Inspection device, method and program

Abstract

To improve determination accuracy of a state of an apparatus in inspection of the apparatus based on hitting.SOLUTION: An inspection device comprises: an acquisition unit 32 that acquires a signal indicating vibration of a lighting device in hitting a pre-determined position of the lighting device which has a lighting fixture joined to a support pillar; a calculation unit 34 that calculates a mel frequency cepstrum coefficient for the acquired signal; a learning unit 36 that learns, for each condition, a probability density function of the mel frequency cepstrum coefficient which is calculated for each of learning signals acquired for each condition determined by a lapse period from use start time of the lighting device, a state of the lighting device and a hitting position; and a determination unit 40 that determines a condition corresponding to an inspecting signal by comparing between the calculated mel frequency cepstrum coefficient and the probability density function for each learned condition, for the acquired inspecting signal in which the condition is unknown.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to an inspection device, an inspection method, and an inspection program. [Background technology]

[0002] Conventionally, hammering inspection is performed to check whether or not there is an abnormality in a structure such as a road or a building based on a hammering signal generated when the structure is hammered. Various techniques related to such hammering inspection have been proposed.

[0003] For example, an inspection device has been proposed that acquires a hammering signal indicating the sound produced when any part of an instrument including a screw joint is struck, performs frequency analysis on the hammering signal, calculates the spectral flatness, and compares the calculated spectral flatness with a predetermined standard to output a prediction result of whether or not the screw is loose (see Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2023-3843 A Summary of the Invention [Problem to be solved by the invention]

[0005] An object of the present invention is to provide an inspection device, method, and program that can improve the accuracy of determining the condition of an instrument when inspecting the instrument by tapping. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, the inspection device of the present invention includes an acquisition unit that acquires a signal indicating vibration of a fixture when a predetermined position of the fixture joined to a pillar is tapped, a calculation unit that calculates Mel frequency cepstral coefficients for the signal acquired by the acquisition unit, a learning unit that learns, for each condition, a probability density function of the Mel frequency cepstral coefficients calculated for each of the signals acquired for each condition determined by the elapsed time since the start of use of the fixture, the state of the fixture, and the tapping position, and a determination unit that compares the Mel frequency cepstral coefficients calculated by the calculation unit with the probability density function for each condition learned by the learning unit for a signal acquired by the acquisition unit for which the condition is unknown, to determine the condition corresponding to the unknown signal.

[0007] According to the inspection device of the present invention, the acquisition unit acquires a signal indicating vibration of the fixture when a predetermined position of the fixture joined to a pole is tapped, the calculation unit calculates Mel frequency cepstral coefficients for the signal acquired by the acquisition unit, the learning unit learns a probability density function of the Mel frequency cepstral coefficients calculated for each of the signals acquired for each condition determined by the elapsed time from the start of use of the fixture, the state of the fixture, and the tapping position for each condition, and the determination unit compares the Mel frequency cepstral coefficients calculated by the calculation unit with the probability density function for each condition learned by the learning unit for the signal acquired by the acquisition unit for which the condition is unknown, thereby determining the condition corresponding to the unknown signal. This makes it possible to improve the accuracy of determining the state of the fixture.

[0008] The striking position may be any one of a plurality of predetermined positions of the tool, and the state of the tool may be a combination of positions of loosened bolts among a plurality of bolts included in the tool, thereby making it possible to predict the position of the loosened bolt.

[0009] The condition of the tool may be the presence or absence or the degree of at least one of corrosion and water retention of the support, and the striking position may be a position of the support at a predetermined height from the ground surface. This makes it possible to predict the conditions of corrosion and water retention.

[0010] The calculation unit may convert the calculated Mel-frequency cepstral coefficients of the first dimension into principal component scores of a second dimension lower than the first dimension by principal component analysis. This can reduce the calculation cost of the determination process and can further improve the determination accuracy.

[0011] The calculation unit may calculate the Mel-frequency cepstral coefficients by using an attenuation portion of the signal acquired by the acquisition unit, thereby making it possible to further improve the accuracy of determining the state of the appliance.

[0012] The striking position may be near a joint between the lighting fixture and the support pole, thereby improving the accuracy of determining the state of the fixture.

[0013] The device may also be a lighting device that is an accessory to a road.

[0014] Furthermore, the inspection method according to the present invention is a method in which an acquisition unit acquires a signal indicating vibration of a fixture when a predetermined position of the fixture, in which a lighting fixture is joined to a post, is tapped, a calculation unit calculates Mel frequency cepstral coefficients for the signal acquired by the acquisition unit, a learning unit learns, for each condition, a probability density function of the Mel frequency cepstral coefficients calculated for each of the signals acquired for each condition determined by the elapsed time since the start of use of the fixture, the state of the fixture, and the tapping position, and a determination unit compares, for a signal acquired by the acquisition unit for which the condition is unknown, the Mel frequency cepstral coefficient calculated by the calculation unit with the probability density function for each condition learned by the learning unit, to determine the condition corresponding to the unknown signal.

[0015] In addition, the inspection program of the present invention is a program for causing a computer to function as an acquisition unit that acquires a signal indicating vibration of a fixture when a predetermined position of the fixture joined to a pillar is tapped, a calculation unit that calculates Mel frequency cepstral coefficients for the signal acquired by the acquisition unit, a learning unit that learns, for each condition, a probability density function of the Mel frequency cepstral coefficients calculated for each of the signals acquired for each condition determined by the period of time since the fixture was first used, the state of the fixture, and the tapping position, and a determination unit that compares the Mel frequency cepstral coefficients calculated by the calculation unit with the probability density function for each condition learned by the learning unit for a signal acquired by the acquisition unit for which the condition is unknown, to determine the condition corresponding to the unknown signal. Effect of the Invention

[0016] According to the inspection device, method, and program of the present invention, it is possible to improve the accuracy of determining the condition of an instrument when inspecting the instrument by tapping. [Brief description of the drawings]

[0017] [Figure 1] FIG. 1 is a diagram for explaining an outline of the present embodiment. [Diagram 2] FIG. 13 is a diagram showing an example of a joint between a lamp and a post. [Diagram 3] FIG. 2 is a block diagram showing a hardware configuration of the inspection device. [Figure 4] FIG. 2 is a block diagram showing an example of a functional configuration of an inspection apparatus. [Diagram 5] FIG. 13 is a diagram showing an example of a pattern of combinations of positions of loose bolts. [Figure 6] FIG. 13 is a diagram showing an example of a striking position. [Figure 7] FIG. 13 is a diagram for explaining an attenuation portion. [Figure 8] FIG. 13 is a diagram for explaining the distribution of principal component scores of MFCC. [Figure 9] 13 is a flowchart showing the flow of a learning process. [Figure 10] 13 is a flowchart showing a flow of an inspection process. [Figure 11] FIG. 13 is a diagram for explaining the significance of the present embodiment. [Figure 12] FIG. 13 is a diagram for explaining the significance of the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will now be described with reference to the drawings. In this embodiment, an example of inspecting the state of a lighting device that is an accessory to a road will be described.

[0019] An outline of this embodiment will be described with reference to Fig. 1. In the inspection of this embodiment, when any part of the lighting device 50 is tapped with a tapping rod 52, a signal indicating vibration generated in the lighting device 50 is measured by a measuring device 54. Then, the measured signal is processed by an inspection device 10 to inspect the state of the lighting device 50.

[0020] The lighting device 50 includes a lamp 50A and a support pole 50B, and the lamp 50A is joined to the support pole 50B by a number of bolts. Fig. 2 shows an example of a joint between the lamp 50A and the support pole 50B. Fig. 2 shows an example in which the lamp 50A and the support pole 50B are joined by four bolts. In the following, the position of the right bolt as viewed from the top of the lamp 50A on the tip side (the side farther from the support pole 50B) of the lamp 50A is represented as FR, the position of the left bolt as FL, the position of the right bolt on the support pole 50B side of the lamp 50A is represented as RR, and the position of the left bolt as RL.

[0021] The measuring device 54 may be a vibration sensor that detects vibrations caused by tapping in the lighting device 50. By using a vibration sensor as the measuring device 54, road noise and the like is prevented from being included as noise in the measurement signal.

[0022] Fig. 3 is a block diagram showing a hardware configuration of the inspection device 10 according to this embodiment. As shown in Fig. 3, the inspection device 10 has a CPU (Central Processing Unit) 12, a memory 14, a storage device 16, an input device 18, an output device 20, a storage medium reading device 22, a communication I / F (Interface) 24, and an input / output I / F 26. Each component is connected to each other via a bus 28 so as to be able to communicate with each other.

[0023] The storage device 16 stores an inspection program for executing a learning process and an inspection process, which will be described later. The CPU 12 is a central processing unit, and executes various programs and controls each component. That is, the CPU 12 reads the program from the storage device 16 and executes the program using the memory 14 as a working area. The CPU 12 controls each of the components and performs various calculation processes according to the program stored in the storage device 16.

[0024] The memory 14 is made up of a RAM (Random Access Memory) and serves as a working area to temporarily store programs and data. The storage device 16 is made up of a ROM (Read Only Memory), a HDD (Hard Disk Drive), a SSD (Solid State Drive), etc., and stores various programs including an operating system and various data.

[0025] The input device 18 is a device for performing various inputs, such as a keyboard or a mouse. The output device 20 is a device for outputting various information, such as a display or a printer. A touch panel display may be adopted as the output device 20 to function as the input device 18. The storage medium reader 22 reads data stored in various storage media, such as a CD (Compact Disc)-ROM, a DVD (Digital Versatile Disc)-ROM, a Blu-ray disc, and a USB (Universal Serial Bus) memory, and writes data to the storage media.

[0026] The communication I / F 24 is an interface for communicating with other devices, and uses standards such as Ethernet (registered trademark), FDDI, or Wi-Fi (registered trademark). The input / output I / F 26 is an interface for connecting the inspection device 10 to peripheral devices. In this embodiment, the inspection device 10 and a measuring instrument 54 are connected via the input / output I / F 26, and a signal measured by the measuring instrument 54 is input to the inspection device 10.

[0027] Next, the functional configuration of the inspection device 10 according to this embodiment will be described. Fig. 4 is a block diagram showing an example of the functional configuration of the inspection device 10. As shown in Fig. 4, the inspection device 10 includes, as its functional configuration, an acquisition unit 32, a calculation unit 34, a learning unit 36, and a judgment unit 40. Each functional configuration is realized by the CPU 12 reading out an inspection program stored in the storage device 16, expanding it in the memory 14, and executing it. In addition, a probability density function DB (database) 38 is stored in a predetermined storage area of ​​the inspection device 10.

[0028] The acquiring unit 32 acquires a signal measured by the measuring instrument 54. In the following, a signal acquired for learning a probability density function for each condition (described in detail later) is referred to as a "learning signal," and a signal acquired by tapping the lighting device 50 to be inspected is referred to as an "inspection signal." When there is no distinction between the learning signal and the inspection signal, they are referred to as "measurement signals."

[0029] The learning signal is a signal measured by tapping the lighting device 50 under each of the conditions determined by the period of time elapsed since the start of use of the lighting device 50, the combination of the positions of the loosened bolts among the multiple bolts, and the tapping position. In this embodiment, the period of time elapsed since the start of use of the lighting device 50 is classified into two types: a new product and a deteriorated product. The deteriorated product is a lighting device 50 that has been in use for a predetermined number of years (for example, 10 years) since the start of use.

[0030] Also, the combination of positions of loose bolts is defined as patterns that are a combination of all possible combinations of whether each of the bolt positions (FL, FR, RL, RR) is fastened or loose, as shown in Fig. 5. In the example of Fig. 5, "X" indicates that the bolt at that position is loose. That is, in the example of Fig. 5, the combination of pattern number 1 (hereinafter referred to as "pattern 1"; the same applies to the other pattern numbers) indicates that all four bolts are fastened, and pattern 7 indicates that all four bolts are loose.

[0031] When inspecting the state of the lighting device 50 for loose bolts, the tapping positions may be set, for example, as shown in Fig. 6. The example in Fig. 6 shows an example in which four tapping positions A to D are set, where A is the tip of the lighting fixture 50A, B is the center of the lighting fixture 50A, C is between B and the base of the lighting fixture 50A, and D is the tip of the support post 50B that is joined to the lighting fixture 50A.

[0032] In the above example, the number of conditions is 2×9×4=72. The conditions are not limited to the above example. For example, three or more stages of time (e.g., 0 years, 3 years, 6 years, 9 years, etc.) may be set as the period of time elapsed since the start of use. In addition, the conditions may include 11 patterns that are combinations of the positions of the bolts that are loose among the five bolts, including the positions of the bolts that support the light bulb inside the lighting fixture 50A. In addition, other positions, such as the side of the lighting fixture 50A, may be included as the tapping position.

[0033] The learning signal includes multiple signals for each condition resulting from multiple taps. For example, for each condition, taps are performed multiple times (e.g., five times each) at multiple levels of strength (e.g., three levels: weak, medium, and strong). In this example, 15 learning signals are obtained for each condition.

[0034] The test signal is similar to the learning signal in that it is a measurement signal obtained by tapping the lighting device 50, but unlike the learning signal, it is a measurement signal whose above conditions are unknown. Among the conditions, the tapping position may be known.

[0035] The acquisition unit 32 extracts the attenuation portion from the acquired measurement signal. In the measurement signal, the vibration of a signal in a specific frequency band appears in the attenuation portion. It has been confirmed by the inventors' experiments that the specific frequency band is a frequency according to the above conditions. Therefore, by performing subsequent processing using the signal in the attenuation portion, it is possible to further improve the accuracy of determining the state of the lighting device 50.

[0036] 7, the acquisition unit 32 extracts a portion of the measurement signal from 0 to 0.005 s after the moment of hitting as a rising portion, and a portion of the measurement signal from 0.005 s onwards as a decay portion. The acquisition unit 32 transfers the signal of the decay portion extracted from the measurement signal to the calculation unit .

[0037] The calculation unit 34 calculates Mel-Frequency Cepstrum Coefficients (MFCC) for the attenuation section signal passed from the acquisition unit 32. Specifically, the calculation unit 34 performs a short-time Fourier transform on the attenuation section signal to calculate a power spectrum for each frequency. The calculation unit 34 multiplies the power spectrum by a 20-dimensional Mel filter bank to convert it into a Mel power spectrum. The calculation unit 34 converts the Mel power spectrum into a logarithmic spectrum, and then applies a discrete cosine transform to calculate a Mel cepstrum, and calculates MFCCs, which are characteristics of the spectral envelope, by removing stationary components and high quefrency components. Here, it is assumed that 12-dimensional MFCCs are calculated.

[0038] The calculation unit 34 converts the calculated 12-dimensional MFCCs into lower-dimensional (e.g., three-dimensional) principal component scores by principal component analysis. The calculation unit 34 passes the principal component scores of the MFCCs calculated from the attenuation parts of the learning signals to the learning unit 36. The calculation unit 34 also passes the principal component scores of the MFCCs calculated from the attenuation parts of the inspection signals to the determination unit 40.

[0039] The Mel frequency cepstrum coefficients of the first dimension may include fluctuations due to factors other than loosening of the bolts, but the principal component scores of the second dimension are reduced by dimensionality reduction for factors other than the fluctuations in the vibration signal due to loosening of the bolts. In other words, by calculating the principal component scores of the MFCCs, the solution can be constrained to a subspace that represents the fluctuations due to loosening of the bolts. This can reduce the calculation cost and further improve the judgment accuracy of the judgment unit 40.

[0040] The learning unit 36 ​​projects the principal component scores of the MFCCs calculated from the attenuation portion of the learning signal passed from the calculation unit 34 onto the principal component space for each condition, and learns a model that represents the distribution of the principal component scores in the principal component space. As shown in Fig. 8, the distribution of the principal component scores of the MFCCs forms a cluster for each condition. In the example of Fig. 8, the distribution is shown for conditions including patterns 2 (only RL is fastened) and 4 (only FL is fastened) in comparison with a condition including pattern 1 (all bolts are fastened).

[0041] Specifically, the learning unit 36 ​​projects the principal component scores for each condition into the principal component space, applies a normal distribution to the distribution of the principal component scores in the principal component space to perform maximum likelihood estimation, and calculates the probability density function shown in the following equation (1) for each condition.

[0042]

number

[0043] μ i is the cluster center (mean vector) of the i-th condition, Σ i is the covariance matrix (distribution variance) of the i-th condition, d is the number of dimensions of the principal components (d = 3), x is the principal component score of the MFCC calculated for the signal of the attenuation part of the inspection signal, y i is the value of the probability density function for the i-th condition. The learning unit 36 ​​stores the probability density function calculated for each condition in the probability density function DB 38.

[0044] The determination unit 40 compares the principal component scores of the MFCCs calculated from the attenuation portion of the inspection signal passed from the calculation unit 34 with the probability density function for each condition learned by the learning unit 36, and determines the condition corresponding to the inspection signal. Specifically, the determination unit 40 reads out the probability density function for each condition from the probability density function DB 38. Then, the determination unit 40 derives the likelihood y for each condition by substituting the principal component scores passed from the calculation unit 34 for x in equation (1). i The determination unit 40 determines the condition with the maximum likelihood as the condition corresponding to the inspection signal. As a result, it is possible to predict from the inspection signal whether the lighting device 50 is new or old, and where the loose bolt is located. The tapping position may be predicted from a probability density, or, if information on the tapping position can be acquired together with the inspection signal, the condition corresponding to the inspection signal may be determined by comparing with a probability density distribution including the acquired tapping position as a condition. The determination unit 40 outputs the determined condition as the inspection result.

[0045] Next, the operation of the inspection device 10 according to this embodiment will be described.

[0046] Fig. 9 is a flowchart showing the flow of the learning process executed by the CPU 12 of the inspection device 10, and Fig. 10 is a flowchart showing the flow of the inspection process executed by the CPU 12 of the inspection device 10. The CPU 12 reads out the inspection program from the storage device 16, expands it in the memory 14, and executes it, whereby the CPU 12 functions as each functional component of the inspection device 10, and each of the learning process and the inspection process is executed. Note that the learning process and the inspection process are examples of the inspection method of the present invention.

[0047] First, the learning process shown in FIG. 9 will be described.

[0048] In step S10, the acquisition unit 32 acquires learning signals measured multiple times for each condition by the measuring device 54. The acquisition unit 32 then extracts attenuation sections from each of the learning signals. Next, in step S12, the calculation unit 34 calculates, for example, 12-dimensional MFCCs for each of the signals of the attenuation sections extracted from the learning signals. Next, in step S14, the calculation unit 34 converts the calculated 12-dimensional MFCCs into, for example, three-dimensional principal component scores by principal component analysis.

[0049] Next, in step S16, the learning unit 36 ​​projects the principal component scores of the MFCCs calculated from the attenuation part of the learning signal into the principal component space for each condition, applies a normal distribution to the distribution of the principal component scores in the principal component space to perform maximum likelihood estimation, and calculates a probability density function for each condition. Next, in step S18, the learning unit 36 ​​stores the probability density function calculated for each condition in the probability density function DB 38, and the learning process ends.

[0050] Next, the inspection process shown in FIG. 10 will be described.

[0051] In step S20, the acquisition unit 32 acquires the inspection signal measured by the measuring instrument 54, and extracts the attenuation portion from the inspection signal. Next, in step S22, the calculation unit 34 calculates, for example, 12-dimensional MFCCs for the signal of the attenuation portion extracted from the inspection signal. Next, in step S24, the calculation unit 34 converts the calculated 12-dimensional MFCCs into, for example, three-dimensional principal component scores by principal component analysis.

[0052] Next, in step S26, the determination unit 40 reads out the probability density function for each condition from the probability density function DB 38. Then, the determination unit 40 calculates the likelihood for each condition by substituting the principal component score of the MFCC calculated from the signal of the attenuation part of the inspection signal into the probability density function for each condition, and determines the condition with the maximum likelihood as the condition corresponding to the inspection signal. Next, in step S28, the determination unit 40 outputs the determined condition as the inspection result, and the inspection process ends.

[0053] As described above, the inspection device according to the present embodiment acquires a signal indicating the vibration of a fixture when a predetermined position of the fixture, in which a lighting fixture is joined to a pole, is tapped, and calculates Mel frequency cepstral coefficients for the acquired signal. The inspection device also learns, for each condition, a probability density function of the Mel frequency cepstral coefficients calculated for each of the signals acquired for each condition determined by the period of time since the fixture was first used, the state of the fixture, and the tapping position. Then, the inspection device compares the Mel frequency cepstral coefficients calculated for the acquired signal, whose condition is unknown, with the learned probability density function for each condition to determine the condition corresponding to the unknown signal. This improves the accuracy of determining the state of the fixture.

[0054] Here, the significance of using MFCCs as the features of a measurement signal will be explained.

[0055] Fig. 11 shows the results of verification of the judgment accuracy of this inspection device, which was performed using a Panasonic roadway lighting fixture YA34194K as the lighting device 50, a test hammer TH-4 as the percussion rod 52, and a Toyo Corporation three-axis accelerometer 356A06 as the measuring instrument 54. Fig. 11 shows the accuracy rate, which is the ratio of the inspection signals for which the judged condition is correct, to the inspection signals for all conditions, for each tapping position. Also, for each tapping position, the bar graph on the left shows the accuracy rate when the condition corresponding to the first likelihood out of the likelihoods calculated for each condition is correct, and the bar graph on the right shows the accuracy rate when any of the conditions corresponding to the first to third likelihoods is correct. In this verification, the verification was performed under 2 x 11 x 4 = 88 conditions, which were combinations of the time elapsed since the start of use of the lighting device 50 (new or old), 11 combinations of the positions of the loose bolts out of the five bolts, and four striking positions.

[0056] As shown in Fig. 11, it was found that the accuracy was high near the joint between the lamp 50A and the post 50B (especially on the post 50B side). Also, as shown in Fig. 12, at the tapping positions C and D, there was no failure to detect loose bolts under 88 different conditions.

[0057] In the above embodiment, the combination of positions where the bolts are loosened is determined as the state of the lighting device, but the present invention is not limited to this. For example, the presence or absence or degree of at least one of the corrosion and water retention of the pillar may be determined as the state of the lighting device. In this case, the condition may include a case where the pillar is corroded or water retention, a case where the pillar is not corroded or water retention, or a state where the degree of corrosion or water retention differs stepwise. In this case, the tapping position may be the position of the pillar at a predetermined height (for example, 15 cm) from the ground surface. In addition, a probability density function group for a condition including a combination of positions where the bolts are loose and a probability density function group for a condition including the presence or absence or degree of at least one of the corrosion and water retention of the pillar may be stored in the probability density function DB. Then, one of the probability density function groups may be selectively used. Furthermore, both the combination of positions where the bolts are loose and the presence or absence or degree of at least one of the corrosion and water retention of the pillar may be included as the condition of the state of the lighting device.

[0058] In the above embodiment, the inspection device is configured as one unit, but the present invention is not limited thereto. For example, when the tapping is performed at a high altitude, the inspection device needs to be moved to the high altitude, and the inspection device needs to be made extremely lightweight. Therefore, the CPU mounted on the inspection device needs to be small and lightweight, and the processing capacity is limited. Therefore, a distributed configuration may be adopted in which the inspection device 10A acquires a signal (waveform data) measured by the measuring device 54 at a high altitude, and the inspection device 10B acquires a signal transmitted from the inspection device 10A on the ground or the like and executes learning and inspection processing.

[0059] In addition, the learning process and the inspection process that the CPU reads the software (program) and executes in the above embodiment may be executed by various processors other than the CPU. In this case, examples of the processor include a PLD (Programmable Logic Device) such as an FPGA (Field-Programmable Gate Array) whose circuit configuration can be changed after manufacture, and a dedicated electric circuit such as an ASIC (Application Specific Integrated Circuit) which is a processor having a circuit configuration designed exclusively for executing a specific process. In addition, the learning process and the inspection process may be executed by one of these various processors, or may be executed by a combination of two or more processors of the same or different types (for example, a plurality of FPGAs, and a combination of a CPU and an FPGA). In addition, the hardware structure of these various processors is, more specifically, an electric circuit that combines circuit elements such as semiconductor elements.

[0060] In the above embodiment, the inspection program is pre-stored (installed) in the storage device, but the present invention is not limited to this. The program may be provided in a form recorded on a recording medium such as a CD-ROM, a DVD-ROM (Digital Versatile Disc Read Only Memory), or a USB (Universal Serial Bus) memory. The program may be downloaded from an external device via a network. [Explanation of symbols]

[0061] 10 Inspection equipment 12 CPU 14. Memory 16 Storage device 18 Input Devices 20 Output Devices 22 Storage media reader 24 Communication I / F 26 Input / Output Interface 28 Bus 32 Acquisition Department 34 Calculation section 36 Learning Department 38 Probability Density Function DB 40 Judgment section 50 Lighting Equipment 50A light fixture 50B Post 52 Percussion Rod 54 Measuring Instruments

Claims

1. An acquisition unit that acquires a signal indicating vibration of a fixture when a predetermined position of the fixture is struck, the fixture being joined to a pole; A calculation unit that calculates Mel-frequency cepstral coefficients for the signal acquired by the acquisition unit; A learning unit that learns a probability density function of Mel frequency cepstral coefficients calculated for each signal acquired for each condition determined by the elapsed time from the start of use of the tool, the state of the tool, and the tapping position for each condition; a determination unit that compares the Mel-frequency cepstral coefficients calculated by the calculation unit with the probability density function for each of the conditions learned by the learning unit, for the signal whose condition is unknown and acquired by the acquisition unit, to determine a condition corresponding to the unknown signal; The inspection device includes:

2. The striking position is one of a plurality of predetermined positions of the tool; The state of the tool is a combination of positions of bolts that have loosened among a plurality of bolts included in the tool.

2. The inspection device according to claim 1.

3. The condition of the device is the presence or absence or the degree of at least one of corrosion and water retention of the support, The striking position is a position of the pillar at a predetermined height from the ground surface.

2. The inspection device according to claim 1.

4. The inspection device according to any one of claims 1 to 3, wherein the calculation unit converts the calculated Mel-frequency cepstral coefficients of a first dimension into principal component scores of a second dimension lower than the first dimension by principal component analysis.

5. 4. The inspection device according to claim 1, wherein the calculation unit calculates the Mel-frequency cepstral coefficients by using an attenuation portion of the signal acquired by the acquisition unit.

6. The inspection device according to any one of claims 1 to 3, wherein the striking position is in the vicinity of a joint between the lighting fixture and the post.

7. 4. The inspection device according to claim 1, wherein the device is a lighting device that is an accessory to a road.

8. The acquisition unit acquires a signal indicating vibration of the fixture when a predetermined position of the fixture in which the lighting fixture is joined to a pole is struck, a calculation unit that calculates Mel-frequency cepstral coefficients for the signal acquired by the acquisition unit; A learning unit learns a probability density function of Mel frequency cepstral coefficients calculated for each signal acquired for each condition determined by the elapsed time from the start of use of the tool, the state of the tool, and the tapping position for each condition; a determination unit, for the signal acquired by the acquisition unit and having an unknown condition, comparing the Mel-frequency cepstral coefficient calculated by the calculation unit with a probability density function for each of the conditions learned by the learning unit, to determine a condition corresponding to the unknown signal; Testing method.

9. Computer, an acquisition unit that acquires a signal indicating vibration of a fixture when a predetermined position of the fixture, in which the lighting fixture is joined to a post, is struck; A calculation unit that calculates Mel-frequency cepstral coefficients for the signal acquired by the acquisition unit; A learning unit that learns a probability density function of Mel frequency cepstral coefficients calculated for each signal acquired for each condition determined by the elapsed time from the start of use of the tool, the state of the tool, and the tapping position for each condition; and a determination unit that compares the Mel-frequency cepstrum coefficients calculated by the calculation unit with the probability density function for each of the conditions learned by the learning unit for the signal acquired by the acquisition unit and whose condition is unknown, to determine a condition corresponding to the unknown signal. Inspection program to function as a.

Images