Metal body inspection device

Abstract

A metal body inspection device comprises: a magnetic detection unit including a magnetic sensor unit; a distance detection unit; and a controller connected to the magnetic detection unit and the distance detection unit, the metal body inspection device being installed on a mobile body that moves relative to a metal body to be inspected. The distance detection unit detects the movement distance of the mobile body relative to the metal body. The magnetic sensor unit includes a first oscillation coil, a second oscillation coil, and a reception coil, and is disposed so as to face the metal body. The reception coil is disposed between the first oscillation coil and the second oscillation coil. The first oscillation coil and the second oscillation coil are configured to generate alternating magnetic fields in opposite directions. The alternating magnetic fields are rectangular waves or ramp waves. The reception coil detects a leakage magnetic flux generated from the metal body excited by the alternating magnetic fields. Thus, the degree of damage in the depth direction of the metal body to be inspected can be detected with high accuracy.

Description

Metal object inspection device

[0001] The present disclosure relates to a metal object inspection device.

[0002] In rail transport, depending on the route, rail vehicles pass over rails as many as 20 times per hour. Rails used in this way are subjected to extremely heavy loads from the traveling rail vehicles, making them prone to defects such as cracks and damage. Therefore, from the perspective of maintaining safety and reliability in rail transport, there is a demand for rail flaw detection technology that can periodically and accurately inspect rails for cracks and other defects.

[0003] Patent Document 1 discloses a rail inspection device that generates inspection data related to deterioration of a railway rail, and includes a first oscillator coil and a second oscillator coil that are arranged on a surface facing the railway rail and generate alternating magnetic fields in opposite directions, and a receiver coil that is located between or near the first oscillator coil and the second oscillator coil and outputs a magnetic field waveform based on the magnetic field received from the first oscillator coil and the second oscillator coil as inspection data. Patent Document 1 also discloses a method of comparing the waveform of the cosine component of the inspection data with the waveform of the square root of the sum of the squares of the cosine component and the sin component of the inspection data based on the inspection data, and identifying parts where these waveforms differ as deteriorated parts of the rail.

[0004] JP 2017-20862 A

[0005] In the rail inspection device described in Patent Document 1, an oscillator coil generates a sine wave as an AC magnetic field.

[0006] Damage to rails for railway vehicles is a natural flaw, and therefore includes a mixture of fine cracks and deep cracks. Such damage to rails cannot be detected with high accuracy even using conventional techniques such as those described in Patent Document 1. In this case, the criteria for determining the level at which rail replacement is necessary must be set high. This necessitates frequent replacement of rails, which increases costs.

[0007] An object of the present disclosure is to detect with high accuracy the degree of damage in the depth direction of a metal object to be inspected.

[0008] The metal object inspection device of the present disclosure comprises a magnetic detection unit having a magnetic sensor unit, a distance detection unit, and a controller connected to the magnetic detection unit and the distance detection unit, and is installed on a mobile object that moves relative to a metal object to be inspected, wherein the distance detection unit detects the distance moved by the mobile object relative to the metal object, the magnetic sensor unit has a first oscillation coil, a second oscillation coil, and a receiving coil, and is arranged to face the metal object, the receiving coil is arranged between the first oscillation coil and the second oscillation coil, and the first oscillation coil and the second oscillation coil are configured to generate alternating current magnetic fields in mutually opposite directions, the alternating current magnetic field being a square wave or a ramp wave, and the receiving coil detects leakage magnetic flux generated from the metal object excited by the alternating current magnetic field.

[0009] According to the present disclosure, the degree of damage in the depth direction of the metal object to be inspected can be detected with high accuracy.

[0010] 7A is a schematic diagram showing a rail inspection device according to a first embodiment of the present disclosure. FIG. 7B is a perspective view showing the magnetic detection unit 2 of FIG. 1. FIG. 7C is a bottom view showing the magnetic detection unit 2 of FIG. 1. FIG. 7D is a partially cutaway plan view showing the magnetic detection unit 2 of FIG. 1. FIG. 7E is a schematic diagram showing an operating state of one of the magnetic sensor units 21A, 21B, and 21C of FIG. 3. FIG. 7F is a schematic diagram showing an operating state when one of the magnetic sensor units 21A, 21B, and 21C of FIG. 3 passes above a singular location on a rail 200. FIG. 7C is an explanatory diagram showing processes 1 to 3 of the monitoring information extraction and analysis process using a continuous wavelet transform in the rail inspection device according to the first embodiment. FIG. 7D is an explanatory diagram showing processes 4 to 6 following process 3 of FIG. 7A. FIG. 7E is an explanatory diagram showing a monitoring information extraction and analysis process using a short-time Fourier transform in the rail inspection device according to the first embodiment. FIG. 7F is a table showing an example of data 222 obtained in process 2 of FIG. 7A. FIG. 7F is a table showing an example of data 333 obtained in process 3 of FIG. 7A. FIG. 7B is a flowchart showing a discrete wavelet transform performed before the continuous wavelet transform of process 5 shown in FIG. 11A. FIG. 11B is a flowchart showing hierarchical decomposition of the detection signal g(t) shown in FIG. 9. FIG. 11C is a block diagram showing the overall configuration of the rail inspection device 1 of the first embodiment. FIG. 11D is a graph showing an example of a digital oscillation signal output from the oscillator 33 of FIG. 11A. FIG. 11E is a graph showing another example of a digital oscillation signal output from the oscillator 33 of FIG. 11A. FIG. 11F is a block diagram showing the detection unit 34 of FIG. 11A. FIG. 11G is a flowchart showing rail monitoring data processing executed by the data processing unit 43 of the evaluation unit 7 of FIG. 11A. FIG. 11H is a bottom view showing the magnetic detection unit 2 of the second embodiment. FIG. 11H is a partially cutaway plan view showing the magnetic detection unit 2 of the second embodiment. FIG. 11I is a block diagram showing the overall configuration of the rail inspection device 1 of the second embodiment.

[0011] Hereinafter, an embodiment will be described with reference to the drawings. In the following embodiment, a rail inspection device that inspects rails used in railway vehicle tracks as the metal object to be inspected is described as an example of a metal object inspection device according to the present disclosure. However, the metal object inspection device is not limited to this example, and the inspection object can also be a metal object having a surface shape made of twisted wires called strands, such as wire ropes used in elevators, cranes, etc.

[0012] First Embodiment <External Configuration of First Embodiment> FIG. 1 is a schematic diagram showing a rail inspection device according to a first embodiment of the present disclosure.

[0013] In Figure 1, a rail inspection device 1 (metal object inspection device) includes a magnetic detection unit 2, a distance detection unit 3, a processing unit 4 (controller), a power supply 88, and connection cables 60 and 61. The rail inspection device 1 is installed on a mobile object 100. The mobile object 100 includes wheels 300 for traveling on a rail 200. The wheels 300 rotate while contacting a rail tread 200a. The mobile object 100 is, for example, a railway vehicle. The object to be inspected (the object to be monitored) by the rail inspection device 1 is the rail 200. The magnetic detection unit 2 and the processing unit 4 are connected by a connection cable 60. The distance detection unit 3 and the processing unit 4 are connected by a connection cable 61.

[0014] The magnetic detection unit 2 is installed on the outside of the bottom surface of the mobile body 100, in a position facing the rail tread surface 200a of the rail 200. The distance detection unit 3, processing unit 4, and power supply 88 are installed inside the mobile body 100. The power supply 88 supplies drive power to the magnetic detection unit 2, distance detection unit 3, and processing unit 4, causing them to operate.

[0015] When the moving body 100 travels on the rails 200, the magnetic detection unit 2 detects the state of the rails 200. A signal representing the detection result is sent to the processing unit 4 via the connection cable 60.

[0016] The rotation axis of the wheels 300 of the moving body 100 is the axle 400. The axle 400 is connected to the distance detection unit 3 via a tachometer generator 500. The operation of the tachometer generator 500 is detected by the distance detection unit 3. The processing unit 4 then collects signals from the magnetic detection unit 2 and the distance detection unit 3 and performs processing such as analysis and display using the signals. In other words, the processing unit 4 monitors the state of the rails 200. Note that if an AC power supply is provided within the moving body 100, the power supply 88 may be a stabilized power supply circuit driven by the AC power supply. If an AC power supply is not available, the power supply 88 may be an uninterruptible power supply unit with a built-in battery.

[0017] Next, the structure of the magnetic detection unit 2 will be described.

[0018] FIG. 2 is a perspective view showing the magnetic detection unit 2 of FIG.

[0019] As shown in Fig. 2, the magnetic detection unit 2 has an integrated structure in which an upper housing 26 and a lower housing 20 are fixed vertically. A flange 25 is provided on the upper side of the upper housing 26. The upper housing 26 and the lower housing 20 have a rectangular parallelepiped shape. A connector 28 is provided on one side of the upper housing 26. A connector 62 provided on one end of a connection cable 60 is configured to be connected to the connector 28. By connecting the connector 62 and the connector 28, the magnetic detection unit 2 and the processing unit 4 (Fig. 1) can communicate with each other via the connection cable 60.

[0020] The connectors 28 and 62 have a dustproof and waterproof structure. This structure prevents foreign matter such as dust or water from entering through the gap when connecting the connectors 28 and 62, which could cause a malfunction of the magnetic detection unit 2 or a poor connection of the connection cable 60. A through-type connector is suitable for the connector 28, which can be inserted into the connector 62 and tightened to prevent detachment or loosening due to vibration and prevent the intrusion of dust or water. However, a connector other than a through-type connector may be used as long as it has a structure that can prevent foreign matter such as dust or water from entering the interior of the magnetic detection unit 2 at the position where the magnetic detection unit 2 is attached.

[0021] Through holes 25a are formed in the four corners of the flange 25. Furthermore, the movable body 100 (see FIG. 1) has screw holes (not shown) at positions facing the through holes 25a at the installation location of the magnetic detection unit 2. Then, a bolt (not shown) is inserted into the through hole 25a via a washer (not shown), and the bolt is tightened into a screw hole provided in the movable body 100, thereby fixing the magnetic detection unit 2 to a predetermined position of the movable body 100.

[0022] The lower housing 20 is made of a non-magnetic material and incorporates a magnetic sensor group 21 that detects the state of the rail 200. The upper housing 26 incorporates a preamplifier 210 that amplifies the magnetic signals detected by the magnetic sensor group 21. The magnetic sensor group 21 and the preamplifier 210 will be described in detail later.

[0023] Because the magnetic detection unit 2 is fixed to the mobile object 100, measures are required to protect it from vibrations and shocks caused by the rocking motion of the mobile object 100 as it moves on the rails 200. Therefore, the internal spaces of the upper housing 26 and the lower housing 20 are molded with a resin by injecting a filler (not shown) having insulating and non-magnetic properties. This structure reduces the vibrations and shocks that the magnetic sensor unit group 21 and the preamplifier unit 210 receive, preventing damage to the magnetic sensor unit group 21 and the preamplifier unit 210 and preventing deviation of the detection position of the magnetic sensor unit group 21. In the example shown in FIG. 2 , the upper housing 26 and the lower housing 20 are rectangular parallelepipeds, but these shapes may be other shapes, such as a cube, a cylinder, or a prism, depending on conditions such as the installation space of the mobile object 100 to be used.

[0024] In this embodiment, three magnetic sensor units are installed in one lower housing.

[0025] FIG. 3 is a bottom view showing the magnetic detection unit 2 of FIG.

[0026] In FIG. 3, the lower housing 20 and flange 25 of the magnetic detection unit 2 are shown.

[0027] The lower housing 20 has magnetic sensor units 21A, 21B, and 21C. Each of the magnetic sensor units 21A, 21B, and 21C has an oscillator coil 5A, 5B, and a receiver coil 6. Each of these coils is fabricated by winding a coated copper wire. In each of the magnetic sensor units 21A, 21B, and 21C, the receiver coil 6 is disposed between the oscillator coil 5A and the oscillator coil 5B. Here, the magnetic sensor unit 21A will also be referred to as the "first magnetic sensor unit," the magnetic sensor unit 21B as the "second magnetic sensor unit," and the magnetic sensor unit 21C as the "third magnetic sensor unit." The oscillator coil 5A will also be referred to as the "first oscillator coil," and the oscillator coil 5B as the "second oscillator coil."

[0028] The lower housing 20 and the lower surface of the flange 25 are rectangular. The magnetic detection unit 2 is disposed so that the long side of the rectangle is parallel to the direction of movement of the moving body 100.

[0029] The bottom surfaces of the magnetic sensor units 21A, 21B, and 21C are also rectangular. The magnetic sensor unit 21A is disposed so that the long side of the rectangle is parallel to the direction of movement of the mobile object 100. The magnetic sensor unit 21B is disposed so that the long side of the rectangle is perpendicular to the direction of movement of the mobile object 100. The magnetic sensor unit 21C is disposed so that the long side of the rectangle diagonally intersects with the direction of movement of the mobile object 100.

[0030] Here, "oblique" means that the angle α between two straight lines (line segments) is greater than 0 degrees and less than 90 degrees. In other words, this means that the following inequality (unit: degrees) holds true:

[0031] 0<α<90 It is preferable that 20≦α≦70 be satisfied, and it is even more preferable that 30≦α≦60 be satisfied.

[0032] An AC current of a predetermined oscillation frequency f (predetermined frequency) flows from the processing unit 4 (see FIG. 1) through the connection cable 60 to the oscillator coils 5A and 5B of the magnetic sensor units 21A, 21B, and 21C, respectively. As a result, an AC magnetic field is generated from each of the oscillator coils 5A and 5B, exciting the rail 200. A magnetic flux generated from the excited rail 200 generates an induced voltage in the receiver coil 6.

[0033] 3 shows a case where three magnetic sensor units 21A, 21B, and 21C are installed, the metal object inspection device according to the present disclosure is not limited to this and includes devices with one, two, and four or more magnetic sensor units. In the second embodiment described below, a case where N magnetic sensor units are installed is shown.

[0034] The orientations of the multiple magnetic sensor units can be summarized as follows.

[0035] The direction of a first straight line connecting the center of gravity of the first oscillator coil and the center of gravity of the second oscillator coil for a first magnetic sensor unit, which is one of three specified magnetic sensor units among the multiple magnetic sensor units, is arranged parallel to the direction of movement of the mobile body, the direction of a second straight line connecting the center of gravity of the first oscillator coil and the center of gravity of the second oscillator coil for a second magnetic sensor unit, which is one of the three magnetic sensor units, is arranged perpendicular to the first straight line of the first magnetic sensor unit, and the direction of a third straight line connecting the center of gravity of the first oscillator coil and the center of gravity of the second oscillator coil for a third magnetic sensor unit, which is one of the three magnetic sensor units, is arranged diagonally intersecting the first straight line of the first magnetic sensor unit.

[0036] Furthermore, at least two predetermined ones of the plurality of magnetic sensor units are arranged so that the directions of the straight lines connecting the centers of gravity of the first oscillator coil and the second oscillator coil for the two magnetic sensor units are different from each other.

[0037] FIG. 4 is a partially cutaway plan view showing the magnetic detection unit 2 of FIG.

[0038] 4 shows the configuration of the preamplifier unit 210 installed inside the upper housing 26. The preamplifier unit 210 has three amplification filter units 22A, 22B, and 22C and a printed circuit board 210a on which they are mounted. The amplification filter units 22A, 22B, and 22C perform amplification and filtering on the induced voltages generated in the receiving coils 6 (see FIG. 3) installed in the magnetic sensors 21A, 21B, and 21C, respectively. The results of this processing are transmitted to the processing unit 4 via the connection cable 60 (see FIG. 1). The processing unit 4 analyzes the received signals and detects magnetic signals generated from the rail 200.

[0039] The upper housing 26, which houses the preamplifier unit 210, is made of a metal such as aluminum, which functions as an electromagnetic shield. By making the upper housing 26 out of a metal, it is possible to shield the electromagnetic field generated by the preamplifier unit 210 and suppress electromagnetic waves from entering the preamplifier unit 210 from the outside in the usage environment. It is also desirable to manufacture the upper housing 26 and the flange 25 as a single unit by metal cutting. This is because the number of components of the magnetic detection unit 2 is reduced and strength is improved.

[0040] Furthermore, when an aluminum alloy is used for the upper housing 26, it is desirable to perform an anodizing treatment or the like to form an oxide film on the surface of the aluminum alloy in order to increase the surface hardness, improve corrosion resistance, and provide insulation properties.

[0041] It is desirable to use a non-magnetic material for the lower housing 20. This is because the lower housing 20 houses the oscillator coils 5A and 5B that generate an AC magnetic field and the receiver coil 6 that detects changes in the magnetic field generated by the rail 200 (metal body). Furthermore, since the magnetic detection unit 2 is installed outside the mobile body 100 and is used under conditions where it is subjected to wind pressure and collisions with muddy water, dust, and the like during the movement of the mobile body 100, it is preferable to use a material with excellent toughness and corrosion resistance, such as fiber-reinforced resin, for the lower housing 20. For example, when using glass fiber-reinforced plastic (GFRP), it is even more preferable to apply a coating film treatment to the surface of the lower housing 20 using a paint made by combining a mixture of liquid unsaturated polyester resin and a hardener with glass fiber or the like. This coating film treatment forms a waterproof coating layer, further improving the weather resistance of the lower housing 20.

[0042] The lower housing 20 is provided with a screw-type insert nut (not shown) on the side of the contact surface with the upper housing 26, and is fixed by screwing to the upper housing 26. The upper housing 26 and the lower housing 20 are bonded at their contact surfaces with a silicone contact agent that has excellent weather resistance, and gaps are sealed to form an integrated unit.

[0043] The connection cable 60 connecting the magnetic detection unit 2 and the processing unit 4 includes a power line for the preamplifier unit 210, an output signal line for the receiving coil 6 (see FIG. 3 ) via the amplification filter units 22A, 22B, and 22C, and an input signal line for an excitation signal for the oscillator coils 5A and 5B (see FIG. 3 ). If the connection cable 60 is subjected to noise caused by electromagnetic waves traveling through space, this will affect the detection of the magnetic signal from the rail 200. For this reason, it is preferable to use a multi-pair shielded cable with excellent noise resistance, such as a twisted-pair shielded cable, for the connection cable 60. Furthermore, as a countermeasure against crosstalk within the connection cable 60, the connection cable 60 may be divided into two cables. Specifically, one cable is a multi-pair shielded cable including a power line for the preamplifier unit 210 and an output signal line from the amplification filter units 22A, 22B, and 22C (a wiring for transmitting an amplified output signal of the receiving coil 6 to the outside). The other cable is another multi-pair shielded cable including input signal lines for excitation signals to the oscillation coils 5A and 5B.

[0044] The connection cable 60 is fixed to the underside of the mobile body 100 facing the rail 200, and is connected to the processing unit 4 via an inlet provided in the mobile body 100. The connection cable 60 is firmly fixed and held in place using fasteners such as cable ties to prevent loosening of the underside of the mobile body 100, in order to prevent interference with the movement of the mobile body 100 and noise generation due to the vibration of the connection cable 60. Furthermore, when fixing and holding the connection cable 60, it is desirable to design a wiring path that makes the length of the connection cable 60 as short as possible. The shorter this wiring path, the more effectively it is possible to avoid the risk of noise being mixed into the detection signal from the magnetic detection unit 2 via the connection cable 60.

[0045] The distance detection unit 3 (see FIG. 1) that detects the moving distance of the moving body 100 will be described.

[0046] The distance detection unit 3 is a rotary encoder. As the rotary encoder's rotation shaft rotates, a pulse signal is generated. By counting this pulse signal, the amount of movement associated with the rotation of the rotation shaft can be calculated. The rotation of the axle 400 of the wheel 300 of the mobile object 100 is transmitted to the rotation shaft of the rotary encoder (distance detection unit 3) via a tachometer generator 500. The tachometer generator 500 includes pulleys and a timing belt attached to the axle 400 and the rotation shaft of the rotary encoder. In other words, when the mobile object 100 moves, the axle 400 rotates in accordance with the movement of the wheel 300, and this movement is transmitted to the tachometer generator 500, causing the rotation shaft of the rotary encoder to rotate simultaneously.

[0047] The pulse signal output from the distance detection unit 3 is sent to the processing unit 4 via the connection cable 61. The processing unit 4 receives the pulse signal in synchronization with the output signal from the magnetic detection unit 2. As with the connection cable 60, the connection cable 61 is preferably a multi-pair shielded cable with excellent noise resistance, such as a twisted pair shielded wire.

[0048] The rotary encoder described above employs an incremental method that counts and measures pulses generated in response to rotations from a measurement start point, but may also employ an absolute method that measures the absolute angular position of the rotary encoder's rotating shaft. Measurement methods for rotary encoders include an optical method in which light is irradiated onto an internal slitted disk to obtain rotational position information as an optical pulse signal that passes through the slitted disk; a magnetic method in which rotational position information is obtained from a rotating disk with a magnetic pattern formed thereon and a pulse signal is output from periodic changes in a magnetic field; an electrostatic method in which a transmitter, receiver, and rotor are fixed inside the encoder and the change in electrostatic capacitance caused by the rotor's rotation is output as distance information; and a contact method in which electrical signals generated by contacts inside the encoder as the rotating shaft rotates are converted into the amount of rotation. Any of these methods may be used.

[0049] <Principle of detecting a peculiar point on a rail, etc.> Figure 5 is a schematic diagram showing the operating state of one of the magnetic sensor units 21A, 21B, and 21C in Figure 3. In the figure, the direction of movement of the moving body, i.e., the direction of movement of the magnetic detection unit, is defined as the "x direction." Furthermore, the direction perpendicular to the underside of the receiving coil 6, i.e., the vertically upward direction on the rail tread surface 200a of the rail 200, is defined as the "y direction."

[0050] As shown in the figure, when current flows from the processing unit 4 (see FIG. 1) in the oscillator coil 5A and the oscillator coil 5B, a synchronized, phase-inverted AC magnetic field is generated. Specifically, the oscillator coil 5A and the oscillator coil 5B are connected in series (or parallel) with the start or end of the coated copper wire windings connected to each other. With this configuration, when an AC voltage is applied from the processing unit 4 to the oscillator coil 5A and the oscillator coil 5B, an AC magnetic field is generated. Then, when the magnetic fluxes ΦA and ΦB generated from the oscillator coil 5A and the oscillator coil 5B are transmitted to the rail tread 200a of the rail 200, a magnetic flux flow is generated on the rail tread 200a.

[0051] This figure shows an example of a case where there are no peculiar points on the rail 200. Here, the absence of peculiar points refers to the absence of defects such as cracks, damage, etc., and the absence of structural changes in the areas where the rails 200 are connected (for example, joints, welds, bolt installation locations, or other structural changes that are not damage to the rail 200).

[0052] The components of the magnetic flux linking the receiving coil 6 are cancelled out because the magnetic flux ΦA and the magnetic flux ΦB are in opposite directions, and depend on the strength balance of the magnetic flux ΦA and ΦB. Therefore, if there is no singularity in the rail 200, the magnetic flux linking the receiving coil 6 will be almost zero, and the induced voltage of the receiving coil 6 will also be almost zero.

[0053] FIG. 6 is a schematic diagram showing an operating state when one of the magnetic sensor units 21A, 21B, and 21C in FIG. 3 passes above a singular point on the rail 200. In FIG.

[0054] In this figure, a peculiar point 202 exists on a rail 200. Here, the peculiar point 202 includes defects such as cracks, damage, structural changes such as at the portions where the rails 200 are connected to each other, etc. In other words, the peculiar point 202 on the rail 200 is a point that has magnetic properties different from those of a normal rail 200. Therefore, the peculiar point 202 can be broadly divided into "damaged points" where defects such as cracks or damage exist, and "structural points" where structural changes exist such as at the portions where the rails 200 are connected to each other.

[0055] In the example shown in this figure, the magnetic flux ΦB of the oscillator coil 5B is affected by the singular point 202, causing the magnetic flux to leak from the rail tread 200a. Therefore, the change in the induced voltage of the receiver coil 6 when the receiver coil 6 passes over the singular point 202 is significantly larger than when there is no singular point as shown in Figure 5.

[0056] The rail inspection device 1 (see Fig. 1) of this embodiment detects the generated leakage magnetic field by utilizing the phenomenon in which the flow of magnetic flux generated in the rail 200, which is the object to be inspected, changes in the vicinity of a singular point 202 as shown in Fig. 6. As an analytical model for this leakage magnetic field, the leakage magnetic field generated in space can be expressed based on a dipole model.

[0057] As an example, let us consider a structure in which the singular point 202 in the model is a joint in the rail 200 that involves a gap.

[0058] The leakage magnetic field component generated from the y direction of the rail 200 exhibits bipolar magnetic flux changes with extreme values ​​(maximum and minimum values) in the x direction, with the center position of the joint as an inflection point when the mobile body 100 moves.

[0059] In the above example, a case has been described in which the singular location is a structure that is a joint with a gap in the rail 200, but the singular location 202 can also be detected in a similar manner if it is damaged. When a crack that is damage exists on the rail tread 200a of the rail 200, the magnetic permeability of the gap at the damaged location differs significantly from the magnetic permeability of the rail 200, causing magnetic resistance and disrupting the flow of magnetic flux. This disruption in the flow of magnetic flux causes the magnetic flux flowing on the rail tread 200a to bypass the crack, generating leakage magnetic flux from the rail tread 200a. Therefore, when the magnetic detection unit 2 passes over the damaged location, the strength balance between the magnetic flux ΦA and the magnetic flux ΦB in the rail 200 is disrupted (see FIG. 6 ), similar to when detecting the above-mentioned structure, and the magnetic detection unit 2 can detect the presence of the damaged location.

[0060] <Rail Monitoring> Monitoring of the rail 200 by the rail inspection device 1 will be described with reference to FIGS. 7A and 7B.

[0061] FIG. 7A is an explanatory diagram showing processes 1 to 3 relating to the monitoring information extraction and analysis process using continuous wavelet transform in the rail inspection device of this embodiment.

[0062] FIG. 7B is an explanatory diagram showing processes 4 to 6 following process 3 in FIG. 7A.

[0063] Note that the processes 2 to 6 shown in FIGS. 7A and 7B are performed in the processing unit 4 (see FIG. 1).

[0064] 7A, the detection signal acquired by the magnetic detection unit 2 is time-series data 111 representing the time series of the detection signal strength (processing 1). In this case, the response waveform of the detection signal changes depending on the moving speed of the moving object 100. In other words, the response waveform has moving speed dependency.

[0065] Specifically, the response time is longer when the moving speed is slower, whereas the response time is shorter when the moving speed is fast. Therefore, the response waveform at a singular location does not always have the same shape. Therefore, even if signals from the same singular location are received, it is difficult to detect them as the same singular location. Furthermore, when noise is mixed into the detection signal, the settings of the filter applied to remove the noise must be changed depending on the moving speed, which makes the application complicated.

[0066] Therefore, the detection signal of the magnetic detection unit 2 is transmitted to the processing unit 4 in synchronization with the output signal from the distance detection unit 3, and the processing unit 4 converts the time-series data 111 of the detection signal into data 222 (correlation data) indicating changes in the detection signal strength with respect to the moving distance (Process 2). If the moving speed of the moving body 100 is constant, sampling of this data 222 can be performed at equal moving distance intervals (i.e., at equal intervals). However, in reality, since the moving speed is subject to acceleration and deceleration of the moving body 100, the sampling intervals are not always constant and are not equal.

[0067] Therefore, the processing unit 4 converts the data 222 obtained in process 2 into data 333 with equally spaced sampling intervals by interpolating the sampling intervals (process 3). The sampling intervals in the data 333 are set to a constant interval of 1 mm, for example. In this case, it is desirable to set the set value to a value that matches the resolution required for monitoring the rail 200 to be inspected.

[0068] Here, an example of sampling conversion accompanying interpolation of detection signals relative to movement distance will be described.

[0069] FIG. 8A is a table showing an example of data 222 obtained in process 2 of FIG. 7A.

[0070] FIG. 8B is a table showing an example of data 333 obtained in process 3 of FIG. 7A.

[0071] 8A shows data in which an ID (identification number) is assigned to each moving distance in process 2 (see FIG. 7A) using signals acquired from the magnetic detection unit 2 and the distance detection unit 3 (see FIG. 1), and the ID, moving distance, and detected signal strength for each moving distance are associated with each other. In this data, the distance intervals are not equal because they depend on the moving speed of the moving object 100.

[0072] Therefore, in process 3 (see FIG. 7A), the detection signal strength of the distance is calculated by interpolation (e.g., linear interpolation) according to the desired sampling interval (e.g., the sampling interval represented by a set value). FIG. 8B shows data calculated by process 3, in which the interpolation process ID, the uniform sampling movement distance, and the detection signal strength for this movement distance correspond to each other. This figure shows an example of data sampled at uniform intervals of 1 mm.

[0073] In this case, for example, the detection signal strength at a distance of 1 mm shown in FIG. 8B (interpolation process ID is 2) calculated from the stored data shown in FIG. 8A is obtained by linear interpolation of the data (IDs are 2 and 3) of the detection signal strength between two points at distances of 0.5 mm and 1.3 mm shown in FIG. 8A, which are located before and after the distance of 1 mm shown in FIG. 8B.

[0074] This linear interpolation is performed by using two points (X ｉ , Y ｉ ) and (X ｉ＋１ , Y ｉ＋１ ) and the approximate value of the function Y for any point X is expressed as a linear function of X by connecting the two points with a straight line as follows:

[0075] Y ａ = Y ｉ + (Y ｉ＋１ -Y ｉ ) × (X ａ -X ｉ ) / (X ｉ＋１ -X ｉ ) where X ｉ ≦X ａ ≦X ｉ＋１ Let us assume that the following holds true.

[0076] 8B shows interpolated data of the detected signal strength for a uniform sampling distance at intervals of 1 mm for the moving distance. Note that the intervals of the moving distance in the interpolation process are not limited to 1 mm, and can be set to any value by the processing unit 4.

[0077] The data 333 obtained in process 3 shown in Fig. 7A has a uniform sampling interval, which allows a spatial filter to be applied to the detection signal.

[0078] The spatial filter is the band-pass filter 444 of process 4 shown in Fig. 7B. The band-pass filter 444 is made up of a high-pass filter and a low-pass filter.

[0079] It is possible to extract only the signal components necessary for extracting the anomalous location by using the band-pass filter 444. For example, a signal change over a distance greater than the size of the magnetic detection unit 2 that performs signal detection, or a signal change over a distance smaller than the size of the receiving coil 6 built into the magnetic detection unit 2, can be considered to be electrical noise caused by the shaking of the mobile object 100 or disturbances during measurement.

[0080] By using the band-pass filter 444 in the process 4, it is possible to remove offset fluctuations and the like from the detection signal, and thus detection signal data 555 obtained by cleansing the data 333 is obtained.

[0081] 7B shows an example of the bandpass filter 444 used in Process 4, which extracts detection signal components for a distance range of 20 mm to 1 m from the cutoff spatial frequency. The cutoff spatial frequency can be freely set depending on the rail to be inspected. In the spatial filter processing in Process 4, data 333 represents the change in detection signal strength relative to the moving distance, so the detection signal can be handled without being affected at all by the moving speed of the moving object 100.

[0082] The detection signal data 555 cleansed by process 4 is subjected to extraction and analysis of the spectral response of the detection signal in the selected test section 666 (process 5 shown in FIG. 7B). A continuous wavelet transform is applied to the spectral response of the test section 666.

[0083] The singular points on the rail 200 are not essentially periodic and do not occur continuously, so the detection signal does not occur as a periodic change but is observed as a sudden change.

[0084] The continuous wavelet transform allows for a variable area to be used to extract the detection signal from the travel distance domain, allowing for processing of the gradual fluctuation components of the detection signal over a long analysis area and processing of the fine fluctuation components of the detection signal over a short analysis area, making it possible to effectively grasp the fluctuation characteristics of the non-stationary or transient detection signal that occur at specific locations.

[0085] In the continuous wavelet transform of process 5 shown in FIG. 7B, Ω is the spatial frequency resolution defined by the wavelet function used, and Δ is the distance resolution defined by the wavelet function used.

[0086] In the continuous wavelet transform, the detection signal (defined as g(t)) subjected to data cleansing in process 4 is converted into a wavelet function ψ ａ，ｂ A continuous wavelet transform T(a, b) is performed by convolution integral at (t). This continuous wavelet transform T(a, b) is expressed by the following equation (1).

[0087]

[0088] Continuous wavelet function ψ ａ，ｂ (t) is expressed by the following formula (2).

[0089]

[0090] The following equation (3) used in the above equation (1) is a continuous wavelet function ψ ａ，ｂ is the complex conjugate of (t), and ψ denotes the base of the wavelet function called the mother wavelet.

[0091]

[0092] From T(a, b), a scale parameter a having information related to the spatial frequency and a shift parameter b having information related to the movement distance can be obtained, and the magnitude of the signal component at each spatial frequency for each position in the movement distance can be determined in the data-cleansed detection signal data 555 (see FIG. 7B ). Typical continuous wavelet functions include the Mexican hat function, the Meyer function, the Morlet function, and the Symlet function. Any of these functions can be used and may be selected depending on the response of the rail to be inspected. For example, when the Mexican hat function is used as the continuous wavelet function, the Mexican hat function has a shape obtained by second-order differentiation of a Gaussian function, and the mother wavelet ψ is expressed by the following equation (4):

[0093]

[0094] In addition, the wavelet function is expressed as the following equation (5).

[0095]

[0096] In addition to the above functions, any waveform shape can be used for the continuous wavelet function. If the shape of the response waveform to be obtained from the rail 200 is known in advance, a wavelet function with a shape similar to that waveform can be created and wavelet transform performed in the processing unit 4. In continuous wavelet transform, two parameters, a and b, are defined for processing. a is a coefficient for expanding and contracting the wavelet function; increasing the value of a stretches the wavelet function in the distance axis direction, and decreasing the value shrinks the wavelet function in the distance axis direction. b is a coefficient for shifting the wavelet function in the distance axis direction.

[0097] In the continuous wavelet transform (process 5 shown in FIG. 7B) shown as an example, the parameter a is set to large, medium, or small (a １ , a ２ , a ３ ) is shown.

[0098] Using the information obtained by the wavelet transform in process 5 shown in Fig. 7B, a graph 777 is created (process 6 shown in Fig. 7B) that displays a color contour showing a signal intensity bar of the detection signal. In graph 777, the vertical axis represents spatial frequency and the horizontal axis represents the moving distance of moving object 100.

[0099] Since the spatial frequency components of the detection signal differ depending on the type of anomalous location, it is possible to monitor including information about the anomalous location from the spectral response (pattern, spatial frequency range in which the signal occurs, etc.) at the anomalous location extracted using the color contour display. This color contour display makes it possible to visualize the anomalous location by using the spectral response in wavenumber space of the detection signal data 555 that has been data cleansed using a spatial filter (band pass filter 444), which is also effective in terms of usability.

[0100] For example, a structure exists over a relatively wide area and therefore exhibits a spectral response in which the signal intensity increases in the low spatial frequency range (see reference numeral 888), whereas damage such as a crack exists locally and therefore exhibits a spectral response in the high spatial frequency range (see reference numeral 999).

[0101] Furthermore, when the rail being inspected is an infrastructure facility, welded joints and structures for maintaining the facility are located at specific intervals, while abnormal areas on the rail due to damage are likely to occur irregularly. Therefore, it is possible to detect (distinguish) structures and damage from differences in spectral response in wavenumber space and the regularity of response patterns that occur in the inspection section.

[0102] The monitoring information extraction and analysis process using the continuous wavelet transform in the rail inspection device described above has been explained using FIGS. 7A and 7B, but some of the process may be replaced with other processes.

[0103] The detection signal of the magnetic detection unit 2 is subjected to data cleansing by spatial filtering in process 4, but there are cases where observation noise originating from the device or usage environment cannot be efficiently removed. In such cases, appropriate processing is required to remove the noise contained in the detection signal obtained by process 4.

[0104] FIG. 9 is a flowchart showing the discrete wavelet transform performed before the continuous wavelet transform of process 5 shown in FIG. 7B.

[0105] The discrete wavelet transform shown in FIG. 9 separates the detected signal into a linear combination of different resolution components.

[0106] In this discrete wavelet transform, the scale parameter a in the continuous wavelet transform is set to 2 ｊ and the shift parameter b is 2 ｊ In this binary division process, the discrete wavelet function is expressed by the following equation (6).

[0107]

[0108] In the above formula (6), j is called a level and represents a doubling of the enlargement or reduction. Note that j and k are integers. The detection signal (above g(t)) from the magnetic detection unit 2, which has undergone data cleansing by spatial filtering in process 4, can be expanded into a series based on a discrete wavelet function, and can be separated as the value of j increases. Note that, hereinafter, g(t) corresponding to level j will be referred to as g ｊ It is expressed as (t).

[0109] The series expansion means a process of separating a signal, and the detected signal is expressed by the following equation (7).

[0110]

[0111] The following equation (8) used in the above equation (7) is a discrete wavelet coefficient.

[0112]

[0113] The discrete wavelet transform S(j, k) is expressed by the following equation (9) in the same format as the continuous wavelet processing.

[0114]

[0115] The above series expansion corresponds to a process of approximation using a linear combination of wavelet functions, and the approximation accuracy of the detection signal g(t) obtained by this approximation process depends on the level, with the approximation becoming coarser as the level value increases. That is, the approximation function of the level j of the signal g(t) is expressed as a series by the following equation (10). The detection signal g(t) can be obtained with the highest accuracy when the level is 0.

[0116]

[0117] The detection signal g(t) is a continuous signal, and is sampled at regular intervals (for example, at intervals of 1 mm for the moving distance) to obtain N values ​​(a １ , ..., a Ｎ ) is obtained. Here, when performing the discrete wavelet processing, the number of data N is a power of 2, and N=2 ｎ The detection signal g(t) is approximated hierarchically.

[0118] A specific example of processing using the above equations (6) to (10) will be described with reference to FIG.

[0119] In this figure, the number of data N when the detection signal g(t) is level 0 is set to 8 (=2 ３ ) is used as an example. This hierarchical approximation is performed using eight data (a １ , ..., a ８ ) is a process of replacing every two consecutive data values ​​with their average value. １ , ..., a ８ ) is four data ((a １ +a ２ ) / 2, (a ３ +a ４ ) / 2, (a ５ +a ６ ) / 2, (a ７ +a ８ ) / 2). Next, a similar process is performed in which the value of every two consecutive data points in the replaced four data points is replaced with the average value of those data points.

[0120] When this hierarchical approximation is performed, the detection signal g(t) is decomposed hierarchically by calculating data h(t) obtained by subtracting data before and after the approximation.

[0121] FIG. 10 is a flowchart showing the hierarchical decomposition of the detection signal g(t) shown in FIG.

[0122] The data at level 0 is g ０ (t) is the approximate data g １ (t) to h １ (t) is obtained. Next, the data at level 1, g １ (t) to h ２ (t) can be decomposed. Similarly, at level 2, g ０ (t) is g ３ (t) and h ３ (t) where the original data g ０ (t) is g ３ (t) and h １ (t) and h ２ (t) and h ３ (t). Therefore, noise removal is performed by calculating the discrete wavelet coefficients of the detection signal g(t) under conditions for each level value within a desired range, and smoothing the calculated discrete wavelet coefficients using a predetermined threshold T. Threshold processing used in this noise removal includes hard threshold processing and soft threshold processing. The wavelet coefficients used in hard threshold processing are expressed by the following equation (11):

[0123]

[0124] When hard thresholding is performed, a decision is made to either keep or remove discrete wavelet coefficients.

[0125] On the other hand, in soft thresholding, the discrete wavelet coefficients are considered to contain both signal and noise, and the noise portion is removed from all discrete wavelet coefficients to separate the signal. The wavelet coefficients used in this soft thresholding are expressed by the following equation (12):

[0126]

[0127] Here, all discrete wavelet coefficients less than the threshold T are set to 0, and all discrete wavelet coefficients having a magnitude equal to or greater than the threshold T are reduced toward 0 by T. Note that the thresholding process is performed byｊ The threshold value T is used for thresholding. The threshold value T at level j is ｊ is expressed by the following formula (13).

[0128]

[0129] Here, n ｊ is the length (number) of discrete wavelet coefficients at level j, and σ ｊ is the target detection signal g ｊ is the standard deviation of the noise contained in (t). ｊ Since the value of is unknown, the median value of the absolute deviation of the discrete wavelet coefficients at the minimum scale is divided by 0.6745 as an estimate of the standard deviation, and a value adjusted to the standard deviation of the Gaussian distribution is used. In addition to the above-mentioned means using a universal threshold, threshold processing may also be performed using the mini-max method, the SURE method, the HYBRID method, the cross-validation method, the Lorenz method, the Bayesian approach, or the like, and any of these may also be used.

[0130] As described above, the discrete wavelet coefficients obtained by the discrete wavelet transform are subjected to threshold processing at each level j according to the threshold, and the detection signal g is obtained by the inverse discrete wavelet transform. ｊ (t) is reconstructed. This reconstructed detection signal g ｊ 7B, the continuous wavelet transform of process 5 shown in Fig. 7B is performed on (t). Then, from the data of the continuous wavelet transform, the spectral response in wavenumber space is outputted by color contour display shown in process 6.

[0131] Additionally, in the rail inspection device 1, instead of Process 5 (continuous wavelet transform) shown in Fig. 7B, a short-time Fourier transform may be used in Process 5. This is applicable when the detection signal constantly exhibits periodic fluctuations and anomalous locations are to be detected within those fluctuations.

[0132] FIG. 7C is an explanatory diagram showing the monitoring information extraction and analysis process using short-time Fourier transform in the rail inspection device of this embodiment.

[0133] In Figure 7C, a short-time Fourier transform is performed on a predetermined section 1111, set to, for example, 1 m, in the region where the detection signal of the section under test exists, shifting the time by 1 mm. Based on the obtained results, a color contour display process of the spectral response similar to process 6 shown in Figure 7B is performed. The continuous wavelet transform is an excellent analytical method that achieves both improved resolution of the moving distance and spatial frequency compared to the short-time Fourier transform. However, for an inspection object having a surface shape made of twisted wires called strands, such as wire ropes used in elevators, periodic fluctuations reflecting the strands occur in the detection signal. For analysis involving such periodic fluctuations, the short-time Fourier transform may be more suitable than the wavelet transform.

[0134] <Circuit Configuration of First Embodiment> FIG. 11A is a block diagram showing the overall configuration of a rail inspection device 1 of this embodiment.

[0135] In this figure, the rail inspection device 1 includes a magnetic detection unit 2, a distance detection unit 3, a processing unit 4, and a power supply 88. The magnetic detection unit 2 includes magnetic sensor units 21A, 21B, and 21C and a preamplifier unit 210. The magnetic sensor units 21A, 21B, and 21C each include an oscillator coil 5A and a receiver coil 6. The preamplifier unit 210 includes amplifier filter units 22A, 22B, and 22C. The amplifier filter units 22A, 22B, and 22C are connected to the magnetic sensor units 21A, 21B, and 21C, respectively.

[0136] The processing unit 4 also includes an amplifier 31 , a digital-to-analog converter 32 , an oscillator 33 , a detector 34 , an analog-to-digital converter 35 , a memory 36 , and an evaluator 7 .

[0137] The oscillator 33 outputs a digital oscillation signal with a predetermined oscillation frequency f. This digital oscillation signal is preferably a square wave or a ramp wave. The oscillation frequency f is selected to a value that can output a sufficient excitation magnetic field and ensure sufficient sensitivity of the receiving coil 6, taking into account the impedance of the oscillator coils 5A and 5B. When selecting the oscillation frequency f, it is preferable to fully consider its influence on rail inspection. Specifically, consideration should be given to being away from 1 / f noise, being sufficiently isolated from line noise (50 Hz or 60 Hz) and its harmonics, and ensuring sufficient magnetic flux flow even with the influence of the skin effect of the rail being inspected. Therefore, it is preferable to select the oscillation frequency f from the range of 0.5 kHz to 100 kHz.

[0138] The digital-to-analog converter 32 converts the digital oscillation signal output by the oscillator 33 into an analog AC voltage. The amplifier 31 amplifies this AC voltage and applies it to the oscillator coils 5A and 5B of the magnetic sensors 21A, 21B, and 21C. As a result, an AC magnetic field with an inverted phase is generated from the oscillator coils 5A and 5B. The magnetic detection unit 2 installed on the mobile object 100 is separated from the rail tread 200a (see FIG. 1 ), which is the running surface, so as not to interfere with the movement of the mobile object 100. Therefore, an AC magnetic field of sufficient strength to sufficiently excite the rail tread 200a is output from the oscillator coils 5A and 5B.

[0139] The amplification filter units 22A, 22B, and 22C in the magnetic detection unit 2 amplify and filter the signal from the receiving coil 6 and transmit the result to the detection unit 34 of the processing unit 4. This "filtering" is a low-pass filtering process that removes frequency components higher than the upper limit frequency of the oscillation frequency f in the variable setting range, taking into account the possibility of allowing flexibility in the setting of the oscillation frequency f, or a high-pass filtering process that removes frequency components lower than the lower limit frequency of the oscillation frequency f in the variable setting range. Note that if the oscillation frequency f to be used is fixed, band-pass filtering may be used that passes only a bandwidth centered on the oscillation frequency f and that includes a frequency range that satisfies the response speed at the maximum moving speed of the mobile object 100. Furthermore, the detection unit 34 uses the reference signal SR1 supplied from the oscillator 33 to generate signals X, Y, R, and θ (details of these signals will be described later) based on the signals supplied from the amplification filter units 22A, 22B, and 22C, and supplies these signals to the analog-to-digital conversion unit 35. The analog-to-digital converter 35 converts each analog signal received from the detector 34 into a digital signal.

[0140] The distance detection unit 3 detects the rotation angle of the wheel 300 of the moving body 100 via a tachograph generator 500 connected to the axle 400 (see FIG. 1 ). The distance detection unit 3 then outputs a pulse signal SP in response to a rotation of a predetermined angle. The analog-to-digital conversion unit 35 converts the pulse signal SP into a digital signal and stores it in the memory unit 36. The digital signal output from the analog-to-digital conversion unit 35 is stored as data in the memory unit 36. The travel distance information of the moving body 100 stored in the memory unit 36 ​​is supplied to the evaluation unit 7. The evaluation unit 7 counts the pulse signals SP to calculate the travel distance of the moving body 100.

[0141] Next, the evaluation unit 7 will be described.

[0142] The evaluation unit 7 is hardware that includes a CPU (Central Processing Unit), RAM (Random Access Memory), ROM (Read Only Memory), and HDD (Hard Disk Drive), similar to a typical computer. The HDD stores an OS (Operating System), application programs, various data, and the like. The OS and application programs are loaded into the RAM and executed by the CPU.

[0143] The evaluation unit 7 includes a control unit 42, a data processing unit 43, an output processing unit 44, an operation input unit 45, a display unit 46, and a storage unit 47. The evaluation unit 7 executes an inspection processing program that identifies a peculiar location 202 (see FIG. 6 ) on the rail 200 based on the rail monitoring data received from the magnetic detection unit 2, the detection unit 34, the analog-to-digital conversion unit 35, and the memory unit 36. In this embodiment, the "rail monitoring data" refers to data at all stages from the receiving coil 6 of the magnetic detection unit 2 to the evaluation unit 7.

[0144] The control unit 42 is, for example, a CPU (an example of a processor) and controls the reading of inspection data from the memory unit 36 ​​and arithmetic processing. The data processing unit 43 performs inspection processing based on the inspection data (details will be described later). The display unit 46 is an LCD (Liquid Crystal Display) or a CRT display (CRT is an abbreviation for Cathode Ray Tube) that displays inspection results, etc. The output processing unit 44 displays rail monitoring results, etc. on the display unit 46. In this case, the output processing unit 44 performs processing to display the results in a visually easy-to-understand display format, appropriately using graphs or tables. The operation input unit 45 is an information input means such as a keyboard, mouse, or touch panel. The memory unit 47 stores data such as inspection results processed by the data processing unit 43. In addition, the data stored in the memory unit 36 ​​is also transferred to the memory unit 47. The data processing unit 43 and the output processing unit 44 are realized by loading programs and data stored in the storage unit 47 into the control unit 42 and executing arithmetic processing.

[0145] FIG. 11B is a graph showing an example of a digital oscillation signal output from the oscillator 33 of FIG. 11A.

[0146] The digital oscillation signal shown in Figure 11B is a square wave. The duty ratio is expressed as A / T, where A is the width and T is the period. In this figure, A / T = 0.5, i.e., the duty ratio is 50%. The duty ratio can be set anywhere in the range of 5 to 95%, but 20 to 80% is preferable, 30 to 70% is even more preferable, and 40 to 60% is particularly preferable.

[0147] When a square wave (e.g., 1 kHz) is used as the digital oscillation signal, not only the fundamental excitation frequency of 1 kHz but also odd-order harmonics (3 kHz, 5 kHz, 7 kHz, 9 kHz, ...) can be detected as leakage flux changes and acquired as flaw detection information, allowing the degree of damage in the depth direction of the metal object being inspected to be detected with high accuracy.

[0148] FIG. 11C is a graph showing another example of the digital oscillation signal output from the oscillator 33 of FIG. 11A.

[0149] The digital oscillation signal shown in FIG. 11C is a ramp wave with 0% symmetry.

[0150] FIG. 11D is a graph showing another example of the digital oscillation signal output from the oscillator 33 of FIG. 11A.

[0151] The digital oscillation signal shown in FIG. 11D is a ramp wave with 100% symmetry.

[0152] When a ramp wave (for example, 1 kHz) is used as the digital oscillation signal, even harmonics (2 kHz, 4 kHz, 6 kHz, 8 kHz, ...) can also be detected as leakage magnetic flux changes and acquired as flaw detection information. In other words, flaw detection information for integer harmonics can be acquired simultaneously.

[0153] In summary, when the AC magnetic field is a square wave, the controller extracts a first signal based on the fundamental frequency of the square wave and a second signal based on odd-order harmonics of the square wave from the leakage magnetic flux, and outputs data of the first signal relative to the signal of the travel distance and data of the second signal relative to the signal of the travel distance.

[0154] When the AC magnetic field is a ramp wave, the controller extracts from the leakage magnetic flux a first signal based on the fundamental frequency of the ramp wave and a second signal based on an integer-order harmonic of the ramp wave, and outputs data of the first signal corresponding to the signal of the travel distance and data of the second signal corresponding to the signal of the travel distance.

[0155] These configurations make it possible to detect with high accuracy the degree of damage in the depth direction of the metal object being inspected.

[0156] FIG. 12 is a block diagram showing the detection unit 34 of FIG. 11A.

[0157] The detection unit 34 shown in this figure includes a delay circuit 72, phase comparators 74 and 76, low-pass filters 78 and 80 (LPF), and calculators 82 and 84.

[0158] The received signal SS from the amplification filter unit group 22 is supplied to phase comparators 74 and 76. Furthermore, the reference signal SR1 supplied from the oscillator 33 (see FIG. 11A) is delayed by a delay circuit 72 by a time corresponding to a phase of 90° of the oscillation frequency f. The delayed reference signal SR1 is referred to as reference signal SR2. The reference signal SR1 is supplied to a phase comparator 76, and the reference signal SR2 is supplied to a phase comparator 74. The phase comparator 76 extracts a component of the received signal SS that is synchronized with the reference signal SR1. The extracted signal is filtered by a low-pass filter 80. The low-pass filter 80 outputs the processing result as a cosine signal X.

[0159] Furthermore, the phase comparator 74 extracts a component of the received signal SS that is synchronized with the reference signal SR2. The extracted signal is filtered by the low-pass filter 78. The low-pass filter 78 outputs the result as a sine signal Y. The calculator 84 calculates the sine signal Y by the function √(X ２ +Y ２ ) that is, (X２ +Y ２ ) ０．５ and outputs the result as an amplitude signal R. Furthermore, the calculator 82 calculates the arc tangent of (Y / X), that is, a tan(Y / X), and outputs the result as a phase difference signal θ.

[0160] The detection unit 34 then supplies the above-mentioned signals X, Y, R, and θ to the memory unit 36 ​​via the analog-to-digital conversion unit 35 (see FIG. 11A). In the example shown in the figure, the detection unit 34 outputs all of the signals X, Y, R, and θ, but the amplitude signal R and the phase difference signal θ may be calculated by the data processing unit 43 (see FIG. 11A) based on the cosine signal X and the sine signal Y, rather than by the detection unit 34.

[0161] Here, the reason why the detection unit 34 detects the sine signal Y in addition to the cosine signal X will be explained.

[0162] Information related to rail monitoring (changes in amplitude in the detection signal) can be obtained using either the cosine signal X or the sine signal Y. On the other hand, if only the cosine signal X is used, adjustment is required to set the phase of the reference signal so that the amplitude of the cosine signal X is maximized during inspection. This phase setting must be adjusted each time an inspection condition changes, for example, when the target rail changes, when the distance between the rail and the magnetic detection unit changes, or when the excitation frequency of the oscillation coil in the magnetic detection unit changes.

[0163] The detector 34 shown in this figure can calculate the phase difference by simultaneously detecting the cosine signal X and the sine signal Y, enabling inspections to be performed without phase adjustment. Furthermore, not only changes in the amplitude of the detection signal but also changes in the phase difference can be obtained simultaneously, allowing for more detailed rail monitoring information to be obtained. Furthermore, the value of the amplitude signal R calculated by the calculator 84 (or the evaluation unit 7 (see FIG. 11A)) is, in principle, constant even when the phase difference signal θ fluctuates, which has the advantage of eliminating the need to optimize the phase of the reference signal.

[0164] <Operation of the Rail Inspection Device of the First Embodiment> FIG. 13 is a flowchart showing rail monitoring data processing executed by the data processing unit 43 of the evaluation unit 7 in FIG. 11A.

[0165] This process is executed at each predetermined control cycle.

[0166] In FIG. 13, the processes of steps S2 and S14 are executed in parallel.

[0167] First, the evaluation unit 7 acquires rail monitoring data from the storage unit 47 in step S2, and then acquires a distance pulse signal SP from the storage unit 47 in step S14. After completing step S14, the evaluation unit 7 converts the distance pulse signal SP into distance data SK in step S16. That is, the evaluation unit 7 calculates the distance data SK based on the number of times the distance pulse signal SP is detected by the distance detection unit 3 until the moving object 100 stops. Here, the distance data SK is the distance from a predetermined reference position (the movement start position of the moving object) and may be an example of position data.

[0168] After completing step S2, the evaluation unit 7 acquires spectral response data in wave number space using the distance data SK in step S4.

[0169] After step S4 is completed, it is determined whether the spectral response data falls outside a predetermined reference range, i.e., a range of signal strength for which a lower limit value is set (step S6). Here, the "spectral response data" used for the determination is, for example, the amplitude signal R shown in FIG. 12. If there is a singular point 202 (see FIG. 6), the amplitude signal R will clearly increase. Therefore, in step S6, the evaluation unit 7 compares the amplitude signal R with a predetermined threshold value R ｔｈ１ Compare "R>R ｔｈ１ If "R≦R", it is determined as "Yes", that is, it is determined that there is a possibility that a singular point exists. ｔｈ１ If "YES", it is determined that there is no singular point.

[0170] If the determination in step S6 is "No", the process proceeds to step S18, the details of which will be described later.

[0171] On the other hand, if the determination in step S6 is "Yes," the process proceeds to step S8, where the structure and damage are discriminated based on the spatial frequency region where an increase in the signal intensity of the signals X, Y, R, θ, etc. shown in Fig. 12 occurs.

[0172] If the determination in step S6 is "No" or if a determination is made in step S8, a rail monitoring result is generated (step S18). Here, if step S8 is performed, the "rail monitoring result" includes a correlation between anomalous locations (structures and damage) and distance data SK. Also, if the determination in step S6 is "No," the "rail monitoring result" may include a result indicating that no anomalous locations are found.

[0173] Next, the data processing unit 43 outputs the rail monitoring results to the display unit 46 or the like (step S20), and ends the processing in this figure.

[0174] <Effects of the Rail Inspection Device of the First Embodiment> As described above, in the rail inspection device 1 of this embodiment, the lower housing 20 is made of a non-magnetic material and houses the magnetic sensor group 21, and the upper housing 26 is made of metal and houses the preamplifier unit 210. This makes it possible to stably detect the condition of the rail 200.

[0175] The magnetic detection unit 2 excites the rail tread 200a of the rail 200 with an AC magnetic field and detects the magnetic flux generated on the rail tread 200a. The processing unit 4 detects the singular point 202 on the rail tread 200a based on the disturbance of the flow of magnetic flux. This makes it possible to accurately detect the state of the rail 200.

[0176] The output processing unit 44 of the rail inspection device 1 displays the type and location of the detected peculiar point 202 as an image including a graph or the like on the display unit 46. This allows the user to visually grasp the condition of the rail 200.

[0177] Furthermore, the discrimination result for the peculiar point 202 can be used to manage the structures and damage present on the rail 200. Furthermore, the processing unit 4 detects the position of the peculiar point 202 based on the distance pulse signal SP from the distance detection unit 3, and displays the result on the display unit 46. This allows the user to visually grasp the position of the peculiar point 202.

[0178] Second Embodiment Next, a second embodiment of the present disclosure will be described. In the following description, parts corresponding to those in Figures 1 to 12 are designated by the same reference numerals, and their description may be omitted. In addition, the following description will mainly focus on differences from the first embodiment, and descriptions of commonalities with the first embodiment will be omitted or simplified.

[0179] <Configuration of Second Embodiment> FIG. 14 is a bottom view showing the magnetic detection unit 2 of this embodiment.

[0180] In this figure, the magnetic detection unit 2 has a magnetic sensor group 21 made up of magnetic sensor units 21-1 to 21-N (N is an integer equal to or greater than 2). The magnetic sensor units 21-1 to 21-N are arranged in a row along the width direction of the lower housing 20, i.e., along the width direction of the rail 200 (see FIG. 1).

[0181] The magnetic sensor unit 21-k (k is any integer between 1 and N) includes an oscillator coil 5A-k, an oscillator coil 5B-k, and a receiver coil 6-k. Each of these coils is made by winding a coated copper wire.

[0182] The oscillator coil 5A-k, the receiver coil 6-k, and the oscillator coil 5B-k are arranged along the direction of movement of the rail 200 (see FIG. 1). The distance between the oscillator coil 5A-k and the receiver coil 6-k is equal to the distance between the receiver coil 6-k and the oscillator coil 5B-k. The oscillator coils 5A-k and 5B-k are supplied with AC currents of different frequencies from the processing unit 4 (see FIG. 1) via the connection cable 60. As a result, AC magnetic fields of different frequencies are generated from the oscillator coils 5A-k and 5B-k, which excite the rail 200, and the magnetic flux generated from the excited rail 200 generates an induced voltage in each receiver coil 6-k.

[0183] FIG. 15 is a partially cutaway plan view showing the magnetic detection unit 2 of this embodiment.

[0184] 14. In this figure, similar to FIG. 4, the central portion of the flange 25 in FIG. 14 is cut away to expose the inside of the upper housing 26. In FIG.

[0185] 15, the preamplifier unit 210 has a printed circuit board on which the amplification filter unit group 22 is mounted. The amplification filter unit group 22 has the same number (N) of amplification filter units 22-1 to 22-N as the magnetic sensor unit group 21 in FIG.

[0186] The amplification filter unit 22-k (k is any integer between 1 and N) amplifies and filters the induced voltage generated in the receiving coil 6-k (FIG. 14), and transmits the results to the processing unit 4 via the connection cable 60 (see FIG. 1). The processing unit 4 analyzes the received signal and detects the magnetic signal generated from the rail 200. There is a 1:1 correspondence between the magnetic sensor unit 21-k (FIG. 14) and the amplification filter unit 22-k.

[0187] <Circuit Configuration of Second Embodiment> FIG. 16 is a block diagram showing the overall configuration of a rail inspection device 1 of this embodiment.

[0188] In the rail inspection device 1 shown in the figure, the magnetic detection unit 2 has magnetic sensor units 21-1 to 21-N and a preamplifier unit 210. In the following description, k is assumed to be any integer between 1 and N.

[0189] The magnetic sensor unit 21-k includes oscillation coils 5A-k and 5B-k and a receiving coil 6-k. The preamplifier unit 210 includes an amplification filter unit 22-k. The amplification filter unit 22-k is connected to the magnetic sensor unit 21-k.

[0190] The processing unit 4 also includes amplifiers 31-1 to 31-N, N digital-to-analog converters 32, an oscillator 33, detectors 34-1 to 34-N, an analog-to-digital converter 35, a memory unit 36, and an evaluation unit 7. In this embodiment, N is an integer of 2 or more, but may also be 1.

[0191] Digital-to-analog converter 32 converts the digital oscillation signal output by oscillator 33 into an analog AC voltage. Amplifier 31-k amplifies this AC voltage and applies it to oscillator coils 5A-k and 5B-k of magnetic sensor unit 21-k. As a result, oscillator coils 5A-k and 5B-k generate AC magnetic fields with inverted phases.

[0192] In this embodiment, the spectral response in wave number space derived by the processing described above using the signals from the magnetic detection unit 2 and the distance detection unit 3 with reference to Figures 7A, 7B, 8A, 8B, 9, and 10 is obtained for the entire area of ​​the rail 200 occupied by the multiple magnetic sensor units 21-k. Therefore, when extracting anomalous locations, it is possible to know the response change of the signal strength at a specific spatial frequency in the measurement area of ​​the rail 200, which makes it possible to grasp, for example, the distribution of damage occurrence, thereby enabling detailed rail monitoring.

[0193] (Modifications) The present disclosure is not limited to the above-described embodiments and various modifications are possible. The above-described embodiments are provided as examples to facilitate understanding of the present disclosure, and the present disclosure is not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace a portion of the configuration of one embodiment with a configuration of another embodiment, or to add a configuration of another embodiment to a configuration of one embodiment. Furthermore, it is possible to delete a portion of the configuration of each embodiment, or to add and / or replace other configurations. Furthermore, the control lines and information lines shown in the figures are those considered necessary for explanation, and do not necessarily represent all control lines and information lines necessary in the product. In reality, it is possible to consider that almost all components are interconnected. Possible modifications of the above-described embodiments include, for example, the following:

[0194] (1) The hardware of the evaluation unit 7 in each of the above embodiments can be realized by a general-purpose computer. Therefore, a program for executing the signal processing shown in FIGS. 7A, 7B, 8A, 8B, 9, and 10 and the processing according to the flowchart shown in FIG. 13 may be installed in a computer from a program source such as a portable storage medium or a server.

[0195] (2) In the above embodiments, the functions shown in FIGS. 11A, 12, and 16 and the processes described based on FIGS. 7A, 7B, 8A, 8B, 9, 10, and 13 have been described as functions and processes realized by a processor executing a program. However, some or all of these functions and processes may be replaced with hardware functions and processes using an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or the like.

[0196] (3) In the above embodiment, the rail inspection device 1 may be installed on any type of mobile object, including a manually operated object such as a handcart, or a self-propelled object such as a motor or diesel engine.

[0197] (4) The present invention can be applied to a self-propelled mobile body having a tachograph that can output distance information other than the distance pulse signal SP using the encoder described in the embodiment. Alternative methods such as estimating the traveled distance based on acceleration information obtained from an acceleration sensor or measuring the traveled distance using a laser Doppler may also be used.

[0198] The above description can be summarized as follows: The following summary may include supplementary and modified versions of the above description.

[0199] The rail inspection device 1 has a magnetic detection unit 2, a distance detection unit 3, and a processing unit 4 connected to the magnetic detection unit 2 and the distance detection unit 3. The magnetic detection unit 2 has a magnetic sensor unit group 21, which is provided on a mobile object 100 that moves relatively on a rail 200 and detects, in a non-contact manner, a leakage magnetic field generated by AC excitation of the rail 200. The distance detection unit 3 is provided on the mobile object 100 and detects the relative movement distance of the mobile object 100 with respect to the rail 200. The processing unit 4 converts time-series data 111 representing a time series of detection signal strength obtained from the magnetic detection unit 2 into data 222 representing the relationship between movement distance and detection signal strength, based on the movement distance at each point in time detected by the distance detection unit 3.

[0200] The processing unit 4 extracts a spectral response in wavenumber space identified from the data 222 and outputs monitoring information based on the extracted spectral response. A peculiar location on the rail 200 can be identified from the monitoring information. Furthermore, since the offset drift caused by the oscillation of the moving body 100, which is a noise factor, varies in distance range, it is easy to remove the offset drift regardless of the speed. The monitoring information may include information representing the results of identifying a peculiar location on the rail 200 (e.g., information representing the travel distance and the nature of the peculiar location). Furthermore, the peculiar locations may be broadly classified into damage locations and structural locations, and it may be possible to identify (estimate) which type of location the structural location is from among multiple types of locations classified as structural locations.

[0201] The monitoring information may be output by displaying the monitoring information (for example, on a display unit). The monitoring information may be information representing the correspondence between spatial frequency and the distance traveled by the mobile object. Since the spatial frequency components of the detection signal differ depending on the type of anomalous location, it is possible to identify the anomalous location from such monitoring information. The correspondence between spatial frequency and the distance traveled by the mobile object may be a color contour in an orthogonal coordinate system with a first axis corresponding to the spatial frequency and a second axis corresponding to the distance traveled by the mobile object, and a color indicator according to the intensity of the detection signal may be arranged in the orthogonal coordinate system. This is expected to provide high visibility.

[0202] The processing unit 4 may perform spatial filtering on the data 222 according to the size required for monitoring the rail 200, and extract a spectral response in wavenumber space from the detection signal data 555 that has been subjected to the spatial filtering for each inspection section 666 that corresponds to monitoring the rail 200. The spectral response in wavenumber space obtained after data cleansing using the spatial filter can provide visualization of the extraction and separation of anomalous locations (distinguishing between structural locations and damaged locations).

[0203] The processing unit 4 may perform interpolation processing on the movement distance samples so that the movement distance samples in the data 222 are spaced at equal intervals. Spatial filtering processing may be performed on the interpolated data 222. Since the movement distance samples are interpolated so that they are spaced at equal intervals in this manner, spatial filtering processing becomes possible.

[0204] The processing unit 4 may perform a continuous wavelet transform to extract a spectral response in wavenumber space for each inspection section 666 of the data 222. This makes it possible to identify which spatial frequency component has changed at which moving distance for each inspection section 666. Note that the processing unit 4 may perform signal separation using a discrete wavelet transform to extract a spectral response in wavenumber space for each inspection section 666 of the data 222, and then use a continuous wavelet transform on the separated signals.

[0205] The processing unit 4 may perform a short-time Fourier transform to extract a spectral response in wavenumber space for each inspection section 666 of the data 222. In this case, the rail 200 may have shapes that are periodically arranged along the relative movement direction of the mobile body 100. This makes it possible to detect periodic structural changes in the rail 200.

[0206] The magnetic detection unit 2 may have a plurality of magnetic sensor units arranged in a direction perpendicular to the direction of relative movement of the moving body 100 .

[0207] The controller converts time series data representing the time series of the detected signal strength obtained from the magnetic detection unit into correlation data representing the relationship between the movement distance and the detected signal strength based on the movement distance at each point in time detected by the distance detection unit, extracts a spectral response in wave number space identified from the correlation data, and outputs monitoring information based on the spectral response.

[0208] The spectral response is extracted using a continuous wavelet transform.

[0209] The controller performs signal separation on the correlation data with a discrete wavelet transform and extracts the spectral response using a continuous wavelet transform on the separated signals.

[0210] The controller outputs image data that displays an indicator and color contours according to the detected signal strength as a graph showing the relationship between spatial frequency and distance traveled.

[0211] 1: rail inspection device, 2: magnetic detection unit, 3: distance detection unit, 4: processing unit, 5A, 5B: oscillation coil, 6: receiving coil, 7: evaluation unit, 20: lower housing, 21: magnetic sensor unit group, 21A, 21B, 21C: magnetic sensor unit, 22: amplification filter unit group, 22A, 22B, 22C: amplification filter unit, 25: flange, 26: upper housing, 31: amplification unit, 32: digital-to-analog conversion unit, 33: oscillation unit, 34: detection unit, 35: analog-to-digital conversion unit, 36: memory unit, 60, 61: connection cable, 88: power supply, 100: moving body, 200: rail, 200a: rail tread, 202: peculiar location, 210: preamplifier unit, 222, 333: data, 300: wheel, 400: axle, 500: tachometer generator

Claims

1. A metal body inspection device comprising: a magnetic detection unit having a magnetic sensor unit; a distance detection unit; and a controller connected to the magnetic detection unit and the distance detection unit, the metal body inspection device being installed on a mobile body that moves relative to a metal body being inspected, wherein the distance detection unit detects the distance that the mobile body has moved relative to the metal body, the magnetic sensor unit has a first oscillator coil, a second oscillator coil, and a receiving coil, and is arranged to face the metal body, the receiving coil is arranged between the first oscillator coil and the second oscillator coil, the first oscillator coil and the second oscillator coil are configured to generate alternating current magnetic fields in mutually opposite directions, the alternating current magnetic field is a square wave or a ramp wave, and the receiving coil detects leakage magnetic flux generated from the metal body excited by the alternating current magnetic field.

2. The metal object inspection device according to claim 1, wherein a plurality of said magnetic sensor units are installed.

3. A metal object inspection device as described in claim 2, wherein at least two of the plurality of magnetic sensor units are arranged so that the directions of the straight lines connecting the centers of gravity of the first oscillator coil and the second oscillator coil for the two magnetic sensor units are different from each other.

4. A metal body inspection device as set forth in claim 2, wherein a first line connecting the center of gravity of the first oscillator coil and the center of gravity of the second oscillator coil for a first magnetic sensor unit that is one of a predetermined three of the plurality of magnetic sensor units is arranged parallel to the direction of movement of the mobile body, a second line connecting the center of gravity of the first oscillator coil and the center of gravity of the second oscillator coil for a second magnetic sensor unit that is one of the three magnetic sensor units is arranged so as to be perpendicular to the first line of the first magnetic sensor unit, and a third line connecting the center of gravity of the first oscillator coil and the center of gravity of the second oscillator coil for a third magnetic sensor unit that is one of the three magnetic sensor units is arranged so as to diagonally intersect the first line of the first magnetic sensor unit.

5. A metal object inspection device as described in claim 1, wherein the AC magnetic field is a rectangular wave, and the controller extracts from the leakage magnetic flux a first signal due to the fundamental frequency of the rectangular wave and a second signal due to odd-order harmonics of the rectangular wave, and outputs data of the first signal for the signal of the travel distance and data of the second signal for the signal of the travel distance.

6. A metal object inspection device as described in claim 1, wherein the AC magnetic field is the ramp wave, and the controller extracts from the leakage magnetic flux a first signal based on the fundamental frequency of the ramp wave and a second signal based on an integer-order harmonic of the ramp wave, and outputs data of the first signal for the signal of the travel distance and data of the second signal for the signal of the travel distance.

7. A metal object inspection device as described in claim 1, wherein the controller converts time series data representing the time series of the detection signal strength obtained from the magnetic detection unit into correlation data representing the relationship between the movement distance and the detection signal strength based on the movement distance at each point in time detected by the distance detection unit, extracts a spectral response in wave number space specified from the correlation data, and outputs monitoring information based on the spectral response.

8. The metallic object inspection device according to claim 7, wherein the spectral response is extracted using a continuous wavelet transform.

9. The metal object inspection device according to claim 8, wherein the controller performs signal separation on the correlation data using a discrete wavelet transform, and extracts the spectral response using the continuous wavelet transform on the separated signals.

10. A metallic object inspection device according to claim 7, wherein said controller outputs image data that displays an indicator and color contour according to the detected signal strength as a graph showing the relationship between spatial frequency and said moving distance.

Images