JP2004325438A - Test state evaluating method, testing machine, and test condition evaluating method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To evaluate the state of machining and mechanical test, the state of machining such as cutting, grinding, and other energy working, the state of a testing machine using a standard substance or material, and the state of a workpiece using a standard testing machine, by performing successive observation/image recording on the same place and image analysis. <P>SOLUTION: In this test state evaluating method, observation/image recording is intermittently performed on changes in the surface state of a test/evaluation object under test/evaluation, and image analysis is performed to extract its characteristics. With respect to the result of observation/image recording and image analysis/characteristic extraction performed by an actual machine and the result on the same items obtained by the testing machine, comparison is performed by finding a change in mutual relation about the characteristics of at least one image or about time change of the characteristics. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a method for evaluating a test state in cutting, grinding, and other energy processing, a method for evaluating a test state for friction and wear, and a method for evaluating characteristics of a processing machine and a test machine reflecting the evaluation method.

Conventionally, the surface of a brake material during friction has been directly observed by irradiating the surface of the brake material with a strobe light every time a certain position is reached on the surface of the brake material (see Patent Document 1).
In addition, the surface of the roll being processed is irradiated with strobe light and light of a specific wavelength to calculate the surface irregularity pattern, obtain the measurement results, and adjust the surface processing conditions based on the results to improve processing accuracy and product yield. It is also known to improve (see Patent Document 2).
Furthermore, using an in-situ observation system for a sliding friction experiment, a time-series change in the form of wear has been observed (see Non-Patent Document 1). In addition, a method of numerically analyzing the form of wear powder generated due to friction and a calculation result thereof have been reported (see Non-Patent Document 2).

JP-A-56-14529 JP 01-162584 A Teruko Kato, Kazuo Horikirikawa, Tsutomu Fukuda, Tadashi Shinooka, Hitoshi Takahashi: Elucidation of sliding wear mechanism in one-sided contact between block and plate, Tribologist, 42,6 (1997) 454 Sugimura, J., Umeda, A., Yamamoto, Y .: Application of Neural Network to Shape Detection of Wear Powder, Transactions of the Japan Society of Mechanical Engineers (C), 61,590 (1995) 4055

However, in the method of directly evaluating the state of processing and test, only evaluation depending on a specific processing method and test method, processing conditions and test conditions, and materials used can be performed. For this reason, comparison of characteristics between test machines and evaluation of the machines themselves could not be performed unless specific processing and tests were actually performed. In general, the results of evaluating the material on an actual machine and the results of testing on a test machine did not agree due to the interaction of complex parameters such as differences in operating conditions and equipment characteristics.
Until now, methods and mechanical (additional) devices for observing a moving solid surface by a strobe method (including a laser strobe method) have been proposed. However, a certain minute area of the solid surface is recorded and accumulated as a still image. There is no way to continue. Analysis of acquired images and still images of other moving objects focuses on pattern recognition for extracting errors, and there is no example of application to processing or test evaluation. In image analysis, there is no analysis example that focuses on shape analysis and extracts features of state change and feature extraction of time-series change and evaluation using these features.
In the present invention, when a new material or a new test device is developed, an accelerated test is performed using a test device that simulates an actual machine. In examining the cause of the discrepancy, we noticed the need to record, accumulate, and analyze changes over time on the machined and test surfaces.
In order to realize this, we deal with still image shooting (strobe shooting) technology using illumination synchronized with motion, high-magnification observation technology for microscopic surface analysis, and feature extraction accompanying the temporal change of the acquired image. The need for analysis has arisen.

In order to achieve the above object, the present invention relates to a state of machining and mechanical testing, a state of machining such as cutting, grinding and other energy processing, a state of a processing machine using standard substances and materials, and a standard processing machine. The state of the workpiece using is evaluated by performing continuous image observation and image recording at the same place, and image analysis.
For example, the evaluation of a test performed before using the machine element material in an actual machine is performed. By extracting the features of the image appearing in the actual machine or processing, the features of the image appearing in the actual machine or processing are compared with the features of the image in the test machine, and the test machine is evaluated.
The state of the processing and the test is evaluated from the change of the feature of the temporal continuous image obtained by the processing and the test.
The state of processing and testing is evaluated from the processing method and test method, the processing conditions and test conditions, and the characteristics of images that do not depend on the test material used.
Using a system that illuminates in synchronization with the movement of processing and testing, and uses a system that forms an image of the observation surface with a CCD camera, the system continuously observes and records images at the same location based on the recorded image. Then, computer processing is performed to extract features of the image and changes over time of the features, and the state of processing and testing is evaluated by comparison. The features that appear in a test machine that imitates the processing and machine conditions actually used are compared with the features that appear in other test equipment, and the other test machines are evaluated.
That is, the present invention relates to a method for intermittently observing and recording images at the same location of a test object, performing image analysis, and performing feature extraction on the surface state change of the test object during test evaluation. By comparing and comparing the results of observation / image recording and image analysis / feature extraction on a machine with the results of the same item on a test machine, at least one image feature or the temporal change of the feature. A method for evaluating the test condition was found.

In the prior art, evaluation of processing and testing was not only possible without actual processing and testing, but was not possible to compare between devices, so it was difficult to evaluate devices. In the present invention, the state of machining or mechanical testing, or the state of machining such as cutting, grinding, or other energy processing, is continuously observed and image-recorded at the same location, and the state is evaluated by image analysis. The evaluation of the test to be performed before using the machine element material in an actual machine is performed. By extracting the features of the image appearing in the actual machine or processing, the features of the image appearing in the actual machine or processing are compared with the features of the image in the testing machine, and the testing machine is evaluated. The state of the processing and the test is evaluated from the change of the feature of the temporal continuous image obtained by the processing and the test. The state of processing and testing is evaluated from the processing method and test method, the processing conditions and test conditions, and the characteristics of images that do not depend on the test material used. Using a system that illuminates in synchronization with the processing and testing movements and uses a CCD camera to form an image of the observation surface, a method of continuously observing and recording images at the same location over time, based on the recorded images Then, computer processing is performed to extract features of the image and changes over time of the features, and the state of processing and testing is evaluated by comparison. The features that appear in a test machine that imitates the processing and machine conditions actually used are compared with the features that appear in other test devices, and the other test devices are evaluated. For this reason, it is applied to the machinery field, precision machining field, cutting / grinding / energy processing field, tribology field, machinery / friction testing machine, etc., also to material testing, observation and analysis of high-speed moving particles, etc. Adaptable. Other tests with time change can be evaluated. Evaluate electrical and mechanical unevenness of mechanical devices and changes in the state of contact surfaces. It can also be applied to optimization of cutting tool shape, selection of lubricant, and defect inspection in joining and cutting.

このような特徴比較を、実際の機械及び複数の試験機において行い、特徴マッチングの工程において、それぞれを比較する。 FIG. 1 shows an example embodying the present invention. The material 20 fixed to the test machine reciprocates or rotates. A test is performed by pressing a pin 15 with a predetermined load 10 applied thereto. Although the type of the test apparatus is not limited, securing a space for mounting the microscope 45 and the trigger sensor 25 is an essential condition. In order to synchronize the pulse light from the laser oscillator 55 and the shutter of the CCD camera 70 with the movement of the material 20, the signal from the trigger sensor 25 that detects the movement of the material 20 is received by the pulse generator 35 via the cable 30 and the laser A signal considering the jitter of the oscillator 55 is sent to the laser oscillator 55 via the cable 80, and a signal considering the shutter opening / closing time and the opening time of the CCD camera 70 is sent to the CCD camera 70 via the cable 75. Light emitted from the laser oscillator 55 passes through the optical fiber 50, enters the microscope 45, and is collected as probe light 40 at a fixed position on the friction surface. The light reflected on the processing surface returns to the microscope 45 through the same optical path, but is separated from the probe light and enters the CCD camera 70 to form an image. The image enters the computer 60 through the cable 65 as an electric signal, and is sequentially recorded and stored as a still image. After the test is completed, the recorded / stored image group is transferred to the analysis computer 90, and after performing position correction and brightness correction, processing for feature extraction is performed. This processing can be shown, for example, in Table 1 below.

Such feature comparison is performed in an actual machine and a plurality of test machines, and each is compared in a feature matching process.

The present invention will be described in more detail with reference to examples, but the present invention is not limited to these examples.
(Example 1)
2-1 Laser strobe applied friction surface observation system The configuration of the friction surface observation system is introduced. The feature of this system is that the probe light necessary for observing the friction surface uses the 2nd harmonic (wavelength: 532 nm, pulse width: 5 nsec) of pulse YAG. This light is introduced into the microscope, the light reflected from the friction surface is returned to the microscope, and the image is captured by a CCD camera into a computer. The trigger signal is taken from the movement of the test piece and sent to the laser oscillator and CCD camera via the pulse generator. This allows the laser to emit light in conjunction with the movement of the test piece, opening the CCD camera shutter. Laser light is transmitted by an optical fiber. After the light exits the microscope objective lens, the light is applied to the friction surface directly or through a mirror from the load direction (perpendicular to the specimen movement direction), and the light reflected from the friction surface returns to the microscope and reaches the CCD camera I do. In this system, the observation magnification is variable from 5 times to 250 times by replacing the microscope.
2-2 Experimental conditions A high-speed reciprocating friction tester was used as a friction tester. The reciprocating speed of the test piece is 600 rpm (10 Hz) and the stroke is 50 mm. The shape of the test piece is a block of 14x17x70mm, and the shape of the mating member is a cylinder having a diameter of 8mm and a length of 23mm. SS400, FC250 and S45C were used as a pair for each material. The test piece was finish-polished with emery paper, ultrasonically washed three times with a mixed solution of acetone and petroleum benzine, and stored in a desiccator. In the friction test, 25 ml of lubricating oil was used. After performing a running-in operation at 60 rpm for 10 minutes under a predetermined load, the rotation speed was increased to 600 rpm, and the image was taken in and the friction coefficient was recorded. The test time was 30 minutes, and during this period, images were continuously recorded and accumulated at 10 Hz at 256 gradations of 640 × 480 pixels. This is equivalent to 18000 still images. Considering the difference in the friction surface after 30 minutes, the load was set to 3 conditions of 24.5, 49, 98N for SS400 and S45C, and 3 conditions of 49, 73.5, 98N for FC25. The magnification of the image of the observation result was 5 times or 35 times.

(Example 2)
An example of a result when a friction test is performed on a test piece of FC250 with a pin of FC250 under a load of 98 N is shown as an example of the result. FIG. 2 shows the specimen surface after the test for 30 minutes.
The test piece was severely worn, and streaks and irregularities were formed. Some of the results of observing the process leading to such abrasion mode at low magnification are shown in FIGS. 3 (a) to 3 (f) in the order of elapsed time from the start of the test. FIG. 3 (a) shows the test piece movement direction during image capture, and FIG. 3 (f) shows the scale. Characteristic streaks are outlined in each image.

Immediately after the start of the friction test (FIG. 3 (a)), it is recognized that countless lines exist parallel to the direction of movement. Relatively characteristic lines ([1] and [2] in Fig. 3 (a)) were recognized as wide and clear lines in the image (Fig. 3 (b)) after 170 seconds. Also, a new line indicated by [3] appears in this figure. Immediately after this, the friction surface changed significantly.
An unprecedented line appears in FIG. 3 (c). Thereafter, the friction surface did not change significantly until the end of the test (FIG. 3 (f): after 1800 seconds). In FIG. 3 (c) to FIG. 3 (f), a common mark is attached to a point or a line to be noted ([4], [5], and [6]). First, at point [4] in FIG. 3 (c), at this point only a few thin lines can be observed. In FIG. 3 (d) after 5 seconds, a black shadow having a size of about 1 mm is observed at this point. The black shadow on the screen is where the reflected light is less than the surroundings, and one of the causes may be a depression formed on the friction surface. This shadow could be observed in FIG. 3 (e), but was hardly recognized in FIG. 3 (f).
The lines [5] and [6] in FIG. 3 (c) appear brighter, contrary to [4]. This may be because, for example, the light is reflected more than its surroundings and is higher or flatter than its surroundings. In this case as well, it changed over time, as in the point [4], and exhibited behaviors such as widening and continuous lines.
It is also important to compare the captured images one by one, but it is also important to track and analyze changes in the overall image and characteristic changes that appear in the images. Therefore, paying attention to the initial changes of these images, an average image of the initial 50 images is obtained by using 180 images every 10 seconds out of the 2000 images from 20 to 200 seconds where a large change is recognized. FIG. 4 shows the result of calculating the cross-correlation between the reference images.

The following equation was used to calculate the cross-correlation.

Equation 1

ただし、ｒ_xyは相互相関係数、Sは分散を示しそれぞれ次式で表される。

Here, r _xy indicates the cross-correlation coefficient, and S indicates the variance, each of which is represented by the following equation.

Equation 2

ここで、Sxは平均画像の各ピクセルに対応した階調列の分散、Syは各画像の各ピクセルに対応した階調列の分散、Sxyは平均画像と各画像の各ピクセルに対応した階調列の共分散を示す。
この相関係数計算式を用いて画像を解析した結果が図４である。図４(上段)は、実画像の相関を示している。この相関係数の変化を見ると安定した変化を示した後、ゆっくりとした相関係数の低下があり、急激な低下へと発展した。これは摩擦係数の変化と極めて類似の変化を示している。
相関係数の変化からも、摩擦開始から175秒後の画像に現れた変化がはっきりと認められる。30秒後までの相関が悪い原因は平均画像がぼやけた画像であったためで、本質的ではないと考えられる。注目すべき点は、100秒後付近から相関が少しずつ落ち始め170秒後付近で大きく落ち込む点である。図3(a) と図3(b)の変化はわずかではあるが、[1]や[2]の線がシャープになり、[3]の線が現れるなどの変化があった。これらの変化が相関の緩やかな低下につながったと考えられる。
一方、図４(下段)は、同じ画像群をそれぞれフーリエ変換し周期性の変化を調べた結果である。フーリエ変換後の実部の相関係数変化を「○」で、虚部を「X」でプロットして示した。相関係数の変化は、実部と虚部、さらには実画像の係数変化ともほぼ同じであるが、フーリエ変換後の相関係数の時間変化には、実画像の相関係数の時間変化で見られたゆっくりとした低下の間に、急激な相関係数の低下が認められ、この間に画像の周期性に乱れがあったことがわかった。

Here, Sx is the variance of the gradation sequence corresponding to each pixel of the average image, Sy is the variance of the gradation sequence corresponding to each pixel of each image, and Sxy is the gradation corresponding to each pixel of the average image and each image. Show column covariance.
FIG. 4 shows the result of analyzing the image using the correlation coefficient calculation formula. FIG. 4 (upper part) shows the correlation between real images. Looking at the change in the correlation coefficient, the correlation coefficient showed a stable change, and then there was a slow decrease in the correlation coefficient, which led to a sharp decrease. This shows a change very similar to the change in the coefficient of friction.
The change in the correlation coefficient also clearly shows the change that appeared in the image 175 seconds after the start of friction. The cause of the poor correlation up to 30 seconds later is that the average image was a blurred image, which is considered to be not essential. The point to be noted is that the correlation starts to decrease little by little after about 100 seconds, and falls greatly around 170 seconds later. Although the changes in FIGS. 3 (a) and 3 (b) were slight, there were changes such as the lines [1] and [2] being sharpened and the line [3] appearing. It is thought that these changes led to a gradual decrease in correlation.
On the other hand, FIG. 4 (lower part) shows the result of examining a change in periodicity by Fourier transforming the same image group. The change in the correlation coefficient of the real part after the Fourier transform is plotted with “○”, and the imaginary part is plotted with “X”. The change of the correlation coefficient is almost the same as the change of the real part and the imaginary part, and also the change of the coefficient of the real image. During the slow decrease observed, a sharp decrease in the correlation coefficient was observed, indicating that the periodicity of the image was disturbed during this time.

1 is an outline of a laser strobe applied friction surface observation system used in the embodiment of the present invention. The surface photograph of the friction test piece using FC250. 2 is an image of a friction surface of FC250 captured by the observation system used in the present invention. FIG. 4 is a diagram illustrating a correlation change of a friction surface image.

Claims (5)

Observation and image recording on the actual machine in the method of intermittently observing and recording the surface condition of the DUT during test evaluation at the same location of the DUT, analyzing the image, and extracting features And a method of evaluating a test state by obtaining a change of a correlation between at least one image feature or a time change of the feature and comparing the result of the image analysis and feature extraction with the result of the same item in the tester. 2. A method for evaluating a testing machine, wherein the testing method according to claim 1 is performed by a plurality of testing machines, and a test machine having a change in mutual relationship selected from the plurality of testing machines. 3. The method according to claim 2, wherein the feature of the image is a periodicity of the intensity of the brightness of at least one pixel or a change in the intensity. A method for evaluating test conditions, wherein the evaluation method according to claim 1 is performed under a plurality of test conditions, and a test condition having a change in mutual relationship is selected from the plurality of test conditions. The method of evaluating test conditions according to claim 4, wherein the feature of the image is a periodicity of a change in brightness or a change in brightness of at least one pixel.

Images