JP2010272250A - Automatic check method for appearance defect of continuous porous electrode base material and winding body of porous electrode base material with its recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an automatic inspection method for appearance defects capable of accurately and effectively performing a continuous inspection of various appearance defects when continuously transferring the porous electrode base material of a long size and a rolled shape, and to provide a winding body of the porous electrode base material capable of immediately fixing inspection results and defective positions. <P>SOLUTION: In the automatic inspection method for appearance defects, inspection light is irradiated on the surface of the porous electrode base material (1), and the transmitted light, specular reflection light, and dispersed light are imaged. The imaged data are analyzed with an image processing section (4), and a recording medium is continuously wound as kinds and existing positions of the defects are recorded on it. The recording medium where the inspection results based on the automatic inspection of appearance defects are recorded is added on the wound winding body. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention automatically inspects the appearance defect of a long continuous porous electrode substrate made of short carbon fibers and carbonized resin, and in particular, the inspection data of the defect portion based on the continuous appearance inspection automatically. The present invention relates to a wound body of a porous electrode substrate having a recorded recording medium.

  The porous electrode base material normally constitutes two electrodes, a hydrogen electrode (fuel electrode) and an oxygen electrode (air electrode), in a polymer electrolyte fuel cell, each of which is interposed between the separator and the catalyst layer It is. A polymer electric field membrane (ion exchange resin membrane) is disposed between the catalyst layers, and these constituent members are integrated to constitute a cell. The porous electrode base material not only functions as an electric conductor between the separator and the catalyst layer, but also functions to distribute a gas such as hydrogen and oxygen supplied from each separator to the catalyst layer, and is generated in the catalyst layer. It is required to have a function of absorbing water and discharging it to the outside. At present, carbonaceous materials are generally effective.

  Conventionally, in order to increase the mechanical strength, methods such as tightly binding short carbon fibers and resin carbide have been adopted. The power generation performance when assembled is often greatly reduced. On the other hand, if the carbon short fiber and the resin carbide are too rough to maintain a high gas permeability, the mechanical strength decreases, and there are various ways to handle it not only when it is incorporated into the fuel cell but also at normal times. Will be subject to various restrictions.

  For example, Japanese Unexamined Patent Application Publication No. 2006-40885 (Patent Document 1) and Japanese Unexamined Patent Application Publication No. 2006-40886 (Patent Document 2) disclose a porous electrode base material that solves such problems. In these porous electrode base materials disclosed in Patent Documents 1 and 2, carbon short fibers dispersed in a random direction in a substantially two-dimensional plane are bound together by an amorphous resin carbide, and the carbon short The fibers are crosslinked with filamentous resin carbide. The carbon short fiber has a fiber diameter of 3 to 9 μm, a fiber length of 2 to 12 mm, and a porous electrode substrate made of a very thin film material having a thickness of 150 μm or less. A porous electrode base material having such a structure is made by making a carbon fiber paper by a wet or dry papermaking method from the papermaking material containing the carbon short fibers, and impregnating and solidifying the carbon fiber paper with a thermosetting resin, It is continuously manufactured by firing in an inert gas atmosphere. The long porous electrode substrate thus obtained is flexible enough to be wound around a paper tube of 3 × 2.54 cm or less, is thin and inexpensive, and has gas permeability and bending strength. Is excellent.

  By the way, these porous electrode base materials are made by impregnating the above-mentioned carbon fiber paper with a thermosetting resin, and then heating and pressurizing to smooth the surface, and then firing and heating in an inert gas atmosphere in a firing furnace. Manufactured by carbonizing the curable resin, but during the production, for example, carbon fibers do not disperse and harden in a bundle, and it remains partially, or impurities that accumulate in the firing furnace and float Due to various causes such as adhesion of impurities or non-uniform impregnation amount of the thermosetting resin, various defects that adversely affect the originally required function of the porous electrode substrate are generated.

  Therefore, in the past, appearance inspection has been performed in order to find various types of defects generated in the porous electrode substrate. However, this inspection is usually made visually. Even in this visual defect inspection, it is possible to scrutinize over time if it is a plate-like porous electrode base material manufactured by, for example, a batch method, but it is continuously transferred as described above. When targeting a long and extremely thin porous electrode base material, the inspection object itself is black, and the defect color is often black, and there is a slight difference in color and gloss on the surface. Therefore, it is not easy to accurately identify these differences, and it is often impossible to inspect these defects.

  On the other hand, for example, Japanese Patent Application Laid-Open No. 2008-89534 (Patent Document 3) proposes a continuous inspection method relating to an opening ratio of a carbon fiber fabric that is continuously transferred. However, the carbon fiber fabric to be inspected is a woven fabric having a structure in which warp and weft intersect each other, and imaging data by transmitted light is used for the woven fabric, and the opening ratio of the woven fabric and the width of the carbon fiber yarn are used. In addition, the arrangement density of the yarns, the skew of the yarns, and the like are subjected to arithmetic processing, and an attempt is made to automatically determine whether or not the fabric is defective based on the arithmetic results.

  However, since the numerical value itself and the numerical range of the inspection target in each specific inspection item according to Patent Document 3 are significantly larger than that of the porous electrode base material, the inspection accuracy is expected to be very high. Absent. Incidentally, paragraph [0099] of Patent Document 3 gives examples of a fabric to be inspected whose shape information is known or a master workpiece of a 12K woven fabric and a 3K woven fabric that are determined to be inspected. The opening width is described as 1.0 mm × 1.0 mm and 0.4 mm × 0.4 mm. On the other hand, the average radius of the pores in Examples 1 to 5 described in Patent Document 2 is 8 μm to 11 μm, and even if they are compared, there is a difference of 50 to 1000 times or more. If there is such a large difference, the carbon fiber fabric continuous inspection method proposed by Patent Document 3 is equivalent to the appearance defect inspection method of the porous electrode substrate described in Patent Documents 1 and 2. Even if the former continuous inspection method is applied to that of the school building, it is impossible to expect a highly accurate inspection result.

  On the other hand, for example, Japanese Patent Laying-Open No. 2005-265467 (Patent Document 4) discloses a thin stain on the surface of an object to be inspected such as a metal plate, film, paper, nonwoven fabric, resin plate, and glass plate that continuously travels on a production line. In addition, a defect detection apparatus that automatically detects low contrast defects such as spots and shallow scratches has been proposed. According to this defect detection apparatus, the reflected or transmitted light of the inspection light irradiated on the surface of the inspection object is used to capture and generate image data, and the image data is based on the pixel density value. To detect defects on the surface of the inspection object. Image processing at this time is performed by integrating the density value of each pixel in the defect detection block constituted by a plurality of adjacent pixels to calculate an integrated value, and comparing the integrated value with a predetermined threshold value. In detecting a defect, the defect detection block is set so as to partially overlap an adjacent defect detection block so that even a low-contrast defect can be detected with high accuracy.

  However, since the surface appearance defect inspection methods disclosed in Patent Documents 3 and 4 are both inspections by a single inspection means (imaging using transmitted light or reflected light), a single type of defect is detected. Although effective if inspected, it is impossible to simultaneously inspect various types of defects having different forms, colors and dimensions with high accuracy. Furthermore, especially for a long and thin porous electrode substrate wound in a roll shape as disclosed in Patent Documents 1 and 2, when cutting into a predetermined dimension and assembling it into a cell, The location of the various types of defects as described above cannot be immediately determined on the spot, which may cause various problems in the subsequent assembly work, and inevitably leads to defects in the final fuel cell. It might be.

JP 2006-40885 A JP 2006-40886 A JP 2008-89534 A JP 2005-265467 A

  In order to solve the above-mentioned problems, the present invention can efficiently and efficiently eliminate various types of appearance defects during continuous transfer of a long, roll-shaped porous electrode substrate that is not marked. In providing an automatic inspection method for appearance defects that can be automatically inspected continuously, and in assembling a fuel cell using the porous electrode substrate, the porous electrode substrate is wound in a wound state at the assembly site. It is an object of the present invention to provide a wound body of a porous electrode base material that can immediately determine the type of defect of a material, its existence position, the inspection result of the size, and the manufacturing history.

  A first main configuration of the present invention is an automatic appearance defect inspection method for a long porous electrode base material composed of carbon short fibers and a carbonized resin that are continuously running. The surface is irradiated with light, and at least the transmitted light, specular reflected light and scattered light are imaged. The image data is analyzed by the image processing unit, and the type, location and size of the defect are recorded on the recording medium. In addition, the present invention is an automatic appearance defect inspection method for a porous electrode base material, which includes continuous winding.

  Further, the second main configuration of the present invention is a carbon short fiber having a recording medium in which the inspection result by the appearance defect automatic inspection method for the porous electrode substrate which is the first main configuration is recorded. It exists in the roll of the elongate porous electrode base material which consists of carbonized resin. Here, it is desirable that the types of defects include at least black defects, white defects, gloss defects, fiber bundle defects, resin shortage defects, and pinhole defects.

  According to the automatic continuous defect inspection method for a porous electrode substrate of the present invention, the defect inspection is continuously inspected by employing at least three kinds of inspection means of transmitted light, specular reflection light and scattered light. Even if it is a defect that occurs in an ultra-thin porous electrode substrate that cannot be detected by normal visual inspection, the use of these inspection means makes it possible to identify the types of defects and many types of defects. Changes in color tone, shape, dimensions, and luminance can be determined simultaneously and accurately, enabling highly reliable and highly accurate inspection. In addition, if a recording medium, for example, a recording paper or a recording chip, on which the data of the inspection result is recorded, is attached to the wound body of the porous electrode base material of the present invention, the type of the defective portion that affects the performance, Since the location and size can be easily identified by recording the recording medium, even in the assembly site of the fuel cell, when assembling the cell by cutting it into a predetermined dimension while rewinding the wound body of the porous electrode substrate, As a result, it is possible to eliminate the defective parts that can be known by the recording, and it is easy to assemble high-performance and high-quality cells.

BRIEF DESCRIPTION OF THE DRAWINGS It is the typical example of an apparatus for enforcing the external appearance defect automatic inspection method of the elongate porous electrode base material which runs continuously of this invention, and its process explanatory drawing. It is explanatory drawing which shows the suitable example of arrangement | positioning of three types of optical inspection means employ | adopted in the said external appearance defect automatic inspection method. It is a block diagram explaining the schematic structure of the image processing part of the said external appearance defect automatic inspection apparatus. It is a judgment table showing the inspection test result obtained by the appearance defect automatic inspection method.

First, prior to describing a preferred embodiment of the method for automatically inspecting appearance defects of a long porous electrode substrate that continuously runs according to the present invention, a long porous film that can be wound up, which is an inspection object of the present invention. The electrode substrate will be briefly described.
This porous electrode substrate is manufactured by the manufacturing method disclosed in Patent Document 1. That is, the porous electrode base material is a fibril shape having a freeness of 400 to 900 ml except for carbon short fibers having a fiber diameter of 3 to 9 μm and fiber fibers dispersed in a random direction in a substantially two-dimensional plane. It is manufactured by impregnating a carbon fiber paper made of a material with a resin and then carbonizing the resin. As the carbon short fiber of the porous electrode base material, any of carbon fiber, pitch-based carbon fiber, rayon-based carbon fiber, and the like is used. In addition, from the viewpoint of production cost of carbon short fibers, dispersibility, and smoothness of the final porous carbon electrode substrate, the diameter of the carbon short fibers is set to 3 to 9 μm, and the fiber length of the carbon short fibers is combined with the resin. From the standpoint of wearability and dispersibility, the thickness is 2 to 12 mm.

  Here, “substantially dispersed in a random direction in a two-dimensional plane” means that the carbon short fibers lie so as to form a single surface, and the resin includes carbon such as phenol resin. Those having strong binding strength with fibers and a large residual weight during carbonization are preferred. This resin carbide has a resin carbide content of 25 to 40 mass% when the porous electrode substrate is taken as 100 mass%. The short carbon fibers are bound together with an amorphous resin carbide.

  Amorphous resin carbide has a net shape with a minimum fiber diameter of 3 μm or less, and short carbon fibers and short carbon fibers are cross-linked with a net resin carbide to mix small holes with a diameter of about 2 μm and large holes with a diameter of about 50 μm. is doing. By mixing the large and small pores in this way, the porous electrode base material is generated not only by the function of efficiently delivering the reaction gas to the reaction part (catalyst layer) but also by water contained in the reaction gas and power generation. It also has a function of efficiently discharging water. By setting the gas permeability of the porous electrode base material to 2000 m / sec / MPa or less, short carbon fibers that are difficult to crack even if the basis weight is small, and that extend in the thickness direction of the base material even if the bulk density is small. Is almost nonexistent. This porous electrode base material is flexible, can be wound around a paper tube having a diameter of 3 × 2.54 cm or less, can be made compact, and is easy to carry and handle. The further specific configuration of the porous electrode substrate is left to the description in Patent Document 1 above.

  Now, in order to carry out the appearance defect automatic inspection method of the present invention on a thin porous electrode substrate wound in a roll having the above-described configuration, as shown in FIG. The roll body 2 which is a wound body is supported so as to be rewound on the gantry 3, and the appearance defect is automatically detected by the inspection means described later while continuously pulling the porous electrode substrate 1 from the roll body 2, The detected data is analyzed in the image processing unit 4, and the radius result based on the analysis result is sequentially stored in a storage unit (not shown) in the computer 5. The porous electrode base material 1 for which the detection of the appearance defect has been completed is rewound into a roll shape by the winding unit 6.

  The basic method of the appearance defect inspection at this time is based on the method described in Patent Document 4 that can be detected with high accuracy even for a low-contrast defect. The most characteristic configuration of the visual defect automatic inspection method according to the present invention, which is different from the technique, is that the appearance defect inspection means is not a single kind as in the literature 4, but adopts three different specific inspection means, The final determination is made by combining the inspection results.

  This is because the porous electrode substrate to be inspected is very thin, so it is difficult to specify the color change and defect size, and the substrate itself is black, and most of the defective parts are also black, As disclosed in Patent Documents 3 and 4 above, the surface of the substrate is irradiated with illumination light, the reflected light or transmitted light is imaged with a single type of inspection means, and the imaging data is analyzed. Then, it is impossible to specify all types of defective portions of the porous electrode substrate having various types and various types of defects and simultaneously determine the degree of the defects.

Therefore, in the present invention, as described above, three kinds of different inspection means are adopted, and the overall determination is performed by combining the inspection results, so that it is impossible to visually inspect. For the first time, it is possible to accurately and accurately specify a change in color tone and a defect size of a porous electrode substrate that is extremely thin and includes other types of defects of various sizes.
In the present invention, the types of defects that appear on the surface of the porous electrode substrate as appearance defects include black defects, white defects, gloss defects, fiber bundles, resin shortages, pinholes, etc. Does not look normal.

Here, there are two types of defects in black defects.
One of them is (1) the black part formed by the impurities deposited in the carbonization furnace and floating impurities adhering to and reacting with the porous carbon electrode substrate, and the appearance is close to burnt. Second, (2) the surface shape of the carbonized resin adhered due to uneven dispersion of the papermaking binder is complicated and looks black because it scatters light well. However, although it is difficult to clearly distinguish between the two by visual inspection and optical inspection, according to the electron microscope, (1) the fibers are cut and the shearing debris is scattered, and (2) the fine mesh fibers are gathered. It is observed that

As with the black defect (2), the white defect is a portion where the carbonized resin ratio is extremely high due to uneven dispersion of the papermaking binder, but the surface shape is smooth unlike the black defect. Therefore, it looks white to reflect light well.
The gloss defect is observed when impurities such as graphite powder and graphite powder are mixed. It looks brighter than the white defect.

The above-mentioned fiber bundle defect occurs due to uneven dispersion between fibers at the paper making stage. It looks black due to the bundle of fibers.
The resin shortage occurs due to poor resin impregnation. It looks black because there is little resin.
Here, the black defect in the case of the fiber bundle defect has a length greater than 10 mm and the ratio of the length to the width is 4: 1 or more, and the black defect in the case of insufficient resin becomes less than 10 mm in length. ing.

  The pinhole defect, as in the case of the black defect (1), is formed by impurities deposited in the carbonization furnace and floating impurities adhering to and reacting with the porous carbon electrode substrate. The black defect (1) occurs when the scratch is shallow, and when the scratch becomes large, it becomes a pinhole.

  In the present invention, as described, three different types of detection means are employed as appearance defect inspection means. One of them is the illumination of illumination light having a required illuminance from one side of the porous electrode base material 1 and the transmitted light imaged by the first imaging device 7 on the other side. The second is imaging means, and the second is to illuminate illumination light having a required illuminance on one surface side of the porous electrode substrate 1 at a predetermined incident angle and to have a second imaging device 8 arranged on the regular reflection angle line. An imaging means for specularly reflected light for imaging specularly reflected light, and the third one is that illumination light having a predetermined illuminance is applied to one surface side of the porous electrode substrate 1 at a predetermined incident angle, and on the regular reflection angle line. The third imaging device 9 arranged on the reflection angle line deviating from the above is a scattered light imaging means for imaging the scattered light. These three types of imaging means are arranged in series along the traveling path of the porous electrode substrate 1 and the imaging data is sent to the image processing unit 4.

  In the case of the second imaging device 8, the irradiation angle of the inspection light and the light receiving angle of the second imaging device 8 are the same, and it is desirable to set the angle to 10 to 30 °. If the angle deviates from 10 to 30 °, the detection accuracy decreases. Since the third imaging device 9 uses scattered reflection, the light receiving angle of the third imaging device 9 is set to 10 to 50 °, and the angle difference between the irradiation angle of the inspection light and the light receiving angle of the third imaging device 9 is set. Is preferably 10 ° or more and 30 ° or less. When the angle is 10 ° or more and 30 ° or less, it is possible to determine the white defect with high accuracy.

Next, an automatic appearance defect inspection method for the porous electrode substrate 1 according to the present invention according to the illustrated embodiment will be specifically described with reference to the drawings.
FIG. 1 is a schematic block diagram showing a typical apparatus example for carrying out the automatic appearance defect detection method of the present invention. This automatic appearance defect detection apparatus corresponds to the first to third light sources 10 to 12 on the unwinding runway of the porous electrode substrate 1 on which the roll body 2 rotatably supported by the gantry 3 is unwound. The first to third imaging devices 7 to 9 are arranged in pairs. Here, the arrangement order of the three different types of imaging means is arbitrary, and is not limited to the above arrangement. Image data captured by the first to third imaging devices 7 to 9 is subjected to image processing in the image processing unit 4.

  Each of the light sources 10 to 12 is constituted by a lighting device capable of generating linear inspection light such as a halogen lamp, an LED, or a fluorescent lamp arranged linearly across the traveling path of the porous electrode substrate 1. Is done. The first to third imaging devices 7 to 9 include an imaging device including a solid-state imaging element such as a CCD camera that converts imaged light into an electrical signal to generate image data. In the present embodiment, a plurality of first to third imaging devices 7 to 9 are linearly arranged in a direction crossing the unwinding running path of the porous electrode substrate 1, and the one-dimensional of the porous electrode substrate 1 that travels is arranged. A two-dimensional image is obtained by sequentially capturing images.

  The image processing unit 4 includes an arithmetic device (CPU), a storage device (memory), and the like, similar to the image processing device described in Patent Document 4. For this reason, the specific configuration of the image processing unit 4 and the processing procedure thereof are left to the detailed description of Patent Document 4 described above, and the following description is limited to a simple description. However, in the present invention, the arithmetic device (CPU), the storage device (memory), and the like described above are provided corresponding to each of the three types of imaging means.

  As shown in FIG. 3, the image processing unit 4 stores an image input unit 4a that converts analog image signals input from the imaging devices 7 to 9 into digital values, and a digital signal obtained by the image input unit 4a. It is used for correction in the line memory 4b, the pixel spot correction unit 4c that corrects pixel offset voltage variations and sensitivity spots of the line sensor included in the digital signal converted by the image input unit 4a, and the pixel spot correction unit 4c. Spot correction coefficient memory 4d that holds coefficients, and correction processing for low-frequency regions such as a decrease in the amount of light at the periphery of the lens and illumination spots included in the digital signal obtained by the image input unit 4a, and a substantially uniform brightness A background processing unit 4e that obtains image data of the line sensor and a background processing coefficient memory 4f that holds coefficients used for correction processing in the background processing unit 4e are provided.

According to the appearance defect automatic detection apparatus having the above-described configuration, first, the porous electrode base material 1 that continuously travels on the conveyance path at a predetermined speed is supplied to each of the first to third light sources 10 to 12. Each is irradiated with inspection light.
At this time, as shown in FIG. 2A, the first imaging device 7 and the first light source 10 place the porous electrode substrate 1 on a vertical plane orthogonal to the traveling surface of the porous electrode substrate 1. The first imaging device 7 images inspection light that is emitted from the first light source 10 and passes through the porous electrode substrate 1.

  On the other hand, the second and third imaging devices 8 and 9 and the second and third light sources 11 and 12 are respectively located above the porous electrode substrate 1 as shown in FIGS. The porous electrode substrate 1 is disposed at a predetermined interval in the traveling direction. Each inspection light emitted from each light source 11, 12 is directed to the surface of the porous electrode substrate 1, and each of the reflected lights is received by the second and third imaging devices 8, 9. Here, the incident angle of the inspection light irradiated from the second light source 11 with respect to the porous electrode substrate surface is the same as the light receiving angle of the second imaging device 8, but the inspection light irradiated from the third light source 12 is porous. The incident angle with respect to the surface of the electrode base material and the light receiving angle of the second imaging device 8 are different. In other words, the second imaging device 8 receives and images the specularly reflected light of the inspection light emitted from the second light source 11, and the third imaging device 9 receives the scattered light of the inspection light emitted from the third light source 12. And take an image.

  As described above, the images picked up by the three different kinds of image pickup means are converted into one-dimensional analog data by the first to third image pickup devices 7 to 9 respectively, and then input to the image processing unit 4. Sent to the unit 4a. In this image input unit 4a, the one-dimensional analog data sent from each of the imaging devices 7 to 9 is converted into digital data indicating the density value of each pixel. Next, this digital data is sent to the pixel spot correction unit 4c. One column in the horizontal direction and one row in the vertical direction of the image data indicating the density value input to the pixel spot correction unit 4c correspond to the density values acquired at a time by the line sensors of the imaging devices 7-9. .

  In a general CCD line sensor, in order to correct the sensitivity variation of pixels (light receiving elements), the average value of density values for each matrix of pixels arranged in a matrix and the average value of density values of pixels in the matrix immediately before that are arranged. The difference from the value is obtained and sequentially used as a spot correction coefficient, and the spot correction coefficient is stored in the spot correction coefficient memory 4d. The pixel spot correction unit 4c adds the spot correction coefficient cumulative value to the input density value to obtain spot corrected data. This corrected data reduces variations in the average value of density values between pixels.

  A background processing unit 4e that corrects a low frequency region such as a decrease in the amount of light at the periphery of the lens and a spot of illumination included in the digital signal obtained by the image input unit 4a, and obtains image data of a line sensor with substantially uniform brightness. Then, in order to eliminate the difference in the output level of the line sensor, the background correction coefficient is calculated and stored in the background correction coefficient memory 4e, and the input image data is multiplied by the background correction coefficient to obtain the output level of the line sensor. Counter the difference. When calculating the background correction coefficient, a smoothing process may be added as necessary in order to remove noise components due to formation or the like.

  Since the imaging data captured by the first to third imaging devices 7 to 9 includes a density value range determined to be not normal, even in this embodiment, it is described in Patent Document 4. As with the defect inspection apparatus, the pixel density limiting unit 4g that excludes values in the density range that is not normal is included, and a 3 pixel × 3 pixel defect detection block (pixel block) An integration operation for integrating the density values of each pixel for each block is performed. Hereinafter, a procedure for performing an integration operation for integrating the density values of each pixel for each block will be specifically described with reference to the procedure described in Patent Document 4.

  Now, if the imaging data of the normal part, the density total value obtained by integrating the density values in the pixel block (defect detection block) and the density total value obtained by integrating the density values in the pixel block including the defect are: , There is a gap in the value. In the block integration unit (not shown), the defect detection block is reliably detected by calculating the integration value while shifting the defect detection block pixel by pixel in the vertical (row direction) or horizontal (column direction). .

  In the vertical movement of the defect detection block, first, the image density for the first to third rows is stored in the image memory 4h by the pixel density limiting unit 4g, and the first to third rows for the block integrating unit 4i. The density values are integrated (step 1). Next, the image data of the fourth row is stored in the image memory 4h, and the density value of the fourth row is added to the integrated value of the first to third rows calculated in step 1 by the block integration unit 4i and the first row. Are integrated to obtain integrated values for three rows of the second to fourth rows (step 2). Further, the image data of the fifth row is stored in the image memory 4h, and the density value of the fifth row is added to the integrated value of the second to fourth rows calculated in step 2 by the block integrating unit 4i and the second row. Are integrated to obtain integrated values for the third to fifth rows (step 3). By repeating such processing, the defect detection block is shifted by one pixel in the vertical direction.

  To move the defect detection block in the horizontal (column) direction, first, image data for three columns of the first to third columns is stored in the image memory 4h, and the block accumulating unit 4i stores 3 of the first to third columns. The density values for the columns are integrated (step 10). Next, the imaging data of the fourth column is stored in the image memory 4h, and the density value of the fourth column is added to the integrated value of the first to third columns calculated in step 10 by the block integration unit 4i and the first value is added. The density values of the columns are subtracted to obtain integrated values for three columns of the second to fourth columns (step 12). Further, the imaging data of the fifth column is stored in the image memory 4h, and the density value of the fifth column is added to the integrated value of the second to fourth columns calculated in step 12 by the block integrating unit 4i and the second column. Are subtracted from each other to obtain integrated values for the third to fifth columns (step 13). By repeating such processing, the defect detection block is shifted by one pixel in the horizontal direction.

  In order to perform an operation on a defect detection block having a spread of 3 pixels in the pixel direction (column direction) of the line sensor and 3 pixels in the movement direction (row direction) of the inspection object S in the block integration unit 4i, 4h needs a capacity to store the number of density values necessary for the calculation. Further, when performing an arithmetic process between blocks adjacent in the moving direction as an emphasis process, a memory capacity corresponding to the block size is required.

  Integration data obtained by the block integration unit 4i is compared with a threshold value set in advance and stored in the threshold memory 4k in the determination processing unit 4j, and the presence or absence of a defect is determined. The threshold value for defect determination is sequentially updated in accordance with the change in formation of the inspection object. In order to enable such updating, the threshold memory 4k is configured to include two threshold holding areas, and one area is set as an area holding a threshold used by the determination processing unit 4j, and the other area is set as the other area. It is an area for holding a threshold value during updating. When the update of the threshold value is completed in the other region, the line for reading the threshold value by the determination processing unit 4k is switched to the other region side, and the defect detection determination is performed with the updated threshold value. With such a configuration, even if the density value of the normal part varies, the defect detection accuracy can be kept constant. Since the integrated value of the density value is obtained while moving the block vertically and horizontally by the block integrating unit 4i, when the defect size is large, a plurality of pieces of information on the same defect are output and the number of data is very large. To be more. For this reason, the aggregation processing unit 4l sets the aggregation range in advance, collects the detection results within the range, and outputs them from the data output unit 4m.

  FIG. 4 is a determination table showing the determination result of the porous electrode substrate determined by the above-described appearance defect automatic inspection method according to the present invention. As can be understood from this table, the inspection results differ depending on the type of inspection means. This is not effective when the inspection object is likely to have other types of defects only by inspection by a single inspection means as in the appearance defect inspection disclosed in Patent Documents 3 and 4 above. This is recognized for the first time through various tests as in the present invention.

  Specifically, for example, when looking at the two types of black defects (1) and (2), even if the same black defect is detected, all of the inspections using specular reflection light are inspected using regular reflection light and scattered light inspection. While it is considered that there is a defect, it may be difficult to specify the type of the defect with scattered light, and no defect is detected by inspection with transmitted light. For example, for a white defect, a defect is discovered by inspection with specular reflection light and transmitted light, but the defect is overlooked by inspection with scattered light. Furthermore, in the case of a glossy defect similar to a white defect, it can be reliably detected by inspection with specular reflection light.

  Furthermore, as for the fiber bundle defect and the resin shortage defect, the resin is still insufficient in both cases, but as described above, it is necessary to determine based on the shape and dimensions of the imaging. As shown in FIG. 4, the resin-deficient defect is found by both inspection using specular reflection light and scattered light, but the fiber bundle defect is overlooked by the scattered light. Further, pinhole defects cannot be determined by inspection with reflected light, but are reliably detected with transmitted light.

In particular, in the appearance defect automatic inspection of the thin porous electrode base material to be inspected according to the present invention, as described above, by combining inspections with specularly reflected light, scattered light and transmitted light, a plurality of types are provided. Presence or absence of appearance defects can be accurately detected and determined, and high-precision inspection is realized.
In addition, in the automatic appearance defect inspection of the porous electrode base material according to the present invention, the record regarding the defect position is naturally left on the recording medium. As a method for detecting the traveling distance, the number of rotations of the shaft of the contact roll or the transfer roll contacting the traveling porous electrode substrate is detected by an encoder. In addition, the defect position in the width direction of the porous electrode base material is obtained by calculation from image data obtained by each image pickup device arranged linearly in the width direction.

DESCRIPTION OF SYMBOLS 1 Porous electrode base material 2 Roll body (of porous electrode base material) 3 Base 4 Image processing part 4a Image input part 4b Line memory 4c Pixel spot correction part 4d Spot correction coefficient memory 4e Background processing part 4f Background processing coefficient memory 4g Pixel density limiting unit 4h Image memory 4i Block integration unit 4j Determination processing unit 4k Threshold memory 4l Aggregation processing unit 4m Data output unit 5 Computer 6 Winding unit 7-9 First to third imaging devices 10-12 First to third light source

Claims (3)

A method for automatically inspecting the appearance of a long porous electrode substrate made of short carbon fibers and carbonized resin and continuously running,
The surface of the porous electrode substrate is irradiated with inspection light, the transmitted light, specularly reflected light, and scattered light are imaged, and the image data is analyzed by the image processing unit, and the type of defect, the position of the defect, and A method for automatically inspecting an appearance defect of a porous electrode substrate, comprising continuously winding the size while recording the size on a recording medium.   A wound body of a long porous electrode base material comprising a carbon short fiber and a carbonized resin, comprising a recording medium on which an inspection result by the appearance defect automatic inspection method according to claim 1 is recorded.   The wound body of the porous electrode substrate according to claim 2, wherein the types of defects include black defects, white defects, gloss defects, fiber bundle defects, resin shortage defects, and pinhole defects.

Images