JP2012225878A - Damage detection device and method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a damage detection device and a method that can accurately detect shape abnormalities such as damage on a surface of a product, etc.SOLUTION: A damage detection device (10) in the invention detects the presence or absence of damage by exposing a surface of a product to a laser beam and measuring intensity of its diffused reflection light. The damage detection device (10) comprises: an intensity curve data detection part (13) for detecting the intensity of the diffused reflection light as intensity curve data; a reference intensity curve data acquisition part (16) for acquiring reference intensity curve data; a continuous wavelet transforming part (17) for transforming continuous wavelets; a variation calculation part (20) for calculating a variation between transform results; and a damage detection part (21) for detecting the presence or absence of damage by determining whether or not the variation is larger than a predetermined value.

Description

  The present invention relates to a technical field of a damage detection apparatus and method for detecting the presence or absence of damage by irradiating the surface of a product rotating at a predetermined cycle with laser light and measuring the intensity of diffusely reflected light.

  In general, detecting the presence or absence of physical damage in a product after manufacture is important for ensuring the reliability of the product. For example, in a product manufacturing factory, an inspection for detecting the presence or absence of appearance damage is performed when determining whether a manufactured product is defective. In addition, when a product is actually used by a user after manufacturing, if it is later damaged in the process of use, it can be detected early to prevent failure or This is important for efficient analysis of the cause even when the error actually occurs.

  Conventionally, the detection of damage in such products has been performed by visual observation, but in recent years, automation has been attempted. As an example, there is a technique for performing image analysis by imaging the surface of a product with a camera using an image sensor such as a CCD or CMOS. However, such image analysis has a large processing burden. In general, since the surface of the product has fine unevenness generated during processing, advanced image analysis is required to accurately distinguish and determine unevenness and damage that are not such damage. Therefore, there has been a problem that the processing burden becomes very large. In order to solve such a problem, a technique has been proposed in which the presence or absence of damage is determined by frequency analysis of vibrations of a product during operation. For example, in Patent Document 1, when a master gear is engaged with a gear to be inspected and rotationally driven, the displacement (vibration) of the master gear is measured with a non-contact sensor, and the measurement data is frequency-analyzed to analyze the gear. Techniques for determining the presence or absence of damage in the are disclosed.

  In such a conventional frequency analysis, a technique of spectrally analyzing vibration signals using Fourier transform is often used. However, the spectrum obtained by Fourier transform is a function of frequency, and there is a problem that time information is lost. In view of this problem, as a new method of frequency analysis, a method of performing vibration analysis using continuous wavelet transform instead of Fourier transform has been proposed (see Non-Patent Document 1).

  The present inventor has been researching and developing a method for detecting damage using such a continuous wavelet transform. When detecting the presence or absence of damage on the tooth surface of a gear as in Patent Document 1, the present inventor irradiates a laser beam. It has been found that it is possible to directly detect damage without using a master gear by performing based on irregularly reflected light generated in this way (Patent Document 2). Patent Document 3 discloses a technique for evaluating the dynamic performance by comparing the wavelet transform of different gears by treating the timing of one rotation of the gear as one cycle.

JP-A-10-3004009 JP 2007-232660 A JP 2001-242009 A

Yuji Oue, Akira Yoshida, “Failure Diagnosis of Gear Tooth Surface by Continuous and Discrete Wavelet Transform”, Tribologist, January 2003, Vol. 48, pp. 828-835

When a plurality of products of the same type are manufactured, the shape of each product varies due to tolerances generated during machining. That is, in reality, there is a considerable variation in shape even between normal products that are not damaged. In any of the above background arts, the presence / absence of damage is determined based on whether or not the spectral intensity difference is greater than a predetermined value. However, in comparison with such a simple threshold value, it is difficult to discriminate from the variation inherent in the normal product as described above, and further improvement in detection accuracy of damage is desired. That is, there is a demand for a technique that can detect damage with high accuracy even between products having variations in shape.
In particular, in Patent Document 3 described above, since the timing of one rotation of the gear is handled as one cycle, in order to detect which tooth is damaged, data before the gear is damaged is acquired in advance. It must be compared with the data. Further, in Patent Document 3, it is possible to confirm the number of teeth in which damage exists, but there is a problem that accuracy for specifying the damage position cannot be obtained sufficiently.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a damage detection apparatus and method capable of accurately detecting a shape abnormality such as damage on the surface of a product.

  In order to solve the above problems, a damage detection apparatus according to the present invention irradiates a surface of a product rotating at a predetermined cycle with a laser beam and measures the intensity of the irregularly reflected light to detect the presence or absence of damage. In the above, an intensity curve data detection unit that detects the intensity of the irregularly reflected light as intensity curve data with a rotation angle or time as a variable, and reference intensity curve data that is intensity curve data obtained when the product is normal is acquired. A reference intensity curve data acquisition unit, a continuous wavelet conversion unit that performs continuous wavelet conversion on the intensity curve data detected by the intensity curve data detection unit and the reference intensity curve data acquired by the reference intensity curve data acquisition unit, and the continuous A deviation calculation unit that calculates a deviation between the conversion results in the wavelet conversion unit, and the deviation calculated by the deviation calculation unit is Characterized by comprising a damage detection unit for detecting the presence or absence of damage by determining whether greater or not than value.

  According to the damage detection apparatus of the present invention, the time (or rotation angle) that the frequency analysis of the intensity curve data detected from the irregularly reflected light from the product is lost in the normal Fourier transform by using the continuous wavelet transform. It can be done while leaving information about. For this reason, it is possible to evaluate deviations that are difficult to handle in the past, which cannot be found by analyzing the intensity curve data using either time (or rotation angle) or frequency as a variable, so that damage can be detected more accurately. . In particular, by preparing reference intensity curve data in advance, it is possible to accurately discriminate and detect variations caused by variations between individuals that occur during machining and damage.

  The damage detection unit may specify a region where damage exists in the product by calculating a variable region in which the deviation calculated by the deviation calculation unit is larger than the predetermined value. According to this aspect, since the area where the calculated deviation is equal to or greater than the predetermined value can be specified as the damaged area, it is possible to accurately determine in which area of the product the damage exists in addition to the presence or absence of the damage. . The predetermined value may be set as appropriate according to normal variations due to product processing tolerances.

  Preferably, the storage device further includes an accumulation unit that accumulates the intensity curve data detected by the intensity curve data detection unit, and the reference intensity curve data acquisition unit averages the intensity curve data accumulated in the accumulation unit. Thus, the reference intensity curve data may be calculated. In this aspect, by calculating the reference intensity curve data by averaging the accumulated intensity curve data, highly reliable reference intensity curve data can be obtained based on the past operation history. Can be detected.

  In addition, an input unit for selecting a mother wavelet function used for continuous wavelet transform in the continuous wavelet transform unit from a plurality of types of mother wavelet functions stored in advance may be further provided. Generally, the waveform of the intensity curve data detected varies depending on the type of damage. In this aspect, the detection accuracy can be further improved by making it possible to specify an optimal mother wavelet that matches the waveform of the intensity curve data to be converted as the mother wavelet used for continuous wavelet conversion.

  Preferably, the intensity curve data detection unit may detect the intensity of the irregularly reflected light through a high-pass filter for removing a component having a predetermined frequency or less. According to the research of the present inventor, it has been clarified that the detected intensity curve data includes an unnecessary signal component due to noise or disturbance on the low frequency side. Therefore, by providing a high-pass filter, it is possible to determine damage with higher accuracy by removing these unnecessary low-frequency components and detecting intensity curve data.

  For example, the product is a gear, and the intensity curve data detection unit receives irregularly reflected light obtained by irradiating a tooth surface of the gear with laser light, and changes the intensity of the received irregularly reflected light over time. Based on this, it is preferable to detect the intensity curve data. The object to be detected by the damage detection apparatus according to the present invention is a product that rotates at a predetermined cycle, and particularly when applied to a mass-produced product that has a large number of regions (tooth surfaces) that should be inspected for the presence or absence of damage such as gears. It is.

  In order to solve the above-described problem, the damage detection method according to the present invention irradiates the surface of a product that rotates at a predetermined cycle with a laser beam and measures the intensity of the irregularly reflected light to detect the presence or absence of damage. In the above, an intensity curve data detection step for detecting the intensity of the irregularly reflected light as an intensity curve data having a rotation angle or time as a variable, and reference intensity curve data that is an intensity curve data obtained when the product is normal is acquired. A reference intensity curve data acquisition step, a continuous wavelet conversion step of performing continuous wavelet conversion on the detected intensity curve data and the acquired reference intensity curve data, and a deviation between conversion results in the continuous wavelet conversion step. It is possible to determine whether there is damage by determining a deviation and determining whether the calculated deviation is greater than a predetermined value. Characterized in that a damage detection step of detecting a.

  The damage detection method according to the present invention can be suitably realized by the damage detection apparatus (including the various aspects described above).

  According to the damage detection apparatus and method of the present invention, by using continuous wavelet transform, frequency analysis of intensity curve data detected from diffusely reflected light from a product is lost (or rotated) by normal Fourier transform. This can be done while leaving information about the angle. For this reason, it is possible to evaluate deviations that are difficult to handle in the past, which cannot be found by analyzing the intensity curve data using either time (or rotation angle) or frequency as a variable, so that damage can be detected more accurately. . In particular, by preparing reference intensity curve data in advance, it is possible to accurately discriminate and detect variations caused by variations between individuals that occur during machining and damage.

It is a table | surface which shows the item specification of the gear which is a target spine of the damage detection apparatus which concerns on this invention. It is a perspective view which shows typically the mode of the gearwheel whose damage is detected by the damage detection apparatus which concerns on this invention. It is a block diagram which shows the whole structure of the damage detection apparatus which concerns on this invention. It is the intensity curve data (FIG. 4A) acquired by the intensity curve data acquisition unit and the continuous wavelet transform result (FIG. 4B). It is the graph which showed the continuous wavelet transformation result shown in FIG.4 (b) in the three-dimensional coordinate system. FIG. 6 (a) is a graph showing the intensity curve data corresponding to the damaged tooth surface and the reference intensity curve data superimposed, and FIG. 6 (b) shows the result of the continuous wavelet transform. It is a graph which shows simply the deviation between intensity curve data and standard intensity curve data. It is an example of the deviation calculated by the deviation calculation part 20 based on the intensity | strength curve data and reference | standard intensity | strength curve data shown in FIG.

  Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the drawings. However, the dimensions, materials, shapes, relative arrangements, and the like of the components described in this embodiment are not intended to limit the scope of the present invention unless otherwise specified, but are merely illustrative examples. Not too much.

  In this embodiment, a gear 1 having a plurality of tooth surfaces on a rotating surface will be described as an example of a target product for detecting the presence or absence of damage by the damage detection apparatus according to the present invention. In particular, in this embodiment, a “helical gear” having a plurality of substantially identical teeth is handled as a gear to be detected, and detailed specifications of the gear are shown in FIG. Needless to say, the helical gear is an example of a detection target and is not limited thereto.

  During the driving of the gear, the tooth surface is brought into contact with (engaged with) the tooth surface of another gear, and thereby the driving force is transmitted. At this time, in the tooth surface of the gear, along with its rotational drive, the surface pressure increases in the vicinity of the pitch circle where it makes rolling contact with the receiving portion of the tooth surface, and the load is repeatedly applied as the gear is driven. Nearby materials can cause fatigue failure and damage (also known as “pitching”). Further, at the gear manufacturing stage, it is assumed that the tooth surface is damaged for some reason during machining, and such defective products are actually generated at a certain rate. In the present embodiment, detection of the presence or absence of various kinds of damage occurring on the tooth surfaces of such gears will be described.

  FIG. 2 is a perspective view schematically showing how damage is detected by the damage detection apparatus 10 according to the present invention. As shown in FIG. 2, the damage detection device 10 uses a laser displacement meter 11 attached to the main body of the damage detection device 10 with respect to the tooth surface of the gear rotating at a predetermined cycle to provide a single wavelength with high directivity. In addition to irradiating the laser beam, which is a light source, and receiving the reflected light from the tooth surface, optically detecting damage. When the tooth surface of the gear is ideal (that is, when the surface has no irregularities formed by machining during manufacturing), when the laser beam is irradiated from a specific direction as shown in FIG. 2, the reflection angle corresponding to the incident angle Regularly reflected light is generated in the direction. However, since the actual tooth surface has fine irregularities formed by machining (cutting or the like) at the time of manufacture, a part of the incident angle is irregularly reflected by the delicate irregularities.

  Such irregularly reflected light is particularly likely to occur in a region where the tooth surface is damaged. For this reason, if the tooth surface is damaged, the proportion of diffusely reflected light in the entire reflected light increases. That is, when there is damage, diffusely reflected light having a large light intensity is observed. The damage detection apparatus 10 according to the present invention can determine the presence or absence of damage with high accuracy by directly detecting the presence or absence of damage by acquiring and analyzing the light intensity of such irregularly reflected light. .

  The gear to be detected is rotationally driven by an actuator (not shown), and is arranged so that the laser light emitted from the fixed damage detection device 10 can scan the tooth surface of the rotating gear. ing. The rotation period of the gear can be arbitrarily set by controlling the actuator, and the rotation angle at each time can be measured. Preferably, the damage detection apparatus 10 is mounted at a position where the irradiated laser beam can be scanned evenly on the tooth surface 2 of the gear 1, and the laser beam is long in the tooth line direction of the tooth surface, particularly preferably, It is even better to irradiate laser light that is approximately as long as the tooth width.

  FIG. 3 is a block diagram showing the configuration of the damage detection apparatus 10 according to the present invention. First, as described with reference to FIG. 2, the damage detection apparatus 10 irradiates the tooth surface of the gear with laser light from the irradiation unit 11 a of the laser displacement meter 11, and irregular reflection from the tooth surface at the light receiving unit 11 b. Receives light. The received irregularly reflected light is photoelectrically converted to create a voltage signal. At this time, according to the research of the present inventor, it has been clarified that the created voltage signal includes an unnecessary signal component due to noise or disturbance on the low frequency side. Therefore, by providing the high-pass filter 12, these unnecessary low-frequency components are removed, thereby improving the damage determination accuracy. Thereafter, the voltage signal is sent to the intensity curve data acquisition unit 13 and detected as intensity curve data. In the intensity curve data acquisition unit 13 of the present embodiment, the intensity curve data is taken in as a voltage signal having a predetermined voltage waveform obtained by photoelectrically converting diffusely reflected light. However, for example, a current waveform may be used. It is not limited.

  The intensity curve data acquired by the intensity curve data acquisition unit 13 is stored in the storage unit 15 provided in the storage unit 14. Such storage operation in the storage unit 15 is performed every time the intensity curve data acquisition unit 13 acquires the intensity curve data, and the past intensity curve data is stored. The intensity curve data acquisition frequency of the intensity curve data acquisition unit 13 may be constant or indefinite, and accordingly, storage in the storage unit 15 is also performed as needed.

Subsequently, the reference intensity curve data calculation unit 16 accesses the storage unit 15 to acquire the accumulated past intensity curve data, and calculates the reference intensity curve data by averaging the acquired intensity curve data. To do. That is, if the accumulated intensity curve data is xi (t), the reference intensity curve data xsi (t) is expressed by the following equation: xsi (t) = Σxi (t) / i (1)
Is calculated by In this way, by calculating the reference strength curve data by averaging the accumulated strength curve data, highly reliable reference strength curve data can be obtained based on the past operation history. Can be detected.

  In the present embodiment, the reference curve data is calculated by averaging the accumulated intensity curve data. Instead of this, reliable reference intensity curve data for a normal product that is not damaged in advance is used. It may be prepared theoretically or experimentally and stored in a storage means such as a memory so that it can be read out.

  Subsequently, the continuous wavelet transform unit 17 performs continuous wavelet transform on the intensity curve data acquired by the intensity curve data acquisition unit 13 and the reference intensity curve data calculated by the reference intensity curve data calculation unit 16, respectively. As described above, the intensity curve data and the reference intensity curve data have the rotation angle (or time) as a variable, and are converted so that the frequency and the rotation angle (or time) are variables by continuous wavelet transform.

  Particularly in this embodiment, a plurality of types of mother wavelet functions used for continuous wavelet transformation are prepared in the mother wavelet function storage unit 18 of the storage unit 14 in advance. These prepared plural types of mother wavelet functions can be specified by the operator via the input unit 19 such as a keyboard, a mouse, or a touch panel. Generally, the waveform of the intensity curve data detected varies depending on the type of damage. In the present embodiment, the detection accuracy can be further improved by making it possible to specify an optimal mother wavelet that matches the waveform of the intensity curve data to be converted as the mother wavelet function used for continuous wavelet conversion.

  FIG. 4 shows the intensity curve data (FIG. 4A) acquired by the intensity curve data acquisition unit 13 and the continuous wavelet transform result (FIG. 4B). Here, the horizontal axis indicates the rotation angle of the gear, and the vertical axis in FIG. 4 (a) indicates the voltage value obtained by photoelectrically converting the light intensity of the received irregularly reflected light, and in FIG. 4 (b). The vertical axis represents the frequency. FIG. 4 shows a waveform corresponding to irregularly reflected light from a plurality of tooth surfaces of the gear, and in this example, data in the case where the tooth surfaces are not particularly damaged is shown. As described above, since there are fine irregularities and accuracy errors formed during machining as described above, even in the intensity curve data shown in FIG. 4, the data waveform obtained for each tooth surface varies. .

  FIG. 5 is a graph schematically showing the result of the continuous wavelet transform shown in FIG. 4B in a three-dimensional coordinate system. As shown in FIG. 5, the result of the continuous wavelet transform is expressed in contour lines, and the region where a signal is generated on the xy plane (x: rotation angle, y: frequency) of the contour map is the same specification without scratches. It has been clarified by the inventors' research that the two gears are almost identical. In the example of FIG. 5, since there is no abnormality (that is, a normal product) as a gear to be detected, the wavelet transform results of the tooth surfaces are almost overlapped. For this reason, the size of the z-axis (intensity) varies slightly depending on each tooth, but the three-dimensional shape representing the wavelet transform result is almost the same.

  Here, FIG. 6A is a graph showing the superposition of the intensity curve data acquired from the damaged tooth surface and the reference intensity curve data, and FIG. 6B shows the continuous wavelet transform result. In FIG. 6, the reference intensity curve data is indicated by a solid line, and the intensity curve data is indicated by a broken line.

  As shown in FIG. 6A, the intensity curve data before the continuous wavelet transform has a periodicity corresponding to the rotation period of the gear, and basically has two variables (horizontal axis x: rotation angle (or time). ), The vertical axis y: voltage value). As compared with the reference intensity curve data, a difference occurs only in a narrow area on the horizontal axis (refer to an area indicated as “damage range” in FIG. 6). Here, as described above, each tooth surface has variations due to individual differences such as processing tolerances, and unevenness existing on the surface is also slightly different. Therefore, even if there is no abnormality in any tooth surface, there are not a few differences in the intensity curve data, and whether or not the slight difference as shown in FIG. 6A is due to damage. It is difficult to evaluate.

  On the other hand, in FIG. 6B, the intensity curve data and the reference intensity curve data are expressed by three variables (x: rotation angle or time, y: frequency, z: intensity) by continuous wavelet transform. As a result, as shown in FIG. 6B, a large deviation appears between the intensity curve data and the reference intensity curve data over a wider rotation angle region.

  FIG. 7 is a graph schematically showing a deviation between the intensity curve data and the reference intensity curve data. The calculated continuous wavelet transform results are distributed in a contour line as shown in FIG. When the waveforms corresponding to the presence / absence of damage are superimposed and displayed, as shown in FIG. 7 (b), the change in the waveform due to the presence / absence of damage becomes clear, and the intensity curve data as shown in FIG. 7 (c). And the deviation between the reference intensity curve data becomes clear. When continuous wavelet transformation is performed in this way, if a deviation from a steady pattern occurs, the deviation is greatly reflected in the transformation result. Therefore, the deviation of the intensity curve data acquired from the damaged tooth surface can be largely extracted from the reference intensity curve data corresponding to the tooth surface without damage.

  Returning to FIG. 3 again, such a deviation calculation is performed by the deviation calculating unit 20. The deviation calculation unit 20 calculates the deviation between the continuous wavelet transform results between the intensity curve data calculated by the continuous wavelet transform unit 17 and the reference intensity curve data. Here, FIG. 8 is an example of the deviation calculated by the deviation calculating unit 20 based on the intensity curve data and the reference intensity curve data shown in FIG.

  And the damage detection part 21 detects the presence or absence of damage by determining whether the deviation calculated by the deviation calculation part 20 is larger than a predetermined value. This predetermined value is set in advance by an experimental value, a theoretical value, or the like for cases where damage is present and not present. In this example, the predetermined value is in the vicinity of 2 to 2.5 kHz according to the research of the present inventors so far.

  More preferably, the damage detection unit 21 may determine that there is damage in a region where there is a deviation larger than a predetermined value, and determine which region of the tooth surface is damaged. As shown in FIG. 8, since a rotation angle having a deviation larger than a predetermined value is found, a rotation angle from a reference value such as an initial value of the gear rotation angle is calculated and geometrically analyzed. Thus, the damaged area can be specified. This geometric analysis depends on measurement conditions such as an angle at which laser light is irradiated, as shown in FIG.

  Then, the detection result (the presence / absence or area of damage) in the damage detection unit 21 is output to the result output unit 22. The result output unit 22 may be a display screen such as a liquid crystal display or a plasma display, or may be notification means for generating a warning sound or the like.

  As described above, according to the damage detection apparatus 10 according to the present embodiment, the frequency analysis of the intensity curve data detected from the irregularly reflected light from the product is lost in the normal Fourier transform by using the continuous wavelet transform. This can be done while leaving information on the time (or the rotation angle). For this reason, it is possible to evaluate deviations that are difficult to handle in the past, which cannot be found by analyzing the intensity curve data using either time (or rotation angle) or frequency as a variable, so that damage can be detected more accurately. . In particular, by preparing reference strength curve data in advance, it is possible to accurately discriminate and detect normal variations caused by processing tolerances and assembly tolerances and abnormal shapes caused by damage.

  In addition, such a damage detection device is attached to the tooth surface of a gear already in operation such as in an apparatus so that the tooth surface can be irradiated with laser light. Detection may be performed.

  In addition, in the unmanned vehicle that works by repeatedly traveling the same traveling route like AGV (automated transport vehicle) and unmanned forklift, the obstacle sensor that detects the obstacle by capturing the landscape with images etc. The landscape acquired by these obstacle sensors usually has a steady pattern or tendency, but it may be applied to detect anomalies when this changes with high sensitivity by continuous wavelet transform. As such an application method, for example, it can be similarly applied to the operation control of an industrial robot that repeats the same work repeatedly.

  The present invention is applicable to a damage detection method and apparatus for detecting the presence or absence of damage in a product by irradiating the surface of the product rotating at a predetermined cycle with laser light and measuring the intensity of the irregularly reflected light.

DESCRIPTION OF SYMBOLS 10 Damage detection apparatus 11 Laser displacement meter 11a Irradiation part 11b Light reception part 12 High pass filter 13 Intensity curve data acquisition part 14 Storage part 15 Storage part 16 Reference | standard intensity | strength curve data acquisition part 17 Conversion part 18 Mother wavelet function storage part 19 Input part 20 Deviation Calculation unit 21 Damage detection unit 22 Result output unit

Claims (7)

In a damage detection device that detects the presence or absence of damage by irradiating the surface of a product rotating at a predetermined cycle with laser light and measuring the intensity of the irregularly reflected light,
An intensity curve data detector that detects the intensity of the irregularly reflected light as intensity curve data with a rotation angle or time as a variable;
A reference intensity curve data acquisition unit for acquiring reference intensity curve data which is intensity curve data obtained when the product is normal;
A continuous wavelet transform unit that continuously transforms the intensity curve data detected by the intensity curve data detection unit and the reference intensity curve data acquired by the reference intensity curve data acquisition unit, respectively;
A deviation calculating unit for calculating a deviation between the conversion results in the continuous wavelet transform unit;
A damage detection apparatus comprising: a damage detection unit that detects whether or not there is damage by determining whether or not the deviation calculated by the deviation calculation unit is greater than a predetermined value.   The said damage detection part specifies the area | region where damage exists among the said products by calculating the variable area | region where the deviation calculated by the said deviation calculation part is larger than the said predetermined value, The Claim 1 characterized by the above-mentioned. The damage detection device described. A storage unit for storing the intensity curve data detected by the intensity curve data detection unit;
The damage detection apparatus according to claim 1, wherein the reference intensity curve data acquisition unit calculates the reference intensity curve data by averaging the intensity curve data accumulated in the accumulation unit. .   4. An input unit for selecting a mother wavelet function used for continuous wavelet transformation in the continuous wavelet transformation unit from a plurality of types of mother wavelet functions stored in advance. The damage detection apparatus according to claim 1.   The damage according to any one of claims 1 to 4, wherein the intensity curve data detection unit detects the intensity of the irregularly reflected light through a high-pass filter for removing a component having a frequency equal to or lower than a predetermined frequency. Detection device. The product is a gear;
The intensity curve data detection unit receives irregularly reflected light obtained by irradiating a tooth surface of the gear with laser light, and detects intensity curve data based on a temporal change in intensity of the received irregularly reflected light. The damage detection device according to any one of claims 1 to 5, wherein In the damage detection method for detecting the presence or absence of damage by irradiating the surface of the product rotating at a predetermined cycle with laser light and measuring the intensity of the irregularly reflected light,
An intensity curve data detection step of detecting the intensity of the irregularly reflected light as intensity curve data with a rotation angle or time as a variable;
A reference intensity curve data acquisition step for acquiring reference intensity curve data which is intensity curve data obtained when the product is normal;
A continuous wavelet transform step for performing continuous wavelet transform on the detected intensity curve data and the acquired reference intensity curve data, respectively;
A deviation calculating step of calculating a deviation between the conversion results in the continuous wavelet conversion step;
A damage detection method comprising: a damage detection step of detecting the presence or absence of damage by determining whether or not the calculated deviation is greater than a predetermined value.