JPS61234353A - Removal of noise in acoustic emission measuring system - Google Patents

Abstract

PURPOSE:To remove noises in a measuring system, by passing a detection signal of an acoustic emission generated from a rotor through a parallely branched waveform processing circuit into which a waveform parameter is inputted. CONSTITUTION:An acoustic emission (AE) signal is detected with a sensor 2 from a bearing 1 or the like. The signal is applied to a multiplexer 7 through a filter 4, a threshold level setting circuit 6 and the like and inputted continuously into a parallely branched waveform processing circuit 8 in distribution. The waveform processing circuit 8 is so formed that a plurality of waveform parallel meters, for example, DELTAt, N, Vp/DELTAt and the like when the time to the peak is represented by DELTAt, the number of cycles N and the peak value Vp. Thus, the AE signal can be distinguished from noises by waveform parameters thereby enabling accurate measurement of the AE value by removing the noises.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application This invention is used to detect AI (hereinafter referred to as acoustic emission) generated from a rotating body with a 1M sensor and measure the amount of Arc. This invention relates to a method for removing noise from outside the measurement system.

Conventional technology Conventionally, A11! How to remove noise in measurement system
The equipment included the following:

(1) As disclosed in Japanese Patent Application Laid-open No. 114980/1983, as shown in FIG.
A function is to install a dummy sensor 1 that detects noise other than that, input the electrical signal output from the dummy sensor f as a gate signal, and conduct or interrupt the electrical signal output from the AX sensor a according to this input. It has an electronic switch k with built-in.

In addition, in Fig. 6, 1) y g...Preamplifier %a
.

h...filter, modified, 1...main amplifier, 1.
j...Detector, 1...Output terminal.

(2) As disclosed in Japanese Unexamined Patent Publication No. 57-100348, an AI sensor aK as shown in FIG.
Therefore, the detected signal is preamplifier b - main amplifier d
Discriminator at a threshold level below the noise level of 1 preamplifier b! When it is discriminated by IIK, it becomes a pulse train. The comparator sets the appropriate value of the count number and threshold level value during a certain period of time of this pulse train to discriminate the Al signal.

In FIG. 7, n: a counter, p: a processing circuit such as a time difference measuring circuit.

Problems to be Solved by the Invention However, in the above-mentioned prior art l, the AΣ sensor &9
f1 preamplifier b, g% In addition, a pair of each of ter C1h, main amplifier d, 1, and detector etJ are required, and while the gate signal obtained from dummy sensor b is input to electronic switch k, Aln There is a drawback that the signal is not processed.O In conventional technology 2, if the AR sensor a has a resonant frequency band, it is difficult to distinguish between the count number and threshold level value of the pulse train of the Al signal and those of noise. There was a problem that there was.

Means for solving the problem By detecting the hx generated from the oscilloscope with the A1 sensor, and passing the Aln signal output from the sensor through a waveform processing circuit branched in parallel into which waveform parameters are input. ) A method for removing noise in an am measurement system is provided.

Al generated from the working rotating body! i is detected by an AIC sensor, and the sensor outputs an Al signal. This Aln signal is passed through a conventional filter, amplifier, and discriminator, and then passed through a parallel branched waveform processing circuit into which waveform parameters are input.
It is possible to discriminate only the Al signal and remove noise.

This is because the waveform of the Aln signal shown in Figure FA2 is clearly different from the waveforms of the electrical noise and mechanical noise shown in Figures 83 and 4, so by utilizing the difference between these waveforms, the waveform is input to the above waveform processing circuit. For example, if the time to reach the peak is Δt1, the number of cycles is N, and the peak value is Vp, then Δt.

Store the 1st prize in vp vPl 's / - / e'',
This is because signals outside the range of waveform parameter values specific to the Δt Δt Ale signal can be determined to be noise and removed.

Embodiments Below, embodiments of the present invention will be explained based on the drawings. In FIG. Low frequency Wk components such as mechanical vibrations are cut and further amplified by the main amplifier 5.

All amplified by main amplifier 5! An upper threshold level setting circuit 6 allows only components exceeding a suitable threshold to pass through, and a multiplexer 7 distributes and inputs them continuously without interruption to a parallel branched waveform processing circuit 8.

The waveform processing circuit 8 has a plurality of waveform parameters in advance, such as the time to peak Δt and the number of cycles N1.
When the peak value is VP, Δt.

Memorize N・su4t・7p″4t・4, etc.

The AI signal 1 electric noise and mechanical noise each have different waveforms as shown in Figures 2 to 4. For example, the time Δt until reaching the peak value is the AI signal 1 electric noise and mechanical noise each value. are respectively Δt1 and Δt3. Δt
-, then Δ1. <Δ1. <△1. There is a relationship.

Therefore, it is possible to distinguish between the fihx signal and noise based on each of the waveform parameters described above. 1. Signals whose waveform parameter values are within a certain range close to those of the A11i signal are recognized as AI multiplied signals, and others are judged to be noise. By continuously inputting only the ˜Am signal to the AID processing circuit 9, noise can be removed and the 51 quantity can be accurately measured.

FIG. 5 shows a comparison diagram between Al amount 2 measured by the conventional measuring method and *xill measured by the measuring method of the embodiment of the present invention. Alj
Ll is smaller than AI amount 2 due to noise removal, but the rising edge appears more clearly and J) sAI
It can be determined that the amount was measured more accurately. Therefore, according to the measuring method of the embodiments, it is possible to predict the life of the bearing more accurately.

Effects of the Invention Since the A1c signal is routed through parallel branched waveform processing circuits into which waveform parameters are input, noise can be removed continuously without interruption and accurate AXI! k can be measured.

Also - A1 because no dummy sensor is required! ! There is no need to provide sensor amplifiers, filters, etc. in pairs, and the number of parts is advantageous.

[Brief explanation of drawings]

Figure 1 is a block diagram of an embodiment of this invention, Figure 2 is an AI
Figure 3 is a measurement diagram showing the waveform of the ++ signal, Figure 3 is a measurement diagram showing the waveform of electrical noise, Figure 4 is a measurement diagram showing the waveform of mechanical noise, and Figure 5 is a measurement diagram showing the AI measurement amount according to the embodiment of this invention and the conventional 6 is a block diagram of conventional example 1, and FIG. 7 is a block diagram of conventional example 2. 0 1... (bearing) - rotating body 2... 1 Sensor 8...Waveform processing circuit Figure 3 Figure 4 Figure 5

Claims (1)

[Claims] The acoustic emission sensor detects the acoustic emission generated from the rotating body,
A method for removing noise in an acoustic emission measurement system by passing an acoustic emission signal output from the sensor through parallel branched waveform processing circuits into which waveform parameters are input.