JP2564189B2 - Machine tool life judgment method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial field of application) The present invention is intended to prevent wear and damage of a machine tool in a lathe or the like.
The present invention relates to a method for detecting and determining life by AE (Acoustic Emission).

(Prior Art) <Method of Judging Life> Conventionally, the method of detecting wear / damage of a machine tool such as a lathe and judging the life is as follows: cutting number, time / cutting resistance, cutting temperature, vibration, cutting noise, There were methods by load current, surface roughness, gloss, etc., and methods by AB.

1) Method based on the number of cuts and time It is currently used in many factories. This is a method in which the state of wear and damage is empirically investigated beforehand, and the number of processed products and the total time required for processing are used for the determination.

In this method, a tool that has not actually reached the end of its life is replaced early, or conversely, a tool that has already exceeded the usage limit is being used, and a defective product may be produced.

2) Method based on cutting resistance This is a method in which a strain gauge is attached to the cutting tool, and the elongation of the gauge is converted into voltage and displayed on a recorder.

In this method, a strain gauge must be directly attached near the cutting edge of the cutting tool, which causes a hindrance to the cutting process. Also, since it is attached to the cutting tool each time, it cannot be used if peeled off.

In addition, since the strain increases little by little, only qualitative judgment can be made. Further, since it is difficult to determine the judgment time, the reliability is poor.

3) Method based on cutting temperature a) When measuring the temperature at the tip of the cutting edge A thin hole must be drilled close to the cutting edge of the cutting tool, and a thermocouple temperature measurement wire must be inserted into the hole. For this reason, it is difficult to make a hole in a commonly used cemented carbide tool because it is extremely hard.

b) Average cutting edge temperature It is necessary to put the work material in an insulating state. This is a method that cannot be done outside the laboratory.

In both a) and b), it is difficult to set the conditions of the measuring device,
This is a method that cannot be put to practical use, and the determination accuracy is qualitative as in resistance measurement, and determination cannot be made.

4) Vibration method Measure with a vibration measurement sensor attached. This is easily affected by the vibration of the machine and the vibration transmitted from other machines, and is difficult to determine.

5) Method using cutting sound This is a method in which a microphone is attached near the cutting tool. If it is used in a factory, noise in the factory will be picked up and it will be impossible to judge.

6) Method using load current A current measuring device is required to check the value of the spindle current flowing during cutting. This method cannot be determined because the influence of the change in the wear amount of the tool is small. Only a considerably large breakage (a tool breaks) etc. will be judged.

7) Method of Finishing Surface Roughness and Gloss A method of visually performing finishing roughness and gloss. When a human being is near the machine and must finish the processing, he must look at the surface of the work material to make a judgment. Currently, one person is seeing about three machines at any factory, so it is not possible to assign people to all machines. In addition, since the liquid is sprayed during normal processing, it is often impossible to see the finished surface, and it is naturally impossible to predict.

8) Method by AE The detection method by AE (acoustic emission = elastic liquid generated by releasing energy stored when material plastically deforms or breaks) is often applied to the microscopic process of plastic deformation or failure. Therefore, it was thought that in-process measurement of tool life would be possible if it could be applied to the determination of tool life in continuous cutting in turning. However, many of them have been investigated only by the effective value (RMS value) of AE, which has not yet been put to practical use.

<Predictive device-AE sensor> Since the conventional AE sensor has the wave receiving plate directly bonded to the tool, the AE signal as a vertical / horizontal composite wave is supplemented, and the AE wave generated when cutting fluid or chips is used. There was a lot of noise, and the desired signal was sometimes disturbed.

(Problems to be solved by the invention) In view of the above situation, the present invention quantitatively measures wear and damage of a tool by a liquid type AE sensor (hereinafter referred to as AE sensor) in an in-process state, It provides a means for stably improving the accuracy of a work piece and appropriately managing a production system by determining an abnormality or a failure in real time and performing preventive maintenance.

In addition, the AE sensor is more effective if it is capable of eliminating noise.

[Structure of the Invention] (Example) <Determination Method> As an AE analysis, a threshold value is set, and a longitudinal wave waveform from when the waveform is sensed to when it is no longer sensed is recognized as one waveform. One event. Further, the number of amplitudes in one event is the number of oscillations. This is the first
Shown in the figure.

Comparing the number of events with the number of oscillations, the number of events tends to gradually decrease as the tool wear increases. In the case of oscillation, the number of events increases gradually as opposed to the number of events. This is shown in FIG.

From these tendencies, the ratio of the number of events and the number of oscillations can be used to quantitatively determine the numerical value of the tool life judgment criterion, which is the most effective judgment.

Below, the minimum delation function, floating function, and real-time analysis software will be described in order in the figure. It has a function for.

Figure 3 shows the effect of changes in minimum deration. The minimum duration is to set a minimum duration, and a signal whose duration is smaller than the set value is regarded as electrical noise and is cut off and not captured. If the setting value is 1024μsec, the waveform detected by the threshold for the duration less than 1024μ is not captured. 0 to these discrimination processing
The experiment was performed by dividing into 9 types up to 32768 μsec. FIG. 4 shows only three types. 0, 10 from the top
24, 32768 μsec.

In the case of 0 μsec, the number of events and the number of oscillations when all AE waveforms are captured are shown.
Since all AE waveforms (including AE waveforms generated in cutting fluid etc.), the amount of change due to tool wear is blurred.

In the case of 1024 μsec, a waveform of 1024 μsec or more is captured, the number of events is greatly reduced and changed, and the number of oscillations is also large. Therefore, it is easy to compare when the respective ratios are taken.

In the case of 32768 μsec, since the number of events is extremely small and the number of oscillations is also small, the error factor tends to increase when the ratio is taken. Also, there are many variations in the number of events that occur within the unit time of the number of oscillations.

From the above, the setting value of 1024 μsec is considered to be optimal, because noise-like waveforms can be removed and the quantity can be sufficiently captured and stabilized.

(2) Floating function The signal extracted by the minimum delation function is
Set thresholds and retrieve events. In this case, if a threshold value is always added to the average value of the waveform, the average value follows even if the amplitude changes in the waveform, so no matter how the waveform changes, it is detected as a waveform. As a result, stable uptake is possible.

FIG. 5 shows a case where the amplitude of the waveform fluctuates when the threshold value is set, and the waveform cannot be sensed at a constant threshold value. (A) shows a state in which when the set value of the threshold value is small, it can be detected when the amplitude of the waveform is small, but cannot be detected when the amplitude of the waveform becomes large. On the other hand, (b) shows a case where the set threshold value is large, which cannot be detected when the amplitude is small, but can be detected when the amplitude is large. In other words, with the normal threshold setting, there will be a situation where the AE waveform cannot be captured when the amplitude fluctuates.

FIG. 6 shows a case where a floating function is added. In the floating function, the threshold value is always added to the average value of the waveform. This is because even if there is a change in the amplitude of the waveform, the average value follows it, so if you set a threshold value that is added to the average value, you can sense it as a waveform no matter how the waveform changes. Stable uptake becomes possible. As a result, even if the amplitude of the AE waveform changes from the beginning of the tool to the end of the tool life, it can be captured without omission and stable analysis becomes possible.

(3) Real-time analysis software This is analysis software that can calculate and display stable AE signal acquisition simultaneously with the functions of (1) and (2).

In order to determine the tool life, it is necessary to quantify the reference numerical value.

In order to quantify, it is necessary to capture the number of events and the number of oscillations at regular time intervals and display the ratio thereof moment by moment. In order to display these series of displays, we have developed software for real-time analysis.

FIG. 7 is a flowchart of the real-time analysis software. A software that processes AE signals in real time with OP-01 (high-speed arithmetic module), requests the number of events that captured the processed data from the computer side within the sampling time, receives that data, and displays it after computation. It is a flowchart.

FIG. 8 shows a real-time analysis screen. The sampling time, the number of events, the number of oscillations, and the calculated value are displayed by setting the sampling time, level, and coefficient value and capturing the AE signal. When the calculated value exceeds the level setting, that value is displayed in the frame of the calculated value of level over. The tool life is determined by the number of displayed numerical values.

(Practical example) A concrete example of real-time analysis will be described below with reference to the figures and tables. As an AE reception analysis device, the AE sensor uses a PZT (lead zirconate titanate porcelain) piezoelectric element, and the AE analysis
Use AE system made by NF circuit design block. When the LS30N numerical control lathe manufactured by Okuma Tekko Co., Ltd. is used, the voltage level of the AE signal that is normally detected is very small, tens to hundreds of microvolts, so the AE signal is amplified by the amplifier by a total gain of 80 dB. Table 1 shows the experimental conditions.

FIG. 9 shows the transition of the detected waveform due to tool wear. (A) is a detection waveform in the initial state of cutting. (B) is a detected waveform when the tool life is approaching, and (c) is a detected waveform when the tool life is approaching. From these results, it can be seen that the detected waveform has a small amplitude in the initial stage, but the amplitude of the waveform increases as the tool wear increases.

Fig. 10 shows the frequency analysis of the waveform that changes with tool wear. (A) is an initial result, (b) is an intermediate result, and (c) is a frequency analysis result when the tool life is approaching. Looking at those changes, there is a peak value at 1 MHz in the initial stage. The state when approaching the tool life is the same 1MHz
Is at the peak. However, the frequency level is increasing overall. From these, it is considered that the data for predicting the life cannot be obtained even if the change in the frequency band is examined.

FIG. 11 shows changes in the number of events and tool surface roughness due to tool chipping. The number of events tended to decrease as the wear of the tool increased, but a large increase was observed after the end of the period considered as the life. Looking at the state of the work material at that time, there is a step, and the one examined by a roughness meter is in the upper circle. From these facts, it can be inferred that a large amount of chipping has occurred when a large increase phenomenon is seen after the tool is considered to have reached the end of its life.

Fig. 12 shows a diagram in which the oscillation increases due to tool wear on the left, and a more detailed diagram on the right. The vertical axis of the detailed diagram is the number of waveforms. The horizontal axis is the number of oscillations in one waveform. The part (2) in the left figure is the initial state of cutting. The number of oscillations is still small. Looking at the detailed diagram (2 ') on the right side at that time, it is noticeable that many of the waveforms have a small number of oscillations. On the other hand, when looking at the detailed diagram (9 ') corresponding to the portion (9) where the number of oscillations is maximum from the left figure, the number of oscillations included in one waveform increases. You can see how it looks. In other words, it can be seen that the number of large waviness waveforms increases as the tool life approaches.

(Reliability of Tool Life Judgment) Table 2 shows the number of events and the number of oscillations and the ratio thereof until the tool life is reached by performing a cutting experiment with real-time analysis software.

It can be clearly seen from Table 2 that the calculated values increase with the passage of sampling time. 3 when the calculated value exceeds 20
The tool life was considered when the number was displayed. The roughness of the finished surface also deteriorates after this time, so it can be considered that the tool life can be considered from the perspective of the finished surface roughness. When the flank wear of the tool was examined, it was about 0.3 mm.

In Table 3, the level setting (value that predicts the life by the ratio of the number of events and the number of oscillations) is set to 20, and values of 20 and above are 3
The time when the individual pieces are displayed is predicted as the service life, the cutting is stopped, and the tool escape wear at this time is displayed.

I tried to estimate the variation of these tool escape wear amounts.

Based on Table 3, estimation was performed by a statistical method in order to investigate the variation state of the lateral flank wear. The difference between the tool wear width and the standard value can be estimated by the following formula.

If the reliability is 95%, then μ ₀ (α) = 1.96. σ =
Calculated with 0.0132, = 0.299, μ ₀ = 0.3mm, and n = 10, with a 95% reliability, -0.009 against the standard value (0.3mm)
It was presumed that it was in the range of 2 to +0.0072. From the above, use the real-time analysis software to set the level setting value.
If the tool life is judged to be 20, the flank wear width is 0.3 mm
It is considered that the tool life can be judged within the range of ± 0.01. Also, when examining the finished surface roughness, when the flank wear width exceeds 0.3 mm in the work material S45C, the finished surface roughness tends to deteriorate, so the flank wear width is reduced to 0.3 mm as a tool life. It is considered suitable to set.

<Prediction device-liquid type AE sensor> The liquid type AE sensor 1 is a liquid container 1c filled with a liquid 1b.
The wave-receiving body 1d of the solid-state AE sensor 1a provided so far is penetrated into the inside, and the liquid container 1c and the solid-state AE sensor 1a are joined in a liquid-tight manner.

An adhesive surface 1f to be adhered to a tool via an adhesive is formed on the outer surface portion of the liquid container 1c facing the wave receiver 1d.

And, as a preferred embodiment thereof, the solid-state AE sensor 1a is liquid-tightly joined to one flat surface of the cylindrical liquid container 1c, and the attachment surface 1f to the tool is attached to the outer surface portion of the opposing flat surface.
The solid-state AE sensor 1a and the cylindrical liquid container 1c which are formed except the adhering surface 1f are covered with an air jacket 1e.

(Operation) The operations (1), (2), and (3) will be described with reference to the diagram of the AE signal measurement system shown in FIG.
The AE signal is transmitted through the tool and caught by the AE sensor.
The AE sensor changes a sound wave signal into an electric signal. This electric signal is amplified by an amplifier, and a certain amount of noise is eliminated by the minimum delay function. The noise-removed AE electrical signal is captured without missing the number of events and oscillations due to the floating function with threshold settings.

This AE signal is processed by OP-01 in real time, the processed data is received and calculated, the number of events and the number of oscillations are taken in at regular intervals, and the ratio is displayed moment by moment. When this calculated value exceeds the level setting, the tool life can be determined by the number.

In addition, the AE sensor can effectively eliminate noise by utilizing the property of transmitting only the longitudinal wave of liquid. FIG. 14 shows a conventional solid-state AE sensor, and FIG. 15 shows a liquid-type AE sensor of the present invention. Comparing the figures, Figure 15 shows that the number of events is more stable and the number of oscillations is also increasing.

[Advantage of the Invention] The present invention uses a liquid type AE sensor and can eliminate noise signals such as cutting fluid and cutting chips by the minimum delation function, and stabilizes the number of events and the number of oscillations by the floating function. The tool life can be determined by quantifying the ratio of the number of events per unit time to the number of oscillations per unit time. By discriminating these signals and simultaneously applying the instantaneous calculation display of data, the tool life can be properly judged in real time.

In this way, the degree of tool wear and damage can always be properly grasped and managed, so the production accuracy and stability of the processed product can be improved and quality can be improved. It enables unmanned driving.

Therefore, the productivity of the integrated production system is dramatically improved.

Also, if a liquid type AE sensor improved for longitudinal waves is used, the life judgment will become clearer and effective.

[Brief description of drawings]

Fig. 1 shows 1 event of AE signal, Fig. 2 shows the number of events and oscillations, Fig. 3 shows discrimination by minimum deceleration, Fig. 4 shows change of number of events and oscillations by minimum deceleration, 5 Fig. 6 shows the relationship between the detected waveform and the threshold value, Fig. 6 shows the relationship between the detected waveform and the threshold value by adding the floating function, Fig. 7 is a flowchart of the real-time analysis software, Fig. 8 is the real-time analysis screen, and Fig. 9 is Changes in detected waveform due to tool wear, Fig. 10 changes in frequency analysis due to tool wear, Fig. 11 changes in number of events due to tool chipping, Fig. 12 changes in oscillation due to tool wear, and Fig. 13 AE signal Fig. 14 shows the number of events and oscillations by the conventional solid-state AE sensor, and Fig. 15 shows the number of events and oscillations by the liquid-type AE sensor of the present invention. Shon number, FIG. 16 is a sectional view of a liquid-type AE sensor of the present invention. 1 ... Liquid type AE sensor, 1a ... Conventional solid state AE sensor, 1b ... Liquid, 1c ... Liquid container, 1d ... Wave receiver, 1e ...
… Air jacket, 1f… Adhesive surface.

 ─────────────────────────────────────────────────── ─── Continuation of front page (56) Reference JP-A-57-205049 (JP, A) JP-A-56-73345 (JP, A) JP-A-60-242364 (JP, A) Actual development Sho-58- 14160 (JP, U) JP 60-49541 (JP, B2)

Claims (1)

(57) [Claims] 1. An AE longitudinal wave signal is obtained by a liquid type AE sensor, and a signal whose duration of this AE longitudinal wave signal is smaller than a set value is regarded as electrical noise and is discarded. Event that exceeds the threshold value by setting a certain threshold voltage to the average value of the waveform extracted from the minimum duration function and the minimum duration function that sets the minimum duration that is not captured. It has a floating function to take out a stable AE by the minimum delation function and the floating function.
Simultaneously with the acquisition of the longitudinal wave signal, the number of events is taken out at regular intervals, and the degree of tool wear / damage is detected by the number of calculated values calculated by dividing the sum of the frequency of each event by the number of events. A method for determining the life of a machine tool, which comprises determining the life.