Fault monitoring method for rotating equipment

Abstract

The present invention relates to the technical field of rotating equipment monitoring, and specifically to a method for fault monitoring of rotating equipment, comprising: step S1: acquiring the acceleration signal; step S2: performing spectrum editing and integration processing on the acceleration signal to generate a velocity signal and a displacement signal; step S3: judging whether the rotating equipment has a fault based on the velocity signal and the displacement signal. The beneficial effect of the present invention is that spectrum editing processing is performed based on sensor characteristics, the background noise in the acceleration signal is effectively suppressed, and interference components are effectively suppressed based on the signal background noise size and peak distribution characteristics, while retaining the effective signal. This method has low computational complexity, accurate results, and can be well adapted to online monitoring scenarios.

Description

Technical Field

[0001] The present invention relates to the technical field of rotating equipment monitoring, and in particular to a fault monitoring method for rotating equipment. Background Art

[0002] Currently, rotating machinery accounts for approximately 80% of my country's metallurgical, petroleum, and chemical industries. Many of these devices operate in harsh environments, such as high speeds and heavy loads, making them prone to failure. Unplanned downtime due to equipment damage causes significant losses to companies, making fault monitoring and diagnosis of these rotating machinery crucial.

[0003] Vibration signal-based monitoring technology is the most direct and effective method for rotating machinery fault diagnosis. Vibration caused by gear and bearing faults is easily captured by accelerometers. Rotor imbalance, misalignment, and looseness, among other issues, can easily excite low-frequency components such as the machine's rotational frequency and harmonics, corresponding to the vibration velocity or displacement signal frequency bands. Due to cost and other factors, the current mainstream data acquisition method uses a single sensor and then uses numerical integration or differentiation to obtain vibration signals of varying dimensions. Single or double integration of the acceleration signal can yield information such as the machine's velocity and displacement. However, existing methods face challenges due to signal distortion caused by DC components, temperature drift, unstable low-frequency sensor performance, environmental interference, and background noise. Time-domain integration cannot completely remove trend terms and is prone to error accumulation during the integration process. Frequency-domain integration methods are simple and can effectively avoid the cumulative amplification effect of time-domain integration on small error signals. However, they still cannot completely remove background noise components and tend to amplify low-frequency background noise, drowning out the true components. Summary of the Invention

[0004] In view of the above problems existing in the prior art, a fault monitoring method for rotating equipment is provided.

[0005] The specific technical solutions are as follows:

[0006] A method for monitoring a fault of a rotating device, wherein an acceleration sensor is provided on the rotating device for generating an acceleration signal, and the method comprises:

[0007] Step S1: obtaining the acceleration signal;

[0008] Step S2: performing spectrum editing and integration processing on the acceleration signal to generate a velocity signal and a displacement signal;

[0009] Step S3: judging whether the rotating device has a fault according to the speed signal and the displacement signal.

[0010] Preferably, the spectrum editing integration method in step S2 includes:

[0011] Step S21: preprocessing the acceleration signal to generate a signal amplitude sequence;

[0012] Step S22: processing the signal amplitude sequence to generate a speed signal;

[0013] Step S23: generating the displacement signal according to the speed signal.

[0014] Preferably, the step S21 includes:

[0015] Step S211: removing a DC component from the acceleration signal to generate a preprocessed signal;

[0016] Step S212: performing high-pass filtering on the preprocessed signal to generate the signal amplitude sequence.

[0017] Preferably, the step S22 includes:

[0018] Step S221: performing fast Fourier transform on the signal amplitude sequence to generate a spectrum amplitude sequence;

[0019] Step S222: performing spectrum editing on the spectrum amplitude sequence to generate a cropping sequence;

[0020] Step S223: performing inverse Fourier transform on the clipped sequence to generate the speed signal.

[0021] Preferably, the step S222 includes:

[0022] Step S2221: generating a first sequence of carpet values ​​according to the spectrum amplitude sequence;

[0023] Step S2222: performing spectrum editing on the spectrum amplitude sequence according to the first sequence of carpet values.

[0024] Preferably, step S23 includes:

[0025] Step S231: performing fast Fourier transform on the velocity signal to generate a velocity spectrum sequence;

[0026] Step S232: performing spectrum editing on the velocity spectrum sequence to generate a velocity edited sequence;

[0027] Step S233: Perform inverse Fourier transform according to the velocity editing sequence to generate the displacement signal.

[0028] Preferably, in step S212, the cutoff frequency of the high-pass filter is the low-frequency response cutoff frequency of the acceleration sensor.

[0029] Preferably, in step S222, the spectrum editing method is:

[0030]

[0031] Where: C j is the cropped sequence, B j is the spectrum amplitude sequence, df is the spectrum resolution, fs is the sampling frequency, M is the frequency offset of the effective peak, α is the coefficient of the effective peak relative to the mean amplitude of the adjacent frequencies, and β is the ratio of the effective amplitude to the noise carpet value.

[0032] Preferably, the frequency offset value of the effective peak is 4, 6 or 8;

[0033] The effective peak relative to the adjacent frequency amplitude mean coefficient has a value range of 1.5 to 3;

[0034] The ratio of the effective amplitude to the noise carpet value ranges from 6 to 10.

[0035] The above technical solution has the following advantages or beneficial effects: Spectral editing based on sensor characteristics effectively suppresses the noise floor in the acceleration signal. Interference components are effectively suppressed based on the noise floor and peak distribution characteristics of the signal, while retaining the valid signal. This method has low computational complexity, accurate results, and is well suited for online monitoring scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The embodiments of the present invention will be described more fully with reference to the accompanying drawings, which are provided for illustration and description only and are not intended to limit the scope of the present invention.

[0037] Figure 1 It is a frequency domain image of a signal after conventional time domain integration processing in the prior art;

[0038] Figure 2 It is a frequency domain image of a signal after conventional frequency domain integration processing in the prior art;

[0039] Figure 3 This is a schematic diagram of the overall method in an embodiment of the present invention;

[0040] Figure 4 This is a time domain diagram of the original acceleration signal in an embodiment of the present invention;

[0041] Figure 5 This is a frequency domain diagram of the original acceleration signal in an embodiment of the present invention;

[0042] Figure 6 Schematic diagram of an integration method based on spectrum editing in an embodiment of the present invention;

[0043] Figure 7 This is a schematic diagram of sub-steps of step S21 in an embodiment of the present invention;

[0044] Figure 8 This is a schematic diagram of sub-steps of step S22 in an embodiment of the present invention;

[0045] Figure 9 This is a frequency domain diagram of a speed signal in an embodiment of the present invention;

[0046] Figure 10 This is a time domain diagram of a speed signal in an embodiment of the present invention;

[0047] Figure 11 This is a partial enlarged view of the speed signal in an embodiment of the present invention;

[0048] Figure 12 This is a schematic diagram of sub-steps of step S222 in an embodiment of the present invention;

[0049] Figure 13 This is a schematic diagram of sub-steps of step S23 in an embodiment of the present invention;

[0050] Figure 14 This is a time domain diagram of a displacement signal in an embodiment of the present invention;

[0051] Figure 15 This is a local enlarged view of the displacement signal in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0053] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0054] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0055] In the prior art, the background noise filtering of acceleration signals is usually performed by time domain integration or frequency domain integration to filter out noise signals, such as Figure 1As shown in the figure, the signal is a frequency domain image of the signal generated after conventional time domain integration processing in the prior art. It can be seen from the figure that the original acceleration signal is 500Hz, but with the time domain integration process, the noise signal below 100Hz also increases, which causes the velocity signal to be seriously distorted. Similarly, Figure 2 The signal shown is a frequency domain image of the signal generated after conventional frequency domain integration processing in the prior art. Below 100 Hz, the noise signal also exhibits the characteristic of infinitely increasing during the integration process. Therefore, the velocity signal obtained in the prior art is difficult to directly read from the acceleration, which in turn affects the accuracy of the fault diagnosis process.

[0056] In view of the above technical problems, the present invention provides a fault monitoring method for a rotating device. An acceleration sensor is provided on the rotating device to generate an acceleration signal. Figure 3 As shown, the fault monitoring method includes:

[0057] Step S1: Acquire acceleration signal;

[0058] Step S2: performing spectrum editing and integration processing on the acceleration signal to generate a velocity signal and a displacement signal;

[0059] Step S3: Determine whether the rotating device has a fault based on the speed signal and the displacement signal.

[0060] Specifically, if Figure 4 and Figure 5 As shown in the figure, the acceleration signal directly obtained by the accelerometer contains a lot of noise. Frequency domain analysis shows that this noise is prevalent in the 0-1200Hz range, making it impossible to filter out the related noise through simple filtering methods. Therefore, further processing of the acceleration signal is required so that the fault monitoring system can effectively determine the fault condition by judging the waveform changes of the acceleration signal based on the waveform diagram.

[0061] In a specific implementation, this solution pre-installs an acceleration sensor on the rotating device. Based on the vibrations generated during operation, the sensor generates a periodic acceleration signal corresponding to the vibrations of the rotating device. It is common knowledge that the vibrations of a rotating device during normal operation differ from those during a fault. Furthermore, depending on the location of the fault, the vibrations generated will also vary, which is reflected in the acceleration signal. Therefore, the acceleration signal can effectively detect operational faults in the rotating device. In practical implementation, a memory can be pre-set with a set of experimentally acquired acceleration, velocity, and displacement signals collected from the rotating device under normal conditions; as well as multiple acceleration, velocity, and displacement signals corresponding to different typical faults of the rotating device. By comparing the velocity and displacement signals processed using the aforementioned method, the current state of the rotating device and the specific fault condition can be directly detected.

[0062] In a preferred embodiment, Figure 6 As shown, the spectrum editing integration method in step S2 includes:

[0063] Step S21: pre-processing the acceleration signal to generate a signal amplitude sequence;

[0064] Step S22: processing the signal amplitude sequence to generate a speed signal;

[0065] Step S23: generating a displacement signal according to the speed signal.

[0066] In a preferred embodiment, Figure 7 As shown, step S21 includes:

[0067] Step S211: removing the DC component from the acceleration signal to generate a preprocessed signal;

[0068] Step S212: Perform high-pass filtering on the preprocessed signal to generate a signal amplitude sequence.

[0069] Specifically, the method for removing the DC component in step S211 is to calculate the mean of the acceleration signal as a whole and remove the mean from the acceleration signal to avoid the influence of the DC component on the acceleration signal;

[0070] The cutoff frequency f of the high-pass filter in step S212 is the low-frequency response frequency of the acceleration sensor, which is adjusted according to the selected acceleration sensor. i}, {i=1,2,3,…,N}, N is the number of sampling points.

[0071] In a preferred embodiment, Figure 8 As shown, step S22 includes:

[0072] Step S221: performing fast Fourier transform on the signal amplitude sequence to generate a spectrum amplitude sequence;

[0073] Step S222: performing spectrum editing on the spectrum amplitude sequence to generate a cropping sequence;

[0074] Step S223: Perform inverse Fourier transform on the cropped sequence to generate a velocity signal.

[0075] Specifically, in step S221, by performing the signal amplitude sequence {A i} Perform fast Fourier transform to generate spectrum amplitude sequence {B j}, {j=1,2,3,…,N}. The fast Fourier transform process is common knowledge and will not be described here. At this time, according to the formula The carpet value of the spectrum signal, that is, the spectrum valley mean value of the spectrum, can be calculated as the background noise in the spectrum. i Satisfy B i >B i-1 And B i >B i+1 , Q is the number of points that meet the above conditions.

[0076] In a preferred embodiment, in step S222, the spectrum editing method is:

[0077]

[0078] Where: C j is the cropping sequence, B j is the spectrum amplitude sequence, df is the spectrum resolution, fs is the sampling frequency, M is the frequency offset of the effective peak, α is the coefficient of the effective peak relative to the mean amplitude of the adjacent frequencies, and β is the ratio of the effective amplitude to the noise carpet value.

[0079] By spectrum editing, we can effectively select the spectrum that meets the peak condition and reaches a certain multiple of the noise floor amplitude, and perform integration operation on it to obtain the following: Figure 9 As shown in the velocity frequency domain diagram, it can be seen that the four effective frequency components in the spectrum are effectively retained, the noise is suppressed, and the defect of unlimited amplification of low-frequency noise in the existing technology is avoided.

[0080] Then, the inverse Fourier transform is performed to generate the time domain image of the speed signal, that is, the waveform of the speed signal. The final speed signal waveform is as follows: Figure 10 、 Figure 11 As shown, it can be seen Figure 10 The speed signal waveform in Figure 11The local enlarged images of the medium-speed signal waveforms overlap, indicating that the velocity waveform generated by this method has a small error.

[0081] In a preferred embodiment, Figure 12 As shown, step S222 includes:

[0082] Step S2221: generating a first sequence of carpet values ​​according to the spectrum amplitude sequence;

[0083] Step S2222: performing spectrum editing on the spectrum amplitude sequence according to the first sequence of carpet values.

[0084] Specifically, in this technical solution, the first sequence of carpet values ​​is used to obtain the mean of the spectral valleys in the spectral amplitude sequence, which is then used as the noise floor in the spectral amplitude sequence. By calculating the first sequence of carpet values, the noise floor in the spectral amplitude sequence can be effectively extracted, making it easier to remove it through spectral editing, achieving better denoising results.

[0085] In a preferred embodiment, Figure 13 As shown, step S23 includes:

[0086] Step S231: performing fast Fourier transform on the velocity signal to generate a velocity spectrum sequence;

[0087] Step S232: performing spectrum editing on the velocity spectrum sequence to generate a velocity edited sequence;

[0088] Step S233: Perform inverse Fourier transform according to the velocity editing sequence to generate a displacement signal.

[0089] Specifically, the displacement signal can be obtained by performing secondary processing on the velocity signal. The specific method includes: processing the velocity signal sequence {X i} Perform fast Fourier transform to generate velocity spectrum sequence {Y j}, {j=1,2,3,…,N}. The fast Fourier transform process is common knowledge and will not be described here. At this time, according to the formula The carpet value of the spectrum signal, that is, the spectrum valley mean value of the spectrum, can be calculated as the background noise in the spectrum. i Satisfy B i >B i-1 And B i >B i+1 , Q is the number of points that meet the above conditions.

[0090] In a preferred embodiment, in step S222, the spectrum editing method is:

[0091]

[0092] Where: Z j To edit the sequence for speed, Y j is the velocity spectrum sequence, df is the spectrum resolution, fs is the sampling frequency, M is the frequency offset of the effective peak, α is the coefficient of the effective peak relative to the mean amplitude of the adjacent frequencies, and β is the ratio of the effective amplitude to the noise carpet value.

[0093] In a preferred embodiment, in step S212 , the cutoff frequency of the high-pass filter is the low-frequency response cutoff frequency of the acceleration sensor.

[0094] In a preferred embodiment, the frequency offset value of the effective peak is 4, 6 or 8;

[0095] The value range of the coefficient of the effective peak relative to the mean value of the adjacent frequency amplitude is 1.5 to 3;

[0096] The ratio of the effective amplitude to the noise carpet value ranges from 6 to 10.

[0097] The beneficial effects of the present invention are:

[0098] 1. The spectrum editing-based integration method proposed in this invention can effectively utilize the sensor's frequency response characteristics to suppress low-frequency components in the signal. It can also adaptively suppress interference components and retain valid signals based on the signal's noise floor and peak distribution characteristics.

[0099] 2. The frequency domain integration method based on spectrum editing proposed in the present invention has a simple process and low computational complexity. In the presence of multiple sinusoidal component effective signals and a mixture of Gaussian noises with different signal-to-noise ratios, the integration effect is accurate and the error is small, making it suitable for real-time online monitoring scenarios.

[0100] The above are only preferred embodiments of the present invention and do not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for monitoring faults of rotating equipment, characterized in that: The rotating device is provided with an acceleration sensor for generating an acceleration signal, and the fault monitoring method includes: Step S1: obtaining the acceleration signal; Step S2: performing spectrum editing and integration processing on the acceleration signal to generate a velocity signal and a displacement signal; Step S3: determining whether there is a fault in the rotating device based on the speed signal and the displacement signal; The spectrum editing and integration method in step S2 includes: Step S21: preprocessing the acceleration signal to generate a signal amplitude sequence; Step S22: processing the signal amplitude sequence to generate a speed signal; Step S23: generating the displacement signal according to the velocity signal; The step S22 includes: Step S221: performing fast Fourier transform on the signal amplitude sequence to generate a spectrum amplitude sequence; Step S222: performing spectrum editing on the spectrum amplitude sequence to generate a cropping sequence; Step S223: performing an integration operation on the clipping sequence to obtain a velocity frequency domain map, and then performing an inverse Fourier transform to generate the velocity signal; In spectrum editing, first based on the formula Calculate the carpet value of the spectrum signal, satisfy > -1 and > +1, Points for meeting the above conditions; Then, based on the formula: Perform spectrum editing to select the spectrum that meets the peak condition and reaches a certain multiple of the noise floor amplitude; in: is the cropping sequence, is the spectrum amplitude sequence, j=1, 2, 3, ..., N; , is the sampling frequency, is the number of sampling points; is the cutoff frequency; is the frequency offset near the effective peak, is the coefficient of the effective peak relative to the mean value of the adjacent frequency amplitude, is the ratio of the effective amplitude to the noise carpet value.

2. The fault monitoring method according to claim 1, characterized in that: The step S21 includes: Step S211: removing a DC component from the acceleration signal to generate a preprocessed signal; Step S212: performing high-pass filtering on the preprocessed signal to generate the signal amplitude sequence.

3. The fault monitoring method according to claim 1, characterized in that: The step S222 includes: Step S2221: generating a first sequence of carpet values ​​according to the spectrum amplitude sequence; Step S2222: performing spectrum editing on the spectrum amplitude sequence according to the first sequence of carpet values.

4. The fault monitoring method according to claim 1, characterized in that: The step S23 includes: Step S231: performing fast Fourier transform on the velocity signal to generate a velocity spectrum sequence; Step S232: performing spectrum editing on the velocity spectrum sequence to generate a velocity edited sequence; Step S233: Perform inverse Fourier transform according to the velocity editing sequence to generate the displacement signal.

5. The fault monitoring method according to claim 2, characterized in that: In step S212, the cutoff frequency of the high-pass filter is the low-frequency response cutoff frequency of the acceleration sensor.

6. The fault monitoring method according to claim 1, characterized in that: The value of the frequency offset close to the effective peak is 4, 6 or 8; The effective peak relative to the adjacent frequency amplitude mean coefficient has a value range of 1.5 to 3; The ratio of the effective amplitude to the noise carpet value ranges from 6 to 10.

Images