CN102857194A - Frequency band noise elimination method of continuous pressure wave signal of drilling fluid - Google Patents

Abstract

The invention provides a frequency band noise elimination method of a continuous pressure wave signal of a drilling fluid. The frequency band noise elimination method comprises the following steps of: constructing an adaptive filter algorithm according to a minimum mean square value of an error signal by adopting an adaptive filter, taking the continuous pressure wave signal, of the drilling fluid, containing noises as a desired signal of the adaptive filter, and taking a carrier extracted from a synchronizing signal as an input signal of the filter; detecting the pressure change of the drilling fluid by a pressure sensor; and sampling and quantizing an electric signal of the pressure sensor into a digital signal by a computer data collection and processing system, automatically regulating a filter weight coefficient matrix, carrying out adaptive filtering processing on the digital signal to enable the output value of the adaptive filter to be approximate to a drilling fluid pressure signal which reflects information of the measurement while drilling in a pit, and realizing the frequency band noise elimination of the continuous pressure wave signal of the drilling fluid.

Description

The band noise removing method of drilling fluid continuous pressure ripple signal

Technical field:

The present invention relates to measurement while drilling/well logging during in a kind of oil/gas drilling process and upload the processing method of signal, particularly a kind of band noise removing method that carries out the drilling fluid continuous pressure ripple signal of downhole data transmission based on drilling fluid.

Background technology:

DHM-MWD/well logging during (MWD/LWD) is a kind of modern drilling ancillary technique of measuring and transmit in real time down-hole information in drilling process.In the drilling process, drilling fluid is pumped to the down-hole by the earth's surface by drill string, is used for the lubrication and cooling of drill bit and upwards returns out well head by the ring-like space of the borehole wall and drill string from the bit port ejection, and the MWD/LWD instrument is installed in the drill collar on drill bit top, sees accompanying drawing 1.In the MWD/LWD instrument, be installed in that transducer in the nearly drill bit drill string obtains measurement data and by drilling liquid pressure information telemetry system transmissions to ground, information telemetry is by modulating the drilling liquid pressure in the drill string and DHM-MWD/well logging during data are transmitted in the propagation of pressure wave in drill string.The pressure information remote measurement adopts base band pressure pulse or continuous pressure ripple to transmit down-hole information usually, and wherein the continuous pressure ripple has the down-hole information transmission rate more much higher than base band pressure pulse, is the developing direction of pressure information telemetry; Continuous pressure ripple signal mainly contains drilling liquid pressure phase shift keying (PSK) signal of modulated carrier phase place, comprise drilling liquid pressure differential phase keying (DPSK) (DPSK) signal and Quadrature Phase Shift Keying (QPSK) signal, signal spectrum is concentrated, and has the characteristics of frequency band transmission.Drilling fluid continuous pressure ripple signal can run into very large noise jamming in the transmission course earthward from the shaft bottom, noise mainly comes from the pressure oscillation that drill bit vibrations, mud motor stall, Tubing Buckling etc. cause, has very large randomness, because frequency is lower, noise spectrum shows the band white Gaussian noise that is limited.Although the frequency of down-hole noise source is not high, but because drilling fluid continuous pressure ripple signal is a kind of band signal, still have partial noise entering signal frequency band, cause the signal to noise ratio of signal seriously to reduce, the down-hole pressure signal that standpipe detects contains the larger random pressure noise of relative amplitude, and noise and signal generation spectral aliasing.Present drilling liquid pressure signal processing method, such as matched filter method, adaptive compensation, wavelet analysis method etc. mainly for the lower pressure signal that transmits in the base band mode of transmission rate; And for the relatively high drilling fluid continuous pressure ripple signal of transmission rate, the processing method of above-mentioned drilling hydraulic force signal all can't effectively be eliminated the noise effect of entering signal frequency band.

Summary of the invention:

Purpose of the present invention is exactly the defective that exists for existing signal processing technology, and a kind of band noise removing method of drilling fluid continuous pressure ripple signal is provided.

The present invention adopts sef-adapting filter to eliminate the band noise of drilling fluid continuous pressure ripple signal.Sef-adapting filter is a kind of digital filter, formed by signal input part, desired signal input, filter output, error signal output, filter weight coefficient matrix and adder, wherein, the matrix coefficient of the weight coefficient matrix of filter is controlled by error signal.Sef-adapting filter is generally used for processing the narrow band signal of radio communications system, and so-called narrow band signal refers to the ratio of signal band and carrier wave much smaller than 1, and the relative carrier frequency variation of signal frequency is little in the frequency band.And drilling fluid continuous pressure ripple signal is as a kind of machinery modulation signal, because mechanical system inertia and pressure signal transmit in drilling fluid, make its carrier frequency be limited in very low tens hertz, the ratio of its modulation signal frequency band and carrier wave is a kind of typical broadband signal usually close to 1; For the application self-adapting filter comes treated drilling fluid continuous pressure ripple signal, the present invention is take the underground signal that contains Gaussian noise as desired signal, the carrier wave that extracts from synchronizing signal is as input signal, the mean-square value of the difference of desired signal and filter output signal forms error signal, sef-adapting filter by adjusting adaptively the matrix coefficient of filter weight coefficient matrix, makes the output signal of filter constantly approach desired signal under the error signal effect; Because the mean-square value of Gaussian noise is non-vanishing, when error signal for hour, error signal approaches the mean-square value of Gaussian noise, this moment, the output signal of filter approached the drilling fluid continuous pressure ripple signal that reflects measurement while drilling information.

The weight coefficient column matrix W (n) of sef-adapting filter is the weight coefficient column matrix of 1 * N dimension, if input signal matrix X (n) is N * 1 dimension row matrix, then the output signal matrix Y (n) of filter is one 1 * 1 dimension matrix, can be expressed as the output signal that a multinomial forms

$Y\left(n\right)=W\left(n\right)X\left(n\right)=y\left(n\right)=\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{w}_i\left(n\right)x(n-i)---\left(1\right)$

Wherein, w _i(n) be the matrix coefficient of weight coefficient matrix W (n); X (n-i) is the matrix coefficient of input signal matrix X (n), is expressed as the unit delay sampled value of input signal; N is discrete variable.

If the carrier signal of discretization is x (n)=A _cSin (ω _cN), the noise of measurement while drilling signal leading is n _w(n), the output signal of filter is

Drilling fluid continuous pressure ripple signal is drilling liquid pressure phase shift keying (PSK) signal s (n)=A _sSin[ω _cN-f (n)], desired signal is d (n)=s (n)+n _w(n), error signal is e (n)=d (n)-y (n), and then the mean-square value of error signal is

$\mathrm{\ϵ}=E\left[e^2\left(n\right)\right]=E\left\{\right[s\left(n\right)-y\left(n\right){]}^2\}+E[n_w^2\left(n\right)]+2E\{n_w\left(n\right)[s\left(n\right)-y\left(n\right)]\}---\left(2\right)$

Since s (n) and y (n) respectively with random noise n _w(n) uncorrelated, so 2E{n _w(n) [s (n)-y (n)] }=0. When error mean square value ε for hour, E{[s (n)-y (n) must be arranged] ²Minimum, therefore have

$\min \mathrm{\ϵ}=\min E{\left[A_s\sin \right[{\mathrm{\ω}}_{c}n-f\left(n\right)]-\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{w}_i\left(n\right)A_c\sin [{\mathrm{\ω}}_c(n-i)\left]\right]}^2+E\left[n_w^2\left(n\right)\right]$

Then

$\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{w}_i\left(n\right)A_c\sin \left[{\mathrm{\ω}}_c(n-i)\right]\→A_s\sin [{\mathrm{\ω}}_{c}n-f\left(n\right)]---\left(3\right)$

The physical significance of formula (3) for when the error signal mean-square value hour, this moment, the weight coefficient matrix of filter reached best, can be with value A of a certain moment of carrier wave _cSin (ω _cN) and N-1 be worth in the past constantly A _cSin[ω _c(n-i)] stack of weighted linear removes to approach the drilling fluid continuous pressure ripple signal value s (n) in this moment=A _sSin[ω _cN-f (n)], wherein N is the digital filter exponent number.Namely can utilize

Come reconstruct drilling fluid continuous pressure ripple signal, this moment, the output signal of filter was

$y\left(n\right)=s\left(n\right)=\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{w}_i^{*}\left(n\right)A_c\sin \left[{\mathrm{\ω}}_c(n-i)\right]---\left(4\right)$

In the formula (4),

Be best weights coefficient matrix W ^*(n) matrix coefficient.The matrix coefficient of filter weight coefficient matrix W (n) is determined by formula (5)

w(n+1)=w(n)+2μe(n)x(n) （5）

Reach minimum in the error signal mean-square value, and satisfy

The time, can obtain best weights coefficient matrix W ^*(n).

In the formula (5), μ is the adaptive step factor of decision systems stability and convergence rate.

Formula (4) and formula (5) consist of adaptive filter algorithm, adopt computer to carry out data acquisition and the processing of pressure signal.Wherein, the reconstruct of drilling fluid continuous pressure ripple signal realizes according to formula (4), because noiseless in the reconstruction signal, thereby reach the purpose of eliminating drilling fluid continuous pressure ripple signal band noise.

The invention has the beneficial effects as follows: adopt sef-adapting filter to eliminate the band noise impact of drilling fluid continuous pressure ripple signal, improve the signal to noise ratio of signal, process and the signal demodulation lays the foundation for further signal, reach the error rate that reduces transfer of data and the purpose that increases the drilling liquid pressure signal transmission distance.

Description of drawings:

Fig. 1 is the drilling rig schematic diagram that includes measurement while drilling/well logging during instrument and drilling liquid pressure input and processing in the prior art.

Fig. 2 is sef-adapting filter structural representation of the present invention.

Fig. 3 is computer data acquiring of the present invention and process software block diagram.

Fig. 4 a is the wave simulation figure of the drilling liquid pressure DPSK original signal of 10 binary elements among the present invention.

Fig. 4 b is the wave simulation figure that drilling liquid pressure DPSK original signal is sneaked into noise among the present invention.

Fig. 4 c is the dpsk signal wave simulation figure of sef-adapting filter output among the present invention.

Fig. 4 d is the signal waveform analogous diagram after the dpsk signal of sef-adapting filter output among the present invention passes through the normal tape bandpass filter again.

Fig. 5 a is the wave simulation figure of the drilling liquid pressure QPSK original signal of 10 binary elements among the present invention.

Fig. 5 b is the wave simulation figure that drilling liquid pressure QPSK original signal is sneaked into noise among the present invention.

Fig. 5 c is the QPSK signal waveform analogous diagram of sef-adapting filter output among the present invention.

Fig. 5 d is the signal waveform analogous diagram after the QPSK signal of sef-adapting filter output among the present invention passes through the normal tape bandpass filter again.

Among the figure: 1-drilling fluid, 2-earth's surface, 3-drill string, 4-drilling fluid pump, 5-drill bit, 6-annular space, 7-stratum, 8-drilling parameter and formation parameter measurement mechanism, 9-drilling fluid continuous pressure wave generator, 10-pressure sensor, 11-signal processing apparatus, 12-drill collar, 13-drilling fluid tank, 14-desired signal input, 15-filter input end, 16-filter weight coefficient matrix, 17-adder, 18-error signal output, 19-filter output.

Embodiment:

Further describe the present invention below in conjunction with drawings and Examples.

Figure 1 shows that the existing drilling rig that includes measurement while drilling/well logging during instrument and drilling liquid pressure input and processing.Drilling fluid 1 in the drilling fluid tank 14 injects the drill string 3 on stratum 7 by the drilling fluid pump 4 on earth's surface 2, arrive the drill bit 5 of drill string 3 bottoms, be back to earth's surface 2 by the annular space 6 between drill string 3 and the stratum 7 after bit port flows out, arrow represents the flow direction of drilling fluid 1 among the figure.Place instrument near in the drill collar 12 of drill bit 5 in drill string 3, drill collar weight is greater than other drill string, and whole drill string applies enough the pressure of the drill to drill bit 5 and is used for formation drilling 7.Be installed in the physical characteristic that drilling parameter in the drill collar 12 and formation parameter measurement mechanism 8 are used for monitoring drill-well operation and evaluation of earth formations.

In order in drilling fluid, to produce pressure oscillation and to transmit downhole data by drilling fluid, a drilling fluid continuous pressure wave generator 9 is installed in the top of drill collar 12, the drilling hydraulic force signal is uploaded to ground by drill string, through being installed in pressure sensor 10 detections on the standpipe and sending into signal processing apparatus 11 and carry out the signal processing.

As shown in Figure 2, contain noisy drilling fluid continuous pressure ripple signal access expectation signal input part 14 from the down-hole, the input signal end 15 of the carrier wave access filter that from synchronizing signal, extracts, carrier wave obtains output signal by multiplying each other with filter weight coefficient matrix 16, from filter output 19 outputs, form error signal with the signal plus of desired signal input 14 again after output signal is anti-phase, send into again filter weight coefficient matrix 16 from the error signal of error signal output 18 outputs and automatically adjust matrix coefficients.

The band noise removing method of drilling fluid continuous pressure ripple signal proposed by the invention, realize by following technical scheme:

Adopt sef-adapting filter to eliminate the impact of noise in the drilling fluid continuous pressure ripple signal band, to contain noisy drilling fluid continuous pressure ripple signal as the desired signal of sef-adapting filter, the carrier wave that extracts from synchronizing signal is as the input signal of filter, LMS least mean square according to error signal makes up adaptive filter algorithm, under the error signal effect, pass through automatically to adjust the matrix coefficient of filter weight coefficient matrix, make the output valve of sef-adapting filter approach the drilling hydraulic force signal that reflects DHM-MWD information, reach the purpose of noise in the erasure signal frequency band.

With reference to accompanying drawing 1, the pressure that pressure sensor 10 detects drilling fluid 1 changes, and signal processing apparatus 11 is computer data acquisition and treatment system, and the signal processing method of employing is adaptive-filtering method of the present invention.

With reference to accompanying drawing 2, desired signal contains noisy drilling fluid continuous pressure ripple signal for coming from the down-hole, filter input signal is carrier wave, the adaptive filter algorithm that through type (4) and formula (5) consist of forms the row input matrix with carrier wave by the time sampling, row input matrix and filter weight coefficient column matrix 16 multiply each other and obtain output signal, the anti-phase addition of output signal and desired signal forms error signal, error signal is sent into filter weight coefficient column matrix 16, automatically adjusts the matrix coefficient of filter weight coefficient column matrix 16 with the LMS least mean square of error signal; The computing that so goes round and begins again will be until till the error signal minimum, the output of filter will approach the drilling fluid continuous pressure ripple signal that reflects DHM-MWD information this moment, and the noise in the signal band is greatly suppressed or eliminates.

Among Fig. 3, the adaptive filter algorithm that computer consists of according to formula (4) and formula (5) computing that iterates obtains the best weights coefficient matrix of filter, comes reconstruct drilling fluid continuous pressure ripple signal by matrix coefficient and the carrier signal discrete sampling long-pending linear superposition on duty of best weights coefficient matrix.

Among Fig. 4 a, the modulating data of drilling liquid pressure DPSK original signal is encoded to [1 11111111 1], carrier frequency 20Hz, amplitude 1Pa.

Among Fig. 4 b, the white Gaussian noise mean-square value is 0.5Pa, and after drilling liquid pressure DPSK original signal was sneaked into white Gaussian noise, the PSNR power signal-to-noise ratio of signal was 1, and noise entering signal frequency spectrum makes the signal waveform indistinguishable.

Among Fig. 4 c, after the adaptive-filtering processing, Signal-to-Noise brings up to 26, and noise is eliminated substantially in the frequency band, and remaining signal noise mainly comes from outside the frequency band, can further eliminate by common band pass filter.

Among Fig. 4 d, after adaptive-filtering and band pass filter processing, out-of-band noise is completely removed, just a little some distortion of waveform of drilling liquid pressure dpsk signal, the distortion factor is 11%, but signal is subjected to the Changing Pattern of code element modulation consistent with the original signal among the accompanying drawing 4a, and signal is better recovered.

Among Fig. 5 a, the modulating data of drilling liquid pressure QPSK original signal is encoded to [0 00110110 0], carrier frequency 20Hz, amplitude 1Pa.

Among Fig. 5 b, the white Gaussian noise mean-square value is 0.5Pa, and after drilling liquid pressure QPSK original signal was sneaked into white Gaussian noise, the PSNR power signal-to-noise ratio of signal was 1, and noise entering signal frequency spectrum makes the signal waveform indistinguishable.

Among Fig. 5 c, after the adaptive-filtering processing, Signal-to-Noise brings up to 24, and noise is eliminated substantially in the frequency band, and remaining signal noise mainly comes from outside the frequency band, can further eliminate by common band pass filter.

Among Fig. 5 d, after adaptive-filtering and band pass filter processing, out-of-band noise is completely removed, just a little some distortion of waveform of drilling liquid pressure QPSK signal, the distortion factor is 7%, but signal is subjected to the Changing Pattern of code element modulation consistent with the original signal among the accompanying drawing 5a, and signal is better recovered.

Claims (1)

1. the band noise removing method of a drilling fluid continuous pressure ripple signal, it is characterized in that: adopt sef-adapting filter, to contain noisy drilling fluid continuous pressure ripple signal as the desired signal of sef-adapting filter, the carrier wave that extracts from synchronizing signal is as the input signal of filter, through type The adaptive filter algorithm that consists of with formula w (n+1)=w (n)+2 μ e (n) x (n) forms the row input matrix with carrier wave by the time sampling, in the formula,

Be best weights coefficient matrix W ^*(n) matrix coefficient, μ is the adaptive step factor of decision systems stability and convergence rate, row input matrix and filter weight coefficient column matrix multiply each other and obtain output signal, the anti-phase addition of output signal and desired signal forms error signal, error signal is sent into filter weight coefficient column matrix, automatically adjusts the matrix coefficient of filter weight coefficient column matrix with the LMS least mean square of error signal; The computing that so goes round and begins again will be until till the error signal minimum, the output valve of filter will approach the drilling fluid continuous pressure ripple signal that reflects DHM-MWD information this moment, and the noise in the signal band is greatly suppressed or eliminates.

Images