JPH0698553B2 - Ultrasonic signal feature extraction device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention relates to an ultrasonic signal generated by friction between a cutting tool (tool) and an object to be cut (work), that is, an acoustic emission signal ( Hereinafter, the present invention relates to an ultrasonic signal feature extraction device that detects, for example, wear and damage of a cutting tool by detecting an AE signal) and detecting the feature of the AE signal.

[Conventional technology]

Conventionally, an attempt was made to estimate the state of the cutting tool, that is, the wear and damage states, by detecting the AE signal generated by the friction between the cutting tool and the object to be cut during cutting with a lathe, milling machine, drilling machine, etc. Has been done a lot.

If the wear and damage of the cutting tool used for cutting, for example, can be determined by detecting this AE signal, it can be replaced with a new cutting tool and cutting can be performed to improve the quality of the product. Also, by providing the machine tool with a device (tool changer) for automatically changing to a new cutting tool, unmanned operation of the machine tool becomes possible.

Since the AE signal generated from the machine tool is generated due to the plastic deformation of the work piece during cutting or the friction between the cutting tool and the work piece, it is considered to have a lot of information on the processing state.

However, the AE signal actually detected by the AE sensor contains not only information about machining, but also mechanical noise generated from the gear mechanism or bearings of the spindle motor or speed reducer, or electrical noise from switches or fluorescent lamps. It is included and only the information signal regarding cutting must be extracted from such AE signal. However, most of the mechanical noise contained in this AE signal has its frequency component distributed in a frequency region considerably lower than the information signal related to the cutting process to be detected, and these are simple high-pass filters. Can be removed.

In this way, even if an AE signal that does not contain noise is obtained, any AE signal that occurs with cutting, such as collision between chips and cutting tools, wear of cutting tools, chipping or cracking of cutting tools, etc. Will be included at the same time,
It is not easy to separate them. In addition, the AE signal changes intricately due to changes in cutting speed, cutting, feed, chip shape, and so on.

However, in steady-state cutting (chips are also stable), when AE signals are observed at short time intervals (about 10 msec), even randomly generated AE signals are statistically stable. There is. Therefore, when considered over a relatively long time (about 1 second), the average value and frequency distribution of the AE signal, the number of AE signals generated per unit time, etc. do not fluctuate significantly, and the AE signal having an extremely large amplitude. Does not occur. At this time, if the cutting tool bit or drill breaks, for example, the AE signal with a large amplitude lasts for a certain period of time, and the AE signal generation state before and after the breakage generally changes significantly with both the average amplitude and the frequency distribution. To do.

As a machining monitoring device for machine tools, appropriate processing must be applied to this AE signal to detect the desired phenomenon. The ideal function of this monitoring device is to reliably detect a desired phenomenon (such as wear or breakage of the cutting tool). However, the device installed at the processing site is required to be simpler to operate and inexpensive.

A commercially available device must detect the target AE signal while satisfying the above conditions in a well-balanced manner.

The first conventional means for detecting such an AE signal is that the AE signal is a few μ
It can be thought of as a superposition of damping oscillations of the order of a second, but when processing such an essentially transient signal,
The FFT analyzer is most used in laboratories. This FFT analyzer is extremely excellent especially as a transient signal processing device, and it can be said that the one to which a computer is connected as an arithmetic device is the most advanced AE signal analysis device at present.

The second conventional means for detecting an AE signal is, as shown in FIG. 13, that an AE signal from an AE sensor a is amplified to an appropriate level by an amplifier circuit b, and then the mechanical noise is removed by a high-pass filter c. After removing the remaining signals and performing full-wave rectification on the full-wave rectification circuit d, an averaging circuit (integrating circuit)
The signal averaged by e is compared with the threshold value preset by the threshold value setting circuit f by the comparison circuit g, and when it exceeds the threshold value, it is judged as an abnormal signal. is there. The advantages of the device configured in this way are
The principle of signal processing is clear, the structure is simple, and the price is low.

The third conventional means for detecting AE signals is to prepare two filters having different characteristics as the high-pass filter shown in FIG. 13, average the signals from both filters, and then calculate the ratio of their magnitudes. For example, breakage of the cutting tool is detected. This is an AE generated from a tool changer etc.
The signal has a different frequency distribution from the AE signal due to breakage of the cutting tool, so by taking the ratio of these AE signals,
The AE signal is being discriminated. Further, even if the absolute value of the AE signal fluctuates to some extent, there is an advantage that its detection capability does not deteriorate, and it can be said that the drawbacks of the device shown in FIG. 13 are considerably improved. In addition, there is also a device for detecting only tool wear by AD conversion of the average value signal of the AE signal and then performing numerical processing by a microcomputer or the like to eliminate fluctuations in the AE signal due to chips and the like (JP-A-60- 228058 publication).

Further, the fourth conventional means for detecting an AE signal is to consider the pulse to randomly generate an AE signal, count the number of pulses exceeding a set threshold value, and There is also a means for calculating the temporal change in the rate of change per unit time with a microcomputer or the like to detect the initial damage or wear of the cutting tool.

[Problems to be Solved by the Invention]

However, in the case of the conventional first means for detecting the AE signal, the price is naturally high, and it is not practical to bring the AE signal to the processing site as it is because it can be detected by this device. Also, the AE signal has a very high frequency (100 KHz to 1 MHz), and this device has dead time because it analyzes the AE signal captured in a short time over time. Can be said to be a fatal drawback. Therefore, generally, based on the means developed in the laboratory using such a device, a more inexpensive device which is easy to operate and has no dead time will be developed.

Further, in the case of the conventional second means for detecting the AE signal, there are many problems in terms of detection reliability. For example, an abnormal signal is generated by picking up an AE signal due to a shock generated from an accessory mechanism that operates intermittently such as a tool changer. Further, considering the simplicity of the operation, the variable resistor for setting the threshold value must be readjusted every time the operator changes the processing conditions. As mentioned above, since the AE signal fluctuates due to many factors, its adjustment is very delicate, and it is hardly practical in the present day when various processing is required.

Also, in the case of the third conventional means for detecting the AE signal, the AE signal having a band of several 100 KHz is too slow at the processing speed of the current microcomputer to directly process the AE signal. I have no choice. In any case, since this device averages the AE signal, it only roughly captures the overall fluctuation of the AE signal, and cannot detect the AE signal finely.

Further, in the case of the conventional fourth means for detecting the AE signal, since the information on the amplitude is not used, it is impossible to detect a sudden signal such as breakage of the cutting tool.

As described above, the conventional means for detecting an AE signal is composed of a "feature extraction device" that takes in an AE signal and extracts some features, and a "post-processing device" that processes the output by a microcomputer or the like. Conventionally, there has been a tendency to enhance the function of this post-processing device.

However, these means lose much of the information originally contained in the AE signal by performing averaging or uniformly handling those exceeding the set threshold value at the stage of the feature extraction device. Therefore, even if the capability of the post-processing device is enhanced, its detection capability is limited. Also, the signal included in the AE signal has a great deal of information, and since the device that processes all of that information is currently in a very large scale, it is necessary to perform some feature extraction. However, in order to develop a monitoring device with even better performance, an AE signal feature extraction device that can process at high speed and low cost without losing as much information as possible is required.

The conventional AE signal detecting means reduces the amount of information to be processed by paying attention to a part of the information contained in the AE signal and discarding the other part. Therefore, the phenomena that can be detected are limited to specific ones.

The present invention, which has been difficult to detect at the same time in the past, monitors a wide range of sudden phenomena and gradually changing phenomena.Therefore, the waveform characteristics related to the frequency component and the amplitude component of the AE signal are monitored over the entire band. By extracting without dead time, the information is thinned out evenly from all of the information contained in the AE signal, so that the entire image of the information contained in the AE signal can be stored and processed by a simple microcomputer as a post-processing device. It is an object of the present invention to provide an ultrasonic signal feature extraction device that reduces information to an appropriate amount and outputs the information in a form that the post-processing device can easily process.

[Means for Solving the Problems]

As a means for solving the above problems, the present invention uses an ultrasonic signal sensor that converts an ultrasonic wave generated by friction between a cutting tool and an object to be cut into an electric signal, and an appropriate ultrasonic wave signal. A high-pass filter that is connected to the output side of this amplification circuit and that extracts only the high-frequency component from the ultrasonic signal, and the ultrasonic signal obtained from this high-pass filter passes zero voltage. Each time the positive or negative maximum voltage is reached and the zero voltage is passed again, the elapsed time is measured and digitized to generate an elapsed time value.A pulse width digitizing circuit and the maximum voltage within the elapsed time are measured and digitized to the maximum voltage value. The amplitude quantification circuit that generates the pulse width and the pulse width quantification circuit and the amplitude quantification circuit respectively connected to the lapsed time value and the maximum voltage value In which the manner selected counter and feature extracting unit of the ultrasonic signals, characterized by comprising a counter array that is updated every elapsed time.

[Action]

The ultrasonic signal feature extraction device of the present invention has the above-described configuration, amplifies the ultrasonic signal detected by the ultrasonic signal sensor to an appropriate level with an amplifier circuit, and passes the ultrasonic wave of about 100 KHz by high-pass. A filter removes mechanical noise and electrical noise. The ultrasonic signal changes irregularly both in pulse width and amplitude, and the pulse width and amplitude are converted into numerical values for the pulse width and amplitude, respectively. At this stage, since the ultrasonic signal has not been subjected to any averaging or integration processing, the information contained in the ultrasonic signal is digitized, but its resolution is reduced depending on the degree of digitization. Since all the signals are processed equally, there is no problem that all the specific components of the ultrasonic signal are lost, such as loss of a high frequency signal which is a problem in signal averaging. In addition, since the ultrasonic signal is digitized by the processing in the pulse width digitizing circuit and the amplitude digitizing circuit, and it is not necessary to handle the analog signal thereafter, most of the circuit is configured only by the digital circuit. Well, therefore, in this device, by the numerical value corresponding to the pulse width and the numerical value corresponding to the amplitude of each pulse signal of the ultrasonic signal, which is numerically converted by the pulse width numerical value circuit and the amplitude numerical value circuit. The number of times a pulse signal having such a pulse width and amplitude is stored in a memory at a two-dimensionally specified position that constitutes a counter array is counted. Then, the characteristics of the ultrasonic signal can be discriminated by reading the above-mentioned information stored in the count array by an appropriate reading means.

〔Example〕

FIG. 1 is a block diagram of an embodiment of the present invention, in which 1 is an ultrasonic signal sensor for converting an ultrasonic wave generated by friction between a cutting tool and an object to be cut, that is, acoustic emission (AE) into an electric signal ( Hereinafter, referred to as an AE sensor), 2 is an amplifier circuit for amplifying the ultrasonic signal (hereinafter referred to as AE signal) to an appropriate level, 3 is connected to the output side of the amplifier circuit 2, and the high frequency component from the AE signal A high-pass filter that extracts only the signal, 4 is a pulse width digitization circuit that extracts and digitizes the pulse width from each AE signal whose pulse width and amplitude fluctuate at any time, and 5 is an amplitude value that also extracts and digitizes the amplitude. A digitizing circuit 6 is connected to the pulse width digitizing circuit 4 and the amplitude digitizing circuit 5, respectively, and is a memory at a position secondarily designated by the pulse width and the amplitude digitized by these. It is a counter array that counts how many times such a pulse is stored.

The pulse width digitization circuit 4 digitizes the time until the absolute value of the amplitude of each pulse of the AE signal exceeds the set value and then falls below the set value again, and the amplitude digitization circuit 4 5 is adapted to digitize the peak value of each pulse of the AE signal.

Further, the counter array 6 is an array aggregate of memories that is two-dimensionally determined by two numerical values of a pulse width and an amplitude that are digitized.

The ultrasonic signal feature extraction device configured as described above,
An AE signal detected by the AE sensor 1 is amplified by an amplifier circuit 2 to an appropriate level, and a high-pass filter 3 that passes an AE signal of about 100 KHz removes basic mechanical noise and electrical noise. This AE signal changes irregularly in pulse width and amplitude, but as shown in FIG.
The pulse width and amplitude of all AE signals are detected independently and digitized. Therefore, at this stage, since neither averaging nor integration processing has been performed, the resolution of the information contained in the AE signal is
All AEs, although decreasing depending on the degree of digitization
Since the processing is equal to the signal, there is no problem that all the specific components of the AE signal are lost, such as loss of a high frequency signal which is a problem in signal averaging. Also,
With the pulse width digitizing circuit 4 and the amplitude digitizing circuit 5, AE is performed.
By processing the signal, the AE signal is digitized, and since it is not necessary to handle the analog signal thereafter, many parts of the circuit may be composed only of digital circuits, and the device can be inexpensively constructed.

Reference numeral 7 is a reading means for reading the number of times the AE signal is stored, which is stored in the counter array 6.

Here, as shown in FIG. 2, when the numerical value corresponding to the pulse width of the AE signal is X and the numerical value corresponding to the amplitude is Y, X, A set of numerical values (digital signals) of Y is generated and sent to the counter array 6. This counter array 6 is sent
Each time an AE signal of the same combination of numerical values is sent to the count contents (count value of how many times the AE signal is stored) of one memory specified two-dimensionally by the combination of numerical values of X and Y. Add 1 to. Therefore, all the detected AE signals are always reflected in the count content of any one of the memories in the counter array 6, and the count content stored in the counter array 6 after a fixed time is used as the reading means 7. When it is displayed on the screen by a microcomputer, it has a two-dimensional numerical value distribution as shown in FIG. 3, for example. This makes it possible to capture the overall characteristics of the AE signal.

Generally, in steady cutting without generating an AE signal with a large amplitude, when the count content of the counter array is read at appropriate time intervals, the count content of the memory corresponding to the large amplitude signal is always 0. The count content of the memory corresponding to the small-amplitude AE signal that is generated as a result becomes thousands to tens of thousands. At this time, for example, if an AE signal with a sudden large amplitude (due to breakage of a cutting tool, etc.) occurs three times, the count content of the memory corresponding to the amplitude and pulse width of the AE signal is 0 until now. Since 3, the breakage of the cutting tool can be reliably detected. Also, the AE that occurs constantly
Since the count content of the memory corresponding to the signal has a value of several thousands to tens of thousands, the statistical fluctuation of the numerical pattern stored in the counter array causes a problem of the overall fluctuation of the AE signal which is constantly generated. A machining phenomenon (for example, wear of a cutting tool or a minute defect) can be correctly determined. From this,
It can be used not only for abnormality monitoring but also as a processing status monitor.

The count content of the counter array can be read out by a microcomputer as a reading means, and by making it possible to read out even during the acquisition of the AE signal, the count from the counter array can be read for subsequent processing. The time to read the contents can be kept to a minimum. In addition, while reading the count contents of the count array and erasing the count contents, a plurality of counter arrays are provided so as not to miss the count of the AE signal, and through the switching means (not shown), By configuring this counter array to be connected to the pulse width digitizing circuit 4 and the amplitude digitizing circuit 5 and the reading means 7 for reading the count content of the counter array, an external command is issued to the switching means via the switching means. It is also possible to instantaneously switch the plurality of counter arrays.

The embodiment of the present invention will be described in more detail with reference to FIGS. 4 to 6. Reference numeral 8 in FIG. 4 denotes an AE pulse detection circuit, which constitutes a counter array 6. Reference numeral 6a denotes a pulse width and an amplitude of an AE signal. An array aggregate of memory that is two-dimensionally determined by two digitized numerical values, 6b is a memory selection circuit interposed between the array aggregate 6a of this memory and the amplitude digitizing circuit 5, and 6c is an array of memory. A memory selection circuit interposed between the aggregate 6a and the pulse width digitizing circuit 4, and 6d interposed between the memory array aggregate 6a and the AE pulse detection circuit +1.
A circuit, 6e, is a read circuit interposed between the memory array assembly 6a and the read means 7 such as a microcomputer.

Next, the operation will be described. As an example of the AE signal, the sixth
Consider a case where a waveform as shown in the figure is input.

Generally, the waveform of the AE signal varies irregularly in both amplitude and pulse width as shown in FIG. The AE signal feature extraction apparatus processes these waveforms as pulses. That is,
When the AE signal passes the 0 level and does not exceed the set signal height, that is, the amplitude, the AE signal is ignored, and after passing the set amplitude, it passes through the 0 level again (see, for example, FIG. 6). A) to b) are regarded as one AE signal,
It is expressed as "one AE signal is detected". Therefore,
In FIG. 6, five AE signals are included.

A circuit that detects the beginning and end of such an AE signal
It is the AE pulse detection circuit 8. The AE pulse detection circuit 8 outputs a measurement start signal at the front end (point a in FIG. 6) of each AE signal and the memory array of the counter array 6 at the rear end (point b in FIG. 6). A "recording command signal" is generated to record the generation of the AE signal in the aggregate. However, the measurement start signal is always output at the front end of the AE signal, but the recording command signal is not output unless the amplitude of the AE signal exceeds the set value (see FIG. 5). The pulse width digitizing circuit 4 and the amplitude digitizing circuit 5 operate in synchronization with the two signals.

The pulse width digitizing circuit 4 comprises a clock generating circuit 4a and a clock counter 4b. The clock counter 4b is reset to 0 by the measurement start signal from the AE pulse detection circuit 8 and starts counting time (pulse width of AE signal) by counting clock pulses from the clock generation circuit 4a.

Further, the amplitude numerical value circuit 5 digitizes the maximum amplitude of the AE signal until it is reset to 0 by the measurement start signal and thereafter the measurement start signal is generated again.

The vertical axis of FIG. 6 shows numerical values from 0 to 3,
This is because when the maximum measurable amplitude is divided into 4 equal parts, the amplitude is 1/4 to 2/4, 2/4 to 3/4, 3/4 to 4/4, 4/4.
It means that the amplitude digitizing circuit 5 generates 0 to 3 as numerical data for each of the above signals (for the sake of convenience of explanation, whether the signal is positive or negative is not considered). Since signals of 1/4 or less are ignored and it is determined that "AE signal is not generated", "Record command signal" is not generated and AE signal is not recorded. At this time, the pulse width digitizing circuit 4 and the amplitude digitizing circuit 5 are reset by the next measurement start signal, discarding the data up to that point and starting to detect a new AE signal. This is to prevent the noise of the circuit from being mistaken for the AE signal. These things, if the maximum recognizable amplitude is 4V (volt) in voltage,
It is summarized as shown in Table 1.

The minimum level of the AE signal that can be detected (the minimum level at which the numerical value 0 is output from the amplitude digitizing circuit 5 and the threshold value) is determined by the noise level of the circuit. If the minimum level is set lower than the noise level, the AE signal is generated. Even if it is not done, the recording command signal is sent to the memory array aggregate of the counter array 6, and noise other than the AE signal is included in the output information of the counter array 6, so that the correct detection of the AE signal is possible. Can not. However, in terms of device sensitivity, we want to detect AE signals with the lowest possible amplitude.
For this purpose, for example, if the noise level is 0.4V, the level corresponding to the numerical value 0 digitized by the amplitude digitizing circuit 5 may be changed to about 0.5 to 2V instead of 1 to 2V. Correspondence between the numerical value output from the amplitude digitizing circuit 5 and the actual amplitude range is possible with a slight modification of the circuit in the case of this device. In addition, each numerical value (0, 1,
Since the amplitude ranges corresponding to (2) and (3) can be set independently, the division intervals need not be equal.

In this way, when the detection of one AE signal is completed and the numerical value signal of the level set by the amplitude numerical value conversion circuit 5 is detected, a "recording command signal" is generated from the AE pulse detection circuit 8, The AE signal is recorded in the array aggregate of the memory of the counter array 6. The following describes in detail how the AE signal is recorded.

The memory array aggregate of the counter array 6 conceptually has a two-dimensional structure as shown in FIG. 7, and many memory array elements (16 elements from C00 to C33 in FIG. 7). Each of them has a function of adding 1 to the numerical value that it currently has as a memory.

Table 2 shows the AE signals P1 to P5 shown in FIG. 6 in the form of pulse widths, amplitudes, and corresponding array elements of the memory of the counter array.

In Figures 7 to 9 and Table 2, the total number of memories is 16
However, the total number of memories is not related to the principle of the present invention.

As an example, when the AE signal P1 shown in FIG. 6 is detected, the array element C21 of the memory shown in FIG. 7 is selected and 1 is added to the currently recorded value. Therefore, when the stored numerical values of the array elements of all the memories are 0 before the AE signal P1 is detected (see FIG. 8), the AE signals from the AE signal P1 to the AE signal P5 in FIG. As a result of detecting (see FIG. 2 and Table 2), the memory array will have the values shown in FIG. So, for example, the memory array element C3
By examining the stored numerical value of 2, it can be seen that two AE signals with a pulse width of 3 and an amplitude of 2 were detected.

Such a function of storing the AE signal may be realized by any means, but in view of the reduction of the circuit scale and the cost, in this device, as shown in FIG. 4, the memory array assembly 6a and the memory selection circuit are arranged. The counter array 6 is composed of 6b, 6c, the +1 circuit 6d, and the read circuit 6e. The memory is practically a memory IC and can store numerical information, but since it does not have the "function of adding 1" as a counter,
This function is realized by adding a +1 circuit 6d.
The +1 circuit 6d has a function of reading the stored numerical value of the memory selected by the memory selection circuits 6b and 6c, adding 1 to the stored numerical value, and writing the numerical value information to the same memory again. Therefore, since the AE signal is recorded every time the +1 circuit 6d is activated, the "recording command signal" is nothing but a signal for activating the +1 circuit 6d. The reading circuit 6e is a circuit for reading the stored numerical value of the array assembly 6a of the memory from the outside, and is used by the reading means 7 for reading information.

Now, two pieces of numerical information of the pulse width and the amplitude of the detected AE signal are sent to the two memory selection circuits 6b and 6c connected to the memory array assembly 6a, and one of the memory array assembly 6a. Is selected two-dimensionally. Next, the "recording command signal" is +
1 is sent to the circuit 6d, and 1 is added to the stored numerical value of the selected memory. In this way, recording of one AE signal is completed.

The outputs from the pulse width digitizing circuit 4 and the amplitude digitizing circuit 5 are numerical data, which are sent to the counter array 6 by signal lines weighted as binary numbers. The number of signal lines depends on the amplitude and the fineness of the pulse width division. For example, if the pulse width is divided into 8 steps and detection is performed, numerical data from 0 to 7 will be transferred. At this time, three digits in binary number are required, and therefore three signal lines are required. When the number of divisions is further increased, the signal line is set to 1 each time the number of divisions is doubled (increase by one digit in binary).
I need more books. The number of divisions does not have to be a multiple of 2, but 16 signals can be sent if there are, for example, 4 signal lines. In this case, it is considered efficient to divide into 16 stages. In the case of amplitude, the relationship between the number of divisions and the number of signal lines is the same as in the case of pulse width.

From the above, for example, when the pulse width and the amplitude are divided into 16 steps, the number of signal lines is 4 for each of the pulse width and the amplitude, and a total of 8 signal lines are connected to the array assembly of the memory. . The total number of memories required is determined by the number of signal lines, and when the number of signal lines is 8, 2 ⁸ = 256 memories are required. Since the number of the memories directly becomes the amount of information to be processed by the reading means (secondary processing device) such as a microcomputer, if the memory is finely divided, more detailed information can be taken out, but the amount of information to be processed increases. Processing takes time. FIG. 7 shows the case where the amplitude and the pulse width are set in four steps, and the size of the memory array (total number of memories) needs to be 4 × 4 = 16. The pulse width and amplitude can be freely determined, and the prototype circuit has a memory array with a size of 16 × 16 = 256.

When a processing state monitoring device is configured as an application of this ultrasonic signal feature extraction device, a set cycle (for example, 1
The phenomenon that can be detected (abnormality or wear of the cutting tool) will be determined by how the secondary processing is performed on the numerical table sent from the ultrasonic signal feature extraction device at 10 sheets per second).
As an example, some secondary processes are shown below, but since considerable processing power is required to realize these functions,
A microcomputer or the like is required as the secondary processing device, but this is not an essential requirement of the present invention.

FIG. 10 shows a monitoring device to which the feature extracting device of the ultrasonic signal is applied for the purpose of notifying the operator of the abnormality of the processing machine.

In this application example, a microcomputer 9 that performs the following operation is used as a means for reading the number of times the AE signal is generated and stored in the counter array 6 that constitutes the AE signal feature extraction device. To explain the operation,
The signal from the counter array 6 is sent to one input end of the judging means 9a and the reference value fetching means 9b which constitute the microcomputer 9. The reference value fetching means 9b fetches the AE signal at the time of normal cutting as a reference value and sends it to the reference value storage means 9c for storage in the subsequent processing for reference. The information stored in the reference value storage means 9c is sent to the reference value selection means 9d. In this reference value selection means 9d, the reference value sent from the reference value storage means 9c is selected in accordance with the cutting condition instructed by the dial 10 or the like outside the microcomputer, and the selected reference value is selected. Is sent to the other input end of the judging means 9a, the AE signal sent to the one input end by this judging means 9a is compared with the reference value sent to the other input end, The quality of the processed state is determined. The determination result is represented by the lighting of a red lamp 11a connected to the outside of the microcomputer, which indicates a machining abnormality by its lighting, a yellow lamp 11b indicating a caution, and a green lamp 11c indicating normality.

In mass production of small varieties, the same processing is repeated many times,
The amount of wear of the cutting tool can be predicted from past results, and the worn cutting tool can be replaced before the machining accuracy deviates from the standard value, but this is not possible in high-mix low-volume production. For this reason, wear of cutting tools can be measured by removing the cutting tool from the processing machine after the completion of processing and observing it with a microscope to measure the amount of wear.In automatic processing machines, the cycle of directly measuring the dimensions of the cutting tool is processed. It was incorporated into the program and measured directly. Therefore, there has been a problem that the processing efficiency is lowered.

Therefore, as another application example of this ultrasonic signal feature extraction apparatus, a wear tool wear display apparatus as shown in FIG. 11 is conceivable. This wear display device does not judge whether the cutting tool is good or bad as shown in FIG. 10, but based on the numerical information of the number of times the AE signal is output from the ultrasonic wave feature extraction device. , As a means for reading out the numerical information, by using the following microcomputer 12, to quantitatively display the current degree of wear of the cutting tool,
A numerical information signal of the number of times the AE signal is generated from the counter array 6 is sent to one input end of the wear degree calculating means 12a and the equation calculating means 12b which constitute the microcomputer 12. This equation is determined by the equation calculating means 12b by the preliminary test cutting, and is stored in the equation storing means 12c by the number of machining types. The equation stored in the equation storage means 12c is sent to the equation selection means 12d. In this equation selection means 12d, the equation sent from the equation storage means 12c is selected according to the cutting conditions instructed by the dial 13 or the like outside the microcomputer, and the selected equation is subjected to the wear. Degree calculation means 12a
Of the wear rate by this equation, the number of times the AE signal is generated sent to the one input end, and the cutting speed and the feed speed of the cutting tool. The calculation means 12a calculates the wear of the cutting tool, and the calculation result is displayed on the display 14 connected to the outside of the microcomputer.

In this way, if the wear of the cutting tool can be quantitatively measured in-process, the maximum amount of wear of the cutting tool that can be processed is displayed as 100. Since it is possible to monitor and display, it is possible to manage the wear of the cutting tool without lowering the machining efficiency and to operate the cutting tool economically.

Since the AE signal reflects the processing state very faithfully, if the AE signal generation state can be displayed in an easy-to-understand format,
It is possible to know subtle changes in the processing state, and it has great utility value in ultra-precision processing.

This means that the lamps whose brightness changes according to the size of the numerical data of the array elements of the counter array are arranged corresponding to the array elements of the memory of the counter array, or the array elements of the memory are visualized on the CRT display. It can be realized by displaying it in a static manner.

An apparatus using this CRT display is shown in FIG. AE
As a means for reading the number of times the AE signal is stored, which is stored in the counter array 6 which constitutes the signal feature extraction device, the following microcomputer 15 and CRT display 1 are used.
6, the signal transferred from the counter array 6 is converted by the display information converting means 15a constituting the microcomputer 15 into a display form signal which is easy for the operator to understand, and this signal is further generated as a video signal. It is converted into a video signal by the means 15b, and this video signal is sent to the CRT display 16 for display. If you look at this display,
Three-dimensional pattern, that is, pulse width, amplitude of AE signal,
Also, the subtle changes in the AE signal can be visually confirmed by the change in the number of AE signals generated classified by them.
As a result, it is possible to accurately manage the degree of wear of the cutting tool, which has hitherto been taken into consideration.

〔The invention's effect〕

As described above, the present invention provides an ultrasonic signal sensor for converting an ultrasonic wave generated by friction between a cutting tool and an object to be cut into an electric signal, and an amplifier circuit for amplifying the ultrasonic signal to an appropriate level. And a high-pass filter that is connected to the output side of this amplification circuit and extracts only the high-frequency component from the ultrasonic signal, and the ultrasonic signal obtained from this high-pass filter passes through zero voltage, and the maximum positive or negative voltage Pulse width digitizing circuit that measures and digitizes the elapsed time every time it reaches zero voltage and passes through zero voltage again, and amplitude digitizing circuit that measures and digitizes the maximum voltage within the elapsed time and generates the maximum voltage value And a counter which is connected to the pulse width digitizing circuit and the amplitude digitizing circuit, and which is selected two-dimensionally by the elapsed time value and the maximum voltage value digitized by them. Since the ultrasonic signal feature extraction device is composed of the counter array that is updated at each elapsed time, all the ultrasonic signals obtained from the ultrasonic signal sensor are two-dimensionally selected by the amplitude and the pulse width. As the numerical distribution of the counter array, it can be extracted without dead time. Therefore, in the past, in order to extract a specific property of the ultrasonic signal, we focused on only a part of the information that the ultrasonic signal has, so if we focused on sudden changes in the ultrasonic signal, we could not capture the overall change, While paying attention to the overall variation, it was not possible to detect the sudden variation, whereas the ultrasonic signal feature extraction apparatus of the present invention can detect these variations simultaneously with a relatively simple configuration.

Then, the information extracted by the feature extracting device of the ultrasonic signal is read by various reading means, for example, a microcomputer or the like, so that it can be easily applied to an advanced processing monitoring device. .

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of an ultrasonic signal feature extraction apparatus of the present invention, FIG. 2 is a diagram showing an example of an ultrasonic signal, and FIG. 3 is an ultrasonic signal feature extraction apparatus. FIG. 4 is a diagram in which ultrasonic signals are displayed on a CRT display via a microcomputer. FIG. 4 is a block diagram of a more specific embodiment of the ultrasonic signal feature extraction apparatus. FIG. FIG. 6 is an explanatory view of the pulse width and amplitude of the ultrasonic signal, and FIGS. 7 to 9 are explanatory views of an array of memories forming a counter array which is one of the constituent elements of the present invention and the stored numerical values thereof. 10 to 12 are views showing an application example of the present invention, and FIG. 13 is a view showing a conventional ultrasonic signal detecting apparatus. 1 ... Ultrasonic signal sensor, 2 ... Amplifier circuit 3 ... High-pass filter 4 ... Pulse width digitizing circuit 4a ... Clock generating circuit 4b ... Time counter, 5 ... Amplitude digitizing circuit 6 ... Counter array 6a …… Memory array assembly 6b, 6c …… Memory selection circuit 6d …… + 1 circuit, 6e …… Reading circuit 7 …… Reading means 8 …… AE pulse detection circuit 9 …… Microcomputer 10 …… Dial 11a, 11b, 11c …… Red, yellow, green lamps 12 …… Microcomputer 13 …… Dial 14 …… Display 15 …… Microcomputer 16 …… CRT display

Claims (2)

[Claims] 1. An ultrasonic signal sensor for converting an ultrasonic wave generated by friction between a cutting tool and an object to be cut into an electric signal, an amplifier circuit for amplifying the ultrasonic signal to an appropriate level, and this amplifier circuit. A high-pass filter connected to the output side of the ultrasonic signal for extracting only the high-frequency component from the ultrasonic signal, and the ultrasonic signal obtained from this high-pass filter passes zero voltage and reaches the maximum positive or negative voltage, and is zero again. Each time a voltage is passed, the elapsed time is measured and digitized to generate an elapsed time value.A pulse width digitization circuit and the maximum voltage within the elapsed time are digitized to generate a maximum voltage value. A counter connected two-dimensionally to the digitizing circuit and the amplitude digitizing circuit and selected two-dimensionally by the digitized elapsed time value and maximum voltage value is updated at each elapsed time. Feature extraction unit of the ultrasonic signals, characterized by comprising a counter array. 2. A counter array comprising a plurality of counter arrays, the counter array being connected to a pulse width numerical value circuit and an amplitude numerical value circuit via a switching means. An ultrasonic signal feature extraction device according to item 1.