 78 Hz. Figure 22(a-1) is the side view modal simulation diagram of the rear part of the monitoring vehicle, and Figure 22(a-2) is the upward view modal simulation diagram of the rear part of the monitoring vehicle.

[0043] In Figure 22(b), the vertical vibration frequency is approximately 103 Hz. Figure 22(b-1) is the side view modal simulation diagram of the rear part of the monitoring vehicle, and Figure 22(b-2) is the upward view modal simulation diagram of the rear part of the monitoring vehicle.

[0044] In Figure 22(c), the vertical vibration frequency is approximately 117 Hz. Figure 22(c-1) is the front view modal simulation diagram of the rear part of the monitoring vehicle, and Figure 22(c-2) is the rear view modal simulation diagram of the rear part of the monitoring vehicle.

[0045] In Figure 22(d), the vertical vibration frequency is approximately 254 Hz. Figure 22(d-1) is the front view modal simulation diagram of the rear part of the monitoring vehicle, and Figure 22(d-2) is the rear view modal simulation diagram of the rear part of the monitoring vehicle.

[0046] In Figure 22(e), the vertical vibration frequency is approximately 339 Hz. Figure 22(e-1) is the front view modal simulation diagram of the rear part of the monitoring vehicle, and Figure 22(e-2) is the rear view modal simulation diagram of the rear part of the monitoring vehicle.

[0047] Figure 23(a) is the modal simulation diagram of the entire vehicle when the vertical vibration frequency is approximately 60 Hz.

[0048] Figure 23(b) is the modal simulation diagram of the entire vehicle when the vertical vibration frequency is approximately 69 Hz.

[0049] Figure 23(c) is the modal simulation diagram of the whole vehicle when the vertical vibration frequency is approximately 74 Hz;

[0050] Figure 23(d) is the modal simulation diagram of the whole vehicle when the vertical vibration frequency is approximately 87 Hz;

[0051] Figure 23(e) is the modal simulation diagram of the whole vehicle when the vertical vibration frequency is approximately 101 Hz;

[0052] Figure 23(f) is the modal simulation diagram of the whole vehicle when the vertical vibration frequency is approximately 152 Hz;

[0053] Figure 24 It is a schematic diagram of the working mode of the inspection vehicle;

[0054] Explanation of reference numerals:

[0055] 1 - driving wheel, 2 - suspension system, 3 - vehicle frame, 4 - hardware box, 5 - attitude sensor, 6 - counterweight connection part, 7 - square groove member, 8 - boarding plate, 9 - excitation part, 10 - slot, 11 - end slot, 12 - center bearing, 13 - excitation wheel, 14 - flange, 15 - sensor installation position, 16 - fixing part. Specific implementation manner

[0056] To enable those skilled in the art to better understand the technical solution of the present invention, the structure of the detection device and the principle of the passive excitation type bridge flaw detection method will be described in detail below.

[0057] The present invention provides a bridge inspection vehicle body structure, including: a driving part, an excitation part and a counterweight connection part. As shown in the schematic diagram of the bridge inspection vehicle body in Figure 1 and 2 , the driving part includes two driving wheels 1 and two 10-inch 500W hub motors installed on the vehicle frame 3, a suspension system 2 and a hardware box 4. The suspension system 2 may include hydraulic nitrogen shock absorbers for reducing the driving wheel noise to reduce the interference with the excitation signal. An installation position for installing an attitude sensor is provided on the top of the vehicle frame 3, and the attitude sensor is used to realize semi-supervised inertial navigation. A battery and a control system may be provided in the hardware box 4.

[0058] As shown in Figure 4 , the counterweight connection part 6 is composed of a square groove member 7 and a boarding plate 8. The boarding plate 8 is welded and fixed on the square groove member 7. As shown in Figure 2 , 3 and 6, the square groove member 7 is connected to the front body driving part and the rear body excitation part through 4 slots 10 and bolts. As shown in Figure 5 , the boarding plate 8 includes a top triangular connection part and a support arm for enhancing the torsional stiffness, and a bottom square steel connection part for enhancing the bending stiffness. Two end slots 11 are provided on the bottom square steel connection part.

[0059] As Figure 7 shown, the excitation part 9 includes an excitation wheel 13 with uniformly distributed excitation teeth on its surface. A flange 14 is arranged inside the hub of the excitation wheel 13. The central bearing 12 of the flange 14 is sleeved on the round shaft and fixed. Fixing pieces 16 with rectangular end faces are fixed at both ends of the central bearing 12. The fixing pieces 16 are fixedly connected to the end slots 11 of the lap plate 8 by bolts. The central bearing 12 can be a deep groove ball bearing. A sensor mounting position 15 is arranged on the round shaft, and the sensor can be an IEPE accelerometer. The excitation wheel 13 can be composed of a rubber tire with a hardness of 75A and an aluminum alloy hub. For example, the parameters of the excitation wheel 13 are as follows: the radius r is 125 mm, the width is 50 mm, and the number of patterns n of the rubber tire is 72. The relationship between the excitation frequency and the above parameters is f = vn / 2πr. Therefore, when the detection speed is 1.5 m / s, the excitation frequency is about 138 Hz.

[0060] After installing the acceleration sensor and the flaw detection signal analysis element on the body structure of the above bridge flaw detection vehicle, by using the fact that the degree of mutation at the stiffness mutation point on the beam is approximately quadratic with the square root of the acceleration amplitude or power spectral density of the vehicle, the characteristic points corresponding to the derivative term are found for disease location.

[0061] Fast bridge stiffness anomaly point detection mode of the bridge flaw detection vehicle:

[0062] The working mode of this method is as Figure 24 shown. According to the instruction transmission timing, each module can be divided into a user terminal, a control mode selection, a host computer, a lower computer, and the underlying hardware from top to bottom.

[0063] For the used scenarios, the user terminal needs to select the control mode according to the needs respectively. For example, when the device needs to be quickly moved at the site to be detected, the lower computer can be directly controlled by using the method of short-range remote control to achieve quick steering or turning around; while during the detection process, the semi-supervised inertial navigation method is adopted, that is, the detection vehicle mainly performs detection at a fixed speed by the inertial navigation method. However, if there is a large deviation in the inertial navigation sensor due to the expansion joint on the way, the user terminal can correct it.

[0064] The functions of the host computer include receiving and processing the instructions of the user terminal, collecting the data of the inertial navigation sensor and the detection data, and analyzing and storing the detection data. In terms of the control flow, after receiving the user instructions, the host computer sends the tasks required during the detection implementation to the lower computer for execution, and accepts the feedback from the lower computer. Finally, the processed detection results are fed back to the user terminal for viewing.

[0065] The functions of the lower computer are mainly to complete the tasks of various hardware controls, including driving the motor, collecting the rotation speed, monitoring the power and temperature, and avoiding obstacles and emergency stops.

[0066] Principle of the stiffness identification algorithm for beam structures considering unsteady effects: Figure 8 is the basic model of the percussion scanning method, where the flexural stiffness of the Euler-Bernoulli beam is EI, the mass per unit length is m, and the damping coefficient is μ B . The trolley is simplified into a rigid body with a mass of M V and a combination of wheels with a mass of M W . The connection stiffness between the body and the wheels is k V , and the damping coefficient is μ V . The contact stiffness between the wheels and the beam is k W , and the damping coefficient is μ W . The road surface roughness is represented by r(x). The excitation teeth of the wheels can be regarded as a contact interface without thickness (i.e., the displacement of the excitation teeth is the displacement of the beam at that place), and its stiffness and damping are reduced to k W and μ W . The inertial force generated by the mass M T of the excitation teeth is the passive percussion force F T . And the force F V acting on the body is the active percussion force. In the following derivation, all displacements and forces are positive in the y direction; the subscripts V, W, T, and B represent the body, wheels, excitation teeth, and beam respectively. For this system, the following equilibrium equations can be established at x = vt in the local coordinate system (x, y) of the beam (the body takes the state balanced with its own weight as the origin):

[0067]

[0068]

[0069]

[0070]

[0071] where: F is the supporting force in the y direction of the beam on the wheels; The dot above this formula represents the derivative with respect to time t; the superscript'

[0072] represents the derivative with respect to x. Only considering the case where the wheels and the beam surface are tightly connected, then y T = y B . According to equation (3), it can be known that:

[0073]

[0074] There are excitation teeth with a mass of M T at the wheel contact interface, then represents the percussion force F T, equations (1) to (3) are the axle-bridge coupling equations of the passive percussion scanning method.

[0075] If there is no dynamic vibration damping phenomenon in the wheel, the above model can be further simplified to a single-degree-of-freedom system, and the equation can be simplified to:

[0076]

[0077]

[0078]

[0079] Because the car moves at a constant speed v, and the teeth of the excitation wheel are evenly distributed, the pulse excitation in each percussion process is approximately a half-sine wave, so F can be assumed to be:

[0080]

[0081] Assume that the surface topography of the beam is composed of a smooth surface superimposed with a rough perturbation, then A and ε s represent the percussion forces generated on the smooth surface and the rough perturbation respectively, and their magnitudes are determined by the axle load and the vehicle speed. ε s is the noise generated by wheel slip. τ0 represents the duration of the excitation force, and the circular frequency of the half-sine wave satisfies ω F = π / τ0, T = 2π / ω0 is the period of the excitation force. When the radius of the excitation wheel is r and the number of teeth on its pattern is n, the percussion circular frequency is determined by the following formula:

[0082]

[0083] Denote the duty cycle of the excitation tooth as η, then:

[0084]

[0085] So:

[0086]

[0087] If the tooth depth of the gear is large enough relative to the road surface roughness, then A >> ε s . Further ignoring the influence of slip, equation (9) can be expanded using Fourier series, and finally we get:

[0088]

[0089] It can be seen that as n increases, the higher-order terms decrease rapidly. So usually the percussion force F T, the frequency is mainly ω0. Moreover, to avoid interference from environmental noise, ω0 is set at a relatively high frequency. At this time, the knocking force of the high-order frequency is far from the sensitive frequency of the trolley. Therefore, only the case of n = 1 will be discussed below. At this time:

[0090]

[0091] In particular, when 2η = 1:

[0092]

[0093] According to equations (6) to (8), the origin displacement impedance of the beam at the excitation wheel can be obtained:

[0094]

[0095] Thus, it can be known that the acceleration of the trolley is related to the origin displacement impedance and therefore contains information about local damage:

[0096]

[0097] Since the rigidity of the bridge is very large, it can be considered that the deformation of the bridge under the knocking action of the trolley is very small and the vibration response is linear. Therefore, the bridge displacement can be expressed as:

[0098]

[0099] Where: is the j-th mode of the bridge; q Bj (t) is the modal coordinate.

[0100] Substitute equation (16) into equations (6) to (8), and then multiply both sides of the equation by and then integrate along the length of the beam to obtain:

[0101]

[0102] Where, ω Bj is the j-th natural circular frequency of the bridge:

[0103]

[0104] On the premise that the trolley moves at a low speed, it is easy to satisfy So the in equation (17) can be ignored. In addition, F T >>M T g, M T g can be ignored. So the decoupled equation can be obtained. Substitute equation (13c) into this equation and use the Duhamel integral to obtain the forced vibration solution (Note: Because slight local damage will not cause ω Bjhas changed greatly, so it is considered that ω in the following formula Bj is a constant; in addition, it is considered that the initial displacement and initial velocity of the beam are zero):

[0105]

[0106] wherein:

[0107]

[0108]

[0109] Substituting Equation (16) into Equation (6), we get:

[0110]

[0111] where: 2n V = μ V / M V ; k is related to the position on the beam, so the equivalent natural frequency is a function of time:

[0112]

[0113] It can be seen that Equation (22) is a second-order linear ordinary differential equation with variable coefficients. To derive the analytical solution, this mechanical model equivalently represents the actual damaged beam + undamaged trolley as an undamaged beam + damaged trolley, that is, it is considered that k is related to the suspension stiffness k V of the trolley and the damage condition of the beam at the contact point. Therefore, the equivalent natural frequency in Equation (22) is a function of time.

[0114] Assume that the stiffness distribution of the damaged beam is as Figure 9 shown. Then, according to the stiffness equivalent model, we know that:

[0115]

[0116]

[0117]

[0118] where: θ = (EI)2 / (EI)1, α = a / L, β = b / L, and γ = c / L. D varies with position, being a quadratic polynomial of γ outside the damage zone and a quartic polynomial of γ inside the damage zone, and taking an extreme value at the center of the damage zone.

[0119] Based on the above analysis, it can be considered that ω V in Equation (22) is a piecewise function. At this time, the forced vibration displacement of the trolley can be solved using the piecewise Duhamel integral. Assume that damage starts to occur at x = a:

[0120] When \(0\leq v_t\lt a\), the beam is undamaged. The initial displacement and initial velocity of the trolley are both zero:

[0121]

[0122]

[0123] When \(a\leq v_t\), the beam is damaged. The initial displacement and initial velocity of the trolley are \(y\) V (a / v) and

[0124]

[0125]

[0126] Next, the influence of stiffness change on acceleration is qualitatively analyzed according to equations (27) and (29).

[0127] Take the beam parameters as \(E = 27.5\) GPa, \(I = 0.12\) m⁴, \(m = 4800\) kg / m, \(L = 25\) m. Only consider the first ten-order modes, and the first two-order frequencies are 2.084 Hz and 8.336 Hz respectively. When the modal damping ratio of these two modes is 0.03, the corresponding Rayleigh damping coefficients are \(\alpha = 0.6285\) and \(\beta = 0.0009165\). Let the trolley parameters in the single-degree-of-freedom model be \(k = 5.5\times10^7\) N / m, \(M\) V \(= 75\) kg. Therefore, the natural frequency of the trolley is \(f\) V \(= 136.29\) Hz; Let the radius of the trolley wheel be 0.125 m, and the damping ratio \(\mu\) V \(= 0.01\); The excitation force parameters are set as the excitation frequency \(f_0 = 141\) Hz, the duty cycle of the knocking force \(\eta = 0.5\), and the amplitude \(A = 2Mg\).

[0128] Considering the case of sudden stiffness change, if the distribution form of the natural frequency of the trolley at the position of sudden stiffness change is a beta function:

[0129]

[0130] Take \(p = 0.99995\), and the curve form of \(x = [0, 5\times10^{-4}]\) describes the change trend of the trolley frequency, but still calculate the change in the natural frequency caused by the beam stiffness change according to equation (23). If we denote \(k\) ed \(= k\) d|max(abs(kd / k-1)) , then in the case of known damage section parameters, \(k_{ed}\) can be calculated, and according to symmetry, the distribution of the natural frequency and its derivative with respect to time can be set as (see Figure 10 ):

[0131]

[0132] When the length of the stiffness change section is 0.5 m, the position parameter is α = 0.5, and the maximum value of the stiffness change of the trolley caused is k ed = 1.01k. The stiffness distribution of the trolley is as above, and the other parameters remain unchanged. When the excitation frequencies are 141 Hz and 131 Hz respectively, ω V and the influence of its derivatives on the acceleration is shown in Fig. 11. Among them, d 2 ω V / dt 2 has the greatest influence on the amplitude, but the smallest influence range (as shown by the arrow position in Fig. 11), and ω V has a smaller influence on the amplitude but the largest influence range. The influence range is mainly determined by Figure 10 the width in the time domain where each quantity in has a large mutation value, that is, this influence only exists within the length range of the stiffness change section. Therefore, the position where the stiffness change occurs can be located by analyzing the change in the amplitude of the trolley acceleration.

[0133] Because N << 1 and ND << 1, it can be seen from equations (23) to (25) that: ω Vd ≈ ω V (1 - ND / 2), Thus, it can be known that the highest power term of N in equations (27) and (29) is quadratic. Combining N = θ - 1, the approximate relationship between the trolley acceleration and the beam stiffness change coefficient can be obtained:

[0134]

[0135] In actual detection, the noise in the low-frequency band has a greater interference on the time-domain signal. Therefore, the STFT processing method is usually used to analyze the signals in the higher sensitive frequency band. According to dimensional analysis, the equivalent acceleration can be denoted as where PSD max is the maximum value of the power spectral density in the frequency domain at each moment. Then Y max is related to the window function used in the STFT analysis. Therefore, the variable θ in equation (32) also needs to be added with the corresponding window according to the actual analysis. At this time, the weighted stiffness value within the window function width corresponds one-to-one with Y max . If the bending stiffness distribution of the beam is as shown in Fig. 12(a), and the bending stiffness at 5 m is denoted as the reference value (EI) ref , when a 1 m Hamming window is taken, the weighted stiffness change is as shown in Fig. 12(b). Since it is impossible to give a reference value to the actual beam, the weighted stiffness of the previous unit is used as the reference value (EI) w1 in the following text, and the weighted stiffness of the current unit is (EI) w2 , and let θ w =(EI) w2 / (EI)w1 , then for different window lengths, the distribution of θ w is shown in Figure 13. The inflection point of θ w is approximately the most sensitive point of the damage area. When the weighted stiffness has symmetry, the weighted stiffness ratio θ w has antisymmetry. Therefore, for the sudden change in stiffness at 6 m in Figure 12(a), two antisymmetric inflection points will appear in the figure. The window length will change the position of such inflection points, mainly affecting the position of the inflection points generated when the sliding window gradually leaves the stiffness mutation point stage, such as the inflection points at 6.75 m, 6.375 m, and 6.25 m in Figure 13. For the sudden change in stiffness, the value of the inflection point position when the window length has less influence and just enters the mutation point is denoted as θ s (the inflection points at 5.75 m, 5.875 m, and 5.937 m), that is, the first inflection point is used as an index to measure the stiffness mutation at that place. For the gradually changing stiffness at about 8.5 m in Figure 12(a), the value of the inflection point (the first inflection point) that is less affected by the window length is also taken as the measurement index.

[0136] Based on the above method of taking θ s , a description of the mutation degree at the inflection point of the stiffness change can be obtained, and the influence of different mutation degrees on the acceleration amplitude and power spectral density can be further analyzed.

[0137] For Figure 9 the case where there is only one damage segment in i , the position of the stiffness change is taken as the mid-span, i.e., α = 0.5, the width of the stiffness change is 0.5 m, and the excitation force parameter is f

[0138] For the time-domain amplitude, without windowing, θ s is equivalent to θ. When θ = 1 / 1.07, the acceleration of the trolley is shown in Figure 14. If the influence of the derivative term on the acceleration amplitude is considered (the edge of the stiffness change, the data points shown in Figure 14), the peak points near the edge of the stiffness mutation can be analyzed (the region controlled by the second derivative term, which can be determined according to Figure 10 ). For the power spectrum diagram, since the width of the mutation signal is small and is smoothed after windowing, it is difficult to directly judge. Therefore, first determine the moment of the characteristic point of the time-domain signal, and then find the inflection point in the power spectrum diagram near this moment.

[0139] For the characteristic points within the influence region of the second derivative, the relationship between the acceleration amplitude and θ, and the relationship between the equivalent acceleration Y max and θs are shown in Figure 15. The solid line is the curve fitted by the quadratic function relationship based on the data points near θ = 1, and the correlation coefficient R 2= 0.9438. When θ > 1, the stiffness of the damaged section increases, and the second derivative within the affected area changes sign. At this time, the non - monotonic relationship between the amplitude of the characteristic point and θ, that is, the inflection point of the approximate quadratic curve, is in the right - hand plane. This is because the term containing the second derivative affects the amplitude of the acceleration at different times in different directions, resulting in the conversion of the characteristic point. As shown in Figure 16, according to Figure 10 it can be seen that the second - derivative term comes into play after 8.425 s. When θ gradually increases, the acceleration of the peak point at 8.431 s gradually decreases, while for the moment of 8.428 s, the acceleration gradually increases. Therefore, there will be an exchange of peaks and valleys within the area affected by the second - derivative term. And there are points during the exchange process where the peak value is less than the peak value when θ = 1. Therefore, the quadratic curve is not symmetric about θ = 1. For Figure 15, it can be found that the symmetry of the data points is weak, but there is still a quadratic - curve relationship within the range of θs ∈ [0.99, 1.06], and the correlation coefficient R 2 = 0.7668.

[0140] Based on the above principle, it can be seen that the degree of stiffness mutation at the stiffness mutation point on the beam is approximately in a quadratic - curve relationship with the square root of the amplitude of the trolley acceleration or the power spectral density. The key lies in finding the characteristic point corresponding to the derivative term. For an actual bridge, the acceleration is easily affected by low - frequency noise. Therefore, the power spectral density can be used for analysis. To distinguish different degrees of stiffness mutation, the following algorithm is proposed for classification:

[0141] 1) Use the short - time Fourier transform (STFT) to obtain n signals, adopt a Hamming window with a window length of 1 m, and the data overlap rate of adjacent windows is 0.875;

[0142] 2) Take the vectors Yi, i = 1, 2, …, n at the sensitive frequency band (K points near the sensitive frequency, usually ±5 Hz), sum up each vector to obtain the total energy E i at this moment, as well as the mean value and standard deviation at all moments, and calculate the mean value of the vector at this moment and the standard deviation Furthermore, obtain the coefficient of variation Q i = σ i / μ i and the mean value of the coefficient of variation at all moments and the standard deviation The coefficient of variation is used to measure the fluctuation of the spectrum at this point. If the fluctuation is large, there may be a stiffness mutation;

[0143] 3) Record the position - point vector loc s after removing noise according to the filtering algorithm:

[0144] loc s = {i|E i > μE + B × σ E ||Q i > μ Q + B × σ Q , i = 1, 2, ..., n} (33)

[0145] 4) Calculate the equivalent acceleration Obtain a two-dimensional curve, select the maximum point, and count its position into the vector loc peak ;

[0146] Then the positions with obvious stiffness mutations are loc f = loc s ∩ loc peak

[0147] 5) For the case where θ is near 1, if the equivalent acceleration is expressed dimensionless and combined with Equation (32), then

[0148]

[0149] where the dimensionless coefficients G and H are calibrated by the test beam. Therefore, for the stiffness mutation positions, the mutation degree can be determined by the following formula:

[0150]

[0151] Example

[0152] The test specimen is a 12.16 m long T-beam simply supported at both ends. As shown in Figure 17, the beam has a middle partition, and the structures on both sides are symmetric. After a 2.55 m long equal-thickness web, there is a 2.15 m long variable-thickness web, and finally a 0.85 m long equal-thickness web connected to the end partition. Assuming that the concrete elastic modulus everywhere on the beam is E = 43.698 GPa, Figure 12 shows the distribution of the equivalent bending stiffness of the test beam cross-section.

[0153] As Figure 3 , the rear part of the inspection vehicle weighs 75 kg and has a single rear-wheel structure, including an excitation wheel, a circular shaft, and a counterweight, etc. The excitation wheel has 72 teeth and is composed of rubber and aluminum alloy. The circular shaft and the counterweight are both made of stainless steel. The acceleration acquisition point is on the circular shaft, 20 mm away from the midpoint of the rear axle. The natural frequency of the inspection vehicle is about 138 Hz.

[0154] During the test, the inspection vehicle travels on the test beam from west to east at a speed of about 1.5 m / s, and the frequency of the excitation force generated is 138 Hz. A total of 20 sets of effective data are recorded. The axle acceleration signal and power spectral density of the first time are shown in Figure 19. It can be seen that the average acceleration of the vehicle is about 0.5g.

[0155] In Fig. 19(a), the marked point indicates the moment when the trolley passes through the expansion joint. On the right side is the position of the expansion joint from the starting point of the trolley. The difference between the two coordinates is approximately 12.1 m, which is consistent with the length of the beam. Therefore, the position of the trolley on the bridge can be located based on the first expansion joint, and the abscissas of the subsequent results are all based on the first expansion joint. According to the algorithm described in the principle part above, when B is taken as 1, the processing result as shown in Fig. 20(a) can be obtained, where the line is Y max , and the circled points are the abnormal points of stiffness mutation obtained after filtering. This result can be compared with the marked points in Fig. 12(b). The statistical values of each position point are shown in Table 1. It can be seen that the detection equipment and method adopted by the present invention can find the stiffness mutation points corresponding to positions 1, 2, 3, and 5 of the test beam with a success rate of 90%, and find the stiffness mutation point corresponding to position 4 with a success rate of 60%.

[0156] Table 1 Detection results when the excitation frequency is 138 Hz

[0157]

[0158] As Figure 21 is the distribution of the equivalent acceleration with the stiffness mutation degree θ s as the abscissa at each position in Table 1, where the data points are the mean and standard deviation of the equivalent acceleration corresponding to the five positions. This distribution form is similar to the distribution in Fig. 15. The data points (positions 1, 3, 4, and 5) are relatively concentrated near θ s = 1, and Y max shows an approximately monotonically decreasing trend. However, when θ s is relatively large (position 2, θ s = 1.068), Y max is relatively large. Therefore, the relationship between Y max and θ s can generally be described by a quadratic function.

[0159] As shown in Figure 22, it is the modal analysis of the rear axle under specific geometric parameters. The boundary conditions are as follows: the contact part between the excitation wheel and the ground is a fixed constraint, and the horizontal movement is restricted at the front end face of the square groove. Among them, the main vibrating elements in Figure 22(c) are the wheel axle and the excitation wheel, and it is a vertical vibration mode with a frequency of 117 Hz, meeting the requirements of the detection method. The modes in Figure 22(a), (b), and (d) are the rotational vibrations of the excitation wheel around its fixed point. This type of mode is mainly related to the shear stiffness of the excitation wheel and the contact condition between the excitation wheel and the ground. Since the rotation of the excitation wheel continuously changes the contact condition, this type of mode is not stable. In addition, in this case, the vibration direction of the signal measurement point is mainly horizontal, with less interference in the vertical direction. The mode in Figure 22(e) is a mode of coupled bending and torsion of the overall structure, with a frequency of 339 Hz, which is much higher than the frequency of the mode in Figure 22(c). When the operating frequency is around 140 Hz, this mode will not interfere with the signal.

[0160] As shown in Figure 23, it is the modal analysis of the vehicle body including the front part. The constraint condition is that the contact parts of each wheel with the ground are fixed constraints. Among them, the mode in Figure 23(e) is a vertical vibration mode that meets the detection requirements, and the mode in Figure 23(d) similar to it is the rotational mode of the excitation wheel, and the interference to the detection is reduced as described above. The frequency of the mode in Figure 23(f) is 152 Hz, with a difference of 51 Hz from the frequency of the mode in Figure 23(e). When the excitation frequency is near the frequency of the mode in Figure 23(e), this mode will not be excited, so it will not interfere with the detection signal.

[0161] The above examples of the present invention have been described in detail in combination with the embodiments. However, the present invention is not limited to the above examples. Within the scope of knowledge possessed by those of ordinary skill in the art, various changes made without departing from the purpose of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A passive excitation type bridge flaw detection device, comprising: An excitation part, a damage location and evaluation part, and a counterweight connection part, characterized in that, The excitation part includes an excitation wheel with uniformly distributed excitation teeth on its surface, which is used for passive excitation of the bridge deck at a fixed excitation frequency. Fixed parts with rectangular end faces are fixed at both ends of the central bearing of the excitation wheel, and the fixed parts are fixedly connected to the end slots of the approach slab through bolts. An acceleration sensor is arranged on the wheel shaft to collect the acceleration signal transmitted from the bridge surface to the excitation teeth; The damage location and evaluation part determines the moment when the stiffness mutation is detected by the maximum point of the equivalent acceleration extracted from the time-frequency analysis result, and determines the bridge damage location and evaluates the damage degree corresponding to the inflection point in the power spectrum diagram; The counterweight connection part includes a square groove member and an approach slab. The square groove member connects the front body drive part and the rear body excitation part through a slot and bolts; the approach slab is welded and fixed on the square groove member, and includes a top character connection part and a support arm for enhancing the torsional stiffness, and a bottom square steel connection part for enhancing the bending stiffness.

2. The passive excitation type bridge flaw detection device according to claim 1, wherein The damage location and evaluation part filters the acceleration signal as follows: First, perform time-frequency analysis on the acceleration signal to obtain n signals. Each signal corresponds to a specific moment and has a corresponding spectral vector Y i , i = 1, 2, …, n. First, sum up each vector to obtain the total energy E at this moment i , and then calculate the mean value of the energies at all moments and the standard deviation Calculate the mean value of the spectral vector at the i-th moment and the standard deviation Furthermore, obtain the coefficient of variation Q that measures the fluctuation of the spectrum at this moment i = σ i / μ i , and the mean value of the coefficients of variation at all moments and the standard deviation The fluctuation of the coefficient of variation represents the sudden change in stiffness caused by damage.

3. The passive excitation type bridge flaw detection device according to claim 2, wherein, The damage location step of the damage location and evaluation part is as follows: First, record the position point vector loc after removing noise according to the filtering result s : loc s = {i | E i > μ E + B × σ E || Q i > μ Q + B × σ Q , i = 1, 2,..., n}; where B is the noise interference degree; Then calculate the equivalent acceleration on the time-frequency diagram of the curve varying with time, select the maximum points and count their positions into the vector loc peak , where PSD max is the maximum value of the power spectral density in the frequency domain at each moment, and the damage location with an obvious stiffness mutation is loc f = loc s ∩ loc peak .

4. The passive excitation type bridge flaw detection device according to claim 3, characterized in that, The evaluation of the damage degree by the damage location and evaluation part is determined by the following formula: where G and H are dimensionless coefficients calibrated in advance.

Images