Defect inspection device and defect inspection method

Abstract

To provide a defect inspection device that is able to infer positions of various kinds of defects.SOLUTION: A defect inspection device includes an acquisition unit and an inference unit. The acquisition unit acquires hammering sound data for each of a plurality of inspection positions in an inspection region from a sound collection device. The inference unit performs the following processes (1) to (6). (1) The hammering sound data is analyzed to calculate a feature value indicating a feature of the hammering sound data for each inspection position. (2) The inspection region is divided into a temporary defect region and a temporary normal region that is a complementary set of the temporary defect regions. (3) Using the feature value, pattern recognition is performed by machine learning so as to distinguish the temporary defect region and the temporary normal region. (4) An error in the pattern recognition is calculated for each inspection position. (5) An inversion region is set based on the error, and a temporary defect region and a temporary normal region in the inversion region are replaced with each other. (6) A defect region is inferred by repeating the processes from (3) to (5) until a result of the pattern recognition converges.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present application mainly relates to a defect estimation device that estimates defects in an inspection object based on hammering sound data. [Background technology]

[0002] Patent Document 1 discloses a method for determining the presence or absence of defects based on the impact sound spectrum obtained when FRP (Fiber Reinforced Plastic) is struck with a hammer. Specifically, the value Δ = dB2 - dB1 is calculated. dB1 is the sound frequency that is characteristic of a defect-free FRP panel. Areas of an FRP panel without defects often have a characteristic mountain-shaped spectrum in this frequency band. Furthermore, dB2 is the sound frequency at which a difference appears when a defect is present. Areas of an FRP panel with defects often have a higher level in this frequency band than areas without defects.

[0003] Patent Document 2 discloses a method for determining the presence or absence of defects based on feature quantities of hammering sound data obtained when a structure is struck with a hammer. In this method, feature quantities are extracted by calculating the sound intensity for every 200 Hz based on the hammering sound data. Next, feature quantities are selected using clustering. In this method, a classifier is generated based on the selected feature quantities to classify the hammering sound data to be determined. In Patent Document 2, this classifier is used to classify the hammering sound data to be determined as hammering sound data of a structure without defects or hammering sound data of a structure with defects.

[0004] Non-Patent Document 1 discloses a method for estimating the approximate size and location of a cavity based on a sound pressure waveform obtained when a test piece containing a cavity is struck with a striking tool. In this method, a clustering map is created based on the sound pressure waveform obtained from a test piece whose cavity size and location are known in advance and the frequency characteristics obtained by applying a fast Fourier transform. Next, the approximate size and location of the cavity are estimated by determining which group in the clustering map the hammering sound waveform of the test piece to be evaluated most closely matches. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Patent No. 4736501 [Patent Document 2] Japanese Patent Application Laid-Open No. 2018-13348 [Non-patent literature]

[0006] [Non-Patent Document 1] Eiki Matsuoka, Yoichi Hirose, Takahiko Kurahashi, Yuki Murakami, Shigehiro Toyama, Fujio Ikeda, Tetsuro Iyama, and Ikuo Ihara, "Application of the Adjoint Variable Method to Improve the Accuracy of Quantitative Evaluation of Defects by Hammering Inspection," Journal of the Society of Materials Science, Japan, Vol. 67, No. 9, September 2018, pp. 860-876 Summary of the Invention [Problem to be solved by the invention]

[0007] The method of Patent Document 1 determines defects based on the Δ value, but the Δ value itself has no physical meaning and is not a versatile indicator of defects. Therefore, depending on the type of defect, the Δ value of a defective portion overlaps with the Δ value of a non-defective portion. Therefore, the method of Patent Document 1 cannot estimate the locations of various types of defects. The method of Patent Document 2 estimates the location of a defect based on a classifier generated through prior training. Therefore, the method of Patent Document 2 cannot recognize untrained defects and therefore cannot estimate the locations of various types of defects. Non-Patent Document 1 discloses a method for estimating the location of void-related defects, but does not disclose a method for estimating other types of defects.

[0008] The present application has been made in view of the above circumstances, and its main object is to provide a defect inspection apparatus capable of estimating the positions of various types of defects. [Means for solving the problem]

[0009] The problem to be solved by the present application is as described above. Next, the means for solving this problem and the effects thereof will be explained.

[0010] According to a first aspect of the present application, there is provided a defect inspection device having the following configuration. That is, the defect inspection device includes an acquisition unit and an estimation unit. The acquisition unit acquires hammering sound data for each of a plurality of inspection positions in an inspection area from a sound collection device. The estimation unit estimates a defect area included in the inspection area based on the hammering sound data acquired by the acquisition unit. The aforementioned The estimation unit performs the following processes (1) to (6): (1) analyzes the hammering sound data and calculates feature values ​​that indicate the features of the hammering sound data for each inspection position; (2) divides the inspection area into tentative defect areas and tentative normal areas that are complements of the tentative defect areas; (3) performs pattern recognition by machine learning to distinguish between the tentative defect areas and the tentative normal areas using the feature values; (4) calculates the error of the pattern recognition for each inspection position; (5) sets an inverted area based on the error, and swaps the tentative defect area and the tentative normal area in the inverted area; and (6) performs a detailed inspection to estimate the defect area by repeating the processes (3) to (5) until the pattern recognition results converge.

[0011] According to a second aspect of the present application, there is provided a defect inspection device having the following configuration. That is, the defect inspection device includes an acquisition unit and an estimation unit. The acquisition unit acquires hammering sound data for each of a plurality of inspection positions in an inspection area from a sound collection device. The estimation unit estimates a defect area included in the inspection area based on the hammering sound data acquired by the acquisition unit. The aforementioned The estimation unit performs the following processes (1) to (3): (1) sets a provisional inspection area and a plurality of comparison areas set based on the provisional inspection area within the inspection area; (2) calculates feature values ​​indicating the features of the hitting sound data for the provisional inspection area and the comparison areas; and (3) calculates a feature value based on the difference between the feature value of the provisional inspection area and the feature value for each comparison area. The aforementioned The defect area is estimated.

[0012] According to a third aspect of the present application, there is provided the following defect inspection method. Specifically, hammering data for each of a plurality of inspection positions in an inspection area is acquired from a sound collection device. (1) The hammering data is analyzed to calculate a feature value indicating the feature of the hammering data for each of the inspection positions. (2) The inspection area is divided into a tentative defect area and a tentative normal area that is a complement of the tentative defect area. (3) Pattern recognition is performed by machine learning to distinguish between the tentative defect area and the tentative normal area using the feature value. (4) An error in the pattern recognition is calculated for each of the inspection positions. (5) An inverted area is set based on the error, and the tentative defect area and the tentative normal area in the inverted area are swapped. (6) A detailed inspection is performed to estimate a defect area by repeating the processes from (3) to (5) until the pattern recognition result converges. [Effects of the Invention]

[0013] According to the present application, it is possible to realize a defect inspection device that can estimate the positions of various types of defects. [Brief explanation of the drawings]

[0014] [Figure 1] FIG. 1 is a conceptual diagram showing how a panel is inspected using a defect inspection device. [Figure 2] FIG. 10 is a diagram showing a provisional inspection area and a comparison area of ​​a panel in an outline inspection. [Figure 3] Flowchart of the overview inspection. [Figure 4] FIG. 10 is a diagram showing an inspection position, a provisional defective area, and a provisional normal area of ​​a panel in a detailed inspection. [Figure 5] Flowchart of detailed inspection. [Figure 6] A conceptual diagram showing the flow of detailed inspection. [Figure 7] 10A and 10B are diagrams showing the results of an experiment using a defect inspection device. [Figure 8] FIG. 10 is a conceptual diagram showing how a panel is inspected using a defect inspection device according to another embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0015] Next, an embodiment of the present application will be described with reference to the drawings. Fig. 1 is a conceptual diagram showing how a panel 21 is inspected using a defect inspection apparatus 10.

[0016] The defect inspection device 10 is a device that estimates defects in a panel 21 based on hammering sound data obtained when the panel 21 is hammered. The panel 21 is an object to be inspected by the defect inspection device 10. The panel 21 of this embodiment is a multilayer panel having a structure in which both sides of a core material are sandwiched between skin materials and bonded together. The core material is a member that includes a cavity, such as a honeycomb core. The skin material is a plate-like member that is thinner than the core material. The defect inspection device 10 can inspect objects of any shape and material that can be subjected to hammering sound inspection. Therefore, the object to be inspected by the defect inspection device 10 may be a solid member without a cavity, or may be an installed structure.

[0017] Defects contained in the panel 21 include, for example, a core recess where the core material is recessed compared to the surrounding area, an adhesive-free area where the core material and the skin material are not bonded, a backsheet-applied area where a backsheet is present as a foreign object between the core material and the skin material, etc. However, the defects detected by the defect inspection device 10 are not limited to the above defects.

[0018] In this embodiment, an inspector performs the tapping test using a hammer 22. Specifically, the inspector taps each position on the panel 21 using the hammer 22 according to a predetermined procedure. Note that the tapping test may be performed using a tapping rod or the like instead of the hammer 22. Furthermore, as will be described later, the panel 21 may be automatically tapped using a tapping device.

[0019] As shown in FIG. 1, the defect inspection device 10 includes a sound collection device 11 and a computer 12.

[0020] The sound collection device 11 is a microphone that receives the hitting sound generated when the panel 21 is hit and converts it into an electrical signal. Hereinafter, this electrical signal will be referred to as hitting sound data. The hitting sound data indicates a change in amplitude over time, i.e., a hitting sound waveform. The sound collection device 11 may be of a diaphragm type or a piezoelectric type. The sound collection device 11 may be configured to have directivity in a specific direction (for example, the direction in which the panel 21 is located) or may be omnidirectional. The sound collection device 11 outputs the hitting sound data to the computer 12 via wired or wireless communication.

[0021] The computer 12 includes an arithmetic unit such as a CPU, a storage device such as an HDD, an SSD, or a flash memory, and a communication device such as a wired module or a wireless module. The arithmetic unit executes programs stored in the storage device, allowing the computer 12 to perform various processes related to the defect inspection apparatus 10. The above-described hardware and software work together to allow the computer 12 to operate as an acquisition unit 12a and an estimation unit 12b.

[0022] The acquiring unit 12a acquires the hammering sound data from the sound collection device 11. The acquiring unit 12a receives the hammering sound data using a connector or an antenna of the computer 12, and performs amplification, signal conversion, and the like.

[0023] Furthermore, the acquisition unit 12a stores the hammering sound data in a storage device in association with the positions where the hammer 22 has been struck. There are various methods for associating the hammering sound data with positions. For example, the positions and order in which the hammer 22 will strike the panel 21 can be determined in advance, thereby associating the hammering sound data with positions. Alternatively, the inspector can strike the panel 21 with the hammer 22 in response to audio or visual instructions from the computer 12, thereby associating the hammering sound data with positions. Alternatively, the inspector can register the striking positions by operating an operating tool such as a keyboard or a mouse, and then actually strike the panel 21 with the hammer 22, thereby associating the hammering sound data with positions.

[0024] The estimation unit 12b estimates a defective area of ​​the panel 21 based on the hammering sound data acquired by the acquisition unit 12a. The estimation of the defective area by the estimation unit 12b may be performed in parallel with the process of the inspector hitting the panel 21 with the hammer 22. Alternatively, the estimation of the defective area by the estimation unit 12b may be performed after the inspector has completed the process of hitting the panel 21 with the hammer 22. Details of the process performed by the estimation unit 12b will be described later.

[0025] Hammering inspections are typically performed by skilled inspectors based on the tone of the hammering sounds. However, this method results in significant variation in inspection accuracy among inspectors. Furthermore, training inspectors requires time and effort, and can place a significant physical burden on the inspectors. Furthermore, the methods described in prior art documents rely on identifying or learning specific features, making it difficult to accurately estimate the locations of various types of defects. In this regard, the defect inspection device 10 of this embodiment quantifies the sound collection conditions, thereby reducing variation in inspection accuracy among inspectors, and by focusing on errors in features that are not dependent on specific features, it is possible to more accurately estimate the locations of various types of defects. The defect inspection method of this embodiment will now be described in detail.

[0026] Inspections performed using the defect inspection apparatus 10 include a general inspection, which estimates the approximate position of a defect, and a detailed inspection, which estimates the shape of the defect. In this embodiment, the general inspection is performed, and then a detailed inspection is performed using the results of the general inspection. However, the general inspection can be omitted. If the general inspection is omitted, both the position and shape of the defect are estimated in the detailed inspection.

[0027] The overview inspection will be described below with reference to Figures 2 and 3. Figure 2 is a diagram showing a provisional inspection area and a comparison area of ​​the panel 21 in the overview inspection. Figure 3 is a flowchart of the overview inspection.

[0028] As described above, the panel 21 contains various types of defects. Furthermore, the sound produced when a defective area is struck differs depending on the type of defect. For this reason, it is difficult to learn all of the strike sounds in advance. On the other hand, the strike sound produced when a defective area is struck differs from the strike sound produced when a normal area around the defect is struck. This point is taken into consideration in the overview inspection, which identifies areas where the strike sound differs from other areas and targets these areas for detailed inspection. The specific processing is described below.

[0029] As shown in FIG. 2, in the general inspection, the surface of the panel 21 is divided into a plurality of square regions. The length of one side of each region is length L1. In the general inspection, the inspector hits one point of each region with the hammer 22 once or multiple times. It is preferable that the position at which the inspector hits with the hammer 22 is the center of each region, as shown by the midpoint in FIG. 2. The sound collection device 11 converts the hitting sounds generated by hitting with the hammer 22 into hitting sound data and outputs the data to the computer 12. The sampling frequency when collecting the hitting sounds and converting them into hitting sound data is 44.1 kHz.

[0030] As shown in Fig. 3, the acquisition unit 12a of the computer 12 acquires hammering sound data for each position from the sound collection device 11 (S101). Next, the estimation unit 12b of the computer 12 sets an initial provisional inspection area (S102). The provisional inspection area is an area for which the presence or absence of a defect is estimated using a rough inspection. In Fig. 2, the provisional inspection area is indicated by diagonal hatching. In the rough inspection, all areas are inspected in order.

[0031] Next, the estimation unit 12b sets multiple comparison areas based on the provisional inspection area (S103). The comparison areas are areas for comparing the provisional inspection area with the hammering sound data. In FIG. 2, the comparison areas are indicated by halftone hatching. Here, areas adjacent to the provisional inspection area are likely to have the same defects as those in the provisional inspection area. Furthermore, areas far from the provisional inspection area may have different hammering sound data even if they are normal areas. Therefore, in this embodiment, areas two spaces away from the provisional inspection area are used as comparison areas. However, any area different from the provisional inspection area may be selected as the comparison area. For example, depending on the characteristics of the panel 21, the length L1, etc., it may be preferable to select an adjacent area or an area three or more spaces away.

[0032] Next, the estimation unit 12b calculates the difference between the feature value of the provisional inspection area and the feature value of each comparison area (S104). The feature value is a value indicating the feature of the hammering sound data. In this embodiment, the AR coefficient of the hammering sound data is used as the feature value. As described above, the hammering sound data indicates a hammering sound waveform. Furthermore, if the same position on the panel 21 is struck multiple times with the hammer 22, correlation analysis may be performed on the multiple hammering sound waveforms to extract multiple hammering sound waveforms having a correlation coefficient equal to or greater than a certain value, and the extracted waveforms may be averaged. An Nth-order AR coefficient is estimated for this hammering sound waveform. The AR coefficient can be estimated using a known method such as the Yule-Walker method, the least squares method, the maximum likelihood method, or the Burg method. The AR coefficient estimated for the hammering sound waveform at the coordinate position (i, j) is expressed by the following equation (1).

number

[0033] In this embodiment, the difference in feature values ​​is calculated by subtracting the AR coefficient of the comparison region from the AR coefficient of the provisional inspection region and squaring the result. If the coordinates of the provisional inspection region are (i1, j1) and the coordinates of a certain comparison region are (i2, j2), the difference d AR is expressed by the following formula (2). The difference in the AR coefficient, which is a feature value, can be considered as a value that indicates the difference in the timbre of the striking sound.

number

[0034] Next, the estimation unit 12b calculates a sum P of the magnitude of the difference between the feature value of the provisional inspection region and the feature value of each comparison region. AR Based on the sum P, it is determined whether or not the set provisional inspection area is to be subjected to detailed inspection (S105). AR is expressed by the following equation (3) based on the above equation (2).

number

[0035] In the general inspection, a provisional inspection area to be inspected for detailed inspection is extracted. For example, the sum of the magnitudes of the differences P AR The larger the difference, the higher the probability of it being a temporary defect area. AR Therefore, the smaller the sum of the magnitudes of the differences P AR Based on this, the estimation unit 12b estimates whether the corresponding tentative inspection area is a tentative defect area or a tentative normal area. Furthermore, if the estimation unit 12b estimates the tentative inspection area as a tentative defect area, it targets this tentative inspection area for detailed inspection. Furthermore, if the estimation unit 12b estimates the tentative inspection area as a tentative normal area, it excludes this tentative inspection area from the target of detailed inspection. However, the estimation unit 12b may also determine that tentative normal areas near the tentative defect area are targets for detailed inspection. A threshold may be used when determining whether or not to target a tentative inspection area. If a threshold is used, the threshold may be appropriately determined based on experiments and experience. Furthermore, for example, for 16 locations in the comparison area in Figure 2, the differences between the feature values ​​of the tentative inspection area and the feature values ​​of the comparison area may be calculated individually, and whether or not to target a region for detailed inspection may be determined based on the number of locations where the difference exceeds the threshold.

[0036] The feature values ​​are not limited to AR coefficients, but can be calculated using different methods. Calculate withFor example, a power spectrum calculated by Fourier transforming the hammering waveform may be used. The power spectrum is a value indicating the intensity of each frequency of the hammering waveform. Specifically, the power spectrum is divided into a plurality of frequency bands, and the energy value of each frequency band is quantified and used as the feature value.

[0037] Next, the estimation unit 12b determines whether inspection of all provisional inspection areas has been completed (S106). If there are any provisional inspection areas that have not been inspected, the estimation unit 12b sets the next provisional inspection area (S107) and performs the processes from step S103 onwards again. The next provisional inspection area may be set by an inspector or the like, rather than by the estimation unit 12b. If inspection of all provisional inspection areas has been completed, the estimation unit 12b ends the general inspection.

[0038] By performing the above-mentioned overview inspection, it is possible to estimate the area containing the defect regardless of the type of defect. Furthermore, since the estimation can be performed objectively without relying on the subjectivity or experience of the inspector, it is possible to estimate the area containing the defect regardless of the inspector's level of proficiency.

[0039] Next, detailed inspection will be described with reference to Figs. 4 to 6. Fig. 4 is a diagram showing the inspection position, provisional defect area, and provisional normal area of ​​a panel in detailed inspection. In Fig. 4, the inspection position of the panel is indicated by a midpoint, provisional defect area by diagonal lattice hatching, and provisional normal area by oblique hatching. Fig. 5 is a flowchart of detailed inspection. Fig. 6 is a conceptual diagram showing the flow of detailed inspection.

[0040] In this embodiment, the detailed inspection is performed after the general inspection. The detailed inspection is an inspection to estimate the shape of the defect in the tentative defect area estimated in the general inspection. Therefore, in the detailed inspection shown in FIG. Panel 21 The length L2, which indicates the interval between the tapping positions, is Panel 21 This is often shorter than the length L1, which indicates the distance between the positions where the keys are struck.

[0041] First, the estimation unit 12b sets an inspection area (S201). The inspection area is a unit area to be inspected in the detailed inspection. The inspection area is set based on the temporary defect area estimated in the general inspection. In this embodiment, the inspection area is set to an area obtained by multiplying one side of the temporary defect area by three. That is, if the length of one side of the temporary defect area is denoted by L3 and the length of one side of the inspection area is denoted by L4, then 3×L3=L4 holds. However, there are various methods for setting the inspection area, and the inspection area may be set to an area obtained by multiplying one side of the temporary defect area by a predetermined factor, for example, between two and four times, or may be set to an area of ​​a predetermined size and shape.

[0042] Next, the computer 12 instructs the inspector to tap the inspection area at intervals of length L2. Hereinafter, the position in the inspection area where the tapping is performed for detailed inspection will be referred to as the inspection position. The acquisition unit 12a of the computer 12 acquires tapping sound data for each inspection position from the sound collection device 11 (S202).

[0043] Next, the estimation unit 12b of the computer 12 calculates a feature value for each inspection position (S203). In this embodiment, the AR coefficient is used as the feature value in the detailed inspection as well as in the general inspection. The AR coefficient is estimated in the same way as in the general inspection. Also, as in the general inspection, a power spectrum can be used as the feature value. Note that, although there may be large differences between AR coefficients of normal regions at distant positions across the entire panel 21, it can be assumed that the AR coefficients of normal regions within a small inspection region are almost the same.

[0044] Next, the estimation unit 12b sets a tentative defect area and a tentative normal area based on the results of the overview inspection (S204). In this embodiment, the tentative defect area and the tentative normal area are set based on the results of the overview inspection. The inspection area is classified as either a tentative defect area or a tentative normal area. In other words, the tentative normal area is the complement of the tentative defect area. Note that if the overview inspection is omitted, any area may be set as a tentative defect area. This is because, as will be described later, even if the tentative defect area is set incorrectly, it can be corrected.

[0045] Next, the estimation unit 12b performs pattern recognition by machine learning so as to distinguish between the provisional defect area and the provisional normal area using the feature values ​​(S205). In this embodiment, pattern recognition is performed by the following method. Here, the provisional defect area is denoted by A, and the provisional normal area is denoted by B. A and B each include a plurality of inspection positions. The AR coefficients, which are the feature quantities of these inspection positions, are linearly separated in a linear space. That is, using the separation function shown in the following equation (4), the provisional defect area is set to 1 and the provisional normal area is set to 0, as shown in equation (5). Then, the weight w of the separation function is set so as to minimize the square error I shown in equation (6). k Let k=1,...,N.

number

[0046] The method for distinguishing between the provisional defect area and the provisional normal area by pattern recognition using the feature values ​​is not limited to the above-described method. For example, pattern recognition may be performed using a method such as nonlinear separation or a support vector machine.

[0047] Next, the estimation unit 12b calculates the error of the pattern recognition for each inspection position (S206). This error is an index for evaluating the accuracy of the pattern recognition, and specifically, is the square error I described above.

[0048] Next, the estimation unit 12b determines whether the pattern recognition result has converged (S207). The method for determining whether the pattern recognition result has converged is arbitrary, and various methods can be used. For example, the estimation unit 12b determines whether the pattern recognition result has converged based on the smallness of the pattern recognition error in the inspection area, for example, the value of the squared error I. In detail, a threshold value is set for the pattern recognition error, and the pattern recognition result is determined to have converged when there are no more areas where the error exceeds the threshold value. However, the pattern recognition result may also be determined to have converged even when there are still a few areas where the error exceeds the threshold value.

[0049] When the estimation unit 12b determines that the pattern recognition result has not converged, it sets an inverted region based on the pattern recognition error and swaps the tentative defect region and tentative normal region of the inverted region (S208). The inverted region can be set, for example, using a threshold value. That is, if a region where the pattern recognition error exceeds the threshold is a tentative defect region, it is changed to a tentative normal region, and if a region where the pattern recognition error exceeds the threshold is a tentative normal region, it is changed to a tentative defect region. In the example shown in FIG. 6, the region where the pattern recognition error exceeds the threshold is the inverted region. Furthermore, the inverted region can be set to a region limited to the boundary between the tentative defect region and the tentative normal region (assuming that the defect portion is one region) among regions with large pattern recognition errors, or to a region with a shape similar to a specific pattern (such as a defect shape that frequently appears).

[0050] Thereafter, the estimation unit 12b performs the processes of steps S205 to S207 again. That is, after swapping the tentative defect area and the tentative normal area, it again performs pattern recognition by machine learning and calculates the difference in error. The estimation unit 12b repeats these processes until the pattern recognition result converges. Then, when it determines that the pattern recognition result has converged, the estimation unit 12b estimates, as a defect area, an area where the separating function shown in equation (4), in other words, the value calculated according to the feature value, exceeds a threshold (S209). Note that instead of this process, the tentative defect area at the time when the pattern recognition result has converged may be estimated as a defect area.

[0051] FIG. 7 shows the results of an experiment in which defects were estimated by performing general and detailed inspections using the defect inspection apparatus 10 of this embodiment. In this experiment, defects of 10 mm, 20 mm, and 30 mm in size were created in multiple panels and then estimated. The types of defects created were the aforementioned non-adhesive and backsheet-applied areas. Furthermore, the skin material located on the upper side of the panels used in the experiment is referred to as the upper skin, and the skin material located on the lower side is referred to as the lower skin. The surface of the upper skin includes irregularities. The surface of the lower skin is flat. Three photographs are arranged horizontally in FIG. 7. The sizes of the created defects differ in each photograph (10 mm, 20 mm, and 30 mm from the left). The panels, surfaces, and defects used in the inspection are described to the left of the leftmost photograph. The rectangles in the center of each figure in FIG. 7 indicate the range of the actual defects. The areas surrounded by curved lines in each figure in FIG. 7 are the defect areas estimated by the estimation unit 12b. Defects that were not detected by the inspector without the defect inspection device 10 are marked with an X in the upper right corner of the photograph. As shown in FIG. 7, when inspector A performed an inspection without the defect inspection device 10, he was unable to detect a defect with a defect size of 10 mm on panel 1. However, by using the defect inspection device 10, he was able to successfully detect this defect. Furthermore, when inspector B performed an inspection without the defect inspection device 10, he was unable to detect defects with defect sizes of 10 mm, 20 mm, and 30 mm on panel 3. However, by using the defect inspection device 10, he was able to successfully detect these defects. As shown in FIG. 7, by using the defect inspection device 10 of this embodiment, defects can be accurately estimated regardless of individual differences between inspectors or defect sizes.

[0052] Next, another embodiment will be described with reference to Fig. 8. In the above-described defect inspection device 10, an inspector hits the panel 21 with a hammer 22. Instead, a defect inspection device 100 of another embodiment automatically hits the panel 21 using a striking device 32.

[0053] The impact device 32 has, for example, a hammer at its tip, and can strike the panel 21 by moving the hammer back and forth relative to the panel 21. The impact device 32 is also configured to be movable, for example, along one side of the panel 21. The panel 21 is placed on a conveyor 31 whose transport amount is adjustable. Therefore, any position on the surface of the panel 21 can be struck by the conveyor 31 and the impact device 32. With this configuration, the impact device 32 can strike a position that is predetermined or determined by the computer 12. Providing the impact device 32 in the defect inspection device 100 can reduce the physical burden on the inspector.

[0054] The impact device 32 may be movable in two directions that are perpendicular to each other in a plan view. In this case, the conveyor 31 can be omitted. The defect inspection device 100 may also be integrally provided with the sound collection device 11, the computer 12, and the impact device 32.

[0055] The sound pickup device 11 and the computer 12 of the defect inspection device 100 are the same as those in the above-described embodiment, and therefore a description thereof will be omitted.

[0056] The defect inspection apparatus 10, 100 of this embodiment includes an acquisition unit 12a and an estimation unit 12b. The acquisition unit 12a acquires hammering sound data for each of a plurality of inspection positions in the inspection area from the sound collection device 11. The estimation unit 12b estimates a defect area included in the inspection area based on the hammering sound data acquired by the acquisition unit 12a. (1) Analyzes the hammering sound data and calculates a feature value indicating the feature of the hammering sound data for each inspection position. (2) Divides the inspection area into a provisional defect area and a provisional normal area, which is a complement of the provisional defect area. (3) Performs pattern recognition by machine learning to distinguish between the provisional defect area and the provisional normal area using the feature value. (4) Calculates the pattern recognition error for each inspection position. (5) Sets an inverted area based on the error, and swaps the provisional defect area and provisional normal area in the inverted area. (6) Performs a detailed inspection to estimate a defect area by repeating the processes (3) to (5) until the pattern recognition results converge.

[0057] This allows the positions of various types of defects to be estimated because defective areas are assumed based on comparison with the surrounding area, rather than on a judgment model created in advance based on training data.

[0058] In the defect inspection apparatuses 10 and 100 of this embodiment, the estimation unit 12b determines that the pattern recognition results have converged when there is no longer any region where the error exceeds the threshold.

[0059] This allows the defective area to be estimated with high accuracy.

[0060] In the defect inspection devices 10, 100 of this embodiment, the acquisition unit 12a acquires hammering sound data of a provisional inspection area and hammering sound data of multiple comparison areas located at positions different from the provisional inspection area from the sound collection device 11. The hammering sound data is analyzed to calculate a feature value of the hammering sound data of the provisional inspection area and a feature value of the hammering sound data for each comparison area. Based on the total value of the magnitude of the difference between the feature value of the provisional inspection area and the feature value for each comparison area, a general inspection is performed to classify the provisional inspection area into a provisional defect area and a provisional normal area.

[0061] Since the feature amounts of the hammering sound data at different positions are compared, the provisional defect area can be set accurately.

[0062] In the defect inspection apparatuses 10 and 100 of the present embodiment, the intervals between positions at which the inspection object is struck in detailed inspection are shorter than the intervals between positions at which the inspection object is struck in general inspection.

[0063] A general inspection can be performed, and a detailed inspection can be performed in detail.

[0064] The defect inspection devices 10 and 100 of this embodiment are provided with an impact device 32 that automatically strikes a plurality of inspection positions.

[0065] This reduces the workload of the inspector.

[0066] The preferred embodiment of the present application has been described above, but the above configuration can be modified, for example, as follows.

[0067] In the above embodiment, a method of performing both a general inspection and a detailed inspection, and a method of performing only a detailed inspection have been described. Alternatively, only a general inspection may be performed and the detailed inspection may be omitted. When the detailed inspection is omitted, an area estimated as a tentative defective area in the general inspection of the above embodiment is treated as a defective area.

[0068] The flowcharts shown in the above embodiments are merely examples, and some processes may be omitted, the contents of some processes may be changed, or new processes may be added.

[0069] The functions of each element, including the estimator 12b, disclosed herein can be performed using circuits or processing circuits, including general-purpose processors, dedicated processors, integrated circuits, ASICs (Application Specific Integrated Circuits), conventional circuits, and / or combinations thereof, configured or programmed to perform the disclosed functions. A processor is considered a processing circuit or circuit because it includes transistors and other circuits. In this disclosure, a circuit, unit, or means is hardware that performs the enumerated functions or hardware programmed to perform the enumerated functions. The hardware may be hardware disclosed herein or other known hardware that is programmed or configured to perform the enumerated functions. When the hardware is a processor, which is considered a type of circuit, the circuit, means, or unit is a combination of hardware and software, and software is used to configure the hardware and / or processor. [Explanation of symbols]

[0070] 10,100 defect inspection equipment 11 Sound pickup device 12. Computers 12a Acquisition part 12b Estimation part Panel 21 22 Hammer 31 Conveyor 32 Percussion Device

Claims

1. an acquisition unit that acquires hammering sound data for each of a plurality of test positions in the test area from a sound collection device; an estimation unit that estimates a defective area included in the inspection area based on the hammering sound data acquired by the acquisition unit; Equipped with The estimation unit (1) analyzing the hammering sound data to calculate a feature value indicating a feature of the hammering sound data for each of the inspection positions; (2) Dividing the inspection area into a provisional defect area and a provisional normal area which is a complement of the provisional defect area; (3) performing pattern recognition by machine learning to distinguish between the provisional defect area and the provisional normal area using the feature value; (4) calculating an error in the pattern recognition for each of the inspection positions; (5) setting the area where the error exceeds the threshold as an inverted area, and swapping the temporary defective area and the temporary normal area in the inverted area; (6) A defect inspection apparatus characterized in that detailed inspection is performed to estimate the defect area by repeating the processes (3) to (5) until the result of the pattern recognition converges.

2. 2. The defect inspection device according to claim 1, The defect inspection device is characterized in that the estimation unit determines that the result of the pattern recognition has converged when there is no longer any region in which the error exceeds a threshold value.

3. 3. The defect inspection device according to claim 1, the acquiring unit acquires, from the sound collecting device, hammering sound data of a provisional inspection area and hammering sound data of a plurality of comparison areas located at positions different from the provisional inspection area; The estimation unit analyzing the hammering sound data to calculate the feature value of the hammering sound data of the provisional inspection area and the feature value of the hammering sound data of each comparison area; A defect inspection device characterized by performing a general inspection to divide the provisional inspection area into the provisional defect area and the provisional normal area based on the difference between the feature value of the provisional inspection area and the feature value of each comparison area.

4. 4. The defect inspection device according to claim 3, A defect inspection device characterized in that the intervals between positions at which the inspection object is struck during the detailed inspection are shorter than the intervals between positions at which the inspection object is struck during the general inspection.

5. 5. The defect inspection device according to claim 1, a defect inspection device comprising an impact device for automatically striking each of the plurality of inspection positions;

6. acquiring hammering sound data for each of a plurality of test positions in the test area from a sound collection device; (1) analyzing the hammering sound data to calculate a feature value indicating a feature of the hammering sound data for each of the inspection positions; (2) Dividing the inspection area into a provisional defect area and a provisional normal area which is a complement of the provisional defect area; (3) performing pattern recognition by machine learning to distinguish between the provisional defect area and the provisional normal area using the feature value; (4) calculating an error in the pattern recognition for each of the inspection positions; (5) setting the area where the error exceeds the threshold as an inverted area, and swapping the temporary defective area and the temporary normal area in the inverted area; (6) A defect inspection method characterized by performing detailed inspection to estimate a defect area by repeating the processes (3) to (5) until the result of the pattern recognition converges.

Images