JP2009002889A - Method, device, and program for compensating errors - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method for compensating motion errors at high precision. <P>SOLUTION: In a method for eliminating measurement errors due to motion errors of a moving section 26 by comparing a measured value signal obtained by measuring the shape of a measuring workpiece 40 by a calibration value signal obtained in advance by measuring the shape of a reference workpiece 32, the calibration value signal is obtained by eliminating a reference workpiece shape component from the output signal of the detector 36 when the shape of the reference workpiece 32 is measured while the moving section 26 is moved relative to the stationary section 30. The measurement value signal is obtained by eliminating a measuring workpiece shape component from the output signal of the detector 36 when the shape of the measuring workpiece 40 is measured while the moving section 26 is moved relative to the stationary section 30. The error compensating method contains a process (S10) of adaptively eliminating, as a motion error component, a component having the highest shape correlation with the calibration value signal among components contained in the measurement value signal by using the calibration value signal. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an error correction method, an error correction apparatus, and an error correction program, and more particularly to consideration of environmental changes between calibration and measurement.

Conventionally, in order to measure the shape, a shape measuring machine such as a contour shape measuring machine has been used.
By the way, in a shape measuring machine such as a contour shape measuring machine, it is necessary to correct the straightness of the measurement axis (see, for example, Patent Documents 1 to 4).
For this reason, conventionally, correction is performed as follows. That is, a calibration measurement is performed in advance using a standard shape such as a known shape such as a high-precision optical flat to obtain a calibration value. Next, measurement is performed using the measurement workpiece to obtain a measurement value. The calibration value at the time of calibration was directly subtracted from the obtained measured value (see, for example, Patent Documents 1 to 3).
Further, conventionally, a surface roughness measuring instrument has also been used to measure the value of the average interval of unevenness on the surface of a workpiece (see, for example, Patent Documents 4 to 5).
Japanese Utility Model Publication No. 61-30814 JP-A-2-75905 Japanese Patent Laid-Open No. 11-118473 JP 2004-325120 A Japanese Patent Laid-Open No. 8-313248 (Japanese Patent No. 3539695)

However, although there is still room for improvement in the correction accuracy even in the conventional method, conventionally, there has been no appropriate technique that can solve this.
The present invention has been made in view of the above-described problems of the prior art, and an object thereof is to provide an error correction device, an error correction method, and an error correction program capable of performing motion error correction with high accuracy.

As a result of intensive studies on the motion error correction by the present inventor, it has been clarified that the cause of the high accuracy that has been unknown in the past is the environmental change between calibration and measurement.
In other words, in actual measurement, the measurement environment is different from the calibration environment due to various factors such as the repeatability of the measuring device itself and the deviation from the calibration value due to the difference in material between the reference workpiece and the measurement workpiece. It was found that there is a limit to the correction capability in the process of simply subtracting the calibration value from.
Based on the discovery that the cause that hinders high accuracy is the change in environment between calibration and measurement, the present inventor has performed the following adaptive processing to perform measurement closer to the true value. The inventors have found that a value can be obtained and have completed the present invention.

An error correction method, that is, an error correction method according to the present invention for achieving the above object, uses a calibration value signal obtained by measuring the shape of a reference workpiece whose shape is known in advance, and the moving portion is moved relative to the fixed portion. In an error correction method for removing a measurement error caused by a movement error of the moving part from a measurement value signal obtained by measuring the shape of a measurement workpiece with a detector while performing relative movement,
The calibration value signal is obtained by removing the shape component of the reference workpiece from the output signal of the detector when measuring the shape of a reference workpiece with a known shape while moving the moving portion relative to the fixed portion during calibration. Is,
The measurement value signal is obtained from the output signal of the detector when the shape of the measurement workpiece is measured while moving the moving portion relative to the fixed portion at the time when the time has elapsed since the calibration. The shape component is removed,
A motion error component removing step is provided.
Here, the motion error component removal step uses the calibration value signal to adaptively remove a component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal as a motion error component. .

The movement error component removal according to the present invention means, for example, the following processing is performed in order to appropriately follow the change even if the frequency or amplitude of the movement error component included in the measurement value signal changes (varies) with time. .
(1) Among the components included in the measurement value signal from the shape measurement start position to the end position, the component having the highest shape correlation with the calibration value signal is estimated as the measurement-time movement error component.
(2) Subtracting from the measurement value signal a value obtained by adjusting the frequency or amplitude of the calibration value signal so as to approximate the shape of the estimated motion error component.

In the present invention, it is particularly preferable that the motion error component removal step adaptively removes the motion error component from the measurement value signal using the calibration value signal by an adaptive noise canceller.
As the adaptive noise canceller of the present invention, for example, the one described in Japanese Patent Application Laid-Open No. 5-191882 can be used.

<Removal of thermal noise component>
Moreover, in this invention, it is suitable to provide a thermal noise component removal process.
Here, the thermal noise component removal step adaptively removes the thermal noise component from the components included in the measurement value signal after the motion error component removal step.
In the present invention, it is particularly preferable that the thermal noise component removal step adaptively removes the thermal noise component from the measurement value signal after the motion error component removal step by an adaptive line enhancer.
As the adaptive line enhancer of the present invention, for example, those described in JP-A-2006-266758 can be used.

<Straightness>
In the present invention, the motion error is particularly preferably the straightness of the measurement axis of the shape measuring machine.
<Shape component>
In the present invention, it is particularly preferable to include a shape component removal step and a shape component recovery step.

Here, the shape component removal step is provided before the movement error component removal step, and from the output signal of the detector when measuring the shape of the workpiece while moving the movement portion relative to the fixed portion, The workpiece shape component is removed.
In the shape component recovery step, the workpiece shape removed in the shape component removal step in the measurement value signal from which the error component has been removed before obtaining a correction value from the measurement value signal from which the error component has been removed. Restore ingredients.

In order to achieve the above object, the error correction apparatus according to the present invention uses a calibration value signal obtained by measuring a reference workpiece having a known shape in advance to move the moving part relative to the fixed part. In the motion error correction device that removes the measurement error caused by the motion error of the moving part from the measurement value signal obtained by measuring the shape of the measurement workpiece with the detector while moving,
The calibration value signal is obtained by removing the shape component of the reference workpiece from the output signal of the detector when measuring the shape of a reference workpiece with a known shape while moving the moving portion relative to the fixed portion during calibration. Is,
The measurement value signal is obtained from the output signal of the detector when the shape of the measurement workpiece is measured while moving the moving portion relative to the fixed portion at the time when the time has elapsed since the calibration. The shape component is removed,
A movement error component removing unit is provided.
Here, the motion error component removal means adaptively removes, as a motion error component, a component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal, using the calibration value signal. .

In order to achieve the above object, the error correction program according to the present invention is a calibration value signal obtained by causing a computer to execute a motion error component removal step and measuring a reference workpiece having a known shape in advance. Error correction to remove the measurement error caused by the movement error of the moving part from the measured value signal obtained by measuring the shape of the workpiece with the detector while moving the moving part relative to the fixed part In the program
The calibration value signal is obtained by removing the shape component of the reference workpiece from the output signal of the detector when measuring the shape of a reference workpiece with a known shape while moving the moving portion relative to the fixed portion during calibration. Is,
The measurement value signal is obtained from the output signal of the detector when the shape of the measurement workpiece is measured while moving the moving portion relative to the fixed portion at the time when the time has elapsed since the calibration. The shape component is removed,
The motion error component removing step uses the calibration value signal to adaptively remove, as a motion error component, a component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal. And

Error Correction Method According to the error correction method of the present invention, the method includes a motion error component removal step that adaptively removes a motion error component from the measurement value signal using the calibration value signal. Can be performed with high accuracy.
In the present invention, the motion error component removal step adaptively removes the motion error component from the measurement value signal using the calibration value signal by the adaptive noise filter, thereby performing the motion error correction with higher accuracy. Can be done.

In the present invention, the correction can be performed with higher accuracy by combining a thermal noise component removing step that adaptively removes a thermal noise component from the measurement value signal.
In the present invention, the thermal noise component removal step adaptively removes the thermal noise component from the measurement value signal by an adaptive line enhancer, whereby the correction can be performed with higher accuracy.
In the present invention, since the motion error is straightness, straightness correction can be performed with high accuracy.
In the present invention, the correction can be performed with higher accuracy by including the shape component removal step and the shape component recovery step.

In addition, according to the error correction apparatus of the present invention, the apparatus includes the motion error component removing unit that adaptively removes the motion error component from the measurement value signal using the calibration value signal. The accuracy of error correction can be increased.
Error Correction Program According to the error correction program according to the present invention, since the motion error component removing step is executed by the computer, the motion error correction can be performed with high accuracy.

Hereinafter, a preferred embodiment of the present invention will be described with reference to the drawings.
FIG. 1 shows a schematic configuration of an error correction apparatus according to an embodiment of the present invention.
In the present embodiment, an example will be described in which a measurement error caused by the straightness of the measurement axis of the contour shape measuring machine is adaptively corrected as a motion error.
A straightness correction device (error correction device) 10 shown in FIG. 1 includes a computer, and includes a straightness component removal unit (motion error component removal unit) 12, a thermal noise component removal unit 14, and a storage unit 16. The straightness component removing unit 12 and the thermal noise component removing unit 14 are connected in cascade.

The straightness component removal unit 12 performs a straightness component removal step (S10). In other words, the straightness component removing unit 12 uses a calibration value signal (reference signal) obtained in advance to determine the component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal. It is removed adaptively as a (motion error component).
In this embodiment, the signal can be processed by converting it into the frequency domain, but it is particularly preferable to process the signal in the shape (time / space domain). That is, even if the frequency or amplitude of the motion error component included in the measurement value signal changes (varies) with time, the straightness component can be estimated and removed more appropriately following the change. Because.
The thermal noise component removing unit 14 performs a thermal noise component removing step (S12). That is, the thermal noise component removing unit 14 adaptively removes the thermal noise component from the components included in the measurement value signal after the straightness component removing step (S10).

Further, the straightness correction apparatus 10 shown in the figure further includes a calibration value signal acquisition unit 20 and a measurement value signal acquisition unit 22.
Here, the calibration value signal acquisition means 20 performs a calibration value signal acquisition step (S14). That is, the calibration value signal acquisition unit 20 moves the stylus arm (movement unit) 26 in the X-axis direction by the X-axis drive unit 24 at the time of calibration, so that the table (fixed unit) 30 of the contour shape measuring machine 28 is moved. From the output signal of the detector 36 when the contour shape of the reference workpiece 32 is measured while moving the trace position by the stylus 34 on the reference workpiece (optical flat) 32 placed on the reference workpiece 32, the reference shape of the reference workpiece 32 is measured. The component is removed and the calibration value signal is obtained.

  Moreover, the measured value signal acquisition means 22 performs a measured value signal acquisition step (S16). The measurement value signal acquisition unit 22 includes a shape component removal unit 38. The shape component removing unit 38 performs a shape component removing step (S18). That is, the measurement value signal acquisition unit 22 moves the stylus arm 26 in the X-axis direction by the X-axis drive unit 24 on the table 30 of the contour shape measuring machine 28 during the measurement after the calibration. The shape component of the measurement workpiece 40 is removed from the output signal of the detector 36 when the contour shape of the measurement workpiece 40 is measured while moving the trace position by the stylus 34 on the measurement workpiece 40 placed, and the measurement is performed. A value signal is obtained.

Further, the straightness correction apparatus 10 shown in the figure further includes a shape component recovery means 42.
Here, the shape component recovery means 42 performs the shape component recovery step (S20). That is, the shape component recovery means 42 removes the measurement value signal from which the error component has been removed by the shape component removal means 38 before obtaining the correction value signal (correction value) from the measurement value signal from which the error component has been removed. The workpiece shape component is recovered.
The storage unit 16 stores a straightness correction program (error correction program) 50. In the present embodiment, the straightness correction program (computer) 10 is converted into a calibration value signal acquisition unit 20, a measurement value signal acquisition unit 22, a straightness component removal unit 12, and a thermal noise component removal unit 14 by a straightness correction program 50. The shape component removing unit 38 and the shape component restoring unit 42 function.

The straightness correction apparatus 10 according to the present embodiment is configured as described above, and the operation thereof will be described below.
In the present embodiment, since each of the means is provided, a straightness component is selected from the components included in the measurement value signal and having the strongest shape correlation with the calibration value signal using the calibration value signal obtained in advance. And then the remaining thermal noise component can be adaptively removed.
That is, in the present embodiment, the straightness component removing unit 12 can adaptively remove the straightness component from the measurement value signal by using the calibration value signal obtained at the time of calibration in advance.
In the present embodiment, the thermal noise component removing unit 14 can adaptively remove the thermal noise component from the measurement value signal after the straightness component removal.

As described above, according to the straightness correction apparatus 10 according to the present embodiment, the straightness component removing unit 12 that adaptively removes the straightness component from the measurement value signal using the calibration value signal, and the straightness component is removed. And a thermal noise component removing means 14 for adaptively removing the thermal noise component from the measured value signal.
As a result, in this embodiment, even if the frequency or amplitude of the straightness component included in the measurement value signal changes with time, straightness component removal that appropriately follows the change can be performed, so that the contour shape measurement is performed. The straightness of the measurement axis of the machine 20 can be corrected with higher accuracy.
In the present embodiment, more accurate correction can be realized by combining straightness component removal and thermal noise component removal that has not been conventionally corrected.
In the present embodiment, the straightness component removing unit 12 and the thermal noise component removing unit 14 are connected in cascade, so that the synergistic effect of the correction can be obtained.

By the way, in this embodiment, in order to perform the correction with higher accuracy, selection of a specific correction means is also very important. For this reason, in the present embodiment, in order to correct the straightness of the measurement axis of the shape measuring machine with higher accuracy, a correction unit as shown in FIG. 2 is selected from among a number of correction units. .
That is, in the figure, as the straightness component removal step, the straightness component is adaptively removed from the measurement value signal by the adaptive noise canceller 60 using the calibration value signal. For this purpose, the adaptive noise canceller 60 includes a calculation means 62, an FIR filter (adaptive filter) 64, and a coefficient correction algorithm 66.

Further, as the thermal noise component removing step, the adaptive noise enhancer 70 adaptively removes the thermal noise component from the measurement value signal from which the straightness component has been removed. For this purpose, the adaptive line enhancer 70 includes a calculation means 72, an FIR filter (adaptive filter) 74, a coefficient correction algorithm 76, and a delay circuit 78.
As described above, in the present embodiment, as the straightness component removing unit 12 and the thermal noise component removing unit 14, an adaptive noise canceller is selected from among a number of correcting units for the purpose of achieving both high-precision correction processing and simplification of the configuration. By selecting a means of cascade connection between 60 and the adaptive line enhancer 70, straightness correction can be performed more accurately and easily in the contour shape measuring machine than in the case of using other devices.

By the way, in the straightness correction processing of the conventional method, a calibration value is obtained in advance by calibration measurement using a reference workpiece, and at the time of measurement, the calibration value previously obtained at the time of calibration is subtracted as it is from the measurement value. It was. That is, the straightness correction of the conventional method is based on the premise that the characteristics of the unknown path do not change between calibration and measurement.
However, in actual measurement, the straightness component included in the measured value signal is obtained by measuring the reference workpiece due to various factors, such as measuring the workpiece repeatability and the workpiece different from the reference workpiece material. There is no guarantee that the calibration values are always the same, and errors may occur due to environmental changes between calibration and measurement.

Here, in order to reduce an error due to a change in environment between calibration and measurement, if the straightness component is measured at the same time when measuring a measurement workpiece, there are the following problems.
That is, since the contour shape measuring machine has only one detector, it is difficult to measure the straightness component simultaneously. Providing a separate detector for simultaneously measuring the straightness component would complicate the configuration and could not be adopted as a solution.
On the other hand, in the present embodiment, since the shape of the reference workpiece is known, the calibration value signal obtained by the measurement using the reference workpiece is equivalent to acquiring the noise that has passed through the unknown path. I focused on that.

In the present embodiment, the calibration value signal obtained by subtracting the ideal flat shape (known shape) of the optical flat from the output signal of the detector at the time of calibration is approximately treated as a noise component, and adaptive as shown in FIG. Since the straightness component can be adaptively corrected by the noise canceller 60, more accurate correction processing can be performed.
That is, in the figure, the characteristics of the adaptive filter 64 are adapted even if the characteristics of the unknown path change between calibration and measurement, and the straightness calibration value (straightness component) included in the measurement value signal changes. Therefore, the change in straightness calibration value (straightness component) can be reliably followed, and a straightness correction value closer to the true value can be obtained.

<Removal and recovery of low frequency components>
In order to perform the straightness component removal more appropriately, removal and recovery of the workpiece shape component are very important.
That is, in the present embodiment, since a steady value is assumed in principle, it is very important to previously remove a work shape component that is a low frequency component (trend component) from the measurement value signal as shown in FIG. Is important to.

That is, in the present embodiment, an error value after collation with the design value is targeted in a highly accurate work whose design value is known.
However, even if the design value is unknown, if it can be considered as a steady value by removing the trend component by some method, it can be applied to high-precision workpieces whose design value is unknown. Is possible.
As an actual processing procedure, as shown in FIG. 5, pre-processing related to the torrent component and post-processing related to the torrent component are required as necessary.

That is, when the design value is used as shown in FIG. 5A, as the shape component removal step (S18), the best fit step (S22) with the design value and the design value component removal step (S24). ).
In FIG. 9A, the straightness component removal step (S10) and the thermal noise component removal step (S12) are performed after the best fit with the design value (S22) and the design value component removal step (S24) are completed. Do. Further, a design value component recovery step (shape component recovery step (S20)) is then performed.

When the design value is not used as shown in FIG. 5B, the shape component removal step (S18) includes a trend component estimation step (S26) and a trend component removal step (S28). Is preferred.
In FIG. 5B, the straightness component removal step (S10) and the thermal noise component removal step (S12) are performed after the trend component estimation step (S26) and the trend component removal step (S28) are completed. Further, a trend component recovery step (shape component recovery step (S20)) is then performed.

  As described above, in the present embodiment, the value including the trend component cannot be directly processed. However, as shown in the figure, the straightness component removal step (S10) and the thermal noise component removal step (S12) are performed. By adding a shape component removal step (S18) and a shape component recovery step (S20) before and after, the straightness component removal step (S10) and the thermal noise component removal step (S12) can be performed more appropriately. Therefore, the correction can be performed with higher accuracy.

<Removal of thermal noise component>
In the present embodiment, in order to further improve the accuracy of the correction, it is also very important to reduce the thermal noise component that has not been conventionally corrected.
That is, when high-accuracy correction processing is realized by removing the straightness component, ideally, the components remaining in the measurement value signal after the motion error correction include the surface property (surface roughness and waviness) component of the workpiece and the heat It becomes a noise component.
In the present embodiment, the surface property of the workpiece can be measured with high sensitivity by reducing a thermal noise component that has not been conventionally corrected.

Here, when a plurality of measurements are repeated and the average is taken, the thermal noise decreases as the number of repetitions increases due to the effect of integration. However, since repeated measurement directly increases the measurement time, it is not particularly advantageous for a contact-type shape measuring machine using a pivot stylus.
Therefore, in the present embodiment, the thermal noise component is adaptively removed by an adaptive line enhancer 70 as shown in FIG. The adaptive line enhancer 70 shown in the figure uses an autocorrelation characteristic that the surface property component of the components included in the measurement value signal has a correlation with itself but does not have a correlation with the thermal noise component. Since the filter coefficient (filter characteristic) of the filter can be adaptively changed with respect to the environmental change from the time of measurement and calibration, the thermal noise component can be reliably removed from the measurement value signal.

<Adaptive filter>
In the present embodiment, the selection of the configuration of the adaptive filter is also very important in order to further improve the accuracy of the correction.
Hereinafter, the adaptive filter will be described more specifically.
In the present embodiment, as the adaptive filters 64 and 74, it is very preferable to select an FIR filter as shown in FIG. 7 having excellent filter characteristic stability.

In the present embodiment, in order to further improve the accuracy of the correction, it is very important to select an algorithm for updating the filter coefficient when the adaptive value processing is performed by the adaptive filters 64 and 74.
Therefore, in the present embodiment, a sequential least square (RLS) algorithm is adopted from among algorithms having a number of filter coefficient updates in order to further improve the accuracy of correction.
In other words, the least mean square (LMS) algorithm updates the coefficient based on the instantaneous value, whereas the RLS method updates the coefficient based on the measured value signal up to a certain time, so that the sensitivity to the abnormal value is low. It can be suppressed. The RLS method is advantageous from the viewpoint of convergence characteristics.
As described above, in the present embodiment, the filter coefficients of the adaptive filters 64 and 74 are updated based on the measured values up to a certain time using the successive least squares (RLS) algorithm. Can be achieved.

<Initialization process>
In the present embodiment, it is also very important to perform filter coefficient initialization processing in the adaptive filters 64 and 74 in order to further improve the accuracy of correction.
That is, an initial value is required for updating the filter coefficient. Usually, since the initial value of the coefficient is set as w _i = 0 (i = 0, 1,..., P−1), an indefinite interval occurs until the update of the filter coefficient sufficiently converges.
On the other hand, in the present embodiment, it is not always necessary to synchronize with the sampling at the time of measurement, and input values having a certain number of samples or more are buffered in advance. Thereafter, processing of the measurement value signal is started. If this coefficient value is adopted as the initial value, the coefficient can be updated in accordance with the change of the value from the converged state, so that the indefinite state at the value start position can be eliminated.

  As described above, according to the straightness correction apparatus according to the present embodiment, since the straightness correction processing is performed in consideration of the environmental change between the calibration and the measurement, the conventional method, that is, the environment between the calibration and the measurement is performed. Therefore, the straightness correction can be performed with higher accuracy compared to the case where the calibration value at the time of calibration is simply subtracted as it is from the measured value.

Experimentally FIGS. 8 to 13 are shown comparison results in the case of using the case of using the straightness correction in consideration of environmental changes according to this embodiment, the straightness correction is not considered conventional environmental changes .
In this experiment, an optical flat and a gauge block were measured with a shape measuring machine.
In this experiment, the optical flat data and gauge block data measured by the shape measuring machine are values obtained in the middle of improving the X-axis motion accuracy (guide mechanism) on the device side. The trend component (shape component) included in the original value has been removed in advance.
In the conventional method, the calibration value obtained at the time of calibration is directly subtracted from the measurement value at the time of measurement to obtain a correction value.
All values are measured for the same part of the same workpiece a plurality of times, one of which is considered as a calibration measurement and the other as an actual measurement. For this reason, it is an ideal correction state that all the values after correction become zero.

<Straightness correction>
FIG. 8 shows the correction results when straightness component removal by the adaptive noise canceller according to the present embodiment and straightness component removal of the conventional method are used.
In the figure, the gauge block was measured with a shape measuring machine. (A) is a calibration value, (B) is a measurement value, (C) is a conventional method, and (D) is an adaptive noise canceller (number of taps of an FIR filter) according to the present embodiment. Is the case using 3).
In FIG. 6C showing the conventional method, periodicity is seen in the error waveform, whereas in FIG. 9D showing this embodiment, random behavior is shown. This indicates that the error amount of the adaptive noise canceller output waveform of the present embodiment can be further reduced as compared with the output waveform of the conventional method when the thermal noise component removal is considered after the adaptive noise canceller.

<Removal of thermal noise component>
FIG. 9 shows the result of the thermal noise component removal process when the adaptive noise canceller output according to the present embodiment is used and when the correction value of the conventional method is used.
In the figure, the gauge block was measured with a shape measuring machine. FIG. 6A shows a case where the correction value obtained by the conventional method is an adaptive line enhancer input, and FIG. 6B shows an adaptive line enhancer when the adaptive noise canceller output of this embodiment is an adaptive line enhancer input. Is the output.
In the figure, the number of taps of the FIR filter mounted on the adaptive line enhancer is 3, and the number of delay elements of the delay circuit is 2.
As shown in FIG. 5B, the present embodiment arranges the thermal noise component removal after the adaptive noise canceller, so that the correction effect of the adaptive noise canceller is compared with FIG. It turns out that it can extract more effectively.

<Initialization of adaptive filter coefficients>
Next, the effect of the adaptive filter coefficient initialization processing is shown in FIG.
In the figure, the optical flat was measured with a shape measuring machine. FIG. 4A shows the result when the filter coefficient pre-optimization process is not performed, and FIG. 4B shows the result when the filter coefficient pre-optimization process is performed in the present embodiment.
As is clear from FIG. 6A, the initial value is set to w _i = 0 (i = 0, 1, 1) without performing special processing for coefficient initialization even though the number of taps of the filter is as small as three. .., P-1), an insufficient convergence portion occurs at the processing start position. On the other hand, when the initialization process of this embodiment is introduced as shown in FIG. 5B, it can be seen that a stable result can be obtained from the process start position.

<Followability for gain change>
FIG. 11 shows a simulation result when it is assumed that the gain at the time of measurement is twice that at the time of calibration. FIG. 5A shows the case where the straightness correction of the conventional method is used, and FIG. 6B shows the case where the adaptive noise canceller of this embodiment is used.
In FIG. 6A showing the conventional method, an error component corresponding to the gain difference appears, whereas in FIG. 9B showing the adaptive noise canceller of the present embodiment, the correction process following the change in gain. Can be seen.

<Followability for positional deviation>
(1) Adaptive noise canceller FIG. 12 shows the result of the adaptive noise canceller following the positional deviation.
FIG. 5A shows the case where the conventional method is used, FIG. 5B shows the case where an adaptive noise canceller with 3 taps is used in the present embodiment, and FIG. 6C shows the adaptive noise canceller with 10 taps in this embodiment. It is a result at the time of using.
In the figure, it was assumed that the positioning at the time of measurement was shifted by 250 μm. The sampling pitch was 25 μm.
In this experiment, since measurement is performed at a sampling pitch of 25 μm, a deviation of 250 μm in positioning corresponds to 10 in the number of filter taps of the adaptive noise canceller. In this embodiment (B) showing an adaptive noise canceller with 3 taps in this embodiment, this deviation cannot be sufficiently followed, but in this figure (C) showing an adaptive noise canceller with 10 taps in this embodiment, this deviation. You can follow the amount.

(2) Adaptive Line Enhancer FIG. 13 shows the result of performing the adaptive line enhancer process on the correction value for positional deviation. FIG. 6A shows a case where adaptive line enhancer processing is performed on the correction value obtained by the conventional method. FIG. 5B shows the correction value obtained by the adaptive noise canceller (10 taps) of the present embodiment. In contrast, the adaptive line enhancer processing is performed.
In the figure, it was assumed that the positioning at the time of measurement was shifted by 250 μm. The sampling pitch was 25 μm. The number of delay elements of the adaptive line enhancer is 2, and the number of filter taps is 3.
According to the same figure (B) which shows this embodiment, compared with the same figure (A) which shows a conventional system, the outstanding correction effect by an adaptive noise canceller was confirmed.

<Arrangement of each means>
In this embodiment, as shown in FIG. 2, it is particularly preferable to arrange the adaptive line enhancer 70 at the subsequent stage of the adaptive noise canceller 60, but an arrangement as shown in FIG. 14 is also possible.
That is, in FIG. 2A, the adaptive line enhancer 70 is disposed in front of the adaptive noise canceller 60. Then, the calibration value signal and the measurement value signal are subjected to adaptive line enhancer processing to remove thermal noise in advance.
Also, in FIG. 5B, the adaptive line enhancer 70 is arranged at the front stage and the rear stage of the adaptive noise canceller 60. Then, the calibration value signal is subjected to adaptive line enhancer processing to remove thermal noise in advance.

In this way, depending on the nature of the signal, it is more effective to remove the thermal noise in advance by applying an adaptive line enhancer to the calibration value signal or measurement value signal, or using a matched filter instead of an adaptive line enhancer can reduce the error. There are cases where it can be dealt with. For this reason, in this embodiment, it is also possible to select the arrangement of the adaptive line enhancer and the adaptive noise canceller so that the optimum calibration can be performed according to the nature of the signal.
In addition, the present invention is particularly preferably applied to the straightness correction of the measurement axis of the contour shape measuring machine, but can be applied to other shape measuring machines and movement errors as long as the movement error correction is required. It is.

<Measurement device>
The present invention can also be applied to, for example, straightness error correction of a measuring axis of a precision measuring machine such as straightness, length, and angle of a linear motion of a linear motion mechanism of a roundness measuring machine.
<Motion error>
For example, it can be applied to a circular motion error of a pivot type stylus of a contour shape measuring machine. Further, the present invention can be applied to the rotational motion error of the mounting table of the mounting table rotation type roundness measuring machine and the rotational motion error of the detector rotating mechanism of the detector rotating type roundness measuring machine.

It is explanatory drawing of schematic structure of the straightness correction apparatus concerning one Embodiment of this invention. 5 is a specific configuration of a straightness component removing unit and a thermal noise component removing unit that are characteristic in an embodiment of the present invention. It is explanatory drawing of the straightness component removal means characteristic in one Embodiment of this invention. It is explanatory drawing of a shape component removal process in one Embodiment of this invention. It is a flowchart which shows the process sequence of the straightness correction method concerning one Embodiment of this invention. It is explanatory drawing of the characteristic thermal noise component removal means in one Embodiment of this invention. It is explanatory drawing of the characteristic adaptive filter in one Embodiment of this invention. It is comparison explanatory drawing of the correction result at the time of using the case of using one Embodiment of this invention, and a conventional system. It is comparison explanatory drawing of the adaptive line enhancer output at the time of using the case of using one Embodiment of this invention, and the conventional system. FIG. 10 is a comparative explanatory diagram of correction results in a case where a filter coefficient pre-optimization process is performed and in a case where the filter coefficient pre-optimization process is not performed in an embodiment of the present invention. It is comparison explanatory drawing of the result of the tracking property with respect to a gain change at the time of using the case where one Embodiment of this invention is used, and a conventional system. It is comparison explanatory drawing of the followable | trackability with respect to position shift at the time of using the case where one Embodiment of this invention is used, and a conventional system. It is comparison explanatory drawing of the result of having performed the adaptive line enhancer process with respect to the correction value with respect to the position shift when using one embodiment of the present invention and using the conventional method. It is a modification of arrangement | positioning of the straightness component removal means and thermal noise component removal means which are characteristic in one Embodiment of this invention.

Explanation of symbols

10 Straightness correction device (error correction device)
12 Straightness component removal means (motion error component removal means)
14 Thermal noise component removal means 20 Calibration value signal acquisition means 22 Measurement value signal acquisition means 24 X-axis drive part 26 Stylus arm (movement part)
28 Contour measuring machine (Shape measuring machine)
30 Table (fixed part)
34 Stylus 36 Detector 38 Shape component removal means 42 Shape component recovery means 50 Straightness correction program (error correction program)
60 Adaptive noise canceller 70 Adaptive line enhancer

Claims (8)

Using the calibration value signal obtained by measuring the shape of a reference workpiece with a known shape in advance, from the measurement value signal obtained by measuring the shape of the workpiece with the detector while moving the moving part relative to the fixed part. In the error correction method for removing the measurement error due to the motion error of the moving part,
The calibration value signal removes the shape component of the reference workpiece from the output signal of the detector when measuring the shape of a reference workpiece with a known shape while moving the moving portion relative to the fixed portion during calibration. And
The measurement value signal is obtained from the output signal of the detector when measuring the shape of the measurement workpiece while moving the moving part relative to the fixed part at the time when the time has elapsed since the calibration. It is the work shape component removed,
A motion error component removing step for adaptively removing, as a motion error component, a component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal, using the calibration value signal. Error correction method. The error correction method according to claim 1,
The motion error component removing step adaptively removes the motion error component from the measurement value signal using the calibration value signal by an adaptive noise canceller. The error correction method according to claim 1 or 2,
An error correction method comprising: a thermal noise component removing step of removing a thermal noise component from components included in the measurement value signal after the motion error component removing step. The error correction method according to claim 3,
The error correction method characterized in that the thermal noise component removal step adaptively removes the thermal noise component from the measurement value signal after the motion error component removal step by an adaptive line enhancer. In the error correction method according to any one of claims 1 to 4,
The error correction method, wherein the motion error is a straightness of a measurement axis of a shape measuring machine. The error correction method according to claim 5,
Shape component removal, which is provided before the motion error component removal step, removes the workpiece shape component from the output signal of the detector when measuring the shape of the workpiece while moving the motion portion relative to the fixed portion. Process,
A shape component recovery step for recovering the workpiece shape component removed in the shape component removal step in the measurement value signal from which the error component has been removed before obtaining a correction value from the measurement value signal from which the error component has been removed; ,
An error correction method comprising: Using the calibration value signal obtained by measuring the shape of a reference workpiece whose shape is known in advance, from the measurement value signal obtained by measuring the shape of the workpiece with the detector while moving the moving portion relative to the fixed portion, In an error correction apparatus for removing a measurement error caused by a movement error of the moving part,
The calibration value signal is obtained by removing the shape component of the reference workpiece from the output signal of the detector when measuring the shape of the reference workpiece while moving the moving portion relative to the fixed portion during calibration. Yes,
The measurement value signal is obtained from the output signal of the detector when measuring the shape of the measurement workpiece while moving the moving portion relative to the fixed portion at the time when the time has elapsed since the calibration. The shape component of
A motion error component removing unit is provided for adaptively removing, as a motion error component, a component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal using the calibration value signal. An error correction device. Using a calibration value signal obtained by measuring the shape of a reference workpiece with a known shape in advance by causing the computer to execute the motion error component removal process, shape the workpiece with the detector while moving the moving portion relative to the fixed portion. An error correction program for removing a measurement error caused by a movement error of the moving part from a measurement value signal obtained by measurement,
The calibration value signal is obtained by removing the shape component of the reference workpiece from the output signal of the detector when measuring the shape of the reference workpiece while moving the moving portion relative to the fixed portion during calibration. Yes,
The measurement value signal is obtained from the output signal of the detector when measuring the shape of the measurement workpiece while moving the moving portion relative to the fixed portion at the time when the time has elapsed since the calibration. The shape component of
In the motion error component removing step, the calibration value signal is used to adaptively remove a component having the highest shape correlation with the calibration value signal among the components included in the measurement value signal as a motion error component. An error correction program.

Images