JPH03221354A - Abnormality predicting device for rolling bearing - Google Patents

Abstract

PURPOSE:To invariably monitor a machine in continuous operation by removing the same frequency component as the reference data from the automatic power spectrum of the measured data which are the output signals of an acceleration sensor, then determining the maximum value to obtain the quantitative abnormal ity predicting danger signal. CONSTITUTION:When the rolling bearing of a machine is in the normal state, a no-load operation is performed at the constant rotating speed (n) rpm of a main spindle 1, and the acceleration signal is sampled N times at the interval DELTAt sec to obtain the automatic power spectrum Poo(i). The maximum value P'oo(i) is calculated by considering errors such as the main spindle rotating speed error and temperature error and stored in the memory of a computer 8 as the reference data. The acceleration signal is detected by quite the same method as the method to obtain the reference data at the time of abnormality monitoring, the automatic power spectrum Pxx(i) is determined, the performance function is determined to remove the same frequency component as the reference data from it, and its maximum value S is determined and quantitatively outputted as the abnormality predicting danger signal of the rolling bearing.

Description

DETAILED DESCRIPTION OF THE INVENTION <Industrial Application Field> The present invention relates to an abnormality prediction device for rolling bearings, and is useful when applied to various machine tools.

<Conventional technology> Diagnosis of abnormalities in rolling bearings has been performed by vibration analysis using acceleration sensors. To further comment, the abnormality diagnosis in the conventional technology involves importing the acceleration signal, which is the output signal of the acceleration sensor, into a computer, performing predetermined signal processing, and converting the resulting auto power spectrum into a CR.
This was done by displaying the information on the T and having the worker monitor this display.

<Problems to be Solved by the Invention> In the above-mentioned prior art, a worker with specialized knowledge must view and make a judgment on the signal processing results of the computer. Therefore, in order to constantly monitor abnormalities in rolling bearings in machines that operate continuously, workers with specialized knowledge must constantly monitor and judge the processing results on the CRT, which is virtually impossible. It's impossible.

In view of the above-mentioned prior art, the present invention provides a rolling bearing abnormality prediction device that enables constant monitoring of machines in continuous operation by determining the presence or absence of abnormalities in rolling bearings through computer processing without human judgment. The purpose is to provide.

<Means for solving WIWIi> The configuration of the present invention that achieves the above object includes an acceleration sensor fixed to a bearing box housing a rolling bearing and detecting vibration of the bearing box, and an acceleration sensor that detects when the rolling bearing is in a normal state. The auto power spectrum of the data obtained by sampling a predetermined number of acceleration signals, which are the output signals of the acceleration sensor during driving, is determined and stored, while the actual measurement data, which is the output signal of the acceleration sensor during driving The present invention is characterized in that it has a computer that obtains an auto power spectrum of , further removes the same frequency component as the reference data from this auto power spectrum, and then obtains the maximum value thereof and uses it as a quantitative abnormality prediction danger signal.

Effects> According to the present invention having the above configuration, the abnormality prediction danger signal becomes a signal representing a peak of a frequency component different from normal vibration when the reference data is measured. Therefore, the abnormality prediction danger signal is a signal whose magnitude differs depending on the degree of abnormality in the rolling bearing.

<Example> Embodiments of the present invention will be described in detail below based on the drawings.

As shown in FIG. 1, a main shaft 1 of a machine tool (the main body is not shown) is rotatably supported by a rolling bearing housed in a bearing box 2, and is rotationally driven by a motor 3. .

The acceleration sensor 4 is fixed to the bearing box 2 and sends out an acceleration signal that reflects vibrations of the bearing box 2. This acceleration signal is supplied to a bandpass filter 6 via a charge amplifier 5, which cuts unnecessary frequency components, and then supplied to a computer 8 via an A/D converter 7. The computer 8 performs signal processing expressed by the following algorithm.

First, when the rolling bearings of the machine are in normal condition, the main shaft 1
No-load operation is performed at a constant rotational speed (nrpm), and N acceleration signals of Δ are sampled at intervals of seconds. Find the auto power spectrum P0° (i) from the t data obtained from this. Tatami i = 1.2, 3゜In this example, abnormality prediction is determined by comparing data measured later with normal data, but the main axis is the difference between the data measured later and the data measured later. 8 errors such as rotation speed error and temperature difference! ! Then, P. . Perform the following transformation on fil and get P'. . Calculate (1).

(Conversion) Maximum value P' thus obtained. . (1) is stored in the memory of the computer 8 as reference data.

When monitoring an abnormality, an acceleration signal is detected using exactly the same method used to obtain reference data. In other words, the spindle rotation speed n rpm
A constant no-load operation is performed, and N acceleration signals of Δ are sampled at intervals of seconds. The auto power spectrum P, , (i) is determined from the data obtained as a result of B. However, i=1.2,3° Next, this P. In order to remove the same frequency components as the reference data from 8(1), the evaluation function P,,X(i) is changed to P,,Jil
-P,,4i1-P'. . (Find it using i+.

Next, the maximum value of the evaluation function P, .

The computer 8 quantitatively outputs this S as a rolling bearing abnormality prediction risk signal.

The controller 9 on the machine side receives the quantitative abnormality prediction risk level signal output from the computer 8 and automatically performs measures according to the risk level, such as decelerating or stopping the machine, by controlling the motor 3.

<Effects of the Invention> As specifically explained above with the embodiments, according to the present invention, an abnormality can be automatically predicted without human judgment,
It should be possible to take measures such as automatically slowing down or stopping the machine. In addition, machine maintenance workers can predict abnormalities before they reach an abnormal state that requires a complete stop of the machine, allowing them to take early measures such as arranging replacement parts, and reducing the amount of time the machine will be stopped due to trouble. can be shortened.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention. In the figure, 1 is the main axis, 2 is the bearing box, 3 is the motor, 4 is an acceleration sensor, 8 is a computer.

Claims (1)

[Claims] An acceleration sensor fixed to a bearing box housing a rolling bearing and detecting vibrations of the bearing box; and an acceleration signal that is an output signal of the acceleration sensor detected when the rolling bearing is in a normal state are transmitted at regular intervals. The auto power spectrum of the data obtained by sampling a predetermined number of items is determined and stored, while the auto power spectrum of the actual measurement data, which is the output signal of the acceleration sensor during driving, is determined, and the above-mentioned standard is further determined from this auto power spectrum. 1. An abnormality prediction device for a rolling bearing, comprising: a computer that removes frequency components that are the same as the data, then obtains the maximum value thereof and uses it as a quantitative abnormality prediction danger signal.