JPS5963529A - How to diagnose rotating machinery - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for diagnosing rotating machines in dog-type plants such as plants.

Rotating machines account for a very high proportion of large plants and equipment such as LNG plants, and their role is also important. When abnormalities or breakdowns occur in these rotating machines, not only will there be economic losses, but if the accidents escalate, it will develop into a major social problem.

For example, LNG plants are used as a supply source for city gas, cold power generation equipment, etc., so if a failure occurs in the LNG pump that pumps LNG from the tank, it will cause a large loss to the user side and also cause damage to general users. become.

Mechanical diagnostic technology in Japan began in 1950, when the Shinkansen began operation and the issue of pollution began to gain attention.
Although full-scale efforts began in the 1960s, the current situation is that the criteria for making such decisions have not yet been generalized.

The present invention was developed against this background, and aims to establish a technology for predicting failures of rotating machinery in plants, improve plant reliability, reduce maintenance costs, and homogenize maintenance techniques. That is.

The present invention will be described in detail below with reference to the drawings.

To carry out the machine diagnosis method of the present invention, an acceleration detector is installed at an appropriate position on a rotating machine, and a microcontroller is detected based on the signal from the detector.

A vibration spectrum in a predetermined frequency range at a certain point in time is created using a user system or the like, and the spectrum is stored and displayed on a required display device such as a monitor television.

If we add the subscript H to the nth spectrum in a given frequency domain, then the nth acceleration spectrum is α(,), the nth velocity spectrum is v,(n), and the nth amplitude spectrum is x(?). I) The nth angular velocity is ω(?L), the n@th frequency is 1 (represented by n3, the spectrum width is Δf, the sampling frequency is fs, the number of sampling data is represented by N8, and the relationship between the above spectra is Power as shown below.

ω(W)=2πf(n) f(n)=zΔf Generally, when vibration of a rotating machine is detected by an acceleration detector, its spectral value increases as deterioration progresses.

The following describes the case where the acceleration spectrum is used as an index among the vibration spectra.

Here, if the total width of the frequency domain required for diagnosis is 10 KHz, the number of sampling data points is 1024 points, and the sampling frequency is 20 KHz, the total number of spectral lines is 512, and the width of the acceleration spectrum is (-y-)/(L!7-ri). : 1
9.53H2, and if an acceleration spectrum is created under these conditions, it is sufficiently effective as an index for fault diagnosis in ordinary rotating machines.

FIG. 1 shows the acceleration spectrum (→ distribution) of the rotating machine obtained under the above conditions, with the horizontal axis representing the frequency (maximum 1'0 KH2) and the vertical axis representing the acceleration (G).

In the present invention, the entire width of the acceleration spectrum (b) is divided as required (10 divisions in the figure), and the average acceleration of each section (the middle 5
The average value of one spectrum is determined, and a partial abnormality in a rotating machine is detected based on the change in the average acceleration over time, and a failure diagnosis is performed.

The section frequency f divided here. -710% The corresponding section average acceleration is L1~LIO, and the average acceleration is 1~LI
Display O as a time series characteristic on the same display screen (second
figure).

In general, deterioration of a machine usually begins after the test run is completed, the machine stabilizes, and has been operated for a long time.As shown in Figure 2, the average acceleration gradually changes from the start of operation, and the figure shows a case where it increases. (test run region), a stable state (stable region) where the value hardly changes, and then an abnormal state (damaged region) where the value increases further.

Therefore, the sequence characteristic of the section average acceleration is given by the section number =
When 1 to 10 are sequentially observed, if a sudden change is observed in the average acceleration of a certain numbered section, it can be quickly diagnosed and predicted that the components related to the frequency of that section have begun to deteriorate.

Also, using the value in the stable region of each section average acceleration as a reference, if the ratio of the section average acceleration (observed section average acceleration) calculated for each predetermined period with respect to the reference standard exceeds the required ratio, this ratio It is also possible to diagnose that there has been deterioration or damage to parts related to the section corresponding to the section average acceleration exceeding .

When such a judgment is made, an alarm is issued and appropriate countermeasures are taken, such as activating the automatic stop system.

As described above, in the present invention, since failure diagnosis is performed based on the change over time in the section average vibration of the rotating machine, it is possible to easily predict the time of failure and also estimate the location of deterioration.

Although the above embodiment uses an acceleration spectrum, it goes without saying that fault diagnosis can be performed in a similar manner using a velocity spectrum or an amplitude spectrum.

FIG. 3 is a block diagram of a preferred apparatus for carrying out the present invention, which will be briefly described below.

In the figure, (1) is a sensor for detecting acceleration, and (2) is a zero converter (
3) is a low-pass filter for reducing aliasing errors when converting to a digital signal, and the signal converted into a digital signal by the A/D converter (3) is inputted to the master microcomputer stem (4). A diagnostic condition setter (5) is connected to the master microcomputer system (4), and the diagnostic interval and the value of the average acceleration of each section in the stable region are used as the reference pupil, and the reference value that is the basis for diagnosis is used. In addition to setting and inputting the ratio to the observation interval average acceleration, the arithmetic processor (6), main memory (7),
Connect the auxiliary memory (8), the diagnostic display section (9), and the alarm device Ql to the master microcomputer system (4).

Master microcomputer system (4) u After performing averaging calculations to extract the characteristics of the signal from the A/D converter (3), digit number adjustment calculations to prevent underflow and overflow, etc. , arithmetic processor (6), auxiliary memory (8) as well as frequency analysis (FF
T) to convert the time axis signal into an acceleration spectrum on the frequency axis, further divide the acceleration spectrum into 10, calculate the average acceleration of each divided area, and compare it with a reference value. Here, the arithmetic processor (6) performs butterfly operations (multiplication, addition) in frequency analysis and square root operations to obtain a spectrum. The auxiliary memory (8) includes a microprocessor, a main memory (7), and an arithmetic processor (
6) Contains an operating system that assists the program, and also displays calculation results (average acceleration for each section)
is stored in memory. The master microcomputer system (4) rtiB displays the calculation result on the diagnostic display section (9) according to the diagnostic conditions inputted by the diagnostic condition setter (5).
), and if necessary, activates the alarm device 0@1 and outputs a signal 0η for starting the display, alarm, and automatic stop system.

In addition, the master microcomputer system (4) receives power via the semiconductor relay 01). ■It is connected to , and is energized and driven only at the required time (diagnosis time).

0:i id: Slave microcomputer system, the master microcomputer system (
4) A memory α and a clock are connected to the slave microcomputer stem o3,
In addition, a slave microcomputer system (l [available]
A battery backup circuit (JQ) connected to the power supply (2) is connected to the clock 00.

The slave microcomputer system α [with] is always operating as a countermeasure against power outage by a battery backup circuit (T), and is based on the diagnostic conditions from the master microcomputer system (4) based on the results measured by the clock α. , controls the diagnostic interval by driving the semiconductor relay αη at an appropriate time, memorizes and stores necessary data in preparation for a power outage after the diagnosis is completed, and then drives the semiconductor relay αη to control the diagnostic interval after the diagnosis is completed. / stem (
4) is stopped and then restarted 30 minutes before the next diagnosis.

As described above, the present invention establishes a failure diagnosis technique for rotating machines, and also makes it possible to reliably grasp the time of failure, thereby increasing the reliability of the plant and significantly reducing economic losses.

[Brief explanation of the drawing]

Fig. 1 is a diagram of the acceleration spectrum, which is an indicator of the failure diagnosis of the present invention, Fig. 2 is a diagram showing the time series characteristics of the interval average value of the acceleration spectrum, and Fig. 6 is a diagram showing the time series characteristics of the interval average value of the acceleration spectrum. 1 is a block diagram of a preferred apparatus for (1) i-sensor, (4) u-master microcomputer stem, (9) a diagnostic display section, and αQ a slave microcomputer system. Ishikawajima Harima Heavy Industries Co., Ltd. Toyosu General Office, 3-2-16 Toyosu, Koto-ku, Tokyo 0 Author: Kenji Sugishita Inside the Ishikawajima Harima Heavy Industries Co., Ltd. Toyosu General Office, No. 3-216 Toyosu, Koto-ku, Tokyo

Claims (1)

[Claims] 1) Detect the acceleration of the rotating machine at appropriate intervals, determine the vibration spectrum in a predetermined frequency range based on the detection results, and further divide the full width of the vibration spectrum as required to determine the section average vibration value for each section. , a method for diagnosing a rotating machine that determines an abnormality in the rotating machine based on a time-series characteristic in which the section average vibration value changes from a test run region, a stable region, and a damaged region.