JP2007101244A - Inspection device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an inspection device capable of discriminating between a nondefective article and a defective article with high accuracy, by using an order zone waveform or a rotational frequency zone waveform. <P>SOLUTION: This device is equipped with a combination part for correlating a time zone waveform of a sound, vibration or the like generated from an inputted inspection object with rotational frequency information of the inspection object; a waveform conversion part for determining the rotational frequency zone waveform and the order zone waveform, based on the time zone waveform and the rotational frequency information correlated with each other; an order filter for extracting and/or removing one or a plurality of set order components, based on the time zone waveform and the rotational frequency information correlated with each other; a characteristics quantity calculation means for acquiring the rotational frequency zone waveform, the order zone waveform and a waveform outputted from the order filter, and determining the characteristics quantity, based on the acquired information; and a determination part for determining whether the inspection object is a nondefective article, based on the characteristics quantity determined by the characteristic quantity operation means. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an inspection apparatus that determines whether or not an inspection object is a good product based on waveform information such as sound and vibration generated from the input inspection object.

  In automobiles and home appliances, a rotating device incorporating a motor is very often used. For example, taking an automobile as an example, rotating equipment is mounted everywhere in the engine, power steering, power seat, mission and others. Household appliances include refrigerators, air conditioners, washing machines and other various products. When such a rotating device actually operates, a sound is generated along with the rotation of the motor or the like.

  Some of these sounds are inevitably generated along with normal operations, and other sounds are generated due to defects. Causes of abnormal noise caused by defects include bearing abnormalities, abnormal internal contact, imbalance, and contamination. For example, every time the gear rotates once, the abnormal noise is generated at a frequency of one time. The gear is missing, the foreign object is bitten, the spot is scratched, and the rotating part and the fixed part inside the motor are rubbed for a moment during rotation. There are things. The abnormal sound accompanying such a failure may cause further failure.

  There are various sounds that humans feel uncomfortable, for example, within a human audible range of 20 Hz to 20 kHz. As an example of the frequency of the uncomfortable sound, there is, for example, about 15 kHz. When the sound having such a frequency component of 15 kHz is generated, it is not a failure of the product, but a product that generates unpleasant sound feels poor in quality when viewed from the user. Therefore, the uncomfortable sound also becomes an abnormal sound. The frequency of abnormal sound is not limited to 15 kHz.

  Therefore, for the purpose of quality assurance for each of these products, the production factory performs “sensory inspection” that relies on the five senses such as hearing and tactile sensation by the inspector to determine the presence or absence of abnormal sounds. Specifically, an inspector listens to a product to be inspected in operation with his / her ear or touches it with his / her hand to check vibrations. Here, the sensory test is a test in which an attribute that can be sensed by a human sensory organ is performed by the human sensory organ itself.

  The sensory test that relies on the five senses of such inspectors not only requires skilled skills but also results in large variations in judgment results such as individual differences and changes with time. Therefore, in order to solve such a problem, an abnormal sound inspection system for the purpose of stable inspection based on a quantitative and clear standard has been developed. This abnormal sound inspection system is an apparatus for the purpose of automating the “sensory inspection” process. It measures the vibration and sound of the product drive unit with a sensor, extracts feature values from the obtained waveform data, and extracts the extracted features. Based on the quantity value, it is determined whether the product is good or defective.

  One of the feature quantities used when making the determination is power in a certain frequency band. As described above, when it is desired to determine that the 15 kHz sound is abnormal sound, frequency analysis is performed using an FFT analyzer, and when the power in the frequency band including 15 kHz exceeds a preset threshold, it is determined as a defective product. .

Another feature amount is focused on the power of the order component. If it is known that the power of the nth-order component of the rotation speed to be inspected increases when a failure or abnormality occurs, the obtained waveform data is converted into an order-domain waveform. When the power of the target n-order component is acquired and the acquired value exceeds a preset threshold value, it is determined as a defective product. Such an abnormal sound inspection system using an order waveform is disclosed in, for example, Patent Document 1 and the like.
Japanese Patent Laid-Open No. 10-82689

  When discriminating between non-defective products and defective products by paying attention to the power of the n-order component described above, the following problems occur. For example, it is assumed that the waveform signals obtained by passing two waveform data through an order filter that extracts n-order components are as shown in FIGS. FIG. 1A shows the waveform data of the non-defective product, and FIG. 1B shows the waveform data of the defective product. In each figure, consider that waveform data is divided into blocks surrounded by squares of wavy lines.

  The two waveforms after passing through the order filter shown in FIGS. 1A and 1B have the same elements and the order of the six blocks constituting the waveforms. That is, in the waveform shown in FIG. 1A, a block made up of a waveform having a large amplitude and a block made up of a waveform having a small amplitude appear alternately, but the waveform shown in FIG. Three blocks having a large waveform appear in succession, followed by three blocks having a waveform having a small amplitude. When the determination algorithm for determining whether or not the product is a non-defective product obtains the power (effective value) for 6 blocks from the waveform data of the n-th order component, and determines that an abnormality occurs when the power exceeds the threshold, the two waveform data The powers are equal and they cannot be separated.

  Furthermore, the conventional abnormal noise inspection system makes a determination based on waveform data acquired in a state where the inspection object is rotated at a constant speed. Therefore, for example, in a car engine or the like, an abnormality that occurs when the rotation is accelerated or decelerated cannot be detected.

  It is an object of the present invention to provide an inspection apparatus that can accurately discriminate between a non-defective product and a defective product by using an order frequency waveform or a rotational frequency waveform. Another object of the present invention is to enable analysis even in an acceleration / deceleration state when the object to be inspected rotates.

  In order to achieve the above-described object, the inspection apparatus of the present invention includes a coupling means for associating time domain waveforms such as sound and vibration generated from an input inspection object with rotation speed information of the inspection object. Waveform conversion means for obtaining at least one of a rotational frequency range waveform and an order frequency waveform based on the associated time domain waveform and rotational speed information, and a setting based on the associated time domain waveform and rotational speed information Acquire at least one of an order filter that extracts and / or removes one or more order components, the rotational frequency range waveform, the order frequency waveform, and a waveform output from the order filter, and the acquired information A feature amount calculation means for obtaining a feature amount based on the above, and a determination means for determining whether or not the inspection object is a non-defective product based on the feature amount obtained by the feature amount calculation means, Symptoms amount calculating means, and configured to have a function of calculating a feature amount from the shape of at least the acquired waveform. The time domain waveform that has not been filtered by the order filter may be subjected to waveform conversion by passing the waveform once received by the feature amount calculation unit to the waveform conversion unit, or the feature amount calculation unit via the waveform conversion unit. You may make it input to.

  In the embodiment, the feature amount calculating means is composed of a frame dividing unit 25, a frame feature amount calculating unit 26, and a final feature amount calculating unit 27. In the embodiment, the frame division is performed. However, the present invention includes one that obtains the feature amount without performing the frame division.

  In the present invention, the threshold value processing is not performed for a certain order component, and the feature amount calculation unit calculates the feature amount from the rotational frequency range waveform and / or the order frequency waveform and the shape of the waveform output from the order filter. Therefore, it is possible to make a more accurate determination. In other words, the inspection apparatus of the present invention can distinguish between non-defective products and defective products that cannot be separated by the threshold processing by appropriately setting the feature amount.

  The waveform converting means may comprise means for normalizing an amplitude value of a given time domain waveform based on the rotation speed information. In the embodiment, this normalization is obtained by dividing the amplitude value of the input waveform by the number of rotations. The present invention normalizes the amplitude value of the time domain waveform based on the rotation speed information even when the rotation speed of the inspection object changes during measurement of waveform data such as sound and vibration generated from the inspection object. Therefore, the amplitude value of each waveform is not affected by fluctuations in the rotational speed. The inspection apparatus can perform analysis even when the inspection target is in an accelerated / decelerated state, and can perform accurate determination.

  You may comprise so that the output of the said order filter may be given to the said feature-value calculating means through the said waveform conversion means. In the present invention, the feature quantity can be obtained based on the output of the order filter, or based on the rotational frequency domain waveform or the order domain waveform obtained by converting the time domain waveform output from the order filter by the waveform converting means. The feature amount can also be obtained. In this way, the inspection apparatus can perform more accurate determination by using various kinds of feature amounts.

  A frequency filter for filtering a time domain waveform associated with the rotation speed information may be provided, and an output of the frequency filter may be provided to at least one of the order filter and the waveform conversion unit. According to the present invention, since unnecessary frequency components are removed or required frequency components are extracted by the frequency filter, various arithmetic processes in the subsequent stages are based on the unnecessary information cut. Therefore, highly accurate determination can be performed.

  The feature amount calculating means is provided with a frame dividing means for dividing the waveform data not divided into frames, and the waveform data divided by the frame dividing means directly or via the waveform converting means. The feature amount to be given to the determination means from a frame feature amount calculation means for obtaining a frame feature amount from waveform data in units of frames and a plurality of frame feature amounts for the same inspection object obtained by the frame feature amount calculation means. And a final feature amount calculating means to be obtained. In the present invention, a given waveform is divided into frames, a frame feature amount for each frame is obtained, and a final feature amount is obtained from a plurality of frame feature amounts for the same inspection object. Conventionally, it is possible to determine a product in which a good product and a defective product cannot be separated.

  It is preferable to divide the frame so that it is equally spaced in the rotation speed range. For example, in the case of a defect that generates an abnormal noise every time the inspection object rotates once, the generation timing of the abnormal noise also changes when the rotation speed changes. In this case, the generation timing of abnormal noise in the rotation speed waveform is constant. Therefore, by dividing the frame into equal intervals in the rotation speed range, the inspection apparatus can perform analysis even when the inspection target is in an accelerated / decelerated state, and can perform accurate determination.

  In the present invention, a non-defective product and a defective product can be discriminated with high accuracy by using an order range waveform, a rotational frequency range waveform and / or a waveform output from an order filter. Further, when the object to be inspected rotates, analysis is possible not only in the constant speed rotation state but also in the acceleration / deceleration state.

  FIG. 2 shows a preferred embodiment of the present invention. First, the microphone 2 is arranged close to the inspection object 1 and the acceleration pickup 3 is brought into contact with the inspection object 1. The microphone 2 detects sound generated from the inspection object 1, and the acceleration pickup 3 detects vibration of the inspection object 1. Signals detected by sensors such as the microphone 2 and the acceleration pickup 3 are amplified by the amplifier 4, changed to digital data by the A / D converter 5, and then given to the inspection device 10.

  The inspection object 1 performs a rotational motion of an engine or the like. The inspection object 1 is provided with a rotation speed detection device that detects the rotation speed. The rotational speed information output from the rotational speed detection device is also given to the inspection device 10. When the rotation speed information output from the rotation speed detection device is analog data, the output of the rotation speed detection device is changed to digital data by the A / D converter 5 and then provided to the inspection device 10.

  In FIG. 2, the A / D converter 5 is described as being provided outside the inspection apparatus 10, but may be provided inside the inspection apparatus 10. There is also a case where a rotation speed detection device that detects a specific rotation speed is not attached to the inspection object 1 and a sensor that outputs a signal (for example, a pulse train) corresponding to the rotation speed is attached like an encoder. is there. In this case, the output of the encoder or the like is configured to be input to the inspection apparatus 10, and a rotation speed calculation unit that calculates the rotation speed from a signal corresponding to the input rotation speed is provided in the inspection apparatus 10.

  The inspection device 10 acquires waveform data based on sound data collected by the microphone 2 and vibration data collected by the acceleration pickup 3 and rotation speed information of the inspection object 1, and uses the waveform data and rotation speed information. Then, the feature amount is extracted and it is determined whether or not the inspection object 1 is a non-defective product. The inspection device 10 includes a computer, and includes a CPU body 10a, an input device 10b such as a keyboard and a mouse, and a display 10c. Although not shown, the CPU main body 10a is generally composed of a central processing unit (CPU), a main storage device (memory), and an auxiliary storage device (memory). The internal configuration of the inspection apparatus shown in FIGS. 3 and 4 is mainly realized by software stored and executed in the CPU main body 10a.

  As illustrated in FIG. 3, the inspection apparatus 10 includes a feature amount calculation unit 11 that calculates a feature amount from waveform data acquired via the A / D converter 5, and a feature amount calculated by the feature amount calculation unit 11. And a determination unit 12 that performs pass / fail determination. The determination result of the determination unit 12 can be displayed on the display 10c in real time or stored in a storage device, for example. Although not shown in the figure, the type of feature amount calculated by the feature amount calculation unit 11, the parameter of the feature amount, and the like are stored in the feature amount parameter storage unit. The feature quantity computing unit 11 performs feature quantity computation according to the information stored in the feature quantity parameter storage unit, and obtains the feature quantity value. The determination unit 12 determines whether or not the product is a non-defective product by executing a determination algorithm such as fuzzy inference based on the feature value given from the feature value calculation unit 11.

  FIG. 4 is a block diagram showing an internal configuration of the feature amount calculation unit 11. The feature amount calculation unit 11 includes a combining unit 21, a frequency filter 22, an order filter 23, a waveform conversion unit 24, a frame division unit 25, a frame feature amount calculation unit 26, a final feature amount calculation unit 27, It has.

  The combining unit 21 associates the given waveform data with the rotation speed information, and provides the associated data to the frequency filter 22, the order filter 23, and the frame dividing unit 25, respectively. In the present embodiment, the inspection object 1 may rotate at a constant speed, or the rotational speed may change from moment to moment as in acceleration operation or deceleration operation. The waveform data given to the coupling unit 21 is composed of a set of digital data (amplitude values) sampled at the sampling period. Therefore, the rotational speed information when the amplitude value is sampled is acquired and associated with the amplitude value. For this association, for example, a table in which amplitude values are associated with rotation information is created and stored. In addition, each amplitude value that constitutes the waveform data is a record that specifies information for specifying the waveform data and the amplitude value at which the waveform data needs to be reproduced on the time axis of the waveform data. If the data structure has a number or the like, it may be converted to a data structure in which the rotation speed information data is added after the amplitude value.

  The frequency filter 22 extracts or removes a predetermined frequency component from the waveform data (time domain waveform) given from the coupling unit 21. The frequency filter 22 is composed of, for example, a bandpass filter, and the frequency band to be extracted is one or a plurality of regions set as parameters. The waveform data that has passed through the frequency filter 22 is given to the order filter 23 and the waveform converter 24. The rotation speed information is also associated with each amplitude value constituting the filtered waveform data given at this time. That is, the frequency filter 22 simply extracts or removes predetermined frequency components from given waveform data regardless of the rotation speed information, and each amplitude value on the time axis after filtering is converted. However, when attention is paid to a certain point on the time axis, the rotation speed information does not change. Therefore, the data in which only the amplitude value is changed by the filtering process among the rotation number information and the amplitude value associated by the combining unit 21 is given to the order filter 23 and the waveform converting unit 24.

  The order filter 23 extracts or removes an n-order component based on given waveform data and rotation speed information. Here, n may be a single value or a plurality of values. Specifically, as an example where n is one value, there is one that performs a filtering process that extracts only the quaternary component. Examples of multiple n values include filtering by specifying multiple order components such as extracting the third order component or higher (removing the second order component or lower) or extracting the sixth order component from the fourth order component. There is something to process.

Here, the order is a value obtained by normalizing the frequency with the basic rotational speed, and is specifically defined by the following equation. In the following formula, the basic rotational speed is the rotational speed of the inspection object 1 given as the rotational speed information.
Order = frequency / basic speed

  When the filtering process by the order filter 23 is performed, for example, waveform data as shown in FIG. 5A is converted into waveform data as shown in FIG. As is apparent from the figure, the waveform data that has passed through the order filter 23 is also waveform data (time domain waveform) in the time axis region. As in the case of the frequency filter 22, the information output from the order filter 23 is information that associates the waveform data with the rotation speed information. The output (waveform data + rotation number information) of the order filter 23 is given to the waveform converting unit 24 and the frame dividing unit 25.

  The order filter 23 performs a filtering process on the waveform data of the predetermined frequency component filtered by the frequency filter 22 and the waveform data of the raw data that does not pass through the frequency filter 22. Further, which order component is extracted and removed is set as a parameter. This is the same as in the case of the frequency filter 22.

  The frame dividing unit 25 divides the given waveform data into frames according to the set conditions, and outputs the divided waveform unit waveform data. The setting conditions for dividing the frame include the frame width (time of one frame), the size of data constituting one frame, the degree of overlap of previous and subsequent frames (including overlap 0), and the like. Similar to the parameters in both filters 22 and 23, the setting conditions for frame division are also stored and held in the feature quantity / parameter storage unit as parameters and are given to the frame division unit 25.

  The data given to the frame dividing unit 25 may be waveform data subjected to filtering processing given from the order filter 23 and raw waveform data given from the combining unit 21. Although the waveform data to be given is different, it is the same in that it is a time domain waveform. Therefore, the frame dividing unit 25 can execute any waveform data in accordance with the setting conditions for dividing the same frame.

  Waveform data divided into frames (associated with rotation speed information) is provided to the waveform converter 24 and the frame feature quantity calculator 26. Which processing unit is passed to is determined by the data path to the frame dividing unit 25. That is, when the waveform data received from the order filter 23 or the combining unit 21 is divided into frames by the frame dividing unit 25, the frame dividing unit 25 passes the divided waveform data in units of frames to the waveform converting unit 24. On the other hand, when the output of the frequency filter 22 or the order filter 23 is given to the waveform converting unit 24 and the data obtained by waveform conversion is received by the frame dividing unit 25, the frame divided data is supplied to the frame feature amount calculating unit 26. hand over. When there is no need to perform waveform conversion, the frame dividing unit 25 may divide the waveform data received from the order filter 23 or the combining unit 21 into frames and pass the result to the frame feature amount calculating unit 26.

  As shown in FIG. 6, the waveform converter 24 receives two pieces of information, a time domain waveform and rotation speed information, calculates a rotation speed waveform and an order frequency waveform, and outputs the calculated waveform. is there. The rotation speed range waveform is waveform data in which the horizontal axis represents the rotation speed and the vertical axis represents the amplitude value. As shown in FIG. 7, the order-domain waveform is obtained by taking the order on the horizontal axis and the amplitude value on the vertical axis. In addition, since the calculation algorithm itself which calculates | requires a rotation speed range waveform and an order range waveform based on two input information is well-known, description of a specific algorithm is abbreviate | omitted.

  The frame feature amount calculation unit 26 is based on the waveform data of each frame divided by the frame division unit 25 and the data (frame unit) converted by the waveform conversion unit 24, and the feature amount of each frame (frame feature amount). Is calculated. As shown in FIG. 8, the frame feature quantity computing unit 26 obtains the rotational speed domain feature quantity when the rotational speed domain waveform is input, and obtains the order domain feature quantity when the order frequency waveform is input. . RMS (effective value), peak-to-peak value, average value, etc. can be used as the rotational speed range feature quantity. As the order range feature amount, for example, as shown in FIG. 7, a shape formed by connecting amplitude values of a plurality of order components (indicated by broken lines in the drawing) can be used. For example, it may be simply composed of a combination of magnitude relationships between the amplitude values of the n-order component and the (n + 1) -order component, or the ratio of a plurality of order components (for example, the ratio of the second-order component to the third-order component). You may make it use, and can take various methods.

  Each frame feature value obtained by the frame feature value computing unit 26 is given to the final feature value computing unit 27 in the next stage. The final feature quantity computing unit 27 obtains a feature quantity representing the waveform data to be inspected based on the frame feature quantity obtained from a plurality of frames. As an example, it is assumed that the rotation frequency waveform obtained by the waveform converter 24 for the waveform data for each frame divided by the frame divider 25 is as shown in FIG. The rotation frequency waveform in FIG. 9A is composed of six frames. Assuming that the frame feature amount of each frame is RMS, as shown by a black circle in FIG. Therefore, the RMS between adjacent frames repeats increasing and decreasing as indicated by arrows in FIG. 9B. In the representative feature amount calculation, for example, the average of the top 3 changes in RMS is obtained, and the obtained average value is used as the feature amount of the waveform.

  Assuming that the rotational speed range waveform data shown in FIG. 9A is a non-defective product and the rotational speed range waveform data shown in FIG. The feature value of the defective product becomes small and can be distinguished from the non-defective product.

  In the above-described example, the frame feature amount calculation unit 26 obtains the frame feature amount for the order frequency waveform and the rotation frequency range waveform obtained by the waveform conversion unit 24. However, the present invention is not limited to this. There is no. For example, a time domain waveform (see FIG. 5B) obtained by performing the filtering process with the order filter 23 is supplied to the frame dividing unit 25 and divided for each frame. Then, the frame feature amount may be obtained based on the shape of the time data for each frame obtained by giving each frame to the frame feature amount calculation unit 26 and performing the order filtering process. As this frame feature amount, for example, RMS, peak-to-peak value, or the like can be used. The frame feature value obtained in this way is given to the final feature value computing unit 27.

  When the rotational speed changes, the detected waveform data may be affected. For example, as shown in FIG. 11 (a), when the number of rotations is increasing, as shown in FIG. 11 (b), the region B having a larger number of rotations is compared with the region A having a smaller number of rotations. However, the generated sound becomes louder. If the amplitude value of the waveform data based on sound is small, it is determined that the product is a non-defective product, and if the amplitude value is large, the product is determined to be defective. FIG. 12A shows the amplitude value of a non-defective product at the rotational speed X, and FIG. 12B shows the amplitude value of a defective product at the rotational speed X. FIG. 12C shows the amplitude value of a non-defective product at the rotational speed Y (Y> X), and FIG. 12B shows the amplitude value of a defective product at the rotational speed Y. As is apparent from the figure, it is assumed that the magnitude relationship that the amplitude value of the non-defective product is smaller than the amplitude value of the defective product is maintained at the same rotation speed. However, when the rotation speed changes from moment to moment, a reverse phenomenon occurs in which the non-defective product amplitude value is larger than the non-defective product amplitude value, as shown in FIGS. 12 (b) and 12 (c). There is a risk of misjudgment.

Therefore, the waveform converter 24 has a function of normalizing the waveform amplitude value. Thus, for example, when a time domain waveform as shown in FIG. 13A is given, a time domain waveform with normalized amplitude values is obtained as shown in FIG. 13B. Specific waveform conversion processing is, for example,
It can be obtained from the amplitude value / rotation speed. That is, since the amplitude value is proportional to the rotation speed, normalization can be performed by dividing the amplitude value by the rotation speed as in the above-described equation.

  The waveform converter 24 may output the time domain waveform normalized in this way, or calculate a rotation frequency waveform or an order frequency waveform based on the normalized time domain waveform and the rotation speed information, You may make it output.

  As shown in FIG. 14, when there is a defect 30 a ′ in a part of the teeth 30 a of the gear 30, abnormal noise is generated every time the gear 30 makes one rotation. When the gear 30 is rotating at a constant speed, abnormal noise is generated at regular time intervals. When the gear 30 is accelerating and rotating, as shown in FIG. 15 (a), the generation interval of abnormal noise generated with the defect 30a 'is gradually shortened. As shown in FIG. 15A, after the frame dividing unit 25 divides the time domain waveform at equal intervals in the time domain, the waveform converting unit 24 converts the time domain waveform of each frame. When converted to a rotation speed waveform, the result is as shown in FIG. As is apparent from FIG. 15B, the frame width of each frame constituting the rotation frequency waveform is different, and if the frame feature amount is obtained based on this frame, the abnormal sound generation phenomenon may not be accurately captured. There is.

  Therefore, as shown in FIG. 16, the frame is divided at equal intervals in the rotation speed range. As a result, as shown in FIG. 16B, the frames can be divided so that different sounds are generated at the same timing in each frame. Therefore, it is possible to accurately detect the phenomenon of occurrence of abnormal noise and perform highly accurate determination.

  For example, the frame dividing unit 24 divides frames at unequal intervals in the time domain as shown in FIG. 16A using the rotation number information. Yes. Specifically, this can be dealt with by dividing the frame by a function that shortens the frame width in the time domain as the number of rotations increases. Alternatively, the waveform conversion unit 24 may convert the rotation frequency range waveform to a frame division unit 24, which may be divided into frames at regular intervals.

It is a figure explaining the conventional problem. It is a figure which shows an example of the test | inspection system containing the test | inspection apparatus of this invention. It is a figure which shows one Embodiment of the test | inspection apparatus of this invention. 3 is a diagram illustrating an example of an internal structure of a feature amount extraction unit that is a main part of the inspection apparatus 10. FIG. It is a figure explaining the function of an order filter. FIG. 3 is a diagram for explaining a waveform conversion unit 24; It is a figure which shows an example of an order region waveform. It is a figure explaining the frame feature-value calculating part 26. FIG. FIG. 6 is a diagram for explaining a function of a frame feature quantity computing unit 26. FIG. 6 is a diagram for explaining a function of a frame feature quantity computing unit 26. It is a figure which shows an example of the waveform data in the case of carrying out the speed increasing operation. It is a figure explaining the influence on the feature-value which a rotation speed gives. It is a figure explaining the function which normalizes an amplitude value based on rotation speed information. It is a figure which shows an example of the gear (defective product) which is a test object. It is a figure explaining the problem at the time of dividing into frames at equal intervals on the time axis. It is a figure explaining the function which divides | segments a frame into equal intervals on a rotation speed axis.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Inspection apparatus 11 Feature-value calculating part 12 Determination part 21 Combining part 22 Frequency filter 23 Order fill 24 Waveform converting part 25 Frame dividing part 26 Frame feature-value calculating part 27 Final feature-value calculating part

Claims (6)

A coupling means for associating a time domain waveform such as sound and vibration generated from the input inspection object and rotation speed information of the inspection object;
Waveform conversion means for obtaining at least one of a rotational frequency range waveform and an order frequency waveform based on the associated time domain waveform and rotational speed information, and a setting based on the associated time domain waveform and rotational speed information At least one order filter that extracts and / or removes one or more order components;
A feature quantity computing means for obtaining at least one of the rotation speed domain waveform, the order domain waveform, and a waveform output from the order filter, and obtaining a feature quantity based on the obtained information;
Determination means for determining whether or not the inspection object is a non-defective product based on the feature quantity obtained by the feature quantity calculation means,
2. The inspection apparatus according to claim 1, wherein the feature amount calculating means is configured to have a function of calculating a feature amount from at least the acquired waveform shape.   2. The inspection apparatus according to claim 1, wherein the waveform converting means includes means for normalizing an amplitude value of a given time domain waveform based on the rotation speed information.   The inspection apparatus according to claim 1, wherein an output of the order filter is configured to be supplied to the feature amount calculation unit via the waveform conversion unit. Providing a frequency filter for filtering the time domain waveform associated with the rotation speed information;
4. The inspection apparatus according to claim 1, wherein an output of the frequency filter is provided to at least one of the order filter and the waveform conversion unit. 5. The feature amount calculating means includes:
A frame dividing means for dividing the waveform data not divided into frame units into frames;
A frame feature quantity computing means for obtaining a frame feature quantity directly from the waveform data frame-divided by the frame dividing means or from the frame-unit waveform data given via the waveform converting means;
And a final feature quantity computing means for obtaining the feature quantity to be given to the determination means from a plurality of frame feature quantities for the same inspection object obtained by the frame feature quantity computing means. The inspection apparatus according to any one of claims 1 to 4.   6. The inspection apparatus according to claim 5, wherein the frame is divided so as to be equidistant in the rotation speed range.

Images