CN115270843A - A fault diagnosis method and device for a reciprocating compressor of a floating platform - Google Patents

Abstract

The invention discloses a fault diagnosis method and a fault diagnosis device for a floating platform reciprocating compressor, which comprises the steps of decomposing a vibration signal to be processed to obtain an IMF component; screening certain orders of IMFs containing obvious impact characteristics through indexes; filtering the screened IMF components respectively to filter non-impact components; reconstructing the IMF component after the adaptive variable-scale morphological filtering to obtain a noise reduction signal; amplifying a sudden change component in the noise reduction signal; calculating the envelope of the abrupt change component, and performing smooth filtering on the envelope by using a filter; carrying out peak detection to extract the phase characteristics of impact components in the signals; and acquiring the phase characteristics of the impact. The diagnostic device includes: the system comprises an acquisition module, a processing module, an extraction module and a diagnosis module, wherein a reciprocating compressor fault characteristic knowledge base is established, and fault diagnosis is carried out according to vibration signal impact characteristics captured by the extraction module. And provides support and guarantee for the predictive maintenance of the reciprocating compressor of the ocean platform in the future.

Description

Fault diagnosis method and device for a reciprocating compressor on a floating platform

technical field

The present invention relates to the field of fault diagnosis of reciprocating compressors on floating platforms, in particular to a fault diagnosis method and device for extracting vibration signal impact features of reciprocating compressors on floating platforms.

Background technique

As a power source for compressing and transporting medium, large reciprocating piston compressors are widely used in petroleum, chemical industry, refrigeration and other fields, and are also one of the key equipment of offshore platforms. Its safe, reliable, stable and long-term operation is very important. When some failures of reciprocating compressors occur, it often leads to an increase in the number of shocks or a shift in the shock phase detected by the vibration sensor at the structure such as the middle body, valve cover, and crankcase. By monitoring the number and phase of shocks in a cycle in real time , a preliminary diagnosis of the fault can be made. For example, when the small head tile is worn out, additional impact will occur near the stop point of the piston reversing; when the intake and exhaust valves have spring failure, valve plate breakage and other faults, it will cause the intake and exhaust valves to open and seat. Advance or delay, which will cause the acceleration shock phase at the valve cover and cylinder block to shift; when the big head shoe wears, there will be two shocks per cycle in the crankcase speed signal. Preliminary diagnosis of faults can be made by monitoring the number, peak and phase of shocks to the compressor body, valves or crankcase within a cycle.

The body acceleration signal of the reciprocating compressor includes the harmonic signal related to the crank rotation period, the shock signal related to the operating state of the equipment and the environmental noise signal. In order to accurately extract the impact components in the signal, in addition to noise reduction, it is also necessary to suppress the harmonic components in the signal. Using traditional linear filters to filter out harmonic components in the signal requires sufficient prior knowledge to accurately predict the frequencies of all harmonics in order to apply an effective notch filter to suppress them. In addition, once the frequency of the harmonics coincides with the frequency components contained in the shock signal, it will cause the loss of shock characteristics, contrary to its goal of accurately extracting shock characteristics.

Contents of the invention

The technical problem to be solved by the present invention is a fault diagnosis method and device for a reciprocating compressor on a floating platform, which extracts the impact characteristics of the vibration signal of the reciprocating compressor on a floating platform and realizes fault diagnosis.

In order to solve the above technical problems, the present invention provides a fault diagnosis device for a reciprocating compressor on a floating platform, including:

The collection module is used to obtain the vibration data of the reciprocating compressor on the floating platform, and collect and transmit the data to the host computer for processing;

The processing module is connected with the acquisition module, and processes the acquired vibration signal through EEMD-adaptive variable-scale morphological filtering;

The extraction module is connected with the processing module, and is used for performing peak detection, extracting the phase characteristics of the shock component in the signal, and capturing the parameter characteristics of the shock.

The diagnosis evaluation module is connected with the extraction module, establishes the reciprocating compressor fault feature knowledge base, and compares the fault feature knowledge base according to the vibration signal shock feature captured by the extraction module to perform fault diagnosis and output diagnostic results;

The visualization module is connected with the diagnostic evaluation module and the acquisition module, and visualizes the diagnostic results and collected data. At the same time, remote data processing and calling can be realized through interactive access to web pages through remote terminals.

Another technical solution is: a fault diagnosis method for a reciprocating compressor on a floating platform, characterized in that it includes the following steps:

S01: Decompose the vibration signal to be processed to obtain several IMF components;

S02: Screen several IMF components through indicators to select certain IMFs with obvious impact characteristics;

S03: Perform adaptive variable-scale morphological filtering on the screened IMF components, and filter non-impact components;

S04: Reconstruct the IMF component after adaptive variable-scale morphological filtering to obtain a noise reduction signal;

S05: Amplify the amplitude-frequency mutation component in the noise reduction signal;

S06: Calculate the envelope of the sudden change component, and use a smoothing filter to smooth the envelope;

S07: Perform peak phase detection to extract the phase characteristics of the impact component in the signal;

S08: After obtaining the phase characteristics of the impact, capture the parameter characteristics to further diagnose the fault of the reciprocating compressor;

S09: Visualize the diagnostic results and collected data, and remotely process and call the data.

Further, the index in the step S02 adopts a kurtosis-correlation index, including:

The first step, the kurtosis index:

For a discrete signal x(i)={xi _i ∣i=1,2,…,n}, where n is the number of signal points, kurtosis K is defined as:

为x(i)的均值，定义为：In the formula

is the mean of x(i), defined as:

Calculate the kurtosis value of each order component separately, and remove the component whose kurtosis value is less than 5.0, and realize the first step of screening of the IMF component;

The second step, the correlation index:

Calculate the correlation coefficient between each order IMF component and the original function, assuming that there are two continuous time domain signals x(t), y(t), their correlation coefficient R(xy) is defined as:

Select the IMF with a correlation coefficient greater than 10% to realize the second step of screening.

Further: in the step S03, the adaptive scaling morphological filtering includes:

和腐蚀算子

为：For a discrete signal f(n) with a total number of points of N, where the independent variable n=0,1,...,N-1 and a sequence of structural elements g(m) with a total number of M points, where the independent variable m=0,1, ..., M-1, N<M, define the expansion operator

and corrosion operator

for:

表示f(n)关于g(m)的膨胀，

表示f(n)关于g(m)的腐蚀。in,

Indicates the expansion of f(n) with respect to g(m),

Indicates the corrosion of f(n) with respect to g(m).

The morphological opening operator (°) and morphological closing operator ( ) are defined by the sequential combination of dilation operator and erosion operator:

Among them, (f°g)(n) represents the morphological opening operator of f(n) with respect to g(m), and (f·g)(n) represents the morphological closing operator of f(n) with respect to g(m).

Further, the morphological opening-closing (OC) and morphological closing-opening (CO) operators are defined by the sequential combination of the morphological opening operator and the morphological closing operator:

OC[f(n)]=(f°g·g)(n)

CO[f(n)]=(f·g°g)(n)

Among them, OC[f(n)] represents the morphological opening and closing operator of f(n), and CO[f(n)] represents the morphological closing and opening operator of f(n).

The cascaded morphological filter Γ[f(n)] of f(n) formed by the combination of morphological open-close operator and morphological close-open operator:

Further, the filter structural element uses a flat structure with a height of 0, and defines a structural element sequence g(n, m, k) (n =0,1,…,N-1; m=0,1,…,k-1), where N is the number of signal points, k is the width of the structural element of the nth sampling point, and its width k is determined as follows:

x(i)＝{x_i∣i＝1,2,…,N}，其中N为信号点数；(1) Record the phase and amplitude of each point of the signal as

x(i)={x _i ∣i=1,2,…,N}, where N is the number of signal points;

(2) Calculate all local extremum points of the signal, record their phase and amplitude as φ(j)={φ _j ∣j=1,2,…,M}, y(j)={y _j ∣j= 1,2,...,M}, where M is the number of extreme points;

(3) Perform linear normalization y _norm (j) on the amplitude of the extreme point of the signal:

Among them, y _j represents the amplitude of the jth extreme point, min{y(j)} is the minimum value of the extreme point sequence y(j), max{y(j)} is the maximum value of the extreme point sequence y(j) value.

(4) Calculate the signal waveform scale s:

将冲击与非冲击成分的界限模糊化，改变归一化局部极值 点幅值分布，则信号中各点所对应的结构元素宽度k(i)＝{k_i∣i＝1,2,...,N}由下式确定：(5) Define nonlinear mapping

Blur the boundary between shock and non-shock components, and change the amplitude distribution of normalized local extremum points, then the width of structural elements corresponding to each point in the signal k(i)={k _i ∣i=1,2,. ..,N} is determined by the following formula:

表示向上取整，映射

定义为：where the symbol

Represents rounding up, mapping

defined as:

In the formula, a is a variable parameter that controls the "bending" degree of the mapping curve, and is used to adjust the amplitude distribution of extreme points. The recommended value range is a∈[2,8]. With k(i) as the element width, cascaded morphological filtering is performed on the signal to achieve effective suppression of non-impact components.

Further, in step S05, using the Teager energy operator to amplify the mutation component in the noise reduction signal:

Specifically, for a discrete signal, for a discrete signal x(i) (i=1,2,…,n) with a total number of n points, a three-point symmetric differential energy operator is constructed to smooth it, and its energy operator Ψ [x(i)] is defined as:

Further, the step 06: obtain the abrupt component Hilbert envelope, and utilize the Savitzky-Golay filter to carry out smooth filtering to the envelope.

定义为：Specifically, for a continuous time signal x(t), its Hilbert transform

defined as:

Construct parse signal

Then the Hilbert envelope of the signal x(t) is:

与均方差σ， 利用3-σ法则去除含有明显冲击特征的小段，剩余的m小段数列

表征信号中相对平稳的成分。设数列r[j]的 平均值为

最大值和最小值之差为Δr，加权系数a用于对阈值进行调整则可以定义冲击检测阈 值为

Divide the envelope filtered by the Savitzky-Golay smoothing filter into n small segments in the angle domain diagram, calculate the average amplitude of each small segment signal, and arrange them into a set of numbers l[i], i=0,1, ..., n-1, calculate the mean of the sequence

and the mean square error σ, use the 3-σ rule to remove small segments with obvious impact characteristics, and the remaining m small segment sequence

Characterize relatively stationary components of a signal. Let the average value of the sequence r[j] be

The difference between the maximum value and the minimum value is Δr, and the weighting coefficient a is used to adjust the threshold, then the impact detection threshold can be defined as

Further, take the segment number n=72, that is, every 5° is a small segment.

The technical effect of the present invention is that: the embodiment of the present application proposes a method of obtaining several eigenmode functions by performing overall average empirical mode decomposition on the vibration signal, and using an adaptive variable-scale morphological filter for each signal Points are configured with appropriate structural element widths, and the excellent performance of cascaded morphological filters in pulse suppression is reversely used to preserve the shock components in the IMF while suppressing non-shock components. Finally, an adaptive shock peak detection method is used to automatically Judging the phase characteristics of the shock component in the signal, the device and method can effectively extract the shock characteristics in the signal, provide an identification basis for the fault diagnosis of the reciprocating compressor, and provide support and guarantee for the predictive maintenance of the reciprocating compressor on the offshore platform in the future.

Based on the EEMD-adaptive variable-scale morphological filter, the fault diagnosis method and device for extracting the shock feature of the vibration signal of the reciprocating compressor on the floating platform use the mathematical morphological filtering method to analyze the harmonic components in the IMF of each order to be reconstructed Compared with the morphological filtering of the noise reduction signal after EEMD decomposition and reconstruction, it has a better harmonic suppression effect, and realizes the accurate impact feature extraction of the vibration signal of the reciprocating compressor on the floating platform and realizes the fault diagnosis.

Description of drawings

Fig. 1 is a flow chart of the fault diagnosis method for extracting the impact feature of the vibration signal of the reciprocating compressor on the floating platform;

Figure 2 is a structural schematic diagram of a fault diagnosis device for extracting vibration signal impact features of a reciprocating compressor on a floating platform;

Figures 3a-3d are the simulated reciprocating compressor impact, harmonics, noise and their mixed vibration signals;

Figure 4 is the reconstruction of the noise reduction signal after the adaptive variable-scale morphological filtering of the simulated reciprocating compressor;

Fig. 5 is the simulation reciprocating compressor shock threshold calculation and peak detection results;

Figure 6 shows the body acceleration signal and dynamic pressure at the crankshaft end (CE) and cylinder head end (HE) in a real reciprocating compressor;

Fig. 7 is the body acceleration signal in the real reciprocating compressor;

Fig. 8 is the reconstructed noise reduction signal of a real reciprocating compressor after adaptive variable-scale morphological filtering;

Fig. 9 is the filtering result of teager energy operator of real reciprocating compressor;

Figure 10 shows the real reciprocating compressor shock threshold calculation and peak detection results.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only It is a part of the embodiments of this application, not all of them. The components of the embodiments of the application generally described and illustrated in the figures herein may be arranged and designed in a variety of different configurations. Accordingly, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative efforts all belong to the scope of protection of the present application.

Based on this, the embodiment of the present application proposes a method of obtaining several Intrinsic Mode Functions (IMF) by performing overall average empirical mode decomposition (Ensemble empirical mode decomposition, EEMD) on the vibration signal, and using an adaptive variable Scale Morphological Filter (Adaptive VariableScale Morphological Filter, AVSMF), configure the appropriate width of structural elements for each signal point, and reversely use the excellent performance of cascaded morphological filters in pulse suppression to achieve the preservation of the impact components in the IMF. Suppress the non-shock components, and finally, use an adaptive shock peak detection method to automatically judge the phase characteristics of the shock components in the signal. This device and method can effectively extract the shock features in the signal, and provide a recognition basis for reciprocating compressor fault diagnosis , and provide support and guarantee for the predictive maintenance of reciprocating compressors on offshore platforms in the future.

Based on the EEMD-adaptive variable-scale morphological filter, the fault diagnosis method and device for extracting the shock feature of the vibration signal of the reciprocating compressor on the floating platform use the mathematical morphological filtering method to analyze the harmonic components in the IMF of each order to be reconstructed Compared with the morphological filtering of the noise reduction signal after EEMD decomposition and reconstruction, it has a better harmonic suppression effect, and realizes the accurate impact feature extraction of the vibration signal of the reciprocating compressor on the floating platform and realizes the fault diagnosis.

Fig. 1 is a flowchart of a fault diagnosis method for extracting vibration signal shock features of a reciprocating compressor on a floating platform provided by an embodiment of the present application. Referring to Fig. 1, the floating platform reciprocating compressor vibration signal shock feature extraction fault diagnosis method provided by the embodiment of the present application specifically includes:

Step 1: Perform EEMD decomposition on the vibration signal to be processed to obtain several IMFs.

Specifically, the EEMD decomposition introduces Gaussian white noise into the original signal, performs multiple EMD decompositions, and averages the IMF components decomposed multiple times to obtain the final IMF.

Each IMF needs to meet the following two conditions:

(1) In the entire signal, the difference between the number of extreme points of the signal and the number of zero-crossing points is less than or equal to 1;

(2) At any time, the upper envelope formed by the local maximum points and the lower envelope formed by the local minimum points are locally symmetrical with respect to the time axis.

The main steps of EMD decomposition:

(1) Find the upper and lower extreme points of the vibration signal, and draw the upper and lower envelopes;

(2) Calculate the mean value of the upper and lower envelopes to obtain the mean envelope;

(3) Subtract the mean value envelope from the original vibration signal to obtain the intermediate signal;

(4) Determine whether the intermediate signal satisfies the two conditions of IMF. If so, this intermediate signal is recorded as an IMF component. If not, then steps (1) to (4) are repeated for this intermediate signal.

In step 2, several IMF components are selected through the kurtosis-correlation index to select some IMF components with obvious impact characteristics.

Specifically, the first step is the principle of kurtosis index:

The kurtosis index is a time-domain statistical dimensionless index, which is often used to detect shock features in signals. For a discrete signal x(i)={xi _i ∣i=1,2,…,n}, where n is the number of signal points, kurtosis K is defined as:

为x(i)的均值，定义为：In the formula

is the mean of x(i), defined as:

The kurtosis value of each order component is calculated separately, and the component whose kurtosis value is less than 5.0 is eliminated to realize the first step of screening of the IMF component.

The second step, the correlation index:

When performing EEMD decomposition, due to factors such as calculation errors and edge effects, the number of decomposed IMF components is more than the actual components of the original signal, and these extra IMF components are called "pseudo components". In signal reconstruction, including these pseudo components will lead to the introduction of new noise, so these pseudo components must be eliminated.

Calculate the correlation coefficient between each order IMF component and the original function, assuming that there are two continuous time domain signals x(t), y(t), their correlation coefficient R(xy) is defined as:

Select the IMF with a correlation coefficient greater than 10% to realize the second step of screening.

Step 3: Perform adaptive variable-scale morphological filtering on the selected IMF components, filter out non-shock components, and improve the signal-to-noise ratio of the shock signal.

和腐蚀算子为

For a discrete signal f(n) with a total number of points of N, where the independent variable n=0,1,...,N-1 and a sequence of structural elements g(m) with a total number of M points, where the independent variable m=0,1, ..., M-1, N<M, define the expansion operator

and the corrosion operator is

表示f(n)关于g(m)的膨胀，

表示f(n)关于g(m)的腐蚀；in,

Indicates the expansion of f(n) with respect to g(m),

Indicates the corrosion of f(n) on g(m);

The morphological opening operator (°) and morphological closing operator ( ) are defined by the sequential combination of dilation operator and erosion operator:

Among them, (f°g)(n) represents the morphological opening operator of f(n) with respect to g(m), and represents the morphological closing operator of f(n) with respect to g(m);

Further, the morphological opening-closing (OC) and morphological closing-opening (CO) operators are defined by the sequential combination of morphological opening and morphological closing operators:

OC[f(n)]=(f°g·g)(n)

CO[f(n)]=(f·g°g)(n)

Among them, OC[f(n)] represents the morphological opening and closing operator of f(n), and CO[f(n)] represents the morphological closing and opening operator of f(n).

Among them, the morphological open-close operator can suppress the positive pulse in the signal and eliminate the sharp "peak" in the signal, while the morphological close-open operator can suppress the negative pulse in the signal and fill in the low-lying "valley" in the signal . In order to remove the positive and negative bidirectional pulses in the signal at the same time, the cascaded morphological filter Γ[f(n)] of f(n) formed by the combination of the morphological open-close operator and the morphological close-open operator can be defined.

Traditional morphological filters use structural elements with a fixed width, while the width of structural elements used in the morphological filter proposed by this method is automatically adjusted according to the amplitude of the local extreme points of each peak (trough) in the signal. adaptable to change. Among them, the structural elements use a flat structure with a height of 0. For a discrete signal f(n)(n=0,1,...,N-1), define a sequence of structural elements g(n,m,k)(n=0,1, ...,N-1; m=0,1,...,k-1), where N is the number of signal points, k is the width of the structural element of the nth sampling point, and its width k is determined as follows:

(1) Calculate the phase and amplitude of all local extremum points in the signal, respectively recorded as φ(j)={φ _j ∣j=1,2,…,M}, y(j)={y _j ∣j ＝1,2,...,M}, where M is the number of local extremum points;

(2) Calculate all local extremum points of the signal, record their phase and amplitude as φ(j)={φ _j ∣j=1,2,…,M}, y(j)={y _j ∣j= 1,2,…,M}, where M is the number of extreme points;

(3) Perform linear normalization y _norm (j) on the amplitude of the local extremum point of the signal:

Among them, y _j represents the amplitude of the jth extreme point, min{y(j)} is the minimum value of the extreme point sequence y(j), max{y(j)} is the maximum value of the extreme point sequence y(j) value.

(4) Calculate the signal waveform scale s:

将冲击与非冲击成分的界限模糊化，改变归一化局部极值 点幅值分布，则信号中各点所对应的结构元素宽度k(i)＝{k_i∣i＝1,2,…,N}由下式确定：(5) Define nonlinear mapping

Blur the boundary between shock and non-shock components, and change the amplitude distribution of normalized local extremum points, then the width of structural elements corresponding to each point in the signal k(i)={k _i ∣i=1,2,… ,N} is determined by the following formula:

表示向上取整，映射

定义为：where the symbol

Represents rounding up, mapping

defined as:

In the formula, a is a variable parameter that controls the "bending" degree of the mapping curve, and is used to adjust the amplitude distribution of extreme points. The recommended value range is a∈[2,8]. With k(i) as the element width, cascaded morphological filtering is performed on the signal to achieve effective suppression of non-impact components.

Step 4: Reconstruct the IMF component after adaptive variable-scale morphological filtering to obtain a noise-reduced signal.

Step 5, use the Teager energy operator to amplify the amplitude-frequency mutation component in the noise reduction signal.

Specifically, for a discrete signal, for a discrete signal x(i) (i=1,2,…,n) with a total number of n points, a three-point symmetric differential energy operator is constructed to smooth it, and its energy operator Ψ [x(i)] is defined as:

Step 6, obtain the Hilbert envelope of the mutation component, and use the Savitzky-Golay filter to smooth the envelope.

定义为：Specifically, for a continuous time signal x(t), its Hilbert transform

defined as:

Construct parse signal

Then the Hilbert envelope of the signal x(t) is:

The Savitzky-Golay filter is a widely used data stream smoothing and denoising method, which is based on polynomials in the time domain and uses the least squares method to perform the best fitting method through moving windows.

与均方差σ， 利用3-σ法则去除含有明显冲击特征的小段，剩余的m小段数列

表征信号中相对平稳的成分。设数列r[j]的 平均值为

最大值和最小值之差为Δr，加权系数a用于对阈值进行调整，则可以定义冲击 检测阈值为

本申请实施例中取段数n＝72，即每5°为一小段。Divide the envelope filtered by the Savitzky-Golay smoothing filter into n small segments in the angle domain diagram, calculate the average amplitude of each small segment signal, and arrange them into a set of numbers l[i], i=0,1, ..., n-1, calculate the mean of the sequence

and the mean square error σ, use the 3-σ rule to remove small segments with obvious impact characteristics, and the remaining m small segment sequence

Characterize relatively stationary components of a signal. Let the average value of the sequence r[j] be

The difference between the maximum value and the minimum value is Δr, and the weighting coefficient a is used to adjust the threshold, then the impact detection threshold can be defined as

In the embodiment of the present application, the number of segments is n=72, that is, every 5° is a small segment.

Step 7: Perform peak phase detection to extract the phase features of the shock component in the signal.

Specifically, by performing zero-crossing detection with the threshold as zero on the smoothed envelope, all "peaks" higher than the threshold line can be extracted, which are shock peaks. False shocks are eliminated through the continuous angle of each shock peak in the angle domain diagram, and the continuous angle is taken as 5°. Finally, it is determined that the angle in the angle domain diagram where the peak of each shock peak is located is the phase characteristic of the shock.

Step 8: After obtaining the phase characteristics of the shock, capture the amplitude, energy and other characteristics of the shock for further fault diagnosis of the reciprocating compressor.

Figure 2 is a schematic structural diagram of a fault diagnosis device for extracting vibration signal impact features of a reciprocating compressor on a floating platform. Referring to Figure 2, the fault diagnosis device for extracting vibration signal impact features of a reciprocating compressor on a floating platform provided in an embodiment of the application specifically includes :

The collection module is used to obtain the vibration data of the reciprocating compressor on the floating platform, and collect and transmit the data to the host computer for processing.

The acquisition module specifically includes: a key phase sensor, a multi-measuring point vibration sensor and a signal conditioning acquisition device. The key phase sensor is used to obtain key phase pulse signals at impact positions such as the cylinder valve cover, middle body, and crankshaft of a reciprocating compressor. The setting of the trigger value can obtain the time interval of one working cycle of the compressor. The multi-measuring-point vibration sensor can capture vibration signals at multiple key locations of the compressor. The signal conditioning acquisition equipment includes: data acquisition hardware using NI 9263 sound and vibration voltage input module, NI Compact RIO 9047 chassis with NI 9263 module for data bottom acquisition and TCP/IP network data transmission. First read the data from the FIFO, and then transmit the data to the upper computer through the TCP protocol. The TCP protocol supports the transmission of data to multiple clients, that is, the transmission to multiple upper computers.

The acquisition module can provide the required original vibration signal data for the subsequent processing module, and provide the multi-source fault diagnosis basis for the subsequent diagnosis module.

The processing module is connected with the acquisition module, including the storage and processing of the collected vibration signal data, and the acquired vibration signal is processed through EEMD-adaptive variable-scale morphological filtering.

The extraction module is connected with the processing module, and is used to perform peak detection and extract the phase characteristics of the shock component in the signal, and capture other characteristics such as the amplitude and energy of the shock.

The diagnostic evaluation module is connected with the extraction module to establish a reciprocating compressor fault feature knowledge base, and perform fault diagnosis according to the shock characteristics of the vibration signal captured by the extraction module:

Specifically, the extraction module performs impact feature extraction based on continuous real-time data collected on site, compares multi-source feature values with the reciprocating compressor fault feature knowledge base, and outputs diagnostic results. The time interval of diagnostic output results can be set, that is, the diagnostic interval of compressor working cycle can be customized.

The visualization module is connected with the diagnostic evaluation module and the acquisition module, and visualizes the diagnostic results and collected data. At the same time, remote data processing and calling can be realized through interactive access to web pages through remote terminals.

Specifically, the module is implanted into the actual 3D model of the compressor, and the collected measurement results are visualized in real time and the location of the diagnostic fault is highlighted. At the same time, in view of the characteristics of remote monitoring and storage of large amounts of data, such as reciprocating compressors on offshore platforms, the diagnostic module based on web page access is set locally, and web page interactive access is performed through remote terminals to avoid bandwidth problems caused by big data backhaul. Provide a remote data download port (on-demand download), so that experts who install the client can conduct in-depth analysis and realize remote monitoring and diagnosis of the compressor system.

Floating platform reciprocating compressor vibration signal impact feature extraction fault diagnosis device can realize acquisition module vibration signal acquisition in each compressor working cycle, through the processing module and extraction module to obtain the vibration signal impact phase eigenvalue, the eigenvalue and The knowledge base in the diagnosis and evaluation module is matched to give the diagnosis results of the compressor status, and the visualization module displays the results and the monitoring and collection values. The whole device has the functions of remote acquisition of data and monitoring of compressor status to ensure safe and stable operation of reciprocating compressors.

Embodiment one:

The continuous time domain model of the body acceleration signal in the reciprocating compressor is established as follows:

In the formula, τ _i and A _i are the time and amplitude of the i-th impact, f _ni is a certain natural frequency of the body-sensor system in the compressor, ζ is the damping ratio of the corresponding system, and f _j is the The multiplier frequency component of , 1(t-τ _i ) is the unit step function, and p(t) is the noise signal.

Figure 3 is the simulated reciprocating compressor shock, harmonics, noise and its mixed vibration signals, referring to Figure 3, set the reciprocating compressor speed to 900RPM, acceleration sensor sampling rate f _s =51200Hz, assuming that there are 3 different types of signals Shock signals, which respectively represent high-frequency high-amplitude shocks, low-frequency low-amplitude shocks, and high-frequency low-amplitude shocks. Their amplitudes A _i are 160, 90, 90 respectively, the natural frequencies f _ni of the system are 3000, 1500, 3000 respectively, the damping ratios ζ _i are 0.1, 0.1, 0.1 respectively, and the phases τ _i of the impact are 60°, 120°, 270°, simulate a cycle signal refer to Figure 3(a), at the same time, in order to simulate the real compressor working condition, introduce the harmonic components with amplitudes of 20, 10 and frequencies of 90Hz and 180Hz respectively, set The phases are the same, refer to Figure 3(b), and finally introduce Gaussian white noise with a signal-to-noise ratio of -3dB, refer to Figure 3(c). Refer to Figure 3(d) for the final synthesized simulation signal.

Since it takes a while for the shock to reach its peak value, the actual phase of the peak value of the shock lags slightly behind the phase of the shock.

The phases of the three shock peaks in this example are shown in Table 1.

Table 1 Simulation shock phase

The mixed vibration signal of the reciprocating compressor is simulated, and the impact feature is extracted. The steps are as follows:

Perform EEMD decomposition on the mixed signal to obtain several orders of IMF;

Calculate the kurtosis of each order IMF and the correlation coefficient with the original signal, and finally screen out the 2nd, 3rd, and 4th order IMFs;

Adaptive variable-scale morphological filtering is performed on the screened IMF, and the shock components in the filtered signal are retained, while the non-shock components are suppressed;

Reconstruct the filtered signal to obtain the noise reduction signal. At this time, the harmonic signal and noise signal of the original signal are well suppressed. See Figure 4, using the Teager energy operator to amplify the mutation component of the signal;

Perform envelope smoothing to obtain the final signal to be detected;

Perform adaptive calculation of the shock threshold and perform peak detection to obtain the shock phase, see Figure 5, and see Table 2 for the detection results.

Table 2 Impact test results

From the shock phase detection results, it can be seen that whether it is a high-amplitude shock or a low-amplitude shock, or whether it is a high-frequency shock or a low-frequency shock, it can be detected accurately, and the phase error of the peak value of the shock is less than 0.5 degrees.

Embodiment two:

For example: the flash gas three-stage double-acting reciprocating compressor that has just been put into use has two first-stage low-pressure cylinders and two second-stage high-pressure cylinders, which are symmetrically arranged at both ends, and the middle body of each cylinder is installed An acceleration sensor model PCB-EX603C01 is used to monitor the vibration signal of the middle body. The acquisition device used is NI Compact RIO9047 reconfigurable embedded measurement and control system, which is matched with NI 9263 sound and vibration input module, which can realize high-speed acquisition of acceleration signals. In this example, the running speed of the compressor is 1200RPM, and the sampling rate is set to 10240Hz. Take the acceleration signal of a certain two periods in the middle body of the #1 first-stage low-pressure cylinder and the dynamic gas pressure in the cylinder head side (HE) and crankshaft side (CE), and transform the phase into the angle domain diagram to be 0°～720°, Referring to FIG. 6 , where the acceleration signal is taken out separately, refer to FIG. 7 . After EEMD-adaptive variable-scale morphological noise reduction processing, the noise-reduced signal is shown in Figure 8. It can be clearly seen that the non-impact components in the signal have been significantly suppressed. After filtering by Teager energy operator (as shown in Figure 9) and envelope smoothing filtering, the impact threshold is adaptively identified, and finally the impact phase is determined by peak phase detection, as shown in Figure 10, where the horizontal line is the impact The detection threshold line, the vertical line is the detection result of the shock peak. The results of the peak phase detection of the shock are shown in Table 3.

Table 3 Impact test results

From the shock peak phase detection results combined with the dynamic pressure in HE and CE cylinders, it can be clearly seen that the phase stability of shock 1 and shock 2 respectively indicates the seat of the exhaust valve at the CE end and the seat of the suction valve at the HE end; the phases of shock 3 and 5 are relatively stable , indicating that the exhaust valve at the HE end is seated and the suction valve at the CE end is seated; while the shock 4 phase fluctuates greatly, combined with the dynamic pressure in the cylinder at the HE end, there is a short-term abnormal increase in air pressure around 280° in a single cycle, which is judged to be The vibration of the exhaust valve at the HE end caused multiple consecutive collisions between the valve plate and the valve seat, thereby introducing additional shocks. In addition, from Figure 11, Figure 12 and Figure 13, it can be observed that there are small impact components around 220° and 380°, combined with the obvious large air pressure fluctuations near these two phases in the dynamic pressure of the gas in the cylinder at the HE and CE ends, It is comprehensively judged that the exhaust valve at the HE and CE ends does not match the exhaust pressure, resulting in the vibration of the valve plate, which leads to a slight collision between the exhaust valve plate and the valve seat during the exhaust process and introduces additional impact.

The above-described embodiments are only preferred embodiments for fully illustrating the present invention, and the protection scope of the present invention is not limited thereto. The equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are all within the protection scope of the present invention. The protection scope of the present invention shall be determined by the claims.

Claims (8)

1. A fault diagnostic device for a floating platform reciprocating compressor, comprising:

the acquisition module is used for acquiring vibration signal data of the floating platform reciprocating compressor and acquiring and transmitting the vibration signal data to an upper computer for processing;

the processing module is connected with the acquisition module and is used for processing the acquired vibration signals through EEMD-adaptive variable-scale morphological filtering;

the extraction module is connected with the processing module and used for carrying out peak detection to extract the phase characteristics of the impact components in the signals and capturing the parameter characteristics of the impact;

the diagnosis and evaluation module is connected with the extraction module, establishes a fault characteristic knowledge base of the reciprocating compressor, compares the fault characteristic knowledge base according to the vibration signal impact characteristics captured by the extraction module, and outputs a diagnosis result;

and the visualization module is connected with the diagnosis evaluation module and the acquisition module, visualizes the diagnosis result and the acquired data, and simultaneously carries out webpage interactive access through the remote terminal to realize remote data processing and calling.

2. A fault diagnosis method for a floating platform reciprocating compressor is characterized by comprising the following steps:

s01: decomposing a vibration signal to be processed to obtain a plurality of IMF components;

s02: screening out certain orders of IMFs containing obvious impact characteristics from a plurality of IMF components through indexes;

s03: respectively carrying out self-adaptive variable-scale morphological filtering on the screened IMF components to filter out non-impact components;

s04: reconstructing the IMF component after the adaptive variable-scale morphological filtering to obtain a noise reduction signal;

s05: amplifying amplitude-frequency mutation components in the noise reduction signal;

s06: calculating the envelope of the abrupt change component, and performing smooth filtering on the envelope by using a smoothing filter;

s07: carrying out peak phase detection to extract the phase characteristics of the impact components in the signals;

s08: after the impact phase characteristics are obtained, capturing parameter characteristics to further carry out fault diagnosis on the reciprocating compressor;

s09: and visualizing the diagnosis result and the acquired data, and remotely processing and calling the data.

3. The diagnostic method of claim 2, wherein: the step S02 uses a kurtosis-correlation index, including:

first, kurtosis index:

for discrete signal x (i) = { x_i| i =1,2, \8230;, n }, where n is the number of signal points, kurtosis K is defined as:

in the formula

Is the mean of x (i) and is defined as:

respectively calculating kurtosis values of IMF components of all orders, and rejecting components with kurtosis values smaller than 5.0 to realize the first-step screening of the IMF components;

step two, correlation indexes are as follows:

calculating the correlation coefficient of each IMF component and the original function, and assuming that there are two continuous time domain signals x (t), y (t), the correlation coefficient R (xy) is defined as:

and selecting IMF with the correlation coefficient larger than 10% to realize the second step of screening.

4. The diagnostic method of claim 2, wherein: in step S03, the adaptive variable-scale morphological filtering includes:

for a discrete signal f (N) with a total number of points N, where the argument N =0,1, \ 8230;, N-1 and the total number of pointsThe sequence of structural elements g (M) for M, where the argument M =0,1, \8230, M-1, N < M, defines the inflation operator

And erosion operator

Comprises the following steps:

wherein,

denotes the expansion of f (n) with respect to g (m),

denotes the corrosion of f (n) with respect to g (m);

defining a morphology opening operator by a sequential combination of an expansion operator and an erosion operator

And a morphological closing operator (·):

wherein,

represents a morphology opening operator of f (n) with respect to g (m), and (f.g) (n) represents a morphology closing operator of f (n) with respect to g (m);

further, a morphology open-close (OC) operator and a morphology close-open (CO) operator are defined by sequential combinations of the morphology open operator and the morphology close operator:

wherein OC [ f (n) ] represents the morphological on-off operator of f (n), and CO [ f (n) ] represents the morphological on-off operator of f (n);

a cascade morphological filter Γ [ f (n) ] formed by combining a morphological open-close operator and a morphological closed-open operator to f (n):

5. the diagnostic method of claim 4, wherein: the filter structure element uses a flat structure with a height of 0, and a structure element sequence g (N, m, k) (N =0,1, \8230;, N-1) is defined for a discrete signal f (N) (N =0,1, \8230;, N-1 m =0,1, \8230;, k-1), wherein N is the number of signal points, k is the width of the structure element at the nth sampling point, and the width k is determined as follows:

(1) The phase and amplitude of each point of the signal are recorded as

x(i)＝{x_i| i =1,2, \8230 |, N }, where N is a number of signal points;

(2) Calculating the phase and amplitude of all local extreme points in the signal, and respectively recording as phi (j) = { phi (j) =_j∣j＝1,2,…,M}、y(j)＝{y_j-j =1,2, \8230 \ M }, where M is the number of local extremum points;

(3) Carrying out linear normalization on amplitude values of local extreme points of signals y_norm(j)：

Wherein, y_jThe amplitude value of the jth extreme point is represented, min { y (j) } is the minimum value of the extreme point sequence y (j), and max { y (j) } is the maximum value of the extreme point sequence y (j);

(4) Calculating a signal waveform scale s:

(5) Defining a non-linear mapping

Fuzzifying the boundary of impact and non-impact components, changing the amplitude distribution of normalized local extreme points, and determining the width k (i) = { k) = of the structural element corresponding to each point in the signal_i| i =1,2, \8230 |, N } is determined by:

wherein the symbols

Represents rounding up, mapping

Is defined as:

in the formula, a is a variable parameter and is used for adjusting the amplitude distribution of extreme points, a recommended value range a belongs to [2,8], k (i) is used as the element width, and the cascade type morphological filtering is carried out on signals to realize the effective inhibition of non-impact components.

6. The diagnostic method of claim 2, wherein: step S05, amplifying amplitude-frequency mutation components in the noise reduction signals by using a Teager energy operator:

specifically, for a discrete signal x (i) with a total number of points n (i =1,2, \ 8230;, n), a three-point symmetric difference energy operator is constructed and smoothed, and an energy operator Ψ [ x (i) ] is defined as:

7. the diagnostic method of claim 2, wherein: the step 06: obtaining a Hilbert envelope of a mutation component, and performing smooth filtering on the envelope by using a Savitzky-Golay smoothing filter;

in particular, for a continuous-time signal x (t), the Hilbert transform thereof

Is defined as:

structure analytic signal

The Hilbert envelope of the signal x (t) is then:

dividing the envelope filtered by the Savitzky-Golay smoothing filter into n small sections of section number in an angular domain diagram, calculating the average amplitude of each small section signal, and arranging the small sections into a group of series l [ i [ [ i ]]I =0,1, \ 8230;, n-1, the mean value of the sequence is calculated

Using 3-sigma rule to remove the small segment containing obvious impact characteristic and the rest m small segment sequence

Characterizing the relatively stationary components of the signal, a series of r [ j ] is set]Has an average value of

The difference between the maximum and minimum values is Δ r, and the weighting factor a is used to adjust the threshold, then the impact detection threshold can be defined as

8. The diagnostic method of claim 7, wherein: the number of segments n =72, i.e. one segment every 5 °.

Images