CN106371092A - Deformation monitoring method based on GPS and strong-motion seismograph observation adaptive combination - Google Patents

Abstract

The invention provides a deformation monitoring method based on GPS and strong-motion seismograph observation adaptive combination. The method is characterized by fusing high-resolution strong-motion seismograph acceleration observation to a GPS-PPP positioning module, and serving baseline drift errors of a strong-motion seismograph as unknown parameters and carrying out real-time estimation together with other positioning parameters; and meanwhile, realizing parameter estimation optimization by adaptively adjusting dynamic noise and calculating a time window, thereby not only solving the problem of correction of baseline drift, but also enhancing GPS solving intensity, reducing high-frequency noise, and realizing advantage complementation of the two observation technologies; and the method can obtain high-precision wide band deformation information in real time and has a wide application prospect. The method has the advantages of advanced algorithm, high precision, high stability and high degree of automation and the like, has overall package of a multi-satellite system of a BDS, a GPS, a GLONASS and the like, is easy to realize and can be suitable for a plurality of application occasions.

Description

A kind of deformation monitoring method being combined with strong-motion instrument observation self adaptation based on gps

Technical field

The present invention relates to a kind of Ground Deformation monitoring method.

Background technology

Gps and strong-motion instrument observation are two kinds of effective means obtaining high accuracy Ground Deformation (displacement, speed, acceleration), They are widely used to Natural calamity monitoring and differ from one another.Gps is easily obtained high precision displacement, but there is sample frequency The defect that low, signal to noise in high frequency is low, signal stabilization is poor；Meanwhile, strong-motion instrument is easily obtained high-resolution acceleration, but because of baseline The presence of drift error, the speed after its integration and displacement are commonly present deviation.Current data processing method is patrilineal line of descent with only one son in each generation sense mostly Device pattern, leads to the observation resource of multisensor not make full use of.How two classes to be observed data and carry out organic assembling, realize Have complementary advantages, the deformation data of high precision broad frequency is provided in real time, preferably serves the deformation monitorings such as disaster and there is important valency Value.

Content of the invention

In order to overcome the deficiencies in the prior art, present invention offer is a kind of to observe what self adaptation combined based on gps with strong-motion instrument Deformation monitoring method, the observation of high-resolution strong-motion instrument acceleration is dissolved in gps-ppp location model, and by strong-motion instrument Baseline drift error carries out real-time estimation as unknown parameter together with other positional parameters, passes through self-adaptative adjustment dynamic simultaneously Noise and solution time window, to realize the optimization of parameter estimation, both efficiently solve the Correction Problemss of baseline drift, simultaneously Enhance the solution intensity of gps again, reduce high-frequency noise it is achieved that the mutual supplement with each other's advantages of two kinds of observation technologies, can obtain in real time Take the deformation data of high precision broad frequency.

The technical solution adopted for the present invention to solve the technical problems is: is resolved by the fusion of two kinds of observation data and obtains Basic object information after, adaptive adjustment solution strategies, realize the excellent of parameter estimation optimization and two kinds of observation methods Gesture is complementary, specifically includes following steps:

The first step, obtains phase place/Pseudo-range Observations, the acceleration observation of strong-motion instrument and the gps data processing of gps The auxiliary product needing, including gps satellite orbit, satellite clock correction and earth rotation parameter (ERP)；

Second step, to the first step obtain the phase place/Pseudo-range Observations of gps, the acceleration observation of strong-motion instrument and The auxiliary product that gps data processing needs carries out data check, elimination of rough difference and Detection of Cycle-slip, obtains clean data；To institute State clean data and carry out the Theory of Relativity, tide, the correction of antenna phase center, troposphere and earth rotation error；

3rd step, sets up the tight integration model of gps and strong-motion instrument observation, the tight integration mould of described gps and strong-motion instrument observation Type is as follows:

${\left[\begin{array}{c}{l}_c\\ p_c\end{array}\right]}_k=\left[\begin{array}{cccc}{e}_k& m_{z,k}& 1& 1\\ e_k& m_{z,k}& 1& 0\end{array}\right]\left[\begin{array}{c}{x}_k\\ z_k\\ {\mathrm{\δt}}_k\\ b\end{array}\right]+\left[\begin{array}{c}{\mathrm{\ϵ}}_{l_c}\\ {\mathrm{\ϵ}}_{p_c}\end{array}\right],\begin{array}{c}{\mathrm{\ϵ}}_{l_c}~n(0,{\mathrm{\σ}}_{l_c}^2)\\ {\mathrm{\ϵ}}_{p_c}~n(0,{\mathrm{\σ}}_{p_c}^2)\end{array}$

$\begin{array}{cc}{a}_k={\stackrel{\·\·}{x}}_k-u_k+{\mathrm{\ϵ}}_{a_k}& {\mathrm{\ϵ}}_{a_k}~n(0,{\mathrm{\σ}}_{a_k}^2)\end{array}$

Wherein, l_cAnd p_cIt is the carrier phase residual sum pseudorange code observation after deducting geometric distance between satellite and receiver Residual error；E is the unit vector between satellite and receiver；Footmark k represents epoch number；m_z,kIt is troposphere wet stack emission zenith direction Projection function；x、It is displacement and the acceleration of receiver respectively；Z, δ t and b is respectively tropospheric zenith delay, receiver Clock correction and phase ambiguity；ε is measurement noise, ε_lcAnd ε_pcRepresent the noise of carrier wave and pseudorange observation, its corresponding variance respectively For δ²；A and u is respectively strong-motion instrument acceleration observation and baseline drift error；Baseline drift error parameter is setting width Each estimates to estimate as constant in window, and described zenith tropospheric delay is estimated as constant or shown as random Walk process, described receiver clock-offsets are estimated by epoch as white Gaussian noise, and described phase ambiguity is in continuous no week Estimate as constant in the case of jump；

The state equation of the motion of receiver and baseline drift error is as follows:

$\left[\begin{array}{c}{x}_k\\ {\stackrel{\·}{x}}_k\\ {\stackrel{\·\·}{x}}_k\\ u_k\end{array}\right]=\left[\begin{array}{cccc}i& 0.5i\mathrm{\τ}& 0.5{\mathrm{i\τ}}^2& 0\\ 0& i& i\mathrm{\τ}& 0\\ 0& 0& i& 0\\ 0& 0& 0& i\end{array}\right]\left[\begin{array}{c}{x}_{k-1}\\ {\stackrel{\·}{x}}_{k-1}\\ a_k-{\hat{u}}_{k-1}\\ u_{k-1}\end{array}\right]+w_k$

Wherein,For the velocity of receiver, i is the unit matrix of 3x3, and τ is the minimum sample rate of gps and strong-motion instrument, w_kIt is dynamic noise, its mathematic expectaion e (w_k)=0, varianceq_uFor baseline drift Power density, k-1 represents an epoch；Post-acceleration is corrected using baseline driftTo replace acceleration predictive value a_k-1；Given observation power ratio and dynamic noise；Parameter estimation is carried out using Kalman filtering, carries out residual error editor and essence simultaneously Degree statistics, obtains position, speed, acceleration, baseline drift, tropospheric delay and receiver clock-offsets information and the precision of receiver Index；

4th step, the solution information according to the 3rd step calculates the time serieses of baseline drift, and builds four study statistics Amount, determines the standard of statistic；Four statistics are respectively as follows:

The Influence of Displacement of baseline drift

The Influence of Displacement of original acceleration

The Influence of Displacement of the acceleration after correction

Signal intensity

Wherein, t₁、t₂It is the initial time of estimation window and the termination time of definition,It is respectively initial time With terminate time corresponding epoch number,It is the average of a, standard a of each study statistic₀, b₀, c₀, d₀It is the static observation phase Between each learn statistic three times standard deviation in setting time sequence；

Baseline drift error parameter is divided into four-stage according to the standard that the 4th step determines, and calculates new by the 5th step Dynamic noise and time Estimate window；Being defined below of four-stage:

Initial phase: meet (a₂&b₁) or (b₁&c₂),

Process segment: meet (a₁),

The instantaneous baseline drift stage: meet (a₂&d₂),

The permanent baseline drift stage: meet (a₂&d₁)&(b₂&c₁)；

In the instantaneous baseline drift stage, dynamic noise adjustment is as follows:

$q_u^{\′}=\sqrt{\frac{|u_{t_2}-u_{t_1}|}{(t_2-t_1)}}$

Wherein, q '_uNew power density for baseline drift；Baseline drift estimates that the length of window changes into per second one Secondary；For other three phases, dynamic noise q_u=0.0001-0.001m^1/2/s^3/2, length of window remains 5 seconds；

6th step, according to the new dynamic noise determining in the 5th step and new time Estimate window, using Kalman's filter Ripple carries out self adaptation combination and resolves, and obtains optimum combination result.

The invention has the beneficial effects as follows:

First, realize the mutual supplement with each other's advantages of two kinds of observation methods, the deformation data of high precision broad frequency is provided.

High accuracy gps and the observation of high-resolution strong-motion instrument are carried out tight integration process by the present invention, compensate for gps technology noise Than difference and strong-motion instrument technology baseline drift limitation it is achieved that their mutual supplement with each other's advantages, a combination thereof result can export in real time The displacement of high precision broad frequency, speed, acceleration information, may be directly applied to the deformation monitorings such as disaster.

Second, effectively identify and correct the baseline drift of strong-motion instrument, be that the correction of baseline drift provides new way.

The identification of the baseline drift error of strong-motion instrument is correction is seismic field stubborn problem, generally adopts the side of experience Method is completed, but the method for experience is unable to the characteristic of accurate description baseline drift error, and the result after correction has different journeys The system deviation of degree, and can not carry out in real time.The baseline drift error of strong-motion instrument is carried out modelling parameters estimation by the present invention, both Can effectively identify and correct the baseline drift of strong-motion instrument, provide new way for the correction of baseline drift, to rotation simultaneously Seismographic research is significant.

3rd, effectively reduce the observation noise of gps, accelerate the convergence rate of gps positioning.

High-resolution acceleration measurement fusion, in gps positioning equation, can constrain gps and solve intensity, accelerate gps fixed The convergence rate of position, can effectively reduce the noise of gps simultaneously, improve gps positioning precision and stability.

Brief description

Fig. 1 is the four-stage schematic diagram during tight integration is processed；

Fig. 2 is that gps combines block diagram with strong-motion instrument observation；

Fig. 3 is that gps combines block diagram with strong-motion instrument self adaptation.

Specific embodiment

The present invention is further described with reference to the accompanying drawings and examples, and the present invention includes but are not limited to following enforcements Example.

The present invention is dissolved in gps-ppp location model by the observation of high-resolution strong-motion instrument acceleration, and by strong-motion instrument Baseline drift error carry out real-time estimation together with other positional parameters as unknown parameter, simultaneously pass through self-adaptative adjustment move State noise and solution time window, to realize the optimization of parameter estimation, both efficiently solve the Correction Problemss of baseline drift, with When enhance the solution intensity of gps again, reduce high-frequency noise it is achieved that the mutual supplement with each other's advantages of two kinds of observation technologies, can be real-time Obtain the deformation data of high precision broad frequency.

Technical scheme mainly includes two big core algorithms:

(1) gps and the tight integration of strong-motion instrument observation are processed

The tight integration of gps and strong-motion instrument observation is using following observational equation:

${\left[\begin{array}{c}{l}_c\\ p_c\end{array}\right]}_k=\left[\begin{array}{cccc}{e}_k& m_{z,k}& 1& 1\\ e_k& m_{z,k}& 1& 0\end{array}\right]\left[\begin{array}{c}{x}_k\\ z_k\\ {\mathrm{\δt}}_k\\ b\end{array}\right]+\left[\begin{array}{c}{\mathrm{\ϵ}}_{l_c}\\ {\mathrm{\ϵ}}_{p_c}\end{array}\right],\begin{array}{c}{\mathrm{\ϵ}}_{l_c}~n(0,{\mathrm{\σ}}_{l_c}^2)\\ {\mathrm{\ϵ}}_{p_c}~n(0,{\mathrm{\σ}}_{p_c}^2)\end{array}---\left(1\right)$

$\begin{array}{cc}{a}_k={\stackrel{\·\·}{x}}_k-u_k+{\mathrm{\ϵ}}_{a_k}& {\mathrm{\ϵ}}_{a_k}~n(0,{\mathrm{\σ}}_{a_k}^2)\end{array}---\left(2\right)$

Wherein, l_cAnd p_cIt is the carrier phase after deducting geometric distance between satellite and receiver and pseudorange code observation residual error, E is the unit vector between satellite and receiver, and k represents epoch.M is the x to flow process wet stack emission projection function,It is to receive respectively Seat in the plane is moved, acceleration, and z, δ t and b is respectively tropospheric zenith delay, receiver clock-offsets and phase ambiguity, and ε is measurement Noise, its corresponding variance is δ².A and u is respectively strong-motion instrument acceleration observation and baseline drift error, and k represents epoch number. Baseline drift parameter estimates to estimate as constant, other parameters, such as zenith tropospheric delay, in the short time in window at each Interior (hour or two hours) can be estimated as constant, or shows as random walk process, and receiver clock Difference is estimated by epoch as white Gaussian noise, and fuzziness parameter is estimated as constant in the case of continuous no cycle slip.

The state equation of station velocity and baseline drift is expressed as follows:

$\left[\begin{array}{c}{x}_k\\ {\stackrel{\·}{x}}_k\\ {\stackrel{\·\·}{x}}_k\\ u_k\end{array}\right]=\left[\begin{array}{cccc}i& 0.5i\mathrm{\τ}& 0.5{\mathrm{i\τ}}^2& 0\\ 0& i& i\mathrm{\τ}& 0\\ 0& 0& i& 0\\ 0& 0& 0& i\end{array}\right]\left[\begin{array}{c}{x}_{k-1}\\ {\stackrel{\·}{x}}_{k-1}\\ a_k-{\hat{u}}_{k-1}\\ u_{k-1}\end{array}\right]+w_k,---\left(3\right)$

WhereinFor velocity, i is the unit matrix of 3x3, and τ is the minimum sample rate of gps and strong-motion instrument, w_kIt is dynamic Noise, its mathematic expectaion e (w_k)=0, varianceq_uClose for the power of baseline drift Degree.K-1 represents an epoch, adopts baseline drift to correct post-acceleration hereTo replace acceleration pre- Measured value a_k-1, it is to avoid acceleration and baseline drift have big dynamic noise.Application formula (1), (2) and (3), give (experience weight to be determined observation power ratio by measurement noise here, and common span is respectively as follows: with dynamic noiseDynamic noise is given as constant, q_u= 0.003～0.01m^1/2/s^3/2, widow time length is the 1-10 second), available Kalman filtering carries out parameter estimation and both can obtain Position, speed, acceleration, baseline drift, tropospheric delay and receiver clock-offsets.In addition, above-mentioned stochastic model can be adaptively Adjustment makes parameter estimation optimization.

(2) self adaptation combined treatment observed by gps and strong-motion instrument

Robust iterative for gps and the tight integration of strong-motion instrument observation is processed, and needs to solve two key issues.First It is whether baseline drift occurs？If it happens, then need to estimate the persistent period of baseline drift.Second be baseline drift change Change situation, then present invention may determine that the optimal dynamic noise of state vector.

In order to solve these problems, process frequently with adaptive-filtering.The statistic defining some study first is detecting Whether baseline drift occurs, and determines whether it is notable, then adjusts dynamic noise to realize the ART network of parametric solution. In field of satellite navigation, these statistics are normally based on state difference, component of variance and prediction residual to define.But they Gps can not be applied to well and macroseism tight integration is processed.On the one hand, gps and strong-motion instrument tight integration are processed, introduce base Line drift and phase ambiguity parameter, the geometry parameter deviation of the finite observation amount estimation of single epoch is larger or even can not estimate Meter.In this case, the statistical property of state difference XOR component of variance can be very poor.On the other hand, the baseline drift of STRONG MOTION DATA Move very sensitive to speed displacement after integration.Although the value that the value of baseline drift compares peak accelerator is typically very little, But they appreciable impact is integrated after speed and displacement.Additionally, baseline drift error is mainly made by ground inclination or rotation Become, and have a key character, be zero or steady state value when not moving, therefore, select baseline drift as study statistic More particularly suitable.Furthermore, high-resolution acceleration signal can be used for detecting kinestate.Here first assume under static state Baseline drift and acceleration are zero, then define four study statistics, to recognize whether baseline drift and baseline Drift situation of change.Because strong-motion instrument sample rate is more much higher than gps, gps sampling interval duration is selected to float as baseline Move the length estimating window.Meanwhile, for avoiding the impact of random error, whole baseline drift estimate window by baseline drift and Integrated acceleration becomes displacement to be analyzed.

Statistic 1:

Statistic 2:

Statistic 3:

Statistic 4:

Wherein t₁, t₂It is the estimation time window (being commonly defined as the multiple in sampling interval) of definition,For epoch Number,It is the average of a, a represents the Influence of Displacement of baseline drift, b represents the Influence of Displacement of original acceleration, c represents after correcting Acceleration Influence of Displacement, d representation signal intensity.Each study statistic is according to standard a of definition₀, b₀, c₀, d₀Have two The situation of kind.These standards are during static observation, and by the time serieses of each study statistic (i.e. a, b, c, d), (3-5 divides Clock) three times standard deviation be calculated.

Based on defined statistic and standard, baseline drift valuation self-adapting estimation can be analyzed.They can divide Become four-stage, as shown in Figure 1.

Initial phase (1): meet (a₂&b₁) or (b₁&c₂), the 0-47 second in Fig. 1 .b.The beginning of data calculation.Static State, original acceleration almost nil (as Fig. 1 .a), but the baseline drift resolving is not zero (as Fig. 1 .b).Resolve baseline drift Require time for restraining, complete to initialize.The dynamic noise of baseline drift should be very little value.

Process segment (2): meet (a₁), the 47-95 second in Fig. 1 .b.During this period, the baseline drift of resolving is almost Zero, the dynamic noise of baseline drift is also very little.

The instantaneous baseline drift stage (3): meet (a₂&d₂), the 95-150 second in Fig. 1 .b.During this period, baseline is detected Drift, and signal is notable；The dynamic noise of baseline drift should be adjusted to larger value.

The permanent baseline drift stage (4): meet (a₂&d₁)&(b₂&c₁), 150 in Fig. 1 .b second are to last.During this period, Also detect that baseline drift, but there is no significant signal, the dynamic noise of baseline drift should be less value.

In the time interval instantaneous baseline drift is detected, dynamic noise adjustment is as follows:

$q_u^{\′}=\sqrt{\frac{|u_{t_2}-u_{t_1}|}{(t_2-t_1)}}---\left(8\right)$

Wherein q '_uFor the new power density of baseline drift, reflect the actual rate of change of acceleration baseline drift.Separately Outward, baseline drift estimates that the length of window can have the adaptive adjustment of situation according to signal: if there is no signal, such as feelings Condition d₁When, baseline drift is mainly caused by environmental change, and changes slow, so the present invention can be at several seconds or longer Time interval is estimating baseline drift；Otherwise, as situation d₂, baseline drift is mainly by inclination and/or the rotation of ground motion Cause, and change very fast, the frequency of therefore estimation should be one second or less time interval, to find baseline drift Change and adjust dynamic noise, the present invention uses one second and estimates once.

Embodiments of the invention comprise the following steps:

The first step, gps is processed with the tight integration of strong-motion instrument observation, as shown in Figure 2.Be divided into data input, data processing and Result exports three nucleus modules.In data input module, it is desirable to provide the phase place/Pseudo-range Observations of gps, the acceleration of strong-motion instrument Degree observation, and auxiliary product such as track, clock correction, the earth rotation parameter (ERP) product that gps data processing needs.In data calculation Module, is the pretreatment of data first, carries out data check, and elimination of rough difference and Detection of Cycle-slip obtain clean data；Subsequently, enter Row the Theory of Relativity, tide, antenna phase center, troposphere, the correction work of earth rotation error；Then carry out tight integration model Set up, the baseline drift error of strong-motion instrument is estimated as unknown parameter together with other positional parameters；Finally carry out residual error volume Collect and precision statisticses obtain combination solution (substantially combining solution).

Second step, gps and strong-motion instrument self adaptation combined treatment, as shown in Figure 3.Self adaptation combined treatment adopts two karr Graceful wave filter is realizing.First wave filter is used for basic combination and resolves, and is the work of the first step, for judging kinestate With the length determining optimal dynamic noise and baseline drift estimation window.Second wave filter is used for self adaptation combination solution Calculate, to obtain sane combined result.By experience weight and dynamic noise, (experience weight to be determined by measurement noise here, leads to Normal span is respectively as follows: such as Dynamic State noise is given as constant, such as q_u=0.003-0.01m^1/2/s^3/2, widow time length is the 1-10 second), take first first Individual wave filter, to estimate baseline drift, then calculates and determines its time series and four kinds of study statistical data standards.Pass through Analysis, if it is determined that being the instantaneous baseline drift stage, dynamic noise is changed according to the change of instantaneous baseline drift, baseline drift Estimate that the length of window changes into once per second.During other states, dynamic noise is limited to a less value (value Scope is q_u=0.0001-0.001m^1/2/s^3/2), length of window remains 5 seconds.Then, according to the optimal dynamic noise determining With time Estimate window, carry out self adaptation combination using second wave filter and resolve, to obtain optimum combination result.

Claims (1)

1. a kind of deformation monitoring method being combined with strong-motion instrument observation self adaptation based on gps is it is characterised in that comprise the steps:

The first step, the phase place/Pseudo-range Observations, the acceleration observation of strong-motion instrument and the gps data processing that obtain gps need Auxiliary product, including gps satellite orbit, satellite clock correction and earth rotation parameter (ERP)；

Second step, the phase place/Pseudo-range Observations of gps that the first step is obtained, the acceleration observation of strong-motion instrument and gps number Carry out data check, elimination of rough difference and Detection of Cycle-slip according to processing the auxiliary product needing, obtain clean data；To described clean Data carry out the Theory of Relativity, tide, the correction of antenna phase center, troposphere and earth rotation error；

3rd step, sets up the tight integration model of gps and strong-motion instrument observation, the tight integration model of described gps and strong-motion instrument observation is such as Under:

${\left[\begin{array}{c}{l}_c\\ p_c\end{array}\right]}_k=\left[\begin{array}{cccc}{e}_k& m_{z,k}& 1& 1\\ e_k& m_{z,k}& 1& 0\end{array}\right]\left[\begin{array}{c}{x}_k\\ z_k\\ {\mathrm{\δt}}_k\\ b\end{array}\right]+\left[\begin{array}{c}{\mathrm{\ϵ}}_{l_c}\\ {\mathrm{\ϵ}}_{p_c}\end{array}\right],\begin{array}{c}{\mathrm{\ϵ}}_{l_c}~n(0,{\mathrm{\σ}}_{l_c}^2)\\ {\mathrm{\ϵ}}_{p_c}~n(0,{\mathrm{\σ}}_{p_c}^2)\end{array}$

$\begin{array}{cc}{a}_k={\stackrel{\·\·}{x}}_k-u_k+{\mathrm{\ϵ}}_{a_k}& {\mathrm{\ϵ}}_{a_k}~n(0,{\mathrm{\σ}}_{a_k}^2)\end{array}$

Wherein, l_cAnd p_cIt is the carrier phase residual sum pseudorange code observation residual error after deducting geometric distance between satellite and receiver； E is the unit vector between satellite and receiver；Footmark k represents epoch number；m_z,kIt is troposphere wet stack emission zenith direction projection letter Number；x、It is displacement and the acceleration of receiver respectively；Z, δ t and b be respectively tropospheric zenith delay, receiver clock-offsets and Phase ambiguity；ε is measurement noise, ε_lcAnd ε_pcRepresent the noise of carrier wave and pseudorange observation respectively, its corresponding variance is δ²；a It is respectively strong-motion instrument acceleration observation and baseline drift error with u；Baseline drift error parameter set width each estimate Estimate as constant in meter window, described zenith tropospheric delay is estimated as constant or shown as random walk Journey, described receiver clock-offsets are estimated by epoch as white Gaussian noise, and described phase ambiguity is in continuous no cycle slip situation Lower as constant estimation；

The state equation of the motion of receiver and baseline drift error is as follows:

$\left[\begin{array}{c}{x}_k\\ {\stackrel{\·}{x}}_k\\ {\stackrel{\·\·}{x}}_k\\ u_k\end{array}\right]=\left[\begin{array}{cccc}i& 0.5i\mathrm{\τ}& 0.5{\mathrm{i\τ}}^2& 0\\ 0& i& i\mathrm{\τ}& 0\\ 0& 0& i& 0\\ 0& 0& 0& i\end{array}\right]\left[\begin{array}{c}{x}_{k-1}\\ {\stackrel{\·}{x}}_{k-1}\\ a_k-{\hat{u}}_{k-1}\\ u_{k-1}\end{array}\right]+w_k$

Wherein,For the velocity of receiver, i is the unit matrix of 3x3, and τ is the minimum sample rate of gps and strong-motion instrument, w_kIt is Dynamic noise, its mathematic expectaion e (w_k)=0, varianceq_uWork(for baseline drift Rate density, k-1 represents an epoch；Post-acceleration is corrected using baseline driftTo replace acceleration predictive value a_k-1； Given observation power ratio and dynamic noise；Parameter estimation is carried out using Kalman filtering, carries out residual error editor and precision system simultaneously Meter, obtains position, speed, acceleration, baseline drift, tropospheric delay and the receiver clock-offsets information of receiver and precision refers to Mark；

4th step, the solution information according to the 3rd step calculates the time serieses of baseline drift, and builds four study statistics, really Determine the standard of statistic；Four statistics are respectively as follows:

The Influence of Displacement of baseline drift

The Influence of Displacement of original acceleration

The Influence of Displacement of the acceleration after correction

Signal intensity

Wherein, t₁、t₂It is the initial time of estimation window and the termination time of definition,It is respectively initial time and termination Time corresponding epoch number,It is the average of a, standard a of each study statistic₀, b₀, c₀, d₀It is each during static observation Study three times standard deviation in setting time sequence for the statistic；

Baseline drift error parameter is divided into four-stage according to the standard that the 4th step determines by the 5th step, and calculates new dynamic Noise and time Estimate window；Being defined below of four-stage:

Initial phase: meet (a₂&b₁) or (b₁&c₂),

Process segment: meet (a₁),

The instantaneous baseline drift stage: meet (a₂&d₂),

The permanent baseline drift stage: meet (a₂&d₁)&(b₂&c₁)；

In the instantaneous baseline drift stage, dynamic noise adjustment is as follows:

$q_u^{\′}=\sqrt{\frac{|u_{t_2}-u_{t_1}|}{(t_2-t_t)}}$

Wherein, q '_uNew power density for baseline drift；Baseline drift estimates that the length of window changes into once per second；For Other three phases, dynamic noise q_u=0.0001-0.001m^1/2/s^3/2, length of window remains 5 seconds；

6th step, according to the new dynamic noise determining in the 5th step and new time Estimate window, is entered using Kalman filtering The combination of row self adaptation resolves, and obtains optimum combination result.