Disclosure of Invention
The invention aims to provide a pumping unit check valve leakage detection method based on adaptive filtering and impact recognition, which can detect the leakage state of a check valve under different working conditions in real time and has reasonable and accurate detection results.
To achieve these objects and other advantages in accordance with the purpose of the invention, a method for detecting a single flow valve leak of a pumping unit based on adaptive filtering and shock recognition is provided, which comprises:
collecting vibration signals to obtain the working cycle of the pumping unit;
identifying a lower stroke period of each work cycle in the vibration signal, and extracting a pulse signal from the lower stroke period;
and if the pulse signal meets a preset rule, judging that the check valve is in a leakage state.
Preferably, the method for detecting the leakage of the check valve of the oil pumping unit based on the adaptive filtering and the impact recognition judges whether the check valve is in a leakage state or not according to the average pulse amplitude and the pulse density of the pulse signal.
Preferably, the method for detecting the leakage of the check valve of the pumping unit based on the adaptive filtering and the impact recognition comprises the following steps:
calculating the effective value of each moment of the vibration signal X (i) to obtain the time sequence x of the effective valuee(i);
For xe(i) Fourier transform is carried out to obtain an effective value frequency spectrum;
according to the effective value frequency spectrum, calculating the frequency f corresponding to the spectral line with the maximum amplitudemWith a period of work of Tc=1/fm。
Preferably, the method for detecting the leakage of the single flow valve of the pumping unit based on the adaptive filtering and the impact recognition comprises the following steps:
step one, let t0Step Δ T ═ 0, step Δ T ═ Tc/3;
Step two, from
Middle intercept t
0,t
0The data sequence between + Δ T constitutes a time sequence of significant values
(
j p1, 2.., l), where l is t
0,t
0The number of effective value data between + deltat,
is the time series of valid values of the p-th work cycle,
maximum value of (A) is V
maxMinimum value of V
minAnd an average value of V
ave(ii) a Threshold value V
d_stroke=V
min+(V
max-V
min)/5;
Step three, calculating
Average value of (V)
p_aveIf V is
p_ave<V
aveAnd V
p_ave<V
d_strokeAnd is also
All data in (1) are less than V
aveThen the down stroke start time t is recorded
d_sIs t
0If the above condition is not satisfied, then t is set
0=t
0+1, repeating the step two;
step four, let t0=td_s+1;
Step five, from
Middle intercept t
0,t
0The data sequence between + Δ T constitutes a new time sequence
(
j p1, 2.., l), where l is t
0,t
0The number of effective value data between + Δ T;
step six, calculating
Average value of (V)
p_aveIf V is
p_ave<V
aveAnd V
p_ave<V
d_strokeAnd is also
All data in (1) are less than V
aveThen set t
0=t
0+1, repeating the step five; otherwise recording the down stroke end time t
d_eIs t
0。
Preferably, the method for detecting the loss of the check valve of the pumping unit based on the adaptive filtering and the impact recognition comprises the following steps:
from the vibration signal X (i), intercept from tsVibration time sequence X between the first p-th work cyclep(j) (j ═ 1, 2.. times, M), M is tsAnd ts+TcThe number of vibration data between; from Xp(j) Middle intercept td_sAnd td_eForm a new time series X by the vibration datas_p(js)(j s1,2, L), wherein L is the total number of vibration data during the downstroke;
determining a band-pass filter B (F)c_kmax,ΔBw) The kurtosis value of the frequency band obtained by the processing of the band-pass filter is maximum, Fc_kmaxIs the filter center frequency, Δ BwThe width of the band pass filter;
using a band-pass filter B (F)
c_kmax,ΔB
w) To X
s_p(j
s) Filtering to obtain
Method for obtaining vibration impact envelope time sequence of pulse signal by adopting digital envelope demodulation technology
Identifying a vibration impact envelope time series waveform
And obtaining the number of the pulses and the amplitude of the pulses.
Preferably, the method for detecting the leakage of the check valve of the pumping unit based on the adaptive filtering and the impact recognition includes:
if Ep_stroke≥Ep_stroke_maxAnd L isp_stroke≥Lp_stroke_maxIf so, judging that the check valve is in a leakage state;
wherein the content of the first and second substances,
T
p_strokethe total accumulated duration of the data used for detecting the pulse in all P working cycles; e
p_strokeThe average pulse amplitude identified in the down stroke; l is
p_strokeThe pulse density identified in the down stroke; e
p_stroke_maxAllowable average pulse amplitude, L, in a loss-free state for maximum energy tolerance
p_stroke_maxIs the allowable pulse density in the maximum tolerable leakage-free state.
Preferably, the pumping unit check valve leakage detection method based on adaptive filtering and impact recognition determines a band-pass filter B (F)c_kmax,ΔBw) The method comprises the following steps:
step A, setting the acquisition frequency as FsLet fc=ΔBw/2;
Step B, constructing a band-pass filter B (f)
c,ΔB
w) To X
s_p(j
s)(
j s1, 2.., L) is filtered to obtain a new time sequence
Step C, calculating
Kurtosis value of
And synchronously recording B (f)
c,ΔB
w) And
step D, if fc<Fs/2-ΔBwThen order fc=fc+ΔBwRepeating the step B, otherwise, carrying out the next step;
step F, from all saved
Finding the band-pass filter corresponding to the maximum value, namely B (F)
c_kmax,ΔB
w)。
Preferably, the method for detecting the leakage of the single flow valve of the oil pumping unit based on the adaptive filtering and the impact recognition is used for collecting vibration signals on the outer wall of a valve block of the oil pumping unit.
The invention at least comprises the following beneficial effects:
the high-frequency vibration acceleration sensor is fixed on the outer wall of the valve block of the check valve in a magnetic adsorption or adhesion mode, so that high-frequency vibration signals of the valve block under different working conditions of the oil pumping unit can be measured and obtained, and the detection and identification of the leakage state of the check valve are realized by automatically analyzing and detecting the vibration signals on the basis of the vibration signals. The invention can detect the leakage state of various pumping units in real time, has high detection efficiency and reasonable and accurate detection result, and does not depend on manual inspection.
Additional advantages, objects, and features of the invention will be set forth in part in the description which follows and in part will become apparent to those having ordinary skill in the art upon examination of the following or may be learned from practice of the invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
It will be understood that terms such as "having," "including," and "comprising," as used herein, do not preclude the presence or addition of one or more other elements or groups thereof.
There are two main reasons for the loss of a single flow valve:
1) because the check valve works in a long-term oil environment, deformation, cracks, fracture, abrasion and the like are generated.
2) The check valve is blocked due to impurities such as sand and wax contained in the oil.
The two reasons can cause the problems of micro leakage and leakage caused by the fact that the oil in the oil pipe reversely flows into the oil pipe in the well due to the fact that the check valve is not closed tightly in the down stroke process of the oil pumping unit.
In the process of upward flushing of the oil pumping unit, the pressure of a front oil pipe of the check valve is higher than the pressure in the oil return pipe, the valve body is jacked open, oil flows into the oil return pipe through the check valve, and then flows to a direct-current metering room and the like. In the process, no matter whether the check valve has leakage or not, because the well head pressure is higher than the pressure in the oil return pipe, the process oil flow always flows into the oil return pipe from the oil pipe.
During the down stroke, the pressure in the oil pipe is reduced, the pressure in the oil pipe is lower than the pressure in the oil return pipe, although the check valve is closed, if the check valve is not closed tightly, oil flows back to the oil pipe from the oil return pipe.
Therefore, whether the single-flow valve leakage problem exists is mainly judged through data of a down stroke process.
Generally speaking, the orifice size of the clearances formed by deformation, cracking, wear of the check valves and seizure caused by small particle impurities in the oil is small, typically below 1 mm. During the downstroke, there is a pressure differential between the return line pressure and the oil line, which causes the return line oil to flow through this small bore gap into the oil line. Under the same pressure difference, the smaller the gap aperture is, the higher the flow velocity of the return oil is, so that the high-velocity return oil can generate stronger high-velocity impact jet. This impinging jet does not act for a long time, but acts on the valve block, causing short-time resonance of the valve block. The vibration acceleration signal measured from the outer wall of the valve stack is a short duration shock resonance signal.
As shown in fig. 1 and 2, wherein fig. 1 is the valve train external vibration (acceleration) signal measured during the downstroke under no-leak conditions, and fig. 2 is the valve train external vibration (acceleration) signal measured during the downstroke under leak conditions, in particular, in fig. 2, there is a significant short-time high-frequency shock pulse signal, while in fig. 1, there is no significant shock pulse signal. If the check valve body is broken and abraded more, the probability of vibration impact is higher, and the vibration (acceleration) signals measured on the outer wall of the valve group are represented as a series of impact pulse signals due to repeated high-speed impact jet flow, and the frequency density and pulse interval of the impact pulse signals are not fixed and are typical random pulse signals.
Therefore, whether the leakage fault exists in the check valve can be identified by identifying a short-time impact pulse signal caused by leakage in the down stroke process, and carrying out statistical analysis on pulse energy and pulse frequency density.
The invention provides a method for detecting the leakage of a single flow valve of an oil pumping unit based on adaptive filtering and impact recognition, which comprises the following steps:
the original vibration acceleration signal measured at the outer wall of the valve bank is a complex signal containing various frequency components caused by noise, vibration of a motor of the pumping unit and fluid pulsation of an oil return pipe, and the impact pulse signal mixed in the original vibration (acceleration) is almost difficult to identify by directly adopting the original vibration. Fig. 3 is an original vibration signal, and transient impact signals mixed in various frequency components are hardly detected.
Therefore, the invention provides a method for extracting transient impact signals in a vibration acceleration signal in a downstroke process by adopting a self-adaptive band-pass filtering method, further solving envelope signals containing the impact signals by using a digital envelope demodulation method, counting the impact envelope signals on the basis of the impact envelope signals to obtain impact energy and impact frequency density in a unit time, and detecting whether the check valve has leakage or not by using the two parameters as indexes. In the invention, a high-frequency vibration acceleration sensor is required to be fixed on the outer wall of the valve group in a magnetic adsorption or adhesion mode, vibration acceleration data acquisition equipment is utilized to acquire vibration acceleration data with the frequency of not less than 30kHz and more than 5 minutes continuously, and detection and identification are carried out on the basis of the vibration acceleration data. The main content of the invention is how to use the vibration acceleration data to detect and identify the leakage state of the check valve.
Setting the time sequence of the original vibration waveform as X (i) (1, 2.. multidot.N), wherein N is the total number of vibration data acquired in an acquisition time period and the acquisition frequency is Fs(Fs≥30kHz),N=Ts·FsWherein T issFor continuous acquisition of total duration, TsMore than or equal to 300 seconds.
The specific steps of the method are described in detail below.
Identifying the stroke frequency and the work cycle of the oil pumping unit according to the collected vibration signals at the outer wall of the valve group;
defining the stroke frequency (the work done times per minute) of the pumping unit as ScBecause the stroke frequency of each pumping unit is not consistent (even the stroke frequency of the same pumping unit configured in different time periods is not the same), in order to identify the down stroke process of the pumping unit, the stroke frequency of the pumping unit needs to be identified from the acquired vibration signal to obtain the period of work of the pumping unit, and the starting time of the up-down stroke is identified according to the change of the signal in the period of work. The specific method comprises the following steps:
the time-of-day effective values are calculated in units of 1 second for the original vibration continuous waveform signal x (i), and a time series of continuous effective values over the entire time is obtained, as shown in fig. 4. FIG. 4 is a waveform of a time series of n consecutive seconds (n ≧ 600) of significant values, one significant value per second, for a total of n data, denoted by xe(i)(i=1,2,...,n)。
For the time series x of the n effective valuese(i) A Fast Fourier Transform (FFT) was performed to obtain spectral data for a time series of significant values, where the acquisition frequency was 1.0Hz, as shown in fig. 5.
Calculating the corresponding frequency of the spectral line with the maximum amplitude according to the effective value spectrum to obtain the main frequency fmAnd converted to counts per minute.
Sc=60fm (1)
As shown in fig. 5 fm0.033Hz, corresponding stroke number Sc0.033 × 60 ═ 1.98. That is to say thatThe stroke frequency of the oil pumping unit to be measured is 1.98, and the oil pumping unit does work for 1.98 times per minute.
After the stroke frequency of the oil pumping unit is obtained, calculating a work cycle according to the stroke frequency:
Tc=1/fm (2)
identifying impact pulse, pulse energy and impact pulse density caused by leakage from the original vibration data;
the continuously measured vibration signal comprises data of a plurality of working cycles of the pumping unit, each working cycle of the pumping unit comprises an upper stroke process and a lower stroke process, in the scheme, the pulse signal is mainly extracted from the data of the lower stroke, and then the amplitude energy and the pulse quantity of the pulse are identified. And (4) performing the identification and statistics on all work cycles of the test data, and finally obtaining the amplitude energy and the pulse number in the average time. The method comprises the following specific steps:
substep one, setting TcFor the working cycle of the pumping unit, TcObtained according to the step one, the set time is ts=0;
Substep two, from the time series x of the continuous effective values of the valve group vibration signal
e(i) In, intercept from t
sThe time sequence data of the effective value between one working cycle (denoted as the p working cycle) is obtained
(m is t)
sAnd t
s+T
cNumber of data in between);
substep three, from the sequence of significant values of the p-th work cycle
In, positioning the down stroke start t
d_sEnd time t
d_e. First from
Middle calculated maximum value V
maxMinimum value V
minAnd the mean value V
ave(ii) a Setting a threshold value:
Vd_stroke=Vmin+(Vmax-Vmin)/5 (3)
the specific flow steps for positioning the starting time and the ending time of the down stroke are as follows:
(1) let t0Step Δ T ═ 0, step Δ T ═ Tc/3;
(2) From
Middle intercept t
0,t
0The data sequence between + Δ T constitutes a new time sequence
(
j p1, 2.., l), where l is t
0,t
0The number of effective value data between + Δ T;
(3) computing
Average value of (V)
p_aveIf V is
p_ave<V
aveAnd V
p_ave<V
d_strokeAnd is also
All data in (1) are less than V
aveThen the down stroke start time t is recorded
d_sIs t
0If the above condition is not satisfied, then t is set
0=t
0+1, repeating step (2);
(4) let t0=td_s+1;
(5) From
Middle intercept t
0,t
0The data sequence between + Δ T constitutes a new time sequence
(
j p1, 2.., l), where l is t
0,t
0The number of effective value data between + Δ T;
(6) computing
Average value of (V)
p_aveIf V is
p_ave<V
aveAnd V
p_ave<V
d_strokeAnd is also
All data in (1) are less than V
aveThen set t
0=t
0+1, repeating step (5); otherwise recording the down stroke end time t
d_eIs t
0And the flow ends.
A fourth substep of intercepting the time sequence of vibrations from t, from the original time sequence of vibrations X (i) measured outside the valve groupsStarting to obtain the original vibration time sequence data X between the p work cyclesp(j) (j ═ 1, 2.. times, M) (M is tsAnd ts+TcNumber of raw data in between); in the process from Xp(j) Middle intercept td_sAnd td_e(i.e. t)s,ts+TcOriginal vibration waveform data of a down stroke process in a work cycle) to form a new time sequence Xs_p(js)(j s1, 2.., L), where L is the total number of vibration data during the downstroke.
Step five, solving the optimal band-pass filter in a self-adaptive manner;
the kurtosis is an index for describing the kurtosis of the waveform, is very sensitive to an impact signal, and can judge the non-stationary strength of the signal. However, in the original signal containing various complex signals or the signal with low signal-to-noise ratio, it is difficult to detect the critical fault signal by kurtosis. For each transient signal, an optimal frequency band B (F) is associatedc_kmax,ΔBw)(Fc_kmaxIs the filter center frequency, Δ BwBandpass filter width). In this frequency band, the kurtosis of this transient signal is greatest. The impulse signal is a very typical transient signal, so in the actual analysis calculation process, if a frequency band B (F) can be foundc,ΔBw) In this frequency band, the kurtosis value reaches the maximum valueTo find information about the transient signal, and the frequency band B (F)c,ΔBw) That is, the optimum band-pass filter B (F) such that its kurtosis is maximum corresponding toc_kmax,ΔBw)。
Therefore, an adaptive fast algorithm can be used to set a fixed width filter B (F) for different objectsc,ΔBw) By varying the centre frequency FcCalculating the kurtosis value of the waveform under different filters, and then the filter B (F) corresponding to the maximum kurtosis valuec_max,ΔBw) Is an optimal band pass filter.
In the scheme, according to the signals actually measured on site, determining to adopt Delta Bw500 Hz. The specific process of adaptively solving the optimal band-pass filter is as follows:
(1) let the acquisition frequency be FsLet fc=ΔBw/2;
(2) Constructing a band-pass filter B (f)c,ΔBw) For time sequence Xs_p(js)(j s1, 2.., L) is performed
Obtaining a new time sequence after the filter
(3) Computing
Kurtosis value of
And synchronously recording B (f)
c,ΔB
w) And
(4) if f isc<Fs/2-ΔBwThen order fc=fc+ΔBwRepeating the step (2), otherwise, performing the next step;
(5) from all saved
Filter B (f) corresponding to the maximum value in the search
c,ΔB
w) Then this filter is the optimal filter B (F) that maximizes the kurtosis value
c_kmax,ΔB
w)。
Performing band-pass filtering based on the optimal band-pass filter to obtain an impulse pulse signal waveform;
determining the optimal band-pass filter B (F) by an adaptive method
c_kmax,ΔB
w) Then, the band-pass filter is used to process the original signal X
s_p(j
s) Filtering to obtain clear impact pulse waveform signal
FIG. 7 is a vibration waveform after filtering an original vibration signal using an optimal band pass filter
From the waveform, a plurality of high-amplitude impact pulse signals can be clearly observed.
Step seven, adopting digital envelope demodulation to solve the impulse envelope waveform;
after obtaining the impulse waveform signal, the envelope time sequence waveform of the impulse signal is obtained by adopting digital envelope demodulation technology such as Hilbert (Hilbert) conversion and the like
Step eight, counting the total pulse number under a given threshold value according to the vibration impact envelope time sequence waveform;
after the vibration impact envelope waveform is obtained, the number of impact pulses and the amplitude of the pulses can be obtained through identification of the envelope signal.
Assuming a given minimum pulse amplitude threshold of Pp_stroke(for identifying valid pulses), that can be calculated by scanning the temporal sequence of the envelope of the oscillation to obtain the following numbersAccording to the following steps:
the vibration impact amplitude exceeds P in the process of the down stroke in the pth working cycle
p_strokeThe cumulative impact amplitude of (d);
the vibration impact amplitude exceeds P in the process of the down stroke in the pth working cycle
p_strokeThe number of the impact pulses is accumulated;
for the total duration of the calculation data for the identification during the down stroke in the p-th work cycle, i.e. for the identification
(units are seconds);
step nine, if all the working cycles are calculated, the next step is carried out; otherwise let ts=ts+TcContinuing to execute the substep two;
and a substep ten, assuming that the total effective work cycle number is P, calculating the following characteristic indexes:
wherein T isp_strokeIs the total accumulated time length of data used for detecting the pulse in all P work cycles, and takes minutes asA unit; ep_strokeThe average pulse amplitude identified in the down stroke; l isp_strokeThe number of pulses per minute (pulse density) identified in the downstroke. And the above-mentioned Ep_stroke、Lp_strokeIt is the main characteristic parameter that identifies the detection of a single flow valve leak.
In summary, the main characteristic parameters in the scheme of the present invention are as follows:
parameter(s)
|
Description of the parameters
|
Ep_stroke |
Average pulse amplitude in the down stroke;
|
Lp_stroke |
number of pulses per minute in the downstroke (pulse density) |
Step three, judging and identifying the leakage state of the check valve;
in obtaining Ep_stroke、Lp_strokeThe leakage state of the check valve is required to be detected and identified. The drop-out determination recognition rule is set as follows:
if E isp_stroke≥Ep_stroke_maxAnd L isp_stroke≥Lp_stroke_maxThen the check valve may be considered to be in a leak-off state;
in the above, Ep_stroke_maxAllowable average pulse amplitude, L, in a loss-free state for maximum energy tolerancep_stroke_maxThe allowable pulse density in a leakage-free state is the maximum tolerable;
based on the test data statistics, thresholds may be employed that determine:
Lp_stroke_maxnot less than 1.0, i.e., at least one pulse per minute;
Ep_stroke_maxnot less than 6000 (mm/s)2) I.e. the average amplitude is not less than 6000 (mm/s)2);
Another calculation Ep_stroke、Lp_strokeAnother threshold value used is Pp_strokeMainly used for identifying effective pulses, and P can be selected according to data statistics actually testedp_stroke=5000(mm/s2)。
The number of apparatuses and the scale of the process described herein are intended to simplify the description of the present invention. The application, modifications and variations of the present invention to a single flow valve drop-out detection method for a pumping unit based on adaptive filtering and shock recognition will be apparent to those skilled in the art.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.