CN104199054A - Preprocessing method for common view data of Beidou satellite navigation system - Google Patents

Abstract

The invention discloses a preprocessing method for common view data of a Beidou satellite navigation system. The method includes the steps of selecting proper common view data according to an observation state of a time frequency laboratory and a satellite state; correcting the delay during signal transmission of the selected common view data; performing gross error elimination on the corrected common view data; smoothing the common view data through a k-order Vondrak filter, selecting an optimal smoothing factor through iterative and least squares estimation, and eliminating random errors. The preprocessing method for the common view data of the Beidou satellite navigation system is low in calculation amount, high in computation speed and capable of quickly obtaining the proper smoothing factor and effectively eliminating the random errors in the common view data.

Description

A kind of preprocess method of looking altogether data for Beidou satellite navigation system

Technical field

The present invention relates to data processing method, particularly relate to a kind of data preprocessing method for Beidou satellite navigation system common-view time Frequency Transfer.

Background technology

In Beidou satellite navigation system common-view time Frequency Transfer process, two time-frequency laboratories that participate in looking are altogether at same big-dipper satellite of synchronization observation, obtain respectively the time difference data of local clock and the spaceborne clock of this big-dipper satellite separately, through communication link, the data that record are separately exchanged mutually, through data processing, remove spaceborne clock and can obtain the time difference between two time-frequency laboratories, thereby realize the transmission of Big Dipper common-view time.

Owing to existing and disturbing when receiving Big Dipper signal, for obtaining more accurate local clock and the spaceborne clock time difference, before carrying out exchanges data, need to carry out in this locality data pre-service, comprise: altogether depending on the choosing of data, the correction of signal transmission delay, the eliminating of stochastic error.Signal transmission delay correction refers to be revised geometric delays, spatial medium time-delay, Relativistic Time Delay, receiver hardware delay etc. in time difference data, can realize by setting up the mode of model and calibration, and the weakening of stochastic error can only realize by the mode of data smoothing.

Data smoothing is very important work of Data processing, and user will obtain effective information from measurement data, must do pre-service to measurement data, reduces as far as possible the impact of stochastic error, the quality of simultaneously all right assessment of metrology data.Conventional smoothing method has curve, running mean method and Vondrak smoothing method at present.The prerequisite of curve-fitting method is to provide functional form, and great majority are polynomial expression, and its exponent number is often rule of thumb determined, and human factor impact is larger; Running mean method require data must be equidistantly and also sampled point than comparatively dense, there is significant limitation.Vondrak smoothing method essence is by selecting smoothing factor, to the absolute matching of measurement sequence with definitely smoothly, seeking a balance, the in the situation that of unknown fitting function form, Vondrak method also can be carried out reasonably smoothly survey data, and its key is to select rational smoothing factor.Conventional selection smoothing factor method has at present: observational error method, frequency response method, cross-certification method.Observational error method requires to know in advance observation data priori standard deviation, and we can not know standard deviation information accurately conventionally; Frequency response method can obtain good filter effect for the observation data of knowing frequency or cycle; Cross-certification method is to select smoothing factor by test repeatedly, and operand is large, and speed is slow, and the smoothing factor randomness of selection is also larger.

Therefore, need to provide a kind of preprocess method of looking altogether data, with fast selecting smoothing factor, reduce operand, improve travelling speed.

Summary of the invention

The technical problem to be solved in the present invention is to provide a kind of data preprocessing method for Big Dipper common-view time Frequency Transfer, solve and to look altogether at present Data processing, in the undesirable and traditional Vondrak smoothing method of conventional smoothing method smooth effect, smoothing factor is difficult to the problem of selecting.

For solving the problems of the technologies described above, the present invention adopts following technical proposals.

For Beidou satellite navigation system, look altogether a preprocess method for data, the method step comprises

S1, according to the observer state in time-frequency laboratory and satellitosis, choose the suitable data of looking altogether;

S2, to revising depending on the delay in data signal propagation process altogether after choosing;

S3, to revised, depending on data, carry out gross error altogether and reject to process;

S4, employing k rank Vondrak wave filter, to carrying out smoothing processing depending on data altogether, are chosen the postfitted orbit factor by least squares estimate iteration, eliminate stochastic error.

Preferably, in described step S1, the condition of choosing depending on data is altogether comprised

Guarantee that two time-frequency laboratories are at same satellite of synchronization observation;

The elevation angle that is observed satellite is not less than 15 °;

According to Beidou satellite navigation system feature, extend total length tracking time.

Preferably, need the delay item of revising to comprise geometric distance delay, ionosphere delay, tropospheric delay in described step S2, sagnac effect and receiver postpone.

Preferably, Deferred Correction method comprises

Utilize dual-frequency data to revise ionosphere delay;

Delay parameter and the satellite ephemeris modeling of according to Beidou satellite navigation system receiver, in row text, extracting, eliminate geometric distance delay, tropospheric delay and sagnac effect;

Mode by calibrating receiver overcomes receiver and postpones.

Preferably, in described step S3, adopt 3 σ criterions to reject and look altogether the gross error in data.

Preferably, adopt least square process of iteration to choose smoothing factor in described step S4, its step comprises

S41, the time difference sequence of making even after sliding are estimated as waiting, by Vondrak filtering ultimate principle, using time difference sequence and time difference sequence k jump divide as two class observed readings, set up respectively error equation, thereby Vondrak filtering be converted into least square adjustment problem;

S42, given initial smoothing factor, utilize least square method to solve the error equation of observed reading, obtains once level and smooth result;

S43, the method for utilizing Helmert component of variance to estimate, carry out estimation variance component according to the correction vector V last time smoothly obtaining, and revises initial smoothing factor;

S44, by iteration repeatedly, obtain the postfitted orbit factor, Simultaneous Iteration finishes to complete elimination stochastic error.

Beneficial effect of the present invention is as follows:

Technical scheme calculated amount of the present invention is little, and fast operation can obtain suitable smoothing factor fast, can effectively eliminate the stochastic error of looking altogether in data.

Accompanying drawing explanation

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail;

Fig. 1 illustrates the schematic diagram of a kind of data preprocessing method for Big Dipper common-view time Frequency Transfer of the present invention.

Embodiment

Below in conjunction with one group of embodiment and accompanying drawing, the present invention is described further.

The invention discloses a kind of data preprocessing method transmitting for Big Dipper common-view time, the method concrete steps comprise:

The first step, choosing depending on data altogether

In Big Dipper common-view time Transfer Technology, choose while looking data altogether, except what receive, be same satellite-signal, also must meet three conditions below simultaneously:

1) tracking is constantly identical, and same satellite must be observed at synchronization in two time-frequency laboratories, so just can eliminate the error of satellite clock completely;

2) elevation angle is not less than 15 °, and when elevation of satellite is too low, ionosphere delay error is difficult to correct, and Multi-Path Effects aggravation;

3) follow the tracks of length, for GPS, one time total length is tracked as 13 minutes, add preset preparation in 2 minutes and the processing of 1 minute data, for the Big Dipper, look altogether, because Beidou satellite navigation system is containing GEO and IGSO satellite, can total length tracking time of proper extension.

Second step, signal transmission delay correction

Choose and look altogether after data, the every delay of reply satellite-signal in communication process, comprises geometric distance delay, ionosphere delay, tropospheric delay, and sagnac effect and receiver delay etc. is revised, and is deducted in altogether depending on data.Wherein, ionosphere delay can utilize dual-frequency data correction, geometric distance delay, tropospheric delay, sagnac effect etc. can be eliminated according to modelings such as the delay parameter extracting from Beidou receiver navigation message, satellite ephemeris, and receiver postpones by the mode of calibrating, to obtain.

The 3rd step gross error is rejected

Because Vondrak is smoothly subject to the impact of gross error, before carrying out data smoothing, first adopt 3 σ criterions to reject and look altogether the gross error in data.Owing to being once per second depending on data altogether, twice measurement in front and back is more or less the same, therefore data mean value substitutes the rough error weeding out before and after can directly utilizing rough error.

The 4th step Vondrak is level and smooth

Adopt method that Vondrak is level and smooth to carry out data smoothing depending on data altogether after signal propagation delays correction and elimination of rough difference, eliminate stochastic error.

The ultimate principle of the level and smooth measurement data of Vondrak method is:

Q＝F+λ ²S＝min (1)

Wherein, F=Σ p _i(y _i'-y _i) ², S=Σ (Δ ^ky _i) ², y _i(i=1,2 ..., N) for time argument is x _isurvey data.Y _i' be smooth value to be asked, p _iweight for measurement data.

This programme adopts k rank Vondrak wave filter as follows to looking altogether smoothed data modeling:

with on interval between 2, with one, pass through i, i+1 ..., the i+k altogether k rank lagrange polynomial of k+1 point defines level and smooth result y '=y ' (t), and corresponding, the k jump of smooth value divides and can be expressed as

$S=\underset{i=1}{\overset{n-k}{\mathrm{\Σ}}}{(a_{i0}{y}_i^{\′}+a_{i1}{y}_{i+1}^{\′}+\·\·\·+a_{\mathrm{ik}}{y}_{i+k}^{\′})}^2---\left(2\right)$

Wherein,

$a_{i0}=\frac{k!\sqrt{t_{i+\left[\frac{k}{2}\right]+1}{-t}_{i+\left[\frac{k}{2}\right]}}}{(t_i-t_{i+1})\·\·\·(t_i-t_{i+k})\sqrt{t_{n-\left[\frac{k-1}{2}\right]}{-t}_{\left[\frac{k}{2}\right]}}},i=\mathrm{1,2},\·\·\·,n-k$

$a_{\mathrm{ik}}=\frac{k!\sqrt{t_{i+\left[\frac{k}{2}\right]+1}{-t}_{i+\left[\frac{k}{2}\right]}}}{(t_{i+k}-t_i)\·\·\·(t_{i+k}-t_{i+k-1})\sqrt{t_{n-\left[\frac{k-1}{2}\right]}{-t}_{\left[\frac{k}{2}\right]}}}$

Make y _ifor observed reading, y _i' be to be valuated, by formula (1), set up two class error equations:

$\begin{array}{c}{V}_1=y^{\′}-y\\ V_2={\mathrm{\Δ}}^k{y}^{\′}\end{array}---\left(3\right)$

Make first kind observed reading power for P ₁, the second kind observation value power is P ₂.

According to least square, have:

$V^T\mathrm{PV}={\left[\begin{array}{c}{V}_1\\ V_2\end{array}\right]}^T\left[\begin{array}{cc}{P}_1& 0\\ 0& P_2\end{array}\right]\left[\begin{array}{c}{V}_1\\ V_2\end{array}\right]=\min ---\left(4\right)$

That is:

$V_1^T{P}_1{V}_1+V_2^T{P}_2{V}_2=\min ---\left(5\right)$

Make P ₁=ε p, P ₂=I:

$\mathrm{\ϵ}\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{p}_i{(y_i^{\′}-y_i)}^2+\underset{i=1}{\overset{n-k}{\mathrm{\Σ}}}{\left({\mathrm{\Δ}}^k{y}_i^{\′}\right)}^2=\min ---\left(6\right)$

Order above formula is consistent with Vondrak filtering ultimate principle, and Vondrak is smoothly converted into the described least square adjustment problem in formula (3)～(5).

Get quantity of state X=[y ₁' y ₂' ... y _n'] ^t, formula (3) can be expressed as:

$\underset{2n-k,1}{V}=\left[\begin{array}{c}\underset{n,1}{V_1}\\ \underset{n-k,1}{V_2}\end{array}\right]=\left[\begin{array}{c}\underset{n,n}{B_1}\\ \underset{n-k,n}{B_2}\end{array}\right]\underset{n,1}{X}-\left[\begin{array}{c}\underset{n,1}{f_1}\\ \underset{n-k,1}{f_2}\end{array}\right]---\left(7\right)$

Wherein, f ₁=[y ₁y ₂y _n] ^t, f ₂=[0 0 ... 0] ^t

According to the principle of least square, solve, obtain the estimated value of quantity of state

$\underset{n,1}{X}=\underset{n,n}{N^{-1}}\underset{n,1}{f_e}---\left(8\right)$

Wherein, $N=N_1+N_2=B^T\mathrm{PB}=B_1^T{P}_1{B}_1+B_2^T{P}_2{B}_2$

$f_e=f_{e1}+f_{e2}=B^T\mathrm{Pf}=B_1^T{P}_1{f}_1+B_2^T{P}_2{f}_2$

The 5th step component of variance is estimated, right value update

From the 4th step, Vondark filtering is converted into the least square adjustment problem with the different observed readings of two classes, its key is the appropriate power of choosing all kinds of observed readings, for this reason, the method of utilizing Helmert component of variance to estimate, according to step 3 gained adjustment result, calculate correction vector V and carry out estimation variance component, revise adjustment weights used.Step is as follows:

For given initial smoothing factor ε=1/ λ ², test accordingly front weights and be respectively P ₁=ε p, P ₂=I, carries out pre-adjustment according to formula (8), obtains p ₁v ₁, p ₂v ₂, can be calculated component of variance estimated value and

${\mathrm{\σ}}_{0i}^2=\frac{V_i^T{P}_i{V}_i}{n_i-\mathrm{tr}\left(N^{-1}{N}_i\right)}---\left(9\right)$

Rear weights are tested in calculating

$\begin{array}{c}{P}_1^{\′}=C/{\mathrm{\σ}}_{01}^2\×P_1\\ P_2^{\′}=C/{\mathrm{\σ}}_{02}^2\×P_2\end{array}---\left(10\right)$

Wherein, C is constant.

The 6th step iteration is selected the postfitted orbit factor

Get smoothing factor can find out, each iteration is all to P ₁, smoothing factor ε is revised.Using revised weights as testing the new adjustment of front weight, repeatedly carry out, adjustment-> component of variance estimation-> right value update is adjustment again, until twice of front and back iteration smoothing factor is consistent.

The 7th step iteration finishes, and obtains the pretreated data of looking altogether.

While utilizing the present invention to adopt k rank Vondrak wave filter to carry out data smoothing, increase along with filter order, its frequency response curve steepening, the effect of cross frequence component is better, but calculated amount is larger, and require sampling density larger, at the Big Dipper, look altogether Data processing, between local clock and the spaceborne clock of the Big Dipper, time difference data is once per second, it is 20min that one secondary tracking duration is set, each tracking period need to be carried out data smoothing to 1200 data, and calculated amount is larger, therefore select three rank Vondrak wave filters to carry out data smoothing.In addition, consider at one and follow the tracks of in the period, because receiver acquisition satellite-signal needs certain hour, before and after each comparison period, the data error of 2min is larger, for reducing rough error for the impact of level and smooth result, each can be compared to the data of period front and back 2min and remove, then do smoothing processing.

In sum, technical scheme of the present invention.

Obviously; the above embodiment of the present invention is only for example of the present invention is clearly described; and be not the restriction to embodiments of the present invention; for those of ordinary skill in the field; can also make other changes in different forms on the basis of the above description; here cannot give all embodiments exhaustive, every still row in protection scope of the present invention of apparent variation that technical scheme of the present invention extends out or change that belong to.

Claims (6)

1. for Beidou satellite navigation system, look altogether a preprocess method for data, it is characterized in that, the method step comprises

S1, according to the observer state in time-frequency laboratory and satellitosis, choose the suitable data of looking altogether;

S2, to revising depending on the delay in data signal propagation process altogether after choosing;

S3, to revised, depending on data, carry out gross error altogether and reject to process;

S4, employing k rank Vondrak wave filter, to carrying out smoothing processing depending on data altogether, are chosen the postfitted orbit factor by least squares estimate iteration, eliminate stochastic error.

2. data preprocessing method according to claim 1, is characterized in that, in described step S1, the condition of choosing depending on data is altogether comprised

Guarantee that two time-frequency laboratories are at same satellite of synchronization observation;

The elevation angle that is observed satellite is not less than 15 °;

According to Beidou satellite navigation system feature, extend total length tracking time.

3. data preprocessing method according to claim 1, is characterized in that, needs the delay item of revising to comprise geometric distance delay, ionosphere delay, tropospheric delay in described step S2, and sagnac effect and receiver postpone.

4. data preprocessing method according to claim 3, is characterized in that, Deferred Correction method comprises

Utilize dual-frequency data to revise ionosphere delay;

Delay parameter and the satellite ephemeris modeling of according to Beidou satellite navigation system receiver, in row text, extracting, eliminate geometric distance delay, tropospheric delay and sagnac effect;

Mode by calibrating receiver overcomes receiver and postpones.

5. data preprocessing method according to claim 1, is characterized in that, adopts 3 σ criterions to reject and look altogether the gross error in data in described step S3.

6. data preprocessing method according to claim 1, is characterized in that, adopts least square process of iteration to choose smoothing factor in described step S4, and its step comprises

S41, the time difference sequence of making even after sliding are estimated as waiting, by Vondrak filtering ultimate principle, using time difference sequence and time difference sequence k jump divide as two class observed readings, set up respectively error equation, thereby Vondrak filtering be converted into least square adjustment problem;

S42, given initial smoothing factor, utilize least square method to solve the error equation of observed reading, obtains once level and smooth result;

S43, the method for utilizing Helmert component of variance to estimate, carry out estimation variance component according to the correction vector V last time smoothly obtaining, and revises initial smoothing factor;

S44, by iteration repeatedly, obtain the postfitted orbit factor, Simultaneous Iteration finishes to complete elimination stochastic error.