JP2016078044A - Processing state determination device and processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make it possible to determine a processing state of a processor with a high accuracy.SOLUTION: A processing state determination device is equipped with: a sound collector group 10 having plural sound collectors which observe a process sound when a projectile for the purpose of processing a work-piece 2 by a processor 3 is projected, and output the sound as an observation signal; a feature extractor 200 into which plural observation signals outputted by the sound collector group 10 are inputted, converts the plural observation signals into a feature amount using a spatial feature and outputs the same; and a processing state determination part 30 into which the feature amount is inputted, and determined a processing state of the processor 3.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technique for determining a processing state of a processing machine.

A non-contact type processing machine (non-contact processing machine) that performs processing by projecting a projectile onto a processing target (processing material), for example, a laser processing machine or a water jet processing machine, the processing machine itself is the processing material. Therefore, there is an advantage that distortion of the workpiece due to cutting or cutting can be avoided.
However, on the other hand, since the workpiece and the processing machine are not in direct contact, it is impossible to detect the state of the workpiece being processed or the processing machine itself. For this reason, the determination of whether or not the machining operation has been completed has a problem that the time required for the machining operation predicted in advance must be sufficiently long and the efficiency is poor.

For example, a laser processing machine has a process called piercing, and one of the problems is to reduce the time required for this process and improve the processing efficiency. Piercing is a process in which a workpiece is irradiated with a laser intensively to form a hole.
Originally, the laser irradiation time is sufficient until it penetrates the workpiece. However, since there is no established means for detecting that the laser has penetrated, in many cases the laser is long enough. We cope by continuing to irradiate.

This irradiation time is generally set to a length approximately twice the average time required for penetration in consideration of safety, and the thicker material that takes time for penetration, the more time is wasted. .
In order to deal with such problems, it is possible to detect the state of the workpiece even in a non-contact processing machine using a sensor capable of sensing without directly contacting the workpiece, such as an optical sensor or a microphone (sound collector). Various processing state determination devices have been disclosed so far.

For example, Patent Document 1 discloses a technique for improving the processing efficiency by attaching a microphone to a laser processing machine, determining penetration of piercing processing based on the observation signal, and reducing the time required for the piercing processing described above.
Patent Document 2 also discloses a technique for detecting a machining state of a workpiece using a microphone or the like.

JP-A-6-277862 JP2009-190095

  A conventional machining state determination device collects observation signals as timbre information using only a single microphone, and determines in which machining state the sound is close. For this reason, there was a problem that the machining state could not be determined with high accuracy.

  The processing state determination apparatus of the present invention observes a processing sound when projecting a projectile for a processing machine to perform processing on a workpiece, and has a plurality of sound collectors that output as an observation signal A group, a plurality of observation signals output from the sound collector group as inputs, a feature extractor that converts the plurality of observation signals into feature amounts using spatial features, and outputs the feature amounts And a machining state determination unit that determines a machining state of the processing machine.

  According to the present invention, it is possible to determine a machining state of a processing machine with high accuracy by using a feature amount including spatial information.

It is a block diagram which shows the whole structure which concerns on Embodiment 1 of this invention. It is a figure which shows an example of the microphone arrangement | positioning in the processing state determination apparatus 1 which concerns on Embodiment 1 of this invention. It is the figure which modeled the sound wave propagation of the process sound in the piercing process (before penetration) in the microphone arrangement | positioning which concerns on Embodiment 1 of this invention. It is the figure which modeled the sound wave propagation of the processing sound in the piercing process (after penetration) in the microphone arrangement | positioning which concerns on Embodiment 1 of this invention. It is the figure which modeled the flow of the assist gas during the process (before penetration) of the laser beam machine concerning Embodiment 3 of this invention. It is the figure which modeled the flow of the assist gas during the process (after penetration) of the laser beam machine concerning Embodiment 3 of this invention.

Embodiment 1 FIG.
FIG. 1 is a block diagram showing an overall configuration according to Embodiment 1 of the present invention. In the first embodiment, the processing state determination device 1 determines the processing state of the processing machine 3 with respect to the workpiece 2.
The processing state determination apparatus 1 in the present embodiment is configured by a microphone group 10, a feature extraction unit 20, and a processing state determination unit 30.

In the present embodiment, a laser processing machine will be described as a specific example of the processing machine 3 in FIG. 1, but other processing machines such as a water jet processing machine and a plasma processing machine may be used.
In addition, although various processes are included in the material processing by the laser processing machine, paying attention to the above-described piercing processing, the processing state in the piercing processing is treated as a “pierce through” state and a “pierce non-pierce” state, A case will be described in which the current machining state is determined from the machining sound of the processing machine 3.
However, the application target of the present invention is not limited to the above-described processing machine and processing steps.

  The microphone group 10 includes at least two microphones (sound collectors). In FIG. 1, the example in the case of m microphones (10a-10m) is shown. The microphones 10 a to 10 m observe the processing sound of the processing machine 2 and output an observation signal to the feature extraction unit 20. The processing sound observed by the microphone may be not only audible sound but also non-audible sound such as ultrasonic waves.

FIG. 2 is an example of microphone arrangement in the machining state determination apparatus 1 according to the first embodiment.
FIG. 2 shows a case where two microphones are arranged around the workpiece 2 with the workpiece 2 interposed therebetween. A microphone disposed on the side irradiated with a laser by the laser processing machine (hereinafter referred to as an irradiation surface) is denoted by 10a, and a microphone disposed on the back surface side of the irradiation surface is denoted by 10b.

These microphones 10a and 10b observe the processing sound of the processing machine 3, and output the observation signal to the feature extraction unit 20 in FIG.
3 and 4 schematically show the sound wave propagation of the processing sound during the piercing processing in the microphone arrangement of FIG.

FIG. 3 shows a “pierce not penetrated” state in which the laser of the processing machine 3 does not penetrate the workpiece 2.
In FIG. 3, the processing sound generated from the laser irradiation point is blocked by the workpiece 2. Therefore, when the space on the irradiation surface side and the back surface side thereof are compared, a larger processing sound is observed in the space on the irradiation surface side.

FIG. 4 shows a “piercing penetration” state in which the laser of the processing machine 3 penetrates the workpiece 2.
In FIG. 4, since the laser penetrates the workpiece 2 and the processing sound leaks from the hole through which the laser penetrates, the sound pressure difference between the irradiation surface side and the back surface side is the unpenetrated state shown in FIG. Compared to the case, it becomes smaller.
As described above, by arranging the microphones 10a and 10b as shown in FIG. 2, it is possible to detect the magnitude of the sound pressure difference between the irradiation surface side and the back surface side of the workpiece 2.

In the present embodiment, the example in which only the microphone 10a and the microphone 10b are arranged has been described, but a microphone may be arranged in addition to this.
As described above, the processing sound of the processing machine 3 observed by the microphone group 10 is output to the feature extraction unit 20 as an observation signal.

The feature extraction unit 20 includes a feature extractor 200 and a feature extractor group 210.
The feature extractor 200 receives at least two observation signals output from the microphone group 10 as inputs, converts them into feature amounts using spatial information, and outputs them to the machining state determination unit 30.
In the present embodiment, an example is shown in which the feature extractor 200 receives the observation signals of the microphones 10 a and 10 b as inputs, converts them into feature amounts using spatial information, and outputs them to the processing state determination unit 30. ing.

The feature amount using the spatial information output by the feature extractor 200 represents the ratio of the sound pressure on the irradiation surface side and the back surface side.
For this reason, this characteristic amount is close to 1 because the difference in sound pressure between the irradiation surface side and the back surface side is small as described above in a state where the laser penetrates the workpiece 2.
Further, this feature value is a value away from 1 because the sound pressure difference is large in the unpenetrated state. That is, the ratio of the magnitudes of the observation signals output from the microphone 10a and the microphone 10b becomes different values between the pierced through state and the pierce-unpierced state, which is a useful feature amount in determining the processing state.

The feature extractor group 210 includes an arbitrary number of feature extractors.
FIG. 1 shows an example in which the number of feature extractors in the feature extractor group 210 is n (210a to 210n).
The feature extractors 210 a to 210 n receive one or more observation signals output from the microphone group 10 as inputs, convert them into feature amounts, and output them to the machining state determination unit 30.
The feature quantity converted from the observation signal by the feature extractors 210a to 210n in the feature extractor group 210 is not necessarily the spatial information described in the case of the feature extractor 200, and timbre information may be used.
The total number of microphones and the total number of feature extractors may be different values or the same value.

  The processing state determination unit 30 determines the processing state of the processing machine based on the feature amount output from the feature extraction unit 20 and outputs the processing state to the processing machine 3.

The processing state determination unit 30 determines the processing state based on the feature amount output from the feature extraction unit 20.
Specifically, a multidimensional vector composed of feature amounts output from the feature extraction unit 20 is identified, and the processing state of the processing machine 3 is determined.
For identification of a multidimensional vector, for example, a neural network or a support vector machine can be used.

The processing machine 3 operates based on the processing state output from the processing state determination unit 30.
During the piercing process, the processing machine 3 continues the piercing while the processing state output from the processing state determination unit 30 is in the “pierce not penetrated” state, and the processing state output from the processing state determination unit 30 is If it is in the “piercing penetration state”, the piercing process is terminated and the process proceeds to the next step.

Further, since the position of the processing head of the laser processing machine changes every moment during processing, the signal level of the observation signal changes according to the position of the processing head when the position of the microphone is fixed.
However, in the case of the microphone arrangement as shown in FIG. 2 described in the present embodiment, it can be expected that the signal level increases and decreases at the same ratio in each microphone when the position of the machining head changes.
Therefore, since the feature extractor 200 outputs the feature amount based on the ratio of the magnitudes of two or more input signals, the amount of change in the signal level accompanying the movement of the machining head is canceled out, so that the generation of machining sound The influence by the change of a point can be reduced and the robustness of the processing state determination with respect to the processing machine 3 improves.

  The feature extraction unit 20 and the machining state determination unit 30 can be realized as software on a personal computer or a microcomputer, for example.

  According to the present embodiment, the feature extractor 200 extracts feature quantities including spatial information based on the observation signal of the microphone group 10 composed of at least two microphones, and processes based on the feature quantities. When the state determination unit 30 determines the processing state of the processing machine 3, it is possible to determine the processing state with higher accuracy compared to the method based only on the timbre information.

  Further, by arranging at least two or more microphones constituting the microphone group 10 with the workpiece 2 separated, a plurality of spaces separated by the workpiece 2 are connected through a hole penetrating the workpiece 2. Can be detected based on the ease of propagation of sound waves, thereby detecting with high accuracy whether the processing machine 3 has completed a processing step such as making a hole in the workpiece 2 or the like. It becomes possible.

The processing state determination unit 30 determines the processing state based on a multi-dimensional vector composed of a plurality of feature amounts output from the feature extraction unit 20, so that timbre information, spatial information, and the like are integratedly used for determination of the processing state. Can be determined with high accuracy.
As described above, in the present embodiment, the laser processing machine has been described as a specific example of the processing machine 3 in FIG. 1, but other processing machines such as a water jet processing machine and a plasma processing machine may be used.

Embodiment 2. FIG.
In the first embodiment, the feature extractor 200 outputs the ratio of the magnitudes of at least two or more observation signals output from the microphone group 10 to the processing state determination unit 30 as a feature amount using spatial information. Explained the case.
In the second embodiment, a case will be described in which the feature extractor 200 outputs a difference in magnitude between at least two observation signals output from the microphone group 10 as a feature amount using spatial information.

In the present embodiment, as in the first embodiment, a case where two microphones (microphone 10a and microphone 10b) are arranged across the workpiece 2 as shown in FIG. 2 and the processing state of piercing processing is determined. A specific example will be described.
In the state in which the laser penetrates the workpiece 2, the difference in sound pressure between the microphone 10 a and the microphone 10 b becomes a value close to 0 because the difference in sound pressure between the irradiation surface side and the back surface side is small as described above. I can expect that.
On the other hand, since the sound pressure difference is large in the non-penetrated state, it can be expected that the value is away from zero.
That is, the difference in the magnitude of the observation signal between the microphone 10a and the microphone 10b differs between the pierce-through state and the pierce-unpierce state, which is a useful feature amount in determining the processing state.

In the second embodiment, when noise coming from outside the processing machine 3 or the processing state determination device 1 is mixed in the observation signal, the noise observed by the microphone 10a and the microphone 10b is almost the same sound pressure level. Can be expected.
Therefore, the feature extractor 200 calculates the difference between the magnitudes of the observation signals of the microphone 10a and the microphone 10b, so that a feature amount in which the influence of noise coming from the outside is canceled is output. Improves robustness against noise.

ただし、ωは周波数、Ｚ（ω）はＹ（ω）の複素共役、Ｅ［・］は期待値を表す。このとき、Ｃ（ω）はコヒーレンス関数と呼ばれ、ふたつの信号の関連性が高いときは１、低いときは０に近い値となる。 Embodiment 3 FIG.
In the first embodiment and the second embodiment, the feature extractor 200 processes the magnitude ratio and difference of at least two observation signals output from the microphone group 10 as feature amounts using spatial information. The case of outputting to the state determination unit 30 has been described.
In the third embodiment, a case will be described in which the feature extractor 200 outputs a correlation between at least two observation signals output from the microphone group 10 as a feature amount.
The correlation is a quantity representing how much a plurality of signals are related to each other. For example, using the frequency spectra X (ω) and Y (ω) obtained from the Fourier transform of two signals, the degree of correlation can be calculated as in the following equation.

However, ω represents a frequency, Z (ω) represents a complex conjugate of Y (ω), and E [•] represents an expected value. At this time, C (ω) is called a coherence function, and is a value close to 1 when the relevance of the two signals is high, and close to 0 when the relevance is low.

In the first embodiment and the second embodiment, as a result of the processing machine 3 performing a processing operation on the workpiece 2, processing is performed using a difference in sound wave propagation depending on the presence or absence of a hole opened in the workpiece 2. Although it determines the state, it is also possible to use the difference in airflow.
In the third embodiment, as in the first embodiment, two microphones (microphone 10a and microphone b) are arranged across the workpiece 2 as shown in FIG. 2, and the processing state of piercing processing is determined. Will be described as a specific example.

A laser processing machine constantly blows a gas called assist gas to a laser irradiation point while processing.
The assist gas has effects such as blowing away chips of the melted workpiece 2 and promoting an oxidative combustion reaction.
FIG. 5 and FIG. 6 schematically show the flow of the assist gas during the processing of the laser processing machine.

As in the case of the sound wave propagation of the processing sound generated from the laser irradiation point described in the first embodiment, the assist gas released from the processing head is blocked by the workpiece in the non-piercing state. The assist gas does not flow into the space on the back side.
At this time, the correlation between the observation signal of the microphone 10a and the observation signal of the microphone 10b can be expected to be low.

On the other hand, in the piercing through state, the assist gas flows into the space on the back side from the hole of the workpiece, and it can be expected that the correlation between the observation signals of the microphone 10a and the microphone 10b is higher than that in the non-penetrating state. .
That is, since the correlation between the observation signals of the microphone 10a and the microphone 10b is different between the pierced through state and the unpierced state, it is a useful feature amount in determining the processing state.

  The correlation of the observation signal can be calculated while avoiding the influence of the setting of the microphone, the amplifier, the AD converter, and the individual difference. Therefore, it is possible to reduce the work cost required for correcting the microphone, the amplifier, the AD converter, and the like.

  DESCRIPTION OF SYMBOLS 1 Processing state determination apparatus, 2 Work material, 3 Processing machine, 10 Microphone group, 10a-10m microphone, 20 Feature extraction part, 30 Processing state determination part, 200 Feature extractor, 210 Feature extractor group, 210a-210n Extractor

Claims (8)

A sound collector group having a plurality of sound collectors for observing the processing sound when projecting a projectile for processing by the processing machine to the workpiece, and outputting as an observation signal;
A feature extractor that receives a plurality of observation signals output from the sound collector group, converts the plurality of observation signals into feature quantities using spatial features, and outputs them.
With the feature quantity as an input, a machining state determination unit that determines the machining state of the processing machine,
A machining state determination apparatus characterized by comprising: The sound collector group includes a sound collector disposed in a projection side space on which the projectile of the workpiece is projected, and a collector disposed in a space opposite to the projection side of the workpiece. A sound device,
The feature extractor receives an observation signal output from a sound collector disposed in the projection-side space and an observation signal output from a sound collector disposed in a space opposite to the projection side. The machining state determination apparatus according to claim 1, wherein A plurality of the feature extractors;
The machining state determination apparatus according to claim 1, wherein the machining state determination unit receives feature amounts output from a plurality of the feature extractors.   4. The processing state determination apparatus according to claim 1, wherein the feature extractor converts the feature amount into a feature amount based on a ratio of magnitudes of a plurality of input observation signals. 5.   4. The processing state determination apparatus according to claim 1, wherein the feature extractor converts the feature amount into a feature amount based on a difference in magnitude between a plurality of input observation signals. 5.   4. The processing state determination device according to claim 1, wherein the feature extractor converts the feature amount into a feature amount based on a correlation between the magnitudes of a plurality of input observation signals. 5.   The machining state determination apparatus according to claim 3, wherein the machining state determination unit determines a machining state based on a multidimensional vector including a plurality of feature amounts.   A processing machine comprising the processing state determination device according to any one of claims 1 to 7.

Images