JPS644611B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to objects that perform periodic motion such as bearings and gears, that is, periodic moving bodies, and a method for monitoring equipment equipped with the same.

In general, when abnormalities such as scratches on parts, eccentricity of the rotating shaft, poor lubrication, etc. occur in periodic moving bodies such as bearings and gears, and equipment equipped with these,
If this is left untreated, it will not only cause damage to parts such as bearings and gears, but also cause failure and destruction of the entire machine equipped with it. Accurately identifying such abnormalities is therefore an extremely important issue in the maintenance and management of equipment including bearings, gears, and the like. Conventionally, in order to accurately grasp such abnormalities, a sensor is attached to a periodic moving body, and from the time series data obtained from the output signal, the effective value, that is, as shown in equation (1) below, is calculated.
Calculate the RMS (Root Mean Square) value and use its analysis to diagnose objects to be monitored, including periodic moving objects such as bearings and gears, as well as equipment equipped with these.
Surveillance was taking place.

However, x(t): Time series data (t=1, 2...N) N: Number of data: Average value of time series data The RMS value is useful for understanding the size of the monitored object, the magnitude of the load,
Since it differs depending on the rotational speed, etc., and it is not possible to set a universal judgment standard, it is necessary to accumulate data on each individual normal condition. Furthermore, when determining subtle abnormalities, low sensitivity is a problem in that even though there may be fluctuations between frequency components, the RMS value of the entire signal does not change.

The present invention was made in view of the above circumstances, and it is possible to grasp the overall degree of deterioration of the monitored object, and also to evaluate it using unbiased judgment criteria, and also to detect subtle abnormalities. It is an object of the present invention to provide a method for monitoring periodic moving objects that can be determined with high sensitivity even when the object is moving.

The method for monitoring a periodic body according to the present invention is based on time series data obtained by sampling the vibrations of a periodic body at a constant period, and calculates its effective value and the fourth moment of its probability density function as its variance. It is characterized by detecting abnormalities by determining the index normalized by dividing the periodic body and the bicoherence determined for an arbitrarily selected frequency, and by combining these to comprehensively judge the degree of deterioration of the periodic moving body. shall be.

Among the above-mentioned values, the effective value is advantageous in understanding the overall degree of deterioration of the monitored object as described above, and the fourth moment of the probability density function of time-series data can be used to calculate the variance of time-series data. The normalized index divided by Anomalies can be detected even when time-series data contains noise that is difficult to distinguish. Therefore, the method for monitoring a periodic moving body according to the present invention effectively combines the respective advantages of these methods to comprehensively determine the degree of deterioration of the periodic moving body.

Next, bicoherence will be explained. Conventionally, power spectra have been used for diagnosis, but the phase information is not used, and therefore it is difficult to detect minute faults in which ambient noise cannot be removed and only the phase relationship between frequency components changes. It is difficult. This is where bispectral diagnosis is performed.

Note: _The n-th spectral density function _is the (n- ₁ )-dimensional Fourier transform ^of the n-th ^correlation _{function .} K _o , x(t ₁
,…,
t _o-1 )×exp{−i2π(f ₁ t ₁ , +…+f _o-1 t _o-1 )}dt ₁
…
_The spectral density function when n=2 is called the power spectrum, and the one when n=3 is called the bispectrum.

When the gear device is driven steadily, its acoustic signal x(t) is x(t) = _∞ 〓 ⁿ⁼⁰ a _o cos(2πnf ₀ t+φn)+n(t) …() where n(t ): Gaussian ambient noise f ₀ : Rotational fundamental frequency The power spectrum of equation () is So, this bispectrum B(f ₁ , f ₂ ) is However, <> is an average operation δ: Delta function () From the () formula, the following properties are extracted.

(1) Unlike the power spectrum P(f), the bispectrum B(f ₁ , f ₂ ) is a complex number and therefore has phase information as well as amplitude.

(2) It is completely unaffected by additive Gaussian noise.

(3) Represents the dependence of three frequency components whose frequencies are f ₁ +f ₂ +f ₃ =0, and whose phase reflects the phase relationship between these components.

Therefore, bispectral analysis allows detection of minute abnormalities that change only the phase relationship between frequency components. This detection usually uses bicoherence, which is obtained by normalizing the bispectrum with the power spectrum {see equation (2)}.

The method of the present invention will be explained in detail below with reference to the drawings. 1st
The figure is a schematic block diagram of a device used in the implementation.A vibration generator 1 is equipped with a vibration detection device 2 that detects the vibration and converts it into an electric signal. The output of this vibration detection device 2 is input to a sampling device 3, where it is sampled at a constant cycle, converted from analog data to digital data, and stored in a storage device 4. Time series data x(t) stored in this storage device 4
is respectively taken into the RMS value calculation device 5, the bicoherence calculation device 6, and the K value calculation device 7, and the calculation processing described below is performed by each calculation device.

In the clogged RMS value calculation device 5, the time series data x(t) is calculated according to the above equation (1) to obtain an effective value, that is, an RMS value. In addition, in the bicoherence calculation device 6, the calculation according to the following equation (2) is performed, and the bicoherence B _ic.XXX
(f ₁ , f ₂ ) are found.

However, f ₁ , f ₂ : Two frequencies whose relationship is to be investigated B _XXX (f ₁ , f ₂ ): Bispectrum obtained by Fourier transform of the cubic correlation function obtained from time series data x(t). That is, S _XX (f): Power spectrum at frequency f, i.e. S _XX (f) = 1/TX(f)・X ^* (f) T: Data acquisition period X(f): Fourier transform of original sequence data x(t) That is, x(f)=∫ ^∞ _-∞ x(t)e ^-j2 〓 ^ft dt *: Conjugate complex number Further, in the K value calculation device 7, calculation is performed according to the following equation (3), and time series data x(t) A normalized index, that is, the K value, is obtained by dividing the fourth moment of the probability density function by the variance of the time series data x(t). Note that the central moment is used as the moment.

However, P(x): Probability density function of time series data x(t) μ ₄ : Fourth moment σ ² : Variance of time series data x(t) The thus obtained RMA value, bicoherence, and K value The related data is inputted to the comparison circuit 9 together with the data related to the normal level and the lifespan level stored in the reference data storage device 8, and as a result of the comparison therein, if the monitored object is comprehensively determined to be abnormal, the data is sent to the alarm device 10. Signals are being sent and warnings are being raised.

When a periodic moving body is monitored using such a device, excellent monitoring of the periodic moving body as described below becomes possible. That is, the RMS value calculated by the RMS value calculation device 5 is advantageous in understanding the overall degree of deterioration of the monitored object as described above, and the bicoherence calculated by the bicoherence calculation device 6 is useful for detecting subtle abnormalities. It is particularly excellent in detecting the K value calculated by the K value calculation device 7.
The value is effective for making an absolute evaluation that is not affected by the size of the object to be monitored, the magnitude of the load, the rotation speed, etc., and when implementing the method of the present invention using the above-mentioned device, it is necessary to combine these values to make a comprehensive evaluation. Therefore, it is possible to monitor periodically moving bodies with high reliability.

In this embodiment, the RMS value calculation device 5, the bicoherence calculation device 6, and the K value calculation device 7 perform calculation processing, respectively, but all calculation processing is performed by one calculation device. It's okay.

Next, an example will be described in which the above-mentioned device is used to monitor a speed reducer that reduces and transmits power using a plurality of gears supported by bearings. Normal condition, no-lubrication condition (Case 1), condition with small scratches on the gear (Case 2), condition with medium scratches on the gear (Case 3), and condition with large scratches on the gear (Case 4) ) RMS value for each state,
Graphs summarizing the results of calculating bicoherence and K value are shown in FIGS. 2, 3, and 4. Looking at the RMS value shown in Figure 2, its absolute value cannot be evaluated unless there is a standard corresponding to the degree of deterioration of the reducer, but the RMS value increases according to the degree of deterioration, and the deterioration of the monitored object This clearly shows that it is advantageous in understanding the overall situation. Also, looking at the bicoherence shown in Figure 3,
The difference in the size of scratches on the gear is clearly visible, and it can be seen that this method is particularly good at detecting subtle abnormalities. Furthermore, looking at the K value shown in FIG. 4, its absolute value is 3.0 in normal times, and tends to deviate from 3.0 when an abnormality occurs, indicating that it is effective for absolute evaluation.

The reason why the K value decreased in Cases 3 and 4 despite the system deterioration was because the torsional natural vibration of the gear of the reducer became large and a significant peak was formed in the power spectrum.

If it is a sine wave with one peak in the power spectrum, the K value is 1.5. Generally, if the power spectrum has a significant peak, the K value is lowered, so in order to properly evaluate it, it is necessary to It is necessary to perform filtering and calculate the K value. Therefore, by making an absolute evaluation using the K value, understanding the overall degree of deterioration using the RMS value, and detecting subtle abnormalities using bicoherence, highly reliable periodic motion can be achieved. It becomes possible to monitor the body.

As detailed above, in the case of the present invention, based on time series data, its effective value (RMS value) and the index (K Abnormality is detected by comprehensively determining the degree of deterioration of a periodic moving object by combining the bicoherence determined for an arbitrarily selected frequency, which enables highly reliable monitoring of periodic moving objects. becomes. Therefore, the present invention makes a significant contribution to the improvement of abnormality detection technology for periodic moving bodies.

[Brief explanation of the drawing]

Figure 1 is a schematic block diagram of the apparatus used to carry out the present invention, and Figure 2 is a diagram showing the calculation results using the method of the present invention.
FIG. 3 is a graph showing RMS values, FIG. 3 is a graph showing bicoherence calculated using the method of the present invention, and FIG. 4 is a graph showing K values calculated using the method of the present invention. 1... Vibration generator, 2... Vibration detection device, 5...
... RMS value calculation device, 5 ... Bicoherence calculation device, 7 ... K value calculation device, 9 ... Comparison circuit.

Claims (1)

[Claims] 1. Based on time series data obtained by sampling the vibrations of a periodic body at a constant period, the effective value and the fourth moment of the probability density function are divided by their variance to be normalized. A periodic moving body characterized in that an abnormality in the periodic moving body is detected by determining the deterioration degree of the periodic body by comprehensively determining the degree of deterioration of the periodic moving body by determining the calculated index and the bicoherence obtained with respect to an arbitrarily selected frequency. monitoring method.