Non-destructive testing equipment, non-destructive testing system, and non-destructive testing method

Abstract

This invention provides a technology to prevent a decrease in inspection accuracy caused by unintended elastic waves in non-destructive testing using the impact elastic wave method. [Solution] A non-destructive testing device (10) that performs non-destructive testing of a structure 6 including concrete by impact elastic wave method comprises a vibration source (2) that generates an input wave, a vibration sensor (3) that detects an output wave, a machine (4) on which the vibration sensor (3) is installed, and a switching unit (5) that switches between a first state in which the machine (4) holds the vibration source (2) and the vibration sensor (3), and a second state in which the vibration source (2) is installed on the surface (6S) of the structure (6) and detached from the machine (4), and the machine (4) holds the vibration sensor (3) and presses the vibration sensor (3) against the surface (6S).

Description

[Technical Field]

[0001] The present invention relates to a non-destructive testing apparatus, a non-destructive testing system, and a non-destructive testing method for structures including concrete. [Background technology]

[0002] As non-destructive testing methods for evaluating the soundness of structures containing concrete, the impact elastic wave method (see, for example, Non-Patent Document 1) and the local vibration test method that applies the impact elastic wave method (see, for example, Non-Patent Document 2) are known.

[0003] In the impact elastic wave method, the elastic wave (also called an impact elastic wave) generated by impacting the surface of a structure with a collision element (such as a steel ball or electromagnetic hammer) is used as the input wave, and the elastic wave that propagates inside the structure and undergoes multiple reflections according to the internal structure is used as the output wave. In the impact elastic wave method, it is common to detect the output wave as a spectral response function using a vibration sensor and perform inspections based on the resonance frequency of the spectral response function.

[0004] Furthermore, in the local vibration test method, which can be considered a form of the impact elastic wave method, for example, an exciter equipped with an electrodynamic excitation coil is used instead of a collision element, and the elastic wave with a wide frequency band generated by the exciter is used as the input wave.

[0005] In the following, the impact elastic wave method is used as a concept that includes local vibration testing methods, and the impactor and exciter are collectively referred to as vibration sources.

[0006] The worker performing the impact elastic wave method carries a complete set of inspection tools, including a vibration source and vibration sensors, and moves sequentially to several predetermined inspection locations, performing inspections at each location. Alternatively, a self-propelled robot could be used to perform the impact elastic wave method on behalf of the worker (for example, Patent Documents 1 and 2). In the case of a self-propelled robot, the vibration source and vibration sensors are fixed to the robot body. After moving to the inspection location, the self-propelled robot presses the vibration source and vibration sensors, respectively, against the surface of the structure at the inspection location and performs the inspection. [Prior art documents] [Non-patent literature]

[0007] [Non-Patent Document 1] Concrete Diagnostic Technology '15, p. 139. [Non-Patent Document 2] Journal of Japan Society of Civil Engineers, Series E2 (Materials and Concrete Structures), Vol. 67, No. 4, 522-534, 2011. [Patent Documents]

[0008] [Patent Document 1] Japanese Patent Publication No. 2020-106440 [Patent Document 2] Japanese Patent Publication No. 2021-85860 [Overview of the project] [Problems that the invention aims to solve]

[0009] However, in the self-propelled robot described above, since the vibration source and vibration sensor are fixed to the robot body, the vibration sensor detects not only the desired elastic waves that have propagated through the structure and undergone multiple reflections according to the internal structure, but also unintended elastic waves that propagate through the robot body as part of the propagation path. In this way, the superposition of unintended elastic waves with the desired elastic waves in the output wave reduces the accuracy of non-destructive testing.

[0010] One aspect of the present invention has been made in view of the above-mentioned problems, and its purpose is to prevent a decrease in inspection accuracy caused by unintended elastic waves in non-destructive testing using the impact elastic wave method. [Means for solving the problem]

[0011] To solve the above problems, a non-destructive testing apparatus according to one aspect of the present invention is a non-destructive testing apparatus that performs non-destructive testing of a structure including concrete by the impact elastic wave method, comprising: a vibration source that generates an input wave; a vibration sensor that detects an output wave; a machine body on which the vibration sensor is provided; and a switching unit that switches between (1) a first state in which the machine body holds the vibration source and the vibration sensor, and (2) a second state in which the vibration source is installed on the surface of the structure and detached from the machine body, and the machine body holds the vibration sensor and presses the vibration sensor against the surface.

[0012] Furthermore, a non-destructive testing system according to one aspect of the present invention is a non-destructive testing system comprising the above-mentioned non-destructive testing apparatus, an information processing apparatus including an estimation unit, and one or more terminals, wherein the information processing apparatus further comprises a communication unit that communicates the degree of abnormality of the structure estimated by the estimation unit to the one or more terminals via a communication line.

[0013] Furthermore, a non-destructive testing method according to one aspect of the present invention is a non-destructive testing method for performing non-destructive testing of a structure including concrete by impact elastic wave method, and includes a machine body that holds a vibration source and a vibration sensor, a separation step of installing the vibration source on the surface of the structure at the measurement position and separating the vibration source from the machine body, and an inspection step of applying an input wave to the structure using the vibration source and detecting an output wave using the vibration sensor. [Effects of the Invention]

[0014] According to one aspect of the present invention, in non-destructive testing using the impact elastic wave method, it is possible to prevent a decrease in inspection accuracy caused by unintended elastic waves. [Brief explanation of the drawing]

[0015] [Figure 1] a is a diagram showing the configuration of a non-destructive inspection device or an autonomous mobile robot according to an embodiment. b is a partially enlarged view showing a vibration source and a switching unit in a state where the vibration source is detached from the device. [Figure 2] It is a diagram showing an example of a method for measuring the vibration of a structure according to an embodiment. [Figure 3] It is a diagram showing an example of a spectral response function according to an embodiment. [Figure 4] It is a flow chart showing an inspection method of a non-destructive inspection device according to an embodiment. [Figure 5] It is a partial perspective view showing a vibration source and a switching unit of a non-destructive inspection device according to an embodiment. [Figure 6] It is a schematic top view showing a vibration source, a support unit, and a switching unit of a non-destructive inspection device according to an embodiment. [Figure 7] It is a partial cross-sectional view showing an embodiment of a non-destructive inspection device provided with a plurality of vibration sources. [Figure 8] It is a schematic cross-sectional view showing the vibration sensor 3 in three modes. [Figure 9] It is a block diagram showing the hardware configuration of a non-destructive inspection device and a cloud server. [Figure 10] It is a graph comparing the spectral data of an autonomous mobile robot, a conventional robot, and manual measurement. [Figure 11] It is a block diagram showing the hardware configuration of a cloud server. [Figure 12] It is a diagram showing the configuration of a non-destructive inspection system according to an embodiment. [Figure 13] It is a flowchart showing an example of a flow of a method for generating a machine learning model and an example of a flow of a method for estimating the degree of abnormality according to an embodiment. [Figure 14] It is a diagram for explaining a method for calculating the degree of abnormality according to an embodiment. [Figure 15] It is a diagram showing an example of an update process according to an embodiment. [Modes for carrying out the invention]

[0016] [Embodiment] <Basic Configuration of Non-Destructive Testing Equipment> One embodiment of the present invention will be described below. Figure 1a is a diagram showing the configuration of a non-destructive testing device or autonomous mobile robot according to the embodiment. Figure 1b is a partially enlarged view showing the vibration source and switching unit with the vibration source separated from the device. The non-destructive testing device 10 is a device for estimating the state of a structure. The structure is constructed of, for example, hydrated solidified bodies such as concrete and mortar, inorganic solidified bodies such as geopolymers, hydrates such as cement, wood, or combinations thereof, and more specifically, for example, RC (Reinforced Concrete) structures and SRC (Steel Reinforced Concrete) structures. Examples of structures include, but are not limited to, road decks, bridge foundations, tunnels, and port facilities such as piers. As an example, the non-destructive testing device 10 is equipped with a general-purpose computer for performing non-destructive testing of a structure 6 containing concrete by the impact elastic wave method.

[0017] The condition of a structure can be described by factors such as the presence or absence of abnormalities or the degree of abnormality (i.e., the degree of abnormality). Examples of structural abnormalities include deterioration (aging) or damage due to external forces (live loads such as earthquakes and wind). Examples of structural deterioration include wear, fatigue, salt damage, freeze-thaw damage, carbonation, ASR, and fire. More specifically, this can include spalling of the concrete cover due to rebar corrosion, or further deterioration leading to subsidence. By detecting the degree of abnormality in a structure before deterioration progresses, preventive maintenance measures can be taken.

[0018] As shown in Figure 1, the non-destructive testing apparatus 10 comprises a vibration source 2 that generates an input wave, a vibration sensor 3 that detects an output wave, a machine 4 on which the vibration sensor 3 is mounted, and a switching unit 5 that switches between (1) a first state in which the machine 4 holds the vibration source 2 and the vibration sensor 3, and (2) a second state in which the vibration source 2 is installed on the surface 6S of the structure 6 and detached from the machine 4, and the machine 4 holds the vibration sensor 3 and presses the vibration sensor 3 against the surface 6S. In Figure 1, the two states that can be switched using the switching unit 5 are referred to as the first state (1) and the second state (2), respectively. The non-destructive testing apparatus 10 is an example of an abnormality estimation apparatus according to the present disclosure. The vibration source 2 is, for example, an exciter that excites vibrations for inspection onto the structure 6. The vibration sensor 3 is an acceleration sensor, a magnetostrictive sensor, or a laser, and the acceleration sensor is, for example, a highly sensitive piezoelectric element that outputs sensing data representing acceleration for vibration detection. Here, the vibration sensor 3 may be detached from the aircraft body 4, and the vibration source 2 may be pressed against the surface 6S, in the reverse relationship.

[0019] The amplitude of the response wave from the vibration sensor 3 becomes less than 1 / 10 of the amplitude of the input wave from the vibration source 2 due to attenuation within the structure 6, for example. Therefore, even a slight vibration transmitted via the autonomous mobile robot 1 can have a significant impact on the vibration sensor 3. While it is possible to generate white noise that takes into account the vibration characteristics of the non-destructive testing equipment or the autonomous mobile robot, it is difficult to completely eliminate the influence of vibrations from the equipment or robot itself. Furthermore, the weight balance and load capacity of the equipment or robot itself change constantly due to working conditions and improvements to the robot, making this approach impractical.

[0020] If the vibration source 2 is not disconnected from the machine body 4 during inspection, it may negatively affect the input wave itself. For example, when using white noise as the input wave, if the vibration source 2 is fixed to the machine body 4, its flatness will be lost. In the second state, the vibration source 2 installed on the surface 6S of the structure 6 is disconnected from the machine body 4. Therefore, it is possible to prevent unintended shock elastic waves from being superimposed on the input wave (i.e., output wave) (e.g., from the machine body 4), thus preventing a decrease in inspection accuracy caused by unintended shock elastic waves. As a secondary effect, since the vibration source 2 is pressed against the surface using its own weight, the load applied when pressing the vibration source 2 against the surface 6S can be brought closer to a constant value.

[0021] (Method for measuring vibrations) The "elastic wave with a wide frequency range" input to vibration source 2 can be white noise or sweep waves. White noise is a noise wave created by randomly combining waves with a wide frequency range and the same energy, and can be input for a short time (about 1 second). On the other hand, sweep is an input wave with a relatively long duration (several tens of seconds) created by sequentially combining waves in the time direction from low frequency to high frequency. In this embodiment, as an example, white noise is used for thin members such as floor slabs, and sweep is used for thick members such as beams to ensure that sufficient vibration energy is transmitted (white noise does not transmit vibration energy).

[0022] Figure 2 shows an example of a method for measuring the vibration of a structure 6 using a vibration source 2. In the example in Figure 2, a vibrator is installed on the surface 6a of the structure 6 as the vibration source 2, and the vibrator generates, for example, white noise to excite local vibrations of the structure 6. During excitation, a vibration sensor 3 is brought into contact with the surface 6a of the structure 6 to measure the response acceleration. Vibration measurements are taken at multiple locations on the surface 6a of the structure 6 (for example, at 300 mm intervals), and data is collected.

[0023] If steel balls or hammers are struck too hard, surface vibrations will occur rather than longitudinal vibrations. Also, with steel balls, the size (outer diameter) of the steel balls must be selected according to the thickness of the structure 6. Therefore, multiple steel balls must be prepared in advance. In contrast, a vibration exciter can generate a wide range of frequencies in advance, so it is advantageous because only one unit needs to be prepared.

[0024] Here, we will explain the difference in vibration measurement results between the case where there is no internal damage to structure 6 and the case where there is, referring to Figure 2. When there is no internal damage (cracks, etc.) to structure 6, most of the elastic waves generated by vibration source 2 (exciter) are reflected at the bottom surface 6b. On the other hand, when there is internal damage (cracks, etc.) to structure 6, of the elastic waves generated by vibration source 2 (exciter), high-frequency waves are reflected at the crack surface 6c, while low-frequency waves are diffracted through the crack.

[0025] In this way, by utilizing the existence of elastic waves (high-frequency waves) reflected at the internal crack surface 4c and elastic waves (low-frequency waves) diffracting through the cracks, the state of the structure 6 can be evaluated by performing a fast Fourier transform on these response accelerations to obtain differences in the spectral response function (a response function that shows the response to vibration at each frequency), thereby capturing the change in the resonant frequency, which is the peak frequency. This evaluation method is also called the local vibration test method, as mentioned above. However, the local vibration test method assumes that the spectral response function in a sound state is known. Here, a sound state refers to a state in which no abnormalities such as cracks have occurred in the structure 6.

[0026] If the spectral response function of a healthy structure is unknown, it is possible to assess internal damage based on changes in the resonant frequency (the specific frequency at which an object vibrates naturally). Figure 3 shows an example of the spectral response function of structure 6. In Figure 3, spectral response function w11 is the spectral response function of structure 6 in a healthy state, and spectral response function w12 is the spectral response function of damaged structure 6. In spectral response functions w11 and w12, the horizontal axis represents frequency (Hz), and the vertical axis represents the amplitude spectrum normalized to the maximum value. The resonant frequency of spectral response function w12 of the damaged structure is lower than the resonant frequency of spectral response function w11 in a healthy state.

[0027] When evaluating internal damage based on changes in resonant frequency, the resonant frequency in a healthy state is calculated, for example, by the following theoretical formula (1), assuming an apparent speed of sound c.

number

[0028] <Inspection methods for non-destructive testing equipment> Figure 4 is a flowchart showing the inspection method of the non-destructive testing apparatus 10 according to this embodiment. This inspection method M10 is a non-destructive testing method that performs non-destructive testing of a structure 6 including concrete by impact elastic wave method, and uses a machine body 4 that holds a vibration source 2 and a vibration sensor 3. The inspection method M10 includes a separation step, an inspection step, a moving step, and a holding step.

[0029] In step S2, the detachment process involves placing the vibration source 2 on the surface 6S of the structure 6 at the measurement position and detaching the vibration source 2 from the machine body 4. At this point, the vibration sensor 3 is also placed on the surface 6S of the structure 6 while being pressed with a predetermined force accompanying the operation of the switching unit 5.

[0030] In step S3, the inspection process applies an input wave to the structure 6 using the vibration source 2 and detects the output wave using the vibration sensor 3.

[0031] The moving process is a moving process (step S1) performed before the detachment process S2 and the inspection process S3, in which the machine body 4, holding the vibration source 2 and vibration sensor 3, is moved to the measurement position. The holding process is a holding process (S4) performed after the detachment process S2 and the inspection process S3, in which the machine body 4 holds the vibration source 2.

[0032] <Mechanism> The mechanism of a non-destructive testing device (autonomous mobile robot) according to one embodiment of the present invention will be described. Figure 1 is a diagram showing an exemplary mechanism of the non-destructive testing device 10 (autonomous mobile robot 1) according to this embodiment. The vibration source 2, vibration sensor 3, and switching unit 5 are arranged considering the weight balance of the body 4. As an example, the body (i.e., vehicle) 4 is assembled with a vehicle frame with a four-wheel structure in which the rear wheels have a larger outer diameter than the front wheels. This body 4 is configured as a gear-motor rear-wheel drive mobile unit (mobility mechanism) 4M with omniwheels at the front and rubber tires at the rear. In this case, the direction of movement is steered by the rotational difference of the rear wheels. Thus, the autonomous mobile robot according to this embodiment employs driving by transmitting power to the road surface via wheels as a mode of movement. However, the mode of movement in the autonomous mobile robot according to this embodiment is not limited to driving, and may be, for example, flying or walking.

[0033] A personal computer 9 is mounted in the center of the body 4 or above the front wheels, and the switching unit 5, which is a lifting mechanism, is mounted on the body frame on the front wheel side, positioned between the front and rear wheels. The waterproof box 8, described later, is mounted on the body frame near the rear wheel axle to lower the center of gravity of the non-destructive testing device 10. As a result, the autonomous mobile robot 1 can travel until the slope of the surface 6S of the structure 6 reaches approximately 7.0%. To further improve driving performance, four-wheel drive may be implemented. Alternatively, an off-road specification including springs and dampers may be implemented to further improve driving performance and reduce vibrations transmitted to the body 4 during inspection.

[0034] (Switching section) The mechanism for holding / detaching the vibration source 2 by the switching unit 5 will be explained with reference to Figure 5. Figure 5 is a partial perspective view showing the vibration source 2 and the switching unit 5 of the non-destructive testing apparatus 10 according to this embodiment.

[0035] The switching unit 5 includes a support unit 5S that moves between a first position with a higher height and a second position with a lower height. In the first position, the support unit 5S supports a portion 2P of the vibration source 2 from below and separates the vibration source 2 from the structure 6 to achieve a first state (1) (see Figure 1a). In the second position, the support unit 5S places the vibration source 2 on the surface 6S and separates it from a portion 2P of the vibration source 2 to achieve a second state (2) (see Figure 1b). This allows the non-destructive testing device 10 to reliably and easily switch between the first and second states.

[0036] In Figure 5, the switching unit 5 raises and lowers the vibration source 2 vertically, but it is not limited to this; it may also be able to move horizontally, for example, like the reach lift function of a forklift. The support unit 5S is, for example, screwed to the switching unit 5.

[0037] The vibration source 2 is provided with a pair of protrusions 2P, 2P that protrude from each of the pair of sides 2S, 2S at a position higher than the center of gravity, and the support part 5S is a pair of members 5B, 5B whose spacing is wider than the spacing between the pair of sides 2S of the vibration source 2 and narrower than the spacing between the tips of the pair of protrusions 2P, 2P. Here, the pair of members 5B, 5B are plate-shaped, but are not limited to this and may be any suitable shape such as rod-shaped (fork-shaped), tubular, or sheet-shaped. This makes it possible to realize a configuration that can switch between the first state (1) and the second state (2) at low cost.

[0038] As shown in Figure 5, each of the pair of members 5B, 5B has the same shape and a first recess 5V, 5V that opens more upward when the main surfaces 5M, 5M are viewed from above. Here, the first recesses 5V, 5V are V-shaped notches in the figure, but are not limited to this and may be U-shaped, trapezoidal notches, or vertically extending key parts (rod parts) (for insertion into the keyhole).

[0039] Each of the pair of protrusions 2P, 2P of the vibration source 2 has a convex portion that engages with the first recesses 5V, 5V. Each of these convex portions (each of the pair of protrusions) has the same shape and a second recess 2V, 2V that opens more downward when viewed from a direction parallel to the main surfaces 5M, 5M of the pair of members 5B, 5B.

[0040] In this embodiment, the second recesses 2V, 2V are inverted V-shapes. Hereafter, the second recesses 2V, 2V will also be referred to as inverted V-shaped structures 2V, 2V. However, this means that the second recesses 2V, 2V are shaped to engage with the first recesses 5V, 5V to position the vibration source 2, so they may be U-shaped, trapezoidal, or shapes corresponding to key parts (for example, an inverted U-shape with the same curvature, a trapezoidal (bobbin) shape with the same engagement size, or a keyhole for a key part). In other words, the support part 5S of the switching part 5 is a forklift structure, and the second recesses 2V, 2V are arbitrarily formed so as to engage with and be held by the shapes 5V, 5V of the support part 5S.

[0041] As a result, even if the surface 6S of the structure 6 is not a horizontal, smooth surface (in other words, if the vibration source 2 and the support part 5S are in a positional relationship that is offset vertically and horizontally in the approximate top view of Figure 6), when transitioning from the second state (2) to the first state (1), the inverted V-shaped structures 2V, 2V of the pair of protrusions 2P, 2P slide and engage with the V-shaped structures 5V, 5V of the pair of plate-like members 5B, 5B in a two-dimensional plane, thereby reliably holding the vibration source 2 (self-centering). Furthermore, even if the machine body 4 moves toward the inspection position in the first state (1) and there is some tilting (for example, a slope with a gradient of 7.0%) or shaking (for example, collision with a pebble) of the surface 6S of the structure 6, the vibration source 2 can be reliably held.

[0042] Furthermore, during measurement, the V-shaped structures 5V, 5V function as stoppers for the inverted V-shaped structures 2V, 2V, preventing the vibration source 2 from tipping over due to strong winds or collisions with obstacles. The support section 5S, for example, when viewed from above (from a direction parallel to the main surfaces 5M, 5M of the pair of members 5B, 5B), is U-shaped (right-angled U-shape) and includes the pair of plate-like members 5B, 5B.

[0043] The switching unit 5 may move (or raise / lower) the vibration source 2 using a gripping mechanism such as a robot hand or crane (catch claw, UFO catcher®), or a container lifting mechanism, or an upper / lower keyhole mechanism (insert type), instead of a lifting support mechanism.

[0044] (Method for generating vibrations) The vibration source 2 is a vibrator comprising an excitation coil, a housing 2H that houses the excitation coil, the housing 2H having a lower bottom surface 2B that faces and is close to the surface 6S when the vibration source 2 is installed on the surface 6S, a vibration transmitter 2T connected to the excitation coil with its tip protruding from the lower bottom surface 2B, and one or more legs 2F that allow the housing 2H to stand upright when the vibration source 2 is installed on the surface 6S.

[0045] Each of these one or more legs 2F is made of an elastic material and is configured such that (1) when the tip is separated from the surface 6S, the lower end is positioned below the tip, and (2) when the housing 2H is placed on the surface 6S and the tip is in contact with the surface 6S, the lower end is in contact with the surface 6S together with the tip.

[0046] This allows the leg portion 2F to be used as a cushion, keeping the load pressing the tip of the vibration transmitter 2T against the surface 6S constant while stabilizing the posture of the housing 2H installed on the surface 6S. In addition, it also enables the suppression of reflected waves. By providing the leg portion 2F, the vibration source 2 can make contact with road surfaces 6S with a gradient of up to 7.5%. The leg portion (elastic body) 2F can be made of natural rubber, silicone rubber, springs, etc. The leg portion 2F also serves to provide a vertical movement stroke for the vibration transmitter 2T of the vibration source 2.

[0047] Figure 7 is a partial side view showing one embodiment of a non-destructive testing apparatus equipped with multiple vibration sources 2. As shown in this figure, multiple vibration sources 2 may be mounted in a single row side by side on the switching unit 5. For example, if the weight of a vibration source 2 is about 2.0 kg and the lifting capacity of the switching unit 5 is about 30 kg, it is preferable to have three vibration sources 2, considering the workability of the robot. The number of vibration sources 2 may be selected according to the width and length of the structure 6, and the pitch distance between each vibration source 2 may be adjustable.

[0048] Alternatively, although not shown in the diagram, a single vibration source 2 may be attached to the switching unit 5 so that it can move by an actuator in the left-right direction (a direction perpendicular to the main surfaces 5M, 5M of the pair of plate-shaped members 5B, 5B) as shown in Figure 7. This allows the autonomous mobile robot 1 (non-destructive testing device 10) to perform inspections of multiple side lines (within its range of motion) without moving itself, thereby improving the inspection movement efficiency of the autonomous mobile robot 1.

[0049] Furthermore, the autonomous mobile robot 1 (non-destructive testing device 10) may be further equipped with one or more cameras (image sensors) to capture the condition of the surrounding surface (road surface) 6S. Using these cameras, for example, an optimal position with fewer pebbles or sand may be determined, and the vibration source 2 may be moved to that optimal position and grounded by driving the autonomous mobile robot 1 (non-destructive testing device 10) or the actuator described above. This can improve the inspection accuracy.

[0050] (Vibration sensor) The vibration sensor 3 is fixed directly or indirectly to the support part 5S. In the first state (when the machine body 4 is moving), the vibration sensor 3 is attached to the surface 6S of the structure 6 at a height away from the surface 6S of the structure 6, for example, by screw fastening to the surface 5S of the machine body 4 to which the support part 5S is attached. In the second state (during inspection), the vibration sensor 3 is positioned relative to the switch part 5 so as to be pressed against the surface 6S of the structure 6 with a predetermined force (detailed below). This makes it possible to press the vibration sensor 3 against the surface 6S of the structure 6 and to detach the vibration source 2 from the machine body 4 simultaneously. This screw fastening may be performed via a positioning spacer or a piston / cylinder mechanism, etc., as described below.

[0051] Figure 8 is a schematic side view showing the vibration sensor 3 in three different configurations. The vibration sensor 3 is attached to the tip of the piston rod 3R of the piston-cylinder mechanisms 3P and 3C. The left side of Figure 8 is a schematic side view when the vibration sensor 3 is in contact with the surface 6S of the structure 6. Here, the non-destructive testing device 10 further includes an elastic cushioning material 3S (spring, rubber, etc.) interposed between the vibration sensor 3 (piston 3P) and the machine body 4 (cylinder 3C).

[0052] As shown in the center diagram of Figure 8, the vibration sensor 3 is configured to be pressed against the surface 6S of the structure 6 with a predetermined force (specifically, 1.0 kgf or more and less than 2.0 kgf) by torque control of the switching unit 5, which is a lifting mechanism. This predetermined force is set so as to be able to reproduce the force with which a skilled inspection worker presses the vibration sensor 3 with their finger. Furthermore, dampers may be provided in the piston-cylinder mechanisms 3P and 3C.

[0053] This makes it possible to keep the load applied to the vibration sensor 3 against the surface 6S as close to constant as possible. Furthermore, by appropriately selecting the elastic modulus of the elastic material constituting the cushioning material 3S, it is possible to more accurately reproduce the load applied by, for example, a skilled worker when pressing the vibration sensor 3 against the surface 6S. Therefore, it is possible to achieve inspection accuracy equivalent to that when a skilled worker performs non-destructive testing (see the lower diagram in Figure 10). Here, the pressing force (a predetermined force) of the vibration sensor 3 with respect to the angle formed between the main shaft of the piston rod 3R and the surface 6S of the structure 6 may be learned, and the elastic modulus may be selected accordingly.

[0054] Furthermore, as shown in the right-hand diagram of Figure 8, the non-destructive testing device 10 may also include a ball joint 7 interposed between the vibration sensor 3 and the support part 5S (piston rod 3R), which is directly or indirectly fixed to the support part 5S, where the vibration sensor 3 is fixed to the ball joint 7.

[0055] This allows the vibration sensor 3 to be brought into close contact with the surface 6S of the structure 6, regardless of the inclination angle of the surface 6S (although the gradient is limited to 10%), thereby suppressing variations in inspection accuracy.

[0056] The above describes the mechanism of a non-destructive testing device (autonomous mobile robot) according to one embodiment of the present invention. The non-destructive testing device 10 (autonomous mobile robot 1) is further equipped with a waterproof box 8, as shown in Figure 1. The waterproof box 8 houses a battery, a data logger for the vibration sensor 3 (an observation device that digitally processes and stores signal records), and the drive circuit of the autonomous mobile robot 1. The autonomous mobile robot 1 is also equipped with a GNSS antenna (not shown) for autonomous movement by referring to pre-input coordinate values, enabling automatic inspection at multiple inspection locations.

[0057] The non-destructive testing device 10 is configured to automatically upload the spectral data obtained by the vibration sensor 3 to the cloud server 100 after performing FFT (Fast Fourier Transformation) processing on the control unit 20 or microcontroller (not shown) of the non-destructive testing device 10. Here, FFT processing refers to the Fast Fourier Transform, which can decompose a signal into multiple frequency components and represent their magnitudes as a spectrum, thereby reducing the size of the raw measurement data.

[0058] (Aircraft self-positioning) For the self-positioning of aircraft 4, GNSS standalone positioning, GNSS relative positioning, or high-precision standalone positioning CLAS will be used. These methods ((1) to (4)) are explained below.

[0059] GNSS standalone positioning ((1) Smartphone / car navigation method) does not require a reference point and performs positioning using four or more satellites with a single receiver (determining the coordinate position). The positional accuracy error is approximately a few meters to 20 meters in the horizontal direction, but measurement is not possible in the vertical direction (altitude).

[0060] In addition to standalone GNSS positioning, there is also GNSS relative positioning, which requires a reference point. Relative GNSS positioning includes (2) D-GNSS (differential) and (3) RTK-GNSS (interferometric positioning). D-GNSS performs standalone positioning (distance) from GPS satellites at both known points (DSPS stations and radio stations) and the observation point, calculates the difference in coordinate values ​​from the position information of the known points, and transmits correction information to the observation point in real time. The position accuracy error is approximately 50 cm to 5 m in the horizontal direction, and measurement is not possible in the vertical direction. RTK-GNSS sets up a known point in addition to the observation point, receives the wavenumber and phase difference of the carrier wave from GPS satellites, and corrects the position information of the observation point in real time. The position accuracy error is approximately 2.0 cm to 3.0 cm in the horizontal direction and approximately 2.0 cm to 4.0 cm in the vertical direction.

[0061] In addition to GNSS standalone positioning and GNSS relative positioning, there is also high-precision standalone positioning CLAS, which does not require a reference point. (4) The PPP method exists as a high-precision standalone positioning CLAS. This method collects position data from geographically densely located reference stations (control stations) and adds local correction information by receiving distance and phase signals transmitted from the Quasi-Zenith Satellite System "Michibiki". The position accuracy error is 5.0 cm or less in both the horizontal and vertical directions.

[0062] In this disclosure, (4) high-precision standalone positioning using the PPP system "Michibiki" is preferred, but (3) the RTK-GNSS system, which has a similar level of accuracy, may also be applied.

[0063] The above methods (1) to (4) are usable outdoors, but for indoor use, SLAM using LiDAR is applied. When considering operation on a pier (outdoors), there may actually be warehouses, containers, etc., which could cause GNSS radio interference, so a combination of GNSS satellite coordinates and a point cloud map obtained by SLAM may be used. Topographic data for inspection reference may also be measured using LiDAR.

[0064] (Aircraft direction detection) An Inertial Measurement Unit (IMU) may be used to understand, predict, and control the behavior (attitude and trajectory) of the moving object (aircraft 4). The IMU is a unit that detects three-dimensional inertial motion (translational and rotational motion in the orthogonal three-axis directions) and is equipped with three-axis angular velocity (gyro) sensors, three-axis acceleration sensors, and a temperature sensor. 2 Translational motion is detected by the IMU, and rotational motion is detected by the gyro sensor (° / s). Inertial information from the IMU may be applied for autonomous driving.

[0065] The gyro sensors mounted on the IMU generally use the following methods in order of increasing accuracy: ring laser gyroscope, optical (FOG; Fiber Optical Gyroscope), and vibration (MEMS; Micro Electro Mechanical). To determine the direction of the aircraft 4, it is preferable to use a GNSS compass using two of the aforementioned GNSSs, but a magnetic compass or a combination of the aforementioned gyro sensors may also be used.

[0066] Furthermore, if there are multiple pebbles or sand accumulated on the surface 6S of the structure 6, these must be removed before inspection. For this reason, a LiDAR, a wire cup brush for leveling the surface, and an air blower for blowing away pebbles, twigs, etc. may be attached to the front of the switching unit 5 of the autonomous mobile robot 1 in the direction of travel, to determine the presence or absence of foreign objects such as pebbles and automatically clean the inspection area (not shown). Alternatively, a cleaning robot may precede the autonomous mobile robot 1.

[0067] Furthermore, a camera such as a GoPro® may be attached to the autonomous mobile robot 1 to capture the state of the surface 6S in real time, and information such as pebbles that could not be removed or surface cracks and damage may be reflected in the training phase of the machine learning model 202 described later. Although not shown in the diagram, the autonomous mobile robot 1 is also equipped with an emergency stop button mechanism.

[0068] The autonomous mobile robot 1 may be not limited to the four-wheeled type described above, but may also be three-wheeled, caterpillar-type, etc. Furthermore, it may not be limited to such wheeled forms of movement, but may be flying types such as unmanned aerial vehicles (drones), snake-type, quadrupedal walking types, disc-type (Roomba®), centipede-type, spider-type, etc. It may also be two independent autonomous cooperative mobile robots, one for vibration source 2 and the other for sensor 3. Alternatively, multiple autonomous mobile robots 1 may be used depending on the width and length of the structure 6.

[0069] <Hardware configuration of non-destructive testing equipment> Figure 9 is a block diagram showing the hardware configuration of the personal computer 9 and cloud server (information processing device) 100 that constitute the non-destructive testing apparatus 10. As shown in Figure 9, the non-destructive testing apparatus 10 (personal computer 9) includes a control unit 20, a storage unit 40, a communication unit 50, an input unit 60, and an output unit 70.

[0070] (Communications Department) The communication unit 50 communicates with an external device (e.g., a cloud server 100) of the non-destructive testing apparatus 10 via a communication line. The specific configuration of the communication line is not limited to this exemplary embodiment, but examples of communication lines include wireless LAN (Local Area Network), wired LAN, WAN (Wide Area Network), public telephone network, mobile data communication network, or a combination thereof. The communication unit 50 transmits data supplied from the control unit 20 or storage unit 40 to the external device, and supplies data received from the external device to the storage unit 40 or control unit 20. Note that the non-destructive testing apparatus 10 (personal computer 9) may be configured without a communication unit 50.

[0071] (Input section) The input unit 60 is configured to receive input to the non-destructive testing device 10 (personal computer 9), and may include input devices such as a keyboard, mouse, touch panel, camera, and microphone. The input unit 60 may also be configured to receive data from input devices via wired or wireless interfaces such as USB (Universal Serial Bus) or Bluetooth (registered trademark). In this embodiment, sensing data from the vibration sensor 3 is input to the input unit 60. However, the vibration sensor 3 may be connected to the communication unit 50 instead of the input unit 60.

[0072] (Output section) The output unit 70 is configured to output from the non-destructive testing device 10. For example, the output unit 70 may have an interface such as a wireless LAN (Local Area Network) and is configured to transmit spectral data, measurement data, etc., to a cloud server 100 or an output device (external terminal or tablet, etc.) via this interface or the communication unit 50. The output unit 70 may also have an output device such as a display, printer, touch panel, or speaker. The vibration source 2 may be connected to the communication unit 50 or the output unit 70 for output control.

[0073] (Storage part) The memory unit 40 stores various types of data that the control unit 20 references. In particular, the memory unit 40 stores instructions for the computer program executed by the control unit 20. Examples of data stored in the memory unit 40 include measurement data 402, inspection position information 403, device position information 404, and spectral data SD, which will be described later. The measurement data 402 is data representing the actual measurement results of vibration testing on the structure 6.

[0074] (Control Unit) The control unit 20 comprises a switching control unit 21, a movement control unit 22, a vibration source control unit 23, a vibration sensor control unit 24, an inspection position information storage unit 25, an inspection position designation unit 26, an equipment position information acquisition unit 27, and a transmission unit 28. The switching control unit 21, the movement control unit 22, the vibration source control unit 23, and the vibration sensor control unit 24 control the switching unit 5, the movement unit 4M, the vibration source 2, and the vibration sensor 3, respectively. Each of the units 21 to 28 of the control unit 20 is realized by the control unit 20 reading and executing instructions from a computer program stored in the storage unit 40.

[0075] <Control system for non-destructive testing equipment> The moving unit (movement mechanism) 4M moves the machine body 4. The non-destructive testing device 10 equipped with this movement mechanism functions as an autonomous mobile robot 1 that performs non-destructive testing.

[0076] The inspection location information storage unit 25 stores inspection location information 403 representing each of the multiple inspection locations. The inspection location information 403 may be stored in a cloud server 100, and the autonomous mobile robot 1 may access this information via an external terminal (i.e., receive instructions from the on-site inspector).

[0077] Assuming the autonomous mobile robot 1 is in its initial state (first state where the vibration source 2 is held by the support unit 5S), the inspection position designation unit 26 designates the first inspection position to be inspected (for example, the inspection position closest to the autonomous mobile robot 1). The device position information acquisition unit 27 acquires device position information 404 representing the position of the autonomous mobile robot 1 (non-destructive testing device 10) from the Global Navigation Satellite System (GNSS). Based on this current position information, the movement control unit 22 moves the robot body 4 to the inspection position designated by the inspection position designation unit 26.

[0078] The movement control unit 22 moves the machine body 4 to its inspection position and then stops the machine body 4, and the switching control unit 21 switches the switching unit 5 from the first state (1) to the second state (2). Here, the downward operation of the switching unit 5 causes the vibration source 2 to be placed on the surface 6S of the structure 6 while detached from the machine body 4, and the vibration sensor 3 is pressed against the surface 6S of the structure 6 with a predetermined force applied by the torque control and pressing mechanism described above.

[0079] After the switching unit 5 switches to the second state (2), the vibration source control unit 23 generates an input wave, and the vibration sensor control unit 24 detects an output wave. At this point, the non-destructive testing device 10 performs vibration measurement in inspection mode, converts the raw measurement data into spectral data SD by FFT processing, and uploads it to the cloud server 100. After the inspection measurement is completed, the control unit 20 switches from the second state (2) to the first state (1) by controlling the switching unit 5 with the switching control unit 21 (upward operation of the switching unit 5). At this point, the autonomous mobile robot 1 (non-destructive testing device 10) returns from inspection mode to mobile mode. During this switching, the vibration source 2 is held by the support unit 5S.

[0080] Subsequently, the inspection position designation unit 26 designates the next inspection position to be inspected (for example, another inspection position closest to the one inspection position mentioned above). Upon receiving this designation, the movement control unit 22 uses a Global Navigation Satellite System (GNSS) to move the aircraft 4 to the inspection position designated by the inspection position designation unit 26.

[0081] By repeating this series of movement and inspection routines, the autonomous mobile robot 1 can automatically perform multiple inspections, thereby reducing the manpower and effort involved in non-destructive testing.

[0082] (Data comparison with conventional models) Figure 10 is a graph comparing spectral data from an autonomous mobile robot 1, a conventional robot (when the autonomous mobile robot 1 inspects and measures with the vibration source 2 fixed to the body 4), and manual measurement. Regarding this spectrum, the theoretical resonant frequency of the structure 6 being inspected is around 6,000 Hz, but the peak in the 8,000-10,000 Hz band is due to the influence of the boundary conditions of the structure 6 (structural shape, stiffness, load conditions, etc.).

[0083] The resonant frequency (natural frequency) of the machine body 4 itself is around 4,000 Hz. When measurements are taken with the vibration source 2 fixed to the autonomous mobile robot 1, a large difference is observed between the spectrum obtained with manual measurement and the measured spectrum, as shown in the upper part of Figure 10. However, when measurements are taken with the vibration source 2 disconnected, the influence of the vibration characteristics of the autonomous mobile robot 1 is almost eliminated, as shown in the lower part of Figure 10, and a spectral pattern almost identical to that obtained with manual measurement is obtained.

[0084] Furthermore, while boundary conditions are difficult to determine manually without expertise, they can be relaxed using the machine learning model (frequency normalization) described later, and this relaxation improves the accuracy of anomaly assessment.

[0085] <Building machine learning models> This disclosure may include the construction of a machine learning model for estimating the degree of abnormality of a structure under inspection. The following describes a machine learning model according to one embodiment of the present invention.

[0086] Conventional local vibration testing methods focus only on the change (difference) in the resonant frequency of modal parameters (equivalent mass, spring constant, damping ratio, etc., for each vibration mode), and therefore, even if a spectral response function is obtained, information in frequency bands other than the resonant frequency is not fully utilized. In contrast, the non-destructive testing device according to this embodiment estimates the state (degree of abnormality) of a structure by comparing spectral response functions (healthy spectrum vs. spectrum of the object under inspection). Here, the healthy spectrum is obtained in advance from an arbitrary (suitable) structure, normalized (amplitude normalized, frequency standardized) to give it generality, and then customized for the object under inspection by transfer learning of the healthy spectrum using a small amount of data from the object under inspection that appears to be healthy.

[0087] Normalizing the spectral response function involves both amplitude axis normalization (0 to 1) and frequency axis normalization. Each axis (amplitude, frequency) offers different advantages. Amplitude axis normalization scales the amplitude spectrum intensity to 0 to 1, eliminating variations due to the contact condition of the concrete surface (roughness), variations by the measurer (how the exciter is pressed, how the vibration sensor is pressed), and individual differences in measuring instruments. On the other hand, frequency axis normalization scales concrete structures with different member thicknesses or stiffnesses, such as beams and floor slabs, using a reference frequency (e.g., resonant frequency). This eliminates the need to learn sound data from various concrete structures, thus increasing versatility.

[0088] (Training data / Measurement data) As shown in Figure 9, the cloud server 100 includes a control unit 101 for estimating the degree of abnormality of the structure under inspection, training data 201, and a machine learning model 202. The cloud server 100 is an example of a training device according to this disclosure. The training data 201 and the machine learning model 202 may be contained in the storage unit 40 of the non-destructive testing device 10. The training data 201 is data used to train the machine learning model 202. As an example, the training data 201 includes a learning spectral response function obtained from the measurement results of vibrations of a healthy training structure.

[0089] (Machine learning model) The machine learning model 202 is a model used by the first estimation unit 211 and the second estimation unit 212, described later, to estimate the degree of abnormality of the structure 6 under inspection. The machine learning model 202 may be an unsupervised learning model or a supervised learning model. In this embodiment, an unsupervised learning model is used as the machine learning model 202. Here, if the machine learning model 202 is stored in the storage unit 40 of the non-destructive testing device 10, it means that the parameters defining the machine learning model 202 are stored in the storage unit 40.

[0090] (Specific example of machine learning model 202) As described above, in this embodiment, an unsupervised learning model is used as the machine learning model 202. Here, an autoencoder is used as a specific example of an unsupervised learning model. An autoencoder reduces the number of nodes in the hidden layer to reduce dimensionality (feature extraction), and then reconstructs data close to the original before dimensionality reduction by the reverse process, with the input layer and output layer having the same number of nodes. In this embodiment, only healthy learning spectral response functions are used as training data, and the autoencoder is trained so that the same learning spectral response function given to the input layer is reconstructed in the output layer. The trained autoencoder created in this way has only learned measurement data of healthy learning structures, so it can reconstruct healthy data but not abnormal data. Taking advantage of this property, separately acquired test data is input to the trained autoencoder, and data with small reconstruction errors between the input and output layers are identified as healthy, while data with large reconstruction errors are identified as abnormal.

[0091] However, the machine learning model 202 is not limited to an autoencoder or an unsupervised learning model. The machine learning model 202 may be, for example, a machine learning model that outputs classification results generated by supervised learning (i.e., a supervised learning model). In this case, the training data used to train the machine learning model 202 is, for example, a set of learning spectral response functions and correct labels (e.g., abnormal / normal) obtained from measurement results of vibrations of a healthy learning structure.

[0092] (Input and output of a machine learning model) The input data to machine learning model 202 is the spectral response function of the structure under inspection, obtained by measuring the vibration of the structure under inspection 6. The output data from machine learning model 202 is either the data used to estimate the degree of abnormality of the structure under inspection 6 or the calculated degree of abnormality itself. If machine learning model 202 is an autoencoder, the output of machine learning model 202 is the reconstructed spectral response function of the structure under inspection. If machine learning model 202 is a classification model, the output of machine learning model 202 is, for example, data showing the classification result of the state (degree of abnormality) of the structure under inspection 6.

[0093] (Cloud Server 100) Figure 11 is a block diagram showing the hardware configuration of the cloud server 100. The cloud server 100 comprises a control unit 101, a storage unit 200, a communication unit 300, an input unit 400, and an output unit 500.

[0094] The control unit 101 comprises a data processing unit 11, a first training phase execution unit 12, a first estimation phase execution unit 13, a second training phase execution unit 14, and a second estimation phase execution unit 15. Each part of the control unit 101 is realized by the control unit 101 reading and executing computer program instructions stored in the storage unit 200. The data processing unit 11 comprises an acquisition unit 111, a specification unit 112, and a standardization unit 113. The first training phase execution unit 12 comprises a training unit 121. The first estimation phase execution unit 13 comprises a first estimation unit (estimation unit) 131. The second training phase execution unit 14 comprises an extraction unit 141 and a retraining unit 142. The second estimation phase execution unit 15 comprises a second estimation unit 151.

[0095] (Acquisition Department) The acquisition unit 111 generates a learning spectral response function for a healthy learning structure based on the response acceleration measured by the vibration sensor 3. More specifically, as an example, the acquisition unit 111 extracts the response acceleration measured by the vibration sensor 3 at a predetermined extraction time, calculates multiple amplitude spectra (for example, 30 amplitude spectra) by performing a fast Fourier transform, and calculates a learning spectral response function by averaging these multiple amplitude spectra. However, the method by which the acquisition unit 111 generates the learning spectral response function is not limited to the example described above, and a learning spectral response function representing the vibration of the learning structure may be generated by other methods. In the following description, the learning spectral response function for a healthy learning structure is also referred to as "healthy data".

[0096] Furthermore, the acquisition unit 111 may acquire the learning spectral response function input to the input unit 400, or it may receive the learning spectral response function from another device connected via the communication unit 300. Alternatively, the acquisition unit 111 may acquire the learning spectral response function by reading it from a storage location specified by the user of the non-destructive testing device 10 (which may be the storage unit 40 inside the non-destructive testing device 10, or a storage device outside the non-destructive testing device 10).

[0097] (Specific part) The identification unit 112 identifies a reference frequency having predetermined characteristics in the learning spectral response function (a learning spectral response function obtained by applying elastic waves with a wide frequency band to a healthy learning structure) generated by the acquisition unit 111, which represents the output wave. The reference frequency is the frequency used by the normalization unit 113, described later, to normalize the learning spectral response function in the frequency axis direction. The reference frequency is set based on existing data and the characteristics of the structure. The reference frequency is, for example, the resonant frequency of the structure, but is not limited to this. The reference frequency may be, for example, the resonant frequency of similar healthy data, or a theoretical resonant frequency, or any arbitrary frequency may be set. Furthermore, the reference frequency may be a frequency identified using the resonant frequency of the structure.

[0098] (Standardization Department) The normalization unit 113 normalizes the learning spectral response function acquired by the acquisition unit 111 in the amplitude direction (vertical axis direction in Figure 3) using its maximum amplitude, and also normalizes it in the frequency direction (horizontal axis direction in Figure 3) using the reference frequency. Here, normalization includes both normalization and standardization. Normalization means normalizing to a number between 0 and 1.

[0099] Normalization refers to establishing a standard and indicating the variation based on that standard. As an example, the normalization unit 113 normalizes the learning spectral response function in the frequency axis direction by dividing the frequency of the learning spectral response function by the reference frequency. However, the method for normalizing the learning spectral response function is not limited to this, and the normalization unit 113 may normalize the learning spectral response function by other methods.

[0100] By normalizing the learning spectral response function in the frequency axis direction (scaling it to a reference frequency), the learning spectral response function becomes independent of a specific thickness L or stiffness K (the reference frequency matches). Therefore, by using healthy data obtained from any learning structure in conjunction with this, a general-purpose machine learning model 202 can be constructed. This makes it possible to estimate the degree of abnormality of the structure under inspection even when there is no healthy learning spectral response function for a similar learning structure.

[0101] (Training Department) The training unit 121 trains the machine learning model 202 using a standardized learning spectral response function as training data. The machine learning method used for the machine learning model 202 is not limited; for example, decision tree-based, linear regression, or neural network methods may be used, or two or more of these methods may be used. Examples of decision tree-based methods include LightGBM (Light Gradient Boosting Machine), Random Forest, and XGBoost. Examples of linear regression methods include Bayesian regression, Support Vector Regression, Ridge Regression, Lasso Regression, and ElasticNet. Examples of neural networks include deep learning.

[0102] (1st estimation part) The first estimation unit (estimation unit) 131 acquires the spectral response function of the structure under inspection. The spectral response function acquired by the first estimation unit 131 is, for example, a spectral response function obtained by applying elastic waves with a wide frequency band to the structure under inspection. For example, the first estimation unit 131 extracts the response acceleration measured by the vibration sensor 3 at a predetermined extraction time, calculates multiple amplitude spectra (for example, 30 amplitude spectra) by performing a fast Fourier transform, and calculates the spectral response function of the structure under inspection by averaging these multiple amplitude spectra.

[0103] Furthermore, the first estimation unit 131 may acquire the spectral response function to be inspected input to the input unit 400, or it may receive the spectral response function to be inspected from another device connected via the communication unit 300. Alternatively, the first estimation unit 131 may acquire the spectral response function to be inspected by reading it from a storage location specified by the user of the non-destructive testing apparatus 10 (which may be the storage unit 40 in the personal computer 9 that constitutes the non-destructive testing apparatus 10, or a storage device outside the non-destructive testing apparatus 10).

[0104] When vibrations are measured at multiple locations within the structure under inspection, a spectral response function for the structure under inspection is generated from the measurement results at each location.

[0105] Furthermore, the first estimation unit 131 identifies a reference frequency having predetermined characteristics in the acquired spectral response function of the structure under inspection. The reference frequency identified by the first estimation unit 131 is, for example, the resonance frequency of a healthy structure corresponding to the structure under inspection. Here, a healthy structure corresponding to the structure under inspection is, for example, the structure in the state before an abnormality occurred in the structure under inspection.

[0106] The first estimation unit 131, as an example, identifies the reference frequency of the structure under inspection. More specifically, the first estimation unit 131, as an example, adds density ρ, member thickness L, and dynamic elastic modulus E to equation (1) above. d The resonant frequency f0 obtained by substituting is identified as the reference frequency. Alternatively, the first estimation unit 131 may identify the reference frequency by referring to a pre-constructed database. In this case, the first estimation unit 131 may, for example, refer to a database in which the stiffness, shape, and / or density of the structure are associated with the reference frequency to identify the reference frequency of the structure under inspection. In other words, the first estimation unit 131 can also identify the reference frequency using the stiffness, shape, and density of the structure under inspection.

[0107] The rigidity of a structure has a significant impact on its resonant frequency. Specifically, the compressive strength of typical concrete and the thickness of the members have a major influence on the resonant frequency.

[0108] The higher the stiffness, the faster the elastic waves propagate through the material; conversely, the lower the stiffness, the slower the elastic waves propagate. In the case of typical concrete, stiffness can be estimated from the compressive strength, with higher strength corresponding to higher stiffness. Within the elastic range, the stiffness of concrete is defined by the elastic modulus. Furthermore, the resonant frequency is the frequency at which an object vibrates intrinsically, and as shown in equation (1), it is determined by the object's stiffness, shape, and density.

[0109] Furthermore, the first estimation unit 131 normalizes the acquired spectral response function under test in the amplitude direction (vertical axis direction in Figure 3) using its maximum amplitude, and normalizes the spectral response function under test in the frequency axis direction (horizontal axis direction in Figure 3) using the reference frequency identified by the first estimation unit 131. The process by which the first estimation unit 131 normalizes the spectral response function under test is the same as the process performed by the normalization unit 113 described above, and a detailed explanation is omitted here. In the example in Figure 11, the normalization unit 113 and the first estimation unit 131 are shown as different components, but the normalization unit 113 and the first estimation unit 131 may be a single component, or the normalization process may be performed using a common library.

[0110] Furthermore, the first estimation unit 131 estimates the degree of abnormality of the structure under inspection by inputting the normalized spectral response function of the structure under inspection into the machine learning model 202. In other words, the first estimation unit (estimation unit) 131 estimates the state of the structure 6 by inputting the spectral response function obtained by the normalization unit 113 into the machine learning model 202, which takes the spectral response function representing the output wave as input data and the data used to estimate the state of the structure 6 including concrete as output data.

[0111] If the machine learning model 202 is an autoencoder, the first estimation unit 131 estimates the state of the structure 6 under inspection based on the normalized spectral response function of the object under inspection and the spectral response function of the object under inspection output from the autoencoder. If the machine learning model 202 is a classification model generated by supervised learning, the first estimation unit 131 estimates the state of the structure 6 under inspection based on the classification result obtained by inputting the spectral response function of the object under inspection into the machine learning model 202.

[0112] (Extraction part) The extraction unit 141 extracts from among the multiple spectral response functions of the subject to be inspected output by the first estimation unit 131 those whose estimation results by the first estimation unit 131 satisfy predetermined conditions. Here, predetermined conditions are, for example, that the rank when the abnormality degree calculated for each spectral response function of the subject to be inspected is sorted in ascending order is higher than a predetermined rank (e.g., top 10%). Alternatively, if the majority of the structure to be inspected is judged to be in a sound state, it is also effective to train the autoencoder using only the measurement data (spectral response functions of the subject to be inspected) obtained from the structure to be inspected, self-evaluate the data used for training with the trained model, sort it in descending order of abnormality degree, and extract data that is higher than the predetermined rank. In other words, the extraction unit 141 extracts the measurement results of locations that are likely to be sound (no abnormalities have occurred) from among the vibration measurement results at each of the multiple locations of the structure to be inspected.

[0113] (Retraining Department) The retraining unit 142 updates the machine learning model 202 trained by the training unit 121 by machine learning using the spectral response function of the object under inspection extracted by the extraction unit 141 as training data. As an example, the retraining unit 142 updates the machine learning model 202 by transfer learning using the spectral response function of the object under inspection extracted by the extraction unit 141 as training data. In this case, in other words, the machine learning model 202 can also be said to be a model updated by transfer learning using training data in which the spectral response function of the object under inspection, representing multiple measurement results that appear to be healthy for the structure under inspection, is normalized in the amplitude axis direction and normalized in the frequency axis direction.

[0114] However, the method by which the retraining unit 142 updates the machine learning model 202 is not limited to the examples described above. The retraining unit 142 may update the machine learning model 202 by fine-tuning using the spectral response function of the subject under test extracted by the extraction unit 141, for example.

[0115] (Second estimation part) The second estimation unit 151 estimates the state of the structure 6 under inspection using a machine learning model 202. The processing performed by the second estimation unit 151 is the same as that performed by the first estimation unit 131, and a detailed explanation is omitted here.

[0116] Figure 12 shows the configuration of a non-destructive testing system according to one embodiment of the present invention. The non-destructive testing system 600 comprises an autonomous mobile robot 1 (non-destructive testing device 10), a cloud server (information processing device) 100, an AI model (machine learning model) 202, and one or more terminals T, T'. Here, terminal T is the terminal of the on-site inspector, and terminal T' is the terminal of the remote inspector.

[0117] The autonomous mobile robot 1 (non-destructive testing device 10) receives control information (such as the coordinates of the measurement points, the number of measurement points, and the measurement range) from the local terminal T (i.e., the local inspector), performs movement and inspection, and transmits measurement data such as measurement spectrum data SD, coordinates, and images (processed using FFT, taking data efficiency into consideration) to a cloud server 100, for example, via AWS (Amazon Web Services). The control information may also be transmitted from the local terminal T or remote terminal T' to the non-destructive testing device 10 via the cloud server 100. Here, parameters for data quality control are transmitted from the local terminal T or remote terminal T' to the cloud server 100.

[0118] Measurement data may be sent to the local terminal T. In addition, parameters for AI analysis are sent from the local terminal T or remote terminal T' to the cloud server 100. The cloud server 100 sends the measurement data and these AI analysis parameters to the AI ​​model 202, and the AI ​​model 202 takes them as input and outputs the AI ​​analysis results (estimated anomaly score) to the cloud server 100.

[0119] The cloud server 100 shares the AI ​​analysis results and measurement data with the local terminal T and remote terminal T' via the communication unit 300. Based on this data, the verifier may send further instructions to the autonomous mobile robot 1 to perform further inspections, or request the dispatch of personnel.

[0120] In other words, one or more terminals obtain the degree of abnormality of the structure 6 estimated by the first estimation unit 131 via the communication unit 300 of the cloud server (information processing device) 100. This allows the inspector to easily grasp the degree of abnormality of the structure 6. Furthermore, on-site inspectors can sequentially check the data on the cloud server 100, for example, using a tablet, and can also check remotely from branch offices or research laboratories. The terminals may be equipped with an emergency stop button for the autonomous mobile robot 1. In addition, there may be multiple cloud servers 100, each with its own assigned role.

[0121] <An example of the process for generating and estimating the state of a machine learning model> Figure 13 is a flowchart showing an example of the process for generating (training) the machine learning model 202 performed by the non-destructive testing system 600, and an example of the process for estimating the state. In the example in Figure 13, the processing flow is explained when the machine learning model 202 is an autoencoder.

[0122] (Data quality assessment, initial screening) In step S10, the control unit 101 obtains a map with pre-entered coordinates of measurement points and tentatively determines the theoretical resonance frequency when representative physical properties are used. The control unit 20 of the autonomous mobile robot 1 performs spectral processing on the actually measured resonance frequency and compares it with the theoretical resonance frequency. If it is determined that the measurement was clearly not successful (for example, if it deviates by ±30% or more from the theoretical value), the control unit automatically performs the measurement again. Alternatively, if three measurements are taken at the same location and even one of them is clearly different (for example, if it deviates by ±10% or more from the measurements in the other two measurements), the control unit 101 performs the measurement again. These thresholds for re-measurement may be arbitrarily set for each target structure based on the uncertainty of the measured value.

[0123] (Obtaining the spectral response function for training) In step S11, the acquisition unit 111 acquires a learning spectral response function obtained from the vibration measurement results of a sound learning structure. The learning spectral response function is acquired by the following method as an example. First, the manager of the non-destructive testing system 600 brings the vibration source 2 and vibration sensor 3 into contact with the surface of the learning structure, which is the test object, and generates white noise, for example, using the vibration source 2 to excite local vibrations of the learning structure, which is the test object. The acquisition unit 111 acquires the response acceleration measured by the vibration sensor 3 during the excitation. The response acceleration is acquired, for example, at a sampling frequency of 50 kHz.

[0124] Furthermore, for example, in step S12, the acquisition unit 111 may augment the training data 201. As an example, the acquisition unit 111 may adjust one learning spectral response function f to accommodate slight changes in the learning spectral response function due to differences in how the vibration source 2 is pressed against the test specimen during vibration measurement. i to f i The data is copied to -s(-3≦s≦+3) and shifted along the frequency axis, resulting in a 7x data augmentation. The acquisition unit 111 may set to zero any frequency components that would result in data loss due to the shift. The acquisition unit 111 may also be configured not to perform the processing in step S12.

[0125] (Normalization of the spectral response function for training) In step S12 (an example of a specific step, an example of a normalization step), the specific unit 112 identifies a reference frequency having predetermined characteristics in the learning spectral response function (a spectral response function obtained by applying elastic waves with a wide frequency band to the learning structure) generated by the acquisition unit 111. The normalization unit 113 normalizes the learning spectral response function in the amplitude direction and normalizes the learning spectral response function in the frequency axis direction using the reference frequency. By normalizing the amplitude axis direction in the learning spectral response function, variations due to the contact state between the vibration source 2 and vibration sensor 3 and the surface of the learning structure, as well as variations due to the measurer, can be reduced. By normalizing in the frequency axis direction using the reference frequency, the spectral characteristics become independent of a specific thickness L or a specific stiffness K. Therefore, sound data (learning spectral response function) obtained from different measurers, surface conditions, and different types of learning structures (member thickness, stiffness, etc.) can be jointly used as training data.

[0126] (Machine Learning) In step S13 (an example of a training step), the training unit 121 trains the machine learning model 202 using a normalized learning spectral response function as training data.

[0127] (Acquisition of the spectral response function under examination) In step S21, the first estimation unit 131 obtains the spectral response function of the structure under inspection obtained by measuring the vibration of the structure under inspection. The spectral response function of the structure under inspection is obtained by the following method as an example. First, the manager of the non-destructive testing system 600 brings the vibration source 2 and vibration sensor 3 into contact with the surface 6S of the structure under inspection 6, and generates white noise, for example, using the vibration source 2 to excite local vibrations of the structure under inspection 6. The first estimation unit 131 obtains the response acceleration measured by the vibration sensor 3 during excitation. The response acceleration is obtained, for example, at a sampling frequency of 50 kHz. For example, the first estimation unit 131 extracts the waveform at a predetermined extraction time (for example, 0.3072 seconds). For example, the first estimation unit 131 performs a fast Fourier transform 30 times consecutively on the extracted waveform and generates the spectral response function of the structure under inspection by averaging the 30 amplitude spectra.

[0128] (standardization) In step S22 (an example of the inspection target normalization step), the first estimation unit 131 normalizes the generated inspection target spectral response function. More specifically, the first estimation unit 131 first identifies a reference frequency having predetermined characteristics. The first estimation unit 131 also normalizes the acquired inspection target spectral response function in the amplitude direction and normalizes the inspection target spectral response function in the frequency axis direction using the identified reference frequency. Here, the first estimation unit 131 normalizes the inspection target spectral response functions measured at multiple locations on a single inspection target structure in the amplitude direction and the frequency axis direction, respectively.

[0129] (Estimation of state) In steps S23 and S24 (an example of an estimation step), the first estimation unit 131 estimates the degree of abnormality of the structure under inspection based on the output data output from the machine learning model 202 by inputting the normalized spectral response function of the structure under inspection into the machine learning model 202. Here, if the machine learning model 202 is an autoencoder, the first estimation unit 131 estimates the state of the structure under inspection according to the spectral response function of the structure under inspection input into the autoencoder and the reconstructed spectral response function output from the autoencoder. Here, the first estimation unit 131 estimates the above state by inputting the spectral response functions of multiple locations measured on a single structure under inspection into the machine learning model 202. Steps S23 and S24 will be described in order below.

[0130] First, the first estimation unit 131 inputs the normalized spectral response function of the object under test to the autoencoder and obtains the reconstructed spectral response function output from the autoencoder. For the sake of explanation, the normalized spectral response function of the object under test input to the autoencoder will be referred to as the "measured spectrum," and the reconstructed spectral response function, which is reconstructed in a normalized state and output from the autoencoder, will be referred to as the "healthy spectrum."

[0131] Furthermore, the first estimation unit 131 performs smoothing and noise reduction processing on both the measured spectrum and the healthy spectrum. As an example, the first estimation unit 131 performs smoothing by calculating a moving average with a predetermined parameter width. Because the spectral response function has a steep multimodal nature, even a slight difference between the peak frequency of the spectral response function under inspection input to the autoencoder and the peak frequency of the reconstructed spectral response function results in a large reconstruction error. Therefore, the peaks of the spectral response function are smoothed by calculating a moving average.

[0132] Also, since most of the frequency components of the plurality of parameters are noise, the first estimation unit 131 subtracts a certain threshold value and sets the components that become negative values to zero to perform noise removal. As an example, the threshold value may be set to a height of 0.2 when the maximum value of the spectral response function before performing the moving average is set to 1.

[0133] Further, the first estimation unit 131 normalizes each of the measured spectrum and the healthy spectrum in the amplitude direction. Although these spectrum data are already normalized, due to the smoothing process and the noise cut process, the amplitude of the spectral response function changes, so normalization is performed again to maintain the amplitude normality.

[0134] Also, the first estimation unit 131 performs an evaluation regarding the state of the inspection target structure based on the measured spectrum and the healthy spectrum. As an example of the evaluation, it is a value indicating the degree of health (abnormality degree) of the inspection target structure. More specifically, as an example, the first estimation unit 131 calculates at least any one of the following abnormality degrees Ansco, Ansco2, and Ansco3 as an evaluation of the inspection target structure.

[0135] FIG. 14 is a diagram for explaining a method of calculating the abnormality degrees Ansco, Ansco2, and Ansco3. In FIG. 14, the reconstructed spectral response function w22 is a healthy spectrum output from the autoencoder. The inspection target spectral response function w21 is a measured spectrum obtained by measuring the vibration of the inspection target structure. In the diagram for explaining this method of calculating the abnormality degree, the amplitude axis uses a normalized one, but for the frequency axis direction, it may be normalized or not normalized (the abnormality degree results are the same). In the figure, the horizontal axis indicates frequency (Hz), and the vertical axis indicates the amplitude spectrum. The frequency fs is the center of gravity of the reconstructed spectral response function w22. The amplitude Am i is the amplitude spectrum of the frequency f of the inspection target spectral response function w21. The amplitude As i is the amplitude spectrum of the frequency f in the reconstructed spectral response function w22. j is the amplitude spectrum of the frequency f in the reconstructed spectral response function w22. j is the amplitude spectrum of the frequency f in the reconstructed spectral response function w22.

[0136] (Abnormality Ansco) The first estimation unit 131 calculates the anomaly score Ansco as an example using the following formula (2). In formula (2), As j Each frequency f of the amplitude-normalized healthy spectrum j This is the amplitude, Am i Each frequency f of the amplitude-normalized measurement spectrum i This is the amplitude. In this example, the anomaly score Ansco is the sum of the absolute values ​​of the amplitude differences between the measured spectrum and the healthy spectrum.

number

[0137] (Abnormal degree Ansco2, Ansco3) The above-mentioned anomaly score Ansco may not reflect the extent of damage. This is because the method using the anomaly score Ansco only considers the amplitude difference between the measured spectrum and the healthy spectrum. For example, even if there is a large decrease in the resonant frequency, the anomaly score may be calculated to be small. Therefore, the extent of damage may be evaluated using anomaly scores Ansco2 and Ansco3, which are based on the area moment of the spectral response function. As an example, the first estimation unit 131 may calculate anomaly scores Ansco2 and Ansco3 using the following equation (3). In equation (3), fs is the centroid of the healthy spectrum (see Figure 14), and the amplitude Am i The frequency f of the spectral response function w21 under examination. i This is the amplitude. Amplitude As j This is the frequency f in the reconstructed spectral response function w22. j This is the amplitude.

number

[0138] In this case, to put it another way, the first estimation unit 131 can also calculate the degree of abnormality regarding the state of the structure under inspection using the area moment Mm of the measured spectrum and the area moment Ms of the healthy spectrum with respect to the centroid fs of the healthy spectrum. More specifically, the first estimation unit 131 can also calculate the degree of abnormality regarding the state of the structure under inspection as the value obtained by dividing the area moment Mm of the measured spectrum by the area moment Ms of the healthy spectrum (Mm / Ms) with respect to the centroid fs of the healthy spectrum. Alternatively, the first estimation unit 131 can also calculate the degree of abnormality regarding the state of the structure under inspection as the square root of the value obtained by dividing the area moment Mm of the measured spectrum by the area moment Ms of the healthy spectrum (Mm / Ms) with respect to the centroid fs of the healthy spectrum.

[0139] (Extraction of relatively healthy data) In step S31 (an example of an extraction step), the extraction unit 141 extracts relatively healthy data. For example, the extraction unit 141 extracts from among multiple measurement spectra those whose estimation results using the machine learning model 202 satisfy predetermined conditions. More specifically, as an example, the extraction unit 141 extracts from among the multiple measurement spectra used to calculate the anomaly score those whose rank is higher than a predetermined rank when the calculated anomaly scores are sorted in ascending order.

[0140] The retraining unit 142 performs an update process (update step) to update the machine learning model 202 using machine learning with the measurement spectrum extracted by the extraction unit 141 as training data. As an example, the retraining unit 142 updates the final layer of the machine learning model 202 using transfer learning with the measurement spectrum extracted by the extraction unit 141 as training data.

[0141] As described above, by normalizing the spectral response function in the frequency axis direction, the condition of the structure under inspection can be evaluated even without sound data for similar structures. On the other hand, if sound data for the same type of structure as the structure under inspection is available, the accuracy of the machine learning model 202, which was trained on a different training structure, can be improved by using transfer learning.

[0142] Figure 15 shows an example of the update process performed by the retraining unit 142. In the example in Figure 15, the retraining unit 142 retrains the final layer 2021 of the machine learning model 202, which was trained in steps S11 to S13 of Figure 13, using data representing the measurement results of the structure under inspection or data of a small amount of healthy structures of the same type as the structure under inspection. The main advantage of retraining is that if a small amount of healthy data of the structure under inspection is available, the pre-trained machine learning model 202 can be customized for the structure under inspection. Hereafter, the machine learning model 202 before the update process is performed by the retraining unit 142 will also be referred to as the "general-purpose model".

[0143] For example, not all parts of a structure under inspection are damaged; often, sound parts remain. Therefore, by acquiring information on the spectral response function of the healthy state of the structure under inspection, even in small amounts, and customizing the general-purpose machine learning model 202 constructed in steps S11 to S13 of Figure 13, the estimation accuracy can be improved. By incorporating actual measured values ​​of the structure under inspection, the type of concrete, boundary conditions, and the influence of the vibrator are incorporated into the model, resulting in improved prediction accuracy of the spectral response function in a healthy state. Note that if no spectral response function of a healthy state of the structure under inspection can be obtained, the retraining unit 142 does not perform the retraining process.

[0144] The second estimation unit 151 estimates the state of the structure under inspection using the machine learning model 202 generated by retraining. More specifically, the second estimation unit 151 performs the processes shown in steps S21 to S24 of Figure 13. The content of the processes performed by the second estimation unit 151 is the same as that performed by the first estimation unit 131, and a detailed explanation is omitted here.

[0145] <Effects of the Embodiment> As mentioned above, the spectral response function of structure 6 varies depending on the member thickness and stiffness, and for example, the spectral characteristics of a floor slab and a beam differ significantly. Therefore, if the spectral response function is used as training data without normalizing it along the frequency axis, there is a problem that the estimation accuracy will be low depending on the type of structure. For example, a model trained using training data for a slab cannot appropriately estimate the state of a beam. Furthermore, there are differences in stiffness and density depending on the type of cement (e.g., ordinary cement or blast furnace cement), and considering that the measured vibration characteristics (spectral response function) also depend on boundary conditions (structural shape, fixing conditions, load conditions) and the characteristics of vibration source 2, the number of required data cases becomes enormous, making it practically difficult to comprehensively obtain sound data.

[0146] In contrast, according to this embodiment, the spectral response function is normalized in the amplitude axis direction using a reference frequency, and the machine learning model 202 is trained using the spectral response function normalized in the frequency axis direction. As a result, the spectral characteristics no longer depend on a specific member thickness or specific stiffness. Therefore, sound data obtained from different measurers, surface conditions, and different types of structures 6 (member thickness, stiffness, etc.) can be jointly used as training data. In other words, according to this embodiment, even when there is no sound data or the sound data is insufficient, a machine learning model 202 capable of more accurately estimating the state of various types of structures 6 can be constructed.

[0147] Furthermore, according to this embodiment, a general-purpose model is trained based only on an arbitrary healthy spectral response function, and the model is updated by transfer learning or the like to match the target structure (concrete stiffness, member thickness, boundary conditions, etc.). This makes it possible to estimate the state of various structures. In addition, according to this embodiment, in order to efficiently utilize important information of the spectral response function as an anomaly score, an index can be provided that effectively utilizes frequency band information other than the resonant frequency.

[0148] [Examples of implementation using software] The functions of the non-destructive testing apparatus 10 (hereinafter referred to as "the apparatus") can be realized by a program that causes a computer to function as the apparatus, and by a program that causes a computer to function as each control block of the apparatus (particularly each part included in the control unit 20).

[0149] In this case, the device includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., memory) as hardware for executing the program. By executing the program using this control device and storage device, the functions described in each of the embodiments are realized.

[0150] The above program may be recorded on one or more computer-readable recording media, not temporary ones. These recording media may or may not be provided by the above device. In the latter case, the program may be supplied to the above device via any wired or wireless transmission medium.

[0151] Furthermore, some or all of the functions of each of the above control blocks can also be realized by logic circuits. For example, an integrated circuit in which logic circuits functioning as each of the above control blocks are formed is also included in the scope of the present invention. In addition, it is also possible to realize the functions of each of the above control blocks by, for example, a quantum computer.

[0152] Furthermore, each process described in the above embodiments may be performed by AI (Artificial Intelligence). In this case, the AI ​​may operate on the control device described above, or it may operate on other devices (for example, an edge computer or a cloud server).

[0153] The present invention is not limited to the embodiments described above, and various modifications are possible within the scope of the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of the present invention.

[0154] 〔summary〕 A non-destructive testing apparatus according to Embodiment 1 of the present invention is a non-destructive testing apparatus for performing non-destructive testing of a structure including concrete by impact elastic wave method, comprising: a vibration source for generating an input wave; a vibration sensor for detecting an output wave; a machine body on which the vibration sensor is provided; and a switching unit for switching between (1) a first state in which the machine body holds the vibration source and the vibration sensor; and (2) a second state in which the vibration source is installed on the surface of the structure and detached from the machine body, and the machine body holds the vibration sensor and presses the vibration sensor against the surface.

[0155] According to the above embodiment, in the second state (2), the vibration source 2 installed on the surface 6S of the structure 6 is separated from the machine body 4. Therefore, it is possible to prevent unintended shock elastic waves from being superimposed on the output wave, and thus prevent a decrease in inspection accuracy caused by unintended shock elastic waves. As a secondary effect, since the vibration source 2 is pressed against the surface 6S using its own weight, the load applied when pressing the vibration source 2 against the surface 6S can be kept constant.

[0156] Furthermore, the non-destructive testing apparatus according to Embodiment 2 of the present invention further comprises a moving unit for moving the machine body in the non-destructive testing apparatus described in Embodiment 1. According to the above embodiment, the non-destructive testing apparatus 10 functions as an autonomous mobile robot 1 that performs non-destructive testing.

[0157] Furthermore, the non-destructive testing apparatus according to embodiment 3 of the present invention further comprises a switching control unit, a moving control unit, a vibration source control unit, and a vibration sensor control unit, which control the switching unit, the moving unit, the vibration source, and the vibration sensor, respectively, in the non-destructive testing apparatus described in embodiment 2, wherein the moving control unit moves the machine to the inspection position and then stops the machine, the switching control unit switches from the first state to the second state, the vibration source control unit generates the input wave, the vibration sensor control unit detects the output wave, and the control unit switches from the second state to the first state by controlling the switching unit.

[0158] According to the above embodiment, the non-destructive testing device 10 can automate the inspection at the inspection location.

[0159] Furthermore, the non-destructive testing apparatus according to aspect 4 of the present invention is a non-destructive testing apparatus according to aspect 3, further comprising: an inspection position information storage unit that stores inspection position information representing each of a plurality of inspection positions; an inspection position designation unit that designates the next inspection position to be inspected from among the plurality of inspection positions; and an apparatus position information acquisition unit that acquires apparatus position information representing the position of the non-destructive testing apparatus from a satellite positioning system, wherein the movement control unit moves to the inspection position designated by the inspection position designation unit.

[0160] According to the above embodiment, the non-destructive testing device 10 can automatically perform inspections at multiple inspection locations, thereby reducing the number of personnel involved in non-destructive testing.

[0161] Furthermore, in the non-destructive testing apparatus according to embodiment 5 of the present invention, the non-destructive testing apparatus according to any one of embodiments 1 to 3, wherein the switching unit includes a support unit that moves between a first position with a higher height and a second position with a lower height, the support unit achieving the first state by supporting a part of the vibration source from below and moving the vibration source away from the structure in the first position, and achieving the second state by installing the vibration source on the surface and moving the part of the vibration source away from the structure in the second position.

[0162] According to the above embodiment, the non-destructive testing device 10 can reliably and easily switch between the first state (1) and the second state (2).

[0163] Furthermore, the non-destructive testing apparatus according to embodiment 6 of the present invention is the non-destructive testing apparatus described in embodiment 5, wherein the vibration source includes, as part of the, a pair of protrusions that protrude from each of the pair of sides at a position higher than the center of gravity, and the support portion is a pair of members whose spacing is wider than the spacing between the pair of sides of the vibration source and narrower than the spacing between the tips of the pair of protrusions.

[0164] According to the above embodiment, the non-destructive testing device 10 can inexpensively realize a configuration that allows switching between a first state (1) and a second state (2).

[0165] Furthermore, in the non-destructive testing apparatus according to embodiment 7 of the present invention, in the non-destructive testing apparatus described in embodiment 6, each of the pair of members has the same shape and has a first recess that opens more upward when the main surface is viewed from above.

[0166] Furthermore, the non-destructive testing apparatus according to embodiment 8 of the present invention is the non-destructive testing apparatus described in embodiment 7, wherein each of the pair of protrusions has the same shape and a second recess that opens more downward when viewed from a direction parallel to the main surface of the pair of members.

[0167] According to the above embodiments (7 and 8), even if the surface 6S of the structure 6 is not a horizontal smooth surface, when transitioning from the second state (2) to the first state (1), the pair of plate-like members 5B, 5B (their V-shaped structure) and the pair of protrusions (their inverted V-shaped structure) interlock, ensuring that the vibration source 2 is held securely. Furthermore, even when the machine body 4 moves toward the inspection position while the first state (1) is adopted, the vibration source 2 can be held securely.

[0168] Furthermore, the non-destructive testing apparatus according to aspect 9 of the present invention is a non-destructive testing apparatus according to any one of aspects 1 to 3, wherein the vibration source is an exciter comprising an excitation coil, a housing for housing the excitation coil, the housing having a lower bottom surface facing and approaching the surface when the vibration source is installed on the surface, a vibration transmitter connected to the excitation coil, the vibration transmitter having a tip protruding from the lower bottom surface, and one or more legs that allow the housing to stand upright when the vibration source is installed on the surface, wherein the one or more legs are made of an elastic body and are configured such that (1) when the tip is separated from the surface, the lower end is located below the tip, and (2) when the housing is installed on the surface and the tip is in contact with the surface, the lower end is in contact with the surface together with the tip.

[0169] According to the above embodiment, the cushion is identified as a leg. The load pressing the tip against the surface can be kept constant while stabilizing the housing placed on the surface.

[0170] Furthermore, the non-destructive testing apparatus according to embodiment 10 of the present invention is the non-destructive testing apparatus described in embodiment 5, wherein the vibration sensor is directly or indirectly fixed to the support portion, and further comprises an elastic cushioning material interposed between the vibration sensor and the machine body.

[0171] According to the above embodiment, the load applied to the vibration sensor 3 against the surface 6S can be made to be as close to constant as possible. Furthermore, by selecting the elastic modulus of the elastic material constituting the cushioning material, it is possible to reproduce, for example, the load applied by a skilled worker when pressing the vibration sensor 3 against the surface. Therefore, inspection accuracy equivalent to that achieved when a skilled worker performs non-destructive testing can be realized.

[0172] Furthermore, the non-destructive testing apparatus according to embodiment 11 of the present invention further comprises a ball joint interposed between the vibration sensor and the support portion, the ball joint being directly or indirectly fixed to the support portion, and the vibration sensor being fixed to the ball joint.

[0173] According to the above embodiment, the vibration sensor 3 can be brought into close contact with the surface 6S of the structure 6 regardless of the angle of the surface 6S, thereby automatically suppressing variations in inspection accuracy.

[0174] Furthermore, a non-destructive testing system according to aspect 12 of the present invention is a non-destructive testing system comprising a non-destructive testing apparatus described in any one of aspects 1 to 3, an information processing device including an estimation unit, and one or more terminals, wherein the information processing device further comprises a communication unit that shares the degree of abnormality of the structure estimated by the estimation unit with the one or more terminals.

[0175] According to the above embodiment, the abnormality level of structure 6 can be shared simultaneously with multiple terminals.

[0176] Furthermore, the non-destructive testing system according to embodiment 13 of the present invention further comprises, in the non-destructive testing system described in embodiment 12, the non-destructive testing being a non-destructive testing by a local vibration test method, and the information processing device further comprising: an identification unit that identifies a reference frequency having predetermined characteristics in the spectral response function representing the output wave; a normalization unit that normalizes the spectral response function in the frequency axis direction using the reference frequency; and an estimation unit that estimates the state of the structure by inputting the spectral response function obtained by the normalization unit into a machine learning model that takes the spectral response function representing the output wave as input data and outputs data used to estimate the state of the structure including concrete.

[0177] According to the above embodiment, vibration characteristics other than the reference frequency can also be considered, so even if there is no output wave obtained by measuring a healthy learning structure, the state of structure 6 can be estimated.

[0178] Furthermore, in the non-destructive testing system according to embodiment 14 of the present invention, the normalization unit normalizes the spectral response function in the amplitude axis direction as well, in the non-destructive testing system described in embodiment 13. According to the above embodiment, by normalizing the amplitude of the spectral response function, it is possible to reduce the variation in output waves that may occur due to differences in the contact state between each of the vibration source 2 and vibration sensor 3 and the surface of the learning structure, or the skill level of the worker performing the test.

[0179] Furthermore, an embodiment 15 of the present invention is a non-destructive testing method for performing non-destructive testing of a structure including concrete by impact elastic wave method, and includes a machine body that holds a vibration source and a vibration sensor, a separation step of installing the vibration source on the surface of the structure at the measurement position and separating the vibration source from the machine body, and an inspection step of applying an input wave to the structure using the vibration source and detecting an output wave using the vibration sensor.

[0180] According to the above embodiment, the same effects as in embodiment 1 can be obtained.

[0181] Furthermore, the non-destructive testing method according to embodiment 16 of the present invention further includes, in the non-destructive testing method described in embodiment 15, a moving step performed before the separation step and the inspection step, wherein the machine body, while holding the vibration source and the vibration sensor, is moved to the measurement position; and a holding step performed after the separation step and the inspection step, wherein the machine body holds the vibration source.

[0182] According to the above embodiment, inspection at the inspection location can be automated.

[0183] Furthermore, the non-destructive testing method according to aspect 17 of the present invention is a non-destructive testing method according to aspect 15 or 16, wherein the non-destructive testing is a non-destructive testing method using a local vibration test, and includes: a identification step of identifying a reference frequency having predetermined characteristics in the spectral response function representing the output wave; a normalization step of normalizing the spectral response function in the frequency axis direction using the reference frequency; and an estimation step of estimating the degree of abnormality of the structure by inputting the spectral response function obtained in the normalization step into a machine learning model that takes the spectral response function representing the output wave as input data and data used to estimate the state of the structure including concrete as output data.

[0184] According to the above configuration, the same effects as in embodiment 13 can be obtained.

[0185] Furthermore, in the non-destructive testing method according to embodiment 18 of the present invention, in the non-destructive testing method described in embodiment 17, in the normalization step, the spectral response function is normalized in the frequency axis direction using the reference frequency, and the spectral response function is normalized in the amplitude axis direction.

[0186] According to the above configuration, the same effects as in embodiment 14 can be obtained. [Explanation of symbols]

[0187] 1. Autonomous mobile robot 2 Vibration source 3. Vibration Sensor 4 aircraft 5. Switching section 6 Structures 7 Ball joint 8 Waterproof Boxes 9. Personal Computers 10 Non-destructive testing equipment 20 Control Unit 11 Data Processing Unit 12. First Training Phase Execution Unit 13. First Estimated Phase Execution Unit 14. Second Training Phase Execution Unit 15. Second Estimated Phase Execution Unit 40 Storage section 50 Communications Department 60 Input section 70 Output section 100 Information Processing Equipment (Cloud Server) 600 Non-destructive Testing Equipment System

Claims

1. A non-destructive testing device that performs non-destructive testing of structures including concrete using the impact elastic wave method, A vibration source that generates an input wave, A vibration sensor that detects the output wave, The aircraft equipped with the aforementioned vibration sensor, A non-destructive testing apparatus comprising: a switching unit that switches between (1) a first state in which the machine holds the vibration source and the vibration sensor, and (2) a second state in which the vibration source is installed on the surface of the structure and detached from the machine, and the machine holds the vibration sensor and presses the vibration sensor against the surface.

2. The aircraft is further equipped with a moving part for moving the aforementioned aircraft. The non-destructive testing apparatus according to claim 1.

3. The system further comprises a switching control unit, a movement control unit, a vibration source control unit, and a vibration sensor control unit, which control the switching unit, the movement unit, the vibration source, and the vibration sensor, respectively. The movement control unit moves the machine to the inspection position and then stops the machine, and the switching control unit switches from the first state to the second state. The vibration source control unit generates the input wave, and the vibration sensor control unit detects the output wave. The control unit switches from the second state to the first state by controlling the switching unit. The non-destructive testing apparatus according to claim 2.

4. An inspection location information storage unit that stores inspection location information representing each of multiple inspection locations, An inspection position designation unit that designates the next inspection position to be inspected from among the aforementioned plurality of inspection positions, A device position information acquisition unit that acquires device position information representing the location of the non-destructive testing device from a satellite positioning system, and further comprises, The movement control unit moves to the inspection position specified by the inspection position designation unit. The non-destructive testing apparatus according to claim 3.

5. The switching unit includes a support unit that moves between a first position with a higher height and a second position with a lower height. The support portion achieves the first state in the first position by supporting a part of the vibration source from below and separating the vibration source from the structure, and achieves the second state in the second position by installing the vibration source on the surface and separating the part of the vibration source from the structure. A non-destructive testing apparatus according to any one of claims 1 to 3.

6. The vibration source includes, as part of the above, a pair of protrusions that protrude from each of the pair of sides at a position higher than the center of gravity, The support portion is a pair of members whose spacing is wider than the spacing between the pair of sides of the vibration source and narrower than the spacing between the tips of the pair of protrusions. The non-destructive testing apparatus according to claim 5.

7. Each of the pair of members has the same shape and a first recess that opens more upward when viewed from above. The non-destructive testing apparatus according to claim 6.

8. Each of the pair of protrusions has the same shape and a second recess that opens more downward when viewed from a direction parallel to the main surface of the pair of members. The non-destructive testing apparatus according to claim 7.

9. The vibration source is, Excitation coil and A housing for the excitation coil, wherein when the vibration source is installed on the surface, the lower bottom surface of the housing faces and is in close proximity to the surface, A vibration transmitter connected to an excitation coil, wherein the tip of the vibration transmitter protrudes from the lower bottom surface, One or more legs that allow the housing to stand upright when the vibration source is installed on the surface, It is a vibrator equipped with, The one or more legs are made of an elastic material and are configured such that (1) when the tip is separated from the surface, the lower end is located below the tip, and (2) when the housing is installed on the surface and the tip is in contact with the surface, the lower end is in contact with the surface together with the tip. A non-destructive testing apparatus according to any one of claims 1 to 3.

10. The vibration sensor is fixed directly or indirectly to the support portion. The system further includes an elastic cushioning material interposed between the vibration sensor and the aircraft body. The non-destructive testing apparatus according to claim 5.

11. A ball joint interposed between the vibration sensor and the support portion, further comprising a ball joint directly or indirectly fixed to the support portion, The vibration sensor is fixed to the ball joint. The non-destructive testing apparatus according to claim 10.

12. A non-destructive testing apparatus according to any one of claims 1 to 3, An information processing device including an estimation unit, One or more terminals, A non-destructive testing system equipped with, The information processing device further comprises a communication unit that shares the degree of abnormality of the structure estimated by the estimation unit with one or more terminals, in a non-destructive testing system.

13. The aforementioned non-destructive testing is a non-destructive test using the local vibration testing method. The aforementioned information processing device is In the spectral response function representing the output wave, a specification unit identifies a reference frequency having predetermined characteristics, A normalization unit that normalizes the spectral response function in the frequency axis direction using the aforementioned reference frequency, The system further comprises an estimation unit that estimates the state of a structure by inputting the spectral response function obtained by the normalization unit into a machine learning model that takes the spectral response function as input data and outputs data used to estimate the state of a structure including concrete as output data. The non-destructive testing system according to claim 12.

14. The normalization unit normalizes the spectral response function in the amplitude axis direction as well. The non-destructive testing system according to claim 13.

15. A non-destructive testing method for performing non-destructive testing of structures including concrete using the impact elastic wave method, Using a machine that holds the vibration source and vibration sensor, At the measurement location, the vibration source is installed on the surface of the structure, and the vibration source is separated from the machine in a separation step, An inspection step in which an input wave is applied to the structure using the vibration source and an output wave is detected using the vibration sensor, Non-destructive testing methods including

16. A moving step performed before the separation step and the inspection step, the moving step of moving the machine, which is holding the vibration source and the vibration sensor, to the measurement position, A holding step performed after the separation step and the inspection step, comprising a holding step in which the machine body holds the vibration source, The non-destructive testing method according to claim 15, further comprising:

17. The aforementioned non-destructive testing is a non-destructive test using the local vibration testing method. A process for identifying a reference frequency having predetermined characteristics in the spectral response function representing the output wave, A normalization step of normalizing the spectral response function in the frequency axis direction using the aforementioned reference frequency, The system includes an estimation step which involves inputting the spectral response function obtained in the normalization step into a machine learning model that takes a spectral response function representing the output wave as input data and data used to estimate the state of a structure including concrete as output data, thereby estimating the degree of abnormality of the structure. The non-destructive testing method according to claim 15 or 16.

18. In the normalization step, the spectral response function is normalized in the frequency axis direction using the reference frequency, and the spectral response function is normalized in the amplitude axis direction. The non-destructive testing method according to claim 17.

Images