Product inspection apparatus and product inspection method

Abstract

To accurately detect a sudden sound.SOLUTION: Disclosed is a product inspection device including a data acquisition unit that acquires time-series data related to sound or vibration during product operation, a first calculation unit that calculates an effective value for each predetermined time from a data portion for each predetermined time in the time-series data or from frequency characteristic data for each predetermined time based on the time-series data, a second calculation unit that calculates a difference value between effective values for each predetermined time adjacent in time series, and an abnormality detection unit that detects an abnormality during product operation based on the difference value.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present disclosure relates to a product inspection device and a product inspection method. [Background technology]

[0002] A known method for inspecting machine operating noise detects the operating noise of a machine as a vibration signal and determines whether the operating noise is abnormal based on the vibration signal. This method divides data that has been frequency analyzed using a fast Fourier transform (FFT) into multiple frequency bands, calculates the intensity level and frequency (number of occurrences) of the operating noise for each frequency band, accumulates the number of occurrences exceeding a predetermined intensity level, and compares the accumulated value with a threshold value to determine whether the sound pressure level of the operating noise is abnormal. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 11-223550 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the above-described conventional method, i.e., the method of evaluating the intensity level of the operating sound for each frequency band, does not take into account the time element, and it is difficult to accurately detect a sudden sound that occurs for only a relatively short period of time (a sudden sound that should be detected as an abnormal sound).

[0005] Therefore, in one aspect, an object of the present disclosure is to enable detection of a sudden sound with high accuracy. [Means for solving the problem]

[0006] In one aspect, a data acquisition unit that acquires time-series data related to sound or vibration during operation of the product; a first calculation unit that calculates an effective value for each predetermined time from a data portion of the time series data for each predetermined time or from frequency characteristic data for each predetermined time based on the time series data; a second calculation unit that calculates a difference value between the effective values ​​for each of the predetermined times that are adjacent in time series; and an abnormality detection unit that detects an abnormality during product operation based on the difference value. [Effects of the Invention]

[0007] According to one aspect of the present disclosure, it is possible to detect a sudden sound with high accuracy. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic diagram showing an entire product inspection system. [Figure 2] FIG. 2 illustrates an example of a hardware configuration of a processing device. [Figure 3] FIG. 2 is a diagram illustrating the functionality of a processing device. [Figure 4] FIG. 1 is an explanatory diagram of a frame obtained by short-time Fourier transform. [Figure 5] FIG. 10 is an explanatory diagram of an RMS change curve. [Figure 6] FIG. 10 is an explanatory diagram of a threshold setting method. [Figure 7] 1 is a schematic flowchart showing the flow of main processes executed by the processing apparatus of the present embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Each embodiment will be described in detail below with reference to the accompanying drawings.

[0010] FIG. 1 is a schematic diagram showing the entire product inspection system 1. As shown in FIG.

[0011] The product inspection system 1 inspects the quality of a product (presence or absence of abnormalities, etc.). The product to be inspected may be any product, but in the following, an in-vehicle electrification device will be used as an example. The in-vehicle electrification device may be an electric sunroof with an electrically operated glass roof that opens and closes, various electrically operated doors (side doors, back doors, etc.), an electrically operated roof that retracts and deploys (such as the roof of a convertible), an electrically operated mirror that retracts and deploys, or an electrically operated shade (such as a shade on the rear window). The in-vehicle electrification device does not have to be electrically operated, and may be a product driven by other power systems, such as hydraulic or electromagnetic. In either case, the product to be inspected may be inspected individually, but in the following, it is assumed that the product is inspected in a state where it is mounted on a vehicle. In this case, the product inspection system 1 may be installed on a vehicle production line and may inspect the quality of the product during or after vehicle production is completed.

[0012] The product inspection system 1 includes an inspection data acquisition device 2 and an abnormal noise inspection device 3.

[0013] The inspection data acquisition device 2 may be in the form of a soundproof room that can accommodate a vehicle, and includes various microphones 21 to 24, such as acoustic microphones. Although the microphones 21 to 24 are used, the type, number, and placement of the microphones are optional. Furthermore, if the product to be inspected is placed inside the vehicle cabin, the microphones 21 to 24 may be temporarily placed inside the vehicle cabin.

[0014] The microphones 21 to 24 pick up sounds when the target product is operating, for example, when the sunroof is opened or closed in the case of an electric sunroof. The sound data picked up by the microphones 21 to 24 is input to the abnormal sound inspection device 3.

[0015] The abnormal sound inspection device 3 includes a sound board 30 , an A / D (Analog-to-Digital) converter 32 , and a processing device 100 .

[0016] Connected to the sound board 30 is a microphone 21, for example in the form of an acoustic microphone.

[0017] The other microphones 22 to 24 are connected to the A / D converter 32. The A / D converter 32 converts the analog signals supplied from the microphones 22 to 24 into digital signals.

[0018] A time-series signal (time-series data) of sound obtained via the sound board 30 and the A / D converter 32 is input to the processing device 100. The function of the processing device 100 will be described later with reference to FIG.

[0019] FIG. 2 is a diagram illustrating an example of a hardware configuration of the processing device 100. As shown in FIG.

[0020] 2, the processing device 100 includes a control unit 101, a main memory unit 102, an auxiliary memory unit 103, a drive device 104, a network I / F unit 106, and an input unit 107. Note that part or all of the processing device 100 may be in the form of a circuit in which various elements (including elements in the form of chips) are mounted on a substrate or the like.

[0021] The control unit 101 is a calculation device that executes programs stored in the main memory unit 102 and the auxiliary memory unit 103, and receives data from the input unit 107 or a storage device, calculates and processes it, and then outputs it to a storage device or the like.

[0022] The main memory unit 102 is a read-only memory (ROM), a random access memory (RAM), etc. The main memory unit 102 is a storage device that stores or temporarily saves programs and data, such as an operating system (OS), which is basic software executed by the control unit 101, and application software.

[0023] The auxiliary storage unit 103 is a storage device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive) that stores data related to application software and the like.

[0024] The drive device 104 reads the program from a recording medium 105, such as a flexible disk, and installs it in a storage device.

[0025] A predetermined program is stored in the recording medium 105. The program stored in the recording medium 105 is installed in the processing device 100 via the drive device 104. The installed predetermined program can be executed by the processing device 100.

[0026] The network I / F unit 106 is an interface between the processing device 100 and peripheral devices having a communication function connected via a network constructed by a data transmission path such as a wired and / or wireless line.

[0027] The input unit 107 includes a keyboard equipped with cursor keys, a numeric keypad (numeric input keys), various function keys, a mouse, a touch pad, and the like.

[0028] 2, various processes described below can be realized by causing the processing device 100 to execute a program. Alternatively, the program can be recorded on a recording medium 105, and the recording medium 105 on which the program is recorded can be read by the processing device 100 to realize various processes described below. Various types of recording media can be used for the recording medium 105. For example, the recording medium 105 may be a recording medium that records information optically, electrically, or magnetically, such as a CD (Compact Disc)-ROM, a flexible disk, or a magneto-optical disk, or a semiconductor memory that records information electrically, such as a ROM or a flash memory. The concept of a recording medium does not include a carrier wave.

[0029] In this embodiment, the processing device 100 can be realized by one computer, but may also be realized by a combination of multiple computers. Also, an external server computer may be used.

[0030] Fig. 3 is a diagram schematically illustrating the function of the processing device 100. Fig. 4 to Fig. 6 are explanatory diagrams of the specific processing of the processing device 100, Fig. 4 is an explanatory diagram of a frame obtained by short-time Fourier transform, Fig. 5 is an explanatory diagram of an RMS change amount curve, and Fig. 6 is an explanatory diagram of a threshold setting method.

[0031] The processing device 100 includes a sound data acquisition unit 150, a Fourier transform processing unit 151, a harmonic percussion sound separation processing unit 152, an effective value calculation unit 153, an effective value change amount calculation unit 154, a threshold setting unit 155, an abnormality detection unit 156, and an inspection result storage unit 160. Each of the units from the sound data acquisition unit 150 to the abnormality detection unit 156 can be realized by the control unit 101 shown in FIG. 2 executing one or more programs stored in the main memory unit 102 and / or the auxiliary memory unit 103 and / or the recording medium 105 shown in FIG. 2. The inspection result storage unit 160 can be realized by the main memory unit 102 and / or the auxiliary memory unit 103 and / or the recording medium 105 shown in FIG. 2.

[0032] The sound data acquiring unit 150 acquires sound data from the test data acquiring device 2 (microphones 21 to 24). The sound data acquiring unit 150 may acquire sound data for processing via the sound board 30 or the A / D converter 32. The sound data is acquired as time-series sound data.

[0033] The Fourier transform processing unit 151 performs a short-time Fourier transform (STFT) process on the sound time-series data acquired by the sound data acquisition unit 150. In this case, the Fourier transform processing unit 151 extracts multiple frames by applying a window (window function) that moves with time to the sound time-series data. This allows frequency characteristic data (frames) to be obtained for each predetermined time period determined according to the window.

[0034] 4 schematically shows a plurality of frames obtained as a result of processing by the Fourier transform processing unit 151. Each frame is frequency characteristic data, and is a power spectrum with frequency on the horizontal axis and power on the vertical axis. In FIG. 4, the power spectra for each frame are indicated in chronological order as 400(t1) to 400(t7). In this case, for example, frame 400(t1) and frame 400(t2) are adjacent to each other in the chronological order, and frame 400(t2) and frame 400(t3) are adjacent to each other in the chronological order, and so on.

[0035] The harmonic percussion sound separation processor 152 separates percussive sound components in each frame using harmonic percussion sound separation (HPSS: Harmonic / Percussive Sound Separation) processing. In this case, it is possible to accurately detect sudden sounds (sudden sounds) that occur when components interfere with each other or with a foreign object.

[0036] The effective value calculation unit 153 calculates an effective value related to the hit sound component. The effective value can be any value, but in this embodiment, it is RMS (Root Mean Square). The RMS is calculated for each frame.

[0037] The RMS change calculation unit 154 calculates the difference between the RMS values ​​of adjacent frames in a time series. That is, the RMS change calculation unit 154 calculates the RMS change in frame order in the time series. For example, in the example shown in FIG. 4, the change between the RMS value from frame 400(t1) and the RMS value from frame 400(t2) is calculated, the change between the RMS value from frame 400(t2) and the RMS value from frame 400(t3) is calculated, and so on. Hereinafter, the RMS change for each frame calculated in this manner will be simply referred to as the time series of RMS change. FIG. 5 shows an example of a time series of RMS change, where the horizontal axis represents time and the vertical axis represents RMS change.

[0038] The threshold setting unit 155 sets a threshold used for determination by the anomaly detection unit 156, which will be described later. The threshold may be constant, but is preferably set for each data (target product) by the threshold setting unit 155. In this embodiment, the threshold setting unit 155 sets a threshold for determination related to the time-series data of sound, based on a statistical value derived based on the time-series data of the sound acquired by the sound data acquisition unit 150.

[0039] However, even for products with the same product number (for example, normal products), the sound time-series data acquired by the sound data acquisition unit 150 may not necessarily be the same due to individual differences and external factors (for example, the environment inside the soundproof room). In other words, the time-series data to be processed (sound time-series data) has fluctuations (variations) that are unrelated to whether it is normal or not.

[0040] In this regard, according to this embodiment, a threshold is set for each target product (i.e., for each test) based on a statistical value derived from the time-series data to be processed, so that such variable factors (variable factors unrelated to whether the product is normal or not) can be effectively eliminated, thereby obtaining highly reliable test results.

[0041] In this embodiment, the statistical value is a statistical value related to the time series of the corresponding RMS variation. The threshold value set for the time series data of the sound acquired by the sound data acquisition unit 150 is set based on the statistical value related to the time series of the RMS variation obtained from the time series data.

[0042] The statistical value may be any value such as a mean value, a median value, a variance, a standard deviation, etc., but in this embodiment, quartiles are used. Specifically, the threshold is set based on the following. Threshold = q3 + n × IQR Here, q3 is the third quartile, n is the coefficient, and IQR is an abbreviation for Interquartile Range. The coefficient n is arbitrary, but for example, 1.5. In this case, it is equivalent to approximately +3σ from the data median.

[0043] The abnormality detection unit 156 detects an abnormality in the target product (i.e., an abnormality during operation of the product) based on the time series of the RMS change amount derived by the effective value change amount calculation unit 154 and the threshold value set (calculated) by the threshold value setting unit 155. In this case, the abnormality detection unit 156 may detect an abnormality based on the value of a parameter indicating the number of times, frequency, or rate at which the time series of the RMS change amount exceeds the threshold value.

[0044] In this embodiment, as an example, the abnormality detection unit 156 includes a threshold determination processing unit 1562 , an abnormality degree calculation unit 1564 , and an abnormality determination processing unit 1566 .

[0045] The threshold determination processing unit 1562 calculates the number of frames in which the RMS change amount exceeds the threshold value, based on the time series of the RMS change amount derived by the RMS change amount calculation unit 154 and the threshold value set by the threshold value setting unit 155. Fig. 6 shows an example of threshold values ​​set for the time series of the RMS change amount shown in Fig. 5. In this case, the range in which the RMS change amount exceeds the threshold value is the range indicated by p1 and p2, and the number of frames (time width) within this range is calculated.

[0046] The abnormality degree calculation unit 1564 calculates the abnormality degree based on the determination result of the threshold determination processing unit 1562. In this embodiment, the abnormality degree is the ratio of the number of frames in which the RMS change amount exceeds a threshold, and is specifically as follows: Abnormality = N1 / N total where N1 is the number of frames in which the RMS change exceeds the threshold, and N total is the total number of frames (the total number of frames for which the RMS change amount is calculated by the effective value calculation unit 153). total Instead of N total It may be set to -1.

[0047] The abnormality determination processing unit 1566 determines whether the degree of abnormality calculated by the abnormality degree calculation unit 1564 exceeds a predetermined reference value. The reference value may be a statistical value derived based on time-series data of similar sounds related to multiple normal products with the same product number. In this embodiment, as an example, the reference value is the average + 3σ. If the reference value exceeds 3σ, the abnormality determination processing unit 1566 determines that the current target product is abnormal; otherwise, it determines that the current target product is normal. This effectively reduces the possibility of erroneously detecting a normal sudden sound as an "abnormal sudden sound."

[0048] The inspection result storage unit 160 stores the determination results (inspection results) by the abnormality determination processing unit 1566. The inspection results may be stored together with operation information and the like in association with the product number.

[0049] Next, the flow of main processing executed by the processing device 100 of this embodiment will be described with reference to FIG.

[0050] FIG. 7 is a schematic flowchart showing the flow of main processes executed by the processing device 100 of this embodiment.

[0051] In step S700, the processing device 100 acquires time-series data of the sound related to the target product at this time.

[0052] In step S702, the processing device 100 performs a short-time Fourier transform (STFT) process on the time-series data obtained in step S700. As a result, a plurality of (here, N total frames are extracted.

[0053] In step S704, the processing device 100 calculates the N obtained in step S702. total For each of the frames, the percussive components are separated using harmonic percussion separation processing.

[0054] In step S706, the processing device 100 calculates the RMS of the hit sound components for each frame obtained in step S704, thereby obtaining the RMS for each frame.

[0055] In step S708, the processing device 100 calculates the RMS variation based on the RMS for each frame obtained in step S706. In this case, the RMS variation is N total -1 is calculated.

[0056] In step S710, the processing device 100 sets a threshold value based on the RMS variation (time series of RMS variation) obtained in step S708. The method for setting the threshold value is as described above.

[0057] In step S712, the processing device 100 counts the total number N1 of frames in which the RMS variation exceeds the threshold, based on the RMS variation obtained in step S708 and the threshold obtained in step S710.

[0058] In step S714, the processing device 100 calculates the degree of abnormality (=N1 / N total ) is calculated.

[0059] In step S716, the processing device 100 determines whether the degree of abnormality obtained in step S714 exceeds a reference value. The reference value is as described above, and may be, for example, 3σ, a statistical value derived based on time-series data of similar sounds from multiple normal products. If the determination result is "YES," the process proceeds to step S718; otherwise, the process proceeds to step S720.

[0060] In step S718, the processing device 100 determines that the current target product is abnormal. In a modified example, the processing device 100 may determine that the current target product may be abnormal, or may calculate the possibility of abnormality as a score. In this case, a re-inspection or the like is prompted.

[0061] In step S720, the processing device 100 determines that the target product at this time is not abnormal.

[0062] In this way, according to the process shown in FIG. 7, for each piece of time-series data of sound related to the target product, it is possible to detect whether or not there is an abnormality (or the possibility of an abnormality) in the target product based on the time-series data.

[0063] Incidentally, abnormal sounds (including strange noises) that occur when a product is abnormal can take a variety of forms, but it is difficult to accurately detect sudden sounds using a method that evaluates the intensity level of operating noise for each frequency band.

[0064] In this regard, according to the present embodiment, by performing a threshold determination on the time series of the RMS change amount, it becomes possible to perform an evaluation (determination) taking into account the time element, and it becomes possible to detect a sudden sound with high accuracy.

[0065] Incidentally, sudden sounds that can be detected by performing threshold determination on the time series of the RMS change amount can include normal sudden sounds (for example, in the case of an electric sunroof, the sound of the glass roof rising when it starts to move from a closed state) and abnormal sudden sounds. Note that in the case of an electric sunroof, an abnormal sudden sound can be, for example, the sound of interference between the glass roof and a rail or the like.

[0066] In this regard, simply performing a threshold determination on the RMS change amount may result in a normal sudden sound being erroneously detected as an "abnormal sudden sound."

[0067] In contrast to this, according to this embodiment, not only is a threshold determination performed on the RMS change amount, but a parameter called the degree of abnormality is further derived and used for determination, so that it is possible to effectively reduce the possibility of erroneously detecting a normal sudden sound as an "abnormal sudden sound."

[0068] Although each embodiment has been described in detail above, it is not limited to the specific embodiment, and various modifications and changes are possible within the scope of the claims. It is also possible to combine all or a plurality of components of the above-described embodiments.

[0069] For example, in the above-described embodiment, sound data is used as input data, but vibration data during product operation may also be used as input data. In this case, the vibration data may be time-series data of acceleration signals detected by an acceleration sensor.

[0070] In the above-described embodiment, the RMS value is calculated for each frame after the short-time Fourier transform is performed on the sound time-series data, but this is not limiting. For example, the effective value (e.g., RMS) may be calculated for each predetermined time period before the short-time Fourier transform is performed on the sound time-series data.

[0071] Furthermore, in the above-described embodiment, harmonic percussion sound separation processing is performed, but depending on the characteristics of the abnormal sound to be detected, the harmonic percussion sound separation processing may be omitted, or other alternative processing or additional processing may be performed. [Explanation of symbols]

[0072] 100 Processing device (product inspection device), 150 Sound data acquisition unit (data acquisition unit), 152 Harmonic percussion sound separation processing unit (processing unit), 153 Effective value calculation unit (first calculation unit), 154 Effective value change amount calculation unit (second calculation unit), 155 Threshold setting unit (setting unit), 156 Abnormality detection unit

Claims

1. a data acquisition unit that acquires time-series data related to sound or vibration during product operation; a first calculation unit that calculates an effective value for each predetermined time from a data portion of the time series data for each predetermined time or from frequency characteristic data for each predetermined time based on the time series data; a second calculation unit that calculates a difference value between the effective values ​​for each of the predetermined times that are adjacent in time series; and an abnormality detection unit that detects an abnormality during product operation based on the difference value.

2. the abnormality detection unit detects the abnormality based on a determination result of whether the difference value exceeds a threshold value; The product inspection device according to claim 1 , further comprising a setting unit that sets the threshold value based on a statistical value derived from the time-series data.

3. The product inspection device according to claim 2 , wherein the abnormality detection unit detects the abnormality based on a parameter value indicating the number, frequency, or rate at which the difference value exceeds the threshold value.

4. a processing unit that separates percussive components from the frequency characteristic data by harmonic / percussive sound separation processing; The product inspection device according to claim 1 , wherein the first calculation unit calculates an RMS (Root Mean Square) that is the effective value for each predetermined time from the hitting sound component.

5. Acquire time-series data related to sound or vibration during product operation, calculating an effective value for each predetermined time from frequency characteristic data for each predetermined time based on the time series data; calculating a difference between the effective values ​​for each of the predetermined times adjacent in time series; and detecting an abnormality during product operation based on the difference value.

Images