Inspection device, inspection method and inspection program
The inspection device and method address noise and cost issues in magnetic material testing by analyzing magnetic field changes to calculate natural frequency, offering accurate and cost-effective assessments of magnetic materials.
Patent Information
- Application Number
- JP2023075174
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-04-28
- Publication Date
- 2024-11-08
AI Technical Summary
Existing non-destructive testing methods for magnetic materials, such as magnetic flux leakage testing and hammering sound inspection, face challenges with noise interference, limited applicability, and high costs, making it difficult to accurately assess the condition of magnetic materials.
An inspection device and method utilizing a magnetic sensor to detect changes in the magnetic field caused by vibrations of magnetic materials, analyzing these changes to calculate natural frequency, and evaluating the material's state based on this frequency, potentially using machine learning models.
The method provides accurate, cost-effective, and noise-resistant assessment of magnetic materials, enabling early detection of deterioration and allowing repeated inspections without magnetizing the object, thus improving safety and reliability.
Smart Images

Figure 2024159289000001_ABST
Abstract
Description
[Technical field]
[0001] The present invention relates to an inspection device, a method, and a program for inspecting an inspection object made of a magnetic material. [Background technology]
[0002] Hammering testing is known as a method for inspecting objects such as structures (for example, a method for detecting defects). Hammering testing is a method for grasping the internal state of a structure by the inherent vibration (sound) induced when the surface of the structure is struck. Hammering testing is a simple non-destructive testing method, but for example, when the object to be inspected is a reinforced concrete structure, the acoustic signal detected after striking includes not only the sound of the concrete, but also the sound of the rebar inside and the reflected sound, as well as environmental sounds far from the object to be inspected. As such, the hammering testing method is vulnerable to noise and has low spatial resolution, making it difficult to obtain the information necessary to inspect the object to be inspected.
[0003] Another non-destructive testing method is magnetic leakage flux testing, which is specialized for testing objects made of magnetic materials. Magnetic leakage flux testing applies a magnetic field to the ferromagnetic material being tested, and detects magnetic flux leaking from defects on the surface or surface layer of the ferromagnetic material to perform the test (for example, to detect the presence of defects). Magnetic flux only leaks from magnetic materials, and magnetic fields decay in inverse proportion to distance. As a result, magnetic leakage flux testing is more resistant to noise and has a higher spatial resolution than hammering testing.
[0004] However, in the leakage flux inspection method, if the defect is small, the magnetic flux leaking from the defect is also small, so the magnetic flux signal of the inspection target may be buried in other magnetic flux signals. Therefore, for example, Patent Document 1 discloses a method of calculating a fitting curve that approximates the output signal waveform of a magnetic sensor by fitting processing, and taking the difference between the output signal waveform and the fitting curve to remove noise caused by stray magnetic fields and extract small leakage magnetic flux signals with high accuracy. [Prior art documents] [Patent documents]
[0005] [Patent Document 1] JP 2011-196863 A Summary of the Invention [Problem to be solved by the invention]
[0006] Although the magnetic leakage flux inspection method is an effective inspection method, it takes time and costs to conduct the inspection because it is necessary to magnetize the demagnetized inspection object with a permanent magnet. Furthermore, there is also the issue that the type of inspection object made of a magnetic material and the inspection location are limited because the size of the magnet unit containing the permanent magnet is large.
[0007] By establishing a non-destructive testing technology that is resistant to noise, has high spatial resolution, is easy and inexpensive to use, and can be applied in a wide range of fields, it will be possible to detect the condition of the object being tested (for example, the presence or absence of deterioration and its degree) early and with high accuracy, and it is expected that the safety and reliability of the object being tested will be dramatically improved. For this reason, there is a demand for a method that can easily and accurately test the condition of an object being tested that is made of a magnetic material.
[0008] The present invention has been devised in view of such problems, and one of its objectives is to provide an inspection device, an inspection method, and an inspection program for inspecting the state of an inspection object made of a magnetic material easily and with high accuracy. However, the present invention is not limited to this objective. Another objective of the present invention is to achieve an effect that cannot be obtained by the conventional technology, which is derived from each configuration shown in the below-mentioned embodiment of the invention. [Means for solving the problem]
[0009] (1) The inspection device disclosed herein is an inspection device for inspecting an inspection object made of a magnetic material, and includes a magnetic sensor that detects magnetic field changes caused by vibration of the inspection object made of the magnetic material, a calculation unit that analyzes the detected magnetic field changes and calculates the natural frequency of the inspection object, and an evaluation unit that estimates and evaluates the state of the inspection object based on the calculated natural frequency.
[0010] (2) The inspection device disclosed herein is an inspection device for inspecting an inspection object made of a magnetic material, and includes a calculation unit that acquires and analyzes magnetic field changes caused by vibration of the inspection object made of the magnetic material and calculates a natural frequency of the inspection object, and an evaluation unit that estimates and evaluates a state of the inspection object based on the calculated natural frequency.
[0011] (3) In the above (1) or (2), it is preferable that the apparatus further comprises an output unit that outputs the estimated evaluation result of the state of the inspection object to an output device. (4) In any one of the above (1) to (3), the change in the magnetic field is preferably obtained by utilizing a tunnel magnetoresistance effect.
[0012] (5) In any of (1) to (4) above, it is preferable that the inspection device repeatedly inspects the object to be inspected, and the evaluation unit compares the currently calculated natural frequency with the natural frequency calculated previously or previously to estimate and evaluate a state of the object to be inspected. (6) In the above (5), it is preferable that the evaluation unit estimates and evaluates a state of the object to be inspected using a model constructed by machine learning the calculated natural frequency.
[0013] (7) The inspection method disclosed herein is an inspection method for inspecting an object to be inspected made of a magnetic material, and includes an acquisition step of acquiring magnetic field changes caused by vibration of the object to be inspected made of the magnetic material, a calculation step of analyzing the acquired magnetic field changes and calculating a natural frequency of the object to be inspected, and an evaluation step of estimating and evaluating a state of the object to be inspected based on the calculated natural frequency.
[0014] (8) In the above (7), it is preferable to further include an output step of outputting the evaluation result of the estimated state of the inspection object to an output device. (9) In the above (7) or (8), it is preferable that the acquiring step detects the change in the magnetic field by utilizing a tunnel magnetoresistance effect.
[0015] (10) In any of (7) to (9) above, it is preferable that the inspection method involves repeatedly inspecting the object to be inspected, and the evaluation step involves comparing the currently calculated natural frequency with the natural frequency calculated previously to estimate and evaluate a state of the object to be inspected. (11) In the above (10), it is preferable that the evaluation step estimates and evaluates a state of the object to be inspected using a model constructed by machine learning the calculated natural frequency.
[0016] (12) The inspection program disclosed herein is an inspection program for inspecting an object to be inspected made of a magnetic material, and causes a computer to execute a process of acquiring and analyzing magnetic field changes caused by vibration of the object to be inspected made of the magnetic material, calculating a natural frequency of the object to be inspected, and estimating and evaluating a state of the object to be inspected based on the calculated natural frequency. Effect of the Invention
[0017] According to the disclosed inspection device, method, and program, the state of the object to be inspected is estimated and evaluated based on the natural frequency of the object to be inspected calculated from the magnetic field change due to the vibration of the object to be inspected, so that the state of the object to be inspected made of a magnetic material can be inspected simply and with high accuracy. [Brief description of the drawings]
[0018] [Figure 1] 1 is a block diagram showing a configuration of an inspection device according to an embodiment; [Diagram 2] 1 is a block diagram showing a hardware configuration of an inspection apparatus according to an embodiment; [Diagram 3] 1 is a flowchart illustrating a procedure of an inspection method according to an embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0019] An inspection device, an inspection method, and an inspection program as embodiments will be described with reference to the drawings. The embodiments described below are merely examples, and are not intended to exclude the application of various modifications and techniques not explicitly described in the following embodiments. Each configuration of the embodiments can be modified in various ways without departing from the spirit of the embodiments. In addition, they can be selected as necessary, or can be combined appropriately.
[0020] [1. Overview] The inspection device, method, and program for an object to be inspected according to the present embodiment analyze the magnetic field change caused by the vibration of the object to be inspected made of a magnetic material, and estimate and evaluate the state of the object to be inspected based on the natural frequency of the object to be inspected. Specifically, when the object to be inspected made of a magnetic material vibrates slightly due to the application of an external force, the static magnetic field around the object to be inspected fluctuates slightly, and this slight magnetic field change corresponding to the natural vibration of the object to be inspected is used to estimate and evaluate the state of the object to be inspected.
[0021] This evaluation may include qualitative evaluation and quantitative evaluation. Examples of the state evaluated here include the presence or absence of deterioration of the inspection object and the degree of deterioration, the estimated dimensions of the inspection object (structure) in a predetermined direction and their changes. The predetermined direction here refers to the direction of displacement when the inspection object vibrates. For example, if the inspection object is a plate material, the predetermined direction is the plate thickness direction, and the estimated dimensions of the thickness and the changes in the thickness can be evaluated. In this way, a method of grasping the state of the inspection object from only information caused by the magnetic body is called a "percussion test method." The "percussion test method" is a method that solves the problems of the conventional percussion test method and the leakage magnetic flux inspection method. The inspection device 1, the inspection method, and the inspection program for realizing the "percussion test method" will be described below.
[0022] [2. Inspection equipment] FIG. 1 is a block diagram showing the configuration of an inspection device 1 according to an embodiment. The inspection device 1 according to this embodiment is composed of a magnetic sensor 20 that detects changes in a magnetic field, and a control device 10 that estimates and evaluates the state of an object to be inspected from the detected (acquired) magnetic field changes. The magnetic sensor 20 is connected to the control device 10 by wire or wirelessly. The control device 10 may be built-in (mounted) or integrated with the magnetic sensor 20, or may be provided separately from the magnetic sensor 20. When the control device 10 is provided separately from the magnetic sensor 20, the magnetic sensor 20 and the control device 10 may be connected wirelessly or by wire, and the magnetic sensor 20 may be remotely operated.
[0023] The inspection object 30 that can be inspected by the inspection device 1 is an object made of a magnetic material, such as a structure made only of a magnetic material (iron pipes and bodies, railroad car rails, etc.) and a structure including a magnetic material (reinforced concrete buildings and bridges, etc.). In the inspection device 1 of this embodiment, the inspection of the inspection object 30 is performed at least once, and preferably repeatedly. When the detection by the magnetic sensor 20 and the processing by the control device 10 are repeatedly performed, the frequency of the inspection may be regular (for example, every month, every few months, or every year) or irregular.
[0024] [2-1. Magnetic sensor] The magnetic sensor 20 detects magnetic field changes caused by vibration of the inspection object 30 made of a magnetic material. The magnetic field changes are caused by vibration of the inspection object 30 when an external force is applied. The external force includes, for example, impact using a piezoelectric element, water pressure, wind pressure, and vibration of surrounding objects. When an external force is applied to the inspection object 30, the static magnetic field around the inspection object 30 fluctuates due to the natural vibration of the inspection object 30, and the distance from the magnetic sensor 20 changes slightly from moment to moment. The magnetic sensor 20 arranged near the inspection object 30 detects the magnetic field changes between the inspection object 30 and the magnetic sensor 20.
[0025] The vibration (natural vibration) of the inspection object 30 occurs in the low frequency band of tens of thousands of Hz or less, more specifically, several tens of Hz to tens of thousands of Hz. Furthermore, the static magnetic field around the inspection object 30, for example, the environmental magnetic field such as the geomagnetism, is very small at 10 nT or less. Therefore, the magnetic sensor 20 that detects the change in the static magnetic field around the inspection object 30 caused by the vibration (natural vibration) of the inspection object 30 is preferably a sensor that can detect low frequency bands and low magnetic fields. Specific conditions for the magnetic sensor 20 are that it can detect frequencies of 20 kHz or less, preferably frequencies in the range of 0.2 Hz to 20 kHz, and 10 -9 T / Hz 0.5 A detection magnetic field resolution of about 10 or less, preferably about -13 T / Hz 0.5 Greater than or equal to 10 -9 T / Hz 0.5 It is desirable to have the following detection magnetic field resolution:
[0026] The magnetic sensor 20 may be, for example, a TMR (Tunnel Magneto Resistance) sensor that detects magnetic field changes using the tunnel magnetoresistance effect. The TMR sensor has a structure in which a thin insulating layer is sandwiched between ferromagnetic layers, and when a voltage is applied perpendicular to the film surface, a current flows through the insulating layer due to the tunnel effect, and the electrical resistance at that time changes significantly depending on the external magnetic field. The TMR sensor is characterized by being less susceptible to temperature changes, having very high magnetic sensitivity, and having a high resistance value. The TMR sensor is a magnetic sensor that satisfies the above-mentioned numerical range conditions and is capable of detecting the above-mentioned low frequency band and low magnetic field.
[0027] The TMR sensor includes a composite circuit including a tunnel magnetoresistance element that detects magnetic field changes and a memory that stores information on the detected magnetic field changes (not shown). Many applications have been filed regarding the TMR sensor, and detailed descriptions are found in, for example, WO2017 / 115839 and JP2022-94518A.
[0028] A tunnel magnetoresistance element is manufactured by depositing each layer on a Si, SiO2 substrate using magnetron sputtering. The high sensitivity of a TMR sensor depends greatly on the combination of materials for the ferromagnetic layer and insulating layer of the tunnel magnetoresistance element. A tunnel magnetoresistance element capable of detecting minute magnetic field changes is, for example, a Si, SiO2 substrate / lower electrode layer / Co 70.5 Fe 4.5 S 15 B 10 (layer thickness 140nm) / Ru (layer thickness 0.4nm) / Co 40 Fe 40 B 20 (layer thickness 3nm) / MgO / Co 40 Fe 40 B 20 (layer thickness 3nm) / Ru (layer thickness 0.9nm) / Co 75 Fe 25 (layer thickness 2nm) / Ir 22 Mn 78(layer thickness 10 nm) / Ta (layer thickness 5 nm) / Pt (layer thickness 5 nm) / Ru (layer thickness 5 nm). In the above film structure, the lower ferromagnetic layer is a first lower magnetic layer Co 70.5 Fe 4.5 S 15 B 10 (layer thickness 140 nm), an insertion layer Ru (layer thickness 0.4 nm) that enables magnetostatic coupling between the upper and lower ferromagnetic layers, and a first upper magnetic layer Co 40 Fe 40 B 20 On the upper magnetic layer, an insertion coupling layer Ru (thickness 0.4 nm) is sandwiched, and a second lower magnetic layer Co 40 Fe 40 B 20 (layer thickness 3 nm), insulating layer MgO, second upper magnetic layer Co 40 Fe 40 B 20 (layer thickness 3 nm) is formed. A more detailed description is given in the paper posted on the following website (https: / / www.nature.com / articles / s41598-022-10155-6). It has been confirmed through experiments that a TMR sensor equipped with this film configuration is capable of measuring minute brain magnetic waves at room temperature, and is also capable of detecting changes in the thickness of the test object 30 with a resolution of approximately 0.3 mm.
[0029] The magnetic sensor 20 detects the change in the external magnetic field as the electrical resistance of the insulating layer due to the tunnel magnetoresistance effect. In other words, the magnetic sensor 20 detects the change in the magnetic field corresponding to the natural frequency of the inspection object 30 as a voltage-time signal (magnetic signal). The distance between the magnetic sensor 20 and the inspection object 30 is not particularly limited, but since the magnetic field attenuates inversely proportional to the distance, a close distance is preferable, and a distance of 2 cm±1 cm is more preferable. In addition, when the magnetic sensor 20 is moved along the inspection object 30, it is preferable that the distance between the magnetic sensor 20 and the inspection object 30 is constant.
[0030] The magnetic sensor 20 stores the detected magnetic signal in an internal memory, or may transmit the detected magnetic signal to the control device 10 wirelessly or via a wire during or after the inspection.
[0031] [2-2.Control device] <Functional configuration> The following describes the functions of the control device 10 according to this embodiment shown in Fig. 1. The control device 10 calculates the natural frequency of the inspection object 30 by acquiring and analyzing the magnetic field change detected by the magnetic sensor 20, and estimates and evaluates the state of the inspection object 30 based on the calculated natural frequency.
[0032] The control device 10 is provided with at least a calculation unit 11 that calculates the natural frequency of the inspection object 30 and an evaluation unit 12 that estimates and evaluates the state of the inspection object 30 as functional elements for carrying out the process. The control device 10 of this embodiment is further provided with a generation unit 13 that generates a model by machine learning, an output unit 14 that outputs the evaluation result to the output device 40, and a storage unit 15 that stores various data. Note that each of these elements may be realized by an electronic circuit (hardware), or some of these functions may be provided as hardware and the other parts may be software.
[0033] The calculation unit 11 acquires and analyzes the magnetic field change detected by the magnetic sensor 20, and calculates the natural vibration frequency of the inspection object 30. As described above, the magnetic field change is acquired as a magnetic signal (voltage-time signal) by the magnetic sensor 20, and therefore the calculation unit 11 performs Fourier analysis on the acquired magnetic signal. In other words, by performing Fourier analysis on the magnetic signal (voltage-frequency signal) originating from the fluctuation of the static magnetic field around the inspection object 30, the natural vibration frequency (natural frequency) of the inspection object 30 can be calculated.
[0034] The evaluation unit 12 estimates and evaluates the state of the inspection object 30 based on the natural frequency calculated by the calculation unit 11. The evaluation unit 12 compares the natural frequency calculated this time with the natural frequency calculated previously or previously, and estimates and evaluates the state of the inspection object 30. That is, if this is the nth evaluation process (n is an integer equal to or greater than 1), the natural frequency calculated this time is compared with the natural frequency calculated previously (n-1). For example, if this is the first evaluation process, the natural frequency calculated this time is compared with the natural frequency calculated previously (0th) (i.e., the initial value). Note that the comparison target (initial value) in this case is the natural frequency of the inspection object 30 when it is new. The natural frequency when it is new can be calculated in advance and may be stored in the storage unit 15.
[0035] Also, for example, if this is the 10th evaluation process, the natural frequency calculated this time is compared with the natural frequency calculated before (0th to 9th). The comparison target in this case may be the previous (9th) natural frequency, an initial value, or any of the natural frequencies before. Alternatively, it may be the average value or median value of the natural frequencies up to now (0th to 9th), or the average value or median value of the natural frequencies from the previous (9th) to a predetermined number of times (for example, 2 times) before (i.e., the natural frequencies of the 7th, 8th, and 9th). All of the natural frequencies calculated so far may be stored in the storage unit 15, or some of them may be stored (updated) in the storage unit 15 depending on the comparison method in the evaluation unit 12.
[0036] If the comparison shows that the natural frequency calculated this time is different from the natural frequency calculated previously or before, the evaluation unit 12 estimates that there is a change in the state of the inspection object 30. As described above, the evaluation standard (comparison target) may be the natural frequency when the object is new, or any of the natural frequencies before or before, or a natural frequency within a certain range including the initial value (the natural frequency when the object is new) (the natural frequency of a non-defective product) may be set as the threshold value.
[0037] The natural frequency represents a change in the thickness of the inspection object 30. The thickness of the inspection object 30 made of a magnetic material is reduced or increased due to corrosion or erosion of the surface. When the thickness is reduced, the peak value (voltage) of the natural frequency becomes smaller and the frequency becomes lower. Conversely, when the thickness is increased, the peak value (voltage) of the natural frequency becomes larger and the frequency becomes higher. Therefore, when the calculated natural frequency is lower than the natural frequency serving as the evaluation standard, it can be estimated that the thickness of the inspection object 30 is reduced, and when the calculated natural frequency is higher than the natural frequency serving as the evaluation standard, it can be estimated that the thickness of the inspection object 30 is increased. On the other hand, when the calculated natural frequency is the same as the natural frequency serving as the evaluation standard, it can be estimated that there is no change in the thickness of the inspection object 30. The evaluation unit 12 qualitatively evaluates the estimated change in the thickness of the inspection object 30. For example, the evaluation unit 12 may evaluate the inspection object 30 as "thinning", "thickening", "no change", or as "deteriorated" or "not deteriorated". In this way, the evaluation unit 12 can qualitatively evaluate the change in thickness of the inspection object 30 by comparing the calculated natural frequency with the natural frequency that serves as the evaluation standard.
[0038] In addition to or instead of the above-mentioned qualitative evaluation, the evaluation unit 12 may quantitatively evaluate the estimated change in thickness of the inspection object 30. For example, the evaluation unit 12 may evaluate the inspection object 30 as "essential thickness reduction of 2 mm," "essential thickness increase of 2 mm," "essential change of 0 mm," or may evaluate it as "essential plate thickness of 10 mm."
[0039] Furthermore, the peak shape of the waveform of the natural vibration represents the size of the defect in the inspection object 30. If a small hole (hole, crack) is formed in the inspection object 30, the peak of the waveform becomes broad. Conversely, if a large hole (hole, crack) is formed in the inspection object 30, the peak of the waveform becomes sharp. Therefore, if the peak of the calculated waveform of the natural vibration is broader than the peak of the waveform of the natural vibration that is the evaluation standard, it can be estimated that the inspection object 30 has a small hole, and the sharper the peak of the waveform is compared to the peak of the waveform of the natural vibration that is the evaluation standard, it can be estimated that the inspection object 30 has a larger hole. On the other hand, if the peak shape of the calculated waveform of the natural vibration is the same as the peak shape of the waveform of the natural vibration that is the evaluation standard, it can be estimated that the inspection object 30 does not have a hole. The evaluation unit 12 qualitatively evaluates the estimated size of the defect in the inspection object 30. For example, the evaluation unit 12 may evaluate the inspection object 30 as “small hole present”, “large hole present”, “no change”, or as “deteriorated” or “not deteriorated”. In this way, by comparing the peak shape of the calculated waveform of the natural vibration with the peak shape of the waveform of the natural vibration serving as the evaluation standard, the state of the inspection object 30 can be qualitatively evaluated.
[0040] In addition to or instead of the above-mentioned qualitative evaluation, the evaluation unit 12 may quantitatively evaluate the estimated size of the defect in the inspection object 30. For example, the evaluation unit 12 may evaluate the inspection object 30 as "having a small hole estimated to be 1 mm" or "having a large hole estimated to be 3 mm."
[0041] The evaluation unit 12 may estimate and evaluate the state of the inspection object 30 using a model constructed by machine learning the calculated natural frequency. The model is constructed in the generation unit 13. The generation unit 13 trains the model by machine learning to generate a trained model. The generation unit 13 may train the model using, for example, teacher data to which a correct answer label is assigned, or without using teacher data.
[0042] The machine learning model is a well-known model, and may be, for example, a deep learning model (deep neural network). The neural network may be a hardware circuit, or may be a virtual network by software that connects layers virtually constructed on a computer program by a processor 10a (see FIG. 2) described later.
[0043] In this embodiment, the generation unit 13 uses the natural frequency (initial value) of the inspection object 30 when it is new as training data, performs machine learning on the natural frequency of the inspection object 30 calculated before or before, inputs the currently calculated natural frequency of the inspection object 30, and generates a model that outputs an evaluation result of the state of the inspection object 30. The machine learning conditions (epoch number, batch size) may be set appropriately. The generation unit 13 may use natural frequencies within a certain range including the natural frequency when it is new (natural frequencies of non-defective products) as training data, perform machine learning on the model, and set a threshold value to evaluate the state of the inspection object 30 estimated based on the natural frequency.
[0044] The output unit 14 outputs the evaluation result by the evaluation unit 12 to the output device 40. The output device 40 may be provided separately from the control device 10, or may be built-in (mounted) or integrated with the control device 10. Examples of the output device 40 include a monitor, a projector, a printer, and the like. For example, a monitor may be provided to output a magnetic signal detected by the magnetic sensor 20, or the evaluation result by the evaluation unit 12 of the control device 10 may be output. The evaluation result may be provided visually on a screen of a monitor connected to the control device 10, or may be audibly notified by a buzzer sound, for example.
[0045] The storage unit 15 stores information such as the magnetic signal acquired from the magnetic sensor 20, the natural frequency calculated by the calculation unit 11, teacher data, the evaluation result by the evaluation unit 12, the threshold value for evaluation, machine learning conditions, the model generated by the generation unit 13, etc. Furthermore, the storage unit 15 may store data such as the distance between the magnetic sensor 20 and the inspection object 30, the number of inspections, information on the inspection object 30 (length, weight, shape, material, etc.), and the number of years of use of the inspection object 30.
[0046] <Hardware configuration> FIG. 2 is a block diagram showing an example of a hardware (HW) configuration of a computer that realizes the functions of the control device 10. Hereinafter, this is also referred to as "computer 10." The inspection device 1 of this embodiment is realized by a general-purpose computer 10 that can execute a computer program 10g for inspection (inspection program). As shown in FIG. 2, the computer 10 illustratively includes a processor 10a, a memory 10b, a storage unit 10c, an IF unit 10d, an IO unit 10e, and a reading unit 10f. These are connected to each other so as to be able to communicate with each other via a bus (control bus, data bus, etc.) provided inside the computer 10.
[0047] The processor 10a is an example of a processing unit that performs various controls and calculations. The processor 10a may be connected to each block in the computer 10 via a bus so that they can communicate with each other. Examples of the processor 10a include integrated circuits (ICs) such as a Central Processing Unit (CPU) and a Graphics Processing Unit (GPU).
[0048] The memory 10b and the storage unit 10c are an example of HW that stores various data, programs, and other information. The storage unit 10c may store a program 10g that realizes all or part of the various functions of the control device 10. The program 10g may be expanded in the memory 10b. The memory 10b and the storage unit 10c are an example of the storage unit 15 shown in FIG.
[0049] The IF unit 10d is an example of a communication IF that controls connection and communication with a network. For example, the IF unit 10d may include an adapter that complies with optical communication such as Ethernet (registered trademark), InfiniBand, Myrinet, or FC (Fibre Channel). The adapter may support one or both of wireless and wired communication methods. The program 10g may be downloaded from the network to the computer 10 via the communication IF and stored in the storage unit 10c.
[0050] The IO unit 10e is a circuit or software that inputs data or signals from the outside to a device or a system, or outputs data or signals to the outside. The IO unit 10e is an example of the output unit 14 shown in FIG. 1. The IO unit 10e may include one or both of an input device and an output device 40. Examples of the input device include a keyboard, a mouse, and a touch panel, and the output device 40 is as described above.
[0051] The reading unit 10f is an example of a reader (adapter, drive device, card reader, etc.) that reads out data or program information recorded on the recording medium 10h. The reading unit 10f may include a connection terminal or device to which the recording medium 10h can be connected or inserted. The recording medium 10h may store the program 10g, and the reading unit 10f may read the program 10g from the recording medium 10h and store it in the storage unit 10c.
[0052] The processor 10a of this embodiment deploys a program 10g (inspection program) stored in the storage unit 10c in the memory 10b and executes it to realize the functions (calculation unit 11, evaluation unit 12, generation unit 13) of the control device 10 illustrated in Fig. 1. The inspection program 10g may be configured to realize a process of acquiring a magnetic field change detected by the magnetic sensor 20. The processor 10a may also construct a machine learning model using a hardware circuit, a virtual network on a computer program by the processor 10a, or general-purpose software.
[0053] The storage unit 10c may further store a program (control program) for controlling the magnetic sensor 20. The computer 10 may expand the control program in the memory 10b and execute it to be connected to an IF unit (not shown) of the magnetic sensor 20 via the IF unit 10d so as to be capable of communicating with each other, and may remotely control the magnetic sensor 20. The whole or part of the program for controlling the magnetic sensor 20 may be included in the above-mentioned inspection program 10g.
[0054] The processor 10a acquires data (hereinafter also referred to as "necessary data") necessary to realize the functions of the computer 10 and to construct a machine learning model based on reception at the IF unit 10d, input from the input device of the IO unit 10e, and reading at the reading unit 10f, or by reading data stored in the memory unit 10c.
[0055] The above necessary data may be stored in advance in the memory 10b and the storage unit 10c, or may be input via an input device of the IO unit 10e. The necessary data includes the magnetic signal acquired from the magnetic sensor 20, the natural frequency calculated by the calculation unit 11, teacher data, the evaluation result by the evaluation unit 12, the threshold value for evaluation, the machine learning conditions, the model generated by the generation unit 13, and the like. Furthermore, the necessary data may include data such as the distance between the magnetic sensor 20 and the inspection object 30, the number of inspections, information on the inspection object 30, and the number of years of use of the inspection object 30.
[0056] Signals necessary for controlling the magnetic sensor 20 may be input via an input device of the IO unit 10e. These signals include signals that trigger the operation of the magnetic sensor 20, such as a detection start (START) or detection end (END).
[0057] The above-mentioned hardware configuration of the computer 10 (control device 10) is an example. Therefore, the hardware in the control device 10 may be increased or decreased (for example, adding or deleting any block), divided, integrated in any combination, or a bus may be added or deleted, as appropriate.
[0058] [3. Testing method] FIG. 3 is an example of a flowchart illustrating the procedure of the inspection method according to the embodiment, and shows the processing of the inspection device 1 described above. The inspection method according to the embodiment is a method for causing the control device 10 to execute the inspection (evaluation) of the inspection object 30 described above. This method can be realized by the inspection program 10g causing the computer 10 to execute the following processing. This flowchart is executed at least once, and is preferably repeatedly executed periodically or irregularly. FIG. 3 illustrates an example of a flowchart that is repeatedly executed periodically (at a predetermined calculation period). Note that the magnetic sensor 20 is controlled separately from this flowchart. This flowchart is started at the timing when the magnetic sensor 20 starts to detect a change in the magnetic field.
[0059] Step S1 is a step (acquisition step) of acquiring a magnetic field change caused by vibration of the inspection object 30 made of a magnetic material, and may be realized by the above-mentioned magnetic sensor 20. The magnetic sensor 20 acquires (detects) the magnetic field change as a magnetic signal (voltage-time signal). The acquisition step may be a process in which the magnetic sensor 20 detects the magnetic field change itself, or a process in which the magnetic field change detected by the magnetic sensor 20 is input to the control device 10. Step S2 is a step (calculation step) of analyzing the acquired magnetic field change and calculating the natural frequency of the inspection object 30, and is realized by the above-mentioned calculation unit 11. The calculation unit 11 performs Fourier analysis on the acquired magnetic signal and calculates the natural frequency (voltage-frequency signal) of the inspection object 30.
[0060] Step S3 is a step (evaluation step) of estimating and evaluating the state based on the calculated natural frequency, and is realized by the above-mentioned evaluation unit 12. The evaluation unit 12 estimates and evaluates the state of the inspection object 30 by comparing the natural frequency calculated this time with the natural frequency calculated previously or previously. Furthermore, the evaluation unit 12 estimates and evaluates the state of the inspection object 30 using a model constructed by machine learning the calculated natural frequency. As the evaluation standard (threshold value), either the natural frequency when new or a single or multiple natural frequencies of a non-defective product may be set.
[0061] If the evaluation unit 12 evaluates that there is a change in the inspection target 30 based on the calculated natural frequency, the process proceeds to a Yes route from step S3, and the output unit 14 outputs the evaluation result to the output device 40 (step S4; output step). In this case, the evaluation result is recorded in the storage unit 10c, and then this flow ends. Note that after outputting step S4, the process may return from step S4 and steps S1 to S3 may be executed repeatedly.
[0062] When the evaluation unit 12 evaluates that there is no change in the inspection object 30 based on the calculated natural frequency, it proceeds to the No route from step S3, returns from step S3 without outputting the evaluation result to the output device 40, and repeats steps S1 to S2. Note that the output unit 14 may be configured to output the evaluation result to the output device 40 even when it evaluates that there is no change in the inspection object 30.
[0063] [4. Actions and Effects] (1) In the above-mentioned inspection device 1, inspection method, and inspection program, a magnetic field change caused by vibration of the inspection object 30 made of a magnetic material is acquired and analyzed, and the state of the inspection object 30 is estimated and evaluated based on the calculated natural frequency. As a result, only information representing the state of the inspection object 30 in the vicinity of the magnetic sensor 20 can be easily acquired, so that it is more resistant to noise and has a higher spatial resolution than the hammering test method. In addition, as a result, magnetization of the inspection object 30 is not required, time and cost can be reduced compared to the leakage magnetic flux inspection method, and the type and inspection location of the inspection object 30 made of a magnetic material are not limited. In this way, the inspection device 1, inspection method (percussion test method), and inspection program of this embodiment are simpler and more accurate than conventional non-destructive inspection methods, so that the inspection object 30 can be inspected multiple times from the start of use. Therefore, for example, early detection (prediction) of deterioration of the inspection object 30 can be realized, and the state of the inspection object made of a magnetic material can be inspected simply and with high accuracy.
[0064] (2) The above-described technology further includes an output unit 14 that outputs an evaluation result of the estimated state of the inspection object 30. This allows the inspector to easily understand the state of the inspection object 30. (3) The above technology further utilizes the tunnel magnetoresistance effect to acquire (detect) magnetic field changes. This makes it possible to detect minute natural frequencies of the inspection object 30 in a demagnetized state. The TMR sensor has a higher magnetic field detection resolution than a Hall sensor that uses electromagnetic induction of a coil. Also, unlike SQUID (Superconducting QUantum Interference Device) sensors that use the quantization of magnetic flux in superconductivity, the TMR sensor can operate at room temperature and does not require any accessories other than the sensor. This allows for a cheaper and more compact device configuration.
[0065] (4) In the above-mentioned technology, furthermore, the inspection of the inspection object 30 is repeatedly performed, and the natural frequency calculated this time is compared with the natural frequency calculated previously to estimate and evaluate the state of the inspection object 30. This makes it possible to grasp, for example, the tendency and progress of deterioration of the inspection object 30 at an early stage, and therefore to deal with the deterioration at an early stage, thereby realizing a longer life of the inspection object 30 and reducing costs.
[0066] (5) In the above-described technology, a model constructed by machine learning the calculated natural frequency is further used to estimate and evaluate the state of the inspection object 30. In this way, by using machine learning, the state of the inspection object 30 can be more easily evaluated.
[0067] In the above-mentioned technique, the acquired magnetic field change is further subjected to Fourier analysis to calculate the natural frequency of the inspection object 30. In this way, since complicated calculations are not required, the calculation cost can be reduced.
[0068] [5.Other] The configurations of the magnetic sensor 20 and the control device 10 described above are merely examples and are not limited to those described above. The magnetic sensor 20 is not limited to a TMR sensor as long as it is a sensor capable of detecting low frequency bands and low magnetic fields. In the above embodiment, the magnetic sensor 20 stores the detected magnetic signal in a memory, and the control device 10 acquires the magnetic signal after the inspection. However, the magnetic sensor 20 may transmit the magnetic signal detected in real time during the inspection to the control device 10. In this case, the calculation of the natural frequency in the control device 10 and the estimation and evaluation of the state of the inspection target 30 may also be performed in real time.
[0069] Even when the evaluation unit 12 performs an inspection using a model, the machine learning of the model may be performed in an unsupervised manner. By learning in an unsupervised manner, it is possible to detect deterioration that cannot be detected by humans, and the condition of the inspection target 30 can be evaluated with a higher degree of accuracy.
[0070] In the above embodiment, the inspection object 30 is a plate material, but the inspection object 30 is not limited to a plate material. For example, when the inspection object 30 is a pipe, the predetermined direction is the radial direction of the pipe, and the estimated dimension of the thickness of the pipe and the change in the thickness can be evaluated. [Explanation of symbols]
[0071] 1. Inspection equipment 10 Control device (computer) 10a processor 10b Memory 10c storage section 10d IF section 10e IO section 10f Reading section 10g Program (Inspection Program) 10h Recording medium 11 Calculation section 12 Evaluation Section 13 Generation part 14 Storage section 15 Output section 20 Magnetic Sensor 30 Inspection object 40 Output Device
Claims
1. An inspection device for inspecting an inspection object made of a magnetic material, comprising: a magnetic sensor that detects a change in a magnetic field caused by vibration of the inspection object made of the magnetic material; A calculation unit that analyzes the detected magnetic field change and calculates a natural frequency of the inspection object; an evaluation unit that estimates and evaluates a state of the inspection object based on the calculated natural frequency; An inspection device comprising:
2. An inspection device for inspecting an inspection object made of a magnetic material, comprising: a calculation unit that acquires and analyzes a magnetic field change caused by vibration of the inspection object made of the magnetic material, and calculates a natural frequency of the inspection object; an evaluation unit that estimates and evaluates a state of the inspection object based on the calculated natural frequency; An inspection device comprising:
3. The apparatus further includes an output unit that outputs the estimated evaluation result of the state of the inspection object to an output device.
3. The inspection apparatus according to claim 1 or 2.
4. The magnetic field change is obtained by utilizing the tunnel magnetoresistance effect.
3. The inspection apparatus according to claim 1 or 2.
5. The inspection of the inspection object is repeatedly performed, The evaluation unit compares the currently calculated natural frequency with the previously calculated natural frequency, and estimates and evaluates the state of the inspection object.
3. The inspection apparatus according to claim 1 or 2.
6. The evaluation unit estimates and evaluates a state of the inspection object using a model constructed by machine learning the calculated natural frequency.
6. The inspection apparatus according to claim 5.
7. An inspection method for inspecting an inspection object made of a magnetic material, comprising: an acquiring step of acquiring a magnetic field change caused by vibration of the inspection object made of the magnetic material; a calculation step of analyzing the acquired magnetic field change and calculating a natural frequency of the inspection object; an evaluation step of estimating and evaluating a state of the inspection object based on the calculated natural frequency; An inspection method comprising:
8. The method further includes an output step of outputting the estimated evaluation result of the state of the inspection object to an output device.
8. The inspection method according to claim 7.
9. The acquiring step detects the magnetic field change by utilizing a tunnel magnetoresistance effect.
8. The inspection method according to claim 7.
10. The inspection of the inspection object is repeatedly performed, The evaluation step compares the currently calculated natural frequency with the previously calculated natural frequency, and estimates and evaluates the state of the inspection object.
8. The inspection method according to claim 7.
11. The evaluation step estimates and evaluates a state of the inspection object using a model constructed by machine learning the calculated natural frequency. The inspection method according to claim 10 .
12. An inspection program for inspecting an inspection object made of a magnetic material, acquiring and analyzing a magnetic field change caused by vibration of the inspection object made of the magnetic material, and calculating a natural frequency of the inspection object; A state of the inspection object is estimated and evaluated based on the calculated natural frequency. An inspection program that causes a computer to carry out processing.
Citation Information
Patent Citations
Leakage magnetic flux flaw detection method, and leakage magnetic flux flaw detection apparatus
JP2011196863A
