Acoustic inspection method and device

Abstract

To identify the sound from a noise source that may emit the sound that is the same or similar to the sound that is emitted from an inspection object and the sound from the inspection object.SOLUTION: An acoustic inspection method collects the horn sound of a vehicle on an inspection line by a microphone and acquires the sound as sound data at a sampling frequency of 96 kHz (S1), separates the acquired sound into the audible sound and the inaudible sound on the high frequency side (S2), performs frequency analysis (S3), determines and extracts the horn sound on the basis of the frequency characteristics of the audible sound component (S4), compares the frequency characteristics of the inaudible region with the characteristics of the inaudible sound of the normal horn and determines whether the acquired sound is the noise (horn sound of the vehicle on an adjacent inspection line) or the correct sound of the inspection object (S5), generates an image displaying a screen of the frequency characteristics and the like of both of the audible sound and the inaudible sound (S6) and displays the image on a display (S7).SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a method and apparatus for acoustic inspection, for example, for performing sound-based inspections as part of an automobile inspection process. [Background technology]

[0002] For example, in the final stage of an automobile production line, in the finished vehicle inspection process, inspectors test drive the finished vehicle on free rollers and inspect numerous items, including the engine, meters, brakes, horn, and lights. Generally, during this inspection process, the presence or absence of abnormalities is determined by the inspector's sensory evaluation of various sounds emitted from various parts of the vehicle. For example, in the case of a horn, the inspector sounds the horn and listens to it to confirm that it is normal. Sounds emitted from the transmission during acceleration and from the brakes during braking are also subject to sensory inspection.

[0003] Instead of such sensory testing, attempts have been made to detect abnormalities by capturing sound with a microphone and analyzing the signal. In acoustic testing using such microphone-captured sound, a major problem is noise, which is sounds generated in places other than the test target or in other equipment.

[0004] For example, Patent Document 1 discloses an audio signal processing device that uses two microphones to remove or reduce zoom operation sounds and the like that get mixed in as noise in the audio captured by the microphone of a video camera, and estimates an operating sound spectrum signal that represents the operating sound by calculating and processing the audio spectrum signals of the audio captured by each microphone based on the relative positional relationship between the sound-generating body that generates the operating sound and each microphone. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 5594133 Summary of the Invention [Problem to be solved by the invention]

[0006] For example, in the finished vehicle inspection process, inspections are carried out in parallel on multiple inspection lines. Therefore, while a specific vehicle is being acoustically inspected using a microphone on one inspection line, sounds generated on other inspection lines can become noise. In this case, a noise source that emits the same or similar sound as the object being inspected (i.e., the same object on another vehicle) may exist on an adjacent inspection line, making it extremely difficult to separate this as noise using frequency analysis or other methods. For example, when an inspector performs a horn inspection by sounding the horn, a horn may sound on another inspection line and mix with the sound picked up by the microphone.

[0007] The technology in Patent Document 1 requires two microphones, and the relative positions of these two microphones and the noise source must be known. Therefore, it is not a technology that can determine noise from the sound itself regardless of the number of microphones, and it cannot handle cases where noise occurs in unknown, unspecified positions. [Means for solving the problem]

[0008] The acoustic inspection method according to the present invention comprises: Acquire the sound emitted from the test object, The acquired sound is separated into audible sound and inaudible sound on the higher frequency side, Analyzing the frequency characteristics of the audible sound and the inaudible sound, Whether or not the acquired sound contains noise is determined from the frequency characteristics of the inaudible sound. When it is determined that noise is included, the noise components are extracted from each of the audible and inaudible sounds, The noise components of inaudible sounds and the noise components of audible sounds are stored as a noise database in which they are linked to each other.

[0009] Generally, a sound source that emits a sound in the audible range also contains high-frequency inaudible sounds, such as higher-order harmonic components. Similar to audible sounds, this inaudible sound can be analyzed to identify certain characteristics. High-frequency inaudible sounds attenuate more with distance in the air than audible sounds, and even within the inaudible sound range, the attenuation with distance increases toward the higher frequencies. Therefore, even if the sound of the test object and the sound of the noise source are similar, it is easy to determine that the sound from a noise source located relatively far away from the test object is noise. [Effects of the Invention]

[0010] According to this invention, in sound-based inspections such as in the finished vehicle inspection process of automobiles, it is possible to easily determine that sounds generated at locations other than the object of inspection are noise, thereby improving the accuracy of sound-based inspections. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a functional block diagram of a first embodiment in which the present invention is applied to horn inspection in the finished vehicle inspection process. [Figure 2] 4 is a flowchart showing the flow of processing in the first embodiment. [Figure 3] FIG. 3 is an explanatory diagram showing a display example on the display unit of the first embodiment. [Figure 4] FIG. 10 is a functional block diagram of a second embodiment. [Figure 5] 10 is a flowchart showing the flow of processing in a second embodiment. [Figure 6] FIG. 10 is an explanatory diagram showing a display example on a display unit according to a second embodiment. [Figure 7] FIG. 10 is a functional block diagram of a third embodiment. [Figure 8] 10 is a flowchart showing the flow of processing according to a third embodiment. [Figure 9] FIG. 11 is an explanatory diagram showing a display example on a display unit according to a third embodiment. [Figure 10] FIG. 11 is an explanatory diagram showing another example of the display on the display unit according to the third embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0012] An embodiment of the present invention applied to the inspection of an automobile horn will be described below. This horn inspection is performed, for example, during the final stage of an automobile production line, known as the finished vehicle inspection process. Typically, during the finished vehicle inspection process, an inspector test-drives the finished vehicle on a free roller and inspects various components, including the engine, meters, and brakes. During this finished vehicle inspection, the inspector sounds the horn by pressing a horn switch on the steering wheel according to a predetermined inspection procedure, and the horn is inspected based on the sound produced. Since anomalies such as an incorrect horn pitch or tone can occur due to anomalies in the horn itself, an incorrect part (horn) model number, poor wiring contact, or anomalies in the horn switch, the inspection is based on the actual sound of the horn.

[0013] The horn inspection in this embodiment is not a fully automated process that determines whether the horn is normal or abnormal, but rather is a type of inspection support system in which an inspector, etc., who looks at the display (described later) makes the final determination of normality or abnormality.

[0014] Figure 1 shows Contains the essential parts of the present invention 1 shows a functional block diagram of an inspection device according to a first embodiment. Note that this inspection device is not an independent device as a horn inspection device, but is configured as part of a finished vehicle inspection device. The inspection device according to the first embodiment includes a sound acquisition unit 10, an audible / inaudible sound separation unit 20, an audible sound frequency characteristic analysis unit 30, an inaudible sound frequency characteristic analysis unit 40, a target sound identification unit 50, a target sound characteristic storage unit 60, a noise determination unit 70, a display image / video generation unit 80, and a display unit 90. The target sound characteristic storage unit 60 includes an inaudible sound characteristic storage unit 61 and an audible sound characteristic storage unit 62.

[0015] The sound acquisition unit 10 includes a microphone that acquires sounds generated from the vehicle under inspection and converts them into electrical signals, i.e., sound data, and a recording unit that temporarily stores the sound data in a digital form. The microphone is placed outside the vehicle so that it can collect sounds from the vehicle, including the horn sound. The microphone's directivity and frequency characteristics are selected according to the measurement target. Typically, a microphone with directivity toward the vehicle is used. Sound data from the horn under inspection may be obtained by localizing the sound source using a microphone array or the like. The sound acquisition is performed for, for example, several seconds, so as to roughly coincide with the timing when the inspector sounds the horn.

[0016] Here, the sound acquisition unit 10 needs to acquire audible sounds and inaudible sounds that are higher in frequency than the audible sounds, so it acquires and digitizes the sounds in a 96 kHz / 24-bit format, which is one standard for so-called high-resolution audio. By using a sampling frequency of 96 kHz, sounds up to 48 kHz, which is beyond the audible range, can be acquired and reproduced. Therefore, microphones and other devices that are compatible with this inaudible range are used.

[0017] The audible / inaudible sound separator 20 separates the acquired sound into audible and inaudible components. The upper limit of audible sound is generally said to be 20 kHz, and here, 20 kHz is used as the boundary for separating audible and inaudible sounds. For example, the two frequency components are separated using a low-pass filter or high-pass filter with a cutoff frequency of 20 kHz. Note that the actual upper limit of audible sound varies depending on factors such as age and gender, and the frequency that should be the boundary between audible and inaudible sounds may also differ depending on the test subject and test environment. Therefore, it is desirable to configure the system so that an inspector or other person can arbitrarily set the frequency that serves as the boundary for separation within a range of, for example, 10 kHz to 20 kHz.

[0018] The audible sound frequency characteristic analysis unit 30 performs frequency analysis on the audible sound separated by the audible sound / inaudible sound separation unit 20, i.e., the audible sound components contained in the acquired sound data. For example, frequency analysis methods such as FFT (Fast Fourier Transform) and wavelet analysis are used to convert the data according to frequency, thereby generating frequency data (audible sound frequency data).

[0019] When frequency data is expressed in two dimensions, for example, it is expressed as a frequency spectrum with frequency on the horizontal axis and power (or sound pressure) on the vertical axis. Note that the vertical axis of the frequency spectrum is not limited to power or sound pressure, and other parameters may be used, such as the product of the A-characteristic function or C-characteristic function of hearing, or loudness, which is a sensory quantity of hearing defined by ISO. Also, all data extracted for a sound of about several seconds may be converted into frequency characteristics all at once. By overlapping the conversion results for each time period in a time series, the frequency data may be handled as a so-called spectrogram, which is three-dimensional data including time.

[0020] The non-audible sound frequency characteristic analysis unit 40 performs frequency analysis on the non-audible sound separated by the audible sound / non-audible sound separation unit 20, i.e., the non-audible sound components contained in the acquired sound data. As with the audible sound frequency characteristic analysis unit 30, it generates frequency data (non-audible sound frequency data) by converting according to frequency using a frequency analysis method such as FFT (Fast Fourier Transform) or wavelet analysis.

[0021] Inaudible sound frequency data is, for example, a two-dimensional frequency spectrum with frequency on the horizontal axis and power (or sound pressure) on the vertical axis. However, unlike audible sound, parameters related to hearing are not used on the vertical axis. It may also be a three-dimensional spectrogram including a time axis.

[0022] The target sound identification unit 50 uses the audible sound frequency data obtained by the audible sound frequency characteristic analysis unit 30 to extract the sound to be tested, thereby identifying whether the acquired sound is the target sound, i.e., the sound to be tested. For example, if the test object is a horn, the unit focuses on sounds in a frequency band of 2 to 3 kHz, which is a frequency band characteristic of horn sounds, and identifies the target sound based on whether its sound pressure is 95 dBA or higher. In other words, before determining whether there is strict noise due to inaudible sound, the unit preliminarily determines whether the audible sound includes a sound emitted by the test object based on the frequency characteristics of the audible sound.

[0023] Unlike a noise-free environment such as a soundproof room, an environment such as a factory where finished vehicle inspections are performed contains a variety of relatively loud noises, and the acquired sound is likely to vary significantly compared to the actual sound of the object being inspected. Therefore, by first identifying the sound to be inspected based on the frequency characteristics and sound pressure of audible sounds, it is possible to prevent the sound to be inspected from being overlooked.

[0024] In addition to the horn inspection, sound inspections of other parts of the vehicle can be performed in parallel. In this case, the target sound identification unit 50 determines which part the sound is coming from. For example, in a completed vehicle inspection, a test is conducted in which the vehicle accelerates on free rollers to a predetermined speed. During this acceleration test, the presence or absence of abnormal sounds from the engine and the transmission is checked, and the sound generated by the brakes when the brakes are applied is also checked. When multiple inspection targets are included in this manner, the target sound identification unit 50 identifies which inspection target the sound is from, and each can be subjected to noise judgment, as described below. Note that since these series of inspections are performed in a predetermined order, the sound acquisition unit 10 can acquire sounds and generate sound data roughly divided into each inspection.

[0025] The noise determination unit 70 uses the non-audible sound frequency data obtained by the non-audible sound frequency characteristic analysis unit 40 to determine whether the acquired sound contains noise. This is done by comparing the frequency characteristics (e.g., two-dimensional frequency spectrum) of the non-audible sound components of the acquired sound with the data stored in the non-audible sound characteristic storage unit 61. The non-audible sound characteristic storage unit 61 is a database that stores a large amount of frequency data (such as frequency spectra and spectrograms) of the non-audible sound components of the sound of a normal test object (e.g., a horn) or data on its characteristic parts. The noise determination unit 70 compares the frequency characteristics of the non-audible sound of the acquired sound with the data stored in the non-audible sound characteristic storage unit 61, which indicates the frequency characteristics of the original non-audible sound of the test object, and determines, for example, using a general statistical method, based on the difference or similarity between the two, whether the acquired sound is the original test object sound or a sound containing noise (including the case where it is noise itself).

[0026] In acoustic inspections based on the sounds emitted by various devices, ambient noise can be a problem. When a microphone picks up sounds from noise sources that may emit sounds identical or similar to those emitted by the object being inspected, it is generally difficult to identify and separate them, even using audible sound frequency analysis. For example, finished vehicle inspections are often carried out in parallel on multiple inspection lines within the same factory, and while sound is being collected for horn inspection on one inspection line, a horn may sound on another inspection line. In such cases, the horn sounds from other inspection lines, which are noise, have frequency characteristics similar to those of the horn sound (target sound) of the object being inspected, making it difficult to distinguish them audibly.

[0027] In contrast, when we focus on high-frequency inaudible sounds (the 20 kHz to 48 kHz component in the example above), these high-frequency inaudible sounds have a greater attenuation of power during air propagation, i.e., greater attenuation over distance, compared to audible sounds. Furthermore, the higher the frequency band, the greater the attenuation over distance. Therefore, the frequency characteristics change or deform when the distance from the sound source is relatively long. Here, the distance from the vehicle (horn) on the adjacent inspection line to the microphone is relatively long compared to the distance from the vehicle (horn) under inspection to the microphone. Therefore, even if the same horn sounds are being heard, it is possible to distinguish whether they are the horn sound being tested or a horn sound that becomes noise.

[0028] Furthermore, high-frequency inaudible sounds have the characteristic of being more directional than audible sounds. Therefore, while sound from the vehicle being inspected, at which the microphone is aimed, propagates linearly to the microphone, sound from other vehicles on the inspection line that reaches the same microphone travels a longer distance due to reflections, etc., making the effect of distance-related attenuation even more pronounced.

[0029] In a preferred embodiment, the cutoff frequency of the inaudible sound (the frequency at which the power or sound pressure falls below a certain lower limit (a level at which an audible sound cannot be heard by the human ear)) shifts to a lower frequency, thereby determining that the sound is the horn sound of a vehicle on another inspection line that is not the target of the inspection. For example, FIG. 3 shows an example of a display on the display unit 90 (described later). The frequency characteristic indicated by reference numeral 90g on the right side of FIG. 3 is the inaudible frequency characteristic of the normal target sound (the horn sound of the vehicle being inspected, toward which the microphone is pointed), whereas the inaudible frequency characteristic of the horn sound of a vehicle on an adjacent inspection line that is relatively far away has a large attenuation on the high frequency side, as indicated by reference numeral 90f. Therefore, depending on the position of the cutoff frequency indicated by reference numeral Fc, it is possible to distinguish whether the acquired sound, including the audible sound, is the horn sound of the vehicle being inspected or noise, such as the horn sound of a vehicle other than the vehicle being inspected. In addition to the cutoff frequency, it is also possible to make comparisons in the inaudible range using the slope angle of the attenuation rate of power or sound pressure, or the amount of energy obtained by integrating the power or sound pressure up to a specified frequency, or comparisons based on some other physical quantity.

[0030] The audible sound characteristics storage unit 62 in the target sound characteristics storage unit 60, like the inaudible sound characteristics storage unit 61, is made up of a database that stores a large amount of frequency data (frequency spectrum, spectrogram, etc.) of the sound of a normal test object (e.g., a horn), particularly its audible sound components, or data on its characteristic parts. Although not described in detail, the audible sound frequency data obtained by the audible sound frequency characteristics analysis unit 30 may be compared with the data stored in the audible sound characteristics storage unit 62 in order to determine abnormal sounds, identify noise, etc.

[0031] The display image / video generation unit 80 generates images or videos to be displayed on the display unit 90, representing the results obtained as described above (such as the frequency characteristics of audible and inaudible sounds and noise determination results) and the processing steps leading up to the results. The display unit 90 is a display means for displaying the generated images or videos to relevant parties, such as the inspector driving the vehicle, the production manager, and the data scientist who utilizes the data. For example, the display unit 90 may be configured with an LCD display, an organic light-emitting diode (OLED) display, a head-mounted display (HMD), a smartwatch, or the like. In addition, if audio is involved, the display unit 90 may also include a sound source, amplifier, speaker, and the like for generating and emitting sound. The display image / video generation unit 80 generates various displays, such as still images, videos, and animations, and performs splitting, compositing, and switching between them to provide appropriate displays. Display timing control also makes it possible to display different images on multiple display units 90 at appropriate times.

[0032] FIG. 3 is an explanatory diagram showing an example of a display on the display unit 90 of the first embodiment. In this example, the final decision is left to the inspector, and the frequency characteristics, which are one of the factors for making the decision, are mainly displayed. As shown in the figure, a two-dimensional graph frame (see reference numeral 90a) is displayed on the display screen, with the horizontal axis representing "frequency (kHz)" and the vertical axis representing "sound pressure / power (dB)." The frequency characteristics of the acquired sound are plotted here. Here, the area to the left of 20 kHz, which is the threshold for audible and inaudible sounds, is labeled "[Audible Range Characteristics]" (see reference numeral 90b), and the area to the right is labeled "[Inaudible Range Characteristics]" (see reference numeral 90c), indicating the audible and inaudible sound ranges, respectively. The inaudible sound area is enclosed by a rectangular frame (see reference numeral 90d).

[0033] In the audible range on the left side, a black characteristic line (see reference symbol 90e) is drawn to represent the characteristics of the audible sound frequency data obtained by the audible sound frequency characteristic analysis unit 30. A red frame line (see reference symbol 90h) surrounding a specific frequency band indicates the frequency range (for example, the range from f1 to f2) used in the target sound identification unit 50 described above, and a red line (see reference symbol 90i) indicating the threshold sound pressure (for example, 95 dB) is added.

[0034] In the inaudible sound range on the right side, the characteristics of the inaudible sound frequency data obtained by the inaudible sound frequency characteristic analysis unit 40 are shown as blue characteristic lines (see symbol 90f). in For comparison, a red characteristic line (see symbol 90g) is drawn for the standard characteristic of a normal target sound stored in the inaudible sound characteristic storage unit 61. The position of the cutoff frequency Fc for the characteristic of the acquired sound (see symbol 90f) is indicated by a red arrow (see symbol 90j).

[0035] This display allows inspectors to read the characteristics of the audible and inaudible sound ranges together. In the example of Figure 3, the characteristic line of the acquired sound indicated by reference numeral 90f in the inaudible sound range deviates from the original characteristic indicated by reference numeral 90g, and as is clear from the fact that the cutoff frequency Fc is relatively low-frequency, this is not the target sound but noise (e.g., horn noise) from a vehicle on an adjacent inspection line. The display shown in Figure 3 makes this easily understandable to inspectors.

[0036] If the acquired sound is from the correct test object and does not contain noise (such as horn sounds from other test lines), the acquired sound will be used to determine whether there is an abnormal sound, but since this abnormal sound determination is not the subject of the present invention, its explanation will be omitted.

[0037] 2 is a flowchart showing the processing flow of the inspection device of the first embodiment described above. First, sounds, including, for example, horn sounds, are collected from vehicles on the inspection line using the microphone of the sound acquisition unit 10 and acquired as sound data (Step 1). As described above, for example, a sampling frequency of 96 kHz is used to acquire sounds in the range up to 48 kHz. Next, the acquired sounds are separated into audible sounds and inaudible sounds at higher frequencies (Step 2), and frequency analysis is performed on each of these sounds (Step 3).

[0038] In step 4, the target sound, for example, a horn sound, is identified and extracted based on the frequency characteristics of the audible sound components.

[0039] Next, proceeding to step 5, it is determined whether the acquired sound is noise (for example, the horn sound of a vehicle on an adjacent inspection line) or the sound of the correct inspection target based on the frequency characteristics of the inaudible sound component. In one embodiment, the determination result is shown by providing a display comparing the acquired sound with the data in the inaudible sound characteristics storage unit 61 described above, as shown in Figure 3.

[0040] In step 6, an image is generated that displays the frequency characteristics of both audible and inaudible sounds on the screen, as shown in Fig. 3. Then, in step 7, these images are displayed on the display.

[0041] Finally, in step 8, it is determined whether the inspector has pressed the stop button to end the inspection. If the stop button has not been pressed after the inspection result is displayed on the display, the processes of steps 1 to 7 are repeated. For example, if the acquired sound is displayed as including the horn sound of a vehicle on an adjacent inspection line, it is possible to operate the horn switch again to repeat the horn inspection. When the stop button is pressed, the inspection ends. When this inspection is complete, a message to that effect is displayed on the display that serves as the display unit 90. A buzzer or voice may be sounded to notify the end of the inspection along with the display.

[0042] Next, based on Figs. 4 to 6 This is an embodiment of the present invention.An inspection device of the second embodiment will be described. The following mainly describes the differences from the first embodiment. The inspection device of the second embodiment differs from the first embodiment in that, when the acquired sound is determined to be noise based on the inaudible sound components as in the first embodiment, the inspection device of the second embodiment extracts the noise components from each of the audible sound and the inaudible sound, and accumulates the noise components of the inaudible sound and the noise components of the audible sound as linked to each other in a noise database. Such a noise database can be used, for example, to estimate the noise of audible sound based on the noise of inaudible sound.

[0043] 4 shows a functional block diagram of the inspection device of Example 2. As with Example 1, the inspection device of Example 2 includes a sound acquisition unit 10, an audible / inaudible sound separation unit 20, an audible sound frequency characteristics analysis unit 30, an inaudible sound frequency characteristics analysis unit 40, a target sound identification unit 50, a target sound characteristics storage unit 60 (inaudible sound characteristics storage unit 61, audible sound characteristics storage unit 62), a noise determination unit 70, a display image / video generation unit 80, and a display unit 90, as well as a noise sound extraction unit 100 and a noise sound storage unit 110.

[0044] When the noise extraction unit 100 determines that the acquired sound is noise based on the inaudible sound as in the first embodiment, it extracts noise components (noise sounds) for both the inaudible sound and the audible sound.Then, the inaudible noise components and the audible noise components are stored in the noise storage unit 110 in a linked form.

[0045] That is, when the acquired sound is determined to be noise, the inaudible sound of the normal target sound stored in the inaudible sound characteristics storage unit 61 is subtracted from the inaudible sound of the acquired sound to extract the noise component in the inaudible range. To explain this using the display example of FIG. 6, when comparing the inaudible sound component of the original horn sound denoted by reference numeral 90g with the inaudible sound component of the acquired sound denoted by reference numeral 90f, the portion with a higher sound pressure or power than the original inaudible sound (90g) can be considered noise. Therefore, by subtracting the inaudible sound of the original target sound (90g) from the inaudible sound (90f) of the acquired sound that has been subjected to frequency analysis, the frequency characteristics of the noise component in the inaudible range are obtained, as shown by reference numeral 90q. Note that in this example, the portions where the acquired sound has a lower sound pressure or power are ignored.

[0046] Similarly, for audible sounds, as shown in Figure 6, by subtracting the audible sound components of the normal target sound marked with symbol 90r (which are obtained from the audible sound characteristics memory unit 62) from the audible sound components of the acquired sound marked with symbol 90e, which have been frequency analyzed, the frequency characteristics of the noise components in the audible range can be obtained as shown with symbol 90s.

[0047] In this way, by separately determining the noise components in the inaudible range and the noise components in the audible range and storing them in a linked form, it becomes possible to estimate one from the other, as in the third embodiment described below. In addition, it can be used to analyze the trend of audible noise from the trend of inaudible noise.

[0048] FIG. 5 is a flowchart showing the processing flow of the inspection device of the second embodiment described above. As in the first embodiment, first, sounds including, for example, horn sounds are collected from vehicles on the inspection line using the microphone of the sound acquisition unit 10 and acquired as sound data (Step 1). As described above, for example, a sampling frequency of 96 kHz is used to acquire sounds in the range up to 48 kHz. Next, the acquired sounds are separated into audible sounds and inaudible sounds on the higher frequency side (Step 2), and frequency analysis is performed on each of these (Step 3). In Step 4, the target sound, for example, a horn sound, is identified and extracted based on the frequency characteristics of the audible sound components.

[0049] Next, proceed to step 5, where, based on the frequency characteristics of the inaudible sound components, it is determined whether the acquired sound is noise (for example, the horn sound of a vehicle on an adjacent inspection line) or the sound of the correct object to be inspected.

[0050] Next, the process proceeds to step 11, and if step 5 determines that the sound is noise, the noise components are extracted for both the inaudible and audible sounds.Then, the process proceeds to step 12, and the extracted noise components in the inaudible sound range and the noise components in the audible sound range are stored in the noise storage unit 110 in a linked form.

[0051] Next, proceed to step 6, where images to be displayed on the screen are generated. Then, in step 7, these images are displayed on the display. Finally, in step 8, it is determined whether the inspector has pressed the stop button to end the inspection. If the stop button has not been pressed after the inspection results have been displayed on the display, the above-mentioned process is repeated.

[0052] FIG. 6 is an explanatory diagram showing an example of a display on the display unit 90 of the second embodiment. As in the first embodiment, the audible sound range characteristics are displayed in the left area of ​​the screen (indicated by reference numeral 90k), and the inaudible sound range characteristics are displayed in the right area of ​​the screen, but the inaudible sound range area is further divided into two, upper and lower. As in the first embodiment, the upper area 90m displays the inaudible sound frequency characteristics 90f of the acquired sound and the inaudible sound frequency characteristics 90g of the normal target sound in different colors in the form of two-dimensional frequency characteristics with frequency on the horizontal axis and sound pressure / power on the vertical axis. Then, the lower area 90p displays the frequency characteristics of the noise component (see reference numeral 90q) extracted as the difference between the two, as described above.

[0053] In addition, in the audible sound range area 90k, three characteristics are displayed in different colors in the form of two-dimensional frequency characteristics with the horizontal axis representing frequency and the vertical axis representing sound pressure / power: the audible sound frequency characteristic 90e of the acquired sound, the audible sound frequency characteristic 90r of the normal target sound, and the frequency characteristic 90s of the noise component obtained as the difference between the two.

[0054] Next, an inspection device of a third embodiment will be described with reference to Figures 7 to 10. The inspection device of the third embodiment differs from the second embodiment in that it removes noise components in the audible range from the acquired sound by utilizing the fact that noise components in the inaudible range and noise components in the audible range are stored in a linked form in the noise sound storage unit 110, as in the second embodiment. Therefore, the third embodiment has a configuration and function similar to those of the second embodiment.

[0055] FIG. 7 shows a functional block diagram of an inspection device according to a third embodiment. The inspection device according to the third embodiment includes a sound acquisition unit 10, an audible / inaudible sound separation unit 20, an audible sound frequency characteristic analysis unit 30, an inaudible sound frequency characteristic analysis unit 40, a target sound identification unit 50, a target sound characteristic storage unit 60 (inaudible sound characteristic storage unit 61, audible sound characteristic storage unit 62), a noise determination unit 70, a display image / video generation unit 80, a display unit 90, a noise extraction unit 100, and a noise storage unit 110. The device also includes an audible noise estimation unit 120 and a noise removal unit 130. Note that the audible sound characteristic storage unit 62 is not particularly necessary if abnormal sound detection based on the audible sound after noise removal is not performed. In this embodiment, the noise extraction unit 100 only extracts noise in the inaudible range.

[0056] When it is determined that the acquired sound contains noise based on the frequency characteristics of the inaudible sound as in the first and second embodiments described above, the audible noise estimation unit 120 refers to the noise storage unit 110 to find the corresponding audible noise components based on the inaudible noise components extracted by the noise extraction unit 100. That is, the noise storage unit 110, which is a database, stores audible noise components and inaudible noise components for a large number of noises in a linked manner, and because noise sources generate noise in both the audible and inaudible ranges, it is possible to find the audible noise components from the inaudible noise components.

[0057] The noise removal unit 130 removes noise by subtracting the estimated audible noise components from the audible portion of the acquired sound. This allows for obtaining a noise-removed audible sound. The audible sound after noise removal is used, for example, to determine whether an abnormal sound exists in the test object using data from the audible sound characteristics storage unit 62.

[0058] FIG. 8 is a flowchart showing the processing flow of the inspection device of the third embodiment described above. As in the first and second embodiments, first, sounds including, for example, horn sounds are collected from vehicles on the inspection line using the microphone of the sound acquisition unit 10 and acquired as sound data (Step 1). As described above, for example, a sampling frequency of 96 kHz is used to acquire sounds in the range up to 48 kHz. Next, the acquired sounds are separated into audible sounds and inaudible sounds on the higher frequency side (Step 2), and frequency analysis is performed on each of these (Step 3). In Step 4, the target sound, for example, a horn sound, is identified and extracted based on the frequency characteristics of the audible sound components.

[0059] Next, proceed to step 5, where it is determined based on the frequency characteristics of the inaudible sound components whether the acquired sound is noise (for example, the horn sound of a vehicle on an adjacent inspection line) or the sound of the correct inspection target.Then, proceed to step 11, and if it is determined to be noise in step 5, the noise components of the inaudible sound are extracted.

[0060] Next, the process proceeds to step 21, where the audible noise components are estimated based on the inaudible noise components by referring to the noise storage unit 110. Then, the process proceeds to step 22, where the audible noise components are removed from the audible components of the acquired sound.

[0061] Next, proceed to step 6, where images to be displayed on the screen are generated. Then, in step 7, these images are displayed on the display. Finally, in step 8, it is determined whether the inspector has pressed the stop button to end the inspection. If the stop button has not been pressed after the inspection results have been displayed on the display, the above-mentioned process is repeated.

[0062] FIG. 9 is an explanatory diagram showing an example of a display on the display unit 90 of the third embodiment. This is similar to the screen display of the second embodiment shown in FIG. 6, with the characteristics of the audible range shown in the left area of ​​the screen (indicated by reference symbol 90k) and the characteristics of the inaudible range shown in the right area of ​​the screen, with the inaudible range further divided into two, upper and lower. The upper area 90m displays the inaudible frequency characteristics 90f of the acquired sound and the inaudible frequency characteristics 90g of the normal target sound, each displayed in a different color, in the form of a two-dimensional frequency characteristic with frequency on the horizontal axis and sound pressure / power on the vertical axis. The lower area 90p then displays the frequency characteristics of the inaudible noise component extracted as the difference between the two (see reference symbol 90q).

[0063] In addition, in the audible sound range area 90k, three characteristics are displayed in different colors in the form of two-dimensional frequency characteristics with the horizontal axis representing frequency and the vertical axis representing sound pressure / power: the audible sound frequency characteristic 90e of the acquired sound, the frequency characteristic 90s' of the audible sound noise component estimated from the inaudible sound noise component, and the audible sound frequency characteristic 90t after noise removal in which the audible sound noise component estimated from the audible sound component of the acquired sound has been removed.

[0064] 10 is an explanatory diagram showing a different display example on the display unit 90 of the third embodiment. This example shows a case where the acquired sound contains two noises (noise No. 1 and noise No. 2). That is, two noises having different characteristics are contained in the inaudible range, and the two extracted inaudible noise components are displayed in an area 90p at the bottom right of the screen. The audible noise estimation unit 120 described above estimates the audible noise components for each noise, and the noise removal unit 130 removes these audible noise components from the acquired sound.

[0065] In the audible sound range area 90k, as described above, three characteristics are displayed in different colors in the form of two-dimensional frequency characteristics with the horizontal axis representing frequency and the vertical axis representing sound pressure / power: the audible sound frequency characteristic 90e of the acquired sound, the frequency characteristic 90s' of the audible sound noise component estimated from the inaudible sound noise component, and the audible sound frequency characteristic 90t after noise removal in which the audible sound noise component estimated from the audible sound component of the acquired sound has been removed. In this example, the frequency characteristic 90s' of the audible sound noise component is displayed in a form that combines the two noise components.

[0066] Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment and can be applied in various ways. For example, in the inspection of a completed automobile, in addition to the horn, the inspection objects can also include the sounds of the engine, transmission, brakes, and other sound-producing parts, as well as alarm sounds, notification sounds, etc. Furthermore, the present invention can be widely applied to acoustic inspections of things other than automobiles.

[0067] Furthermore, the screen display on the display in the above embodiment is merely an example, and various information may be displayed sequentially by switching the screen, for example. However, the screen display is not essential to the present invention. [Explanation of symbols]

[0068] 10...Sound acquisition section 20...Audible sound / inaudible sound separation section 30...Audible sound frequency characteristics analysis section 40...Inaudible sound frequency characteristics analysis section 50...Target sound identification section 60...Target sound characteristics memory section 61...Inaudible sound characteristics memory section 62...Audible sound characteristics memory section 70...Noise determination unit 80...Display image / video generation unit 90...Display section 100...Noise sound extraction section 110...Noise sound storage unit 120...Audible noise estimation unit 130...Noise removal section

Claims

1. Acquire the sound emitted from the test object, The acquired sound is separated into audible sound and inaudible sound on the higher frequency side, Analyzing the frequency characteristics of the audible sound and the inaudible sound, determining whether the acquired sound contains noise based on the frequency characteristics of the inaudible sound; An acoustic inspection method comprising: When it is determined that noise is included, the noise components are extracted from each of the audible sound and the inaudible sound; The noise components of inaudible sounds and audible sounds are linked together and stored as a noise database. Acoustic testing methods.

2. and displaying at least information regarding the frequency characteristics of the audible sound and information regarding the presence or absence of noise on the display unit.

2. The acoustic inspection method of claim 1.

3. determining whether the acquired sound contains noise by comparing the frequency characteristics of the inaudible sound with the frequency characteristics of the original inaudible sound to be tested; 3. The acoustic inspection method according to claim 1 or 2.

4. before determining whether or not there is noise due to the inaudible sound, determining whether or not the audible sound includes a sound emitted by the test object based on the frequency characteristics of the audible sound; The acoustic inspection method according to any one of claims 1 to 3.

5. When it is determined that noise is included, the noise component of the inaudible sound is extracted, Based on the noise components in the inaudible sound, the noise components in the audible sound are determined by referring to the noise sound database. Remove this noise component from the acquired or audible sound, The acoustic inspection method according to any one of claims 1 to 4.

6. This applies to acoustic testing where a noise source that can emit sounds identical or similar to the sound emitted by the test object may be located at a distance from the test object. The acoustic inspection method according to any one of claims 1 to 5.

7. The test object is the vehicle horn, and it is applied to sound inspection in which the horn is sounded at each of multiple inspection lines.

7. The acoustic inspection method of claim 6.

8. a sound acquisition unit that acquires sounds emitted from the test object; an audible / inaudible sound separation unit that separates the acquired sound into audible sound and inaudible sound on the higher frequency side; a frequency characteristic analysis unit that analyzes the frequency characteristics of the audible sound and the inaudible sound, respectively; a noise determination unit that determines whether the acquired sound contains noise based on the frequency characteristics of the inaudible sound; a noise extraction unit that extracts noise components from each of the audible sound and the inaudible sound when it is determined that the acquired sound contains noise; a noise storage unit that associates inaudible noise components with audible noise components and stores them as a noise database; An acoustic inspection device comprising:

Images